5
8107 Saturday 13 January 1979 OUTCOME OF ANOREXIA NERVOSA L. K. G. HSU A. H. CRISP BRITTA HARDING Department of Psychiatry, St. George’s Hospital Medical School, London SW17 Summary 100 females with anorexia nervosa were followed up 4-8 years after first presen- tation. All but 12 had had refeeding and/or psycho- therapy. 48 had a good outcome (weight at least near normal, regular menstruation, largely satisfactory men- tal state and psychosexual and psychosocial adjust- ments) but outcome was intermediate in 30, and poor in 20 patients. 2 had died. Poor outcome could be posi- tively associated with clinical data such as longer duration of illness, older age of onset and presentation, lower weight during illness and at presentation, presence of symptoms such as bulimia, vomiting, and anxiety when eating with others, poor childhood social adjust- ment, and poor parental relationships. Introduction THE outcome of anorexia nervosa is much de- bated.’ The recovery rate of this illness has been reported as 35%2 and 88%.3 It is likely, however, that differences in methodology such as diagnostic criteria, duration of follow-up, and outcome criteria account for at least part of this discrepancy.4 Lately, claims for the efficacy of certain methods of treatment have been made3 5 6 without the influence of these factors on out- come being taken into account. We describe here the outcome of a consecutive series of female anorexia nervosa patients on follow-up at least four years after their first presentation at St. George’s Hospital, London, and we assess clinical predictors of outcome from data gathered at first presentation and subsequent treatment. Patients and Methods Patients 105 female patients were diagnosed as having anorexia ner- vosa, according to criteria already defined,’ between May, 1968, and December, 1973. All patients were under the consul- tant care of A.H.C. The clinical features on presentation are summarised in table I. Treatment Just under half of the patients (A.M.H. group, n=49) were subsequently admitted to Atkinson Morley’s Hospital for refeeding and major psychotherapy. The treatment regimen has been described in detail elsewhere.8 The rest of the patients were subdivided into three groups: one had outpatient psycho- therapy at St. George’s Hospital (S.G.H. group, n=31); a second group had subsequent psychiatric treatment in other hospitals (O.H. group, n=13); and a third group had no psy- chiatric treatment at all after the first interview (N.T. group, n=12). 1 patient who disappeared without trace after the first interview was included in this last group. Original Clinical Data A total of 57 items of clinical data were extracted from the original case records. They fall into six categories: clinical features at presentation; history of present illness; personal history; relationships with family; family history and parental relationship; and treatment and outcome 1 year after presenta- tion. These items served as the prognostic indicators in sub- sequent analysis. Follow-up 102 patients were followed-up 4-8 years (mean 5.9 years, s.D. 1.3, s.E. 0-11) after first presentation. 3 patients could not be contacted despite all efforts. 75 patients were interviewed TABLE 1--CLINICAL FEATURES AT PRESENTATION Where appropriate, values are mean±s.D. *6 patients were prepubertal

OUTCOME OF ANOREXIA NERVOSA

  • Upload
    britta

  • View
    215

  • Download
    2

Embed Size (px)

Citation preview

8107

Saturday 13 January 1979

OUTCOME OF ANOREXIA NERVOSA

L. K. G. HSU A. H. CRISPBRITTA HARDING

Department of Psychiatry, St. George’s Hospital MedicalSchool, London SW17

Summary 100 females with anorexia nervosa werefollowed up 4-8 years after first presen-

tation. All but 12 had had refeeding and/or psycho-therapy. 48 had a good outcome (weight at least nearnormal, regular menstruation, largely satisfactory men-tal state and psychosexual and psychosocial adjust-ments) but outcome was intermediate in 30, and poor in20 patients. 2 had died. Poor outcome could be posi-tively associated with clinical data such as longerduration of illness, older age of onset and presentation,lower weight during illness and at presentation, presenceof symptoms such as bulimia, vomiting, and anxietywhen eating with others, poor childhood social adjust-ment, and poor parental relationships.

