CHAPTER 4

THE EPIDEMIOLOGY OFSMALLPOX

Contents

INTRODUCTION

how the body reacted and what symptomswere produced, and how the virus was shed .

Chapters 1 and 3 were concerned with Since smallpox was an infectious disease,smallpox as a disease of individual human there is another dimension to its study besidesbeings, discussing such problems as how the its effect on individual patients-namely, itsvirus entered the body and spread through it,

behaviour in human populations . This di-169

PageIntroduction 169The morbidity and mortality of smallpox in the 20th

century 170Secular changes in the global occurrence of smallpox 170Smallpox incidence and incidence rates 173The effect of vaccination on the incidence of smallpox 175Case-fatality rates in smallpox 175Smallpox epidemics in endemic countries 178Seasonal variations in incidence 179

Infection and infectivity 182Routes of exit and entry of virus 182Shedding of variola virus 183Routes of infection 186The infectious dose of variola virus 187The incubation period of smallpox 188The infectivity of cases of smallpox 189The infectivity of case contacts not exhibiting disease 190

Transmission 191Contact infection 191Indirect transmission 191

Factors affecting the spread of smallpox 194Individual susceptibility 194Social factors 196Demographic factors 196Political and economic factors 198Variolation and laboratory-associated smallpox 198

Patterns of spread in populations 199Spatial distribution 199The rate of spread within small groups 199Spread in institutions 201Dispersal by travellers 202Illustrative case studies in rural areas 203Illustrative case studies in urban areas 206

Summary 207

1 70

mension constitutes the epidemiology of thedisease, which involves examination of theincidence of smallpox and the mortalityattributable to it, over time and in differentcountries and regions ; the various determi-nants of its occurrence and spread ; its main-tenance as an endemic disease ; and the causesof epidemics. These considerations were offundamental importance in its control andeventual eradication.

THE MORBIDITY AND MORTALITYOF SMALLPOX IN THE

20TH CENTURY

In Chapter 5 we describe how smallpoxspread throughout the world, first throughAsia, beginning in the 1st century AD, and theninto Europe and northern Africa from aboutAD 700 onwards. With European colon-ization the disease was carried to Central,South and North America and southernAfrica in the 16th and 17th centuries. Bythe time vaccination was introduced, at theend of the 18th century, the distribution ofsmallpox was world-wide. It was endemiceverywhere, except in remote areas withsparse populations, such as Australia, NewZealand and the islands of the Pacific, Atlan-tic and Indian Oceans. Periodically it causeddisastrous epidemics in many smaller islandcommunities and then died out, as, forexample, in Iceland, the islands of the Carib-bean, Hawaii, Tahiti, Mauritius and thesmaller islands of what were then the Nether-lands East Indies (Hirsch, 1883) .

Secular Changes in the Global Occurrenceof Smallpox

By the early years of the 20th centuryendemic smallpox had been eliminated from afew of the countries of northern Europe withsmall populations, although importationsfrom more densely populated neighbouringcountries occurred almost every year (seeChapter 8) . However, until current statisticsbegan to be published by the Health Organ-isation of the League of Nations in 1922(Howard-Jones, 1975), there was no mech-anism for systematically collecting and dis-seminating information on the incidence ofsmallpox in different countries . National datafor the early years of the 20th century werecompiled by Low (1918) and consolidated

SMALLPOX AND ITS ERADICATION

figures for the years 1920-1947 were pub-lished by the Interim Commission of theWorld Health Organization (Fabre, 1948) .

Weekly and consolidated annual figures forreported cases of smallpox in different coun-tries were published by the World HealthOrganization after 1948. These figures,supplemented by investigations carried outby the WHO Smallpox Eradication unit, wereused for the tabulation of reported cases ofsmallpox by country and by WHO regionprovided in the Final Report of the GlobalCommission for the Certification of SmallpoxEradication (World Health Organization,1980). Further supplemented by studies con-ducted by the authors of the present book,these data were used for the detailed tabula-tion of reported cases of smallpox in the largercountries of the world presented in Chapter 8 .These official data reflect only reported cases,and, as will be shown in a later section,smallpox was grossly underreported in mostcountries. The figures convey an impressionof the ebb and flow of the disease in variouscountries, but they give little idea of its trueincidence.

More relevant in the present context is aglobal overview of secular trends in thenumbers of countries in which smallpox wasendemic during the period from 1920 until itseradication (Fig . 4.1) . To obtain the data forthis, all available sources of information wereconsulted, and the assumption was made thatsmallpox was persistently endemic betweenthe 1920s and the 1940s in countries in whichin the 1950s it appeared to be a long-established endemic disease, even thoughsome of these countries had failed to report itsoccurrence to the Health Organisation of theLeague of Nations during that period . Thedata are plotted by 5-year intervals prior to1958 and annually thereafter ; they show foreach continent the secular changes in thenumbers of countries in which smallpox wasendemic. In terms of the global eradication ofsmallpox, important administrative decisionswere taken in 1959, when the World HealthAssembly first launched a programme forglobal eradication, and in 1967, when theIntensified Smallpox Eradication Programmewas initiated (see Chapters 9 and 10) .

The period 1920-1958In 1920 smallpox was endemic in most

countries, being present in 124 and absent

so-

40N

0CA

wL

30-0u

EW

20-

0L_

EZ 10-

AFRICA (47)

ASIA (38)

.\

\AMERICAS (25)

EUROPE (30)

- -N

OCEANIA (16)0 , ,

1920

1925

1930

1935

1940

1945

Fig . 4 .1 . Numbers of countries and territories inwhich smallpox was endemic between 1920 and 1978,arranged by continent . The figures in brackets indi-cate the numbers of countries and territories involved .Their names (as of 1986) are listed below .

Africa : Algeria, Angola, Benin, Botswana, BurkinaFaso (Upper Volta), Burundi, Cameroon, CentralAfrican Republic, Chad, Congo, Cote d'lvoire,Djibouti, Egypt, Equatorial Guinea, Ethiopia, Gabon,Gambia, Ghana, Guinea, Guinea-Bissau, Kenya,Lesotho, Liberia, Libyan Arab Jamahiriya, Madagascar,Malawi, Mali, Mauritania, Morocco, Mozambique,Namibia (South West Africa), Niger, Nigeria, Rwanda,Senegal, Sierra Leone, Somalia, South Africa, Sudan,Swaziland, Togo, Tunisia, Uganda, United Republic ofTanzania, Zaire, Zambia, Zimbabwe .

Americas : Argentina, Belize, Bolivia, Brazil,Canada, Chile, Colombia, Costa Rica, Cuba, Ecuador,El Salvador, French Guiana, Guatemala, Guyana,Honduras, Jamaica, Mexico, Nicaragua, Panama,Paraguay, Peru, Suriname, USA, Uruguay, Venezuela .

only from all the countries of Oceania and 16countries in the other 4 continents . The onlycountries of Oceania with populations largeenough to have supported endemic smallpoxwere Australia and New Zealand, which wereprotected by distance and an effective systemof seaport quarantine from all but a very fewimportations, most of which were controlledat the port of entry .

By 1920 endemic smallpox in Europe hadbeen eliminated only from Denmark, theNetherlands, Norway and Sweden . Subse-quently the situation improved quite rapidly,

4. EPIDEMIOLOGY

1959: Resolution on globalsmallpox eradication

1-1

N

1

171

1966 : Resolution on theIntensified SmallpoxEradication Programme

• I • I I I I I \7'7`1\I 1

1950

1955 58 60 62 64 66 68 70 72 74 76 78

Asia : Afghanistan, Bahrain, Bangladesh, Bhutan,Burma, China, Democratic Kampuchea, DemocraticYemen, Democratic People's Republic of Korea,Hong Kong, India, Indonesia, Islamic Republic of Iran,Iraq, Israel, Japan, Jordan, Kuwait, Lao People's Demo-cratic Republic, Lebanon, Loro Sae (East Timor),Macao, Malaysia, Mongolia, Nepal, Oman, Pakistan,Philippines, Qatar, Republic of Korea, Saudi Arabia,Singapore, Sri Lanka, Syrian Arab Republic, Thailand,United Arab Emirates, Viet Nam, Yemen .

Europe : Albania, Austria, Belgium, Bulgaria,Cyprus, Czechoslovakia, Denmark, Finland, France,German Democratic Republic, Federal Republic ofGermany, Greece, Hungary, Iceland, Ireland, Italy,Luxembourg, Malta, Netherlands, Norway, Poland,Portugal, Romania, Spain, Sweden, Switzerland,Turkey, USSR, United Kingdom, Yugoslavia .

Oceania : Australia, Fiji, French Polynesia, Guam,Kiribati, Mariana Islands, Marshall Islands, Nauru, NewCaledonia, New Zealand, Papua New Guinea, Samoa,Solomon Islands, Tonga, Tuvalu, Vanuatu .

so that by 1935 endemic smallpox had beeneliminated from 25 countries, although itremained relatively common in Portugal,Spain, the USSR and the United Kingdom .The last endemic case in the USSR wasreported in 1936 (see Chapter 8) . Elsewhere inEurope the downward trend continued, andthe continent was free of endemic smallpox by1953 .

In the Americas, tens of thousands of casesof smallpox were reported annually up to thelate 1930s, variola major in Mexico and someSouth American countries and mostly variola

1 7 2

minor in other South American countries andthe USA. Smallpox was eliminated from theCentral American countries between 1920and 1951 and from Mexico in 1951. Endemicvariola major had been eliminated from theUSA in 1926 but variola minor persisted untilthe late 1940s. Smallpox continued to beendemic in the larger countries of SouthAmerica until 1950, when a steady declinebegan. In that year a regional eradicationcampaign for the Americas was initiated bythe Pan American Sanitary Organization(later renamed the Pan American HealthOrganization) (see Chapter 9) and the inci-dence continued to fall until, by 1958, small-pox remained endemic in only 6 countries .

The situation was not nearly as good as thisin Asia and Africa, the continents in whichreporting was the least reliable . The numberof endemic countries in Asia declined slowly,endemic smallpox being eliminated fromIndonesia, Malaysia, the Philippines and SriLanka by 1938, only to be reintroduced intothese countries during the disruption causedby the Second World War . After that therewas a slight but steady improvement, mainlyin the smaller countries of eastern Asia, butthe disease appeared to be largely uncheckedin the Indian subcontinent and was stillendemic in some 40% of the countries of Asiain 1958.

In Africa progress was even slower, exceptfor the gradual reduction in the number ofreported cases and the elimination of endemicsmallpox in the countries of northern Africaand some of the small sparsely populatedcountries in southern Africa. During thisperiod successful country-wide eliminationdid not always mean that subsequent importa-tions would be successfully contained . Forexample, after the elimination of endemicsmallpox from Egypt in the mid-1930s,variola major was reintroduced during theSecond World War and became endemicagain, producing several thousand reportedcases in 1943-1945 (Tulloch, 1980). Therewere similar exacerbations after the SecondWorld War in Algeria„ Libya and Morocco . Inmost countries in Africa south of the Saharasmallpox was apparently almost as common in1958 as it had been two decades earlier .

The period 1959-1966In 1959 the Twelfth World Health Assem-

bly adopted a resolution introduced by theUSSR calling for the world-wide eradication

SMALLPOX AND ITS ERADICATION

of smallpox (see Chapter 9), confident in theknowledge that eradication had by then beenachieved in Europe and in North and CentralAmerica. The situation in South Americaappeared to be manageable, and indeed thenumber of endemic countries decreased until,by 1967, smallpox remained endemic only inBrazil. The real problem was posed by thedeveloping countries of Africa and Asia .During the few years between 1959 and 1966there were substantial gains in Asia, the mostimportant of which was the elimination ofendemic smallpox from China. This wasachieved by 1961 and was due entirely to anational initiative . Smallpox, however, re-mained rampant in the Indian subcontinent .Some progress was made in western andsouthern Africa, but in 1966 the disease wasstill endemic in 27 of the 47 countries ofAfrica.

The period 1967-1978

As outlined in Chapter 9, by 1967 itwas clear that-the Twelfth World HealthAssembly resolution notwithstanding-theglobal eradication of smallpox could not beachieved without greater financial, adminis-trative and scientific support . In that year theIntensified Smallpox Eradication Programmewas instituted, supported by funds from theWHO regular budget, and a Smallpox Eradi-cation unit was established at WHO Head-quarters in Geneva . At the same time, theWHO regional offices responsible for coun-tries in which endemic smallpox was presentassumed an active role in organizing andcoordinating country programmes. The rapidfall in the number of countries with endemicsmallpox over the period 1967-1978 is illus-trated in Fig . 4.1 and described at greaterlength in Chapter 10 .

By the time that the Intensified Pro-gramme was launched, smallpox had beeneradicated in South America, except forBrazil, in which the last case in the Americasoccurred in 1971 . Results were achievedrapidly in the countries of western and centralAfrica in a campaign which began in 1967 .Steady progress occurred in eastern, south-eastern and southern Africa . The biggestchallenge confronting the Intensified Small-pox Eradication Programme was the eradica-tion of variola major from the Indian subcon-tinent, a relatively slow process, which was,however, achieved by 1975. By 1976 the onlyremaining endemic country was Ethiopia, in

which variola minor persisted and spread toSomalia, eventually to be eradicated from theHorn of Africa, and the world, by the end of1977 .

Smallpox Incidence and Incidence Rates

General considerations

The countries and territories listed beneathFig. 4.1 vary enormously in size and popula-tion, from China and India on the one hand,each with populations of several hundredmillions, to Bahrain and Swaziland, withpopulations of a few hundred thousand, onthe other .

The numbers of cases of smallpox reportedannually to the international health authori-ties by the governments of different countriesvaried greatly in accuracy . Those obtainedfrom non-endemic countries with well-estab-lished health services, reporting small out-breaks due to importations, were probably themost accurate, although even here distortionssometimes occurred, owing to the suppres-sion of reports (see Chapter 23) . The reportednumbers of cases in endemic countries for theearlier years of the period under review-often countries with grossly inadequate re-porting systems-greatly understated thetrue incidence of the disease. Even in coun-tries in which smallpox was an importantpublic health problem and there were reason-able health services, reporting was incomplete

4. EPIDEMIOLOGY

173

before and even after national eradicationcampaigns had been launched (see below) .However, the incidence of reported cases (Fig .4.2) provides some indication of the trueincidence of smallpox in the countries inwhich it was endemic in 1967. At the start ofthe Intensified Smallpox Eradication Pro-gramme, the reported incidence amounted toperhaps 1-2% of the true figure .

During the Intensified Programme severalefforts were made to determine what propor-tion of cases was actually reported in coun-tries with endemic smallpox . Two kinds ofassessment were made. The first contrastedthe reported figures with the incidence asestimated by pockmark surveys carried out inselected age groups ; the second was based on acomparison, in India and Ethiopia, of thereported incidence before and after efficientsurveillance and case-reporting systems wereestablished, during the late stages of theeradication campaigns in each country .

Assessment by facial pockmark surveys

First employed by Dr Jacobus Keja inNepal and then in Indonesia (see Chapter 13),this method was later used by Foster(WHO/SE/72.34) and Hughes et al. (1980)for assessing the efficiency of reporting inwestern Africa and Bangladesh respectively .The method involved the examination ofpersons in appropriate age groups for facialpockmarks (Table 4.1) . For survey purposes a

Fig . 4 .2 . Incidence of reported cases in the 31 countries and territories in which smallpox was endemic in 1967 .

1 7 4

(1)

SMALLPOX AND ITS ERADICATION

Table 4. I . Efficiency of reporting of smallpox in Kano (an urban area in Nigeria), asestimated from faci al pockmark surveys in the age group 0-14 yearsa

Estimated population in the age group 0-14 yearsTotal population-500 000, of which 223 000 (44.6%) were 0-14 years old .

(2) Estimation of total number of cases of smallpox in the age group 0-14 yearsObserved frequency of smallpox pockmarks in a sample survey of the age group 0-14 years-3 .7% .Estimated number of children aged 0-14 years with facial pockmarks-3.7/ 100 x 223 000 = 8250 .Proportion of children aged 0-14 years who retained pockmarks for at least 1-4 years-55%.

Estimated number of children aged 0-14 years who had smallpox and survived- 8250 = 15 000.55/100

(3)

Smallpox case-fatality rate in children aged 0-14 years-13% .

Estimated total number of cases of smallpox in the age group 0-14 years-IS 000

= 17 250 .

Estimation of total number of cases of smallpox during the last /S yearsProportion of smallpox cases which normally occur In the age group 0-14 years-50% .

