Previous attempts to decide whether there is significant year-to-year variation in the birth rate of people who subsequently develop schizophrenia have given conflicting results, probably because of differences in the statistical methods employed.

To determine whether there is significant year-to-year variation in the birth rate.

Variation in the birth rate for the period 1921–1960 was studied in three separate national data sets – English, Danish and Scottish – using cubic splines to smooth the distribution curve before calculating residuals from a Poisson distribution.

Over-dispersion was found in all three data sets, particularly in the Danish and Scottish data. However, the correlation between the sets of standardised residuals derived from the three data sets was only statistically significant for Denmark v. England.

There was statistically significant year-to-year variation in the birth rate of people who subsequently developed schizophrenia in three countries in north-west Europe in the years 1921–1960. This is potentially a clue to the nature of the environmental determinants of schizophrenia, but better data will be needed before useful explanatory hypotheses can be generated and tested.