Introduction

THE outcome of anorexia nervosa is much de-bated.’ The recovery rate of this illness has been

reported as 35%2 and 88%.3 It is likely, however, thatdifferences in methodology such as diagnostic criteria,duration of follow-up, and outcome criteria account forat least part of this discrepancy.4 Lately, claims for theefficacy of certain methods of treatment have beenmade3 5 6 without the influence of these factors on out-come being taken into account.We describe here the outcome of a consecutive series

of female anorexia nervosa patients on follow-up at leastfour years after their first presentation at St. George’s

Hospital, London, and we assess clinical predictors ofoutcome from data gathered at first presentation andsubsequent treatment.

Patients and MethodsPatients

105 female patients were diagnosed as having anorexia ner-vosa, according to criteria already defined,’ between May,1968, and December, 1973. All patients were under the consul-tant care of A.H.C. The clinical features on presentation aresummarised in table I.

Treatment

Just under half of the patients (A.M.H. group, n=49) weresubsequently admitted to Atkinson Morley’s Hospital for

refeeding and major psychotherapy. The treatment regimenhas been described in detail elsewhere.8 The rest of the patientswere subdivided into three groups: one had outpatient psycho-therapy at St. George’s Hospital (S.G.H. group, n=31); asecond group had subsequent psychiatric treatment in otherhospitals (O.H. group, n=13); and a third group had no psy-chiatric treatment at all after the first interview (N.T. group,n=12). 1 patient who disappeared without trace after the firstinterview was included in this last group.

Original Clinical DataA total of 57 items of clinical data were extracted from the

original case records. They fall into six categories: clinicalfeatures at presentation; history of present illness; personalhistory; relationships with family; family history and parentalrelationship; and treatment and outcome 1 year after presenta-tion. These items served as the prognostic indicators in sub-sequent analysis.

Follow-up102 patients were followed-up 4-8 years (mean 5.9 years,

s.D. 1.3, s.E. 0-11) after first presentation. 3 patients could notbe contacted despite all efforts. 75 patients were interviewed

TABLE 1--CLINICAL FEATURES AT PRESENTATION

Where appropriate, values are mean±s.D.*6 patients were prepubertal

62

jointly by G.H. and B.H. 2 patients had died. 12 were

followed-up by postal questionnaire and 13 by interviewing aclose relative. These indirect methods of follow-up were forcedon us either because the patients had gone abroad (7) orbecause the patients (10) or relatives (8) objected to an inter-view.

Assessment of OutcomeFor assessing outcome we used the two outcome measures

described by Morgan and Russell:4 an average outcome scoreand a general outcome category. The average outcome score is,based on the outcome in five areas: nutritional status, men-strual function, mental state, sexual adjustment, and socio-economic status. The ratings on the five scales were based onthe 6-month period immediately preceding the interview andthe final score for each patient represents an average of theratings of these five scales. The general outcome category wasbased on the patients’ body-weight and menstrual functionduring the 6 months preceding follow-up. Three categories ofoutcome were defined: (1) good : weight within 15% of matchedpopulation mean weight (mt.r.M.w.)9 and regular menstrualcycles; (2) intermediate: body-weight within 15% of M.P.M.W.but not constantly sustained, and/or presence of menstrual dis-turbances ; (3) poor: body-weight below 85% of M.P.M.W. andabsence or near absence of menstruation.

Statistical AnalysisEach of 57 items obtained from the patients’ original case

records was used to test its significance as a prognostic indi-cator.

One-way analysis of variance (F testlO) was used for testingthe relationship of variables measurable on a ratio scale (e.g.,age, weight) with outcome categories. For testing their rela-tionship with the average outcome scores the Mann-WhitneyU ranking testll was used when the subjects could be separatedinto two groups (e.g., single vs married, poor vs satisfactoryparental relationship) and the Kendall rank correlation co-efficient12 was used for items of which at least ordinal measure-ment was possible (e.g., social class, childhood social adjust-ment-good, intermediate, poor).

