a Department of Epidemiology Research, Danish Epidemiology Science Centre, Statens Serum Institut, Copenhagen, Denmark.
b Projecto de Saúde de Bandim, Bissau, Guinea-Bissau.
Nete Munk Nielsen, Department of Epidemiology Research, Danish Epidemiology Science Centre, Statens Serum Institut, Artillerivej 5, 2300 Copenhagen S, Denmark. E-mail: NMN{at}SSI.dk
Abstract
Background According to the polio model, severity of disease increases with age at infection. Firstborn children and people belonging to small families are generally infected later and should accordingly have a higher risk of severe polio. However, this model does contradict other explanations of severity of childhood infections including the intensive-exposure model.
Methods To evaluate the deductions from the polio model we performed a study based on medical records from 5590 historical polio cases from the county of Copenhagen 19401953. The relative risk (RR) of polio according to age, birth order and sibship size was evaluated using census data from 1940 and 1950.
Results Severity of polio measured as frequency of paralysis or mortality did not show a steady increase with age, but a U-shaped curve being highest for the youngest as well as the oldest patients. The incidence of polio and paralytic polio was higher in families with several children compared with single children (RR = 1.13, 95% CI : 1.01.3). Furthermore, the incidence was higher in laterborn children (Ptrend < 0.0001). However, as predicted from the intensive-exposure model, secondborn children aged 14 years in two-child families had a higher risk of paralytic polio than firstborn children (RR = 1.47, 95% CI : 1.12.0), whereas the opposite relationship was found for those aged 5 years (RR = 0.65, 95% CI : 0.50.9).
Conclusion The polio model's prediction about the impact of age, sibship size and birth order on polio incidence and severity found only limited support. A model emphasizing intensity of exposure as a risk factor for severity may account better for the epidemiology of polio infection.
Keywords The polio model, severity, risk factors, intensity of exposure
Accepted 11 May 2001
During the first part of this century poliomyelitis or infantile paralysis became increasingly severe and marked by distinct epidemics separated by intervals of approximately 10 years until polio was controlled by immunization in the 1950s and 1960s. The majority of polio virus-infected individuals develop no symptoms or only unspecific symptoms, i.e. sore throat, nausea, vomiting, and headache. Less than 2% of those infected develop non-paralytic polio with specific symptoms such as severe headache, pain and stiffness of neck, back, and leg. Even more rare is paralytic polio characterized by muscle pain and flaccid paralysis.1
The variation in clinical symptoms and the shift from endemic to epidemic presentation has puzzled epidemiologists. To explain this pattern, it has been postulated that in the pre-epidemic era most infections occurred in infancy, when the case/infection ratio is very low, and very few paralytic cases were thus observed. Due to changes in sanitation, personal hygiene, family size and crowding during the present century, children were no longer exposed to polio virus during infancy, and many remained susceptible as teenagers and adults. Becoming infected as a teenager or adult seemed to increase the risk of developing paralysis.2,3 According to this so-called polio model, laterborn children should be exposed early in life and acquire immunity without paralysis whereas firstborn children are exposed later in childhood and therefore have a higher risk of developing paralysis.49 Furthermore, age at infection is lower in large families and the risk of paralytic polio should therefore be lower in such families.10,11 Belonging to a lower social class usually means having a large family and living under less favourable conditions, resulting in early exposure and reduced risk of being infected with severe polio, and poliomyelitis is therefore believed to be a middle and upper class phenomenon.12
The polio model is often used to explain the epidemiology of diseases of unknown aetiology, suggesting that these may be rare complications of common infections, which have changed age distribution in modern society. For example, exposure at older ages to an otherwise harmless virus has been suggested as a possible cause of chronic diseases such as Hodgkin's disease, ulcerative colitis, Crohn's disease and multiple sclerosis.1015 However, studies have been inconsistent.16
Recent studies have shown that the severity of many viral infections such as measles and chickenpox, seems to be dependent on the intensity of exposure rather than just on age at infection.1719 Children attending kindergarten or school are usually infected outside the home (index cases). Parents, older and younger siblings are more likely to be infected as secondary cases at home by a sick sibling or a child. Infected outside the home means being less intensively exposed to the infecting agent, presumably due to a lower dose of virus.18 Due to more intensive exposure within the home, secondary cases absorb a larger dose of virus and experience a more severe progression of infection. As a result of the process of disease transmission, the proportion of secondary cases and severity by age will reveal an U-shaped curve, and not as predicted by the polio model, a curve increasing with age. Furthermore, being a member of a large family should increase the risk of becoming a secondary case, whereas only children should be more likely to become index cases. Accordingly severity should increase by family size and not decrease as predicted by the polio model.