Estimated total number of smallpox cases during the last 15 years-17 250

= 34 500 .50/100

(4) Calculation of the efficiency of reporting smallpoxReported number of smallpox cases during the last I S years-2805 .

Reporting efficiency-2 805

= 8.1% .34 500

a Based on Foster (WHO/SE/72 .34) .

positive case was defined as one which had atleast 5 characteristic round, depressed facialscars, a millimetre or more in diameter. Bycorrecting for the observed rate of smallpoxmortality in a selected age group, for mortal-ity due to other causes and for the disappear-ance of facial scars over time (which variedaccording to the age at which the personacquired smallpox and was determined foreach age group and area) it was possible toestimate the incidence rates of smallpoxwithin the sampled population . The smallpoxincidence in the entire population could beestimated by correcting for the age distribu-tion of cases, and this figure could then becompared with the number of officiallyreported cases .

Using this procedure, Foster calculatedthat the efficiency of reporting was 1 .3% inrural areas of Nigeria and 8 .1 % in Kano, anurban centre . Hughes et al . (1980), using asimilar procedure, found that as late as 1972reporting efficiency in Bangladesh was onlyabout 12%, but rose in the succeeding years,when active searches for unreported caseswere intensified, to over 80% (see Chapter 16,Fig. 16.9). The health services in Bangladeshand Nigeria were more extensive and betterdeveloped than those in most endemic coun-tries at that time, so that these estimatesreflect better-than-average situations.

Changes in the numbers of reported cases afterimproved surveillance

The highest numbers of cases for manyyears were often reported during the periods

just before country-wide elimination wasachieved, since by this time good surveillancesystems were in operation . For example, thehighest figure for India since the 1950s was188 003 in 1974, the year before the attain-ment of eradication. The results of activesearches in India in 1973 (Fig. 4.3), describedin detail in Chapter 15, suggested that at thattime the reporting efficiencies in the highlyendemic states of Madhya Pradesh and UttarPradesh were less than 1% and 5% respec-tively-and this a decade after the launchingof the Indian national smallpox eradicationprogramme and 6 years after the initiation ofthe Intensified Smallpox Eradication Pro-gramme. Likewise, with the development of

L

240

&0 2000

160A

s. 120

c80-

40-

01960

1965

1970

1975

Fig . 4 .3 . Evidence of the underreporting of variolamajor. Increase in the number of reported cases inIndia following the introduction of active searches inthe highly endemic states in 1973 (see also Chapter 15,Fig . 15 .15) . NSEP = National Smallpox EradicationProgramme of India .

87/100

NSEP

Intensifiedstarts

search

240

200

120

40-

01960

1965

Smallpoxprogramme

starts

1970

1975

Fig . 4 .4 . Evidence of the underreporting of variolaminor. Increase in the number of reported cases inEthiopia following the initiation of intensified sur-veillance in 1971 .

better surveillance in Ethiopia in 1971 (seeChapter 21), there was a dramatic increase inthe number of reported cases, 722 beingnotified in 1970 and 26 329 in 1971 (Fig . 4 .4).

For China, the most populous country inthe world, with endemic variola major forcenturies and poor vaccination coverage priorto 1950, the cases reported to the interna-tional health authorities were in the thou-sands only, the highest figure (68 101) beingreported in 1951, just after the nationaleradication campaign had been launched.

The overall pictureIt is clear that the reporting of cases of

smallpox was the most efficient in countriesin which the health services were well de-veloped, which was usually where the diseasewas least common. It was very inefficientelsewhere. Since the countries in which re-porting was good must have provided only asmall proportion of the total number of casesin the world, it is not unreasonable to regardthe official figures reported to WHO asrepresenting only 1-2% of the true inci-dence-probably nearer 1 % for the yearsbefore the initiation of the global eradicationprogramme . In the early 1950s,150 years afterthe introduction of vaccination, there wereprobably some 50 million cases of smallpox inthe world each year, a figure which had fallento perhaps 10-15 million by 1967, when thedisease had already been eliminated in 125 ofthe 156 countries and areas listed in thelegend of Fig . 4 .1 .

4. EPIDEMIOLOGY

175

The Effect of Vaccination on the Incidenceof Smallpox

As has been described in Chapter 1, anattack of smallpox in persons who had beenvaccinated was usually less severe than in theunvaccinated. Much more important, fromthe point of view of the ultimate eradicationof smallpox, was the fact that vaccinationwithin the previous 5 years usually com-pletely prevented disease. Some vaccinatedpersons experienced subclinical infections, asjudged by their serological responses (Heineret al., 1971a). These subjects did not transmitthe disease to others, although the subclinicalinfection substantially increased their level ofimmunity. Even if their immunity had notbeen boosted by subclinical infection orrevaccination, many vaccinated individualswere protected against clinical smallpox for amuch longer period than 5 years . However, nodata exist which allow an accurate assessmentto be made of the proportion of those at riskwho were protected against smallpox atvarious intervals after vaccination. Such dataas are available suggest that vaccination ininfancy protected most of those under 10years of age and at least half of those aged 10-20 years from overt disease.

Case-Fatality Rates in Smallpox

Data from the Indian subcontinent showthat Asian variola major had always been adisease with a high case-fatality rate, of theorder of 20% or more in unvaccinatedpersons . Higher case-fatality rates were ob-served in urban hospitals than in the popula-tion as a whole, as documented in Bangladesh(Koplan et al., 1978 ; see Chapter 1), and incities than in rural areas (see Table 4.6).Historical documents suggest that high case-fatality rates had prevailed everywhere in theworld for many centuries . Indeed, epidemicsin previously unexposed, subsistence-farmingpopulations appeared often to have producedmuch higher mortalities than those seen inAsia, owing to the social disruption andconsequent starvation that they caused .

From about the end of the 19th centuryanother variety of smallpox, variola minor,was recognized, which although it producedsimilar skin lesions had a case-fatality rate ofonly about 1 % (see Chapter 1) . Subsequently,observations made in Africa during the Inten-

1 76

sified Smallpox Eradication Programme sug-gested that smallpox of intermediate severity,with case-fatality rates of 5-15%, was en-demic in many parts of that continent .

In addition to differences in the virulenceof the virus, several other important variablescan be identified that played a role in theobserved variations in case-fatality rates : theage distribution of cases, the effects of pre-vious vaccination, nutritional status, and theinfluence of reporting accuracy, since severe(hospitalized) cases and deaths were morelikely to be reported than milder cases .Pregnant women were especially susceptibleand showed higher case-fatality rates for bothvariola major and variola minor than did menand non-pregnant women of the correspond-ing age groups (see Chapter 1) . Further, therewere interactions between age and vaccina-tion status, since in persons receiving onlyprimary vaccination, susceptibility increasedwith the interval since vaccination and thuswith age. Relevant data from India and fromwestern and central Africa are set out inTables 4.2 and 4 .3 .

Table 4.2 Age distribution and case-fatality rates of smallpox in India and in western and central Africaa

a includes both vaccinated and unvaccinated subjects .b Based on Basu et al. (1979) .c Based on Foege et al . (1975) .

Table 4 .3 Age distribution and case-fatality rates of smallpox in vaccinated and unvaccinated subjects in 6states of India, 1974-I975a

SMALLPOX AND ITS ERADICATION

a Based on Basu et al . (1979).b Thought to be cases vaccinated during the Incubation period .

Foege et al. (1975) noted that the agedistribution of cases in western and centralAfrica in 1968 followed closely the agedistribution of the population, reflecting thepredominant occurrence of cases in isolatedand poorly vaccinated rural areas . In India, onthe other hand, with its much denser andmore mobile population, smallpox was pre-dominantly a disease of children (over 70% ofcases in individuals under the age of 15 years :Table 4.2) ; the situation was similar inBangladesh, Burma, Indonesia and Pakistan .Since the immunity produced by vaccinationin childhood steadily waned, the majority ofcases in vaccinated subjects occurred in olderage groups (54 .3% in persons over 14 years ofage : Table 4.3), whereas in unvaccinatedsubjects 82.3% of cases occurred in childrenless than 15 years old. The age-related trendsin case-fatality rates were similar in India andAfrica, the rates being highest in the veryyoung and in older persons . Children andadolescents (10-19 years of age) had lowercase-fatality rates than any other age group, afeature which was accentuated when only

India, 1974-1975b Western and central Africa, 1968c

Age group(years)

Cases Case-fatalityrate (%)

Age group(years)

Cases Case-fatalityrate (%)Number Number %

< 1 1 373 5 .8 43 .5 < 1 102 4.8 29.41-4 5 867 24 .9 24 .5 1-4 417 19 .6 11 .55-14 9 501 40 .4 11 .4 5-14 494 23 .2 7.715-39 5 698 24 .4 9 .0

15-44 1 009 47 .5 15.240-49 695 2 .9 20 .1

45 103 4.8 32.0X50 412 1 .7 37 .4

Total 23 546 100 17 .4 Total 2 125 100 14.2

Vaccinated UnvaccinatedAge group(years) Cases Case-fatality Cases Case-fatality

Number % rate (%) Number % rate (%)

0-4 114 13 .3 10 .5b 725 36 .8 45.75-14 277 32 .4 5 .1 897 45 .5 12.415-39 348 40.7 4.9 265 13 .4 20.7>40 116 13 .6 8 .6 84 4 .3 29.8

Total 855 100 6 .2 1971 100 26.5

unvaccinated subjects were considered. How-ever the data are analysed, the case-fatalityrates were higher in India than in western andcentral Africa. Nevertheless, the age distribu-tion of cases clearly influences the overallcase-fatality rate for a particular geographicalregion, and critical comparisons of geo-graphical variability in case-fatality rateswould have to assess separately the figures forvaccinated and unvaccinated subjects andconsider the age distribution of cases ofsmallpox. Unfortunately, there are very fewlarge series of cases, from different geographi-cal areas, that give such details ; the analysesthat follow are the best available .

Shafa (WHO/SE/72 .35) analysed data from11 areas for which he considered that reason-ably reliable information was available, mak-ing adjustments for age-specific case-fatalityrates by adjusting the data to a standard agedistribution. The case-fatality rates for 9 areasin which variola major was endemic areshown in Table 4 .4. They were always higheramong infants and children aged 0-4 yearsthan among individuals in older age groups.

Table 4.4 Case-fatality rates of varlola major indifferent geographical areas for the totalpopulation and the age group 0-4yearsa, b

a Based on Shafa (WHO/SE/72 .35) .b Includes both vaccinated and unvaccinated subjects .c More recent data (Joarder et al ., 1980) suggest that the lower

figures (26 .8% and 18 .5%) represent more accurately the truesituation In Bangladesh .

4. EPIDEMIOLOGY

Table 4 .5 Age-specific case-fatality rates in variola minor (data collected during the Intensified SmallpoxEradication Programme)

The effect of age on mortality was evenmore pronounced in variola minor, in bothBrazil and Africa, in which the case-fatalityrate was much higher for infants (less than 12months old) than for any other age group(Table 4.5) . Indeed, in Brazil, in which datawere available for infants less than 3 monthsold, the case-fatality rate in this group was16.7% .

Geographical variations in crude case-fatality rates, based on data obtained duringthe Intensified Smallpox Eradication Pro-gramme, are shown in Table 4.6. There arestriking differences between the case-fatalityrates in large cities and those in the surround-ing countryside (see Indonesia and Pakistan) .This may have been due partly to the betterrecording of deaths and partly to the fact thatdata for the cities were derived in largemeasure from hospitals, in which more severeand neglected cases were more common(Koplan et al., 1978) .

There are also large differences between thecase-fatality rates in countries in which var-iola minor was endemic and those in othercountries. Variola minor itself varied inseverity, the form present in Botswana in1972 (which may have been characteristic of"amaas" of southern Africa) being particu-larly mild . Smallpox other than variola minorwas associated with case-fatality rates thatvaried in different geographical areas fromslightly more than 5% to over 20% . Therewere suggestions in the mid-1960s thatstrains of virus from eastern African countries(Uganda and the United Republic of Tanza-nia) could be differentiated from Asian strainsof variola major virus by laboratory tests, andit was suggested that the disease there shouldbe called "variola intermedius". However, the"characteristic" laboratory reactions of "var-iola intermedius" were not invariable forstrains from that area, and case-fatality ratesonly slightly higher than those found ineastern Africa were also recorded for Indone-

177

Age group(years)

Brazil (1968-1969) Botswana (1971-1972) Ethiopia (1971-1976) Somalia (1977)

Number ofcases

Case-fatalityrate (%)

Number ofcases

Case-fatalityrate (%)

Number ofcases

Case-fatalityrate (3's)

Number ofcases

Case-fatalityrate (3's)

< I 387 5 .2 17 5 .9 1 322 7 .9 47 12.81-4 2 322 0 .7 195 0.5 13 501 1 .8 506 0.65-14 4 389 0 .2 505 0.0 26 087 0.6 1 061 0.115 2 683 1 .0 365 0 .0 14 081 2.5 1 408 0.1Total 9 781 0 .7 1 082 0 .2 54 991 1 .5 3 022 0.4

Age groupCountry or area 0-4 years

(%)All ages(%)

Bangladeshc 47 3626.8 18 .5

India (Tamil Nadu) 43 26Burma 23 17Afghanistan 19 16India (Punjab) 18 15Indonesia (Jakarta) 18 13Indonesia (West Java) 11 8West Africa 14 13Togo 10 8

1 7 8

a Includes both vaccinated and unvaccinated subjects .b Data in italics refer to urban areas .CExcluding cases due to variolatlon.

sia and western Africa. As has been suggestedin Chapters 1 and 2, it is possible to differen-tiate between variola major and variola minorby considering clinical features and epidemi-ology and, for some strains, the laboratorycharacteristics of the virus, but it is not worthwhile to attempt to designate "variola inter-medius" or to differentiate further betweenstrains of variola major. The data shown inTables 4.4 and 4 .6 illustrate the fact that bothvariola major and variola minor were causedby strains of variola virus which differed intheir virulence for man, and that some strainswere predominant in some geographical areasand other strains in other areas.

Smallpox Epidemics in EndemicCountries

It was widely observed that in countries inwhich smallpox was endemic there wereperiodic episodes of much higher incidence :"epidemic years". The history of smallpoxbefore the 20th century is replete withaccounts of great epidemics occurring againsta background of endemic smallpox (seeChapter 5) . An unusually complete series ofdata on smallpox incidence in the blandIslands, Finland, extending from 1751 to1890, showed a 7-year cycle of epidemics inthe 18th century, which changed to an 8-yearcycle, with a higher proportion of casesamong adults, after vaccination became avail-

SMALLPOX AND ITS ERADICATION

Table 4.6 Case-fatality rates of smallpox in different geographical areas (data from country reportssubmitted prior to certification of eradication of smallpox)a

4rvc 3000-

v

1000-

01910

1920

1930

1940

1950

1960

Fig . 4 .5 . Cyclic fluctuation in the incidence of small-pox in the city of Madras, India : annual numbers ofcases and deaths between 1910 and 1960 . (Based onRao et al ., 1960 .) A similar cyclic fluctuation, withepidemics every 5-10 years, occurred in India as awhole and most other countries in which smallpoxwas endemic . (See Chapter 5, Fig . 5 .2 and Chapter 6,Fig . 6 .1 .)

able in 1805 (Mielke et al ., 1984). Thephenomenon of epidemic years is well illus-trated by more recent data from India . In thecity of Madras, for example, Rao et al . (1960)found that there was a regular peaking ofincidence every 4-6 years, epidemics extend-ing over a period of 3 years, the middle yearbeing the peak year (Fig . 4 .5) . This periodic

Number of

Number ofcasesb

deathsbCase-fatalityrate (%)b

1 127

209 18 .523 546

4 103 17 .41 898

306 16 .117 491

1 646 9 .4587

140 23.911 966

930 7.8405

82 20.2

5 628

540 9.61 629

150 9.21 045

54 5.2

9 854

75 0.8

54 991

838 1 .53 019

3S 1 .23 022

12 0.41 059

2 0.2

5000 -

Cases.4000- Deaths

Country or areab Period

Variola major: AsiaBangladesh 1975India 1974-1975Afghanistan c 1969-1973Pakistan (Sind Province) 1972-1974Pakistan (Karachi) 1973-1974Indonesia (West Java) 1969Indonesia (Jakarta) 1968Variola major: AfricaWest Africa 1967-1969United Republic of Tanzania 1967Uganda 1966-1970Variola minor : South AmericaBrazil 1966-1969Variola minor : AfricaEthiopia 1971-1976Sudan 1970-1972Somalia 1977Botswana 1972

behaviour is a consequence of the size of thepool of susceptible individuals in the popula-tion and the rate of their removal by infectionor vaccination. The phenomenon occurs in allcommon infectious diseases in which infec-tion is followed by long-lasting immunity ; ithas been exhaustively analysed in measles(Bartlett, 1957) and is critically dependent onthe rate of addition of new susceptible indi-viduals to the population. A factor whichlengthened the interval between epidemics ofsmallpox was that the occurrence of anepidemic led to greatly increased vaccinationrates, because of increased activity by publichealth officials and increased demand forvaccination by the general public .

