INATIONAL CENTER Se@- 1

For HEALTH STATISTICS Number 1

VITAL and HEALTH STATISTICS

PROGRAMS AND COLLECTION PROCEDURES

Origin, Program, and

Operation of the U.S.

National Health Survey

A description of the developments leading to en-

actment of the Nationa I Health Survey Act ~ and

a summary of the ~licies, initial program, and

operation of the Survey.

Washington, D.c. Reprinted April 1965

U.S. DEPARTMENT OF

HEALTH, EDUCATION , AND WELFARE

Anthony J. Celebrezze

Secretary

Public Health Service

Luther L. Terry

Surgeon Genero I

.-.

.-

1.1

NATIONAL CENTER FOR HEALTH STATISTICS

Forrest E. Linder, Ph. D., DirectorThecdore D. Woolsey, Assistant Director

O. K. Sagen, Ph. D., Assistant DirectorCarl C. Dauer, M. D., Medical Advisor

Louis R. Stolcis, M.A., Executive Officer

OFFICE OF ELECTRONIC SYSTEMS

Charles E. Greene, Chief

OFFICE OF HEALTH STATISTICS ANALYSIS

Iwao M. Moriyama, Ph. D., Chief

NATIONAL VITAL STATISTICS DIVISION

O. K. Sagen, Ph. D., Chief

NATIONAL HEALTH SURVEY DIVISION

Theodore D. Woolsey, Chief

E

Public Health Service Publication No. 1000-Series 1-No. 1

,

-.

I

.

CONTENTS

The Need for Health S~tisrics ------------------------------------------The Dvelopmenr of Healti SEtisucs ------------------------------------Problems of hfeasurement ---------------------------------------------Data Available Prior tothe National Health Survey ------------------------The Experience in Health Statistics Collection ----------------------------

Important Early Srudies ---------------------------------------------Activities of the Years 1949-55--------------------------------------

The Nariollal Health Survey Act-----------------------------------------

Policies and Organization of the Survey ----------------------------------Program of tie U.S. National Health Survey ------------------------------The Health Interview Survey --------------------------------------------The Health Examination Survey -----------------------------------------The Health Records Survey ---------------------------------------------Ikvelopmental and Evaluation Studies -----------------------------------Reporting the Results --------------------------------------------------Potentials and Limitations of the Survey Program -------------------------

Appendix IRecommendations for the Collection of Data ontheDistriburion and Effectsof Illness, Injuries, and Impairments in the Unired States ---------------

Pm-t ISummary ---------------------------------------------------------

Part 11The Work of Previous Committees and tie Task of The Present Sub-comminee ---------------------------------------------------------

Part HICurrent and Potential Uses of hlorbidity and Related Dara ---------------

Broad Areas of Need for Morbidity Statistics and Related Data --------

Parr I\’Types of Dara, Detail, and Frequency of Collection Required ------------

General h40rbid~ Data -------------------------------------------Frequency of Collection ---------------------------------------------Geographical Dtati -------------------------------------------------hd\'idual Smdies ---------------------------------------------------

i

CONTENTS-Continued

Page

Part V

Review of Present Sources of National Morbidity Statistics -------- ------The National HealthSurvey ----------------------------------------Advances in Sampling and Survey Techniques -----------------------Other Recent Studies ---------------------------------------------Llorbidity Statistics Applicable to Large Segments of the Population---How Other Countries Have Met This Problem -----------------------

Part VIRecommendations: Types of Data to be Collected in Order to Meet U.S.

Needs ------------------------------------------------------------Function of National Morbidity Surveys -----------------------------Function of Special Smdies ----------------------------------------Requests for Data That are not Feasible to Meet --------------------

Part VIIRecommendations on SurveyDesign -----------------------------------

Major Specifications of Accuracy and Derail ------------------------Major Specifications of the Survey Design --------------------------

Special S~dies --------------------------------------------------

Part VIIISuggested hfethods for Obtaining Dataon Need for Medical Care ----------

Appendix II

National Health Survey Act------------------------------------------An Acr----------------------------------------------------------National Health Surveys and Studies --------------------------------

252526262831

33333435

35353636

37

404041

U.S. NATIONAL HEALTH SURVEY

A cotxtnumg survey of the United States to ob-tain information about health conditions of the gen-eral population was authorized by legislationsigned by the President in July 1956. The new lawestablished, within the Public Health Service, theU.S. National Healti Survey which soon launcheda program to produce statistics on disease, injury,impairmem, disability, and related topics on auniform basis for the Nation. This marked a sig-nificant step in the efforts of the Federal Govern-ment to provide tools for improving the health ofthe American people.

THE NEED FOR HEALTH STATISTICS

Comprehensive heal~h statistics are neededin this country because a healthy people is per-haps the Nation’s greatest resource. Thehighpo-sition of health in the national scene is justifiedwhether the evaluation of it is smictly in monetaryterms or in terms of less well-defined but moreappropriate scales of national vitality, morale, in-dividual well-being, and other human values. Themeasuremem of this important fac~or—the peo-ple’s health —is found in the statistical measure-mem of well-being.

Primary users of general health statistics,particularly of morbidity data, include the direc-tors of operaung health agencies. An accurateappraisal of the extent and character of diseaseand the dismiburion and mends of morbidity areessential to the effective planning and evaluationof their programs and for extending the scope andimproving the balance of their work.

Public and private health agencies, whichfirst devoted their efforts primarily to the controlof infectious diseases, have seen their responsi-bilities broaden in several dimensions. As empha-sis shifted to the conmol of chronic diseases,many health deparunents began to be concerned

methods and- coordinatingv a. .G.y “1 DLL GG1lL1lg

~heir medical care

efforts. An adequa~e supply of beds for ~he care ofchronic disease patients in special and generalhospitals and nursing homes and the establishmentof home nursing care and rehabilitation programs,and in some areas public medical care programsfor tie medically indigent, all became of int-crestto these healrh deparunems. The nature of occu-pational health programs and maternal and childhygiene programs has changed in recent years,and the importance of health education and of thetechniques of the behavioral sciences has in-creased,

In me United States this evolu~ion in the na-ture of activities of official health agencies hasbeen accompanied by a growth in tie number andvariery of voluntary health agencies and by grow-ing efforts in the area of medical research.

New types of statistics were required toidentify the problems which arose and to developand appraise the new programs. Quantitative in-formation on tie prevalence of chronic diseasesand impairments, “the volume and degree of dis:ability, the use of medical services and hospitalbeds, and tie patterns of behavior of chronicallyill persons are but a few of the types of factsneeded by modern health agencies.

The needs of the social security and voca-tional rehabilitation agencies for health statis~icsare similar to those of the heal~h agencies. Forexample, total benefit payments for unemploy-ment and disability ,ae related to the volume andnature of illness, and tie question of how manycould benefit by vocational rehabilitation is par-ticularly important.

In a more tangential way, s~ati.wits of mor-bidity contribute IO the field of medical research.Clues to the etiology and pathogenesis of diseaseoften emerge horn studying the association be-tween the incidence or prevalence of disease andva’rious demographic, social, and geographicfactors.

An increasing general awareness of economicproblems has brought to the fore some of the

1

. .

more direct economic applications of morbiditystatistics. Some of these applications relate toimproved evaluation of manpower resources forcivilian and defense purposes. The economic lossof productive capacity due co illness is recognizedas an important factor in the changing economicpicture. The more stable component of this loss,attributable to chronic diseases and impairments,represents a potential source of extra productivecapacity if available methods for the rehabilita-tion of workers can be used effecriveIy.

In another application, morbidity statisticsrelated to demographic factors form the actuarialcornerstone in developing voluntary hospiml andmedical insurance plans. As these plans increasein number and scope, health data covering allsegments of the population become indispensable.Closely related to these uses of health statisticsis the interest of drug and appliance manufacturersin estimating potential markets for new productsand the use of such statistics as general guide-lines in market analysis and production schedul-ing.

THE DEVELOPMENTOF HEALTH STATISTICS

Just as the original activities of health agen-cies were focused largely on controlling infectiousdiseases, so the early health statistics werecon-fined largely to reportable diseases and mortality.The expmding prugrmns of the health agenciesindicated a serious need for a corresponding ex-tension of the field of health statistics. In view oftiis, it is remarkable that the health aspects of~he national economic and social life remained forsuch a long time without the systematic statisticalmeasurements that were generally available foragriculture, finance, m mu factoring, employment,foreign trade. ~nd population.

-PROBLEMS OF MEASUREMENT

The comparatively slow development of sta-tistics of illness for the general population can beascribed to inherently difficult problems of meas-urement. In the case of health and vital statistics,death is a clearly defined event, occurring oncefor each person and m a closely determinablepoint in time. ‘The existence of a morbid condition,on the other hand, is by no means always clearlydistinguishable from acceptable good health. Like-wise, the beginning and end of the condition ofmorbidity cannot always be accurately specified,even by the person affected.

2

As has often been pointed our, there is a con-tinuous scale of well-being extending from goodhealth to severe sickness. The point on the scalewhich appropriately represents the dividing line

between illness and heakh depends on the purposefor which tie statistics are to be used, as do theappropriate srandards of measuremem. For somepurpcses this point must be based on medicalevidence, and the whole scale must be marked offin terms of clinicaI and pathological criteria. Forother purposes it is more appropriate to considera person ilI when he considers himself to be illand to mark off the scale in terms of the actionshe takes as a result of his ilIness-the extent towhich he cuts down on his usual activities becauseof his condition, rakes to his bed, calls upon aphysician for help, and so on. The application oftwo such different means of measurement will,with many individuals, produce quite disparateresults in terms of a health-illness scale.

Difficulties in dealing with morbidity statis-tics aIso srem from the many ways in which theevents may be counted and classified. The basicmeasures are point p~-eualence, or simply prez’a-Le)zce, meaning the number of a specified type ofconditions existing at any given time, and inci-deuce, meaning the number of new cases of a cer-tain type starting in a defined period of time.There also are variants and combinations of thesemeasures. The classifications may be numerous:such as whether the condition can be consideredas acute or chronic; or it may be classified ac-cording to the diagnosis, the prognosis, the sever-ity, the kind and amount of care received or re-quired, or the consequences of the condition to theindividual’s family and to the community. Each ofthe possible units of tabulation and each of thevarious axes of classification may provide thestatistics required for some important applicationof the data.

The variety of rypes of data is paralleled bythe number of sources from which morbidity datacan be obtained. Three principal types of sourcesare: (1) those that provide data as a by-productof the operanon of some medical care or insur-ance plan; (2) those that derive data from othertypes of existing medical records, such as therecords of physicians and hospitals; and (3) thosethat are based on records created specifically forthe morbidity statistics, such as interview recordsand special health examinations. Because of thegreater ~ssibilities for relating the data to samp-les of the general ppulation, the last of theseclasses has notable advantages in the compilationof health data. However, with regard to any of

these $only be

Fcmationfrom winformnot titsource.datumoral, blthe pu:

Elishectics otiomble diSuitshealthparattand hcand a,whichchara,uniforsentiasinglewas t,

- tality.the diled ccsyste:registion 5of detber otial Emedi~1956distr.the SLtherestati.sof neeity. Fratesvolumeases.

I ,,con-

1 13~scaleg linerposeio thesome

edical:ed offa. Fornsiderbe ill

ctionstent co?causeupon ation oft will,parate

-tatis -.ch the

basicoreva-:ype ofi inci-a cer-time.

f theseerous:idereded ac-3ever -or re-n to tie:ach ofof tie

.de tieication

sled byity dataotircesmoductinsur -

n otheras the

3) thoseally for‘ecords: of the[0 sam->f thesepilation

any of

m

these sources of data, the full potentialities areonly beginning to be explored.

For example, there are certain types of infor-mation which can best be obtained tiom the sourceshorn which persons seek health care. To gatherinformation of this kind it is desirable to samplenot tie general population but the health caresources themselves. For the most part the basicdarum in such sampling might be an exisdng rec-ord, but often a record may have to be created forthe purpxes of the survey.

DATA AVAILABLE PRIOR TO THE

NATIONAL HEALTH SURVEY

Before the National Healfi Survey was estab-lished, information on the extent and characters-tics of illness was limited largely to that derivedfrom physicians’ reports of selected communica-ble diseases as required by State laws; to the re-sults of a large number of specialized and localhealth studies and surveys; and to scattered, dis-parate reports from hospitals, clinics, and healthand hospital insurance plans. Whatever the qualityand adequacy of these data for the purposes forwhich they were developed, their ad hoc and variedcharacter provided no basis for consmucting auniform and valid nationwide measure of the es-sential features of the health of Americans. Asingle exception in the field of health statisticswas the statistical information relating to mor -tal.iry. In tiis field legal demands for conmollingthe disposition of lmdies and for proof of deathled to the early establishment in every State of asystem requiring medical certification and officialregistration of every death. The death-registra-tion system has provided data for an annual seriesof detailed, narional mortality statistics for a num-ber of years. These statistics serve many essen-tial scientific and adminis~ative purposes in themedical, demographic, and actuarial fields. Until1956 most of the exisdng evidence concerning thedismibution and mends of morbidity was based onthe study of tiese mortality statistics. However,there was a constant awareness tiat mortalitystatistics alone could not serve the meat varieqrof needs for information akut illness and disabil-ity. For example, it was well known thar deathrates could be misleading indexes of the reladvevolume of disability resulting from different dis-eases.

THE EXPERIENCE IN HEALTH

STATISTICS COLLECTION

As the health statistics exisdng in 1956 fellfar short of meedng national needs, so the tech-niques of data collection in the health fieldsuffered from underdevelopment. Nevertheless,the experience of a number of earlier studies,surveys, and commirtee projects was of greatbenefit to the National Health Survey in the plan-ning of its first activities. Without this experi-ence the new program would have had to spend anumber of years in experimentation before anyextensive collection of statistics for publicationcould begin.

Important Early Studies

‘The last previous effort to obtain comprehen-sive illness statistics for the general populationof the Nation was tie nationwide health survey of1935-36. This was a memendous undertaking inwhich interviewers visited 737,000 urban house-holds to learn which members of the householdhad experienced disabling illness and which hadspecific chronic diseases or impatiments.

While this survey was by far the largest thathad ever been devoted to le~g the facts of ill-ness and injury in the general population, it wasnot the first of its kind in the United States. Anumber of smaller studies had demonstrated thatthe interviewer method could provide useful in-formation about @e amount and distribution ofdisease, the circumstances of injury, the loss oftime tiom work or other usual activities resultighorn disease and injury, and tie use of medicalcare in connection with morbidity. Besr known of

these are the Hagerstown, Md., studies of theearly 1920’s and the survey made during the years1928-31 by the Commirtee on the Costs of MedicalCare. The smaller community -rype studies con-tinued after 1936 and additional refinementsweremade in techniques. An important example of anintensive community study was the Eastern HealthDisrncr Study conducted in Baltimore from 193Sto 1943 by the Public Health Service, the MilbankMemorial Fund, tie Johns Hopkins UniversitySchcml of Hygiene and Public Health, and the Bal-timore City Health DeparQnent.

At the same time, great advances were madein the science of population sampling, with tieBureau of the Census leading the way in develop-

—.—— ....-. .—_ —---- -.. . ___,.J

ing practical methods of applying the theory ofprobability sampling in the field. In 1943, the Bu-reau of the Census, building on am earlier surveyby the Works Progress Administration, begancollecting information on the labor force by con-ducting monthly interviews in a national sampleof households. Now known as the Current Popula-tion Survey, it was used almost from the begin-ning to satisfy some of the demand for nationalmorbidity statistics by the addition from time torime of special questions or supplements to thebasic questionnaire. The Current Population Sur-vey, however, originally was designed to have andstill has as its foremost task the collecrionof in-formation on tie characteristics of the laborforce.1

Activities of the Years 1’949-55

In January 1949 the United States NationalCommittee on Vital and Health Statistics wasestablished. Recognizing the inadequacy of avail-able sources and the obsolescence of exisring da-ta, the committee imme&ately gave its attentionto the problem of obtaining adequate national mor-bidiry statistics. Two successive subcommitteeswere appointed by the chairman of the nationalcommittee “to frame the problems in morbiditystatistics, including chronic diseases and medicalcare staustics, in order tha~ morbidity data maybe directly related to demographic factors.” Thesesubcommittees recommended the study of a num-ber of methodological questions, but even as rherecommendations were being made steps were be-ing taken in several parts of the counmy to getsome of the answers in community surveys. Aboutthe same time legisladon calling for an 18-monthstudy of medmds of measuring illness passed theSenate but failed to pass the House of Representa-tives.

Within the next few yeas there was an up-sur_ge of interest in illness surveys as variousgroups set out to fill the gaps in available statis-tics or to answer particular questions with whichthey were faced.

lIn 1959 tic ~=sPnsibLLcy for malysis and publication of la-

bor force data from LIIe surrey wma m-ansferrcd to be Bureau of La-

tar StaUsUcs. Tim Bureau of tbe Census continues m collect and

rabulatc rhe d~ca under canmcc wicb tbe BuremJ of Lab Sratis-

UCS. ~ 1961 tbe BIXCmJ ❑f dm Census initked s new ssmple pan-

el czkd cbc Quarterly Household Smmple, which fmm time to time

will be used ta collect, among I rmriccy of subjects, heakb dam of

a specialized nature.

4

Tn New York City a committee undertook tocompare the health of members of the Health In-surance Plan of Greater New York with that of thepeople of the city in which these members lived,by parallel surveys in the NO populations.

The California Deparfmem of Public Healthfirst launched a study of methods to measure ill-ness in San Jose and then initiated a statewidesurvey to compile statistics to guide the activitiesof the health deparment.

The Graduare School of Public Heahh of theUniversity of Pittsburgh established a series ofcommunity studies in the Arsenal Health Dismictand in the city as a whole, partly to train scien-tists and partly to study patterns of illness in re-lation to demographic and social facmrs.

The Commission on Chronic Illness sponsor-ed two impntant surveys to obtain data on theprevalence of chronic illness and disability andto estimate what rhese data meant in Ierms ofmedical care. One was carried out in a rural area(Hunterdon County, N. J.) and the other h an urbanlocality (Baldrnore, Md.). These surveys wereunique in that they marked the first attempt tocombine results of household interviews with com-prehensive medical examinations offered to rep-resentative samples of the population.

In Kansas City, Mo., Community Studies, Inc.,began a survey to identify a sample of personswiti handicapping conditions who were examinedIO determine whether they could benefit from re-habilitation. Those that could benefit were offeredthe services which they required. These personsrhen were followed to evaluare their rehabilitation.

These and many other studies initiated duringthe late 1940’s and early 1950’s conmibuted to theknowledge of how morbidi~ data could k mademore accurate and useful. The Bureau of the Cen-sus acted as the collecting agent in several of thesurveys, and the Public Health Service served ina consultant capacity in nearly all of them. Bothagencies thereby gained experience which was tobe of great value later.

Meanwhile, anticipating progress in the solu-tion of the methodological problems raised by theearlier subcommittees, the chairman of the U.S.National Commirtee on Vital and Healti Statisticsestablished a new sulxommittee in February 1951.This was the Subcommittee on NationalMorbiditySurvey which was charged with dra~g “a planfor a nationaI morbidi~ survey keeping in viewthe interests of local areas.” The re~rt of thisgroup, ‘rProposal for Collection of Data on ILlnessand Impairments: United States ,“ submitted inOctober 1953, was the basis for specific legisla-tive authorization for a continuing national health

5urvey wtext of th’.~rt set rSurvey, i

Othei

prior to tvey Act 1cepts of tmeetingswhich thcussed t!that a :meetingsSurgeoning GrouLic Healwhich w

of :ence tiof locaInationwon frorIsurveymade h-

Inand Wegeon Cbe autlillnessdationincludeon heaposakHouseafter [healthHealLtgressJuly ;

%s.sof sicknf

accompa

GOvcmm

3LLSS.umy r

seas. V:

,$

-trek todth In-]tof the

lived,lations.

Healthure ill-atewide:tivities

h of themies oflismictt scien-sin re-

x3nsor -on the

lity and:rms of.-al aream urban~6 weresm t 10P,th com-to rep-

:. S,Inc.,persons;aminedmom re-. offeredpersonsIitation.d during,ed to the~ made:he Cen-slof theroved inm. Both,1 was to

-.

he solu-?d bytietie U.S.tatisticsry1951.orbidity, “a plan, in view“t of thisnIIJness,litted inlegisla-al health

9

survey program. (See Appendix I for the completetext of the report.) In view of the fact tiat the re-port set the initial pattern for the National HealthSurvey, it is a basic document of the program.

Other activities during the years immediatelyprior to the enacmnent of the National Health Sur-vey Act helped to clarify tie purposes and con-cepts of the Survey. hong these were a series ofmeetings held within the Public Health Service atwhich the staff of each of the major programs dis-cussed their needs for the type of statistical datathat a survey might supply, the results of themeetings being summarized in a report to theSurgeon General; and the discussions of tie Work-ing Group on General Illness Statistics of tie Pub-lic Healfi Conference on Records and Statistics,which were devoted to the definition of terms.

