Clinical survey techniques to estimate prevalence and annual incidence of poliomyelitis in developing countries*

Bulletin of the World Health Organization, 58 (4): 609-620 (1980) Clinical survey techniques to estimate prevalence and annual incidence of poliomyelitis in developing countries* F. M. LAFORCE,1 M. S. LICHNEVSKI,2 J. KEJA,2 & R. H. HENDERSON3 Recent epidemiological data suggest thatpoliomyelitis isa more importantpublic health problem in developing countries than was originally suspected. Because ofthe availability of a cheap, effective, but labile oral vaccine, it is important that countries determine the incidence of poliomyelitis in order to determine the necessity and speed with which a vaccination programme should be introduced. Sequelae of poliomyelitis are distinctive and can be used to study the incidence of residual paralysis in children above the age of5 years. Prevalence datafor lameness due to poliomyelitis can then be translated into rates of annual incidence of paralytic disease. Lameness surveys can be effciently done in schools and data are reliable as long as a similar search for lame children is done amongst children of the same age group who do not attend school but are in the geographical area served by the school. House-to-house surveys, while more time-consuming, are more reliable. Until recently poliomyelitis has not been considered as a major public health problem in developing countries (1); this impression has been supported by the low incidence rates typically reported. The usual reason given is that children are exposed to poliovirus early in life when they are partially protected by ma- ternal antibodies and at low risk of paralytic disease, although such exposure permits infection with the de- velopment of solid immunity. The validity of this as- sumption is now coming into question as more careful studies of paralysed children are showing that polio- myelitis is a far more important public health problem than was formerly believed. * From the Expanded Programme on Immunization, World Health Organization, 1211 Geneva 27, Switzerland. Requests for reprints should be addressed to Dr R. H. Henderson. I Consultant. Present address: Chief, Medical Service, Veterans Administration Medical Center, Denver, CO, USA. 2 Medical Officer. 3 Director. In 1974 Nicholas and coworkers reported on lame- ness due to poliomyelitis in Ghanaian schoolchildren (2, 3). In an area where poliomyelitis was felt not to be a problem they documented that over 7 per 1000 schoolchildren were lame due to poliomyelitis. The annual incidence was estimated to be at least 28 per 100 000 population, a rate comparable to incidence rates in the United States of America prior to the de- velopment of poliomyelitis vaccine. Similarly, intensi- fied surveillance of poliomyelitis cases in Yaounde, United Republic of Cameroon, documented an aver- age annual incidence rate of48 per 100 000 population from 1973 to 1975 (4). Lameness surveys in Burma, Egypt, Indonesia, the Philippines, and Thailand have all shown prevalence rates of residual paralysis due to poliomyelitis that were higher than initially suspected (5-9). Thus, the very worrisome possibility has arisen that 'the mean annual incidence rate of poliomyelitis under endemic circumstances in tropical countries 3980 -609- F. M. LAFORCE ET AL. may have always been as great, if not greater, than those experienced by temperate countries during epidemic periods' (3). Because of limited hospital and rehabilitation services many of these victims of poliomyelitis remain untreated. In Uganda, with a population of about 10 million, about 30 000 persons are severely paralysed while another 90 000 have some residual paralysis due to poliomyelitis. The number of untreated paralysed victims of poliomyelitis in Nigeria has been estimated at 200 000- 300 000. Extrapolating from the data of Nicholas and coworkers, about 2800 Ghanaian chil- dren under the age of 4 years are permanently crippled each year from poliomyelitis (2). Thus, a sizeable reservoir is being generated of incapacitated ado- lescents and adults who require, but for the most part do not receive, rehabilitation services. Immunization is the sole effective means of pre- venting poliomyelitis. Oral poliomyelitis vaccine (OPV) is a labile, live vaccine and large-scale use requires a well-developed cold chain to ensure potency and vaccine delivery services capable of administering several doses of OPV. Thus, it is important that countries assess the incidence of poliomyelitis in order to determine the need and the speed with which OPV should be included in an expanded programme on immunization. Experience from many countries has made it clear that official statistics (reported cases) grossly underestimate the number of cases since these statistics are limited, for the most part, to hospitalized cases. The techniques to be described, while reflecting the earlier impact of poliomyelitis on a population, can be used to gather information that may help in making a decision whether or not to launch a vaccin- ation programme against poliomyelitis. The clinical sequelae of paralytic poliomyelitis are