Seasonal Variations in Incidence

Most infectious diseases show character-istic seasonal variations in incidence . Intemperate climates, where there are pro-nounced seasonal differences in temperature,arbovirus infections usually occur in summer,enteroviral infections in summer and autumn,influenza and other infections of the respir-atory tract mainly in winter, and measles,chickenpox and mumps mainly in winter andspring. Smallpox showed a seasonal incidencesimilar to that of measles and chickenpox ; itwas mainly a disease of winter and spring . Forseveral of these diseases the seasonal varia-tions are blurred in tropical climates, wherethe seasonal changes in temperature andhumidity are often much less marked . How-ever, smallpox showed a response to seasonaleffects in many tropical regions ; in Bangla-desh the seasonality was so pronounced thatsmallpox was called, in Bengali, guti bashunto,the spring rash (joarder et al., 1980) .

The most detailed studies of the seasonalityof smallpox were those reported by SirLeonard Rogers, using mortality figures fromBritish India (Rogers, 1926,1948) and data onreported cases from England and Wales(Rogers, 1928) and certain parts of Africa(Rogers, 1948) . Although many of his datawere poor, comprising deaths rather thancases, both being grossly underreported,Rogers' observations of the seasonal incidenceof smallpox have been confirmed by the moreaccurate data on case incidence obtainedduring the global smallpox eradicationcampaign.

Fig. 4.6 shows the monthly incidence ofreported cases of smallpox in representative

4. EPIDEMIOLOGY 179

Plate 4.1 . Sir Leonard Rogers (1868-1962) . Dis-tinguished British physician who worked in India formany years and was Professor of Pathology atCalcutta Medical College from 1900 to 1920 . Besidesmaking important contributions to many aspects oftropical medicine, he devoted particular attention toa study of seasonal effects on the incidence of small-pox .

countries in the Northern and SouthernHemispheres and in the humid tropics. InBangladesh the maximum reported incidencewas in the period from January to the end ofApril ; in Brazil, in the Southern Hemisphere,the maximum reported incidence occurred inthe spring, from August to the end ofOctober, with well-marked minima in Sep-tember to the end of December and inFebruary to the end of May. The maximumincidence in Indonesia occurred in Januarybut thereafter the monthly incidence showedlittle variation . Bearing in mind the delays inreporting (see box), the maximum transmis-sion rate probably occurred in December inBangladesh and in June in Brazil-i .e ., duringthe early winter. Old records for Europeancountries (Low, 1918 ; Mielke et al., 1984)show that smallpox there was also mainly awinter disease.

Within India, Rogers (1928) observed thatthe seasonal fluctuation was the least markedand the incidence the most uniform from yearto year in the State of Tamil Nadu (formerlyMadras), in South India, which experiencedlittle seasonal variation in temperature andhumidity ; these observations were confirmedfor the city of Madras by Rao et al. (1960) .Other countries in the tropics, which also

1 80

Interpreting Seasonal Fluctuations in Incidence

The analysis of seasonal factors in smallpox is based on curves of monthly incidence ofreported cases or deaths, as illustrated in Fig. 4 .6 . The extent of reporting may affect suchcurves in a dramatic way, as illustrated during the "search weeks" in India (see Fig . 4 .3) andafter the initiation of the Intensified Smallpox Eradication Programme in Ethiopia (seeFig. 4.4). However, if it is assumed that the extent of reporting was reasonably uniformthroughout each year, over a period of several years, clearly there were large differences inthe incidence of reported cases at different times of the year . It is reasonable to ascribe suchseasonal effects on the incidence of smallpox to factors, biological and/or social, that affecttransmission. It is necessary, however, to determine the average time-lag betweentransmission and the recording of cases in the statistics .

As the WHO system reported smallpox cases by date of detection rather than date ofoccurrence, the observed pattern of disease represents a delay from actual transmission byperiods of up to 2 months . Factors contributing to this time-lag include the incubationperiod, delays in case detection, and delays in reporting. As many outbreaks were notdetected until the 2nd, 3rd or 4th generation of transmission, all cases detected on theinitial investigation would have been recorded in that week-the week of detection .

experience less pronounced seasonal differ-ences in humidity and rainfall-e .g., Indone-sia and Zaire-likewise exhibited much lesspronounced seasonal differences in the inci-dence of smallpox (Fig . 4 .6) . But where therewere clearly distinguishable hot and coolseasons, with high and low absolute humid-ities respectively, the incidence was alwaysmuch higher in the cool, dry season . Data onimportations of smallpox into Europe duringthe period 1961-1973 support the findings inthe endemic countries (see Chapter 23) . Therewere 3 times as many importations during themonths December-May as in the following 6months, which is explained by the higherincidence of smallpox in the main "expor-ting" countries in the Indian subcontinent atthat time. Further, each case imported duringDecember-May gave rise to an average of 24subsequent cases (median 4 .5), whereas eachcase imported in the period June-Novembergave rise to an average of 1 .6 cases (median1 .0) (Henderson, 1974) .

A variety of factors can be envisaged thatcould have contributed to the seasonal inci-dence of smallpox : viability of the virus in aninfectious state, social factors and possibly thephysiological susceptibility of the host .

Viability of the virusExperiments described in Chapter 2 show

that the viability of variola and vaccinia

SMALLPOX AND ITS ERADICATION

viruses was less prolonged at high than at lowtemperatures and at high than at low hu-midities. This was true with virus in scabs(Huq, 1976), on raw cotton (MacCallum &McDonald, 1957) and in an aerosol (Harper,1961). The correlation of low temperatureand humidity with a higher incidence ofsmallpox was observed in many differentcountries-e.g., India (Rogers, 1928 ; Basu etal., 1979), Bangladesh (Joarder et al ., 1980),western and central Africa (Foege et al .,1975), Brazil (Morris et al ., 1971), the USSR(Low, 1918) and Nyasaland (Malawi) (Rog-ers, 1948)-suggesting that the effect ofenvironmental conditions on the viability ofvariola virus was probably an importantfactor in determining the seasonal incidenceof smallpox . In some situations, this physicaleffect was supplemented and amplified bysocial events that increased opportunities fortransmission during the dry season (seebelow) .

Changes in susceptibility of the hostAnother possibility, often invoked to ex-

plain the winter incidence of influenza, is thatthere may be seasonal changes in individualsusceptibility, due perhaps to changes inmucous membrane permeability or to alter-ations in resistance associated with dietarychanges. There is no convincing evidence thatsuch factors were important in smallpox .

20-

5-

0

mm350-

300-

250-

200-

1S0-

100-

50-

0

BANGLADESH1968-1970(12627)

J F M A M J J A S O N D

J F M A M J J A S O N D

Fig . 4 .6 . The monthly incidence of reported cases of smallpox in countries situated in the Northern Hemisphere(Bangladesh), mainly in the Southern Hemisphere (Brazil) and in the tropics (Indonesia), as observed during theIntensified Smallpox Eradication Programme . Figures in brackets indicate numbers of cases . The lower graphsshow the average monthly rainfall (mm) and maximum temperatures (°C) for Dhaka (Bangladesh), Rio de Janeiro(Brazil) and Jakarta (Indonesia) . There was a pronounced seasonal incidence in countries with well-marked wetand dry seasons, but this feature was not as obvious in localities in tropical countries with sustained highhumidities, such as Java (Indonesia), or Madras (India) .

Social factors

Social factors of various kinds amplified theeffects of temperature and humidity on theviability of variola virus. In western andcentral Africa, Foege et al. (1975) observedthat the seasonal variation-low in the rainyseason (September, October) and reaching apeak in the late dry season (March andApril)-was less marked in the humid coastalareas and more pronounced in the dry savannaor Sahel. They suggested that in part, at least,this was due to the effects of climate on socialactivities . Planting during the rainy seasonsresulted in the dispersion of the population inthe fields, whereas in the dry seasons therewere festivals and many more social contactsin the agricultural areas and the long-distancemovement of nomads in the sub-Saharan

4. EPIDEMIOLOGY

181

J F M A M J J A S O N D

Rainfall

JAKARTA

Temperature

Z

J F M A M J J A S O N D

'C- 35

30

25

20

15

10

5

0

areas. These activities facilitated the dissemin-ation of the virus .

In Tamil Nadu (Madras), where theseasonal variation in the incidence of small-pox was not pronounced, Rao (1972) suggest-ed that such variation as was seen was more"man-made" than natural . For example, twopopulation groups belonging to two differentcommunities and professions (Harijans andGounders) lived very closely together in thesame villages, but an outbreak of smallpox waslimited to the Gounders (WHO/SE/68 .6,Rao). Also, there was greater crowding and,because of the seasonal fluctuation in births,greater numbers of unvaccinated infants inthe winter and rainy months, which Raobelieved was enough to explain the limitedseasonal fluctuation in incidence seen in

DHAKA 'C mm35 350- RIO DE JANEIRO 'C mm

35 350-

-30 300- 30 300-

-25 250- 25 250-

-20 200- 20 200-

-15 150 15

ISO-

-10 100- 10 100-

-5

50- 5

50-

0

0 0

1 8 2

Forecasting Epidemic Years

The prime motive for Roger's studies was to forecast years when smallpox might causemajor epidemics. Bearing in mind that seasonal conditions alone were not sufficient toproduce epidemics, which could not occur if the pool of susceptible individuals wassubstantially reduced by recent epidemic prevalence, Rogers (1928, 1948) found aremarkably good association between failure of the monsoon rains and consequently lowabsolute humidity during the rainy season and epidemic prevalence in the following dryseason. Transmission always continued through the rainy season, but there were morechains of transmission when the monsoon rains were poor than with a good monsoon andsustained high humidity. Thus, if the monsoon failed, the population was alreadyextensively seeded with virus when the cooler dry season began, and the stage was set for amajor epidemic. Using this criterion, Rogers (1948) found that the great majority of majorepidemics could be correctly predicted ; most exceptions were satisfactorily explained bythe arbitrary nature of his criterion of "epidemic levels" . Murthy et al . (1958) noted that inUttar Pradesh more severe epidemics were likely if an unduly high incidence of smallpoxoccurred in the post-monsoon months of October and November .

Another factor, on which Rogers did not comment, was that when the monsoon rainswere poor, the harvests were also poorer and hunger was more widespread. The extensivemovements of refugees and others in search of food led to increased opportunities for thetransmission of smallpox .

Madras city . On the other hand, in Pakistan,Mack et al. (1972b) found that there was noseasonal variation in the incidence of trans-mission within compounds ; almost all sus-ceptible subjects were infected when exposedto a case of smallpox, regardless of season. Theinterval between successive generations ofcases was about 3 days longer in the coolseason, a result compatible with longer viralsurvival but not explained by changes insocial activity .

Seasonal variation in incidence in relation toeradication

Whatever its causes, the seasonal variationin the incidence of smallpox was an importantfactor in the planning of national eradicationcampaigns. For example, in India before 1973it had been customary to utilize the end of themonsoons and the beginning of the coolseason (September-October), when relativelylittle transmission occurred and there was nopublic pressure on health workers to vacci-nate, as the leave period for smallpox pro-gramme workers. With the intensification ofthe eradication programme in 1973, ratherthan merely reacting to the high frequency ofdisease during the dry season, the policy waschanged to concentrate on reducing the foci

SMALLPOX AND ITS ERADICATION

of infection during the wet season, when caseswere the least numerous. In several countries,the efforts of smallpox workers during therainy season were concentrated in the urbanareas, where smallpox foci were the mostnumerous. When the rural areas became moreaccessible at the end of the wet season, effortswere made to eliminate persistent foci ofinfection during the seasonal lull intransmission .

INFECTION AND INFECTIVITY

Routes of Exit and Entry of Virus

The pathogenesis of smallpox, outlined inChapter 3, is concerned in the main with howvariola virions moved between different cells,organs and tissues within the human bodyand how the body responded . Epidemiologyis concerned with the movement of virusparticles between human beings-i .e., withthe spread of infection in populations.

There are three principal routes of viralinfection, corresponding to the three princi-pal surfaces of the body : the respiratory tract,the alimentary tract, and the skin . Minorroutes of infection include the urinary andgenital tracts and the conjunctiva. Althoughcongenital infection occasionally occurred in

smallpox (see Chapter 1), it was of noepidemiological importance and will not bediscussed further in this chapter .

There is no evidence that infection everoccurred via the alimentary tract in smallpox .In experiments in monkeys, A . R. Rao (per-sonal communication, 1981) fed concentratedsuspensions of variola virus in food, orintroduced concentrated viral suspensionsdirectly into the stomach with a stomach tube .The monkeys were not infected, nor couldvirus be recovered from their stools. Thelatter result may have been due to the fact thatvariola virus probably behaves like vacciniavirus, which is rapidly inactivated in sewage(K. R. Dumbell, personal communication,1982) .

Mechanical transmission by arthropods isimportant in some poxvirus infections (myx-oma virus in rabbits, fowlpox virus in birds) .Sarkar et al . (1973c) showed that mosquitoscould contaminate their probosces with var-iola virus by feeding on viraemic mice, andthey speculated about the relevance of this tohuman smallpox. Houseflies could and prob-ably did transport variola virus mechanically,especially in tropical countries, but theirimportance was probably trivial comparedwith other sources of infection and in any casethere is no evidence that the ingestion ofvirus-contaminated food would producesmallpox. There is no epidemiological evi-dence to suggest that arthropods of any kindwere involved in the transmission ofsmallpox .

Natural infection in smallpox usually oc-curred via the oropharynx or nasopharynx,sometimes via the lower respiratory tract, inspecial cases (inoculation variola and variola-tion) via the skin, and possibly, but in any casevery rarely, via the conjunctiva . Before de-scribing the processes of infection by theseroutes it is useful to consider how variolavirions were shed from the infected patientand entered the environment and the way inwhich their mode of exit from the patient mayhave influenced their infectivity for man .

Shedding of Variola Virus

As we have shown in Chapters 1 and 3, afteran incubation period of 12-14 days duringwhich there was no evidence of viral shed-ding, lesions developed in the skin and themucous membranes of the nose and mouth .

4. EPIDEMIOLOGY 183

Oral, nasal and pharyngeal secretions

Because there is no tough stratum corneumin the mouth and pharynx, the lesions thereulcerated very soon after their formation (seeChapter 3, Plate 3 .7) and released largeamounts of virus into the oral and pharyngealsecretions. Several reports are available on theviral content of oropharyngeal secretions invariola major, all carried out with materialobtained from cases occurring in the Indiansubcontinent (Madras : Downie et al., 1961a ;Calcutta : Sarkar et al ., 1973a ; Bihar : Kita-mura et al ., 1977b ; Karachi : Shelukhina et al .,1973). Table 4 .7 sets out the results of the twomost extensive studies. The higher propor-tion of positive results recorded by Sarkar etal. (1973a) probably reflects the differentmodes of collection of the specimens in thetwo studies : throat swabs in Sarkar's series (ofwhich the whole extract in 1 ml of fluid wasassayed) and 5 ml of mouth washings (ofwhich only 0.2 ml or 0 .3 ml was assayed) inDownie's series. Besides the volumetric dif-ferences, throat swabs could be expected todetach virus particles from superficial lesionsas well as including those free in the saliva .Both groups of investigators recorded posi-tive results throughout the period of rash(3rd-14th days).

The results of titrations carried out bySarkar et al. (1973a), shown in a condensed

Plate 4 .2. Jaladhi Kumar Sarkar (b. 1916) . FormerlyProfessor of Virology and Director of the School ofTropical Medicine, Calcutta, India . He carried outmany investigations into the virology and epidemiologyof smallpox as it occurred in Calcutta .

1 8 4

SMALLPOX AND ITS ERADICATION

Table 4.7 Variola virus In the oropharyngeal secretions and urine of cases of variola major

a Based on Sarkar et al . (I973a).b Titres expressed as pock-forming units per ml .

Table 4.8 Variola virus in the oropharyngeal secretions of cases of variola majora

form in Table 4.8, reflect the high viralcontent of the oropharyngeal secretions dur-ing the first week. The viral titres reachedtheir highest level on the 3rd and 4th days ofthe disease (i.e., just after the appearance of therash) ; they were highest and persisted for thelongest period in the most severe cases(confluent, which in Sarkar's terminologyincluded confluent ordinary-type and flat-type smallpox). In fatal cases virus was usuallystill present in the throat swabs at the time ofdeath ; in non-fatal cases it was found for aslong as the 14th day from the onset of fever in

confluent ordinary-type smallpox and untilthe 7th-9th days in discrete ordinary-typesmallpox .