Of interest also was the health survey experi-ence in other coumries. In Great Britain a numberof local or special studies were followed by thenationwide Survey of Sickness which was carriedon from 1944 to 1952, Canada made a nationwidesurvey in 1950-51, and Japan and Denmark havemade health surveys on a national basis.

THE NATIONAL HEALTH

SURVEY ACT

In 1955 the DepsrIment of Health, Education,and Welfare proposed a plan under which the Sur-geon General of the Public Health Service wouldbe authorized to conduct a continuing survey ofillness and disability in the Nation. A recommen-dation that Congress enact such legislation wasincluded in the President’s legislative programon health matters. Bills incorporating the pro-posals were inmoduced in both the Senate andHouse of Representatives in February 1956, andafter hearing testimony on the needs for improvedhealth s~atistics~13 Congress passed the NationalHealth Survey Act (Public Law 652, 84th Con-gress). The bill was signed by the President onJuly 3, 1956, and later in the same month funds

were appropriated for the first fiscal year ofoperation. The tea of the National Health SurveyAct is given in Appendix II.

Several salient points of this legislation de-serve special mention since they determine someof tie unique aspecm of the program of the Na-tional Health Survey.

The fact tha~ previously etistig data on thehealth of tie Nation’s population was seriouslyout of date and tiat current information wasurgently needed for many purposes is explicitlyrecognized in the Act. But in the stated broadterms of reference for the authorized Survey pro-gram, it is clear that Congress had in mind notlimited or sporadic attempts to collect betterhealti information bur an earnest and sustainedprogram directed toward tie solution of the prob-lem of producing current health information.

Anotier point of interest is that the subjecc ofmethodology is explicitly mentioned in this legis-lation. The methods to be used in accomplishingthe purposes of the Act are specified only in gen-eral terms, bur the need for studying the technicalproblems of methodology is smessed. It might beassumed that in any authorized program the ad-mfistering agency would give attention to themethodology relating to the work being undertaken,but the National Health Survey law goes beyondthis and specifically provides for the study ofmethods and survey techniques in the health sta-tistics field with a view to tieir continued im-provement.

The Act also recognizes that otier agencieshave an interest in health survey information ormay have available technical resources or ma- -rerials useful to the program. Thus, the Act givesbroad authority for cooperation with such agen-cies m-d for the use, by written agreement, of fa-cilities tiar can be made available. In a similarsense, inasmuch as the program of rhe NationalHealth Survey cannot serve all needs for healthdata, it is recognized that other agencies and or-ganizations will carry on their own special-pur-pose studies, and the Surgeon General is author-ized to make available to them technical adviceand assistance in dte application of statisticalmethods to their surveys or studies in the healthand medical fields.

%1.s.Gn~ress, senate: ~ntinuing =mey snd specinl smdics

of sickness xnd disability in he United States. Repofl No. 171S co

accompany S. W76, B4dI timng. 2d. SCSS. Vksbingmn, D. C., U.S.

Government ptinting office, 1956.

%S. Cmgre.s, House OF Represencsciwm Nntiond Heal~

key Act Report No. 210S to sccompsny S. N76, 84ch (3mg. Id.

scss. Wmshingmn, D. C., U.S. Government Prim&g Office, 1956.

POLICIES AND ORGANIZATION

OF THE SURVEY

The scope of the authbrizalion in the NationalHealth Survey Act is such that to be productive

5

—. . . .— _ —_—. .,-.

and economical the operating plan of the Surveyprogram had to be carefulIy delimited.

Any artempt to cover mo broad a range ofhealth statistics could only result in a programwhich would be not only cumbersome butwouldbescattered in character. For this reason outlinesof the prObTaM had to be clearly drawn.

Yet the need for a great variety of differentrypes of data also had to be recognized. Thismeant that within definite outlines the goals andmethods had to be diversified, concenmating onthe most productive objectives and procedures,and the program had to be flexible enough to dealconsecutively with many areas of interest and torespond readily to changing requirements.

The unique responsibility of the Survey pro-gram is to collect ~ypes of data which are basedeither on a sample of the general population of theUnited States or are relatable to separatelygathered general population statistics. This is inconmast to studies based on the records of par-ticular health agencies or health adminismativeprograms which cannot be relaredto a population-at-risk or which have as a population-at-risk agroup of members of a plan or beneficiaries of aprogram, lmth of which are in whole or in partself-selected or otherwise not representative.

The importance of data relatable to a knowngeneral population stems from the fact that sta-tistical measures of health or morbidity take ontheir scientific meaning, contrasted with certainadminismative applications, only when these canbe expressed as propor~ions or rates.

It is not sufficient to know, for example, thatat a given time so many persons of such-and-suchpopulation group are ill with such-and-such dis-ease. The importance of this aggregate figuretakes on a fuller meaning when it is expressed inrelation to the total number of persons of tiesame ~pulation group.

To serve the purposes of general health sta-tistics, the collection of data on sick persons byage, sex, occupation, area of residence, etc.,must be accompanied by a parallel collection ofcorresponding data for the population as a whole.Focusing the Survey program onpopulation-basedtypes of studies meets this technical requirement.

Using a population sample as the origin of thedata takes advantage of the fact that there is onlyone point from which au the needed informationis generated—this point is the individual.

Starting with a known sample of individualsdata can be gathered horn a variety of sources,These include the person himself or members ofhis family, along with hospital records of physi-cians and dentists, as well as the results of spe-

6

cially conducted measurements, tests, and exam-ination procedures.

The particular smength of such surveys de-rives from the ease with which healti data con-cerning the individual can be related to social,economic, and ocher demographic characteristicswhich are determined for the sick and well alikeas part of tie same survey plan. Hence, many oftie data-collecung activities of the Survey startwith samples of the population.

However, there are other sampling methodswhich can produce valid statistical informationrelatable to a general population-at-risk andwhich may be more efficient for some purposes.Much of the data in which the Survey has an in-terest originates in the places where people obtainhospital, other medical, dental, and nursing care.Consequently, when the objective is to producestatistics on care received, it is often preferableto sample the sources of the care directly. In thistype of survey the population-at-risk is no~ de-termined as part of the same collecting process.It must be determined separately, usually fromgeneral-purpose-population eratistics. Naturally,care must be taken to insure that both bcdies ofdata relate to the same universe.

Both of these general types of survey pro-cedures are used. However, in all cases the tar-get universe is the general population or someimportant segment of that population.

The Survey program has another basic char-acteristic-one which distinguishes it from anum-ber of other studies in the health field. Except forthe general mortality and natality data compiledby the National Vital Statistics Division, moststatistical projects of the Public Health Serviceare designed to serve a particular program in-terest. Such programs collect detailed dara on aparticular disease entity, test specific hypothesesrelated to the etiology or pathogenesis of certainconditions, or measure or assist in the admin-istrative conmol of some heakh-operating pro-gram. Other types of statistical projects, especial-ly those carried on by State, county, or cityhealth agencies, are designed to determine healthconditions or study some problem related to aselected population group or to a limited localarea.

In contrast, the National HeaIth Survey is notdesigned to serve any single health-program in-terest nor to meet the needs for detailed Iocaldata. Its task is to provide general backgrounddata which present the overall health situation andwhich show various components of the healthprobIe,m in proportionate relation to each otherand in relation to important population variables.

Thegeneralplernentsratisticor a reftit canncwhich ur

Therer intelicy of plinformssearch Irow vie’

Thethe direeven ththe parservedhealti zthe mecagency.sions,gist, thinsurarand oh.numerothe hea:

Arthe Surneeds oversionmultitu(cal, ti~

Tht. sible odrroughof the :Surveyning c’-suggeseral a{their ameetinquestinposed c

mservicenot onlconsidespecialon a rzthat aretific St&

Thtdetail blease 01

dexam.

reys de-lta con-

social,erisucs:Ll alikemany ofey start

methodsrmationisk andreposes.s an in-ie obtain,ngcare.produceeferable.’. In this

not de-3rocess.lly frommmally,-mdies of

vey pro-the tar-

ar some

jic char-n anum-<cept formrnpiledm, most1 Service-~am in-iata on apothesesf certain5 adrnin-ing pro-?4pecial-or city

ne health,ted to ated local

:ey is notgram in-led localckground]ation andIe health~ch Oth-

ariables.

The definition of the Survey program in thisgeneral sense means that its activities will sup-plement but till not duplicate special or localstatistical projects. Ir can serve as a backgroundor a reference point for various local studies, butit cannot replace nor serve the specific needswhich underlie such studies.

The absence from the Survey of subjectmat-~er interesta of its own is consonant with the pol-icy of providing service, in the form of statisticalinformation, to those responsible for health re-search or operating programs. Here, too, a nar-row view is not admissible.

The program must not be unduly weighted titie direction of selected subject marter imerests,even though these may be the most demanding atthe particular moment. The entire group to beserved comprises the legislator and the publichealth administrator-Federal, State, and local—the medical research scientist, the privatehealthagent y, the range of health and related profes-sions, the demographer, economist, and sociolo-gist, the teaching institution, the life and healthinsurance agency, the manufacturer of drugsand other health supplies, the market analyst, andnumerous other activities and imeresta related tothe health of tie population.

A major task in designing and administeringthe Survey is tie consideration of the statisticalneeds of this heterogeneous audience and the con-version of the major common elemems of thesemultitudinous needs into a clearly defined, pracu-cal, timely, and economical survey program.

The proper design of such a program is pos-sible only when there etist effective channelsthrough which these needs can be expressed. Oneof the first steps taken in tie organization of theSurvey was to devise a system whereby the plan-ning could proceed with the benefit of advice andsuggestions from many quarters. To this end sev-eral advisory committees were appointed andtheir advice is sought, not only through periodicmeetings for general discussions but through re-questing their review and opinion of specific, pro-posed courses of action.

The objective of the Survey—to provide aservice to a range of heahh interests-requiresnot only that the data collection be designed inconsideration of the various needs, but a 1so thatspecial emphasis be given LOthe speedy release,on a rapid time schedule, of compiled statisticsthat are consistent with high technical and scien-tific standards.

The publication program, described in moredetail below, is designed to facilitate early re-lease of the compiled data togetier with such an-

cillary information as is needed to make the datareadily usable by consumers.

This publication policy requires that statisti-cal tables be accompanied by technical notes,qualifications, definitions, and such illustrativeanalyses as are necessary for a clear apprecia-tion of the meaning, uses, and limitations of thestatistical information.

The analyses included in the publications ofthe Survey must, however, be limited in scope,consisting primarily of exposition and illusma-tive uses of the data. This limitation is desirablesince the function of the Survey is to provide ob-jective and accurate facts but not to interpretthese facts so as to indicate any par~icular courseof action or to support any particular health poli-cy or program. Policy implications of the statis-tical data are the responsibility of the legislatorand the administrator.

The activities of the Survey in collectin~ datathrough various channels throughout the Nationobviously require an understanding and acceptanceof ~hem by the public as well as by all health pro-fessions. This is particularly important in a data-collecting program based on the voluntary coop-eration of the respondents, and the Survey recog-nizes the need to develop rhis general understand-ing of the purpose of rhe program and the methodsto be used.

In developing the organizational arrange-ments for the Survey, it was considered advanta-geous to make mtimurn use of the provisions of theNational Heakh Survey Act which authorized theSurgeon General to conmact for the services orfacilities of any agency, organization, group, or -individual. The use under conmact of existing~echnical facilities and materials makes availableto the Survey valuable, special, and professionalralents and resources which could not otherwisebe duplicated, and permits the limitig of the Sur-vey staff to a relatively small group primarilyconcerned with planning, research, and analysis.

The fact that the Survey operates by con-tacting for many phases of its work with otieragencies does not &Iply, however, that it is inany sense a gram-in-aid program. Contracts aremade with other agencies for conducting specifiedstudies or operations which are aspects of orconmibute to the Survey program as an integralpart of its own work.

Authority for grants-in-aid for prosecutionof surveys by other groups is not included in theNational Health Survey Act, nor are funds avail-able for this purpose. There is in The Act, how-ever, authoriry to make available to appropriatepersons and agencies technical advice and assis~-

7

ante in the application of statistical methods tohealth surveys.

Any active statistical program should also bean evolving program. It is significant that a pro-portion of tie total resources is allocated to eval-uating Survey results, to assessing the reliabilityof data, and to exploring possible improvementsin method.

Modifications of the Survey program may beexpected as experience pints the way to improve-ments in method or to different emphasis on ob-jectives. This means that occasionally compara-bility with earlier data collected by the Surveywill be sacrificed if by doing so it appears highlylikely that the quality or usefulness of the datacan be raised. For example, certain features ofboth the questionnaire and the sampling structureof the household-interview ,,survey were being re-designed almost from the beginning, and tabulatingplans are subject to change each year. The meth-odological work and the continuing evaluation ofresults undoubtedly will lead to further revisions.

PROGRAM OF THE U.S.

NATIONAL HEALTH SURVEY

The varied nature of the dara desired forplanning and evaluating all phases of public andprivate heaIth work and the requirement that cor-responding information be obtained for both theill and the well population determirze the mainlines of content of the program of the NationalHealth Survey.

This is not a single survey with only onemethod and a fixed set of objectives. Rather, it isa program of surveys, using different approachesand changing its end objectives as both the tech-niques and needs for data evolve. But basic to allthe present and funzre surveys is the fundamentalidea that the data colkction must refer IO a rep-resenra~ive population.

. . The activities of the National Health Surveyma-y be divided into three parts: (1) the Health In-te~view SZmJey —a continuing nationwides amplingand interviewing of households; (2) the HealthExamination Survey —physical examination andtesting of samples of individuals proceeding in aseries of separate surveys or cycles; and (3) theHealth Records .%m’ey —another series of samplesurveys in which the sources of information areestablishments which provide hospital, other med-ical, dental, nursing, and other types of health-re-lated care to the general population.

These three survey activities are supportedby a program known as Developmental and Evalu-

8

ation Studies. Referred to previously, these arestudies through which rhe quality of data is evalu-ated and improved methods are devised and tested.

A brief description of each of these parts ofthe program follows. Later publications in thisseries will consider the major surveys separatelyand will describe rhe statistical design, concepts,and definitions associated with each survey.

The order in which the surveys have beenlisted above is the order in which these parts ofthe programs were initiated, except that the De-velopmental and Evaluation Studies have proceed-ed concurrently and were undertaken as needed.

The Health Interview Survey, planned duringthe fall and winter of 1956, was pretested in Feb-ruary 1957 in Charlome, N. C., and nearby areas,was tested again in a national “dress rehearsal”in May-June 1957, and was officially launched onJuly 1 of that year. Data collection has been con-tinuous since that time.

The first cycle of the Health ExaminationSurvey was plamed over a period of more than ayear. Ic was tested in three successive pilot stud-ies which took pIace in Washington, D. C., in June1958, and in Fort Wayne, Indiana, and HowardCounty, Iowa, in the spring of 1959. Examinationsof the first cycle sample were begun in Philadel-phia in October 1959. Work was carried on slowlyat first because of a lack of equipment and staff.Full-scale data collec~ion on this survey datesfrom April 1961.

The variety of survey activities making upthe Health Records Survey began to take shape in1961. Consbnzction of a sample frame, or list, ofmedical, nursing, and personal care institutionsin the United States was he first field task in thissurvey. After existing lists were consolidated ona preliminary basis a mail inquiry was sent outin April 1962. This inaugurated the third majorsubdivision of the National Health Survey.

In all of these works of the National HealthSurvey, the Bureau of the Census played and con-tinues to play an important role. Participation byche Bureau takes four forms: (1) the givingof ad-vice on survey design and methods; (2) the actualselecting and identifying of units for the sample;(3) the conducting of importam par~ of the fieldwork, including the selection, training, and super-vision of interviewers in the Health InterviewSurvey; and (4) the performing of many large-scale tasks of processing.

The paztern of participation by the Bureau ofthe Census varies in the different surveys. Whileadministratively the arrangement resembles acontract on the part of the Bureau of the Censusto do specified work for ~he Public Health Service,

I

I

the relala joint sinterestmibutesvey meti

THE

Theian, noStates.data onjuries;ments;kinds 01

ceived;The

selvesdemogriIlnessSuit ofgroups.

Haually M

a probzfor succsampleklongercontinzihigh fr~solidatiacteris iclassesyear cz

“are freTlm

muItist,consist1,900units (v.lded.Counuearea. “ple hasthere a

Se.seriesthe cho,steps ZTess of Gor addreludedfirst 5:rain anceptiozxsiding a

,.

e are3valu-ested.rts ofn thisrately:epts,mvey.

been‘rts of]e De-Jceed-:eded.~uring1 Feb-meas,arsal”led onn con-

nationthan at stud-n JuneIowarclLathns

Lladel-; 10W1y

i staff.dates

:ing upiape inlist, oftutionsin this~ted onem outmajor

HealthId con-uon by~of ad-actual

ample;le fieldsuper-.erviewlarge-

reau ofWhile

ibles aCensuservice,

,

m

tie relationship actually is much more likethar ofa joint staff. The Bureau of the Census has equalinterest in the quality of data collected and con-tributes greatly horn its fund of knowledge of surv-ey methods.

THE HEALTH INTERVIEW SURVEY

The Health Interview Survey covers the civil-ian, noninsutuuonal population of the UnitedStates. The purpose of tie Survey is 10 providedata on the incidence of illness and accidental in-juries; the prevalence of diseases and impati-mems; the extent of disability; the volume andkinds of medical, dental, and hospital care re-ceived; and other health-related topics.

The data are obtained from the people them-selves and therefore measure the social anddemographic dimensions of health-the impact ofillness and disability and actions taken as a re-sulr of these conditions, in various populationgroups.

Household interviews are conducted conun -ually with interviewing taking place each week ina probability sample of the population. Samplesfor successive weeks can be combined into largersamples to obtain data for a quarter, a year, or alonger period of time. The design thus permimcontinuing measurement of characteristics ofhigh frequency to show trends and to show, by con-solidation into larger samples, analysis of char-acteristics of low tiequency or of smaller sub-classes of the population. Data collected over ayear can be presented as annual estimates whichare free of seasonal biases.

The interview sample is a highly stratified,multistage probability design. The firs~ stageconsists of an area sample drawn born about1,900 geographically defined primary samplingunits (PSU’S) into which the Nation has been di-wded. A PSU is a counry, a group of contiguouscounties, or a standard metropolitan statisticalarea. The number of PSU’S selected inthe sam-ple has varied with che sample design. Currently,there are 357 such areas.

Second and subsequent stages consist of aseries of steps for further subsampling withinthe chosen PSU’S. For discussion purposes, thesesteps may be combtied and described as a proc-ess of drawing clusters of neighboring householdsor addresses. Such clusters, called segments, in-cluded an expected six households during thefist 5 years of the Survey; currently, these con-tain an expected nine addresses. With minor ex-ceptions the sample encompasses all persons re-siding at the selected places.

Statistics from the interviews are producedthrough two stages of ratio estimation. In the firststage tie ratio factor is the count of population inthe 1960 decennial census divided by the estiated1960 population for tie first-stage sample PSU’S.This adjustment is applied separately to resi-dence and color cells, currently 25 in number. Inthe second stage, ratios of the official Bureau ofthe Census current population figures to sample-produced estimates of the current population areapplied to some 60 age-sex-color classes.

Through contractual arrangements, the Bu-reau of the Census provides assistance and coop-eration on many phases of the Survey design andadrninis~ation. This enables tie Survey to takeadvantage of the long experience, comprehensivestatistical organization, and field resources of theBureau.

In keeping wirh specifications and require-ments established by the Public Health Service,Lhe Bureau of tie Census designs and selects thesample, cooperates in the development of mequestionnaire and related manuals, conducts thefieId interviewing, codes the data, and carries outquality -conmol procedures for interviewing andcoding. The Public Health Service edits, tabulates,analyzes, and publishes the results of the Survey.

Interviews are conducted with each individualadult who is at home at the time of the interview.For persons not at home the interview is con-ductedwith a responsible family member — spouse,parent, or adult son or daughter residing in thehousehold. The noninterview rate is almut 5 per-cem, including 1 percent from refusals and theremainder horn households in which no one was .at home during repeated calls.

Variation and bias among interviewers arepotential hazmds to data on health conditions de-rived from household interviews. For this rea-son, interviewers are carefulIy selected, aregiven a comprehensive initial maining course, ,and are given group retaining semiannually andhome maining exercises about eight times a year.In addition, programs of regularly scheduIed ob-servation and independent reinterviews by region-al field supervisors are conducted.

During processing in the cenmal office, allquestionnaires are examined for interview errorsof question coverage and consistency. Results ofall of tie procedures for quality con~ol aremanumitted through the supervisors to the inter-viewers as guides to self-improvement or to theneed for individual retraining.

