Problem posing:According to a number of psychiatrists, the decrease in the number of suicides can almost exclusively be ascribed to the increasing use of new antidepressants (ADs). Several ecological studies have been carried out to lend support to this claim; unfortunately, many of these started out from either methodologically or statistically flawed assumptions. The purpose of the current study is to demonstrate the examined relationships using complex time-series techniques on our national sample.
Method: When investigating the relationships between our time series, first we ensured their stationarity using several methods. We used two methods for the analysis involving several independent variables.
Results: The models fitting our data showed somewhat mixed results. When using dynamic regression to ensure stationarity, the residuals of the suicide and AD time series showed a significant negative correlation. At the same time, when using the more robust technique of time series differentiation, the stationary time series showed no significant relationship between the use of antidepressants and suicide rates. Moreover, those regression models in which several independent variables were entered to control for the AD data showed no significant relationship between the variables either. A further detail is provided by the fact that according to our calculations, suicide rates are not Granger caused by the amount of antidepressants sold. However, the vagueness of ecological models is well demonstrated by the fact that even those sociological variables (number of divorces, alcohol consumption) failed to show a significant relationship with suicides here, which are usually significant in analyses using micro data.