Skin lesions

The titres of virus in vesicular fluid and inscabs were assayed by Mitra et al . (1974) andKitamura et al . (1977b). On about the 18thday of the disease (15th or 16th day of rash)the skin lesions scabbed and subsequently thescabs separated . Assay of the viral content ofsuch scabs revealed that they contained a large

Oropharyngeal secretions Urine

Day ofdisease(from onsetof fever)

Downie et al .(1961a)

Sarkar et al .(1973a)

Downie et al .(1965a)

Sarkar et al .(1973a)

Numbertested

Numberpositive

Numbertested

Numberpositive

Numbertested

Numberpositive

Numbertested

Numberpositive

I 1 0 -2 4 0 2 2 - -3 8 4 10 10 5 54 20 8 22 22 - - 9 95 39 18 20 20 I 0 16 156 49 32 22 22 3 I 14 127 51 33 22 21 5 3 10 68 39 24 18 14 4 2 9 69 34 22 18 II 3 2 8 6

10 10 0 13 7 4 2 5 4II 15 4 10 6 3 2 5 412 12 3 8 4 5 2 3 313 12 1 9 4 I I 4 414 5 1 5 4 2 I - -15 3 0 2 0 I 0 3 316 I 0 1 0 I I I 017 2 0 2 118 4 0 I 019 - - 2 120 1 0

Day ofdisease(from onsetof fever)

Results Number of positivespecimens with titre of:b

10 6Numbertested

Positive10 10 2 103 10 4 10 5Number

2 2 2 100 - I I3 10 10 100 2 I 6 14 22 22 100 7 7 85 20 20 100 - - 10 7 36 22 22 100 1 8 7 6 -7 22 21 95 2 8 9 28 18 14 78 2 7 3 29 18 II 61 1 7 310 13 7 54 3 3 III 10 6 60 1 4 I12 8 4 50 1 313 9 4 44 4 -14 5 4 80 2 215 2 0 016 I 0 0

amount of virus and that the viral titre, asassayed by extracting the scabs in saline,remained at a high level throughout convales-cence (Mitra et al., 1974) .

Epidemiological observations, describedbelow, abundantly confirmed the higher fre-quency of infection after face-to-face contactwith a patient during the 1st week of rash .Exposure to patients in the late stages of thedisease, when large amounts of virus werebeing released into the environment in thescabs, was much less likely to produce infec-tion in susceptible contacts. The reasons forthis difference in frequency of transmissionwere difficult to study experimentally . Oro-pharyngeal secretions expelled early and scabsreleased late in the disease contained virionsthat were equally infectious by ordinary assaymethods, which involved suspending ma-terial from swabs in saline, or homogenizingthe scabs by grinding, and assaying thesuspension or the homogenate on the chorio-allantoic membrane or in cultured cells .Perhaps the main reason for the differencesobserved lay in the physical state of viralparticles shed by the patient from the enan-them and in scabs.

Virions released from lesions of the enan-them into the oropharyngeal secretions inthe early stages of the rash were expelled bythe patient in liquid droplets of various sizes,which might be inhaled by persons in closecontact with the patient or transferred di-rectly to the nose or oropharynx by fingers orobjects contaminated with infected saliva ornasal secretions (Knight, 1973 ; Gwaltney &Hendley, 1978). Many of the large dropletswould rapidly fall and dry on the bedclothesor the floor, being readily dispersed again inthe immediate vicinity of the patient. Virionstransferred in these ways would readily comeinto contact with susceptible cells . In hospi-tals in non-endemic countries, nurses whomade up the beds of patients suffering fromundiagnosed smallpox were often infected ;others were protected by vaccination butdeveloped "smallpox-handler's lung" due tothe inhalation of virus .

In contrast, virions in scabs were enclosedwithin the inspissated pustular fluid and werepresent in flakes which contaminated thepatient's skin and bed-linen and the dust onthe bed and the floor of the room . It is likelythat virions in scabs would rarely have comeinto direct contact with susceptible cells,since inspired particulate matter of this kindis usually swept up in the mucous secretions of

4. EPIDEMIOLOGY

185

the oropharynx and swallowed, or expelledagain into the environment . However, suchvirions could retain infectivity for long per-iods, as judged by laboratory assay.

In spite of this virological evidence, thepolicy of isolating patients adopted duringthe Intensified Smallpox Eradication Pro-gramme followed that traditionally taught-i .e ., patients were not removed from isolationin hospital or at home until the last scab hadseparated. As smallpox eradication becameimminent in Bangladesh and India, villageswere considered to house infective persons for6 weeks after the onset of the last recognizedcase. This policy of isolation and surveillanceprovided a safety margin, since sometimesinfants or other persons became infected andwere hidden from surveillance officers, thusconstituting a missed generation of cases .

Several studies have been made of thepresence of variola virions in the air and onfabrics and skin in the vicinity of smallpoxpatients. In evaluating these reports it must beborne in mind that the ventilation rate ofadult human beings is about 10 litres perminute, so that the air-sampling device wouldhave had to test a volume of about 600 litres inorder to be comparable to an hour's exposureof a susceptible adult person . Unfortunately,rather inefficient methods of air samplingwere used in the earlier studies . Using smallglass funnels tightly packed with dry cottonwool through which air was drawn, Meikle-john et al . (1961) obtained only one positiveresult in the wards of the Infectious DiseasesHospital in Madras. Subsequently, Downie etal. (1965a) analysed the immediate environ-ment of smallpox cases in the same hospitalusing three devices : (1) a fluid impingerwhich excluded particles of 18 ,um or more indiameter, held near the patient's mouth forperiods of 10-15 minutes that included talk-ing and coughing ; (2) settling-plates placedbelow air samplers ; and (3) swabbing of skinand bedclothes . The results, summarized inTable 4.9, show that virus was rarely found inthe small airborne droplets or droplet nuclei,although the saliva and the patient's pillowcovering were heavily contaminated .

Westwood et al. (1966) studied the problemexperimentally by assaying the air in theenvironment of rabbits infected with rabbit-pox virus with sampling devices similar tothose used by Downie et al . They recordedsome positive results, but sampling was oftennegative at times when the rabbits wereclearly infectious. Thomas (1970a) repeated

1 86

SMALLPOX AND ITS ERADICATION

Table 4.9 Recovery of variola virus from the vi-

Urinecinity of patients with variola majora

a Based on Downie et al . (I 965a).

these experiments using a slit sampler (An-dersen, 1958 ; Thomas, 1970b) that sampled alarger volume of air and did not discriminateagainst the collection of larger particles froma heterogeneous aerosol, as did the fluidimpinger and electrostatic precipitator. Asmight be expected, he obtained more frequentpositive results and recorded higher titres ofvirus than the earlier workers. SubsequentlyThomas (1974) compared impingers andsedimentation plates of the type used byDownie et al. (1965a) and a slit sampler likethat used in his rabbitpox experiments in thesampling of the air in a hospital in England.His subjects were a few patients in a ward whowere recovering from variola minor, a situa-tion in which much less environmental con-tamination with virions would have beenexpected than in the Infectious DiseasesHospital in Madras . The impingers gaveuniformly negative results, but virus wasdetected with the slit sampler in air near thebed on several occasions, associated with thephysical activity of patients with active le-sions. Sedimentation plates exposed for somehours about 20 feet from the occupied bedsalso yielded virus on several occasions .

An important feature of smallpox in rela-tion to long-distance airborne spread was thatpatients with uncomplicated smallpox usuallyhad no respiratory signs or symptoms ; cough-ing and sneezing, which generate large cloudsof infective aerosols in the acute stages ofmeasles, for example, rarely occurred. Whenthey did, as in the index cases in the Monschauand Meschede Hospital outbreaks (Anders &Posch, 1962 ; Wehrle et al., 1970 ; see later inthis chapter) airborne infection within build-ings could occur.

The urine was examined for variola virusby Downie et al. (1965a) and Sarkar et al .(1 973a), with the results shown in Table 4 .7 .Sarkar et al. recorded the titre of virus in caseswhich at some time had viruria (21 out of the39 cases examined) . The titre was highestearly in the disease (days 5 and 6) ; it washigher and more persistent in severe cases(haemorrhagic smallpox and those with aconfluent rash) than in milder cases . No viruswas detected in the urine of some cases whoseseverity was comparable to that of cases withviruria. Shelukhina et al. (1973) reported thata few patients excreted small amounts of virusin the urine during convalescence (23rd and25th days of the disease). However, bothbecause of the small quantities of virusinvolved and because of the mode of excre-tion, it is unlikely that urine was an importantsource of infectious virus.

Routes of Infection

Variola virus usually gained entrance to thebody via the oropharynx or respiratory tract,and occasionally through the skin, usually bydirect inoculation, as practised in variolation .

Entry via the respiratory tract

Particles enter the nose or mouth, otherthan in food or drink, either by inhalation, orby implantation on the oral or nasal mucousmembrane by contact with contaminatedfingers. A great deal of experimental work hasbeen carried out on the fate of inhaled bacteriaand viruses (reviewed in Bacteriological reviews,1966 ; Gregory & Monteith, 1967 ; Hers &Winkler, 1973) . Some relevant aspects can besummarized as follows. The ventilation rateof an adult human being is about 10 litres perminute. Most particles larger than 15 ym indiameter and about half those 6 µm indiameter are retained in the nose. Largeparticles deposited in the nasal cavity ororopharynx are usually carried to the back ofthe throat and swallowed . Particles depositedin the lower respiratory tract may be trappedin the mucus and borne upwards from thelungs to the back of the throat by ciliaryaction. However, smaller particles may reachthe lungs, those 1 µm or less in diameterreaching the alveoli, where they may cause

Source of sampleNumber ofpatients

Number ofspecimens

Positive

%Number

Impinger, nearmouth 29 47 5 II

Settling-plates, nearmouth 13 30 12 40

Circumoral swab 32 58 42 72Pillow swab 40 67 41 61Impinger, near

bedclothes 9 15 5 33Settling-plates, near

bedclothes 13 20 II 55Bedclothes swab 11 16 15 94Back swab 35 66 25 38Urine 16 34 17 50

infection of cells of the alveolar walls, oralveolar macrophages .

Size and physical state are also important indetermining whether expelled particles fallquickly to the ground or remain airbornefor a long time. Large particles-particularlyscabs-fall quickly to the ground. Smallerliquid particles dry rapidly and float for a longtime in the air, thus increasing the chance thatthey may enter the respiratory tract . Thevigorous shaking of heavily contaminatedbed-linen may disperse clouds of small par-ticles of infected "dust" in the air (Duguid &Wallace, 1948) .

The other critical factor in determining theinfectivity of both large- and small-particleaerosols generated by talking, coughing andsneezing is the concentration of virus in thesecretions of the nose and mouth. Experi-ments with bacteria showed that many of theexpelled particles were sterile unless theconcentration of bacteria in the secretionsreached 10 6 per ml or more (Duguid, 1946).

Organizations that some years ago wereconcerned with research on biological war-fare carried out a great deal of work onrespiratory infection with viruses, includingorthopoxviruses ; some investigations of thiskind have also been conducted in otherlaboratories (Noble & Rich, 1969) . Rel-evant publications on the subject deal withtwo systems : rabbitpox virus in rabbits(Westwood et al ., 1966 ; Lancaster et al., 1966 ;Thomas, 1970a) and variola virus in monkeys(Hahon & McGavran, 1961 ; Noble & Rich,1969). These experiments were described inmore detail in Chapter 3 ; their importance inthe present context is the demonstration thatsusceptible laboratory animals can be readilyinfected by aerosols of orthopoxvirions . How-ever, neither in smallpox nor in the exper-imental infection of animals with orthopox-viruses has evidence of a primary lesion beenfound in either the oropharynx or the lowerrespiratory tract ; it seems likely that infectioncould occur at any site from the nasal mucosato the alveoli, but that no "local lesion"developed.

Inoculation through the skinFor the genus Orthopoxvirus as a whole, the

skin is a relatively common portal of entry,usually through breaches in the surface,which may be minute. It is the usual portal ofentry of ectromelia virus in mice (Fenner,1947b), as well as the common route of

4. EPIDEMIOLOGY

187

infection with cowpox virus in both cows andhumans, and was deliberately used for induc-ing vaccinia infection in man-i .e., vaccina-tion against smallpox . Infection through theskin with variola virus occupied a specialposition in the history of smallpox . Acci-dental inoculation with variola virus oc-curred in certain occupational groups-e .g.,among hospital personnel or mortuary atten-dants dealing with unrecognized smallpox .However, a much more common situation forcenturies in India and China, during the 18thcentury in Europe, and up to recent times insome parts of Asia and Africa (see Chapters 6,14 and 21), was the practice of variolation, or"inoculation" as it was called before theintroduction of vaccination (Miller, 1957) . InChina a "snuff" of dust containing variolavirus was originally used, but in other coun-tries, and in China since the early 19thcentury, the virus was introduced into theskin .

Inoculation smallpox showed importantdifferences in pathogenesis and symptomsfrom smallpox acquired by the respiratoryroute (see Chapters 1, 3 and 6) . In the case ofthe former, there was always a local lesion atthe inoculation site ; "daughter pustules"often developed in the skin near the inocula-tion site ; the generalized rash appeared two orthree days earlier and the course of the diseasewas usually much less severe than in naturallytransmitted smallpox. However, oropharyn-geal lesions occurred as part of the generalizeddisease and these lesions, in particular, consti-tuted sources of infection of contacts .

The Infectious Dose of Variola Virus

During the 1940s and 1950s substantialexperimental work was undertaken to deter-mine the number of particles of vaccinia virusneeded to infect selected cells or animals byvarious routes of inoculation . It was foundthat for a highly susceptible animal or cellsystem, a single viable virus particle couldinitiate infection (Parker, 1938), althoughusually the ratio of particles to "takes" wasmuch more than 1 to 1 (Sharp, 1965) . Incontrast, in man a relatively high concentra-tion of vaccinia virus was needed to produceconsistently successful primary vaccinationtakes . It was recognized that only a very smallproportion of the virus in the inoculum wasactually introduced into the epithelium byscarification or multiple puncture vacci-

1 88

SMALLPOX AND ITS ERADICATION

nation. However, in assays by scarification, the50% infectious dose was found to be 3 x 10 5pock-forming units per ml for the Listerstrain (Cockburn et al., 1957) and about 10 7pock-forming units per ml for the attenuatedCVI strain (Cherry et al ., 1977) .

Man was the natural host of variola virus,and a priori it might be expected that a singleviable virus particle, lodged in an appropriatesite, could be infectious, although because ofnon-specific protective mechanisms a largerdose would usually be required . However,assays could obviously not be carried out andthis must remain a supposition based onanalogy with vaccinia virus in highly suscep-tible hosts .

The form in which the infectious virus waspresented to susceptible cells was a moreimportant factor than the actual viral contentin determining whether virus in variousexcretions or secretions would cause infec-tion. For example, when variola virions werepresent in fresh oropharyngeal secretions,there was a high probability that unvacci-nated subjects, exposed by inhalation or directcontact, would be infected . On the otherhand, much larger amounts of infectious virus(as judged by laboratory assay of scab extracts)enclosed within inspissated pus in scabsusually failed to produce infection. Suchfragments were handled as foreign bodies bythe mucociliary cleansing apparatus of therespiratory tract and excreted or swallowed,without causing infection .

The Incubation Period of Smallpox

It was difficult to measure the incubationperiod accurately in situations in whichsmallpox was endemic, for exposure usuallyoccurred over a period of several days . How-ever, precise observations were possible in im-portations of smallpox into Europe. Downie(SE/72.3) summarized data from 16 outbreaksof variola major and 2 outbreaks of variolaminor, occurring in several European coun-tries between 1927 and 1960 and involving898 cases. Among these the exact incubationperiod (from a single exposure up to the 1stday of fever) could be determined in 83 cases .Mack (1972) investigated 49 importations ofsmallpox into Europe over the period 1950-1971 and published histograms of the incuba-tion periods (to the onset of symptoms) forindividuals whose possible exposures ex-tended over 1, 2 or 3 days. Subsequent to these

so-

40-

v

" 30-0vE 20-

z

EJ Variola majorQ Variola minor

8 9 10

II

12

13

14

15

16

Incubation period (days)17 18 19

Fig . 4 .7 . The incubation period of variola major andvariola minor, from day of exposure to the onset ofsymptoms. (Based on data from SE/72.3, Downie ;Mack, 1972; and WHO/SE/73.57, Litvinjenko et al .)

analyses, the 1972 epidemic in Yugoslavia(WHO/SE/73.57, Litvinjenko et al . ; Stojko-vic et al., 1974) provided data for the calcula-tion of the incubation period in 171 personswith variola major. Fig. 4.7 is a histogrambased on the information provided by Mackfor 1-day exposures in outbreaks occurringover the period 1950-1971, on Downie's dataprior to 1950, and on data from the Yugoslavoutbreak . In all cases the incubation periodwas taken as the interval between exposureand the onset of symptoms, rather than thetime of appearance of the rash, which wasusually 2-3 days later.