All maining, observation, and review proce-dures are oriented to a detailed Interviewer’s Man-ual. This includes instructions on the question-

9

-_ —.... . . _______ . ..—.-_._1—----

1’

.,

naire, on interviewing techniques, and on adequacyof population coverage. The manual is carriedonall assignments as an on-the-spot gdide in thehandling of unusual questions and situations.

The questionnaire contains items to identifyhouseholds and persons and to classify personsaccording to meir demographic characterisdcs.It also includes questions on illnesses, impair-ments, or other conditions; on the occurrence ofaccidental injuries; and on the use of hospitalcare. For each person to whom these questionsapply, additional questions are asked about thenature of reported conditions, number of days ofdisability, degrees of chronic limitation of acriv-i[y, ~es of accidents, and the duration of andreasons for hospitalization.

Items of the type described above are contin-uously included in the quesdonnaire as a basic coreof inquiries. Other items, such as the use of med-ical and dental care or the extent of health insur -ante coverage, are repeated at intervals of sev-eral years. Still others, of specialized or timelyinterest, are included for a single ye= with nodefinite plans to repeat these at a later date.

Annual modifications of the questionnairepermit collection of data on a larger number oftopics than could be included in a fixed single-visit interview. Other methods are also used toextend the scope of the survey. At times, a self-enumeracion questionnaire is left with the re-spondent when the basic interview is completed.This is mailed in by the respondent, with norrre-sponse follow-ups conducted by letter, telephone,or personal visit. Self-enumeration question-naires are usually employed on topics for whichthe respondent may wish to consult records, ontopics for which more accurate and complete in-formation can be obtained horn persons who maynot be at home at the time of the interview, orsimply to shorten the interview.

Users of data often require more detail abou~pe:sons with a particular health characteristicthan can be included within the interviewMg timeor the space limits of the basic questionnaire. kthese cases, persons with such a characteristiccan be identified &om the basic questionnaire.These persons, or a sample of them, are thens em a second questionnaire to obtain additionaldata which are imeg-rated with other health anddemographic information obtained in the firstvisit.

All new questions or topics are subjected toone or more pretests to assess alternative meth-ods of collection, under standtig by respondents,and the length of time for which respondents canrecall events without excessive memory loss.

10

Wherever ps.sible, such pretests utilize alter-nate measures and data obtained horn records tocompare with interview responses. Concepts,definitions, and coding specifications for newtopics are forrmdated in consultation with per-sons who are specialists in the subject. Planningand conducting of pretests are carried out imp-eratively with the Bureau of the Census.

Household ques~onnaires are mamsmittedweekly from field supervisors to the cenmal officewhere these are checked for completeness of fieldcoverage, omissions, and other errors. “Codingthen is done according to specifications set forthin detailed coding and Uanscription manuals.Code categories for health and medicaI care itemsare established by the Public Health Service tomaintain consistency with the concepts and defi-nitions and to fulfill requirements for health in-formation. Illnesses, diseases, and injuries arecoded, with certain modificauons, according tothe International Classification of Diseases. Codecategories for demographic and economic itemsare, in most cases, established by the Bureau ofthe Census to conform with standard classifica-uons used in other operations of the Bureau. Thecoding process is routinely subjec[ed to quality-conmol procedures requiring defined standards ofproficiency.

Coded data are manscribed on magnetic tapefor compmer processing by the survey staff.These steps include assigning weights and ratioadjusunents; combhing weekly and quarterly datainto longer periods; carrying out a program ofinternal edits and consistency checks among vari-ous items of data; regrouping of ages, diagnosticcodes, and other information to facilitate tie prep-aration of later tabulations; conducting operationsnecessary to derive national annual estimates;and finally, producing the statistical tables.

Tabulations include not only data on healthand medical caxe items but population figures forthe various groups for which the survey is de-signed to produce estimates. In addition, samplingvariances are computed for a Iarge number andrange of statistics. These tabulations form thebasis for analyses and publication of a series ofreports on separate health topics and on healthand medical care ti differ em population groups.

THE HEALTH EXAMINATION

SURVEY

The basic purpose of the Heikh ExaminationSurvey is to conduct those studies of tiehealth ofthe population of the Nation which require, or can

best beclinicalments.data onulation tin the s:

Suctics onnery of’of the :only-v:nosed,eases :lation cmeasurcholest.

mExdrsegmerseparaplannernationoperatifindingon omcollect-anothe:and (cprocedtions ~mentparticLphase3-year

lxductirgabilitybasisfor c:consicwith tl-

Tlciallyconsiswhich,providAt thewhethe.tally sined. 1dicatesCooperrools- c,Continti

TISurvey

lter-rds m:epts,- new

per-,nningcoop-

nittedofficeif fieldZoding

forthnuals.items

~ice to1 defi-lth m-es areling to. Codeitems

:eau ofsifica-U. ThelaZity -ards of

ic tapestaff.

d ratio-ly dataY= of,gvari-~ostic:prep-rationsmates;tables.health

mes foris de-

rnplingber andrm the:ries ofI healthgroups.

4

nination,ealti of., or can

m

best be done by, direct physical examinations,clinical and laboratory tests, and other measure-ments. The survey involves collection of healthdata on small, representative samples of the pop-ulation through direct exarrdnations of the personsin ~he sample.

Such a method is necessary to provide statis-tics on the medically defined prevalence of a va-riety of specific diseases and on the dentalhealthof the population. This is the best—perhaps theonly-way of obtaining data on previously undiag-nosed, unattended, and nonmanifest chronic dis-eases and of obtaining distributions of the popu-lation concerning such physical and physiologicalmeasurements as blood pressure, serumcholesterol, and auditory and visual acuity.

The broad plan of operation of the HealthExamination Survey calls for the study of specificsegments of the total population in successive,separate cycles of examinations. The program isplanned so that at any one time the Health Exami-nation Survey will be engaged in a three-leveloperation: (a) the analysis and publication of thefindings from a completed cycle of examinationson one segment of the population; (b) the fieldcollection of data for another cycle, dealing withanother defined problem and population segment;and (c) the planning and development of detailedprocedures for the succeeding cycle of examina-tions dealing with still snother population seg-ment and defined problem. The goal is to planpmticul~- cycles so fiat the data-collectionphase can be completed in approximately a 2-to-3-year period.

During the first cycle, the feasibility of con-ducting direct, single-visit examinations onprob-ability samples of the population on a nationwidebasis was confirmed. Procedures and methodsfor carrying out this task were developed, andconsiderable experience was gained in dealingwith the problems encountered.

The Health Examination Survey has two spe-cially designed mobile examination centers. Eachconsists of several mailers, drawn by trucks,which, when set up and connected by passageways,provide the setting for examination and testing.At the outset of the program, it was uncertainwhether a sufficiently large portion of a scientifi-cally selected probability sample could be exami-ned. The experience gained in the first cycle in-dicates that it is possible to obtain the necessarycooperation. The experienced field staff and thetools developed make ir possible ro plan on thecontinued application of this survey techtiique.

The first cycle of the Health ExaminationSurvey was designed to collect information on the

,

medically defined prevalence of certain chronicdiseases which are fairly common in the adultpopulation. PrimariIy, these were cardiovasculardiseases, arthritis and rheumatism, and diabetes.

‘The population studied in the first cycle waslimited ro the civilian, noninstitutional populationof the 48 contiguous States between the ages of 18and 79 years, inclusive. In addition to collectinginformation on the selected diseases, the first-cycle survey obtained a variety of measurementdata, including data on height, weight, skinfolds,visual acuity, blood pressures, elecmocardio -graphic tiacings, and the results of other Testsand procedures. The program also included a de-tailed dental examination. The plan and initialpro~am were published in 1962.4 Collection of thedata on some 6,700 persons examined in this pro-gram was completed late in 1962. Analysis of thedata is now in process.

The second cycle of the Health Examin-ationSurvey will focus attenuon on children betweenthe ages of 6 and 11 years, inclusive. Since theprevalence of chronic disease in this populationis quite low, attention will be directed to a varietyof measurements and to characteristics associ-ated with growth and development. A physicalexamination will be given, and a selected batteryof psychomemic tests will be adrntiistered. Care-ful examinations will be made for visual andauditory acuity. Considerable anthropometiic datawill be collected. A dental examination will alsobe made. Finally, tests of respiratory functionand exercise tolerance will be included, as willX-rays and certain other tests and procedures.

A key characteristic of tie Health Examina- -tion Survey is that each of its programs is basedon a probability sample of the population of theNation. The method by which this sample is se-lected closely parallels that described above forthe Health Lnterview Survey. The publication ofdata from the Health Examination Survey will in-clude detailed descriptions of the sampling andestimating processes, including the nature of allsignificant adjustments. The reports will also in-clude information on the sampling error and themeasurement variation.

4u-~. N~&nd Hcnkb S-cy.P[an ad Initia[ Program of tb e

Hea[tb Exmninatz”on 5mveY, Hedtb SmmisUcs. Series A-4. Public

Hemltb Service Publication No. SS4-A4. Public Heal& Serwice.

msbgmn, D. c., M~Y196z.

II

.—. ..—.-. ___,.l

Much effort is devoted to problems of stand-ardization of observations, validadonofthe meas-urement processes, and other aspects ofqualityconmol in connection with the HealthExaminationSurvey; likewise, much amennonis giventoprob-lems of sampling and problems of response. Inall of these areas, as in all of the work of theHealth Examination Survey, interest arises fromtwo different viewpoints. First, attention is givento these matters so as to improve the quality ofthe output of the survey. Second, attention is givento conmibucing generally to the knowledge ofmethods and survey techniques for securing sta-tistical information on healtn. In connection withthis latrer goal, the Health Examination Surveyundertakes various methodological studies whichshould benefit not only the. National Health Surveyprograms but all research workers in this fieId.This methodology work is the subjec~ of more de-tailed comment elsewhere in this report.

Future programs of rhe Healti ExaminationSurvey will consider other population segmentsand other problems. The first of these srudies(cycle 3) is expected to concern children andyouth from the ages of 12 to 17, inclusive. Thiscycle will complement the findings of cycle 2 toprovide broad and comprehensive data on thehealth of the children and youth of the Nation.

THE HEALTH RECORDS SURVEY

The Health Records Survey is a family ofundertakings with two characteristics: (1) tiefirst stage of sampling is the facility or estab-lishment which provides hospital, medical, nurs-ing, personal, or residential services; and (2) amajor part of the information collected has itssource in records of the sample facilities, in con-trast to such origins as the interviewing or thedirect examination of persons.

The main objective of the Health Records. . Survey is the production of sta~istics on the char-

acteristics of health services received by theAmerican people and the characteristics of thosereceiving the services. As an important by-product, the program will also produce statisticson the institutions and the other health facilitiesproviding the services—their numbers and types,their staffs, their charges, and their generalpolicies.

Thus, the Health Records Survey is broad inscope. The concept of the Survey encompassesrecords not only horn hospitals and nursing homesbut also tiom such facilities as clinics, physi-cians’ and dentists’ offices, indus~ial and union

12

dispensaries, and a v&-iety of residential andcorrectional institutions.

Among the several surveys in the program,early emphasis has been ~laced on five projects:(1) a Master Facility List of establishments; (2) anarea sample survey, known as the ComplementSurvey, to identify establishments which are norincluded on the Master Facility LisE (3) a Hos-pital Discharge Survey; (4) an T.nsticution Popu-lation Survey; and (5) a series of ad hoc surveys.Since all but the first of these are still in the de-velopmental stage, their descriptions must beconsidered preliminary. A brief description ofeach of these activiues follows:

1. The Master Facility List. —-A keysmne ofthe Health Records Survey will be the creationand maintenance of a master list of all or a majorproportion of all establishments or institutionswithin the scope of the program.

To establish this inventory, a number of listspreviously assembled by the Public Health Serv-ice, Bureau of tie Census, and lists published bynongovernment agencies were collated and merged.Then, to bring this file up to date, a questionnairewas mailed to each establishment listed. The in-formation requested varied with the type of placequeried, but in general, it related to size, owner-ship, type of service, and number and type ofstaff. This file is called the Master Facility List(MFL).

Thus constituted, the Master FaciIity Listserves two major functions. First, in itself itprovides significant data on the number, types,size, and geographical distribution of the subjectestablishments, and on the” numbers of personswhich can be served. Second, it provides a frameor list which can be stratified and sampled for awider variety of allied purposes.

It is desirable that the Master Facility Listbe both comprehensive and current. Currency isprovided by additions and deletions from the listusing several sources of information concerningnew and discontinued establishments, including ad-ministrative documents horn the Social SecurityAdminismation and Public Health Service pro-grams.

2. The CornpLement Surwey.-Despite the ad-vamage of a comprehensive and current list, itwould be uneconomical to attempt to maintain acontinuing inventory in the Master Facility Listtiat was really complete. It is more efficient rokeep the Master Facility List ar a high level ofcoverage and to augmem it with art auxiliar ys am-ple procedure. For this purpose, the ComplementSurvey is made a part of the total program.

I

r..

Thmake nber ofconducdesignthe su:addresbIocksthe inestablivey thzaskedescabliCompl.for ccterest

3.will b,:(0 obthospiton chthe Hcrea. sto prclarly L

pitaIsfinanc:use 01WhiIepital.izthe scand onfacts .detailt

- the gelAc

involv~stageand tiIn forxia starthe pa

Tundertit willperiodsons hsamplepopular

4.Iation(descriof resithat neof gathfor thi

I

.’

tial and

-ogram,rejects::s; (2) anplementare not

) a Hos-n Popu-wrveys.the de-

must bepriori of

:stone ofcreationa major

umtions

:r of.listslth Serv-lished byi merged.tionnaire. The in-, of pIace~, owner-J rype ofility List

dity Listitself it

:r, types,(e subject

persons~ a frame)Ied for a

:ility Listmrency is.n tie listmc erning: uding ad -\ Securityvice pro-

ite the ad-.nr list, it11aimain acility List:fficient ro$h level ofiiary sam->mplernent

program.

The design for the Complement Survey vdlmake maximum use of existing resources. A num-ber of periodic interview surveys are routinelyconducted across the Nation. Using the probabilitydesign of an ongoing survey and interviewers forthe survey, a list will be made of the names andaddresses of all establishments in specifiedblocks or segments of PSU’S during the course ofthe interviewer’s regular assignments. Eachestablishment identified in the Complement SW-vey that is not on the Master Facility List will beasked zo complete a questionnaire so tiat theestablishment can be properly classified. TheComplement Survey will dms become the basisfor completing the universe of facilities of in-terest in the Health Records Survey.

3. The Hospital Discharge Survey. —Thiswill be a continuing survey of short-sta yhospitalsto obtain comprehensive statistics on tie use ofhospitals, on services provided in hospitals, andon characteristics of their patients. Initiation ofthe Hospital Discharge Survey will greatly in-crease the potential of the National Health Surveyto provide statistics on hospitalization, parucu-larly on the kinds of cases being cared for in hos-pitals of various sizes and types, on methods offinancing the care, and on differing patterns in theuse of care in various sections of the Nation.While the Health Interview Survey data on hos-pitalization provide satisfactory information onthe socioeconomic correlates of hospitalizationand on the relationship of hospitalization to otherfacts shut the illness, they cannot provide thede~ailed characceris~ics of the patient’s care orthe geographic detail.

According to present plans, this survey willinvolve a two-stage probability sample, the firststage being a sampling of short-term hospitalsand the second stage, a sampling of discharges.Information horn the discharges will be copied ona standard absu-act form which will not includethe patient’s name.

The project will differ &om most of the otherundertakings of the National Heahh Survey in thatit will meat the universe of all discharges over aperiod of tie, rather than the universe of per-sons living at a point in time. Estimates tiom tiesample will be related 10 independently compiledpopulation statistics tiom census sources.

~. institution Po@zdatiOn Suwey.—Thepopu-lation sampled in the Health Interview Survey(described above) does not include the populationof resident institutions. This is due to the facttiat neither the rype of sampling nor the mannerof gathering the information is entirely suitablefor this segment of the population of the Nation.

While the institutionalized population constitutesonly about 1 percem of the total ciWian popula-tion, it is an important 1 percent from the stand-point of health. Many of the most severely dis-abled persons, for example, are found in residentinstitutions of one type or another, and certain ofthe published estimates from the Health InterviewSurvey and the Hea.kh Examination Survey areless useful because of the exclusion of this group.

For tiese reasons it, has been consideredimportant to collect data of certain types on tieinstitutionalized ~pulation that would add to in-formation obmined in the Health Interview Survey.Hence. a special procedure for surveying this~pulation (making use of the Master FaciliwList as a basis for the sample) is being planned.This Institution Population Survey would cover alltypes of resident institutions in the Uni~ed States,both medical and nonmedical. This includes alltypes of long-stay hospitals, penal institu~ions,orphanages, nursing homes, homes for the aged,and other types of facilities providing domiciliaryor personal care. It does not include such placesas convents, summer camps, hotels, student dor-mitories, and other places where presidents havethe responsibility of caring for themselves. Theselatter are not considered to be institutions andare therefore part of the universe covered by theHealth Interview Survey.

5. Special ad hoc surveys. —The surveyswhich have just been described are designed tocollect data for specific purposes. However, it isanucipated that tiom time to time there will be aneed for special data regarding the institutionalpopulation which cannot be met from thesesources. Consequently, special one-tie surveys -of specified parts of the Health Records Surveyuniverse will aLso be conducted.

To illusmate, much current kterest exists indata on the health of the aged, infirm, and chron-ically ill who are being cared for in resident in-stitmions of various types. More specialized in-formation is required than the Health InterviewSurvey ordinarily collects. Thus, the proposedInstitution Population Survey would not be a suit-able resource. For example, there is need for in-formation abut the characteristics of this groupwhich bear upon the problems of nursing or per-sonal care, such as, hearing and vision problems,mental status, awareness of surroundings, andabiliry to control feces and urine. In addition,other data are needed which relate to tie healthof these institutionalized older individuals, suchas information on the sraff available to care forrheir needs and the rype of establishment in whichthe persons are domiciled. For this purpose, there

13

..—. —-. —-— ..—- ._________ .,-!’,

. .

has been undertaken an ad hoc survey entitledSurvey oj Resident Places Providing Nursing OrPersonal Care, which will obtain statistic softhiskind.

DEVELOPMENTAL ANDEVALUATION STUDIES

The primary purposes of tie methodologicalstudies are to appraise the effectiveness andefficiency wirh which various aspects of tie Sur-vey program are meeting their objectives, and todevelop new and improved methods of artaining~he goals of the Survey; to design studies and de-vise data-collection methods in connection withnew types of morbidity measurement; and to con-duct pilot projects wherein new methods and newundertakings are prepared for full use in theSurvey.

Some methodological studies are conductedentirely by the staff of the National Health Survey,but studies which involve the special field collec-tion of data and which extend over any long periodof time are usually carried on jointly with theBureau of the Census or other government agen-cies, or are performed by nongovernmental re-search organizations chosen for their competenceand interest in the particular areas of study.

The fact that some of these studies are car-ried out as research contracts with nongover-nmental research organizations has two major ad-vantages for the Survey. First, it provides greaterflexibility in conducting the program by permitdngthe staff working on the methodological researchto be, in effect, expanded and contracted on fairlyshort notice as the need arises. Second, an evengreater advantage is thar it stimulates freshthinking about the many difficult problems chatare encountered. Already, at the time of thiswriting,a number of highly competent statisti-

cians around the countiy have begun to focustheir attention and bring their experience to bearon improving the quality of the surveys whichmake up this program.

At the same time, it is essential that membersof the National Health Survey staff associate them-selves closely with the work being carried onunder research contract. They must acquaint thecontractor with the viewpoint and objectives ofthe Survey, and they mus~ have a fullunderstand-ing of the results, since they will be responsiblefor putting the findings into effect. Consequently,at least one and usually two members of the Sur-vey staff are assigned LOwork with the contractoron each project. In many of the. studies they are

14

responsible for part of tie work, and they areusually responsible for editing the contractor’sreport to produce a brief version for publicationin the regular methodological series of the Na-tional Center for Health Statistics.

The Bureau of the Census likewise has an un-derstandably deep interest in this part of the pro-gram. Eventually, the Bureau may be asked toapply in the field newly developed methods fromthe studies. Furthermore, some of the resultscan be applied to other Census projects. Hence,s tacisticians from that agency often are involvedin the planning and carrying out of tie investiga-tions.

Not all methodological research is performedunder contract, however. A number of studies aredone with the Survey staff or jointly with tie Bu-reau of the Census. This is often the case whenthe srudy relates to improving interview-surveytechniques as these apply to health. Such studiescan often be done by adding some special pro-cedure in a subsample of households in the regu-lar Health Interview Survey. In such cases, thePublic Health Service makes a separate financialarrangement with the Bureau of the Census cocover rhe extra costs.