distinctive and, when observed, can be attributed to poliomyelitis with a high degree of probability. By studying cases of paralysis, particularly in school-age children, it is possible to determine the prevalence of paralysis due to poliomyelitis and to estimate the annual incidence in recent years. This paper will summarize the clinical and epidemiological charac- teristics of poliomyelitis in developing countries and will describe clinical survey techniques in keeping with these characteristics that can be used to determine the prevalence of residual paralysis due to poliomyelitis. EPIDEMIOLOGICAL CHARACTERISTICS OF POLIOMYELITIS IN DEVELOPING COUNTRIES Because of poor sanitary conditions in most devel- oping countries, polioviruses circulate freely among infants. Therefore, almost all children develop anti- bodies to at least one of the three poliovirus types before 5 years of age (10, 11). The solid line labelled A in Fig. 1 shows the percentage of sera drawn in 1948-49 from children in Cairo that contained neu- tralizing antibody to type 2 poliovirus. This curve, showing an initial decline in maternal antibody fol- lowed by a rapid rise in seropositivity, is similar to those found in all seroepidemiological surveys of poliomyelitis done in tropical developing countries. More than 500o of sera from children less than 6 months of age are positive because of transplacental transfer of maternal antibodies but the level of such antibodies wanes so that infants are fully susceptible to poliomyelitis infection by 6-7 months of age. Fifty per cent of serum samples from children between 1 and 2 years of age have poliovirus antibodies. This percentage continues to increase so that, by age 4 years virtually all children have antibodies to at least one poliovirus type. For a given population, the shape of this curve does not change until improvements in sanitation decrease the level of contact children have with poliomyelitis virus. The solid line labelled B in Fig. 1 represents serological data from children living in Miami Beach, USA, during 1949-51. Whereas 75% of children in Cairo had antibody to type 2 poliovirus by 2 years of age, only about 10% of Miami children were positive. By age 5 years about 500o of Miami children had anti- bodies, whereas more than 90% of Cairo children had been infected. Exposure to poliomyelitis virus in Cairo children was very early, intensive, and virtually universal, whereas infection in Miami children was delayed, presumably because of better sanitation which decreased the level of exposure. Consequently, clinical poliomyelitis in tropical de- veloping countries is almost exclusively an infantile disease, with well over 9007o of cases occurring under the age of 5 years (Fig. 1, dotted line labelled A). This is in sharp contrast to the distribution of clinical polio- myelitis cases in developed countries prior to the wide- spread use of poliomyelitis vaccine, where the disease commonly affected adolescents and young adults (Fig. 1, dotted line labelled B). CLINICAL CHARACTERISTICS OF POLIOMYELITIS Infection with poliovirus can go completely un- noticed or, at the other extreme, cause a fulminating, rapidly progressing, fatal bulbar paralysis. Clinical features of acute poliomyelitis have been well de- scribed (12), and will not be reviewed. Rather, we will summarize those clinical characteristics that would help the field epidemiologist determine which type of paralysis in an older child is due to poliomyelitis. 610 ESTIMATING PREVALENCE OF POLIOMYELITIS 18 20 22 24 26 28 30 Y e a r s A= Cairo, Egypt antibodies B= Miami, Fla, USA = cases Fig. 1. Comparison of type 2 neutralizing antibody pattern with cumulative attack-rate of poliomyelitis in populations from two sub-tropical areas.' a From PAUL, J. R. Epidemiology of poliomyelitis. In: Debr6, R., (Monograph Series No. 26). Natural history of clinical poliomyelitis There are a number of possible outcomes after a child develops paralytic poliomyelitis. These out- comes are important since it is only the child with Table 1. Poliomyelitis cases in Greater Bombay: clinical out- comea (1971-75) Outcome Number Percentage No recovery 342 11.4 74.6% Partial recovery 1 895 63.2 Complete recovery 506 16.8 25.4% Fatal 255 8.6 Total 2 998 a 2 months after onset of disease. ed. Poliomyelitis. Geneva, World Health Organization, 1955, p.24 residual paralysis who can be identified at a later date as having had poliomyelitis. Table 1 summarizes clinical data on 2998 cases of poliomyelitis that oc- curred in Greater Bombay, India, from 1971 to 1975 (K. H. Dave, unpublished observations, 1978). Each of these cases was followed clinically and recovery was assessed 2 months after the onset of paralysis. One- quarter of the cases, including deaths and patients with complete recovery, would not have been detected in a later clinical survey. Thus, prevalence surveys of residual paralysis due to poliomyelitis will always significantly underestimate the true incidence of the disease. A rough estimate of all clinical cases of polio- myelitis could be made by multiplying the prevalence rate of residual paralysis due to poliomyelitis by 1.33. In addition, such surveys will not take into account excess mortality in children after an outbreak of polio- myelitis. While this is not a problem in countries with high standards of living and well-developed rehabilita- tion services, paralysis may be an important contribu- tory factor to increased mortality, particularly in times of food shortage. 