Results

The clinical features of the illness in this series will bepresented elsewhere.13 In analysis of the outcome the 3patients who could not be contacted were excluded; 2patients had had a leucotomy, and they too were

excluded. The subsequent progress of these 5 patientswill be described later.

100 patients were analysed (table n). 48 had a goodoutcome. Their overall mean weight was 97% ofM.P.M.W. and all had regular menstrual cycles. 3 hadbeen pregnant and/or had had babies in the 6 months

preceding follow-up. The majority also had a satisfac-tory mental state and psychosexual and psychosocialadjustments. However, a high proportion (11/48, 23%)admitted to being overconcerned about their weight andshape. 2 of these patients had had what appeared to bea schizophrenic breakdown with auditory hallucinationsand thought disorder after apparent recovery from anor-exia nervosa. Their sexual and social adjustment wasmuch less satisfactory than that of the rest of the group.

30 patients, including 2 who had become obese (122%and 117% average weight), had an intermediate out-come. 5 had menstrual irregularity and 9 hadamenorrhcea although all were of normal body-weight.

20 patients had a poor outcome, all retaining the cen-tral symptom of severe "weight phobia". 7

2 patients died in general hospitals of inanition (body-weight at last admission: 51% and 59% M.P.M.w.) andsevere electrolyte disturbances due to vomiting. 1 pa-tient had had repeated psychiatric admissions and died5 years after presentation. The second patient refusedpsychiatric treatment and died 3 years after presenta-tion.

2 patients had leucotomy. Both had had severaladmissions to A.M.H. before operation. 1 had becomeobese (130% average weight) after leucotomy but herweight had become normal during the 6 months beforefollow-up although menstruation remained irregular.The second patient made a remarkable recovery afterthe operation and for the next 4 years led a normal lifeand finished a nursing course. However, she lost weightsoon after working as a staff nurse and became amenorr-hoeic again.

3 patients could not be contacted for follow-up. 1

disappeared without trace soon after her first interviewand not even her mother knew her whereabouts. Thesecond patient was admitted to another psychiatric hos-

I

pital and subsequently improved. When last heard of 4years after presentation, she was separated from herhusband and making plans to move abroad. The thirdpatient would not answer our repeated requests for aninterview and did not return the postal questionnaire,Her general practitioner informed us that she had

apparently recovered from her anorexia nervosa, wasmenstruating regularly, and was working as a yogateacher.

FURTHER DETAILS OF OUTCOME

Nutritional Status

Body-weight had become normal (within ±15%

TABLE II-DISTRIBUTION OF THE 100 PATIENTS AMONG THE CATEGORIES OF OUTCOME

’excluding 2 obese subjects; t weight at last admission shortly before deàth; t excluding the two obese patients and the two patients who died;§ present at least half the time.

63

M.P.M.w.) in 64 patients and in 13 had risen to above85% of M.P.M.w. in the last six months although notmaintained. Only 2 patients had been overweight (122%and 117% of M.P.M.w.) in the 6 months before fol-

low-up. 14 had a body-weight persistently below 75%and 6 had a body-weight 75-85% of M.P.M.w. Food in-take was normal in only 37 patients. Bulimia (morbidvoracious appetite for food) was present in 20 patients,vomiting in 22, and excessive use of purgatives in 36; 33had persistent anxiety on eating with others.