‘The types of projects undertaken under thegeneral heading of methodological research arehighly varied. However, certain characteristicsgive unity to the program.

First, the problems dealt with are those aris-ing in the course of developing and evaluating theHealth Interview Survey, the Health Examinationsurvey, and the Heakh Records Survey. That is,these problems represent applied research. Whilethe use of resources of the National Health Surveyfor general health-data-collection research is notbarred, the urgency of the many problems facedin the program until now has led to using thoseresources solely to develop and improve the sur-veys which were about to be stained or were al-ready in process.

Second, the methodological research can besubdivided quite easily into developmentals tudiesand evaluative srudies —hence, the title of thissection, which is the name used to describe thispart of the work of the National Health Survey.Occasionally, a study contains both elements: (1)an evaluation of the methods currently in use, and(2) the trial of a new method. However, mostsrudies can be identified as predominantly one orthe other.

Third, because of the inherent difficulties ofthe subject matter, the studies tend to be con-cerned with that component of the error of sur-veys known as measurement bias., Evaluating the

size of >them arheakhexist e’.due tO Londary !ing ranportantoften er

Bebiases.as: me[motivattake p:measu7standa:same cspecialreliablinsurarnemsvey.

Ataken

1.

2

5

. .

;hey areractor’sbliCaIiOII

the Na-

M an un-the pro-sked to

)ds fromresults

. Hence,involvedvestiga-

rformed]dies arethe Bu-

~se when‘-surveyI studies.ial pro-,le regu-wes, thefinancialensus to

,nder thearch areteristics

~se aris-ating the.minationThat is,

:h. Whileti Surveyrch is notms faceding those

tie sur-were al-

:h can be21srudiese of thisribe thisI Survey.lents: (1)I use, ander, mostly one or

:ulties ofbe con-

r of sur-lating the

size of such biases and learning how to minimizethem appears to be the first order of business inhealth surveys. These are biases which wouldetist even if no sampling was employed. Biasesdue to tie sample design are of distinctly sec-ondary importance. (On the other hand, minimiz-ing random errors of sampling can be most im-portant because of the low relative frequenciesoften encountered in healti statistics.)

Because of this concern with measurementbiases, much attention is given to such mattersas: memory of health events, clarity of concepts,motivation of persons falling into the sample totake part or to report accurately, replication ofmeasurements, calibration of insmuments, andstandardization of measurement techniques. Thissame concern also leads to a constant search forspecial situations where independent sources ofreliable information are available, such as healthinsurance plans, or where replicate measure-ments can be made to evaluate results of the sur-vey.

A few examples of research projects under-taken will illusmate the general description above:

1. A single-tisit cardiovascular examination,using controlled and standardized proced-ures, was designed and compared wirh amuch more extensive examination as wellas records of previous examinations forthe same group of persons. Interobserverdifferences were also studied throughrepetition of the examination on the sameexaminees.

2. Atitudes and other factors influencing ac-ceptance or rejection of an offer of a tieeheakh examination were studied with aview to discerning problem areas of non-response in the Health Examination Surveyand suggesting means for reducing nonre -sponse.

3. The reporiing of meated chronic conditionsin the standard version of the Health Inter-view Survey was studied by administeringthe intertiew to families for whom medi-cal-visit data were available from therecords of a health insurance plan.

4. Reporting of hospitalization in the HealthInterview Survey and tie effect of a sec-ond interview were studied in a group offamilies containing at least one memberfor whom hospital-records data wereavailable.

5. l%e results of using a paticular piece ofequipment for measuring visual acuitywere calibrated against a newly stand-

6.

ardized version of the Snellen charc bycarrying out both procedures in a sampleof patients attending an optome~ic clinic.lle interviewer-contributed component oftotal mean square error in the Health In-tertiew Survey was studied by using apart of the total sample within which in-terviewer assignments had been random-ized.

In addition to special methodological researchstudies of tie types described above, a number ofsmaller investigations are made, sometimes con-ducted through exploitation of data collected foranother purpose. For example, frequent analysesof the yield of specific illness probe questions inthe Health Interview Survey are conducted bymeans of hand tallies. Furthermore, every majornew topic in the Health Interview Survey, eachnew examination protocol for the Health Exami-nation Survey, and each questionnaire empl~yed inthe Health Records Survey is pretested in thefield. In these pretests, evaluation of the relia-bility of the new information is carried out bywhatever means appears feasible.

REPORTING THE RESULTS

A basic premise governing the disseminationof the Survey results is that findings should bemade available to interested persons as rapidlyas possible.

In keeping witi this principle, tie publicationplan provides for issuance of most Survey re-ports topic by topic. This conmasts with the hold-tig of tabulated material for use in a more tom-.prehensive but necessarily delayed periodic pub-lication such as an annual or semiannual volume.

Reports must be in a form usable for thelargest number of consumers of health statistics.This requirement dictates the arrangement of re-ports in series to the extent that this is applica-ble; reports within each series are then relatedin terms of the source horn which the informa-tion is drawn, the uses of the data, or other ap-propriate criteria. ,

Through 1962 four different series of publi-cations had been initiated by the National HealthSurvey: Sem.es A (program descriptions, surveydesigns, concepts, and definitions); Series B(Health Interview Survey results by topics); Se-?+es C (Health Interview Survey results for pop-ulation groups); and Series ~ (Developmental andEvaluation Reports).

Late in 1962, however, a unified system ofpublication series was designed for the National

IS

—._ -—. ..’r

Center for Health Statistics of which the NationalHealth Survey is a component. Under this sys-tem, Series A and D were redesignated as NCHSSeries 1 and 2; these cover the same subjectmatter as they had previously but will now beused for the Center as a whole. The Series num-bered 10-19 have been set aside for other publi-cations which will be issued by the NationalHealth Survey. Number 10 will be a condnuationof former Series B and C. Numbers 11 and 12will include reports from the Health ExaminationSurvey and the Health Records Survey, respec-tively. Other series will be originated as needed.

The timetable for issuing publications de-pends, of course, on the availability of resultsfrom the surveys. Since in most cases tie resultsare estimates derived from samples, the publi-cation depends on the accumulation of a sufficiemamount of data to permit the making of estimateswith a useful sampling reliability. For example,for most topics covered by the household inter-view questionnaire, data from a full year of in-terviewing are needed to compile national statis-tics in a reasonable degree of detail and a 2-yearsample is needed 10 publish selected estimates bygeographic region. Occasionally, as much as 4years of sample data are combined to providemore detailed tabulation.

Frequency of reporting on a particular topicalso depends on the rapidity of the change in tieof the health characteristics included under thartopic. For example, reports on the incidence ofacute illness will appear more frequently thanreports on the prevalence of physical impair-ments. ACtie same time, early reporting of pre-liminary results is not intended to preclude amore intensive treaanent of the same topic at alater date.

Thus, it may be seen that the system of re-porting results of the surveys is based on a divi-sion imo several publication series, according tothe source of the data or type of report, with in-

. .di;iduaI issues usually devoted to a single topic.Periodic summaries of data on a number of topicsfor a particular period of time are not etiinatedrbut the reporting of all results in a regular-issueamual or biennial volume is not part of tie plan.

The system of issuing reports in establishedseries does not preclude preparation of such ad-ditional reports as circumstances and feasibilitiesmay warrant. In addition to the reports issued inseries, special publications may be preparedhorn time to dine. The weekly Provisional Tab-ulation— Current Slafistics on Res#irat@y fis -eases is an example

Partly because the household-interview sam-ple is not large enough to yield current esdmatesfor items of low incidence or prevalence, weeklyor monthly tabulations were not planned. How-ever, in the fall of 1957 when it appeared thatcases of influenza-like diseases would have ahigher than usual incidence, an investigation wasmade to learn if in this instance the household-interview data, obtained infrequent time intervals,would yield useful information.

A special tabulation of questionnaires, show-ing upper respirator y diseases severe enough tocause persons to go to bed for a day or more, in-dicated thar the data were an adequate basis forissuing provisional tabulations on new cases whichappeared each week and on the average number ofpersons contied to bed each day. These two in-dexes were incorporated in a weekly publicationbeginning in November 1957. A special publica-tion such as this would be condnued only throughthe period of its topical interest.

In addition to the publications of the Survey,Public Health Refioyts and professional journalsserve as a means of circulating information ob-tained from the Survey program. These mediaare particularity appropriate for publication ofsome of the research papers emanating from theDevelopmental and Evaluation Studies. The con-tractors and cooperating agencies working onthese aspects of the Survey often are in a positionto contribute notably to the literature in special-ized subject areas.

Finally. the Survey attaches importance to theobligation s~ated in Public Law 652 to ‘rMakeavailable, to health officials? scientists, and ap-propriate public and other nonprofit institutionsand organizations, technical adtice and assistanceon the application of statistical methods to ex-periments, studies, and surveys in health andmedical fields. ” The consultation which tiis re-sponsibility implies is an important part of theoverall plans to disseminate the knowledge whichthe Survey produces.

Potentials AND LIMITATION:

OF THE SURVEY PROGRAM

Effectively pursued, the Survey programshould provide a comprehensive cross-sectionaland changing picture of illness, accidental inju-ries, disability, and the use of hospital and othermedical and dental services. However, recogni-”don of the value of the program in defining more

clearly t:of the N:variousexist 130t’veys gro

Con(exampleness orthe per~sician,househcThe sarand socrelate I

HoCy of ti-the houviewerh~s ginot hahave rcian s.

FdiagncrespmSympthouse]

Fare areporof sioccasof ccone r

7requiof inreco:dazathateverIimithethe :thison tof tter~

[6

I

# sam-imatesweekly

Hew-ed thatIave acm was;ehold -srvals,

, show->ugh tore, in-sis for

whichnber oftwo in-dicationublica-hrough

:urvey,mrnalsion ob-media

ition of

‘om thele con-king onmsitionpecial-

:e to the“Make

.nd ap -,Sutionsistanceto ex-th and

his re-: of the: which

y$

M

rogramsctionalal inju-id otherecogni-”ig more

clearly tie health and factors affecting the healthof the Nation should not obscure the fact that thevarious surveys have definite limitations. Theseexist both in the individual surveys and in the sur-veys grouped collectively as a system.

Concerning the Health Interview Suey, forexample, the facts of the circumstances of fll-ness or injury and the resulting action taken bythe person, such as going to bed or seeing a phy-sician, can be more accurately obtained hornhousehold members than from any other source.The same is mue of the demographic, economic,and social information with which it is useful torelate morbidity information.

However, there are limi~ations to the accura-cy of the diagnostic information collected. At best,the household respondent can pass onto the inter-viewer orly the information which the physicianhas given to the family; and the respondent maynot have been told about a condition or he mayhave misunderstcmd or forgorten what tie physi-cian said.

For conditions not medically amended, thediagnostic information supplied by a householdrespondent may be no more than a description ofsymptoms, and for nonmanifest diseases thehousehold interview will report no condition.

Furthermore, it is known fiat by no meansare all chronic conditions treated by a physicianreported by tie family in an interview. Onlythoseof significance to the family are reported, andoccasionally a respondent, despite the assuranceof confidentiality, will conceal information forone reason or anotier.

Thus, when clinical or diagnostic detail isrequired or when the statistics mustpresenttypesof information which are available in physicians rrecords, interview data are insufficient. Thesedata must be supplememed by information such asthat produced by the health examinations. How-ever, the Health Examination Survey has its ownlimita~ions. Because of the high cost of performingthe examinations and difficult logistic problems,the sample must be relatively small. As a result,this survey canner supply the volume of materialon the demographic, social, and economic aspectsof health which is available from household in-terviewing.

A disadvantage of giving the health examina-tion only once is that it cannot provide diagnosesof the many conditions which require repeatedand continuous observation and tests before theycan be identified. Also, the limited character ofthe examination precludes the use of differentialdiagnostic tests or complicated procedures.

A further general limitation ~f the HealthExamination Survey is the size of thenonresponse.However, ir is hoped that appropria~e use of theinformation elicited in the household interviewfor the health examination sample will minimizethe impact of the nonresponse on final results.

Together, the three types of surveys whichmake up the program—interviewing, examina-tions, and sampling of the sources of care-pro-vide useful information on a wide range of vari-ables. However, under present plans, the samplesizes are such as to give estimates for majorgeographic areas only. No estimates for individ-ual States or cities (other than the largest metro-politan areas) can be made.

A similar resmiction of the system as awhole, as it is now contemplated, is its inabilityro provide statistics for persons in small groupsof the population such as, for example, a small,hazardous occupational group. The same hoIdsmue for diseases of rare occurrence. These oftenare of interest to a degree far greater than thatwhich their frequency migh~ warrant.

Statistics from the program do not providecrirical tests of clinical and epidemiologicalhypmheses. For example, the program could notrest the hypthesis fiat a special vaccine wouldprevent a certain disease. For this, an experi.mental design, a control group, and similar con-ditions would be required. However, the programmay suggest hypotheses that can be tested byother appropriate means.

Aside horn these limitations imposed bymedmds being used and resources available, theprogram is free 10 collect any statistics on theincidence, prevalence, or other measures of dis-ease, injury, or impairment; the disability orother effects of this morbidity; and the medicalcare used in its treatment. The sole guide, withinthe technical and administrative possibilities, isthe usefulness of the data.

I 17

APPENDIX I

(Raprint=d below is tha text of Public Henlth %rvice publication No. 333. A list of the members of the U. S.

Notional Commiiiae on Vital and H,alth Statistics ond of the Subcommittee on Notionol Morbidity Survey ap-

p.ars en page 39. This R-port W.S p.blishcd in October 1953.)

RECOMMENDATIONS FOR THE COLLECTION OF

DATA ON THE DISTRIBUTION AND EFFECTS OF ILLNESS, INJURIES,

AND IMPAIRMENTS IN THE UNITED STATES

A REPORT OF THE SUBCOMMITTEE ON NATIONAL M@l?BIDITY SURVEY

U. S. NATIONAL COMMITTEE ON VITAL AND HEALTH STATISTICS

PART I

Summary

This report contains a reveiw of the needs forstatistics on illness in the general population, andthe current and potential uses of morbidity statis-tics and related data. On the basis of this reviewand study of the present sources of morbidity andmedical care utilization statistics, a proposal ismade for a national morbidity survey 10 provideneeded information on illness in the United States.The report describes the data collection mecha-nism with specifications as to the types of data tobe secured, the geographic area to be covered,and accuracy of data.

Part H of this report provides background in-formation and a general frame of reference for tieparts which follow. Included are statements re-garding the work of the two earlier ad hoc com-mittees which did certain exploratory work in thisgeneral area, and the specific charge to this Sub-committee by the U.S. National Committee on Vitaland Health Statistics.

Part 111 presents art omline of some of themore important currem and potential uses of mor --bidity statistics and related data.

In Part IV there is discussion of the generaltypes of data, the degree of detail and the frequen-cy of collection which appear necessary to serverhe purposes outlined in Part 111.

Part V is largely devoted to consideration ofthe question of whether etistig programs of datacollection can serve to satisfy the needs for mor-bidity statistics which have been identified. First,there is discussion-of recent improvements in themethodology of data collection in this field. Next,the presem sources of morbidity and medical careutilization statistics are reviewed and evaluated.Finally, the conclusion is reached that adequatecurrent morbidity data on a national basis cannocbe derived from any of the present sources.

Parts VI and VII contain the recommendationthat a continuing national morbidity survey be con-

19

— . ..—— — ---,.J

ducted together with proposals regarding the typesof data which should be collected and a proposeddesign for such a survey. The recommendationsmay be summarized as folIows:

1. That a continuing national morbidity sur-vey be conducted which is adequate to provide sta-tistical estimates for 50 regions of the counmy atintervals of two years and estimates for the narionas a whole at quarterly intervals. The puqmse ofthis survey would be to obtain data on the preva-lence and incidence of diseases, injuries, and im-pairments, on the nature and duration of theresulting disability, and on the amount and typeof medical care received. The data should be

obtained tiom a probability sample of house-holds .

2. That a series of special studies be under-taken in addition to the continuing national survey.The principal purposes of these studies would beas folIows:

(a)

(b)

To obtain data on undiagnosed and non-manifest disease, by means of labora~oryscreening, detection, and physical exam-inations of subsamples drawn from thegeneral surveys.To provide other types of auxiliary data,and to study methodological problems re-lating to the measurement of morbidity.

PART II/

The Work of Previous Committees and the Task of the Present Subcommittee

Since its establishment in January 1949, theUnited States National Committee on Vital andHealth Statistics has given attention on severaloccasions to the need for more adequate nationalmorbidity statistics. Two successive adhoc com-mittees were established by the Chairman of tieNational Committee “to frame the problems inmorbidity statistics, including chronic diseasesand medical care statistics, in order that morbid-ity data may be directly related to demographicfactors.”’

In December 1949, che first of these commit-tees made its report which was a review of thebackground of the problem. It also contained anoutline of the major sources of morbidity and oth-er vital and demographic statistics. The broad ap-proach taken by this committee is indicated by theface that recommendations were made in the fol-lowing fieIds:

1. Study and revision of reporting system fornotifiable diseases.

2. Adaptation of survey techniques for healthpurposes.

3. Chronic disease statistics.4. Development of standards and definitions

for hospital morbidity statistics.5. Study of means of obtaining adequate cur-

rent population data for small areas.6. Longitudinal studies.7. Place of case registers in morbidity sta-

tistics.8. Development of mechanism for providing

civilian casualty statistics in time of war.

The second ad hoc commirtee submitted itsreport in October 1950. This covered an almost ‘equally broad area and recommendations weremade for further study in each of the fields listedabove except the Iast but with the addition of arecommendation regarding the development of

I

standards and definitions for general morbiditystatistics. Moreover, by this time the committeewas able to take note of several new facts: (1) thata thorough overhauling of the notifiable &seasereporting system had been initiated by the PublicHealth Service; and (2) that a special Subcommit-tee on Hospiral Morbidity Statistics had been secup to which the problem of standards and defini-tions for hospital morbidity statistics could be re- ,ferred.

Both committees had urged an investigation of ~the methodological problems in the collection ofstatistics on chronic illness and disability bymeans of surveys. The types of investigationsneeded were listed under three major headings: :.

A. Study of survey techniques.B. Investigation of specialized health, hos -

pital, and medicaI records for utilizationon subsamples of the population.

C. Investigation of specialized respondentssuch as physicians for utilization on sub- .samples of the population.

Even as these recommendations were being .made, moves were under way to study all three of :these matters in conjunction with surveys in sev-eral parts of the coun~. As will be indicated in IPart V of this report, answers to some of the

]:

v

J

methodfrom STNew J,Pittsbuthe Putbeing PIshould I

Wi [for spelectioncare b’Nation:commilthe tas’ity SIX

areas.’term,sider ,

mthis taad hocthesestudie~presertype owould

11 Nmeansdeviat’in spedenot.whichnot be

Tof themedicmorbiized Tthe SFtail ncuses 1has m;

Tbroadexist,mernhthe Pt

::

20

ouse-

rider -mvey.lld be

i non-‘ator y>xam-m the

data,IS re-lidity.

ttee

:ed its.lmost

werelistednofamc of‘biditynittee1) thatsease

Public,nmit-‘en setiefini-be re-

.tion of[ion ofity byationsdings:

, hos-?arion

ndentsn sub-

beingree of.n sev-xed inof the

methodological questions will soon be availablefrom studies in New York City; Humerdon Coumy,New Jersey San Jose County, California; andPittsburgh, Pennsylvania. Other experiments bythe Public Health Service and new surveys nowbeing planned will provide further information thatshould lead to improvements in techniques.

With the knowledge that the time was at handfor specific recommendations regarding the col-lection of national data on morbidity and medicalcare by means of surveys, the Chairman of theNational Committee established the present Sub-committee in February 1951 and charged it withthe task of drafting “a plan for a national morbid-ity survey keeping in view the interests of localareas. ” Because of ambiguity in the use of theterm, the Subcommittee was also asked to con-sider a definition of “morbidity.”

The Sulxom.mittee has attempted to fulfillthis task. It reviewed carefully the reports of thead hoc committees and attempted to take up wherethese groups had left off. The Subcommittee hasstudied the needs for morbidity statistics, tiepresent sources and their inadequacies, and thetype of survey mechanism and special studies thatwould meet the apparent needs. The conclusions

PART

Current and Potential Uses of

“hiorbidity statistics ,“ as used in this report,means quantitative data usable as measures ofdeviations from health in the general ppulation orin specific population groups. The term does notdenote any statistics regarding health or sicknesswhich, while meaningful for other purposes, can-not be related to a particular population at risk.

The purpose of this section is to outline someof the practical questions relating to health andmedical care which require accurate and detailedmorbidity statistics for their solution. It is real-ized that an important element of the problem istie specification of the degree of accuracy and de-tail needed in morbidity statistics for each of theuses listed. However, discussion of these mattershas not been attempted in this outline.

