A number of writers have published data suggesting that schizophrenics are particularly likely to have been born in the winter months. Lewis and Griffin have argued that this "seasonally" effect may be an artifact resulting from (a) the use of inappropriate expected values in the calculation of chi-squares in previous studies; (b) the age-incidence effect-the tendency of risk to increase with time in disorders whose incidence increases with age; and (c) the ageprevalence effect-the fact that people born early in a reporting period have been at risk for any disease longer than those born later in that period. We studied the birth months of 3,556 schizophrenics at a Minnesota Veterans Administration hospital before and after instituting corrections for year-to-year across-month variations in birthrates in our expected values and the age-prevalence bias toward the January-March seasonality effect described in some earlier studies. Finally, we reanalyzed our data on a subset of patients in whom the age-incidence effect should be minimal. Even after these corrections the results supported the contention that the winter birthrate for schizophrenics is excessive, at least in severe climates.More than a dozen papers have appeared describing seasonality effects in schizophrenia-reported tendencies for an excessive number of schizophrenic births to occur in the winter months. This research is particularly important because it suggests that season-related factors-winter cold, summer heat, infectious diseases, prenatal nutritional deficiencies, and so forth-may play a role in the development of schizophrenia.Recently, Lewis and Griffin (1981) have argued that the seasonality effect may be an artifact resulting from three weaknesses in the designs of studies in which it has been reported. First, they point out (as had Hare, 1975) that the expected values used in the chi-squares employed to determine whether the frequency of schizophrenic births in the winter months is proportionately higher than that for the general population are typically based on a pool of many years and do not

No abstract

No abstract