Menstruation54 patients (48 with normal weight) had had normal

menstruation during the 6 months before follow-up, in-cluding the 6 patients who had been pregnant and/orhad had babies within the 12 months before follow-up.28 patients had amenorrhoea although the body-weightwas normal in 11. In 16 menstruation was sporadic andof these 8 had a normal body-weight.Mental State

Only 28 patients claimed that they were no longerover-concerned about their weight. 44 worried con-tinuously or almost continuously about their weight andfatness even though 26 of them had returned to normalor near normal weight. 47 patients were free from otherpsychiatric disturbance. 27 had only a mild disturbance,while 24 had symptoms severe and persistent enough tointerfere with their work and relationships. The com-mon symptoms were depression (38 patients), and obses-sion (22 patients). 1 patient had severe mood swingswhich responded well to lithium but she remained inten-sely weight phobic. 1 patient who had recovered fromanorexia nervosa showed obvious thought disorder dur-ing the follow-up interview but was not being treated.Another 2 patients also had what appeared to be schizo-phrenic symptoms, after apparent recovery from anor-exia nervosa in one and with concurrent symptoms ofanorexia nervosa in the other. Both responded to treat-ment with phenothiazines. Neither showed any psy-chotic features during follow-up and both were in full-time employment.

Psychosexual Outcome29 patients were married at follow-up and 4 were

engaged to be married. 15 of the 20 patients who weremarried after apparent recovery from their illness had a

satisfactory marital relationship. In contrast, of the 14(excluding the untraceable patient) who were married atfirst presentation, 2 had died, 2 were divorced and 1divorced and remarried; and sexual activity within themarriage was minimal or non-existent in 5. 14 patientshad had babies since first presentation, and 1 was 30weeks pregnant at follow-up. However, in 1 patient ill-ness relapsed 6 months after childbirth. 2 patientsdemanded that their husbands-to-be have vasectomiesbefore marriage.On a compound scale4 (attitude towards sexual mat-

ters, professed aim in sexual relationships, level of activesexual behaviour, and attitude to menstruation), 17 pa-tients were clearly abnormal (e.g., aversion to heterosex-ual contact, decreased libido, and satisfaction at con-tinuing amenorrhoea or disgust at return ofmenstruation). 12 of these patients had very low body-weight, but poor adjustment was also found in 2 patientswho had an otherwise good outcome in terms of weightand menstruation.

Psychosocial OutcomeThis was’ assessed on four subscales:4 relationship

with family; establishment of independence from family;social contact outside the home; and occupationalrecord.

,

, 39 patients had difficulties in family relationships, 8of whom had good outcome in terms.of weight and men-struation. An excessive emotional dependence coupledwith resentment and hostility was the commonest find-ing, and it was our impression that the family often te-ciprocated and reinforced this dependence. 21 patients(7 with good outcome) expressed fears that they wouldnot be able to move away from their parents. 25 patients(including 8 with good outcome) had excessive anxietyon meeting people outside the home, leading often toavoidance of such occasions. The outcome according tooccupational record is the most encouraging: 82 patients(including 6 full-titne housewives with babies) were infull-time employment, 12 of them with a low body-weight. Only 8 patients were unemployed for the entire6 months preceding follow-up.

PREDICTION OF LONG-TERM OUTCOME FROMCLINICAL DATA

The 57 items extracted from the original clinicalrecords were used in the search for prognostic indi-cators. The factors which were positively associated witha poor outcome (P<0.05) are summarised in table in.

Childhood social adjustment was considered poor ifthe patient had few or no friends. The disturbed rela-tionship with parents before onset of illness was charac-terised by excessive dependence on the mother in 20cases, and hostility towards an often remote, inaccessible

TABLE III-FACTORS FROM ORIGINAL CLINICAL RECORDSASSOCIATED WITH POOR OUTCOME AT FOLLOW-UP

.Social class assessed from father’s occupation. 14t7’==0-209; P<0-02 if frequent vomiters (more than once per day)were compared with occasional vomiters.

-

64

TABLE IV-DISTRIBUTION OF TREATMENT GROUPS AMONG CATEGORIES OF OUTCOME

A.M.H.: admission to Atkinson Morley’s Hospital; S.G.H.: outpatient treatment at St. George’s Hospital; O.H.: treatment at other hospitals,N.T.: no treatment. -

* Including 1 who had a leucotomy.tAll inpatients and includes 3 male cases as well as 2 females who had subsequently had a leucotomy.

father in 8 cases, while in another 2 cases the relation-

ship with father was judged to be an intensely close one.Relationships between parents were considered poor

if, before most of the illness, there had been overt dis-cord and/or marked apathy.