There will be found below a summary of thebroad areas in which demands for morbidity dataexist, as determined from the experience of themembers of the Sukommittee, from the files ofthe Public Health Service, horn the responses of a

and recommendations are contained in the sectionsthat follow.

To the matter of the definition of “morbidity”the Subcommittee gave careful artention. No pre-cise limited definition of the term has been madefor the following reasons:

1. The term “morbidity” is and should re-main a general word to be used to designate ill-ness (manifest and nonmanifest), injuries, and im-pairments.

2. The definition of The particular morbidconditions that one wishes to count or describe instudies and surveys must vary according to theobjectives of the particular study or survey.Hence, no one definition of a “morbid condition”can easily be adhered to.

3. For any regular program of national mor-bidity surveys an operational definition of what isto be counted or described must be devised. Thisoperational definition must be shaped to suit rheparticular purposes tiat the statistics are to serveand yet must be a definition that can actually beapplied in the field.

4. The task of devising an operational defini-tion will be left to those who will carry our themore detailed phases of planning for the programof national morbidity surveys.

Ill

Morbidity cnd Related Data

limited number of &tential users of morbidirydata who were quesuoned by members of the Sub:committee, and from otier sources. I It must beemphasized that there is no implication intendedthat all of the data needed can feasibly be obtained,nor =at collection of data on all the items out-lined below would necessarily be warranted in anational morbidity survey.

l%e categories in the outline are not intendedto be mutually exclusive in every in.mmce.

IIC j~ &]i=red hat ● mom complere and represcnmciw SLUVCY

of the needs of pnmntinl consumers for morbidity dmra should be

mmdc. To da c.his usk tbomughly would require more time -d rc-

MUCCS dmn the !Mtcommimee h~s ntilrnble. I-fowemr, cermh

letters mnd oher c7idcnces of ❑eeds snd uses of morbidiry sc.tis-

ucs on r.hc generml ppukion collecred by AC SuLxommimee Ue

on file.

21

.,.1

Broad Areas of Need for

Morbidity Statistics and Related Date

A. Statistics for use as a guide to adminis-trative planning and evaluation of official and vol-untary programs in the field of health.

In this area morbidiry statistics would beused direcLly for the purposes at hand.Examples in tiis area are:

(i)

(2)

(3)

(4)

(5)

B.

Use of various measures of morbidityand disability for ranking of problems ofpublic health in order of importance and,in general, for determining how re-sources should be divided among nu-merous programs.Checking on the adequacy of reporting ofnotifiable diseases and establishing base-lines of endemic prevalence.Provision of quantitative data for planningnew programs of contiol and for estimat-ing probabIe maximum expendituresneeded for such programs.Determination of trends in incidence andprevalence of specific diseases for evac-uation of the effect of preventive andtherapeutic innovations.Cohort-t~e studies for evaluation of pre-ventive and therapeutic measures.

Statistics for the evaluation of the currentmorbidity experience in relation to the provisionof medical and dental services, facilities andper-sonnel for meeting the heaIth needs of the nationor of a community.

(1)

(2)

(3)

(4)

Provision of quantitative information onwhich co base sound estimates of needsfor general, chronic, and other specialhospital facilities, nursing home beds,and home care facilities, either in thenation or in particular localities.Estimation of che numbers of persons re-quiring particular rehabilitation serv-ices, to assist in planning for the provi-sion of such services.

Statistics on the frequency of chronic dis-ease, injuries, and resulting disability,for use in considering the cost of extend-ing the scope of medical care insuranceor compensatory insurance of varioustypes.Numbers of persons covered by varioustypes of insurance plans, and amountsspent and methods of payment for medi-cal care, so that study of the economicsof medical care may reflect modern con-ditions.

(5)

(6)

Statistics on the use of medical servicesand personnel in various income andoccupation groups, for establishing theactuarial base of new insurance plansdesigned to cover the medical expensesof particular groups of the population,such as the members of a tiatemalorder.Statistics on the various categories ofdisease and on the receipt of medicalcare, to be used to guide the planning ofmedical and nursing education by indi-cating where emphasis in education canmost usefully be placscl.

c. Statistics for medical research. Whilemorbidity staustics from surveys of the generalpopulation cannot usually be used to make conclu-sive tests of hypotheses in medical research, theycan be useful in suggesting hypotheses for furthertesting and in providing other aids to research.Examples in this area of use are:

(i)

(2)

(3)

(4)

(5)

(6)

D.

The provision of rosters of cases for theintensive investigation of disease etiol-ogy.Information on the association betweenthe incidence or prevalence of variousdiseases and demographic faccors, suchas age, sex, marital status, occupation,and economic status.Provision of data to help research work-ers ti the field of preventive medicineselect segments of the population havingthe greatest risk of developing specifiedmorbid conditions, thereby permitigthe selection of smaller groups for fol-lowup in tests of preventive measures.Statistics on the prevalence of certaindiseases, for use in estiating gene fre-quencies in the population.Geographic distribution of diseases, suchas allergic conditions, numiuon~ dis-eases, and nephritis, to provide clues asto etiology.Description of the natural history of dis-eases, both acute and chronic, in repre-sentative samples of the population toprovide more quantitative lmowledge ofthe preventive aspects of the disease asit occurs in nature.Statistics relating to certain manpower

problems and civilian defense. Both civilian andmilitary agencies concerned wizh problems ofmanpower civil defense need information on thephysical fimess of the population and the numbersof persons falling into various categories by type

and dein this

(1

(2

,

(

GerrE

tisti;sitatt

22

n,“

mhcesne anding thee plansTensesnation,sternal

ries of,ledicalning of)y indi-ion can

While~eneral:onclu-:h, theyfurtherearth.

for the? etiol -

]etweenrarious;, suchpation,

I work-edici.nehaving

]ecifiedmimgor fol-asures.certainme tie-

.S , suchal dis-:lues as

.,of dis -repre-tion toedge ofsase as

.npower

.ian andems ofI on thembersby type

and degree of disability and occupation. Examplesin this area are:

(1)

(2)

(3)

(4)

(5)

Information on absenteeism resulnngfrom disease and injury, for use in esti-mating economic loss to industry throughvarious types of morbidity.Information on the numbers of personswith chronic diseases and handicappingconditions by type of disease or handicap,and the employment status of such per-sons, for use in estimating potential addi-tions to tie labor force.Similar information on young men ofdraft age, for estimates of numbers whomight be drawn into the Armed Servicesin limited duty categories.Data on the morbidity rates of handi-capped persons as compared with thenonhandicapped; data on the personneland facilities required to correct handi-caps that are subject 10 correction, forthe use of indusmy and the Armed Serv-ices.Statistics to answer the following types ofquestions for civilian defense agencies:(a) What proportion of physicians andnurses are physically fit so as to beavailable for duty in the event of bombingartacks or for other types of disaster?(b) Whar proportion of the general popu-lation in an area would need special at-tention because of disability in makingplans for evacuation of the population inanticipation of, or following, bombingartacks? (c) What proportion of thepopu-lation could be considered suitable as apotential source for providing blood?(d) What proportion of the populationwould require special attention such asextra rations or vitamin supplies, inplanning for the nutrition needs of abombed area? (e) In the event that evac-

PART

(6)

E.

uation camps are established, what pro-portion of the population of these campswould require such drugs as insulin fordiabetes or liver exmact for perniciousanemia? (f) What are the endemic levelsof the incidence of nonreportable acmediseases, such as gasmoenteritis, inplanning for defense against biologicalwarfare?Incidence of chronic disease in olderworkers, for the use of indusmial firmsinterested in knowing the risk in em-ploying such workers.Statistics for the use of dru~ firms and

appliance manufacturers in estima&g marketsfor particular preparations or appliances. Thisuse requires not only the usual types of informa-tion on incidence and prevalence of diseases, in-juries, and impairments, but aLso statistics on Lhecurrent utilization of medical services of Wrioustypes, for example, frequency of particular opera-tions, and also frequency of prescriptions and ofuse of hearing aids, artificial limbs, and so forth.Incidence and prevalence of disabilities by part ofthe bcdy affected are Requently needed.

F. Statistics for public health education pro-grams. This is one of the most tiequent of tietypes of request for statistical information re-ceived by health agencies. Examples in this areaof use are:

IV

(1)

(2)

Types of Data, Detail, and Frequency

Estimates of the national ticidence of ac-cidental injuries by type and degree ofdisability caused, for accidem preventionagencies; estiates of tie prevalence ofcerebral palsy, multiple sclerosis ,blind- -ness, deafness, and many other diseasesand impairments, for voluntary healthagencies concerned with these condi-tions.Similar information for use in advertisingcampaigns with a public health educationobjective.

of Collection Required

General Morbidity Data dence, prevalence and the duration of disabilityfor the major categories of disease and impair:

A minimal adequate program to provide sta- ment. The data should allow subclassification bytistics for the uses listed in Part HI would neces- sex, by several age groups, and by employment,sitate collection of data of national scope on inci- educational, income and occupational status, and

23

——

B...’ 9

should have sufficient regional geographic detailto be usable by the Health Officers of States orcities.

The amount of diagnostic detail requiredv=ies greatly wirh the particular application. Aclassification by major diagnostic categorieswould probably be sufficient for a general apprais-al of tie principal health problems in an area, asmentioned under heading A of Part 111. With somediseases, much more detail would be needed tiorder to provide background data for estimates oftie professional services and facilities needed ina program of prevention and care. For many ofthe uses described in Part 111, it appears that thedata should include undiagnosed illness. Nonrnani-fest disease or heretofore unrecognized cases ofdisease should also be included, at least in so faras appropriate followup measures can be speci-fied.

In order to be most useful, the general mor-bidity data must include more than simple inci-dence and prevalence rates. The data must de-scribe:

1. Some of the facts about the natural historyof morbid conditions, such as duration, type,, andextent of incapacities (hth primary and second-ary).

2. Some of the facts almut the social conse-quences of morbid conditions, such as: changes inemployment status, source of support, housekeep-ing assistance and nursing service required.

3. Morbidity indices should be related 10 da-ta concerning membership in prepaid health in-surance plans, and the types and amoums of serv-ices received from such plans.

4. The data should include information onudlization of medical services and facilities. Thefollowinz woes are of interest:

(a)

(b)

(c)

(d)

Physician’s services (classified as tohome visits of general practitioners, of-fice visits 10 general practitioners, oficevisits TO specialists, hospital visits ofphysicians of all types).Dental services (classified as to visits togeneral practitioners, visits to special-ists).Nursing services (classified as to homevisits, services in hospitals, services inother institutions).Hospital services (classified as to serv-ices in general, tuberculosis, mental, andother types of hospitals, and “near hos-pital” care, such as nursing home careand convalescent home care).

5. Another ads of classification of the medi-cal services and facili~ies is also frequentlyneeded:

24

(a)

(b)

(c)

Preventive services (classified as pri-mary - prevention of disease occurrence;and secondary - prevention of diseaseprogression by periodic examinations,prevendon of sequelae, and other pre-ventive services).Diagnosis and ~eaunent services (clas-sified as to inpatient and outpatient serv-ices).Rehabilitation services (classified as toinpatient and outpatient services).

Frequency of Collection

The requests for data cited in Parr HI do notspecifically mention any particular frequency ofcollection that would be desired. It is to be noted,however, that practically all the data requestedare for the purpose of future planning of somekind. Thus, the requests are predicated on tieassumption that the data would be current, or atleast sufficiently up to dare so that pIans for thefuture could safely be based upon them. How up todate the data need be in practice depends, ofcourse, on the rapidity witi which changes in mor-bidity or in the major health problems occur. ThiswiII vary from item to item.

An efficient program of data collection tomeet the requests in Pare III should have the fol-lowing features. It should permit the publication ofbasic data on the prevalence and incidence of dis-ease, injuries, and impairmems, on tie natureand duration of the resultig disability, and on theamount and type of medical care received, at in-tervals of two years. ‘The program should alsohave considerable flexibility for the foHowingreasons:

1. For some data, publication at less fre-quent intervals than two years will be adequate.The program should permit such data to be ob-tained at relauvel y little exma expense.

2. Although it is ~ssible to rank the variousrequests for data, at least roughly, in their orderof importance at the present time, this rankingwill change as time proceeds. The program shouldmake it possible to obtain, at relatively shortnotice, data which become of paramount impor -~ance.

3. In the event of an intensificaaon of the in-ternational crisis, there may be a sudden demandfor data at more frequent intervals or for certainspecific dara at very short notice.

As will be discussed in Part VI, these con-siderations point toward a continuous programof data collection, rather than toward a rigid, in-termittent schedule of surveys any two or fiveyears apart.

IiGeogr

~whichare erlevelsSuffic:of thEgeogrfor thofacdema!speci”ther~aphbidit>bornHeal:by pcownin wgrapwillrela:man.mentjudgrmiseand

knowlrapid

statitimeof eeasecialsr epozare i

the rthes~is ofof itfindil1936

m..as pri-:rence;LseaseiatiOnS,.r pre-

(clas-Ltserv-

,d as to

,1do notlency of? noted,ques~ed)f some

on theit, or at

for the3Wup tomds, ofinmor -lr. This

crion tothe fol-

cation of? of dis-

natureId on the1, at in-lld also>Ilowing

~ss &e-iequate.o be ob-

Various:irorderrankingm should~y short

impor-

.f tie in-I demandr certain

ese con-program.-i@d, in-J or five

Geographical Detail

Most of the needs for morbidiry data, ofwhich the examples in Part III are illustrative,are encountered on national, regional, and locallevels; therefore data are necessary that will havesufficient geographic specificity to be useful at allof these levels. The designation of tie degree ofgeographic detail in which sta~istics are requiredfor the purposes of, let us say, the Health Officerof a cicy is, of course, a function of the accuracydemands of the particular job. However, thespecification of desirable geographic detail is fur-ther complicated by a lack of data on the gm-graphical variation of the common indices of mor-bidity. If a particular index is relatively stablehorn one area to another, men obviously theHealth Officer’s needs may be adequately servedby population rares for all cities of the size of hisown in the broad geographic region of the counmyin which the city lies. On the other hand, if geo-graphic variation is very great, perhaps nothingwill serve the Health Officer but data specificallyrelating to his own city. For the me being thismatter must be semled on the basis of the judg-ment of experts unsupported by evidence. Suchjudgment will seek to find a reasonable compro-mise between the ideal of ~eat geographic detailand the practical dictates of cost. As is wellknown, the cost of obtaining the data increasesrapidly with the geographic detail required.

Individual Studies

Several of the examples cited in Part 111would require specialized studies which havelittle relation to one another or to a national pro-gram of data collection. Examples are: tie pro-vision of rosters of cases for investigation of dis-ease etiology [C(l)], and statistics to indicatethe propcruon of physically fit physicians andnurses [D 5(a)]. Such studies would be best con-ducred individually by the interested bodies.

Interest has also been expressed in studiesshowing: (1) the modifications brought about in theusual course of chronic diseases by adequate pro-fessional care; and (2) the manner in which someof the social consequences of prolonged illnessmay be avoided and economic savings broughtabout by adequate medical, social, and rehabilita-tion services. However, the Subcommittee feltthat these were beyond tie scope of its assign-ment. Such investigations should be made bymeans of independent studies.

The use of’ the indices of morbidity and medi-cal care produced by this survey to estimate thepersonnel, facilities and services required inmeeting the health needs of the nation or of acommunicy presents difficult problems. No singleapproach to such estimation is ideal but severalmethods may be considered. These are discussedin this Appendix.

PART V

Review of Present Sources of

Those unfamiliar wirh sources of morbiditystatistics relating to the general population some-ties have the impression that there is no deartiof estimates of the prevalence of various dis-eases and impairments. Speeches by public offi-cials, testiony before Congressional committees,repcms of conferences and statistical’ ‘fact” booksare full of estimates that might seem to satisfythe needs for such statistics. If the genealogy ofthese estimates is mated back (and their lineageis often complex), a remarkably high proportionof them are found to have descended from thefindings of the National Heslrh Survey of 1935-1936.

National Morbidity Statistics

The National Health Survey

This survey was an attempt on the part of thePublic Health Service to find out tie number, age,sex, income level, and occupation of people in tieurban population of the United Stares who had ex-perienced disabling illness witiin the 12 monthsprior 10 the visit of an interviewer or who at rhetime of the interview were believed to have aspecified chronic illness or impairment (whetherdisabling or not). The Survey, despite its magni-tude (737 ,000 households) and complexity was avery well-planned project. The staff includedmany competent statisticians and, because of the

25

. ,

prevailing lalmr market and the cooperation of theU.S. Employment Service and the W. P. A., it waspossible to obtain a reladvely high grade of work-ers for interviewing, coding, and so ford’t. Thoughcritics have lately pointed to a number of short-comings of the survey plan, the knowledge of sur-vey design upon which these criticisms are basedwas, for the most part, not available to those whodid the planning. The period, 1935-1950, was oneof very rapid development in all phases of the sci-ence of data collection in ~pulation surveys.

There is no question but that the data fromthe National Health Survey have been put to verywide use. Owing to the lack of really appropriatestatistics, the information has been made to servepurposes for which it was not well suited. Dr.Leonard A. Scheele, Surgeon General of the PublicHealth Service, in testimony before the Subcom-mittee on Health of the Senate Committee on Lalmrand Public Welfare (August 23, 1951), stated asfollows:- “In the past 15 years, findings from cheNational Health Survey of 1935-1936 have formedtie basis for about two hundred reports, articles,and comparative studies. The survey data havebeen used to project estimates for more recentyears and for individual communities in anattempt to measure needs for hospital and otherfacilities and community services. Many of theNational Health Survey figures are of somewhatdoubtful applicability to present-day conditions.”

The extensive use of that material right up tothe present tie, even by people who were awareof tie risks involved in using it, is of itself asmong indication that morbidiry statistics are ingreat demand.

Advances in Sampling

and Survey Techniques

No survey of national scope undertaking toobtain more dmn a few very limited items on ill-

. . negs and its consequences has been made sincethe National Health Survey. However, public healrhand sampling statisticians are now in a much bet-ter posiuon to say how a comprehensive surveyor surveys should be planned in order to produceuseful and accurate data. The following areas ofadvance are illustrative:

1. From the uses to which the results of theNational HeaI~h Survey and other studies havebeen put we have learned more about the types ofdata it is necessary to have in a well-roundedprogram of collection.

2. Much has been learned from researchdone in Census work and in non-Gove~ental

fields about berter methods of el.icidng accuratedata by means of questionnaires and interviews.In the design of schedules there have been somemarked improvements.

3. The developments in mass X-ray surveysand the use of multiphasic screening techniqueshave taught us a great deal about the problems ofdetecting nomnanifest disease. Experiments withthe use of symptom schedules have had some suc-cess in picking out persons who are in need ofmedical care but who for one reason or anotherhave not sought medical assistance.

4. Longitudinal and cohort-type studies haveprovided new information about the characteri-stics of illness, particularly chronic illness, thefrequency of disabling attacks, the clusteringwithin households, and similar data. The experi-ence in these intensive studies will prove usefulin designing the more exrensive type of survey.

5. The very rapid advances that have beenmade in the development of applied samplingtheory have “showed how it is possible to designprobability samples of human populations that pro-vide much greater sampling precision per dollarexpended, and during this development of theorythere has been a wholesome emphasis on makingthe theory and the field application coincide. Thegrowing realization that aU sources of error, andnot alone that which is due to sampling, must beconsidered in planning and kept under con~ol isparticularly important for illness surveys.

Other Recent Studies

Several srudies recently completed, in plan-ning, or under way will add “some rather specificmethodological results in the field of the meas-urement of illness. Among these studies are:

1. The Baltimore Eastern Health Dis&ict= in which the white population of a sample of35 blocks in the health dis&ict was studied forperiods up to 5 years (1938-1943). Actually 17 ofthe blocks were included in the study for 5 years,another 17 were included for 3 years, and oneblock was dropped early in the third year of study.A small corps of well-trained interviewers mademonthly visits to all families residing in theblocks at the time of the visits. Information wasobtained on illnesses or injuries that had occurredsince the previous visit and on all medical careincluding preventive services received in the in-terval. hy responsible member of the family whohappened to be present was accepted as the re-spondent. Diagnostic information on cases of ill-

ness that had been artended by a physician waschecked with the physician or with hospital or

Inartcec’

Milba~large 1among

[

01.li kr

veys cfound,

1 eases.1 first

interl, or mcI a disr

d manyof tirr

I that Twereotherthat !the cand cof [htinue

d

Qim emSusServstud;andobj e(usinin C(specCalirieddiarCorn.hav

. terbee.Sanundodsmet

Heapro’of irec(Thi.ConFOIJWittund,viel

26

I 1. .:curate‘views.I some

rrveys,niquessms of;s withle .suc-leed ofnother

!s have:teris-3s, theteringxperi-useful=vey.s beennplingiesignIt pro-dollarfieorylatig?. Thejr, andust berol is

plan-ecificneas-: are:~

ple of?d for17 of

rears,i one.tudy.maden thei wasmredcaree in-YwhoE re-f ill-I was,al or

clinic records. The study, which was jointly fi-nanced by the Public Health Service and theh’lilbank Memorial Fund, has provided data for alarge number of special statistical investigationsamong which are several which will contribute toour knowledge of how to make more accurate sur-veys of the less intensive type in the future. It wasfound, for example, that many of the chronic dis-eases present in the population at the time of thefirst visit were not brought ~o the attention of tieinterviewer until the family had been visited sixor more times. It was probably the occurrence ofa disabling attack that led to the first reporting ofmany of these diseases, but because of tie lengthof time that the families were followed it is likelyrhat nearly all serious chronic diseases presentwere eventually picked up. To mention only oneother of the numerous varieties of investigationThat has been made, it has been ~ssible to studythe course of chronic illness in terms of frequencyand dismibunon of disabling attacks. New studiesof these Eastern Health District data will con-tinue to smear for manv vears.