100 611 a, -ocnCD 0, c c(U -J -C a,- 0~ 100 wa, 0c (U80 -' a) 0 a, 60, a) cm a, a +-A XCD 20 3 0 10 36 38 40 I WHO 80079 F. M. LAFORCE ET AL. Clinical presentation of residual paralysis due to poliomyelitis Paralysis due to poliomyelitis in children under 3 years of age usually affects the lower legs. Clinical data from Nigeria, India, and Liberia, where more than 95% of poliomyelitis cases occurred in children less than 5 years of age, showed evidence of involve- ment of the lower limbs in 78-86% of cases (13-15). Thus, any prevalence study of poliomyelitis that uses lameness as a criterion for diagnosis will identify about 80% of cases. The total prevalence of residual paralysis due to poliomyelitis could be estimated by multiplying the prevalence of lameness due to polio- myelitis by 1.25. Clinically, residual paralysis due to poliomyelitis is relatively easy to recognize. Muscles that extend hips and knees and dorsiflex the ankle are commonly in- volved. The diagnosis of poliomyelitis can be made with a high degree of confidence in the presence of flaccid paralysis with atrophy where there is no de- crease in sensation and a history ofacute onset with no progression. A STABLE ENDEMIC No of Cases B ENDEMIC PLUS RECENT EPIDEMIC No of cases C ENDEMIC PLUS OLD EPIDEMIC No of cames Yar 65 67 69 71 73 75 Age at time of premaencesurvey 15 13 11 9 7 5 77 79 3 1 ESTIMATING THE ANNUAL INCIDENCE OF POLIO- MYELITIS FROM THE PREVALENCE OF RESIDUAL PARALYSIS 2linical surveys of poliomyelitis seek to determine a valence rate, usually expressed as the number of es of residual paralysis due to poliomyelitis per 0 population, in a particular age group. In eloping countries, virtually all cases of polio- elitis have an onset before the fifth year of life. suming a stable endemicsituation, the paralysis due )oliomyelitis in the age group above 5 years should ect all the non-fatal cases with residual paralysis t occurred in this cohort between the ages of 0-4 rs. Thus, surveys of residual paralysis in 10-year-olds reflect the incidence of paralytic cases urring from 1 to 10 years before. Fig. 2 and 3 describe results of a series of theoretical valence surveys. A careful study of each of these draw attention to some of the limitations of clini- surveys. Panel A in Fig. 2 shows the results of a valence survey done in 1980 which reflects a stable 1980 PREVALENCE SURVEY 5 - 15 sampled 5-9ratel-5rate ~ ~ L I L L L L I 5 -15 sampled /5-9rate= 10- 15 rate 5-15sampled 5 - 0 rate>10- 15 rate I -A .I I I I I L .. .I I I I I 1 3 5 7 9 11 13 15 Age WHO 75185 Fig. 2. Prevalence rates under various conditions. 0 0 0 0 a a 0 a 612 I ESTIMATING PREVALENCE OF POLIOMYELITIS disease incidence from 1965-80, as shown on the left. Thus, a prevalence survey of residual poliomyelitis in all age groups would be represented by a sharp in- crease in cases in years 1, 2, and 3, a levelling off of cases in the fourth year and stable prevalence rates from 5-15 years of age. Thus, if a 'lameness survey' were to be done in 5- 15-year-olds, the prevalence rate in 5-9-year-olds would be similar to that noted in 10- 15-year-olds. Panel B in Fig. 2 now superimposes a sharp polio- myelitis outbreak in 1979 and 1980 and the effect of this outbreak on a prevalence survey. Higher than normal prevalence rates would be present in the 2-4-year age group, reflecting cases from the epidemic years. However, cases in the 5-15-year-old group would not reflect this recent epidemic since children affected by the outbreak would not be old enough to be counted in the 5-15-year-old survey group. Thus, prevalence rates in 5-9-year-olds would again equal those seen in 10- 15-year-olds. On the other hand, if the epidemic were to have occurred in A STABLE ENDEMIC No of c 1972 and 1973, increased prevalence rates would be expected in the 5-9-year-olds (Fig. 2, Panel C). Conversely, if an effective vaccination plan were introduced in 1979 and 1980 that resulted in a marked decrease in the number of paralytic poliomyelitis cases, this would have no immediate effect on the prevalence of lameness in 5-15-year-olds (Fig. 3, Panel B). However, if this vaccination effort had been started earlier and had resulted in a sustained decrease in new cases, this would be reflected by lower rates of residual paralysis in cohorts that were vaccinated and that are now old enough to be included in lameness surveys (Fig. 3, Panel C). Surveys of residual paralysis due to poliomyelitis will be reported as cases per 1000, usually in children older than 5 years. This prevalence rate represents the sum of all cases, almost all of which occurred from 0-4 years of age. The prevalence rate of residual paralysis can be used to estimate annual incidence of paralytic cases. For example, if a prevalence rate of 10 cases per 1000 is found in a cohort