These precise data support the widelyexpressed opinion that the incubation periodwas usually 10-14 days but, in rare instances,as short as 7 or as long as 19 days . Downieconcurred with the observation that acutefulminating cases of smallpox, especially hae-morrhagic-type smallpox, may have had ashorter incubation period than the average,but he did not believe that previously vacci-nated individuals had unusually long incuba-tion periods, as some writers have suggested .

Variola minor appeared to have had essen-tially the same incubation period as variolamajor.

The incubation period in inoculation va-riola (variolation) was about 2 days shorterthan in contact infection-i .e., 8 or 9 days tothe onset of fever ; or 5 days if the productionof a local skin lesion at the inoculation sitewas taken as the first sign of disease (Dims-dale, 1767).

The Infectivity of Cases of Smallpox

Smallpox patients, especially those with asevere rash and enanthem, were surroundedby their infected oropharyngeal secretionsand later by scabs, both of which contami-nated their skin, bedclothes and to a lesserextent the air in their immediate vicinity .Susceptible individuals could be infected bycoming into direct contact with the patientand his contaminated bedclothes or by inhal-ing infective droplets ; or, more distantly,laundry workers could sometimes be infectedafter handling contaminated bed-linen . Veryrarely, longer-range airborne infection oc-curred, when the contaminated dropletsoriginating in the mouth and pharynx weresmaller and more numerous, because thepatient coughed or sneezed, and air currentstransported infective droplet nuclei awayfrom the immediate vicinity of the patient.

Period of infectivityThe maximum infectivity of cases of small-

pox was during the 1st week of rash, corre-sponding to the period when the lesions of theenanthem had ulcerated and were releasingvirus into the secretions of the mouth andpharynx (see Table 4.8) . At this stage the skinlesions were intact ; the large amounts of viruslater shed from the skin were not highlyinfectious because of the physical state of thevirus particles, enclosed within hard dry scabs .For example, Mack (1972) found that the vastmajority of cases infected after contact withcases of smallpox who had arrived in Europefrom endemic countries while incubating thedisease occurred within 3 weeks of the initialtime of exposure (Fig. 4.8) . In the hospitalenvironment, the mean interval between thepoint at which a case first became infectiousand the onset of symptoms in contacts was inthe same range as estimates of the incubationperiod. In households, transmission tookplace a few days later, but in both situationsthe great majority of cases appeared within 3weeks of the initial exposure.

It was difficult to obtain evidence of theinfectivity of patients during the latter part ofthe incubation period or during the pre-eruptive fever ; one series of assays of theoropharyngeal secretions of case contacts (seebelow) suggested that occasionally virus waspresent before the eruptive stage. However,epidemiological experience suggested thattransmission very rarely occurred before the

4. EPIDEMIOLOGY

5 Family contacts

189

Fig . 4 .8 . The interval between the first possibleexposure to a case of smallpox imported into Europeby air and the onset of symptoms in first generationindigenous cases, in family and hospital environments .(Based on Mack, 1972 .)

1st day of rash. For example, in rural PakistanMack et al . (1972b) found no evidence ofsignificant transmission during the pro-dromal period. Theoretically, cases remainedinfectious until the last scab had separated(usually from the soles of the feet) but thelevel of infectivity fell off greatly as thelesions of the enanthem healed during thelatter part of the 2nd week of the disease .

The relative infectivity of different clinical types ofsmallpox

Two distinct variables were involved indetermining the relative infectiousness ofdifferent cases of smallpox : the quantity ofvirus excreted in the oropharyngeal secretionsand the number and extent of the face-to-facecontacts between the patient and susceptiblepersons, especially during the period whenlarge quantities of virus were being excreted .Cases with a severe rash and enanthem weremore infectious than those with a slight rashand enanthem, which was why most cases ofmodified-type smallpox were not highly in-fectious . However, cases with early severetoxaemia-that is, most cases of variola majorin unvaccinated persons-were usually con-fined to bed during the prodromal stage andwere thus segregated from the community .On the other hand, although patients withmodified-type variola major excreted lessvirus, they often moved about the communityrelatively freely and were thus in contact with

o

nn

nn~~0

7

14

21

>24

10] Hospital contacts

w

n10

.0

7

14

21

>24Days

1 90

SMALLPOX AND ITS ERADICATION

Table 4 . 10 Frequency of transmission of smallpox to household contacts in villages in Pakistan by casesoccurring in unvaccinated and vaccinated subjectsa

All contacts

a Based on Heiner et al. (1971 b) .

many more persons. Such cases were fre-quently blamed for importations into theUnited Kingdom in the period after theSecond World War, and in the 1972 outbreakin Yugoslavia 2 mild ambulant cases, notrecognized as smallpox, generated 11 and 16secondary cases respectively (Stojkovic et al .,1974). In general, however, persons who wereinfected despite previous vaccination wereonly one-fourth as likely as unvaccinatedcases to transmit infection to householdcontacts (Table 4 .10).

Rao (1972) found that the most highlyinfectious cases in Madras were those ofordinary- and flat-type smallpox in unvacci-nated subjects. Studies in rural Bangladesh(Thomas et al., 1971b) showed that patientswho died had been more infectious to familycontacts during their illness than those whosurvived, and in rural Pakistan Mack et al .(1972b) found that while almost all suscep-tible persons within the relatively confinedenvironment of compounds were infected,regardless of the severity of the index case,spread between compounds was much greaterif the index case was severe .

Because the toxaemia was so much less invariola minor, cases often remained ambula-tory during their period of highest infectivityand were thus likely to spread smallpox in thecommunity, a feature which partially explainsthe persistence of endemic variola minor inBrazil, Great Britain and the USA aftervariola major had been eliminated from thosecountries.

Absence of recurrent infectivity

Another important epidemiological fea-ture of smallpox was that recovery from anacute attack was followed by elimination ofthe virus ; recurrence, with excretion of virus,never occurred (see Chapter 3). The epidemi-ological importance of this aspect of thepathogenesis, in relation to eradication of thedisease, can be judged by comparing thissituation with the recurrent infectivity found

Vaccinated contacts

in varicella-zoster and herpes simplex infec-tions, which accounts for the survival of thesediseases in remote and isolated populations(Black et al., 1974) .

The Infectivity of Case Contacts notExhibiting Disease

Epidemiological observations before andduring the eradication programme gave riseto a widely held belief that subclinical cases,patients still incubating the disease, andcontacts of cases did not constitute importantsources of infection . Although infection wassometimes associated with handling inani-mate objects (fomites) and, rarely, seemed tobe airborne over a considerable distance (seebelow), the vast majority of cases of smallpoxcould be traced to face-to-face contact of asusceptible person with a patient with overtdisease, usually during the 1st week of rash .However, careful laboratory studies by Sarkaret al. (1 973b, 1974) showed that about 10% ofhousehold contacts of cases of smallpox har-boured detectable amounts of variola virus intheir oropharyngeal secretions . Only about10% of such carriers subsequently developedsmallpox. These authors cautioned against tooready an acceptance of the view that asymp-tomatic carriers were not a significant factorin the epidemiology of smallpox .

Looking at the situation after smallpox hasbeen eradicated, it seems probable that infec-tion might occasionally have occurred fromcase contacts or those incubating the disease .Such events might have been responsible forwhat appeared to be abnormally long or shortincubation periods. Nevertheless, the opinionof epidemiologists engaged in the globaleradication campaign is that the vast majorityof cases could be traced to close contact with apatient with an overt smallpox rash ; had itbeen otherwise eradication of the disease frompopulous countries such as Bangladesh andIndia would have been much more difficultthan proved to be the case .

Status ofIndex case Number Contracted

smallpox Number Contractedsmallpox Vs

Unvaccinated 390 83 21 .3 271 IS 5 .5Vaccinated 52 3 5.8 48 I 2 .1

4. EPIDEMIOLOGY 191

Use of the Word "Contact" in Epidemiology

To clarify the discussion which follows it is useful to review the definitions used byepidemiologists for infection via the oropharynx or respiratory tract . Differentconventions exist in different countries . In some countries the word "contact" is used onlywhere true physical contact has occurred . In most English-speaking countries the word isused more broadly (Langmuir, 1973), and this was the common usage by epidemiologistsengaged in the Intensified Smallpox Eradication Programme . According to thisconvention, infection which occurs only or mainly at short range, owing to theimplantation of infectious droplets expelled from the mouth or nose of the patient duringtalking, coughing or sneezing, is said to occur by contact, even though physical contactdoes not occur. Thus "contact" includes physical contact either direct or via the fingers, aswell as the direct implantation of large-particle aerosols on the oral, nasal or pharyngealmucosae or the inhalation of small particles into the alveoli, as long as the infector andinfected are in face-to-face contact .

If infection occurs at longer ranges, between persons not in sight of each other, it isregarded as being caused by airborne "droplet nuclei"-the dried particles that result fromthe evaporation of water from droplets that remain suspended in the air for a period. This iscalled "airborne infection" .

However, the most common use of the word "contact" in this book is to signify a personwho has been exposed to the risk of infection with smallpox, whether as a member of ahousehold with an infected case, as a casual visitor to a hospital ward or as a co-traveller inan aeroplane, bus or train .

TRANSMISSION

The shedding of virus from an infectedsubject and its transfer to a susceptible persontogether constitute the transmission of small-pox. Transmission was ordinarily direct, bythe implantation of infective droplets on tothe nasal, oral or pharyngeal mucous mem-brane, or the alveoli of the lung, or lesscommonly indirect, as an airborne infectionor from fomites .

Contact Infection

Contact infection could result from theinhalation of either large-particle aerosoldroplets or droplet nuclei at close range orfrom transfer via the fingers or variousobjects .

Epidemiologically, the important questionwas the relative frequency of infection of veryclose contacts of the patient and infection, byaerosol, at short distances, such as occurs intuberculosis (e .g ., schoolroom epidemics) andinfluenza. Comparison of the intrafamilialand extrafamilial spread of smallpox (e.g., Raoet al., 1968a ; Thomas et al., 1971 a,b, 1972)demonstrated that the overwhelming ma-

jority of secondary infections occurred - inclose family contacts of overt cases of small-pox, especially in those who slept in the sameroom or the same bed . Next in frequency werethose who lived in the same house ; residentsof other houses, even in the same compound(who would often have visited the house ofthe patient), were much less likely to becomeinfected.

Negative evidence was also important .Especially in India, long-distance movementsby train or bus of patients suffering fromsmallpox, with an overt rash, used to occurfrequently, yet infection of casual fellow-travellers was rare indeed-so rare that in-stances of it were deemed worthy of specialreport (e.g., WHO/SE/72.41, Suleimanov &Mandokhel).

Indirect Transmission

Rather than being propelled direct fromthe patient on to the oropharynx or nasalmucosa of a susceptible contact or transferredby contaminated fingers or various objects(contact infection), infective particles mayreach the same sites by indirect routes, aftertravelling considerable distances either in the

1 9 2

air or on fomites . Because experience hadshown that most patients had been infectedby face-to-face contact, indirect transmissionwas implicated only after all possibilities ofsuch contacts had been eliminated .

Airborne infection

There is good epidemiological evidencethat some viral infections, usually thoseassociated with coughing and sneezing, arespread by aerosol over short distances (Cox-sackievirus disease : Couch et al., 1970 ; influ-enza), while a few, notably foot-and-mouthdisease, may be spread by the wind overdistances of several kilometres (Donaldson,1979 ; Gloster et al ., 1982) . Towards the end ofthe 19th century there was a great controversyin Great Britain about the siting of smallpoxhospitals, based on the view propounded bysome (e.g., Power, 1886 ; Buchanan, 1905)that infection could be carried by aerial spreadfor considerable distances from such hospitals .Collie (1912) and Dixon (1962), in reviewingthe evidence, pointed out the difficulty ofproving the occurrence of this type of spreadin situations in which smallpox was endemicor not under control ; mechanisms other thanaerial spread could have been responsible inmost instances. However, Christie (1980)describes a particular case which he claims isbest explained by aerial spread from a hospitalsituated 400 yards (about 365 metres) away .

In India also it was sometimes said that thehospital was the focus from which the wind-borne spread of smallpox occurred . Hospitalswere usually located in densely populatedareas inhabited by persons of low income,whose vaccination status was often poor .Careful investigation showed that outbreaksof smallpox in these communities were asso-ciated with importations from outside thearea-e.g., from distant construction sites-and not with spread from the hospital . InMadras, A . R. Rao (personal communication,1981) noted that the hospital was surroundedby a zone relatively free of smallpox, probablybecause of more efficient vaccination in thatlocality. Further, Rao (1972) noted that,although wards for infectious diseases otherthan smallpox and corridors used by patientswith such diseases were located within a fewfeet of the smallpox wards of the MadrasInfectious Diseases Hospital, only 7 cases ofsmallpox presumably infected in the hospitaloccurred among 130 000 non-smallpox pa-tients over the 10 years 1959-1968 .

SMALLPOX AND ITS ERADICATION

Nevertheless, airborne infection over shortdistances did sometimes occur . Two hospitaloutbreaks in the Federal Republic of Ger-many, at Monschau (Anders & Posch, 1962)and Meschede (Wehrle et al., 1970), seemcertainly to have been airborne. In the Me-schede outbreak an electrician who had justflown back to the Federal Republic of Ger-many from Karachi, Pakistan, and who, ittranspired, had never been successfully vac-cinated, was admitted 10 days later to theisolation ward of a large general hospital witha feverish illness that was suspected to betyphoid fever. He was confined to his room,and 3 days after admission developed a rash .Smallpox was confirmed by electron micro-scopy 2 days later and the patient was thentransferred to the smallpox hospital . In spiteof rigorous isolation of the patient (because ofthe suspicion of typhoid fever) and, aftersmallpox was suspected, the vaccination of allpatients and nurses in the general hospital (inseveral cases with inactivated vaccine), orinoculation with vaccinia-immune globulin,19 further cases of smallpox occurred there,on all three floors of the building in which theindex case had been nursed. The dates of onsetof these cases are shown in Fig . 4.9A. Seven-teen cases occurred within one incubationperiod, counting from the 3 days duringwhich the index case was in the hospital andinfectious ; the last 2 were room contacts ofearlier cases, as indicated . The locations ofthese patients within the hospital are given inFig . 4 .913, which also shows the results of testson the carriage of smoke through the hospital .A number of circumstances favoured theairborne transmission of variola virus in thisepisode

(1) the index patient had a densely conflu-ent rash and severe bronchitis and cough ;

(2) the relative humidity in the hospitalwas very low, a situation that promotes thesurvival of vaccinia virus (Harper, 1961) andpresumably therefore of variola virus ; and

(3) the design of the hospital led to theraising of strong air currents when thebuilding was heated, as it was during thewinter, when the episode occurred .

A visitor to the hospital during the periodin question (Case No . 8) spent only 15 minutesinside the hospital, away from the patient careareas and the isolation unit corridor, but henevertheless developed smallpox with onsetof fever 11 days after this visit . Case No. 15was a nurse who was located on the top floor ;

4-w

U3-

I

Presumed infectiousperiod of index case while

at Meschede Hospital18

3 2 8 4 7

10

12

14

16

18

20

22

24

26January

® Case number

Stairs

Smoke pattern

9

she had not left her sick-room while the indexcase was in hospital nor had she been incontact with persons other than those in-volved with her case, but she contracted thedisease, which manifested itself after anincubation period of about 16 days. Case No .20 was infected after contact with thispatient .