2, tie California M&bidity Research Prol-~ is being conducted by the State Health Depart-ment with the assistance of the Bureau of the Cen-sus and under a grant from the PubIic HealfiService. The objectives of the project are thesrudy ofi (a) methods of obtaining data on illnessand disability from population surveys based onobjective sampling methods, and (b) methods ofusing data arising from administrative programs,in conjunction with population sample surveys, forspecjal purpose studies or statistics. In San Jose,California, an intensive pilot study has been car-ried out in which the interview method and thediary merhod of collecting illness data have beencompared, a number of aspects of diary-keepinghave been investigated, and the effects of the in-terviewer and the respondent upon the data havebeen evaluated. Making use of the findings of theSan Jose study a statewide survey is soon to beundertaken, both as a demonstration of the meth-ods and also for the purpose of making furthermethodological tesm.

3. ‘l%e Special Research Project of theHealth Insurance Plan of Greater New York willprovide as a by-product some useful comparisonsof illness data horn an interview and from therecords of a broad-coverage medical care plan.This project is financed by grants &om theCommonwealth Fund and horn rhe RockefellerFoundation. It includes: (a) a household interviewwith each of a sample of 5,000 families insuredunder the plan; (b) an identical household inter-view with each of a sample of 5,000 househohis

representing a cross-section of the city of NewYork; and (c) an analysis of accumulated experi-ence of individuals and families with utilizationof medical services as revealed by the recordsof the plan. Diagnostic information obtained in theinterview can be compared with records of theplan, and a similar comparison can be made forrates of utilization of medical services, thus pro-viding a basis for calibration of ~he results ob-tained in household surveys. Variation among in-terviewers can be studied, and aLso the effect ofmemory failure of the respondent can be meas-ured for illnesses attended by a physician. Fur-thermore, since the cost of the service has beenremoved as a determining factor, the utilizationrates for persons covered by the plan will be ofgreat value in helping to determine what medicalcare is required for persons with specific condi-tions.

4. Surveys sponsored by the Commission onChronic Illness will make use of all methods ofmeasuring chronic illness in the population. InHunterdon County, New Jersey, a symptom ques-tionnaire was provided for each adult to be filledout by himself, a sample of families has been in-terviewed, and subsamples of those with and with-out manifest chronic disease will be screened bymeans of diagnostic tests. Thorough ph~sical ex-aminations will also be given and histories taken.These should supply some particularly useful ex-perience on the measurement of needs for medi-cal care. A similar study to be conducted inBaltimore, Maryland, is now getting under way.

5. ml -Studies, being carried on by the Graduate Schoolof Public Healrh of the University of Pittsburgh,till concentrate upon the familial aspecm of ill-ness but will also investigate methodologicalproblems in household illness surveys. The fieldwork for the first of these surveys, involving asample of several thousand households from asection of the Arsenal Health Dismict and a one-percent sample of the remainder of the city ofPittsburgh, has now been completed and analysisof some of the results have begun to appear.

6. Other studies that conmibute to the meth-odology of morbidity surveys are: (a) the seriesof small experiments being conducted by the Di-vision of Public Health Methods on memory forillness, respondent bias, interviewer variation,accuracy of household report of diagnosis, andexistence or nonexistence of records carrying adiagnosis in the offices of general practitioners;(b) the very intensive study of home injuries car-ried out by the School of Public Health of the Uni-versity of Michigan in Washtenaw County; (c) the

27

interesting mial of collection of national data onthe incidence of attended illness horn a country-wide sample of physicians, which is being under-taken by the Research Depsrunent of ModernMedicine Publications, Inc.; and (d) the surveysof unmet medical care needs in rural areas ofMichigan, Ohio, and other areas, making use ofthe symptom schedule developed by Hoffer,Schuler, and others.

When the results of the above-mentionedstudies and projects become available we shouldknow a great deal more than we do now almut tierelative value of” various procedures for themeasurement of illness.

Morbidity Statistics Applicable to

Large Segments of the Population

Before considering the types of additionalda-ta that should be collected to meet the needs fornational statistics described in Pam III of this re-port, ir is essential to review the present sourcesand to determine the extent to which these sourcesare capable of filling the needs. The NationalHealth Survey of 1935-1936 has already been men-tioned. 1[ has a number of inadequacies for cur-rent use. In tie first place, tie survey was basedalmost entirely on urban populations. Hence, na-tional estimates may be quite far in error be-cause of the failure to take into account differ-ences between urban and rural morbidity. In rhesecond place, many important changes have takenplace in the dismibution and characterisucs of theppulation since 1936, and there have been altera-tions in tie parterns of provision of medical careand in medmds of prevention and ~eanmentofill-ness. These alone would be sufficient to make theresults of the National Heakh Survey out of date,buc in addition a world war has intervened leavingin its wake not only a large number of service-connecmd disabilities but also other less obviousscars which may not even be detected for a num-ber of years. Finally, current needs for statisticsare different and data horn the National HealthSurvey are not available in the proper form toanswer today’s questions. To suggest rhaz tiecountry continue to base important decisions onstatistics that are over 15 years old would be apoor recommendation, even if the statistics werewell suited in other respects to tie problems athand.

One other source of national statistics hasalso been mentioned briefly-the Modern Medi-cine Physicians Panel on Medical Treatment(see“Other Studies” above). This must be considered

for the time being a completely unproven mech- “’anism which may turn out to be capable of provid-ing national estimates (andpossiblysomeregiona.lestiates as well) of the incidenceofattendedill-ness by diagnosis in some detail. It may alsosupply useful information on the types of attendingphysician, and the types of drugs and other treat-ment prescribed. Whether it will show rates for 1

population characteristics other than age and sex ~is not yet lmown. The statistics are based upon a !report of the practice of a national sample of phy- ~sicians for one full week in each three-month pe-riod.

Despite their vintage, mention should be made .:of the studies conducted by the Committee on the ,Costs of MedicaI Care in 1928-1931. The detailed ~information on morbidity and utilization of medi- :cal services horn the survey of 9,000 families, ;and the estimates of medical care needs made by :Lee and Jones were not duplicated in the later and tlarger National Health Survey. Consequently, thestudies of the Committee on the Costs of LMedical FCare conti.ue to provide the only source of cer- !

1

tain very useful types of statistics. The 23 papers ‘by S. D. Collins on morbidity and medical care .utilization rates born the househoId survey havebeen widely used. The Lee-Jones estimates arestill consulted although their inapplicability topresent conditions is obvious.

Other significant sources or potential sourcesof morbidity statistics may be classified as fol-lows :

A. The notifiable disease reporting system(including industrial diseases, cancer, rheumaticfever, and other diseases in some Scares).

B, Data accumulated as a by-product of in-surance and prepaid medical care plans.

1.

2.

3.

Sick benefit associations, group health andaccident insurance.Prepayment medical care plans (Health mInsurance Plan of Greater New York, Per-manence Health Plan, indusmial plans,union plans, and others).State disability insurance plans (Rhode

4.

5.

6.

Island, California, New Jersey, New York,and Washington).Life insurance companies (results ofphys-ical examinations).Hospitalization insurance plans (Blue

Cross, etc.).Railroad Retiement Board plan.

L. Tax-financed public assistance andmedi-cal care plans.

1. Programs providing all or a part of themedical services required by recipientsoh old-age assistance, aid to dependent

2.

3.

D.dischzbidity

E12F12G

tive SH

ing prI.

CensuJ,K

of bir iporrs ,

Teludethat a.poses.of thel

. Sourcttheir

Tthe ncknowheresysterconstiexperwhichfor thI

Tfirst kState ~a comseIeccvirtuelarge ~still c:mates

Inage isof tie

28

.

IL—

.’

I mech-provid.i egionalnded ilI-ay alsottendingr treat-ates forand sex1 upon aof phy -mth pe-

‘~emades on the~etailedf medi-‘milies,nade bywer and?tly,” the~iedicalof cer -papers

‘al careq have(es are,ility to

sourcesas fol-

systemwm atic

‘t of in-

llrh and

{Health-k, Per-

plans,. .

(RhodevYork,

f phys-

(Blue

cimedi-

t of thecipients:pendem

.L.

3.

D,

.1

children, aid to the blind, aid IO the per-manently and totally disabled, and generalpublic assistance (whetier by cavitation orfee-for-service).Armed Services sickness and medical carestatistics.Admission or discharge or other hospitalstatistics for Veterans Adminismation,Public Health Service, and Indian Servicehospitals.

Other hospital and clinic admission anddischarge records ~New York City Hospital Mor-bidity Reporting Project).

E. Absenteeism records1. In indusmy2. In schoolsF. Routine physical examinations1. In industry2. In schcdsG. Records of physical examinations of Selec-

tive Service registrants.H. Case-finding programs, multitest screen-

ing programs, case registers.I. National surveys using tie schedule of the

Census Bureau’s Current Population Survey.J. Local surveys.K. Miscellaneous sources (data on the back

of birth certificates, motor-vehicle accident re-ports, industrial injury reports).

This list may not be complete, but it does in-clude the major sources now producing recordshat are or might be usable for statistical pur-poses. It would be impossible to discuss each oneof them here, but certain characteristics of thesesources can be considered from the s~andpoint oftheir usefulness for serving national needs.

The nature and scope of the data supplied bythe notifiable disease reporting system are wellknown; hence, this source may be passed overhere with the comment that, though the reportingsystem is indispensable, the diseases reportedconstitute only a small fraction of the total illnessexperienced by the general population, no matterwhich of the common utits of measurement is usedfor the comparison.

Turning to the sources in Caregory B, it mustfirst be stated that, with the possible exception ofState disability insurance plans, all of these havea common limitation. This is the factor of selfselection in the coverage. In some of the plans, byvirtue of tie fact that coverage is automatic forlarge groups, tiis facror is less important, but onestill cannot use the data to make probability esti-mates for the general population.

In the State disability insurance phns, cover-age is complete for c~tain legally defined classesof the population.

Anotier more serious limitation of thesources in Category B arises horn the restric-tions in the type of illness covered by the insur-ance and the termination of benefirs under theplans after a certain number of weeks haveelapsed or a certain number of dollars have beenpaid. Only the broad-coverage prepayment plans,of which the Health Insurance Plan of GreaterNew York and Permanence Health Plan are givenas examples, do not suffer from this limitation.

For the most par~ ,data from these insuranceplans relate to artended illness. Despite the limi-tations, however, some sources in this categoryhave proved very valuable. As an example we maycite the Public Health Service statistics on indus-trial sickness absenteeism in a group of reportingorganizations made up of mutual sick benefit as-sociations, group health insurance plans, andcompany relief deparunents. The data are limitedto sickness and nonindusmial injuries causingabsence from work for 8 consecutive calendardays or longer. With the exception of the reportsof illness in the armed services, these representthe longest series of comparable illness statisticspublished in the United States, going back as theydo 101920.

The value of the records of broad-coverageprepayment plans for special studies of attendedillness and the utilization of medical care hasbeen referred to previously.

Sources in Category C are limited to specialgroups of che population which are entitled to fi-nancial assistance or free inedical care becauseof special legislation. The Army and Navy sick-ness and medical care statistics are tie moqtcomprehensive statistics of the sort available onany group in the United States. Many specialstudies of importance have originated from thesesources, bm, in general, the lack of represent-anveness of the groups included in Category C isa major limitation 10 rheir use for most of thepurposes we have listed in Part 111.

When hospital admission or discharge rec-ords are collected for all of the hospitals in alarge city (as was done in New York Ciry by theWelfare Council in 1933) so that it is possible torelate hospitalization to a population-at-risk, thestatistics can be useful in indicating differentialsin frequency of hospitalization in different seg-ments of the population. While hospitalized illnessrepresents only a small h action of all illne ss re-ceiving medical care and there is undoubtedlygreat variation from one economic or socialstratum to another in the types and severity ofcases that are hospitalized, nevertheless, hos -pitaliza~on rares by diagnosis, age, sex, andother characteristics of the population are much

29

.,,

-.

in demand for use in planning for facilities, serv-ices, and personnel.

It should be remembered, however, thachos-pital admission or discharge races reflect onlyusage of hospital facilities and must be supple-mented by some kind of independent estimates ofdemand in order to be helpful in planning.

hlemal hospital starisdcs are currently pro-viding our only national data on the magnitude ofthe problem of mental illness. Strenuous effortsare now being made by hospital adminismatorsand statisticians under the leadership of the Na-tional Institme of Mental Health to improve thestatistics. Only a grea~er number of beds andmore medical personnel will remove one basiclimitation which is fiat the rates are Iargely de-termined by the availability or lack of availabilityof hospital care.

bother use of hospitalization statistics re-lated 10 a Imown population base is for obtainingrough rates of “incidence for diseases that arerare and severe. Recent titerest in aplastic ane-mia, for example, has led to a search for hos-pitalized case data in a Imown population.

Absenteeism records and routine physicalexaminations vary greatly in quality and useful-ness. Only a few studies of school absemeeismcould be said to have conmibuted to the kinds ofneeds for statistics mentioned in this report. Thepirfalls are: inaccurate diagnostic information,lack of complete seasonal coverage, difficulty indistinguishing berween absence for illness and ab-sence for other reasons, peculiar distiibucions bydays of tie week, and the interruptions caused byweek ends and holidays. It should be possible,however, to use absenteeism on a day-to-day bas-is as an early index of epidemics, though tie nec-essary degree of cooperation has in the pastproven difficult to maintain. (The cause of the epi-demic would probably have to be determined byindependent investigation.)

School physical examinations are usually apoor source of information on the health of chil-dren. Here and there one hears of communitiesthat have improved the examination procedure andmade real use of the resuIts, but recent efforts touse tie records from the routine examinationsfor statistical purposes are rare.

Physical examinations in some indusmies andalso in other programs, such as that of the FarmSecurity Administration, have been the basis of alarge number of special studies. When the exami-nations are uniformly and thoroughly carried out,and when there is not tw great a factor of selec-tivity determining which individuals come for ex-amination, the compilation of statisucs from the.

results is very worthwhile. With a few exceptions,however, the programs of physical examinationsare resmicted to the employees of a particularplant or indusmy, and here is no basis for gen-eralization of the results. Furthermore, the rypeof information that is obtained horn the usualphysical examinauon is not particularly valuablefor the uses listed above. This is because theexamination identifies various conditions that maybe indicative of the existence of disease, condi-tions such as hypertension, anemia, overweight,refrac~ve errors, and so forth, but the resuksof followup to determine diagnoses are onlyrarely available. Nor is it usually possible to de-termine from the records what medical care hasbeen received for the conditions discovered.

The same can be said for the Selective Serv-ice examination records. The interpretation of thevoluminous data from this source has been a sub-ject of contiovers y. To what extent do they reflectpoor healzh in the age groups of the regismants ?What is the effect of changing standards for ad-mission on tie recorded information? What is theeffect of differing practice from one induction cen-ter to another? The consensus of statisticiansseems to be that onIy the most limited generali-zations are possible.

Category H is intended to cover the programswhich have case-fhding and case-following astheir objective. Some of these, such as the masschest X-ray screening, have reached millions ofpeople. In all of them the use of the results forpurposes of measurement is plagued by one greatquestion: What almut the people who chose not tobe examined? Recently there have been severalstudies aimed at determinirig the characteristicsof those who did not appear for screening. Therehad not been any instance prior to 1952 of an inte-gration for measurement purposes of multitestscreening with other measuring techniques. Be-fore long, however, we shall have the results ofthe surveys of the Commission of Chronic Illness.(See almve.) It seems likely that the combinationin one survey or series of surveys of multitestscreening with physical examinations and inter-viewing (i. e., disease detection examinations) of-fer the greates~ promise for providing morbiditydata of the breadth required to meet the nationalneeds. Such studies as are now under way are lo-cal in nature, but they will teach us a great dealabout tie feasibility of applying the same methodsin more extensive surveys.

The number of diseases for which thereexists a screening test that can be carried out ona mass scale is still small, and the efficiency ofthe screening tests now in use (their ability to

pick upout toois notdone irscreenially exstatisti

Suin maltics onices.per.sivcoverton anwhichof iterCensu(C.P.S

proxkBureaeconosuch :laborpossilby arrmum cject o-Of Coumatiorhouse’meansdata ,C.P-s.of mo

. tionalhas, 1sampltimat~at le:formis VLvaricmereup winstitshoulseparficult:speci:and, chave tregul:last dtailinf

‘resti

30

,’

Sptions,nations:ticularor gen-fie type? usualaluableuse thelat maycondi-

weight,r esulrs‘e only? to de-me has,vered.>Serv-noftheI a sub-reflectrants 7’or ad-r is them cen-ticianslerali-

gramsing as? massions of~LS for? greatnor to

everalristics‘herei inte-Ilutesrs. Be-Ults of,lness.nationltitestinter-ns) of-:bidity]tionalre lo-N deal:thods

thereout on,ncy ofity to

pick up a large proportion of existing cases with-out too many false positives and false negatives)is no~ accura~ely known. Much work is yet to bedone in this field, but the technique of massscreening, devised for case-finding, may eventu-ally extend greatly the usefulness of morbiditystatistics.

Surveys remain the most useful, flexible, andin many ways the most reliable source of statis-tics on morbidity and utilization of medical serv-ices. T%ey are also, unfortunately, the most ex -per,sive source in cost per case of illness dis-covered. Nevertheless, there is one survey basedon an accurately representative national samplewhich can provide estimates for a limited numberof items at a very reasonable cost. This is tieCensus Bureau’s Current Population Survey(C. P.S.). The monthly visiting of a sample of ap-proximately 25,000 households, carried on by theBureau, is primarily for the purpose of collectingeconomic and demographic data on the laimr force,such as unemployment, movement in and out of thelaker force, hours of work, and so forth. It ispossible to add to the schedule of this interview,by arrangement with the Census Bureau, a maXi-mum of, perhaps, six to ten questions on any sub-ject of importance to the Government, providing,of course, that the questions only require infor-mation that the household respondent, usually thehousewife, can supply. On several occasions thismeans has been employed to obtain some limiteddata on morbidity. Over a period of tirm theC.P.S. could be used to accumulate a fair volumeof morbidity statistics relating to the noninstitu-tional populauon of the United States. The sourcehas, however, the following drawbacks: (1) thesample was designed to produce only national es-Iirnates whereas morbidity statistics are neededat least on a regional basis; (2) the amount of in-formation that can be collec~ed in any one surveyis very small and the calls upon the C.P .S. byvarious agencies of the Government are so nu-merous that the monthly survey is usually bookedup well in advance; (3) the population of residentinstitutions is not covered in the surveys, but itshould be possible to sample such institutionsseparately; (4) the Census Bureau would have dif -ficulty mairdng and maintaining a staff of thespecialized coders who code causes of illnessand, consequently, the studies conducted so farhave had to be designed to avoid this coding. (If aregular program of surveys were conducted, thislast difficulty could probably be overcome by de-tailing coders to the Census Bureau.)

The outstanding advantage of the morbidityestimates supplied by the C.P.S. is that an esti-

mate of sampling error can also be provided, ow-ing to the fact that tie 25,000 households consti-tute a probability sample. Furthermore, inter-viewers are carefully mained, and the Censusfield offices provide good supervision. Resultsare available relatively promptly.

In conclusion, one opinion sometimes heardis that it should be possible to combine data fromthese many existing souces, including the nou-fiable disease reports, to make a comprehensivebody of statistics on disease incidence and prev-alence in tie United States. A little study indicatesthat this cannot be done. Not only are there seri-ous gaps (particularly in information for ruralpopulations), but also t!he difficulties inputting to-gether statistics born such a wide variety ofsources with differing definitions, comple~eness ofcoverage, sysrems of classification, and so forth,are overwhelming.

Nevertheless, if the tiagmentary sourceslisted almve are Iooked upon as supplementz-y in-formation to be used for improving the accuracyand scope of data obtatied from national surveys,they offer many pcmsibilities. Some of the poten-tialities of the use of existing records as auxiliaryto a survey are being explored in the CaliforniaMorbidity Research Project. (See above.)