Discussion

The outcome in the present series of patients com-pares favourably with those in other studies of similardesign.4 15 Nearly two-thirds of the patients are at nor-mal weight, and menstruation has become cyclical in

.

over half of the patients. However, a high proportion ofpatients (49%) still retain the central symptom of

"weight phobia"; pattern of food intake remains vari-able and sometimes, of course, highly restricted, in

nearly two-thirds of the patients. However, a daily dietwhich varies in quality and quantity also characterisesnormal teenage females.’6 The psychosexual and psy-chosocial outcome are satisfactory for the series as a

whole, although a disturbed relationship with familywas found in 40% of patients at follow-up.The methodology of our study was very similar to

that of Morgan and Russell’s study&mdash;i.e., all the pa-tients were of a consecutive series and were under thecare of one consultant psychiatrist, diagnostic criteriawere very carefully defined, the same outcome measureswere used, follow-up was of long duration (theirs fromtime of discharge, ours from time of presentation), anda large proportion of patients were interviewed at fol-low-up. Morgan and Russell considered that patientselection factors were crucial in the outcome of anorexianervosa. The outcome in our series is marginally betterthan that in theirs (table iv), but their patients weremales as well as females admitted to a psychiatric hospi-tal, and on the whole they had a lower illness weight(63% of average) than ours (68% of M.P.M.w.). How-ever, duration of illness was similar and in both series

nearly half the patients had had previous psychiatrictreatment in other units.

31 of our patients had had only outpatient treatmentafter first presentation (table iv). The S.G.H. group andthe A.M.H. group differed in body-weight at presenta-tion (F=ll-79; p<0-001). Because lower body-weight atpresentation (as % M.P.M.w.) was correlated with pooreroutcome, it was not surprising that the outcome cate-gories of the S.G.H. group were marginally better thanthose of the A.M.H. group.Our series confirmed the view of Crisp et a1. 17 that the

more intractable the illness (as indicated by longer

duration of illness, lower body-weight during illness andat presentation, failed previous psychiatric treatment)the poorer the outcome. Furthermore, we confirmedTheander’s finding that vomiters did poorly,15 althoughMorgan and Russell4 did not confirm this. We alsofound that bulimia and anxiety when eating in the

presence of others indicated a poor outcome, and this

anxiety, bulimia, and vomiting correlated significantlywith chronicity (longer duration) of illness (F=5.56;p<0.05; Fez3; r<0001; and F=10.88; p<O.01, re-

spectively).We also confirmed the view of Crisp et al. 17 that being

married and coming from a lower social class back-

ground were correlated with poorer outcome. Butbecause being married was correlated significantly withan older age of onset and a longer duration of illness(F=29-94 and 17-71, respectively, p<0-001) the exact

significance of this factor as a prognostic indicatorremained uncertain. Like Theander15 and Morgan andRussell4 we found that older age of onset was associatedwith a poorer outcome, but why this was so is uncertain.A poor childhood social adjustment (difficulties in

mixing with other children and few or no friends) corre-lated with poorer outcome. Morgan and Russell4 foundthat school phobia or difficulty in coping with school in-dicated poor outcome. Bruch,’8 Selvini,19 and Crisp20have emphasised that patients are often compliant, over-conforming, yet socially inept when young, often in re-sponse to an overcontrolling and experience-denyingfamily environment. Of the many family factors that weinvestigated, disturbed parental relationships and dis-turbed relationships between the patients and the par-ents before onset of illness both correlated with pooreroutcome. Morgan and Russell also found that the secondof these factors correlated with poor prognosis, whileCrisp et a1.17 20 expressed the view that the patients’ ill-ness often served to lock the parental marriage or inother ways to minimise conflict in one or both parents.Quinton et al.,22 in their study of 137 consecutive fami-lies referred to an inner London psychiatric clinic, foundthat high initial marital discord correlated significantlywith behavioural deviance in the children. We merelyemphasise here the importance of the family environ-ment on the prognosis of anorexia nervosa.