There are several reasons to believe that intensity of exposure may influence the severity of poliomyelitis. Sabin20 performed studies with Cynomologus monkeys according to which a higher dose of polio virus increased the risk of developing paralysis. Furthermore, previous studies have noted the very high incidence of poliomyelitis not only among the old but also among the youngest children.4,2124 The assumption in the polio model about case/infection ratio increasing with age has not been reproduced in animal polio models, as the youngest animals usually have the most severe infection.25 Finally, according to the polio model, paralytic polio should be very rare in developing countries. However, more thorough studies have found very high polio incidences in developing countries, indicating that the dogma about the absence of paralytic poliomyelitis in the tropics was based on incomplete reporting.26
We have therefore decided to test the deductions of the polio model regarding the impact on severity of age at infection, family size, birth order and social status using the register of hospitalized polio cases from Copenhagen.
Subjects and Methods
The present study is based on hospital records from Blegdamshospitalet in Copenhagen. From 1876 until the hospital was closed in the middle of the 1970s, this was the main infectious disease hospital in the city of Copenhagen.
Polio cases
More than 80 000 consecutive records from the period 1940 to 1953 were reviewed, leading to the identification of 5839 cases of polio patients. However, the present study only included polio patients (n = 5590) living in the county of Copenhagen. Only cases with a discharge diagnosis of non-paralytic polio, paralytic polio, suspected polio or primary lymphocytic meningitis were included in the analysis. Occurring during the same season as polio, primary lymphocytic meningitis was generally considered to be non-paralytic polio, after other possible viral infections such as mumps had been excluded according to the clinical symptoms.27
Besides standard information about name, address, age, family relation and provider's occupation, the hospital records contained detailed information about symptoms, treatment and progression of the disease. The diagnosis was mainly based on a clinical investigation as no paraclinical tests besides lumbar puncture were available. Clinical symptoms of non-paralytic polio were normally fever, headache, vomiting, pain and rigidity of the back and neck. The diagnosis of paralytic polio was verified by the existence of the characteristic asymmetrical flaccid paralysis. Patients admitted as polio cases but discharged with the diagnosis meningeal irritation (normal cerebrospinal fluid) or meningitis were omitted from the analysis, even though many of them might have suffered from non-paralytic poliomyelitis.
Severity was graded as non-paralytic polio (non-paralytic polio, suspected polio or primary lymphocytic meningitis), paralytic polio, respiratory paralysis (given artificial respiration), and death. All polio patients were coded according to municipality of residence and for children, social status was classified according to occupation of the provider as described elsewhere.28
To evaluate the completeness of the material we compared the number of polio cases registered by the National Board of Health in the county of Copenhagen with the number of patients from the county of Copenhagen. Overall 100% of the fatal cases, 85% of the registered paralytic cases, and 92% of the non-paralytic cases were treated at the Blegdamshospital from 1940 to 1953. However, many non-paralytic cases may not have been reported to the National Board of Health.
Census data
To evaluate the relative risk (RR) of symptomatic polio according to birth order and sibship size, we obtained information on the distribution of birth order and sibship size in the population of the municipality of Copenhagen (census from 1950).29 The evaluation of RR of poliomyelitis is therefore restricted to polio cases living in the municipality of Copenhagen and not the whole county. Census data on the distribution of sibship size according to age allowed us to evaluate the RR of symptomatic polio according to sibship size in three age groups (06, 714, 1520 years). Similar information was not available for birth order. Therefore, in order to evaluate the RR of polio according to birth order in different age groups, we used the original registration forms from the 1940 census. We would have preferred to use the data from the 1950 census, but the original data from this census are not yet available to the public. In the 1940 census, every household had to fill in a form providing information on date of birth, family status and occupation for all family members. We randomly selected 120 streets in Copenhagen and studied registration forms from families living in these streets. For all two-child families with at least one of two children under 15 years of age, we tabulated the children according to age and birth order. A total of 3805 children from two-child families were identified from the 120 streets. According to the census of 1950, there were in total 80 260 children under the age of 15 years in the municipality of Copenhagen who had just one sibling. We used the sample from the 1940 census to estimate the age and birth order distribution of the 80 260 individuals in two-child families.