We must conclude, however, that i~ is notpossible to fill the urgent needs for national sta-tistics on the incidence and prevalence of dis-eases, injuries, and imphents and on the utili-zation of medical services by relying solely uponrecords now being collected or data currentlyavailable.

How Other Countries

Have Met This problem

Three orher coun~ies are now engagd in thecollection of national morbidity data by means ofsurveys. These counmies are: Canada, Denmark,and Great Britain. In the case of the first Iwocountries the surveys have a limited durationwhile the survey of Siclmess in England andWales is a cominuing program.

The Canadian Siclmess Survey, for which thefield work is now complete, involved the visitingat intervals of a month of some 10,000 householdsscattered throughout Canada. The survey lasted afull year and was expectedto costinthe neighbor-hood of $500,000. The following quo~ations froman article in the bulleti of the Canadian Depart-ment of Health and Welfare (December 1950) de-scribe the type of data to be obtained and the gen-eral objectives of the survey:

ii

t

:t7

r,.

4(

,1

+

31

L.,4

“Some of the important information that is ex-pected to emerge horn the survey will includetie amount and dismibution of acute illnessfor the survey year in various sections of ourpopulation-how much tie is lost and howmuch it costs. Much information on the extentand nature of chronic illness such as arthri -tis, diabetes, and cancer will also be de-termined along with the amount of disabilityand cost involved in treating and caring forsuch conditions.“As well as learning much almur variouscommunicable diseases (measles, influenza,etc.) and noncommunicable disease such asarthritts, cancer, and high blood pressure,the groundwork will be laid for more detailedresearch projects imo various aspects of thediseases . . . .“Public health amhorities have realized formany yearn tie shortcomings “of our currentImowledge of acute and chronic sicImess anddisability, so that a need for such a surveyhas existed for some time. Some statistics onillness have been available for years in tieform of communicable disease reports, in-dusmial sickness reports, and hospital rec-ords, bur these figures have often been farfrom complete. ‘They have not, in any case,represented an accurate total of Canadianillness. Health authorities feel that a moreaccurate Imowledge of the health and siclaessof Canadians is indispensable for mappingpublic health plans Imth for present and forfuture needs. The present survey is designedto produce this knowledge. It is also realizedthat the present project is not a survey to endall surveys and that other applications ofsimilar methods may be used in the future toprovide health indices in provinces and com-munities. ”The objectives of the 3-year survey being

conducted by the National Health Service of Den-mark are very simikr to those stated for the

-, Canadian survey. The Danish project began inJune 1951 and it, too, involves monthly visits to asample of households. A manslation of a s~ate-ment by the National Health Service gives the fol-lowing types of data that will be produced for thepopulation of the coun~y:

1:

2.

3.

32

Diseases not trea~ed by a doctor with thepatient’s own unverified diagnoses.Diseases treated by a doctor with diag-noses verified by the doctors themselves.Hospitalized diseases reported not onlyaccording to the usual exiracts from hos-pital records, but specified in regard to

sex, age, duration, personal and socialconditions, etc.

4. The relation between patient’s diagnosesand doctorts diagnoses in cases t~eatedand verified by doctors.

IIIring the course of tie survey, informationon the illnesses of approximately 100,000 differ-ent persons over 15 years of age will be obtained.In addition to this survey there will be a specialstudy of hospitalized cases in Denmark for oneyear beginning in 1952.

In Great Britain periodic surveys of illnessamong persons over 16 years of age are beingmade by a government social survey organizationon behalf of r~e Minismy of Health. They wereinitiated in October 1943 and with a few inter-ruptions have provided statistics for every monthsince that tie. Although the Sickness Survey wasoriginally set up to meet the emergency needs ofa counay in the depths of total war, it was foundto be so useful that it was not abandoned at theend of the war. So far as can be ascertained it isa permanent fixture.

The questions on the scheduJe relate ro allillness and injuries experienced during a two-month period whether a doctor was consulted ornot, the amount of time the person was confinedto bed and unable to go to work, the number ofvisits of or visits to the doctor or dentist, and theIengrh of stay in the hospital. The cause of illnessis coded in some detail.

In reply to a Ietter from a member of thisSubcommittee regarding possible gaps in the datanow being collected in the survey of Sichess, Dr.Percy Stocks, Medical Statistician, in the GeneralRegister Office of England and Wales, referredro his paper, “Measurement of the Public Health”(British Medical Bulletin, 1951, Vol. 7, No. 4, pp.312-316) and stated as follows:

“The major gaps, and how they can be filled,are mentioned in the paper. ‘The diagnosticdetail is of course inadequate for many pur-poses such as srudies of neoplasms, but forlarge fields of siclmess which cannoc be as-sessed berter at present in any other way itis adequate. Geographic detail can be obtainedby aggregating records of one or moreyears; it would be useful of course if thatcould be done more quickly for short periods,but that would involve a sample size whichwould be too costly; and in England and Walesit may be possible to meet that need partlyby National Insurance certicauons of unfit-ness for work. I do not know of any immswhich could be profitably added to the sched-ule for regular use, but from time to time

i

Spepe~arithe‘’77mepmmebutSicd-l:St:

der titobtaineuonalrandon -sickne -indust~analys ~claim:diagnc

A.existseach Lsegmeeases.

“ sourcemater:of disftional

Funct

vsurvegiona.estiminter’.the prand ir.the r.type oobtain,(with :tution.

T[betwee

1,1

5 social

,agno6estreated

rrn ationI differ-btained.specialfor one

illnesse betiglizationy Were‘ inter-y monthvey waseeds ofs foundd attheed it is

3 to alla two-

dted orxulnednber ofand theillness

of thishedata55, Dr.eneral‘ferredieakh”

4,pp.

filled,.nosticy pur->qt forbe as-way irtainedmore

.f rhatriods,whichWalesPartlyunfit-temsched.I time

special questions are appended over a limiredperiod for research purposes as the occasionarises (e. g., concerning deafnessto assessthe need for provision of hearing aids).“The only n~ed for earlier diagnosis andmedical attention which I think to be im -portam with the present arrangemems formedical care relares to malignant neoplasms,but I am not of the opinion that a Survey ofSickness should be used to attempt~o improvethat posiuon.”Statistics on sickness and injury claims Wn-

der tie National Insurance Acts are also beingobtained in Great Britain. “The Minismy of Na-tional Insurance has arranged for a 10-percentrandom sample to be taken of all claims to asictiess benefit, a 20-percent random sample ofindustrial injury benefit claims, and a 100-percentanalysis to be made of tie industrial diseaseclaims. In the sample for sickness benefit thediagnosis is associated with age, sex, marital

PART

status, duration of incapacity, industry, occupa-tion, locality, and other relevant fac~ors; for in-jury benefit cases the circumstances in which theaccident or disease arose is also associated.”(“Note on Morbidity Statistics in England andWales ,“ World Health Organization MorbidityConference, WHO/HS/Morb. Conf./3, 3 October1951.)

To provide a population-at-risk for thesestatistics from claims there will also be a 3-per-cent sampling of the records on all persons reg-istered under the National Insurance Acrs.

Other developments in Great Britain include:(1) a pilot study for what may become a nationalsample of discharge or death records for hos -piral inpatients; and (2) a pilot study to explorethe possibility of collecting information aboutparients and illnesses meated by general prac-titioners, based on records maintained by thesepractitioners.

VI

Recommendations:

Types of Data to be Collected in Order to Meet U. S. Needs

As the discussion in Part V indicates, thereexists a multiplicity of possible sources of data,each dealing more or less adequately with somesegment of tie population and with some dis-eases. For tie reasons stated, however, thesesources cannot be brought rogether TOprovide thematerial for an objective picture of the amountof disease, disability, or medical needs at thena-tional, regional, or local level.

Function of National Morbidity Surveys

We recommend tha~ a continutig nationalsurvey be conducted, adequate to provide re-gional esttiates at intervals of two years andestimates for the nation as a whole at quarterlyintervaIs. Its purpose would be to obtain data onthe prevalence and incidence of disease, injuries,and impairments, on the nature and duration ofthe resulting disability, and on the amount andtype of medical care received. The data would beobtained from a probability sample of households(with special provision for the sampling of insti-tutions including hotels).

The suggested maximum interval of 2 yearsbetween publications of results from tie continu-

ing national survey seems adequate to us at tiepresent tie to serve the major needs outined inPart III. However, we recommend that the fieldwork be organized in such a way that more fre-quent summarizations may be made for the nationas a whole if the need should arise. We fur tier -recommend the use of a relatively small corps ofinterviewers for the field work with interviewingconducted during a part of each month, ratherthan a larger corps of interviewers working atless frequent intervals. This will have rhe ad-vantage of permitting more intensive training ofinterviewers at no exma cost. It also should makepossible closer supervision of the field work.Furthermore, the special studies will require theestablishment of a permanem government unit foreffective planning and administration. Instead ofexpanding and conmacdng this staff before andafter large intermittent national surveys, it wouldbe better ro have continuous field work under thecharge of a small staff of fixed size and highcompetency.

Current interests center in chronic diseasesrather than in acute diseases with a marked sea-sonal patiern. Nevertheless, seasonal changesmake it desirable to have survey results which

33

● L ,,

apply to a whole calendar year, or calendaryears. Field work conducted during a partof eachmonth can be planned in such a way that resulrsmay be summarized ac the end of a year or twoyears. With slight modifications quarterly changesmay also be watched if necessary. Experience in-dicates that two visits to each household will im-prove the completeness of repordlg of chronicdiseases present. ‘The panel of households can bechanged several times during a year in order togive a better sample coverage for annual orbiennial averages.

In a survey of national scope, it will presum-ably be necessary to employ part-time lay inter-viewers for ~he most part. For diagnosed ill-nesses, experience in previous morbidity surveysindicates that lay interviewers, if competentlytrained, can obrain data which permit a classifica-tion into meaningful diagnostic categories in theInternational Statistical Classification of Dis-eases, Injuries, and Causes of Death (6th reti-sion). Diagnoses reported by the household re-spondent will be checked with the medical prac-titioner if tests now in progress show that thismethod is practical and useful.

The importance of providing data that areusable at the local level has been noted earlier inthis report. From the efisting knowledge aboutdifferentials in morbidity rates and in availabilityof medical care in different parts of the counmy,it is our opinion thar satisfactory data for use atlocal levels will be obtained from a survey largeenough so that dara can be published separatelyfor each of some 50 regions into which the coun-my can be divided. Several methods of makingthis subdivision have been examined by the Sub-committee. It is sugges~ed that 10 of these regionsbe the 10 largest metropolitan areas. For there-maining 40 regions, one of the foIlowing alterna-tives is suggested:

A geographic subdivision imo 10 areas, andwithin each area, a s~atification by citysize into 4 slrata.A geographic subdivision into 13 areas, andwithin each area, a stratification by cicysize into 3 strata.A geographic subdivision into 20 areas,and within each area, a smatification into2 strata-urban and rural.

The best choice depends on the relative mag-nitude of geographic differentials in morbidity andmedical care, as compared with differentials thatare associated with size of city. By the use of oneof these pIans, or some combination of them, webelieve that each region will be sufflcientlyhomo-

geneous so that a local area can safely utilize forits purposes the survey data for the region inwhich the area lies.

Function of Special Studies

Ahhough the general surveys will furnishbasic data not at present available, there will re-main numerous important gaps. Not only will diag-nostic detail be insufficient for many purposes butno information will be obtained from these gen-eral surveys about nonrnanifest illnesses. Fur-thermore, accurate data will be obtained only fordiseases of relatively high prevalence and inci-denc e.

Undiagnosed disease will appear in ~he sched-ules either as symptoms, or as diagnoses not ac-companied by evidence that they were made by aphysician. It would be desirable to follow throughon cases of this rype by a diagnostic appraisal.Probably, only a subsample of such cases wouldneed to be studied.

The identification of nonrnanifest diseases,i.e., diseases which can be identified by diagnosucmeasures but which have not yet caused symptomsrecognized by the patient as such, is an importantpart of modern preventive medicine. This can bedone either by the complete examination method,by prior application of screening tests, followedby more definitive diagnostic examinations inthose with positive tests, or by a combination ofthese methods, with a short physical examinationserving, in effect, as one of the screening tests.There is already sufficient experience at hand(notably in the five pilot srudies which were spon-sored by health departments in cooperadon withthe Public Health Service 2) to indicate that suchmethods of finding nonmanifest disease are prac-ticable and fruirful. However, up to the present,it has not been possible to apply these methods toa representative sample of the adult population.(School health examinations, in areas where theyare adequately performed, represem, in part,case-finding of nonmanifest disease in the schoolpopulation, but not of the toral population of schoolage.) The services requtied to provide &eannent

2m= ~m~== “=r= ~onducc=d in: Richmond, Vtiginia; AdMu,

Ga.; 22 coundcs in AIabsmn; Bosmn, !@ss.; and Indismpolia, Ind.

This is by m mcsns a complete list of the screening scudics hathave conuibuced m the dcwlopmcnr of the technique.

k

.[

for the (the pre’.eral pO~diseaseshould ::

Conof specigeneralspecial

1.manif e:ing, de”sample

2.ing, oreffecn’progrr

3.paymeaccur:sourct

4.srudie-be usnation

5.latingrelatilvision

Maio

Accu

contibidi~and :spec;(3 l-rholdsoverumatrelat-perccthantiCS i

ationarea.-

34

i;

1

.

ze forion in

..

mnishdlre-diag-es but2 gen-

Fur-ly forinci-

zhed-)t ac-?byaroughtisal.would

ases,10SUCXornsmtantan bexhod,owedsin.on oflation:ests.hand

+Km-withsuch

mac-?sent,ds toition.

tieypart,:hool;hool:ment

,dmm,is, Id.

Cs dmc

.*

for the conditions found cannot be estimated untilrhe prevalence of nonmanifest disease in the gen-eral population is known. If a study of nonmanifestdisease is made, a study of undiagnosed diseaseshould also be included.

Consequently, it is recommended that a seriesof special smdies be conducted in addition to thegeneral surveys. The principal functions of thesespecial studies will be as follows:

1. To obtain data on undiagnosed and non-manifest disease, by means of laboratory screen-ing, detection and physical examinations of sub-samples drawn from the general surveys.

2. To obtain necessary data, where it is lack-ing, on the natural course of a disease or on theeffectiveness of medical care or rehabilitationprograms.

3. To collect statistics on the methods ofpayment and costs of medical care and to test theaccuracy of these by making use of ou=idesources.

4. To assemble any data from the existingstudies and sources described in Parr V as maybe useful for supplementing rhe findings of thenational survey.

5. To attack methodological problems re-lating LO tie measurement of morbidity and therelationship of such measurements to the pro-vision of medical care.

6. To evaluate existing medical care pro-grams and local studies such as the various mul-titest programs, the Framingham Project, etc., incases where such evaluations will contribute toknowledge of techniques for obtaining desired data.

Requests for Date That

are not Feasible to Meei

Needless to say, a combined program of gen-eral and special smdies will fail to meet numer-ous requests for dara by particular bodies. In gen-eral, the methods proposed are not suitable wheregreat diagnostic detail is demanded orwhere dis-eases or conditions that are rare are LO bestudied.

Investigations to determine the extent of men-tal illness, alcoholism, narcotic addiction, uge ofhabit-forming drugs, and special disease entitiesof rare frequency could not feasibly be includedin the general morbidity survey or any of the spe-cial studies that would be a part of the generalprogram. Accordingly, it is recommended thatsuch investigations be undertaken independently bythose groups and agencies which are prim arilyconcerned with each specific problem.

PART Vll

Recommendations on Survey Design

1

E

Maior Specifications of graphic region should be less than ~ percent for..

Accurncv and Detail characteristics that occur in more than Ypercent i

The national survey will be conducted on acontinuous basis providing quarter-yearly mor-bidity statistics for each of 11 metropolitan areasand 3 population-size classes in 13 regions. Thespecifications would provide for rwo interviews(3 months apart) with each of tie sample house-holds and, when the sample results are cumularedover a two-year period, they should permit of es-timates (for each of the 50 areas) such that therelative sampling error would be less than 50percem for characteristics that occur in morethan 1 percent of the population. For characters -

~tics of greater frequency, the coefficients of vari-ation of estiates in each of the memopolitan

I

areas and population-size classes within each geo-

of the national population as shown in the followingtable:

x Y— —

50 115 1012 159 255 503 752 90

Separate estimates are needed for these 11memopolitan areas: (1) New York, (2) Chicago,

35

,lize for:gion in

furnishwill re-,lldiag->ses butse gen-5. Fur-mly forId illC~-

sched-not ac-de by afmough]raisal.; would

seases,gnosticnptomsportantcan be

nethod,>llowedms inwion ofination: tests.~t hand= spon-n withat suche prac-resent,lods 10Ilalion.re they1 part,schoolschoolatmem

Atlam,

mlis, Ind.

dies Ltlar

for rhe condi~ions found cannot be es~imated untilthe prevalence of nonmanifest disease in the gen-eral population is known. If a srudy of nonmanifesrdisease is made, a study of undiagnosed diseaseshould also be included.

Consequently, it is recommended that a seriesof special srudies be conducted in addition to thegeneral surveys. The principal functions of thesespecial studies wilI be as follows:

1. To obtain data on undiagnosed and non-manifest disease, by means of laboratory screen-ing, detection and physical examinations of sub-samples drawn from the general surveys.

2. To obtain necessary data, where iris lack-ing, on the natural course of a disease or on theeffectiveness of medical care or rehabilitationprograms.

3. To collect statistics on the methods ofpayment and costs of medical care and to tesr tieaccuracy of these by making use of outsidesources.

4. To assemble any data from the existingstudies and sources described in Part V as maybe useful for supplementing the findings of thenational survey.

5. To attack methodological problems re-lating to the measurement of morbidity and therelationship of such measurements to the pro-vision of medical care.

6. To evaluate existing medical care pro-grams and local studies such as the various mul-titest programs, the Framingham Projec~, etc., incases where such evaluations will contribute toknowledge of techniques for obtaining desired data.

Requests for Date That

are not Feasible to Meet

Needless to say, a combined program of gen-eral and special studies will fail to meet numer-ous requests for data by particular bodies. In gen-eral, the methods proposed are not suitable wheregreat diagnostic detail is demanded orwhere dis-eases or conditions that are rare are to bestudied.

Investigations to determine the extent of men-tal illness, alcoholism, narcotic addiction,.use ofhabit-forming drugs, and special disease entitiesof rare tiequency could not feasibly be includedin the general morbidity survey or any of the spe-cial studies that would be a part of r-he generalprogram. Accordingly, it is recommended thatsuch investigations be undertaken independently bythose groups and agencies which are primarilyconcerned with each specific problem.

PART Vll

Recommendations on Survey Design

Maior Specifications of

Accuracy and Detail

The national survey will be conducted on acontinuous basis providing quarter-yearly mor-bidity statistics for each of 11 memopolitan areasand 3 population-size classes in 13 regions. Thespecifications would provide for lwo interviews(3 months apart) with each of the sample house-holds and, when the sample results are cumulatedover a two-year period, they should permit of es-timates (for each of the 50 areas) such that therelative sampling error would be less than 50percem for characteristics that occur in morethan 1 percent of the population. For characteris-tics of greater frequency, the coefficients ofvari-ation of estimates in each of the memopolitanareas and population-size classes within each geo-

graphic region should be less than ~ percent forcharacteristics that occur in more than X percentof the national population as shown in the followingtable:

x Y— —

50 115 1012 159 255 503 752 90

Separate estimates are needed for these 11metropolitan areas: (1) New York, (2) Chicago,

35

. . .(3) Los Angeles, (4) Philadelphia, (5) De&oit,(6) Boston, (7) San Francisco, (8) Pittsburgh,(9) St. Louis, (10) Cleveland, and (11) Washington,D. C. Separate estimates are needed also for thefollowing geographic regions: (1) New England,(2) New York, Pennsylvania, and New Jersey,(3) Maryland, Virginia, North Carolina, and Del-aware, (4) Georgia, Florida, and %uth Carolina,(5) Alabama, Mississippi, and Louisiana. (6)Illinois, Indiana, and Ohio, (7) Wisconsin andMichigan, (8) North and SoUth Dakota, Nebraska,Kansas, Minnesota, and Iowa, (9) Kenntcky, Ten-nessee, and West Virginia, (10) Missouri and Ar-kansas, (11) Oklahoma, Texas, New Mexico, andArizona, (12) Montana, Idaho, Wyoming, Nevada,Utah, and Colorado, (13) Washington, Oregon, andCalifonia. And within each of the geographic re-gions, estimates are needed for these three pop-ulation groups: (1) rural, including places under2,500, (2) cities of from 2,500 up to 50,000, and (3)cities of 50.000 or over.