of children 5-10 Im8 PREVALENCE SURVEY 5- 15 ampled B ENDEMIC PLUS RECENT VACCINATION Vaccination I No of cam C ENDEMIC PLUS CONTINUED VACCINATION Vaccination No of cam Yer 66 671 69 71 73 75 77 79 5 - 15 sampled 5 - 9 rate = 10 - 15 rate 5-15smpled 55-9 rate -' 10 -15 rate i I, 1 3 5 7 9 11 13 15 Ap WHO 70186 Fig. 3. Prevalence rates under various conditions (continued). L---V-M . - wirm ___. I 613 I 614 F. M. LAFORCE ET AL. years of age, correction for those cases not involving the lower extremities is done by multiplying the prevalence rate by 1.25 per 1000. An estimate of the annual incidence could be obtained by dividing the prevalence rate (12.5 per 1000) by the number of years at risk (five), giving an annual incidence of 2.5 per 1000 in the 0-4-year age group. If the proportion of the population made up of 0-4-year olds is known (usually 16-2007o), the incidence of paralytic poliomyelitis for the whole population can be deter- mined as follows: Annual incidence in 0-4-year age group x Annual incidence per distribution 0-4-year = 1000 in total popu- age group in general lation. population Continuing with our previous example, if the 0-4-year-old population makes up 200%o of the popu- lation, the annual incidence is 2.5/1000 x 0.2 = 0.5 per 1000 population or 50 per 100 000 population. (An estimate of all cases of paralytic poliomyelitis can be made by multiplying annual incidence by 1.33 to correct for those cases that completely recovered or died within 2 months after onset of disease.) Table 2 converts prevalence data to estimates of annual incidence per 100 000 population. It is worth Table 2. Relationship between prevalence rate and estimate of annual incidence of poliomyelitis Prevalence rate per 1000 Annual incidence per 100 000 5-15 years of age populationa 0.5 2.0 1.0 4.0 2.0 8.0 3.0 12.0 4.0 16.0 5.0 20.0 6.0 24.0 7.0 28.0 8.0 32.0 9.0 36.0 10.0 40.0 11.0 44.0 12.0 48.0 13.0 52.0 14.0 56.0 15.0 60.0 a Assumes 0-4-year-old population as 20% of total. noting that from 1945 to 1954 the annual incidence rate of poliomyelitis in the United States of America ranged from 15-30 per 100 000 population. CLINICAL SURVEY TECHNIQUES Although not essential, it is useful first to determine the age distribution of poliomyelitis cases at the onset of disease. This information is usually available from a review of data from hospitals or rehabilitation centres. If more than 900o of poliomyelitis cases have had their onset before the age of 5 years, one can feel relatively confident that the prevalence survey tech- niques to be described can be used accurately. In the absence of this information, prevalence survey tech- niques can still be used, and will provide information on the age of onset from interviews of parents of lame individuals. The prevalence of residual poliomyelitis can be arbi- trarily described as being low (<1 per 1000), moderate (1-4 per 1000), or high (>4 per 1000). The most important levels to determine reliably are the low and high ones; a prevalence rate less than 1 per 1000 would indicate that paralytic poliomyelitis is a relatively minor public health problem, while a rate greater than 4 per 1000 would indicate a major problem. Since virtually all of the prevalence surveys of re- sidual poliomyelitis done in developing countries have shown rates in the moderate or high range, a sample size of 6000 could be considered as suitable for initial studies. As in any study, carelessly acquired data will threaten the validity of the conclusions and it is prefer- able to limit the sample to a number that can be evalu- ated well rather than to sacrifice quality in order to achieve large numbers. Prevalence surveys in schools A survey of schoolchildren for evidence of residual poliomyelitis paralysis is the simplest technique for determining prevalence rates. Briefly, lame children attending school are identified by a teacher and examined.' Further clinical and epidemiological data are obtained from those children whose lameness is felt to be secondary to poliomyelitis. Despite its apparent simplicity, such an approach does have pitfalls. First, accurate school enrolment data and general demographic data from the area to be sampled are necessary. Secondly, it is important to know trends in school attendance in the higher grades. If school attendance of well children has tended to drop off for any reason, an artificial concentration of crippled children in school may have developed since they are not capable of working. Thus, apparent a Parents are contacted and invited to attend the examination of the child whenever possible. ESTIMATING PREVALENCE OF POLIOMYELITIS 615 increases in poliomyelitis prevalence with age may simply reflect school attendance patterns. Conversely, lame children may not attend school for any number of reasons, thus giving a falsely low prevalence rate. Therefore, it is important to establish a prevalence rate for residual poliomyelitis in non-school-attenders in the same age group as the school population being studied. Choosing schools. The selection of schools for the survey is not easy. Evidence suggests that endemic poliomyelitis is not a focal disease (2, 3). Such a view is consistent with previously-mentioned seroepidemio- logical studies that show early, virtually universal infection with a relatively low infection to case ratio. Lastly, although this point is still controversial, data also suggest that in tropical developing countries there is no difference between prevalence rates of residual poliomyelitis in rural and urban areas. In trying to emphasize simplicity, we would suggest the following approach: Step 1: Prepare a list of all urbanb and rural schools in the area to be sampled. Step 2: Aim to sample 6000 children attending urban schools and an equal number attending rural schools. Step 3: Choose 3 schools at random from each of the rural and urban school lists. Step 4: Plan to collect data from at least 2000 students from each of these 6 sites (if the school chosen does not have 2000 children enrolled, the next-closest school is also visited and so on). The school sample containing the 2000th child should be completed. Thus, for each sample area, 6 groups (3 urban and 3 rural) of over 2000 will have been sampled. Age group to be sampled. This is an important consideration since it fixes the group in whom lame- ness will be looked for and recorded. Rather than arbi- trarily fixing an age group, such as 5-10 years or 5-15 years, it would seem better to discuss with principals and teachers which group above 5 years old is best for such a sample. Local factors, such as seasonal employ- ment, may also help specify the best time for doing the survey. Visiting schools. The principals of the schools to be included in the survey should be visited about one month prior to the planned visit of the survey team in order to explain the purpose of the study and to leave forms for the teachers to complete with the names of all lame children (Form 1).c While at the school, class b For the purposes of this presentation, urban units are those with a population greater than 2000. c Examples of Forms 1 and 2 are shown at the end of this article on pp. 619-620. sizes, drop-out rates, catchment areas, and other information should also be noted. The studies in Ghana have shown that teachers can prepare accurate lists of lame children and thus, on a second visit, it is necessary only to examine each child identified by the teachers as lame and complete a clini- cal examination form for each one (Form 2).' One approach already field-tested in Ghana is as follows: the child is first asked to walk to ensure that he or she is, in fact, lame. The child is then comfortably seated and muscle tone is determined in the involved area and a normal extremity by observing the passive range of motion of a relaxed extremity. Muscle mass is then estimated by examination and palpation; if necessary, mid-calf and mid-thigh measurements can be taken. Sensation is tested by the ability to distinguish the sharp and dull ends of a pin. The degree of disability is recorded as: I-able to walk without mechanical aid; II-able to walk, but only with mechanical aid; III-unable to walk. A history of the acute illness should be obtained, if at all possible, from one or both parents. Age of onset, sex, evolution of defect, and residence at onset should be noted. Lameness among non-school-attenders. If demo- graphic and school enrolment data suggest that school attendance is very high (greater than 90070) most of the information needed to complete the prevalence survey can be obtained from the schools. However, school attendance in many developing countries ranges from 30 to 50/o and decreases significantly each year after school is begun. Therefore, a large percentage of the population in whom prevalence is being determined might not be available for sampling. This can intro- duce bias by falsely elevating the prevalence of polio- myelitis in schoolchildren if children who are physi- cally sound leave school to work. On the other hand, getting to school may prove to be very difficult for lame children, particularly in a rural area, and they may stop going and stay at home. Thus, it is important to know the prevalence of residual poliomyelitis in non-school-attenders in the age group being sampled. Estimating the number of children not attending school in the age group being sampled may be difficult and in most instances requires two steps. The first is an estimate of the total population in the area from which the schools being sampled draw their students. This may be relatively easy in towns when one school serves a well-defined population. However, in cities such an estimate may be difficult, particularly if the catchment areas of schools are not clear or are differ- ent from those used to generate census data. The total number of children in the age group being sampled can be estimated by multiplying the total population by the percentage of children in the sample group. For example, if children 5-10 years of age F. M. LAFORCE ET AL. make up 20% of a total population of 10 000, there will be a total of 2000 children in the target age group. If 1200 of these children attend school, then 800 do not and thus form the denominator for calculation of the prevalence rate of residual