Because this was not designed as a treatment study,it was not possible for us to gauge the effects of treat-ment on outcome. However, the low mortality rate inour very severely ill patients justifies treatment effortsaimed at weight restoration. We’ also believe it justifies

65

attempts to influence the family and individual psycho-pathological factors which appear to influence the evolu-tion of the illness. The long-term effectiveness of suchapproaches remains to be systematically investigated.We thank Dr Gethin Morgan, department of psychiatry, Bristol

University Medical School, for advice on the use of the rating scales;Mr Martin Bland, department of social medicine, St. George’s Hospi-tal Medical School, for advice on statistics; and Mr Jonathan Hart-shorn, computer centre, St. George’s Hospital Medical School, forsupervising the computer analysis of the data. We also thank the pa-tients and their general practitioners for cooperation.

This article is partly based on material submitted by L.K.G.H. inan M.D. thesis to the University of Hong Kong. Requests for reprintsshould be addressed to A.H.C.

REFERENCES

1 Russell, G. F. M. in Anorexia Nervosa (edited by R. A. Vigersky); p. 277.New York, 1977.

2 Cremerius, J. Arch. Psychiat. Zeit. Neurol. 1965, 207, 378.3. Rosman, B. L., Minuchin, S., Baker, L., Liebman, R. m Anorexia Nervosa

(edited by R. A. Vigersky); p. 341. New York, 1977.4. Morgan, H. C., Russell, G. F. M. Psychol. Med. 1975, 5, 355.5. Silverman, A. J. in Anorexia Nervosa (edited by R. A. Vigersky); p. 331. New

York, 1977.6. Bhanji, S., Thompson, J. Br. J. Psychiat. 1974, 124, 166.7. Crisp, A. H. Proc. R. Soc. Med. 1977, 70, 686.8 Crisp, A. H. Br. J. Psychiat. 1965, 11, 505.9. Kemsley, W. F. F. Ann. Eugen. 1952, 16, 316.

10. Siegel, S. Non-Parametric Statistics; p 19. New York, 1956.11. Siegal, S. ibid, p. 116.12. Siegal, S. ibid. p. 213.13. Hsu, L. K G., Crisp, A. H., Harding, B Unpublished.14 Classification of Occupations, Office of Population Censuses and Surveys.

H.M. Stationery Office, London, 1970.15. Theander, S Acta psychiat. scand 1970, suppl. p. 214.16 Lacey, J. H., Chadbund, C., Crisp, A. H., whitehead, J., Stordy, J. Br. J.

Nutrit. (in the press).17 Crisp, A. H., Kalucy, R. S, Lacey, J. H., Harding, B. in Anorexia Nervosa

(edited by R. A. Vigersky); p. 55. New York, 1977.18. Bruch, H. Eating Disorders; p. 621. London, 1974.19 Selvini Palazoli,M. Self Starvation; p. 242. London, 1974.20 Crisp, A H Hospital Med. 1967, 1, 713.21. Crisp, A H., Harding, B., McGuinness, B. J psychosom. Res. 1974, 18, 16722. Quinton, D., Rutter, M., Rowlands, D Psychol. Med. 1976, 6, 577.

DIETARY CARBOHYDRATE ANDMETABOLISM OF INGESTED PROTEIN

ROBERT A. GELFAND ROSA G. HENDLERROBERT S. SHERWIN

Department of Internal Medicine, Yale University School ofMedicine, New Haven, Connecticut 06510, U.S.A.

