The effects of the Chernobyl explosion on induced abortion in Italy

Four regression models have been fitted to data of the monthly number of induced abortions in Italy between January 1984 and April 1986, in order to predict the number which would have occurred in the 5 months following the Chernobyl explosion. In model I the average number of abortions per day in each month was the dependent variable and calendar months, a linear time trend and previous month's value were the independent variables. Model II included a quadratic time trend term in addition to the independent variables used in model I. Models III and IV were like models I and II except that the dependent variable was the average number of abortions per working day in each month and the effect of the previous month's value was omitted. The 4 models all implied that an excess number of abortions were performed in the 5 months following the Chernobyl accident. The mean daily excess was estimated to be 28 and 52 per day for models I and II and the mean excess per working day was estimated to be 20 and 30 by models III and IV, respectively. Clearly the estimated magnitude of the excess depends on whether the quadratic time trend is included among the explanatory variables, but these results imply that the excess is unlikely to be merely due to chance.