Statistical methods
The RR of polio and paralytic polio shown in Tables 2 and 4 were estimated using log-linear Poisson regression.30 Using this approach the CI do not depend on the population size. Test for trends in Table 2
was performed with family size and birth order treated as continuous variables. Test for interaction in Table 2
was performed with age and family size categorized as shown in the Table. Test for trends in Figure 1
was performed using logistic regression with age treated as a continuous variable using medians for each age group. In a case-case analysis within the group of polio patients, we calculated the relative case-fatality rates of paralytic polio according to birth order, sibship size, and social class (Table 3
) using log-linear binomial regression.31 Adjustment was made for age (04, 59, 1014, 1519 years), gender, epidemic period (19401949, 19501953), and social class (high, low). All regression analyses were performed with the SAS procedure PROC GENMOD.32
|
|
|
|
Among the hospitalized polio cases, the risk of developing paralysis was highest for the youngest children and the adults (Table 1, Figure 1
). A U-shaped age pattern was observed in both 19401949 and 19501953. Fifty-three per cent of the children less than 2 years old had paralytic polio decreasing to 20% among those aged 89 years (P < 0.0001) increasing again to 43% among people over 35 years (P < 0.0001) (Table 1
). The type of paralysis varied markedly by age; respiratory difficulties being more frequent among children aged 01 year decreasing to those aged 45 years, and increasing again thereafter (Table 1
). The case-fatality ratio varied in a similar way, being lowest among children aged 45 years (Table 1
).
|
Children hospitalized with polio belonging to the three highest social classes (landowners, academics, farmers, tradesmen, officials) did not have a higher mean age at infection than children belonging to the lowest social classes (6.62 versus 6.62 years) (Table 3). Restricting the analysis to paralytic children, children from the highest social classes were slightly younger than children from the lowest social classes (5.42 versus 5.62 years, P = 0.6). Adjusted for age, gender and period we found no increased risk of dying of paralytic polio for the 389 children belonging to the three highest social classes compared with 967 children of skilled and unskilled workers (RR = 0.94, 95% CI : 0.61.5). However, children from the three highest social classes did have a slightly increased risk of developing paralytic polio compared to children from the two lowest social classes (RR = 1.14, 95% CI : 1.01.3) (Table 3
).
Table 4 shows the risk of being hospitalized with paralytic polio in firstborn compared to secondborn children in two-child families. Controlling for age, the youngest had higher risk than the oldest child among children aged 14 years (RR = 1.47, 95% CI : 1.12.0), whereas the opposite relationship was found for those aged
5 years (RR = 0.65, 95% CI : 0.50.9). The same tendency was seen for all polio cases (non-paralytic + paralytic).
Discussion
The main observations in the present analysis were the U-shaped curve for paralysis and increasing severity of poliomyelitis by increasing family size. Furthermore an interaction between age and relative birth order was seen. Being lastborn and under 5 years in a two-child family increased the risk of developing poliomyelitis, whereas firstborns older than 5 years, had a higher risk of developing polio. This pattern implies that children aged 45 years presumably were important in the transmission of polio in Copenhagen. All these observations are consistent with the intensive-exposure model.
However, in line with the polio model, a steep increase in severity among young adults was observed. This appears, to be associated with a simultaneous change in the pattern of paralysis, with respiratory paralysis being much more common among older individuals.7 While this could be related to unidentified age-related changes in susceptibility or immune response, it could also be related to differences in the mode of virus transmission for adults and children. The faecal-oral route of transmission is usually considered the main mode of virus spread, but as virus is present in the pharynx, droplet spread from person to person is possible.23 Better sanitary practices might increase the importance of infection by inhalation of droplets.33 Accordingly adults might more often be infected by droplet spread. Furthermore it has been suggested that bulbar symptoms might result from infection with the virus through the tonsillo-pharyngeal route, whereas gastro-intestinal infection might initially result in paralysis of the legs.34 Thus the higher case fatality among adults could possibly be related to the route of transmission. Similarly, the increasing importance of number of siblings with increasing age (Table 2) could also indicate that intensity of exposure and crowding played a larger role for respiratory transmission than for faecal-oral transmission.
