203 General Discussion that increasing the number of psychologists working in primary care local groups did not change GPs’ antidepressant prescription. This finding comes with three additional considerations. First, it should be put in the context of small increases in psychologists available during our study – insufficient to solve the challenge of low psychotherapy supply in Portuguese primary care setting. Second, our method cannot address potential confoundment due to time-variant unobservables. Third, our results changed when studying heterogeneity by local groups with different average levels of psychologists. In this secondary analysis, we found that adding one psychologist/100,000 patients led to statistically significant reductions in antidepressant prescription overtime for the GPs in the lowest and highest quartiles of the average number of psychologists. A potential explanation for the heterogeneous results is the fact that GPs can refer patients to psychological support either as a substitute for antidepressants or in combination therapy. While both approaches are recommended by clinical guidelines depending on patient severity, international literature suggests that most GPs still prefer to use psychotherapy in a combined treatment approach [17-19]. The use of a within-between random effects model provided an additional finding: a measure of association between the different GPs’ antidepressant prescription and the average number of psychologists each GP was exposed to during the study period. This association shows that GPs working in local groups with one-unit higher average number of psychologists/100,000 patients had a lower antidepressant prescription share. In contrast with the results within-GP, this finding is not aimed at causally capturing the effect of psychotherapy supply in prescription, and it is most certainly endogenous to GP and local groups’ characteristics and processes. It contributes to our conclusions by flagging the existence of unmeasured variables that explain why GPs prescribing less work in local groups with higher average number of psychologists. We hypothesise that the reasons lie in the characteristics of the local group, which are primary care governance structures that bring together several GP practices and manage their shared resources. Local groups with more psychologists could be those where mental health services were better integrated and in which local protocols, training or clinical leadership favoured practices reducing the prescription of antidepressants. Interpreting our findings together we conclude that, on average, local groups with more psychologists have their GPs prescribing fewer antidepressant drugs, but hiring psychologists does not necessarily lead to a reduction in antidepressant prescription across all levels of psychologists supply. While follow-up research is essential to confirm these results’ heterogeneous and causal nature, they suggest that reducing antidepressant use requires an in-depth understanding of individual and organisational factors that influence the channelling of increased psychotherapy supply to appropriate antidepressant substitution . 7
RkJQdWJsaXNoZXIy MjY0ODMw