There is no reason to doubt that airborneinfection, sometimes over considerable dis-tances, did occur, but it appears to have been

4. EPIDEMIOLOGY

12

II

10 14

28

30

17

16

5

13

3

5

7

9

II

13February

19iI5

20

193

17

19

Fig. 4 .9 . Airborne spread of smallpox in the Meschede Hospital, Federal Republic of Germany, in 1970 . A :Sequence of infections . The numbers in the histogram indicate case numbers, representing patients who werelocated as shown below. No. I was the index case, No. 8 was a visitor who had spent no more than I5 minutesin a ground-floor office, and No . 19 and 20 were second generation cases who had been in contact with No . 17and 15 respectively. B: Floor plan and rear elevation of the hospital building, showing the location of all casesof smallpox . The shaded area indicates the movement of a cloud of smoke generated in the room in which thepatient who was the index case (No . I) had been nursed before being recognized as having smallpox . (Based onWehrle et al ., 1970 .)

very rare. Short-range aerosol infection wasprobably responsible for most non-familialcases apparently infected in hospitals, busesand trains . Many laboratory manipulations,such as blending, centrifugation and pipet-ting, can generate substantial clouds of aero-sols ; airborne infection from a laboratory issuspected to have been the cause of the lastoutbreak of smallpox in the world, in Bir-mingham, England, in August-September1978 (Shooter, 1980 ; see Chapter 23) .

1 9 4

SMALLPOX AND ITS ERADICATION

Infection via fomites

The clothing and bed-linen of cases ofsmallpox were heavily contaminated withvirus derived from oropharyngeal secretionsand later from ulcerated pustules or scabs (seeTable 4.9) . Such objects sometimes served as asecondary source of infectious virus for casecontacts. There are several documented in-stances of the infection of laundry personnelwhose work-places were located at somedistance from the homes or hospitals in whichsmallpox patients had been nursed, and ofchambermaids in hotels (Stallybrass, 1931) .Dixon (1962) has summarized reports oninfection via fomites and commented that forlaundry workers the risk of infection wasgreatest among those who sorted the in-coming laundry and could thus inhale con-taminated dust ; once wetted the garmentsand bed-linen were not infectious .Rao (WHO/SE/72.40) carried out experi-

ments on the persistence of infectivity ofvariola virus on various objects that had beendeliberately contaminated . The most impor-tant finding was that, in Madras, virus onsuch objects was rapidly inactivated, evenwhen they were heavily contaminated ; heconcluded that fomites were of little import-ance in the transmission of smallpox in India .In temperate climates virus in scabs couldsurvive on fomites such as cotton for longperiods of time (MacCallum & McDonald,1957) ; however, virus in scabs rarely appearsto have been a source of infection .

As might be expected, the corpse of a fatalcase of smallpox and the associated clothingremained heavily contaminated, and smallpoxsometimes occurred as an occupational diseaseamong pathologists and mortuary attendants,particularly from unrecognized cases of hae-morrhagic smallpox (Dixon, 1962). In ende-mic countries cases were sometimes associatedwith funerals (Hopkins et al ., 1971 c), but thesecould have been due to exposure by contactwith unrecognized cases as often as to infec-tion from the corpse or clothing . From timeto time other fomites, such as letters, havebeen suspected, but incriminating evidencewas hard to obtain and the subject is ofhistorical interest only ; Dixon (1962) pro-vides a summary .

Another purported association of smallpoxwith infection from inanimate objects ismentioned only to be dismissed . This relatesto folklore in several parts of Africa and Indiawhich held that European missionaries who

Plate 4 .3 . Cyril W . Dixon (b . 1912) . Worked in theDepartment of Preventive Medicine at the Universityof Leeds, England, in 1947-1959, and was Professorof Preventive and Social Medicine at the Universityof Otago, New Zealand, from 1959 to 1977 . He wasthe author of the standard textbook on smallpox inthe English language, which was published in 1962 .

occupied a house in which smallpox hadoccurred months or years before sometimescontracted the disease, supposedly because ofresidence in that house . However, in suchendemic areas there were always opportuni-ties for infection by contact with active casesof smallpox . During the Intensified SmallpoxEradication Programme repeated efforts weremade to investigate such stories, but noinstances were found in which infectionappeared to have been caused by residence in a"contaminated" house .

FACTORS AFFECTING THE SPREADOF SMALLPOX

Individual Susceptibility

In the absence of immunity due to vaccina-tion or other prior infection with an ortho-poxvirus, human beings appeared to be uni-

versally susceptible to infection with variolavirus. By far the most important factoraffecting individual susceptibility was theimmunity provided by vaccination . Indivi-dual susceptibility to the effects of infectionin unvaccinated subjects was influenced bygenetic factors and physiological factors suchas age and pregnancy . The risk and severity ofexposure was influenced by certain occupa-tional factors .

Genetic factors

Although racial differences in susceptibil-ity probably did exist, they were never con-vincingly demonstrated . Inhabitants of India,a country subject to endemic variola major forsome two thousand years, remained highlysusceptible to the effects of the virus, case-fatality rates of over 20% among unvacci-nated individuals being common right up tothe time of eradication. However, it is impos-sible to read the accounts of smallpox amongthe indigenous inhabitants of the Americas(North America : Stearn & Stearn, 1945 ;Mexico : Crosby, 1967 ; Peru : Hemming,1970 ; Brazil : Hemming, 1978) without sus-pecting that the Amerindians, an unexposedpopulation when smallpox was first intro-duced into the Americas, were more suscep-tible than unvaccinated whites or Negroes .When variola minor was introduced intoNew Zealand in 1913 from the USA, therewere 114 reported cases among Europeans,with no deaths, and 1778 reported casesamong the local Polynesians (Maoris), with55 deaths (Dixon, 1962). People of the Maorirace had never before been exposed tosmallpox .

It seems likely that a disease as lethal assmallpox must have exerted some selectionfor more resistant genotypes within popula-tions in which it had been endemic forcenturies. Such selection was readily demon-strated among European rabbits exposed tothe poxvirus disease myxomatosis in as short aperiod as 10 years (Fenner & Ratcliffe, 1965),but this was initially a much more lethaldisease and it was possible to test rabbitsexperimentally to determine their geneticresistance. As noted in Chapter 3, the onlytests of genetic susceptibility that were con-ducted in man related to the possibility thatdeaths due to smallpox had some effect on thepresent-day distribution of genes of the ABOblood group system. The results reported wereunconvincing but do not exclude the possibil-

4. EPIDEMIOLOGY

195

ity that the selection of more resistant geno-types had indeed occurred in countries such asIndia . The linkage between cytotoxic T-cellactivity, which is important in the process ofrecovery from orthopoxvirus infections, andthe major histocompatibility antigens (HLAin man) suggests that the latter may have beenimportant in influencing resistance to small-pox (Chapter 3). However, very few relevantstudies were carried out and no definitive datawere ever obtained .

Age

The age incidence of smallpox dependedmainly on the acquired immunity of theexposed population, whether due to vaccina-tion, variolation or prior natural infection . Inthe absence of protective immunization, theage incidence reflected the level of endem-icity . Thus, when populations were firstexposed to smallpox, persons of all ages andboth sexes were affected . However, if small-pox was endemic, as in the larger cities ofEurope in the 17th and 18th centuries and inmodern India, it was mainly a disease ofchildhood . On the other hand it was lessexplicitly a children's disease than measles oreven chickenpox. In the Intensified SmallpoxEradication Programme it was found thatmore than one-quarter of the cases occurredamong the adult population, even in India(see Table 4 .2) and Bangladesh (Chapter 16,Table 16.17) .

Rural areas and small villages often escapedinfection for several years, and when thedisease was introduced it usually affectedmost of the large proportion of susceptiblepersons that had accumulated, thus causingmuch greater devastation in terms of econ-omic disruption than was the case in places inwhich the disease was always present .

Apart from special physiological factorssuch as pregnancy (see below), and in theabsence of immunizing infections with eithervariola virus or a related orthopoxvirus (e.g.,vaccinia or cowpox), the severity of smallpox,as indicated by the case-fatality rate, wasgreatest in the very young and the elderly (seeTables 4.2-4 .5) . The great susceptibility of thevery young is shown in all series of figures ; itwas more difficult to obtain clear-cut evi-dence of the level of resistance in elderlypersons because of the opportunity that theyhad of being vaccinated or of becominginfected with variola virus earlier in life .What is clear from all data is that case-fatality

1 9 6

rates in the age group 5-14 years were muchlower than in any other age group.

Physiological factorsApart from severe immunological defi-

ciency states (see Chapter 3, Fig. 3 .8), which,for most of history everywhere and in thecountries in which smallpox had been en-demic after 1967, would usually have led todeath from infection during infancy or earlychildhood, pregnancy was the physiologicalstate associated with the highest susceptibilityto severe disease and death . Rao's data (Rao,1972), summarized in Chapter 1, illustratewell the severity of smallpox in pregnantwomen ; other investigators report similarfindings (e .g ., Dixon, 1948) .

It is difficult to obtain data on the effects ofnutritional deficiencies on the severity ofsmallpox, apart from the suggestion (Dixon1962) that blindness due to smallpox occurredmost frequently in ill-nourished subjects .Epidemiologists working in India during theeradication campaign formed the impressionthat cases of smallpox were more severe anddeaths more frequent among ill-nourishedvillagers and refugees than in better-nour-ished subjects, but no reports have beenpublished which document a relationshipbetween malnutrition and smallpox compar-able to that claimed for measles (Gordon,1976) .

Occupational riskOccupational risk was related directly to

the possibility of unvaccinated or inad-equately vaccinated persons coming into closecontact with cases of smallpox . In Europe andthe USA in the period since the Second WorldWar, nurses in particular, and doctors to alesser extent, were often inadequately vacci-nated, although they could be exposed tomassive doses of virus when treating im-ported cases, especially cases of unrecognizedhaemorrhagic smallpox. Perhaps because ofthe effects of large doses of virus, the case-fatality rate in nurses exposed to importedcases was sometimes very high.

Social Factors

The spread of smallpox in a communitydepended not only on the biological factorsjust elaborated, the excretion of virus and the

SMALLPOX AND ITS ERADICATION

infectiousness of cases, the mode of transmis-sion and individual susceptibility, but also ona variety of social factors that affected theopportunities for susceptible persons to comeinto close contact with sources of infection,primarily overt cases of smallpox . Thesevaried from one society to another . In manycountries with well-marked dry and rainyseasons, the population was much less mobileduring the rainy season, and smallpox wasmore confined . In the dry season, travel tomarkets, religious festivals, fairs and similargatherings enhanced the opportunities forthe widespread seeding of cases into pre-viously unaffected communities . Another fac-tor, prevalent in Indonesia, was the habit ofexhibiting children infected with smallpoxand taking them on visits to relatives (seeChapter 13) .

In contrast to diseases such as leprosy,syphilis and gonorrhoea, no social stigma wasattached to smallpox, which lessened thetendency to conceal cases . Nevertheless, in anumber of communities in India and in partsof Africa, cases were hidden from the publichealth authorities because families distrustedthe local infectious disease hospitals, orbecause religious beliefs generated oppositionto vaccination, which would have been admin-istered to family contacts if a case had .been discovered. Such concealment causedsome of the most persistent outbreaks inregions otherwise free of smallpox-e.g., theFaith Tabernacle outbreak in Nigeria(WHO/SE/68.3, Thompson & Foege), andthe outbreak among the Mazezuru people inBotswana (WHO/SE/74.69, Presthus & Si-biya ; see Chapter 20). During national eradi-cation programmes, cases were sometimesconcealed by both local and national healthservice staff, the former because they fearedreprimands based on accusations of inade-quate vaccination and the latter for reasons ofnational pride. The reward system did muchto minimize concealment by local health staff.

Demographic Factors

The size and density of the population atrisk affected the chances of contact betweensusceptible and infectious persons, and there-by the extent and rapidity of the spread ofsmallpox. These factors have been betteranalysed with measles than with smallpox,but the same principles hold for both diseases,although the transmission of measles is more

20-

4. EPIDEMIOLOGY

197

Iceland. ,,-

0

I

2

3

4

5

6

7 1

\/input / km 2

Fig . 4 .10 . Relationship between population densityand the duration of epidemics of measles . The averagedistance, in kilometres, between newborn infantsadded to the respective populations each year is givenin terms of the inverse square root of the number ofnew susceptible individuals added to the populationper square kilometre . In island populations of roughlycomparable size, the duration of epidemics was in-versely proportional to the density of the population .(From Black, 1966 .)

efficient and rapid than was the spread ofsmallpox. Black (1966) has shown that inpopulations of about the same size, measlesspread more rapidly and outbreaks terminatedsooner in areas of greater population density(Fig. 4.10) .

During the smallpox eradication cam-paigns WHO epidemiologists found thatpopulation density, when translated into thedensity of the susceptible population-i .e .,the number of unvaccinated persons persquare kilometre-was a useful concept forassessing the difficulty of controlling small-pox by mass vaccination in different endemiccountries or different regions within acountry (Arita et al ., 1986) . The effects ofpopulation density on the number and den-sity of unvaccinated persons remaining aftervaccination coverage at various levels isshown in Table 4 .11. It is clear that avaccination coverage of 80%, which was oncerecommended by WHO as being adequate tointerrupt the transmission of smallpox, mayleave a population density of 100 unvaccinat-ed subjects per square kilometre or more incountries with a high population density,such as Bangladesh. This is very much higherthan the overall country-wide populationdensity of any African country . Arita et al.(1986) provide an example . In 1967 there was

Table 4.11 Effects of population density on thenumbers of unvaccinated persons re-maining per unit area (km 2 ), at variouslevels of vaccination coveragea

a Rounded to nearest Integer .

a vaccination coverage of 80 .8% in theMatlab Thana, an area of about 194 squarekilometres in Bangladesh. The populationdensity was 582 per km 2 , so that the density ofunvaccinated persons was 112 per km 2 . Inthat year 119 cases of smallpox were reported(and probably many more occurred) . This isnot surprising ; it would be highly unlikelythat smallpox could have been eradicatedfrom Nigeria (population density 54 per km 2 )or the Philippines (population density 116per km 2 ), for example, with no vaccination atall. Clearly, mass vaccination alone wasdoomed to failure in countries with highpopulation densities ; control could only beachieved when it was supplemented by avigorous campaign of surveillance and con-tainment (see Chapter 10) . Even in countrieswith low population densities, surveillanceand containment greatly accelerated theachievement of eradication .

The age structure of the population af-fected the average severity (and case-fatalityrate) of smallpox, since the very young, theelderly and pregnant women had particularlyhigh case-fatality rates . Males aged between 5and 20 years were important in extrafamil-ial spread, since they moved between familygroups more frequently and readily thanother segments of the population and oftensustained relatively mild infections, so thatthey were important both as "exporters" and"importers" of the disease into previouslyunaffected family groups . The destruction ofwhole tribes that occurred when smallpox wasfirst introduced into the populations ofAmerindians in the Americas and Hottentotsin South Africa (see Chapter 5) was due to acombination of the lethal effects of the dis-ease itself in what were immunologicallyand genetically "naive" populations, and the

Persons

Number of unvaccinatedpersons remaining whenvaccination coverage Is :

per km 2

20% 95%40% 80% 90%

500 400 300 100 50 25300 240 180 60 30 15100 80 60 20 10 525 20 15 5 3a 2a10 8 6 2 1 la

VE

a)72 10-0.

Vanuatu . N ew Caledonia

w

0

Samoa

'Solomon Islands•

French Polynesia0

Tonga

r)0

1 98

SMALLPOX AND ITS ERADICATION

social disruption and famine that ac-companied the simultaneous illness of manyadults of active age, as well as persons in otherage groups, in these subsistence societies .

Political and Economic Factors

The level of economic development ofcommunities generally determines the levelof health services (Fig . 4.11). The higher thelevel of economic development, the moreeffectively did surveillance and containmentprinciples apply and the earlier was variolamajor, in particular, eliminated from thecountry (see Chapter 8) .

Variola minor posed an unusual problem .Because of its low case-fatality rate comparedwith variola major and the lack of sequelae(facial pockmarks), it had long been toleratedas an endemic disease even in affluent coun-tries such as the United Kingdom and theUSA. In poor developing countries, such asEthiopia, variola minor ranked very low as anational public health problem, and in the1970s it was much more important for the restof the world than for those countries toeliminate the disease . This justified the muchlarger input of international manpower andfunds into the programmes in the poorercountries, in order to achieve globaleradication .

Variolation and Laboratory-AssociatedSmallpox

Although close contact with overt caseswas responsible for the vast majority of casesof smallpox, there were two other potentialsources of infection : variolation and thesmallpox laboratory. For many centuries andas recently as August 1976, some cases ofsmallpox originated from the practice ofvariolation (see Chapters 6, 14 and 21). Suchcases of inoculation variola themselves consti-tuted potential sources of infection for sus-ceptible subjects in close contact with them .