There may well have been non-paralytic polio cases who were not hospitalized and not registered in official health statistics. However, it seems unlikely that many paralytic cases were not registered. As more than 85% of paralytic polio cases and all fatal polio cases registered in the 19401953 period were hospitalized and included in the study, a recruitment bias is unlikely to have had a major impact on our observations. Since diagnostic problems with non-paralytic cases could have introduced a bias, it should be emphasized that most tendencies remained the same if we analysed paralytic cases only.
A limitation of the present study was our inability to determine who is susceptible and who is immune. It is likely that more individuals in large families would be immune due to a higher risk of previous exposure, and therefore the relative increase in clinical disease would have been even larger in families with several children. This co-variation between number of children and degree of immunity may explain the lack of clear difference in incidence for children aged 06 years in families with 2, 3 and 4 children, respectively (Table 2). Young children in large families may be at lower risk because the older siblings are likely to already have had the infection. On the other hand, older children in large families are likely to have younger siblings in the susceptible age group, explaining the increasing incidence of polio among older children in large families.
If, indeed, intensive exposure is a critical factor, part of the reason for the changing epidemiology of polio in this century may have been a gradual improvement of hygienic conditions, transforming a disease from endemic to epidemic with longer intervals between outbreaks. With longer intervals between epidemics, increasing accumulation of simultaneously susceptible individuals will occur in a family, thereby increasing the risk of intensive exposure and intra-family spread. It is a corollary of this process that severity increases as the age at infection goes up because a high median age at infection reflects the presence of many susceptible individuals in the same family.35 Using this model, measles case fatality has been shown to be highest in isolated communities with a high age at infection and accumulation of susceptibles.36 In the case of polio, this pattern may have been further aggravated if there was a simultaneous increase in the risk of respiratory transmission and bulbar paralysis among older individuals. Only from this exposure model is it possible to understand both the age pattern in severe polio as well as the fact that indicators of crowding, including family size and birth order, are risk factors for severe polio infection.
Polio may soon be eradicated and interest in the disease will therefore decline. However, polio was one of the major infections shaping public health in the 20th century and it may still be important how we understand the epidemiology of polio and the history of infectious diseases. Previous attempts to understand the epidemiology and severity of polio gave rise to the polio model, which has shaped the analysis of the interaction between childhood infections and chronic diseases. If the polio model has only limited validity for polio, there may be a need to re-consider our understanding of how infections are linked to chronic conditions.
Acknowledgments
This study was supported from the Danish Medical Research Council, the Danish Development Research Council, the Danish National Research Foundation and The National Polio Society (PTU). We are very grateful for the assistance from the staff at Copenhagen City Archives and the Danish National Archives.
References
1 Horstmann DM. Epidemiology of poliomyelitis and allied diseases. Yale J Biol Med 1963;36:526.[ISI][Medline]
2 Olin G. The epidemiologic pattern of poliomyelitis in Sweden from 1905 to 1950. In: Poliomyelitis; Papers and Discussion. The Second International Poliomyelitis Conference 1952. Philadelphia: Lippincott, 1952, pp.36775.
3 Bodian D, Horstmann DM. Polioviruses. In: Horsfall FL (ed.). Viral and Rickettsial Infection in Man. Philadelphia: Lippincott, 1965, pp.43073.
4 Nathanson N, Martin JR. The epidemiology of poliomyelitis: enigmas surrounding its appearance, epidemicity, and disappearance. Am J Epidemiol 1979;110:67292.[ISI][Medline]