poliomyelitis in non- school-attenders. To determine the prevalence in non-school- attenders, it is probably not necessary, in most instances, to do a house-to-house survey. An intensive search for cases in non-school-attenders should suffice. For this purpose, several sources of infor- mation should be used, such as village chiefs, phys- icians, rehabilitation centres, hospitals, traditional healers and, most important, lame children attending school. It is important that only those lame children within the school catchment area be included. Each child so identified is then examined, using the same clinical criteria as in the school survey. Analysis of data. On the completion of the study, the survey team should have the following data for each of the urban and rural areas sampled: -total population within area served by school -number of children in the age group sampled -number of children attending school -number of children not attending school -list of lame children attending and not attending school -clinical and epidemiological data on all lame children. From these data, the number of cases of lameness due to poliomyelitis and the rate per 1000 can be calcu- lated for children in urban and rural areas, both school and non-school attenders. The urban and rural prevalence rates can be corrected to include cases of paralysis not involving the legs by multiplying by 1.25 to give the prevalence rate of residual poliomyelitis. An estimate of the total number of cases of polio- myelitis (including those who died or recovered completely) can be made by multiplying the preva- lence rates of residual poliomyelitis by 1.33. Annual incidence in 0-4-year age group can be estimated by dividing the total prevalence by 5. The annual incidence for the total population can be deter- mined by multiplying annual incidence in the 0-4-year age group by the percentage of children 0-4 years old in the total population. Other important data that should be collected for both urban and rural populations include: -age of onset -year of onset -sex (since several authors (4, 6, 16, 17) have reported higher rates of poliomyelitis in males than in females, sex-specific rates should be calculated if either sex is significantly under-represented in the sample) - ratio of poliomyelitis to all forms of lameness -degree of disability due to poliomyelitis -age of school attenders with residual poliomyelitis -age of non-school-attenders with residual polio- myelitis. House-to-house surveys The most accurate technique to measure the preva- lence of poliomyelitis is a house-to-house survey. Such surveys are time consuming but have the great advan- tage of accurately specifying the population being sampled. In general, house-to-house surveys should aim to collect data on about 6000 urban and 6000 rural children in the 5-15-year age group. Individuals organizing such a survey may also wish to collect data on children less than 5 years old, par- ticularly if there is some suspicion of an epidemic of poliomyelitis within the last 2-3 years (Fig. 2, Panel B). Assuming that endemic poliomyelitis is not a focal disease, it is not necessary to have many sampling sites. We would suggest that 3 urban and 3 rural areas within a specified geographical area be randomly chosen and that 2000 children in the target group be sampled per site. Once the areas where surveys are to be done have been chosen, the starting point of the survey can be determined as follows: Step 1: Prepare a list of houses in the village to be evaluated. Step 2: Number the households on the list. Step 3: Select a random number between I and the highest numbered household on the list. This number will correspond to the first house to be visited. (When determining where to begin in an urban area, it may be simpler to begin first by randomly choosing an area in the city and then proceeding to list the houses.) Upon reaching the house selected, the household head should be seen. All children residing in the house whose ages fall within those chosen for the survey (5-10 or 5-15 years) should be checked and counted. Lame children in the age group chosen for study should be examined using the criteria previously detailed in the school survey (Form 2). House-to-house examination should continue until 2000 children in the age group desired have been counted. It is not necessary that all of these children be seen, but it would be of interest to know how many of the children go to school. In house-to-house surveys in rural areas it will be necessary to go out from the initial house in ever- widening circles, but urban areas falling within such circles should be excluded. 