Until the 1970s, laboratories in endemiccountries and in countries liable to importa-tions conducted studies in the diagnosis ofsmallpox and often held stocks of variola virusto assist them in this work . Laboratory pro-cedures are associated with some risk of infec-tion both for those working with viruses thatare pathogenic for man (accidental injection,trauma) and for visitors to such laboratories,which may use procedures that generateaerosols . Of all the dangerous human patho-gens, variola virus was regarded by virologistsas among the safest with which to work in thelaboratory, since regular vaccination andrevaccination provided complete protectionto the workers involved . However, adminis-trative mistakes sometimes occurred and afew cases of laboratory-associated infection

I

Annual per capitahealth budget244.0 (US$)

Fig . 4 .11 . National levels of economic development, as determined by the United Nations on the basis of averageper capita income for the year 1982 . Average per capita expenditures on health in US$ are also shown .

were documented in non-endemic countries.It seems likely that cases may occasionallyhave occurred in endemic countries also, butthe source would have been more difficult toidentify with certainty. The importance ofthe laboratory as a potential source of infec-tion is discussed in Chapter 30 .

PATTERNS OF SPREAD INPOPULATIONS

In the earlier sections of this chapter theinfectivity of cases of smallpox and thesusceptibility of individual human beingswere discussed. The final section is concernedwith various aspects of the spread of smallpoxin populations, looking first at the situationin countries as a whole and then at smallcommunities, including institutions. Severalcase studies are analysed which describe theepidemiology of smallpox in rural and urbansettings .

Spatial Distribution

Secular trends in the occurrence of small-pox dealt with earlier in this chapter weredrawn from the numbers of cases reportedannually by different countries. The fact thatcases were grossly and continually under-reported almost everywhere meant that thesefigures greatly underestimated the incidenceof the disease . Further, by ascribing cases tocountries they fail to express the spatialdistribution of smallpox. Even when small-pox was essentially uncontrolled it was neveruniformly distributed within a country . An"epidemic year", for example, usually indi-cated a high incidence of the disease notthroughout a country, but in a number ofscattered villages and towns. Although small-pox remained endemic throughout the year inthe large cities of endemic countries, itoccurred as many small outbreaks in differentsuburbs and districts within the city . As themaps showing the falling incidence of small-pox in India from 1972 onwards illustrate(Basu et al., 1979), the distribution of thedisease by district was patchy, and it waspatchy even within affected districts .

Effective public health measures in non-endemic countries limited the spread ofsmallpox after importations, usually to thefamily and close contacts of the index case, butsometimes, as in Yugoslavia in 1972, unrecog-

4. EPIDEMIOLOGY

199

nized cases spread the disease more widelybefore its nature was recognized .

The Rate of Spread within Small Groups

The devastating effects of smallpox in non-immune populations in the Americas, south-ern Africa and the islands of the Pacific Ocean(see Chapter 5) gave rise to the view that itwas one of the most infectious of all diseases .It is true that very brief exposures of suscep-tible persons to a case did occasionally lead toinfection. For example, one case in theoutbreak at Meschede, described above, was aperson who had never been in face-to-facecontact with a case of smallpox and had beenexposed to possible airborne infection for aperiod of 15 minutes or less . The last person tocontract endemic smallpox in the world-theunvaccinated hospital cook Ali Maow Maa-lin-was exposed to the disease for a fewminutes only, while he directed a vehiclecarrying two active cases of smallpox from thehospital in Merca, Somalia, to the home of thelocal smallpox surveillance team leader (Deriaet al ., 1980) . Sometimes one index case infect-ed a dozen or more people, as in the outbreakin Yugoslavia in 1972 (see Chapter 23). Theseepisodes, however, were exceptional . Epide-miologists engaged in the global smallpoxeradication campaign in Africa, South Amer-ica and Asia agree with Dixon (1962) thatsmallpox usually spread rather slowly .

Attempts have been made to expressquantitatively the infectiousness (or conta-giousness) of different human diseases bycalculating in various ways the proportion ofcases among exposed contacts. The mostwidely used figure is the secondary attackrate-i .e ., the proportion of susceptible indi-viduals exposed to an index case within ahousehold who became infected within theexpected transmission interval of the disease .It is surprisingly difficult to obtain suitableseries of data to make this calculation ; oftenthe term "secondary attack rate" has been usedto include all subsequent attacks (first, secondand third generation cases) . Hope Simpson(1952) developed a measure that makes use ofresults of second and later generation attacksto provide a figure comparable to the firstgeneration secondary attack rate-what hecalled the susceptible-exposure attack rate .

Some figures for measles, chickenpox andsmallpox are set out in Table 4 .12. Thenumbers available for the calculation of the

2 00

SMALLPOX AND ITS ERADICATION

Table 4 .12 First generation secondary attack rates (or the equivalent) in measles, chickenpox and smallpox

a . . = not applicable; + = scar present ; - = scar absent.

secondary attack rate were often quite smalland the reported secondary attack rates variedwidely. For variola major, the overall averagesecondary attack rate was 58 .4% in unvacci-nated family contacts, and 3 .8 % in vaccinatedcontacts . Vaccination protected most closefamily contacts from overt disease (but notnecessarily from subclinical infection-seeHeiner et al., 1971a). The highest secondaryattack rates were those reported for Pakistanby Heiner et al. (1971a) and Mack et al .(1972a) . These workers suggested that one ofthe causes of this phenomenon might havebeen the dry meteorological conditions pre-vailing for most of the year in the Punjab,compared with other parts of the Indiansubcontinent. On the basis of secondaryattack rates in susceptible subjects, smallpoxappears to have been somewhat less infectiousthan either measles or chickenpox, since thefigures for chickenpox exclude subclinicalinfections, which occur in at least 5% ofvaricella virus infections but are extremelyrare in unvaccinated persons infected withvariola virus.

This comparison of the secondary attackrates of smallpox and other common infec-

tious diseases fails to take cognizance of animportant feature of variola major, whichsubstantially decreased the rate at which itspread outside the household . Patients in theprodromal stage of variola major, before therash had appeared and before they couldtransmit infection, were usually quite ill, withtoxaemia, headache and backache . Most tookto bed, so segregating themselves from thegeneral community, although not from theirhousehold contacts . In contrast, patients inthe early and highly infectious stages ofchickenpox and measles have few symptomsand are usually mobile, and thus spreadthese diseases to school and street contacts aswell as to members of the household . Al-though its inherent transmissibility was prob-ably lower than that of variola major, becauseless virus was excreted in the oropharyngealsecretions, variola minor resembled chicken-pox in the mobility of infectious patients,which partly accounts for its persistence inmany countries after variola major had beeneliminated .

Another measure of infectiousness is theduration of outbreaks in particular com-munity units . There are many records in the

Disease andlocality

Vaccination

(for smallpox)

scar

Total numberof household

contacts

Contacts whodeveloped disease Reference

Number

Measles

. . 266 201 75 .6 Hope Simpson (1952)

Chickenpox

. . 282 172 61 .0 Hope Simpson (1952)Chickenpox

. . 888 771 86 .8 Ross (1962)

Variola minorBrazil 38 20 52 .61Brazil

+ 56 8 14 .3Angulo et al. (1967)

Brazil 674 466 69 .1Brazil

+ 204 7 3 .4Suzart de Carvalho Fllho et al . (1970)

Variola majorNigeria 27 12 44 .4Nigeria

+ 45 12 26 .2Foege et al . (1975)

Benin 17 8Benin

+ 13 247.0115 .4

Henderson & Yekpe (1969)

Madras 103 38 36 .91Madras

+ 1 146 14 1 .2Rao et al . (1968a)

Pakistan 45 33 73 .31Pakistan

+ 190 6 3 .2Heiner et al. (1971a)

Pakistan 22 10 45.51Pakistan

+ 338 3 1 .3Heiner et al. (1971b)

Pakistan 43 38 88 .41Pakistan

+ 180 13 7 .2Mack et al . (1972a)

Calcutta 80 61 76 .31Calcutta

+ 661 47 7 .1MukherJee et al. (1974)

Bangladesh 21 9Bangladesh

+ 57 442.917 .0

Thomas et al. (1971 b)

Average 58 .4(variola major)

+ 3.8

literature of the slow spread of smallpox-forexample, the outbreak of variola minor inNew South Wales (Australia) in 1913-1917(Cumpston & MacCallum, 1925 ; see Chapter8) and among nomads (see below) . Smallpoxsometimes took several generations of infec-tion to spread through quite small popu-lations .

Spread in Institutions

Because of the importance of face-to-facecontact in the transmission of smallpox, thehousehold or family unit was by far the mostfrequently affected group . Nevertheless, someinstitutions, especially hospitals and schools,and public events, such as fairs and religiousfestivals, played an important part in thedissemination of smallpox .

HospitalsThe role of hospitals was particularly

significant in amplifying outbreaks of im-ported smallpox, for undiagnosed and mis-diagnosed cases were often admitted intogeneral wards, and the proportion of unvac-cinated or poorly vaccinated hospital person-nel and patients was usually high (Millar,1965). Over half the cases that occurred afterimportations of variola major into Europebetween 1950 and 1971, and all the largeoutbreaks, were to be found among personsassociated with hospitals, either occupation-ally or as patients or visitors (Mack, 1972) . Anoutbreak in Glasgow in 1950 (Laidlaw &Horne, 1950) was characteristic. An Asianseaman with 4 vaccination scars and a historyof revaccination 3 years earlier was admittedto hospital with what was regarded by anexperienced consultant as chickenpox, butwas in fact modified-type smallpox. Thirteenof the 18 cases which ensued were infected inhospital, including 10 members of the hospi-tal staff. Six died, all of whom were unvacci-nated or had failed to respond to vaccinationafter exposure.

The experience in the epidemic in Yugosla-via in 1972 (Stojkovic et al ., 1974) wasparticularly dramatic : an undiagnosed case ofhaemorrhagic-type smallpox infected 2 con-tacts in a bus and then a total of 36 persons in 3hospitals to which he was admitted .

Hospitals sometimes played a role as ampli-fiers in countries with endemic smallpox.Almost everywhere, they were important

4. EPIDEMIOLOGY

Imported case

-----------------

First generation

Secondgeneration

Thirdgeneration

Fourthgeneration

0 Infected in Fever Hospital

Infected in Sabah Hospital

History of contact unknown

Fig . 4 .12 . The pattern of transmission of smallpox inan outbreak in Kuwait in which two hospitals werethe principal foci of infection . (From Arita et al .,1970 .)

disseminators of smallpox. Obviously, if allstaff were regularly vaccinated and patientsand visitors vaccinated on entry-as ideallythey should have been-the risk of spreadwould have been very low . However, with afew exceptions, such as the Infectious Dis-eases Hospital in Madras (Rao, 1972), this didnot occur. Two outbreaks exemplify the risk .

Kuwait is a small country which in 1967had been free of endemic smallpox for theprevious 10 years, but which was constantlyexposed to the risk of importations fromnearby endemic countries. Smallpox misdiag-nosed as chickenpox spread in the FeverHospital over a period of 2 months before asecond generation fatal case was correctlydiagnosed, after it had caused 9 third genera-tion cases. One of these, unrecognized, wastransferred to another hospital (Sabah), whereit produced several more cases (Fig . 4.12) .Twenty-one of the 39 cases were in childrenunder 4 years of age ; only 8 patients, 6 of themaged 15 years or more, had ever been vaccin-ated. In contrast to the usual situation inEurope, all cases, except one vaccine-modi-fied attack, occurred among patients rather

∎

∎

∎∎∎∎∎∎∎∎∎∎on

∎

∎

∎∎EM∎//

NAMEMEMOa

--------------

-------------

-------------

201

2 02

than nurses or other hospital personnel,whose immune status had been maintained byregular revaccination (Arita et al ., 1970) .

An outbreak in a children's hospital inBrazil described by Morris et al . (1970a)demonstrates clearly the importance of thehospital as a potential focus for transmissionin countries in which variola minor wasendemic. The epidemic had been smoulderingin the community and the hospital for at least10 months before it was extinguished and 40of the 51 cases that occurred in the hospitalduring 1967 were certainly or possibly (4cases) hospital-associated . Over the period,there were at least 11 and possibly 15 separateintroductions of smallpox into the hospital,which formed a continuous source of infec-tion because visiting was not restricted andsmallpox cases were treated in the outpatientclinic. Only one hospital employee was in-fected-a kitchen worker who had never beensuccessfully vaccinated .

Schools

Because of their accessibility, schoolchil-dren in endemic countries were usually wellvaccinated. Unvaccinated children in theinfectious stages of variola major were usuallytoo ill to attend school. Even with variolaminor, contacts between children in thehousehold or, to a lesser extent, in play-grounds appeared to have been more impor-tant than their contacts at school (Angulo etal., 1968) . Space-time cluster analysis ofvariola minor within two schools suggestedthat transmission was most likely to occurfrom infected children to susceptible childrensitting next to them (Klauber & Angulo,1974a) ; studies in schools when two or threeshifts of pupils occupied the same classroomeach day revealed that contaminated desks orother fomites played no part in spread(Klauber & Angulo, 1974b) .

In a study in Mali, Imperato (1970) foundthat smallpox spread much more rapidly andextensively among schoolchildren thanamong equally inadequately vaccinated chil-dren of the same age who did not attendschool. The epidemic was confined to a groupof straw buildings in which there was severeovercrowding and did not spread to childrenin less crowded cement-block buildings in thesame school compound . However, even in thestraw buildings transmission, mostly betweensusceptible and infected persons in close

SMALLPOX AND ITS ERADICATION

contact, was slow and went on for a long timebecause of the absence of an effective responseby the local health personnel .

Markets, fairs and religious festivals

Markets are regular institutions at whichhouseholds mingle ; fairs and religious festi-vals, although held less frequently, drawattendance from much wider areas . Both areimportant as mechanisms for the wide dis-semination of infectious diseases, because ofthe dispersion of participants during theincubation period . Great international reli-gious festivals, such as the pilgrimage toMecca, are kept under strict medical surveil-lance and appeared to have played only aminor role in the international spread ofsmallpox during the latter part of the 20thcentury, although it seems likely that someoutbreaks did occur but were concealed . Thenational religious festivals in India werealways important for the dissemination ofsmallpox, and in the later stages of theeradication campaign they served as conven-ient observation posts for sampling the vacci-nation status and awareness of smallpox invillage populations (see Chapter 15).

On rare occasions funeral ceremonies pro-vided an opportunity for the spread ofsmallpox. When death due to unrecognizedsmallpox occurred, relatives, friends andother associates who were obliged to attendthe funeral could be exposed to a variety ofsources of virus (Hopkins et al ., 1971c) . Likemarkets or fairs, funerals were importantbecause the participants subsequently dis-persed and could then produce secondarycases among widely separated family groups.

Dispersal by Travellers

Because of the long incubation period ofsmallpox, travellers of various kinds-mi-grant workers, nomads, tramps, bus and trainpassengers, and air travellers-could coverlong distances while apparently healthy andintroduce the disease into areas far removedfrom the place in which they had acquired theinfection. The situation was dramaticallyillustrated by importations into Europeancountries by air travellers (Hagelsten & Jes-sen, 1973 ; see Chapter 23), but the movementof train and bus travellers in endemic coun-tries was no less important . Fellow-travellers

were sometimes infected when ambulantpatients, perhaps with vaccine-modifiedsmallpox, travelled in buses or trains (see, forexample, WHO/SE/72.41, Suleimanov &Mandokhel) . Usually the travellers becameinfectious after arriving at their destination,where they initiated new outbreaks. Perhapsthe most dramatic of such instances of disper-sal of smallpox to distant areas were thosefrom the Jamshedpur industrial complex inBihar, India, in 1974 (see Chapter 15, Fig .15.20). During a 6-week period, betweenthe end of February and mid-April 1974,travellers from Jamshedpur caused nearly 300outbreaks in other Indian states, dispersaloccurring mainly through train passengerstravelling from the Tatanagar railway station .During 1974 there were many exportationsfrom India to Nepal and in 1975 smallpox wasreintroduced into India from Bangladesh on32 occasions (see Chapter 15) .

Because of their greater mobility, adultmen were much more likely than women orchildren to acquire smallpox outside thehome ; the movement of sick peasants orworkers from the city back to their villagesappeared to be important in the maintenanceof smallpox in some rural areas in the Indiansubcontinent (Thomas et al ., 1971a ; Thomaset al ., 1972 ; Sommer & Foster, 1974) . Indeed,it was suggested that in Pakistan the citieswere the reservoir of smallpox during themonsoons, the disease being carried back tothe rural areas after the rains. However, wheneffective active surveillance was introduced, anumber of affected localities were found inrural areas immediately after the monsoons, inspite of reports of "no transmission" duringthe rains (see Chapter 14). Not surprisingly, incountries such as India, in which 80% of thepopulation was rural, the predominant travelpattern (74%) was from one rural area toanother, and smallpox transmission followedthe pattern of population movement .

