Epidemiologic Study of Infants With Biliary Atresia

Abstract

To the Editor. Yoon et al1 present an epidemiologic study of 57 infants with biliary atresia born over a 25-year period in the Atlanta area. This data was analyzed using seven variables including month of birth, from which a predilection for the winter months December through March was shown.King's College Hospital, London, is a tertiary referral unit for hepatobiliary diseases predominantly but not exclusively from the south of England. We have specialized in the treatment of biliary atresia since the 1970s and our experience has been reported recently both in general2 and specifically with the biliary atresia splenic malformation syndrome (or sequence).3 Since 1973, we have treated 380 infants with biliary atresia (nonsyndromic, n = 346; syndromic, n = 34) and have data on their month of birth. Using Yoon's own criteria for season (April through July, August through November, and December through March), we found neither predilection nor trend for season (χ2 = 0.56, df = 2:P = 0.75) in these two groups. The Figure illustrates our cases by month of birth in the two types of biliary atresia.Although our series is not from a well-defined area, is heterogenous (but predominantly white), and has no data on the population from which the cases arose, it has the irrefutable advantage of being large. We feel that it is too easy in such studies to infer statistical relationships where none actually exist. A statistical test for seasonality was applied to an average of only 4.7 cases per month and this is clearly a major weakness. Indeed, the wide confidence intervals quoted support this interpretation. The evidence for a viral etiology of biliary atresia is contradictory and, it is true, would be advanced if seasonality could be shown. However, this is not the study that has shown it.In Reply. Using data from a hospital-based case series of patients with biliary atresia, Drs Davenport and Dhawan show that there is no seasonal variation in the births of these patients with biliary atresia. However, in order to meaningfully examine seasonal variation, one needs to look at variation in birth prevalence rates because seasonal variation in births has been shown to exist in most populations.1-1 Drs Davenport and Dhawan may find that because seasonal variation is likely in the population (denominator), the lack of seasonality in the births of their case infants (the numerator) will result in prevalence rates for biliary atresia that are seasonal. In our study,1-2 we used the Walter and Elwood test for seasonality of events with a variable population at risk1-3 to show that there was seasonal variation in the birth rates of infants in metropolitan Atlanta with biliary atresia. To determine the magnitude of the seasonal variation and whether the difference we found was significant, we divided the year into three seasons and used these in a multivariate logistic regression model. Drs Davenport and Dhawan suggest that because the number of cases we studied was small and our confidence limits were wide, we should not infer a statistical relationship between the occurrence of biliary atresia and season. However, we would like to point out that confidence intervals account for the statistical uncertainty, which is related to sample size and any result with a confidence interval that excludes 1.0 is by definition statistically significant.1-4 Our study, which clearly showed that the occurrence of biliary atresia in metropolitan Atlanta was 2.6 times greater from December through March than from April through July, had a 95% confidence interval of 1.2 to 5.9 even when we controlled for other covariates.We would also like to point out that in addition to geographic and climatic variation, the metropolitan Atlanta population differs considerably from the British population with respect to race and other socioeconomic factors. All of these factors could affect the susceptibility of fetuses and infants to biliary atresia or to an environmental agent that might precipitate biliary atresia. Although much can be said for large sample sizes, population-based studies such as we conducted have a number of advantages over hospital-based studies, particularly when it comes to using study estimates to make inferences about the population at large. In summary, although the data presented by Drs Davenport and Dhawan are interesting, we believe the findings from our study support further investigation of the association between biliary atresia and environmental agents such as viruses.