616 ESTIMATING PREVALENCE OF POLIOMYELITIS At the termination of the surveys, the following data should be available: -number of children sampled -list of all cases of lameness -clinical and demographic data on all cases of residual poliomyelitis. Prevalence and incidence should be calculated as for the school survey. Other survey techniques Rehabilitation centres. In certain areas, it may be possible to gather accurate data on acute poliomyelitis cases as they occur by collecting information from treatment centres that are well known to the general population as places where lame children are taken for therapy. For example, rehabilitation centres, particu- larly those operating at no charge, are likely to see a high proportion of acute poliomyelitis cases from the surrounding area. Data on paralytic poliomyelitis from a rehabilitation centre are likely to be far more sensitive to recent changes in the epidemiology of poliomyelitis than prevalence data calculated from survey figures. Diagnosis and dates of onset in rehabilitation centre data are also likely to be quite accurate. Traditional healers. In some countries children with various types of palsies may be taken to a traditional healer such as an acupuncturist or a masseur. Some of these individuals keep accurate records and can provide lists of lame children, as well as serving as sentinel sites. Continuous sampling ofa single age group as a baro- meter of disease activity In many countries surveillance techniques are not sensitive enough to detect even major changes in the incidence of a disease like poliomyelitis. One possible means of measuring the impact of an immunization programme is to concentrate on developing a 'lame- ness register' for all children entering school from the time an immunization programme is begun. It might take 2-3 years to develop a satisfactory system, but one would not expect to see a decrease in the preva- lence rate of lameness in 5-year olds until an effective vaccination programme had been in operation for 3-4 years. Such an approach would have the advantage of being both simple and inexpensive. ACKNOWLEDGEMENTS The authors gratefully acknowledge the assistance of Dr F. Assaad, Dr S. Foster, Dr Z. Islam, and Dr D. Tarantola who critically reviewed earlier drafts of this manuscript. RESUME TECHNIQUES D'ENQU-TE CLINIQUE EN VUE D'ESTIMER LA PRtVALENCE ET L'INCIDENCE ANNUELLE DE LA POLIOMYELITE DANS LES PAYS EN DEVELOPPEMENT Des observations recentes ont montre que la poliomyelite est un important probleme de sante publique dans plusieurs pays en developpement, malgre le faible nombre de cas notifies. Le present article decrit des techniques d'enquete fond&es sur l'identification d'enfants boiteux, destin&es A aider les pays en developpement A determiner l'ordre de priorite qu'ils doivent accorder aux programmes de vacci- nation contre la poliomyelite. Dans les pays en developpement, la plupart des cas de poliomyelite surviennent avant l'age de 5 ans. Des enquetes cliniques sur la claudication chez les enfants de plus de 5 ans peuvent donc servir A estimer l'incidence de cette maladie. Une claudication provoquee par une paralysie flasque avec atrophie musculaire mais sans diminution de la sensibilite, ayant eu un debut brusque, et non progressive, est tres probablement due A la poliomyelite. En divisant le nombre de cas de boiterie attribuables A la poliomyelite par le nombre d'enfants enquetes, on obtient un taux de prevalence des sequelles paralytiques de la polio- myelite dans le groupe d'Age etudie. Le chiffre ainsi obtenu doit etre multiplie par 1,33 pour tenir compte des enfants morts de la maladie ou ayant completement gueri, puis encore par 1,25 pour tenir compte des enfants chez qui la paralysie n'interesse pas les jambes, ce qui donne un taux de prevalence pour l'ensemble des cas de poliomyelite. En divisant ce taux par 5, on obtient une estimation du taux d'incidence parmi les enfants de 0 A 4 ans, car on suppose que tous les cas ont e contractes au cours des cinq premieres ann&es de la vie. Finalement on peut obtenir un taux d'incidence annuelle pour la population totale en multi- pliant le taux d'incidence dans le groupe d'Age 0-4 ans par la proportion d'enfants de cet age dans la population totale. 617 618 F. M. LAFORCE ET AL. Les enqutes visant A identifier les enfants atteints de claudication due A la poliomyelite peuvent etre conduites parmi les ecoliers ou sur la base du porte A porte. La premi&e methode demande moins de ressources mais dans les regions ou la scolarisation est inferieure A 900%o, elle doit etre completee par une recherche des enfants boiteux parmi ceux qui ne frequentent pas l'ecole. Trois &oles urbaines et trois &coles rurales doivent etre choisies au hasard dans des listes de toutes les &coles urbaines et rurales de la zone cible. 11 faut s'efforcer d'obtenir des donnees concernant 2000 enfants de chaque &ole ou groupe d'ecoles. Au cours de la premiere visite A l'&cole, on explique l'objet et les methodes de l'enqute et l'on demande aux enseignants de remplir des formules fournissant des renseignements sur la totalite des enfants de leur classe et d'identifier les enfants boiteux par leur nom. Au cours d'une deuxieme visite, ces enfants sont pries d'etre presents, avec leurs parents si possible, et les caracteres cliniques de la claudication ainsi que l'histoire de son debut sont enregistres. Lorsqu'il est necessaire de rechercher les cas chez des enfants ne fre- quentant pas l'&ole, il convient de le faire parmi les popula- tions des secteurs desservis par chacune des ecoles comprises dans l'enquete. Bien que la methode du porte A porte puisse etre utilisee, il suffit parfois