Summary Six normal people were fed a lean beefmeal while on a normal diet and after

seven days on a very low carbohydrate (<25 g/day)2000 kcal/day diet. After carbohydrate restriction, theprotein-induced rise in branched chain aminoacids was40-50% greater than the rise after the control diet. In-travenous leucine also produced a 40% greater rise inplasma-leucine after carbohydrate restriction. Three

days of fasting exaggerated protein-induced increases inplasma branched-chain aminoacids by 55-77%. Hypo-caloric, pure carbohydrate refeeding restored thebranched-chain aminoacid responses to normal. Severe

carbohydrate restriction thus leads to increased accumu-lation of plasma branched-chain aminoacids after

protein feeding which is at least in part due to reducedutilisation of these aminoacids.

Introduction

LoBB carbohydrate, high protein diets in the treat-

ment of obesity have become a focus of considerable in-terest.1-3 They have also been recommended for themanagement of reactive hypoglycxmia .4 The suggestionthat carbohydrate restriction may confer a unique "pro-tein-sparing" advantage on hypocaloric diets5 6 hasdrawn attention to possible interactions between dietarycarbohydrate and protein metabolism. In spite of thesediets’ popularity, little is known about how carbo-

hydrate restriction affects the metabolism of ingestedprotein.

Muscle tissue, the major reservoir of body protein, isin negative nitrogen balance in the fasted state and res-toration of muscle nitrogen depends on the net uptakeof aminoacids after protein feeding. The branched-chainaminoacids (leucine, isoleucine, and valine)-the majorsubstrates for restoration-are unique in their ability toescape hepatic uptake after intestinal absorption. Theythus account for most of the hyperaminoacidsemia afterprotein feeding and for 60-90% of the aminoacid uptakeby muscle tissue.8 What role, if any, dietary carbo-hydrate has in the utilisation of ingested protein has notbeen established.We examined the effect of carbohydrate restriction on

the plasma aminoacid response to protein feeding

SubjectsTwo groups of subjects were studied. In the first group were

healthy, nonobese adult volunteers aged 23-25 (five males andone female) within 15% of ideal body-weight (based on 1959Metropolitan Life Insurance Company tables). None had a pri-mary family history of diabetes mellitus and none had an ele-vated fasting plasma-glucose. They were studied while on aweight-maintaining diet (1900-2300 kcal/day; at least 200 gcarbohydrate) and again after 7 days on a low carbohydrate(<25 g/day) diet unrestricted in fat, protein, and total calories.Subjects regulated their own diets in accordance with anextensive list of low carbohydrate foods,9 and were observeddaily to ensure they kept to the diet. Their records showed amean +S.E. daily intake of 18+3 g carbohydrate and 1970&plusmn;75kcal.

In the second group were healthy obese subjects aged 27-38(one male, three females) who had volunteered for therapeuticstarvation and diet control at the Clinical Research Center ofthe Yale-New Haven Hospital. They were 54-151% (mean89+22) above ideal body-weight. Each had a normal fastingplasma-glucose and a normal plasma-glucose response to 100g of oral glucose. All had normal thyroid, renal, and hepaticfunction tests. None were taking any drugs. For at least 10days before the study they were on weight-maintaining dietscontaining at least 250 g carbohydrate/day. Their response toprotein ingestion was studied in the postabsorptive state; aftera 3-day fast; and after 5 days of refeeding with a hypocaloric,500 kcal diet containing 120 g carbohydrate (96% of totalcalories) administered as fruit juices. Daily intake during thefast was 2000 ml (or more) water, 1 multivitamin tablet, 1 mgfolic acid, and 1 - g g sodium bicarbonate (sugar-free).

All subjects were informed of the nature, purpose, and pos-sible risk of the study before their written voluntary consentto participate was obtained.

Methods

Protein FeedingThe non-obese subjects were studied in the postabsorptive

state after a 12-14 h overnight fast and the obese group afteran overnight fast and after 3 days of fasting. After the subjects