5 Paul JR. A History of Poliomyelitis. New Haven: Yale University Press, 1971.
6 Klingman J, Chui H, Corgiat M, Perry J. Functional recovery: a major risk factor for the development of postpoliomyelitis muscular atrophy. Arch Neurol 1988;45:64547.[Abstract]
7 Weinstein L. Influence of age and sex on susceptibility and clinical manifestations in poliomyelitis. N Engl J Med 1957;257:4752.[ISI][Medline]
8 Gruffermann S, Delzell E. Epidemiology of Hodgkin's disease. Epidemiol Rev 1984;6:76106.[ISI][Medline]
9 MacMahon B. Is acute lymphoblastic leukemia in children virus related? Am J Epidemiol 1992;136:91624.[Abstract]
10 Gilat T, Hacohen D, Lilos P, Langman MJS. Childhood factors in ulcerative colitis and Crohn's disease. An international cooperative study. Scand J Gastroenterol 1987;22:100924.[ISI][Medline]
11 Newell GR, Mills PK, Johnson DE. Epidemiologic comparison of cancer of the testis and Hodgkin's disease among young males. Cancer 1984;54:111723.[ISI][Medline]
12 Gutensohn N, Cole P. Epidemiology of Hodgkin's Disease in the young. Int J Cancer 1977;19:595604.[ISI][Medline]
13 Gutensohn N, Cole P. Childhood social environment and Hodgkin's disease.N Engl J Med 1981;304:13540.[Abstract]
14 Newell GR. Etiology of multiple sclerosis and Hodgkin's disease. Am J Epidemiol 1970;91:11922.[ISI][Medline]
15 Nathanson N, Miller A. Epidemiology of multiple sclerosis: critique of the evidence for a viral etiology. Am J Epidemiol 1978;107:45161.[ISI][Medline]
16 Jarrett RF. Viruses and Hodgkin's disease. Leukemia 1993;7(Suppl.): S7882.[ISI][Medline]
17 Aaby P, Coovadia H, Bukh J et al. Severe measles: a re-appraisal of the role of nutrition, overcrowding and virus dose. Med Hypotheses 1985;18:93112.[CrossRef][ISI][Medline]
18 Aaby P, Bukh J, Lisse IM, Smits AJ. Risk factors in subacute sclerosing panencephalitis: age and sex dependent host reactions or intensive exposure? Rev Infect Dis 1984;6:23950.[ISI][Medline]
19 Aaby P. Determinants of measles mortality: host or transmission Factors? In: De la Maza LM, Peterson EM (eds). Medical Virology 10. New York: Plenum Press 1991, pp.83116.
20 Sabin AB. Paralytic consequences of poliomyelitis infection in different parts of the world and in different population groups. Am J Public Health 1951;41:121530.[ISI][Medline]
21 Horstmann DM. Poliomyelitis: severity and type of disease in different age groups.Ann NY Acad Sci 1955;61:95667.[ISI][Medline]
22 Dauer CC. The changing age distribution of paralytic poliomyelitis. Ann NY Acad Sci 1955;61:94355.[ISI][Medline]
23 Dömök I. Enterovirus infection: poliomyelitis. In: Cruickshank R (ed.) Epidemiology and Community Health in Warm Climate Countries. London: Churchill Livingstone, 1976, Ch. 13.
24 Siegel MM. Influence of age on susceptibility to virus infections with particular reference to laboratory animals. Ann Rev Microbiol 1952;6: 24780.[CrossRef][ISI][Medline]
25 Lipton HL. Theiler's virus infection in mice: an unusual biphasic disease process leading to demyelination. Infect Immun 1975;11:114755.[ISI][Medline]
26 Sabin AB. Paralytic poliomyelitis: old dogmas and new perspectives. Rev Infect Dis 1981;3:54364.[ISI][Medline]
27 Lassen HCA. Sammenlignende undersøgelser over primær serøs meningitis og paralytisk polio-myelitis. Ugeskr Læg 1939;3:7380.
28 Enevoldsen B, Michelsen N, Friis-Hasch E, Kamper-Jørgensen F. Sociale Klassifikationer. Ugeskr Læg 1980;142:54450.
29 The Statistical Department 1957. Non-Institutional Households Population. Census 1950.
30 Breslow NE, Day NE. Statistical methods in cancer research. Vol. II. The design and analysis of cohort studies. IARC Sci Publ 1987;82:1406.
31 McCullagh P, Nelder JA. Generalized Linear Models. London: Chapman & Hall, 1989.
32 SAS Institute Inc. SAS Technical Report P-243. SAS/STAT Software: The GENMOD Procedure, Release 6.09. Cary, NC: SAS Institute Inc., 1993.
33 Eklund CM, Larson CL. Outbreak of type 3 poliomyelitis on St Paul Island, Alaska. Am J Hyg 1956;63:11526.[ISI][Medline]
34 Swan C. The anatomical distribution and character of the lesions of poliomyelitis. With special reference to the type of cell affected and to portal of entry of the virus. Aust J Exp Biol Msc 1939;17:34564.
35 Aaby P. Malnutrition and overcrowding/intensive exposure in severe measles infection. Review of community studies. Rev Infect Dis 1988; 10:47891.[ISI][Medline]
36 Pison G, Aaby P, Knudsen K. Increased risk of death from measles in children with a sibling of the opposite sex in Senegal.Br Med J 1992; 304:28487.[ISI][Medline]