Dispersal by refugeesAlthough they constituted a special type of

traveller, refugees were of particular impor-tance in the spread of smallpox, because of thevast numbers involved in the flight from war-and famine-stricken areas. Such dispersal wasdirectly responsible for the re-establishmentof endemic variola major in Bangladesh in1972 (see Chapter 16). Many other instancesoccurred, both between and especially withincountries. Movement of refugees provided

4. EPIDEMIOLOGY 203

almost ideal conditions for the promotion ofspread-large numbers of persons of all agesliving in close proximity under very unsatis-factory conditions, often suffering from mal-nutrition and often poorly vaccinated . Theisolation of cases was difficult or impossible .

Illustrative Case Studies in Rural Areas

In the developing countries in whichsmallpox was most recently endemic the vastbulk of the population lives in villages in therural areas . In India, for example, the 1971census figures showed that 80% of thepopulation of 548 million lived in 580 000villages, of which 318 000 had a population ofless than 500. There were only 148 towns andcities with a population of more than 100 000 .In Bangladesh and in most countries of Africathe proportion of population in the ruralareas was even higher. In some countries, suchas Afghanistan and Somalia, some of the ruralpopulation lived a nomadic life ; these nomadspresented special problems in smallpox eradi-cation programmes .

Studies in AfricaHenderson & Yekpe (1969) described the

spread of smallpox in a village in southernBenin, in which about half the population of300, as well as about half the members of theinfected households, had old vaccinationscars. The chronology of the epidemic and theprobable chains of transmission which startedwith the movement of an infected woman andher children back to the village are shown inFig. 4.13. The 3 cases infected outside thevillage lived in house A and the disease spreadto 5 contiguous houses (B-F) by moving fromone infected household to the uninfectedhousehold nearest to it. Seventeen of the 28persons living in that cluster of houses (A-F)became ill . A 17-year-old boy living in houseG, who carried food to persons in the infectedcluster of households, became ill but producedno secondary cases ; both his parents hadvaccination scars. Infection was brought tohouse H by the "grandfeticheur" (Case No. 22),whose occupation involved frequent contactswith all cases of smallpox, and he wasprobably responsible for the other 6 cases inthe village, in houses H and I . The 2 patientsin household I joined the grandfeticheur in thefuneral ceremonies of smallpox victims. Fre-quent casual contact occurred between the

2 0 4

Case 1-2

3

Road

n- H

BCase 4

7-8

21HOUSE B and C

HOUSE G

5

9-12

22'' 23-26HOUSE A

. HOUSED

HOUSE H,'n-n~nn`i1

I

Case 6

13

15

17 18

, '27-28

HOUSE E

HOUSE I

nnno n n}-

Case14

16 19 20

HOUSE F

25 30 5 10 15 20 25 30 5

10 15 20 25 30 5 10 15MAY-I

JUNE-JULY

1-AUG

in 00 0 0

0

00 0

O0

° 0 0 00 0

0

0 0O

0 0 0 0 00

O 0WHO

0ISOLATION

0

0 BG HUNT

0F

0 o E_ W∎Do

0 O 0 Bath AO 0

0

10 metres

Fig . 4 .13 . The spread of smallpox in a village in Beninof 300 inhabitants, living in 59 houses, extendingfrom May to the end of August 1967 . A: Chronologyand spread . Solid lines = routes of transmission ;broken lines = probable routes of transmission . B :Location of households. (Based on Henderson &Yekpe, 1969 .)

villagers in general and the smallpox suffer-ers, although once the rash had developed thepatients were generally avoided .

The striking feature of this epidemic wasthe spatial localization of the disease, whichreflected intimate social contact between atleast some members of the affected house-holds. Once smallpox had been introducedinto a household, secondary cases occurredwithin it. The transmission of smallpoxappeared to stop when the number of suscep-tible individuals in the village who were incasual contact with cases was still large but thesupply of such persons in prolonged andintimate contact with cases was virtuallyexhausted. Even among the household con-tacts smallpox was not highly contagious ;prolonged or intimate exposure was usuallynecessary. Because of this pattern of spread, itis not surprising that smallpox was eliminatedfrom many African countries much morerapidly than had originally been anticipated(see Chapter 17) .

SMALLPOX AND ITS ERADICATION

AStudies in Brazil

Similar observations emerged from a studyof the spread of variola minor in a semi-ruralschool district in Sao Paulo, Brazil, in whichthere were two types of social agglomerates :school classes (Angulo et al ., 1964) andhouseholds (Angulo et al., 1967) . All thesecondary cases in the 12 infected householdsresulted from a single introduction by aknown primary case of smallpox, who wasusually a child infected in the school . Infec-tion occurred only among persons in intimateand face-to-face contact, either in the house-hold, in the school, or in shared householdfacilities (Angulo et al., 1968) .

Studies in the Indian subcontinentThomas and his colleagues have reported

detailed studies of smallpox in rural areas ofPakistan and Bangladesh (see Chapters 14 and16). Their investigations in Bangladesh(Thomas et al., 1971a, b) showed that small-pox did not remain indefinitely endemic ina rural district in which 113 000 inhabitantsresided in 132 villages scattered over an areaof about 200 square kilometres. During aperiod of 12 months, 119 cases of smallpoxoccurred in 30 outbreaks in 27 widely dis-persed villages . Inter-village spread occurredin only one case ; in the other 21 outbreakswhose source could be determined, smallpoxwas introduced by a landless peasant from thevillage in question who had contracted small-pox during a visit to a city to obtainemployment. The probability of an introduc-tion was correlated with the population sizeof the village, and thus with the probablenumbers of landless peasants from each vil-lage seeking employment in the cities . Withthe introduction of smallpox into a village,secondary cases appear to have occurred morefrequently outside than inside families.Nevertheless, the secondary attack ratesamong unvaccinated persons were highestwithin families.

A further, substantially larger study wascarried out in Pakistan (Mack et al ., 1972a, b ;Thomas et al., 1972) . The most strikingdifferences between the two studies werethe very high secondary attack rate (88%)among unvaccinated contacts within com-pounds and the high proportion of cases(78%) resulting from spread between ratherthan within compounds. The explanation for

the very high secondary attack rate in Paki-stan is not clear ; possibly it was due to the lowhumidity prevailing in the Punjab for most ofthe year. As in Bangladesh, smallpox was notmaintained as an endemic disease within therural areas, even in one as large as 6000 squarekilometres with 1 .2 million inhabitants. Overhalf the outbreaks occurring in the study areawere ultimately traced to cities, emphasizingthe significance in Pakistan of these largeaggregations of population for the main-tenance of the disease .

Investigation in Pakistan by Heiner et al .(1971b) confirmed the high secondary attackrate in the Punjab, which reached 77% inunvaccinated compound contacts and 4 .8in previously vaccinated contacts, all of thelatter having been exposed to severe or fatalindex cases. Further, vaccinated persons gen-erated many fewer cases among their contacts(4%) than did unvaccinated index cases(11 .5%) (see Table 4.10), reflecting thegreater severity of the disease in unvaccinatedsubjects.

Another investigation in rural villages inPakistan, which was mainly concerned withinapparent infections (Heiner et al ., 1971a),has been discussed at length in Chapter 1 .Here we would note that in this study, villageresidents who were not household or com-pound contacts of cases were taken as one ofthe control groups. In contrast to vaccinatedcompound contacts, over half of whom hadhad overt or inapparent smallpox, only 6 .5%of these other vaccinated villagers in thiscontrol group had sustained inapparent infec-tions (none had overt smallpox), again show-ing the importance of intimate face-to-facecontact in the transmission of smallpox .

Locality

Population Numberof cases

DartaOridanLoalaBllaheyMadhareBerdebloleBoldweneAbdilelibMandeeloShafa

4. EPIDEMIOLOGY

205

Prolonged transmission among nomads

Nomads constituted a special problem inthe Horn of Africa, sub-Sahelian westernAfrica and Afghanistan . On general princi-ples, it could be predicted that smallpoxwould rarely persist for long in such smallisolated populations, even in the absence ofcontrol measures . This was indeed usually thecase. Outbreaks among nomads constituted68 % of 843 outbreaks in Somalia in 1977, butin only 10 of them did prolonged transmis-sion occur (Table 4.13). The best documentedof these outbreaks, in Mandeelo village (Fos-ter et al ., 1978), is described in Chapter 22 . Asindicated in Table 4.13, transmission wasinterrupted very soon after outbreaks innomads were detected, and in some instancesthe last case had occurred before the outbreakwas detected. Similar prolonged persistence ofsmallpox in nomadic groups of less than 15persons have been described in both Camer-oon and Niger (Henderson & Yekpe, 1969) .

This slow and prolonged transmission wasdue to the balance between the supply ofsusceptible persons and the transmission rate .Nomads were usually poorly vaccinated andpatients with variola minor were often ambu-lant throughout their illness. The transmis-sion rate was low because most of theactivities of nomads occurred in the open,where opportunities for face-to-face trans-mission by large-particle aerosols were greatlyreduced. As noted above, these circumstancesusually led to the spontaneous termination ofoutbreaks (in 98.2% of outbreaks), but iftransmission was maintained it continued formany weeks, even in these very smallpopulations .

Table 4.13 Prolonged transmission of smallpox in several nomadic encampments in Somalia in 19772Number ofdays fromfirst tolast case

a The great majority of outbreaks in such groups were rapidly detected and controlled, or the chain of transmission was Interruptedspontaneously . (Based on Jetek et al ., 1981 .)

Date of:

Firstcase

Lastcase

Detection

17 Feb . I Aug. 20 Aug .21 Jan. 23 June 28 June22 March 6 July 22 June14 March 8 June 19 June6 March 24 May 18 May

20 April 4 July 5 July26 March 8 June 17 May25 Feb. 7 May 24 April23 April I July 23 June28 June I Sept. 12 Aug .

55 14 16335 24 15298 20 10675 23 9535 9 8065 9 7450 6 73I I 6 7046 21 6860 12 65

2 06

Illustrative Case Studies in Urban Areas

Outbreaks of smallpox in non-endemiccountries since 1950, most of which occurredin urban settings, are discussed in Chapter 23 .Of more significance than these episodes, inrelation to the global smallpox eradicationprogramme, was the pattern of spread inurban areas in countries in which smallpoxwas still endemic. Only a few reports of suchsituations have been published, mainly refer-ring to cities in the Indian subcontinent(Madras : Pandit et al ., 1959 ; Rao et al., 1968a ;Rao, 1972 ; Calcutta : Mukherjee et al., 1974 ;Lahore and surroundings : Ali & Heiner,1971 ; Thomas et al., 1972 ; BangladeshSommer & Foster, 1974 ; Brazil (variolaminor) : Rodrigues-da-Silva et al., 1963 ; Azer-edo Costa & Morris, 1975) .

Transmission to household contacts insome cities, e.g., Madras (Rao, 1972), showedmuch the same pattern as that in rural areas ;transmission was usually more commonamong intrafamilial than extrafamilial con-tacts, even when all those studied lived incompounds in which families shared anentrance, toilet facilities and a bath . Theimportance of the local social arrangements isshown by the fact that in Calcutta Mukherjeeet al . (1974) found that the attack rates amonghousehold and compound contacts were verysimilar ; in this situation the mixing of theinhabitants was so intimate that each com-pound could be regarded as a single largefamily.

In many endemic countries the larger townsand cities constituted the major reservoir ofsmallpox during the wet season . Sarkar et al .(1970) studied the epidemiology of off-seasoncases in Calcutta in 1967 and 1968 . Most of thefew cases that occurred during the monthsJune-November of each year were in beggarsand footpath dwellers or in manual labourerswho did not work in fixed places and lived inthe bustees (slum areas). Successive cases weresometimes widely dispersed, but the majorityof cases could be grouped into 6 outbreaks inthe bustees, different groups being involved in1967 and 1968 . Several cases were probablyassociated with introductions from villagesoutside Calcutta, and in one instance there wasevidence that infection was introduced fromCalcutta into a village 30 kilometres away inMay 1968, maintained there by serial transmis-sion, and reintroduced into another part ofCalcutta in September 1968 .

SMALLPOX AND ITS ERADICATION

Several reports refer to the maintenance ofsmallpox in small pockets of unvaccinatedindividuals in otherwise well-vaccinatedpopulations. In Lahore, for example, Ali &Heiner (1971) found that in a situation inwhich 93% of the population had vacci-nation scars or a history of smallpox, thedisease could persist for most of the year,mainly because of the high concentration ofunvaccinated children under 3 years of ageand of older persons in whom the immunityconferred by primary vaccination many yearsbefore had waned .

Experience in Indonesia (see Chapter 13)provided one of the earliest and clearestdemonstrations of the ineffectiveness of massvaccination alone as a means of eliminatingsmallpox from populous countries . It used tobe held that a vaccination level of 80% wouldsuffice to raise the herd immunity to such anextent that transmission would cease (seeChapter 9) . However, in Indonesia in 1969and 1970 chains of transmission of variolamajor were maintained in communities inwhich the overall vaccination rate was about90%. This happened because of the highpopulation density (see Table 4.11), andbecause a large proportion of the unvacci-nated were in the age group 0-4 years . TheIndonesian practice, mentioned earlier in thischapter, of taking children with smallpox onsocial visits to their neighbours and relativesmaintained a chain of transmission in thissection of the population .Thompson & Foege (WHO/SE/68.3)

demonstrated that the existence of specialgroups who, for social or religious reasons,refused vaccination was of significance inmaintaining endemic smallpox within anotherwise well-vaccinated urban communityin Nigeria. The concealment of cases playedan important role in the development of theepidemic, but the disease spread quite slowly(32 recognized cases over a period of 11weeks), even in a compound where nobodyhad been vaccinated. In spite of frequentsocial mixing on religious occasions, trans-mission occurred most frequently as a result offamily and compound contacts. Such situa-tions were not restricted to urban areas, forthe last outbreak of smallpox in southernAfrica occurred among a religious sect inBotswana, who lived in closed communitiesin 9 small towns and likewise refused to bevaccinated (Presthus, 1974 ; WHO/SE/74.69,Presthus & Sibiya) .

SUMMARY

The basic facts of the epidemiology ofsmallpox may be summarized as follows :

(1) Smallpox was a specifically humandisease ; there was and is no known animalreservoir of variola virus (see Chapter 30) .

(2) Compared with the infectious agents ofmany other human viral diseases, variola

4. EPIDEMIOLOGY 207

Plate 4 .4. A : Thomas M . Mack (b . 1936), B : David B . Thomas (b. 1937) and C : Gordon G . Heiner (b . 1924),with Pakistani colleagues, conducted some of the most important and certainly the most comprehensive studiesof the epidemiology of smallpox to be undertaken during the Intensified Programme . Working in what was thenEast and West Pakistan between 1965 and 1968, they demonstrated how surveillance and containment measurescould be highly effective even in some of the most heavily infected areas .

virions are relatively resistant to inactivationby physical and chemical agents ; neverthe-less, infection almost always involved theface-to-face contact of a susceptible subjectwith a person suffering from clinicalsmallpox .

(3) The detection and recognition of caseswere relatively simple, since the rash wasusually quite distinctive and occurred mainlyon uncovered parts of the body .

mf0m

20 8

(4) Subclinical infections with variolavirus seldom occurred, except in vaccinatedclose contacts of cases . These individualsrarely, if ever, transmitted smallpox to othersand were of little or no importance epidemio-logically ; almost all new cases could be tracedto contacts with overt cases .

(5) An attack of smallpox was followed bydeath or recovery ; persistent, latent or recur-rent infection did not occur and cases werenever infectious after the rash had gone .

(6) Different strains of variola virus dif-fered in their virulence . Strains from out-breaks with case-fatality rates of about 1 % inunvaccinated persons are designated variolaminor virus. Most strains, for most of history,were associated with much higher case-fatality rates (5 %-15 %, and more commonly

SMALLPOX AND ITS ERADICATION

about 25%) ; these strains are designatedvariola major virus .

(7) All strains of variola virus are indistin-guishable antigenically. In the vast majorityof cases, immunity to reinfection was abso-lute. Similar protection could be reproduced,for a period of several years, by immunizationwith a live virus vaccine prepared from arelated orthopoxvirus, such as cowpox orvaccinia virus.

(8) There was a pronounced seasonal inci-dence : smallpox was essentially a winter-spring disease.

(9) Smallpox spread rather slowly . Therewas an interval of 2-3 weeks between eachgeneration of cases, and even during thetransmission season an index case rarelyinfected as many as 5 other persons .