de rechercher specifiquement les enfants boiteux d'apres des renseignements obtenus aupres de personnes bien informees dans la collectivite et notamment des renseignements fournis par les enfants boiteux qui frequentent l'eole. Les enqu&es faites par la methode du porte A porte apportent des donnees plus fiables que les enquetes dans les &oles. II convient de choisir trois secteurs urbains et trois secteurs ruraux, au hasard, a P'interieur de la population cible et de reunir des renseignements sur 2000 enfants de chacun d'eau, ainsi que des donnees cliniques et demo- graphiques concernant tous les enfants consid&es comme boiteux A la suite d'une poliomyelite. Les centres de readaptation peuvent, dans certaines regions, fournir des donn&es capables de completer ou de remplacer celles des enquetes dans les &oles ou par le porte A porte. II est probable que de telles donnees seront plus sensibles A des changements recents dans l'incidence de la poliomyelite que celles des enqutes, et plus pr&cises que ces dernieres en ce qui concerne le diagnostic et les dates du debut. Dans certains pays, les gu&risseurs traditionnels peuvent constituer egalement une source de renseignements. Une methode simple et peu couteuse pour evaluer l'impact d'un programme de vaccination antipoliomyelitique consiste A etablir, au moment ou le programme de vacci- nation debute, un registre de la claudication. I1 faudra pro- bablement deux ou trois ans pour mettre au point ce systeme de registre, qui doit etre capable de fournir la docu- mentation sur la diminution des cas de boiterie, prevue parmi les enfants de cinq ans, trois ou quatre ans apres le commencement d'un programme efficace. REFERENCES I. PAUL, J. R. Endemic and epidemic trends of poliomyelitis in Central and South America. Bulletin of the World Health Organization, 19: 747-758 (1958). 2. NICHOLAS, D. D. ET AL. Is poliomyelitis a serious problem in developing countries?-The Danfa experi- ence. British medical journal, 1: 1009-1012 (1977). 3. OFUSU-AMAAH, S. ET AL. Is poliomyelitis a serious problem in developing countries? Lameness in Ghanaian schools. British medical journal, 1: 1012-1014 (1977). 4. GUYER, B. ET AL. Surveillance de la poliomyelite de forme paralytique a Yaounde, Cameroun 1973-1975. Afrique medicale, 15: 697-704 (1976). 5. Expanded Programme on Immunization-Burma. Weekly epidemiological record, 52: 145 -146 (1977). 6. Survey for residual paralysis-Egypt. Weekly epidemiological record, 52: 269-271 (1977). 7. Expanded Programme on Immunization. Poliomyelitis in Indonesia. Weekly epidemiological record, 54: 177-178 (1979). 8. Expanded Programme on Immunization-Philippines. Weekly epidemiological record, 53: 144-146 (1978). 9. Expanded Programme on Immunization. Poliomyelitis in Thailand. Weekly epidemiological record, 54: 202-203 (1979). 10. PAUL, J. R. ET AL. A survey of neutralizing antibodies to poliomyelitis virus in Cairo, Egypt. American journal of hygiene, 55: 400-413 (1952). 11. PAUL, J. R. Epidemiology of poliomyelitis. In: Debre, R., ed., Poliomyelitis, Geneva, World Health Organ- ization, 1955, pp. 9-30 (Monograph Series No. 26). 12. DEBRE, R. & THIEFFRY, S. Symptomatology and diagnosis of poliomyelitis. In: Debre, R., ed., Polio- myelitis, Geneva, World Health Organization, 1955, pp. 109-136 (Monograph Series No. 26). 13. GUPTA, A. K. & GULATI, S. K. Poliomyelitis in Kanpur during the year 1964. Journal of the Indian Medical Association, 51: 230-233 (1968). 14. BADA, J. L. Paralytic poliomyelitis in Monrovia (years 1964-67). Journal oftropical medicine and hygiene, 71: 212-215 (1968). 15. COLLINS, W. R. F. ET AL. Poliomyelitis in Nigeria. The West African medical journal, 10: 217-222 (1961). 16. BASU, S. N. Epidemiology of paralytic poliomyelitis. Indian journal ofpediatrics, 3: 62-72 (1966). 17. ARORA, R. R. ET AL. Epidemiology of poliomyelitis in Delhi. Indian journal of medical research, 67: 11-18 (1978). ESTIMATING PREVALENCE OF POLIOMYELITIS Form 1. School lameness survey- Poliomyelitis Date: Evaluator: Principal: Ages of children Class size Male Female Total Name of teacher Catchment area/area from which students are drawn (rough map on reverse) Listing of all lame children by class and teacher 1 2 3 4 5 6 7 8 9 10 11 12 It is very important that the above lame children be available for examination on If possible one or both parents should also be present. Country: Region: District: School: Class No. 1 2 3 4 5 6 7 8 Yes No D D D D II D LI L LI L L L L L L LI 619 F. M. LAFORCE ET AL. Form 2. Clinical data -lame children Country: Region: District: Evaluator: School Yes Cl No g Date: If YES: Class No. Teacher Name: Sex: Dateol Date of birth: Residei Place of birth: Age at, Care: Physician Health centre Traditional healer Other Clinical: Affected leg: R L Character of paralysis: 1. flaccid 2. intact sensation 3. atrophy 4. acute onset 5. progression Degree of disability: f onset: nce at onset: s onset: Acupuncture both Yes Yes Yes Yes spastic No No No No I pronounced but can walk without mechanical aid 11 walk with prothesis or stick Ill unable to walk Final diagnosis (circle): 1. Poliomyelitis 2. Trauma 3. Other Examiner 620 Date