Maior Specifications

of the Survey Design

1. General. During a two-year period, 45,000dwelling units will be sleeted for the sample, andeach sample dwelling unit will be enumerated intwo consecutive quarters resulting in a total of90,000 enumerations. On tie average, three sam-ple dwelling units will be subsampled from the ex-pected 9 dwelling units within each of the 15,000sample segments. The sample segments will belocated within 500 primary sampling unim se-lemed from the approximately 2,000 such primarysampling unirs in the United States. Within thesample dwelling unit any responsible adult 18years of age or older will serve as respondent forhimself and all related persons in the household.Assuming a questionnaire similar to the one usedin the San Jose Health Survey and tie HIP Survey,a household enumeration would require almut onehour-to complete.2. Metropolitan areas. The 11 metropolitanareas are primary sampling units selected for thesample with certainty. About 9,000 separate dwell-ing units, one-fifth the sample dwelling units, willbe enumerated in metropolitan areas. The enu-merations will be approximately equally dividedamong the 11 areas so that within a two-yearperiod about 820 separate dwelling units or 1,640enumerations will be made in each area. Sincethe enumerations will be taken monthly, there willbe about 65-70 enumerations per month in eacharea. However, it is specified that during the

first 3 enumeration momhs only one half thatnumber be made, and that each month thereafterdwelling units that were enumerated 3 monthsearlier for the first time be reinterview, andan equal number of new sample dwelling units beenumerated for the first tie.3. Geographic regions, exclusive of metropolitanareas. There will be 489 primary sampling unitsin the 13 geographic regions. The 200 odd primarysampling units in the Current Population Surveywill be supplemented by the necessary number ofnew primary sampling units. To reduce tie scat-ter between enumeration units, it is planned thata given primary sampling unit will be visited onlyonce a quarter. This will be accomplished by as-signing one-third the number of primary samplingunits in each region to sets A ,B, and C, respec-tively. The enumeration of dwelling units in pri-mary sampling unirs in set A will bemade duringthe first, fourth, and seventh months and everythird month thereafter. The enumeration of dwell-ing units in primary sampling units inset B will bemade during the second, fifth, and eighth monthsand every third month thereafter. And, likewise,the dwelling units in primary sampling units inset C will be enumerated during the third, sixth,and ninth months and every third month there-after.

During a two-year period about 72,000 enu-merations will be made in the 36,000 sampledweHing units in the geographic regions. There-fore, on the average, about 3,000 enumerations

will be made in each month in 163 primary sam-pling units— or akout 18-20 enumerations per

primary sampling unit per quamer. However, it isspecified that during the initial enumeration in aprimary sampling unit only one half the regularnumber of monthly enumerations be made, andthat each quarter thereafter, the dwelling unitsenumerated for the first tie a quarter earlierbe reinterviewed, and an equal number of newsample dwelling units be enumerated for the firsttime.

Special Studies

‘The national survey would not meer all of themost pressing needs for morbidity statistics un-less it is supplemented by special studies. Themost impormnt special studies are those designedto obtain data on undiagnosed and nonmanif escdiseases by means of physical examinations of asample of the U.S. population.

The surveys of undiagnosed and norunanifescdiseases discussed in Part VI, can best be under-taken as .Parr of the national survey but on a sub-

I

1

sample. S1’cal exami’U.S. natiol-quire a rwould be I

ple casesmean eitht

(i.e., by ITtits) orthe past,selectivity’confined

c.

Thefaciliriemedicalbateswith re.numbercians, :numberbuilt; tkfor theforth. Fmarilyquired ta speci+estimat,particul

Thneededlems. 1

Neth:peheinfablim;obtTh(

timatesfrom o:conside

1.

36

lalf thatsreafter. months.’ed, andmits be

2PQ!Wilgunitsmimary

Surveynber ofle scat-ned thatted ordyi by as-ampJingrespec-in pri-

I duringd everyfdwell.will bemonthskewise,units ini, sixth,

there-

)0 enu-samplellere-rationsT sam-ns per‘er, itis;on in aregularde, andg unitsearlierof new

he first

Ilof thetics un-es. The,esigned~anifestms of a

lanifest?under-1 a mb-

sample. Such a survey would involve actual physi-cal examinations for a random subsample of theU.S. national sample. The examinations would re-quire a medical “crew” and equipment, and itwould be necessary to “bring together” the sam-ple cases and the examining facilities. This maymean either bringing the facilities to the subjects(i.e., by means of completely mobile examini~units) or bringing the subject to the facilities. Inthe past, screening tests have run into a highselectivity factor. Even in screening programsconfined to a single industrial or governmental

PART

establishment, it has not been possible to get 100

percent of the group to come in for examination.The success of a program calling for the exami-nation of a representative sample of the U .S. pop -ulation depends upon the ability to get practicallym (at least 90 percent) of the population to comein for examination. Experience with methods forachieving this goal is not yet availabIe and exten-sive experimentation is necessary with mobileexamining units, publiciry campaigns, specialtransportation for examinees, etc.

Vlll

Suggested Methods for Obtaining Data on Need for Medical Care -

The need for well-founded estimares of thefacilities and services required to furnish goodmedical care is generally recognized. Such es-timates are indispensable for formulating policywirh regard to questions such as the desirablenumber of physicians, nurses, dentists, techni-cians, and related personnel to be mained; thenumber of general and special hospitaI beds 10 bebuilt; the number and rype of other institutionsfor the care of the sick to be established, and soforth. For some purposes interesr centers pri-marily in the total services and facilities re-quired to provide medical and preventive care toa specified group of people; for others, separateestimates are wanted for people known to have aparticular disease or impairment.

The estimation of the amount of medical careneeded in a given situation presents difficult prob-lems. No single method of measurement is ideal.

Needed medical care may be described asthat care (in the broadest sense) which aperson would seek, if it were available, ifhe knew it was available, if he were well-informed as to what was considered desir-able medical care, and if there were noimportant economic or other barriers 10 hisobtaining it.The following three methods for obtaining es-

timates of need for medical care, as distinguishedfrom ordinarily obtained medical care, may beconsidered:

1. By expert appraisal of the needs either of(a) persons with one or another disease or

group of diseases,1 or of (b) a sample ofpersons, boti well and sick, from the gen-eral population. In the former procedurethe total picture of needs for medical carewould be obtained by studying diseases,group by group, and adding the estimatedneeds for these groups, incIuding separateestimates for preventive services and forcare required by persons with la~ent ornonmanifest disease, usually chronic. Datahorn morbidity surveys would also be re-quired. to provide a basis for combiningestimates in tie different groups. The-validity would depend in par~ upon therepresentativeness of the samples of per-sons with the various diseases,

In the latter procedure representativesamples from the general populationwould be srudied by experts to determineall services needed, including diagnostic.SUrnmarizauon might be in terms of cat-egories of services rather than diagnoses,

1~~ would & ~~ilti ~ tie merhod used in *G =-cdkd Lce-

]onea esrimmtes from tbe sndies of tie Committee on rb. C3sts ofMedical -e. See “The Fundamentals of Good Medical Care” by

Roger I. Lee and Lewis Webster Jones, assisted by Bubara Jones;publicnrion No. 22 of die Cnmmitree on the Costs of Medical @e.

Tbc Uniwrsiry of Chicago Press, Chicago, Illinois, 1933.

37

. .%.

7-.

38

In bmh procedures the validity woulddepend upen the consistency among expertsin their appraisal of needs, and also uponthe extent to which people with medicalneeds would actually use facilities pro-vided for those needs.By study of the services actually obtainedby groups under the most favorable con-ditions now existing. (a) Insurance coveredgroups (such as the Health Insurance Planof Greater New York). To this would haveto be added an estimate for the servicesnot covered by the insurance plan and anadjustmem for selection of the insuredgroup (empl~yed, age, etc.). (b) Physician’sfamilies. (c) Highest income group. Studyof services could be secured either fromthe recipients by the journal method or byperiodic visirs, or by studying a record ofservices from vendors of such services,e.g., physicians, hospitals, ourpatientclin-ics, etc., classified by economic status ofthe recipient. The last named method re-quires, of course, that the population serv-ed by the vendors be known and that allsources of medical service to that popula-tion be included.

3. By collection of evidence regarding claimedor apparent shortages of services and fa-cilities. Examples of probable shortagesfor which fairly good evidence could beobtained are (a) psychia~ists; (b) publichealth physicians; (c) nurses; (d) hospitalbeds (general and chronic): (e) dentists.

It should-be noted that these-&e methodsdiffer in their assumptions. Metiod 1 is dependentupon expert opinion as to what care is needed.

Method 2 assumes that the best care now obtainedby favored groups is a reasonable approximationof the care which would actually be sought andneeded by all groups, if economic and other bar-riers were minimized. Method 3 would probablyyield the most conservative estimates of unmetneeds, since it would be based for the most parton those unmet needs already manifest as a re-sulr of existing demand, rather than on demandthat would exist if economic and other barrierswere at a minimum.

A combination of a morbidity survey withmethod 1 would seem to be the most convincingand feasible method. From the standpoint ofmethodology, however, it would seem desirablethat all rhree methods be studied. It may veryweIl appear that each method would supply infor-mation not obtainable from any of the others.

I

IIaimed,nd fa-

)rtagesuld bepublicxspital,ntists.Iethodsendentleeded.xainednationThtandr bar-]bablyunmetst part

a re-xnandrriers

v withincingint ofirableT veryinfer -‘fiers.

-.

Subcommittee on National Morbidity Survey

Dr. W. Thurber FaIcs, atirman ---------- Baltimom City Health &,artmcnt

(Deceased) Baltimore, MaryInnd

Dr. Genrgc F. Bndger ------- ---------- Western Reserve university

Clewclmnd, Ohio

Mr. William G. ~chran -------------- - The Johns Hopkins Universi~Bdr.imore, Maryland

Dr. Edward Holmes -------- ---------- Cicy HcaIch DcpartmencRichmond, Virginia

Dr. Mormn I. Lcvin --------- ---------- New York State Dcparcmcnc of HealcbAlbany, New York

Dr. Eli Mm.rks-- . -------------------National Opinion Research Ccntcz

Universim of Chicago

&icago, ‘Illinois

Mr. Tbeodore Woolsey ------- --------- Division of Public Hcalch Me*ads

Public Hcafch Service

u. s.

Dr. Lowell J.

Washington, D. C.

National Committee on Vital and Health Statistics

Reed, chairman ------------ The Johns Hopkins UnivcrsiW.,

Dr. Halbert L. Dunn, Vice Chairman -------- Nar.ionaf &ffic; of Viral Statistics

Dr. I. M. Moriyama, S=crctary ----------- -

Dr. George Baehr ------------ --------

Dr. Edwin L. Crosby --------- ---------

h. Edwin F. Daily ---------- ---------

Dr. Paul M. Densen ------------- . . . . . .

Dr. Harold F. Dam ----------- --------

Dr. W’. Thurber FaIes (Deceased) ----------

Mr. Eugene L. Hamilton --------- -------

-.——.Public Healfi .%rvicc

Washington, D. C.

National Office of Vital statisticsPublic Health Service

Washington, D. C.

Hcalfn Insurance Plan of Greater New York

New York,New YorkJoint Commission on Accreditation of Hospitals

Chicago, Illinois

HeaIfi Insurance Plan of Greater New- York

New York, New York

Un iversi~ of Pittsburgh

Pittsburgh, Pennsylvania

National Institutes of HealthPublic Health Service

Washington, D. C.

Balrimore C&Y Healrh Deparzmcnr

Baltimore, M-land

Offi Ce of Surgeon General

rlcparu-nenr of the Army

Washington, D. C.

Dr. Philip M. Hauser ------------------ University of Chicago

Dr. Rohm H. Hu[chesan ------- --------

Sir. P. K. Wb@mu ---------- ---------

Chicago, illinois

Scare Deparcrnenr of Public Health

Nashville, Tennessee

Scripps Foundation for

Research in Population pmblcms

Miami Universi w

oxford, Ohio

39

)

i

\1

1’

APPENDIX II

NATIONAL HEALTH SURVEY ACT

PUBLIC LAW 652- 84TH CONGRESS

CHAPTER 510- 2D SESSION-S. 3076

AN ACT

To provide for a continuing sumcy and special studies of sickness and disability in tie United .%ares, andfor periodic qarts of the results chercof, and for ocher purposes.

Be it enacted by the Senate and House of Representatives of the UnitedStates of America in Congr ess assembled, That this Act may be cited as the“National Health Survey Act”.

Sec. 2. (a) The Congress hereby finds and declares—(1) that the latest information on the number and relevant charac-

teristics of persons in the countiy suffering from heart disease, cancer,diabetes, ardmitis and rheumatism, and other diseases, injuries, and hand-icapping conditions is now senousIy out of date; and

(2) that periodic invemories providing reasonably current informs.tion on these matters are urgently needed for purposes such as (A) apprais-al of the true state of health of our population (including bah adults andchildren), (B) adequate planning of any programs to improve their health,(C) research in the field of chronic diseases, and (D) measurement of thenumbers of persons in the working ages so disabled as to be unable to per-form gainful work.(b) It is, tierefore, the purpose of this Act to provide (1) for a continuing

survey and special studies to secure on a non-compulsory basis accurate andcurrent statistical information on the amount, dismibution, and effects of illnessand disability in tie United States and the services received for or because ofsuch conditions; and (2) for studying methods and survey techniques for securingsuch statistical information, with a view toward tieti continuing improvement.

Sec. 3. Part A of title ILI of the Public Health Service Act (42 U. S. C. ch.6A) is amended by adding after section 304 the following new section:

,

40

“NATIONAL HEALTH SURVEYS AND STUDIES

Sec. 305. (a) The Surgeon General is authorized (1) to make, by samplingor other appropriate means, surveys and special studies of the population of theUnited States to determine the extent of illness and disability and related infor-mation such as: (A) tie number, age, sex, ability to work or engage in otheractivities. and occuDaaon or activities of Dersons afflicred with chronic or otherdisease or injury o; handicapping cohditi~n; (B) the rype of disease or injury or 1handicapping condition of each person so afflicted; (C) tie length of time thateach such person has been prevented from carrying on his occupation or activ-ities; (D) the amounts and types of services received for or because of such con-ditions; and (E) the economic and other impacts of such conditions; and (2) inconnection therewith, to develop and test new or improved methods for obtainingcurrent data on ilIness and disability and related information.

“(b) The Surgeon General is authorized, at appropriate intervals, 10 makeavailable, through publications and otherwise, to any interested governmental orother public or private agencies, organizations, or groups, or to tie public, theresults of surveys or studies made pursuant to subsection (a).

“(c) For each fiscal year beginning after June 30, 1956, there are authorizedto be appropriated such sums as the Congress may determine for carrying outthe provisions of this section.

“(d) To assist in carrying out the provisions of this section the SurgeonGeneral is authorized and directed to cooperate and consult with the Depart- ?

ments of Commerce and Labor and any other imerested Federal Deparmnentsor agencies and with State health depamnents. For such puqmse he shall utilize 1

insofar as possible the services or facilities of any agency of the Federal Gov-ernment and, without regard to section 3709 of the Revised Statutes, asamended, of any appropriate State or other public agency, and may, without re-gard to section 3709 of the Revised Statutes, as amended, utilize the services orfacilities of any private agency, organization, group, or individual, in accord-ance with written agreements benveen the head of such agent y, organization, orgroup, or such individual, and the Secretary of Health, Education, and Welfare.Payment, if any, for such services or facilities shall be made in such amountsas may be provided in such agreement. ” i

Sec. 4. Section 301 of the Public Health Service Act (42 IJ. S. C. 241) isamended by smiking out the word “and” ac tie end of paragraph (f), redesignat-ing paragraph (g) as paragraph (h). and inserung immediately following para-graph (f) the following new paragraph:

“(g) Make available, to health officials, scientists, and appropriate publicI

and other nonprofit instimtions and organizations, technical advice and assist-ance on the application of statistical methods to experiments, studies, and sur-veys in health and medical fields: and”.

Approved July 3, 1956.. . I

* us. aavunu[n’r mlmlma OFFK1 : 1- 0-aM-ua41

I

. .

OUTLINE OF REPORT SERIES FOR VITAL AND HEALTH STATISTICS

Public Health service Publication No. 1000

SERIES 1-4. GEN13RAL sERIES. Pmgrem descriptions, mathodologiml rmsmrch, sod smslyiicnl studies of vital nd health sutistics.

Ewlism repds or this kind htve ~ppesred in “ViUl ~’jsfics-Special Reports” snd in “Heslth Statistics frum the Nnticmal Pleslth survey,” Sericc

A smd D, PHS Publication No. 5S4.

series1:

series 2,

series 3:

series 4:

Pmmsms mrd collection procedures. -Reper& which deimnba the geneml pmgrsms CJ tie NationBl Center for Health Ststi9tic9 mrdib o[ficcs snd divisions, dats callsction methods used, definitias, snd other mttenal necess~ for understanding or the techni-cnl cbsmcteristics of published dints.

fhf.s avslumtion md me~ods rsse~rch. -Studies of new stdistical methodolo~ inchding rnxperim~tsl tests of new suwey meth-

ads, studies of vitsl statistics collection methods, new SCIdfi-Cd tmhniques, objective evstuations of relimbili~y or collected dmta,

contributions to etatiw,ical thency.

Anslyticsl Studies. -Tbia series comprises mp presenting mnlyticsl or intacpmtive studies bmcd on vitsl and health statis~ics.

Documents snd committse m- ts.- Finsl reports of nmjor committees mncsrned with vital and henlth etsti.eticg snd documentsnuch -S remmmended mndel rid registmtion laws snd revised bhth md death ccrtificmtcs.

SERIRS 10-12. D.4T.4 FROM TFIE NATIONAL HEALTH SURVEY

Etrlier repo~ of the kind sppeming in Series 10 have been issued m “Health SUtistics frnm the Nm.ionsf Health Survey,” Series B snd C, PHS

Publication No. 5B4.

Series 10: Statistics on illness, mccidnntsl injuries, disability, use of bospitcl, medical, dental, tnd other services, md -her health-;elmted

tnpics, bsscd on datm collecvsd in the continuing Nationnl Herdth Interview Survey.

Series 11: Dscs [corn the HerJth Exsminhtion 5urvey bsssd on the direcl examinr.tion, testing, snd measurement of nstional scmples of the

population of the United Shtas, including the medically definsd prevalence of spscific disecses, snd distributions of the Ppulu-

tion with rasped ta vsrious pbysicnl snd physiological messurmrents..

Series 12: Ds~ from the Health Records Survey relsting to the hetlth chtmcteristics of persons in institution, snd cm hospiml, mf=’ic~Inursing, and personsl cam rsceived, bmsd on nntiarnl scnrples of estsblishmen~ prnviding these semices snd r$cmples or the

rssidsn~ of patients, or of mmrds of Lhe eskblishments.

SERIES 20-23. DATA FROM THF NATIONAL VITAL STATT5TICS SYSTEM

Emrlier reprts of this Kind hmve been iasusd in ‘Vitsl .Statistics-Specid Repnti. ”

Series 20:

Ssries 21:

~~:

%rics 23:

Vmious repnti on motility, tabulations by cause or death, ” ige, ccc., time series of mtes, data for g-gcsphic srems, SUms,

cities, etc. -oLher lbmr M included in mntml m rnmthly repnrr.%

D~L- on rdslity such AS bid by nga of moLhsr, birth order, geogr~phic sre~, StAtes, cities, time series of rates, e&-compila-

tions or d-te not includsd in the mgulsr mrnusl volumes or monthly reperk

Dtti on mmriage md divorce hy vsriou9 demo~phic [nc@rs, geographic nreas, etc.-other I.hmi tlmt included in mnuml or monthly

reports.Dau frnm Lhe pmgmm or ssmple suneys cclmted tn vitsl recmds. The suhjec~ being cmmmd in these surveys are vmicd includ-ing tnpics such m mortslity by cocioecmomic clmsse.s, hospitalization in the last yesr d life, X-my exposure during pre.gnmcy, .mLc.

Catalag Card

U. 1. Nanand Cmt,r @ Hedib ~1411,sf,cS

Origin, prngmm, and npemtion or the U.S. National Henlth Survey. + description orLhe developments Iemding ta enaconent or the Nsrionnl Health Survey Act, md n sum-mmy of the policies, initinl pmgrsm, md opcmtion or the Suwey. Wmshingtan, L1.S.

Depertmmrt or Hedtb, Education. md Welfare. Public HealtJ Service, 1963.

41 p. Z7cm. (IIS Viul md HcddI .%cimics. .Smim 1, no. 1)U.S. Public Hmkh %micc. Publicmion no. 1000, Smies 1, ❑ . 1 I

I L U-S Nsrionml Hemlrh 51mT - Adminimmtian. L U.S. - SC&tics, !kdicml.

1. Tide. II Orslnled by Dcpumcnc O( Hmlrh, E&cmuB., mndWlfUc Libruy.

I