173 Antidepressant therapy prescription, do psychologists help? Evidence from Portugal Robustness analysis To exclude the potential effect of an additional psychologist affecting the GP diagnosis capacity, we regressed the incidence rate of depression at the GP level on the local group psychologists’ supply. Additional diagnosis capacity could have implications in our ability to measure changes in antidepressant prescription (numerator), because our prescription share outcome is standardised for the prevalence of depression (denominator). It could also influence the prescription patterns by changing the average severity level of the patients diagnosed with depression. The results suggest that an increased supply of mental health professionals was not associated with changes in GP capacity to diagnose depressive disorders (non-statistically significant coefficients, Table A1 in the appendix). Figure A4 supports this conclusion by providing a visual clue of a non-systematic relationship between the incidence of depression and the number of psychologists in the local group. Secondly, we explored whether the local groups would hire psychologists as a response to (changes in) GPs antidepressant prescription. Table A1 shows non-statistically significant coefficients for both 1 and 2 year-lags of prescription in psychologists supply, addressing reverse causality-type of concerns. Thirdly, the similarity between findings in the main model and the model without potentially endogenous variables (list size and utilization share) in Table A2 suggests that potential bias from time-varying utilization of GP visits and list size is not a salient issue. This supports our choice of a preferred specification with the full set of covariates. Last, we confirmed that our results remain the same when estimated in a balanced panel of GPs over the 4 years (Table A3). DISCUSSION Our study aimed to investigate the relationship between the number of psychologists working in Portuguese primary care and the prescription of antidepressants by GPs between 2015 and 2018. We found no overall effect of increasing one psychologist/100,000 patients on GP prescription overtime (within term). Additional analysis uncovered heterogeneity in this finding: adding one psychologist/100,000 patients led to statistically significant reductions in prescription overtime for the GPs in the lowest and highest quartiles of average psychologist’s supply. Our study also found that GPs working in local groups with one-unit higher average number of psychologists/100,000 patients have a lower antidepressant prescription share (between term). We thus conclude that on average local groups that have more psychologists have their GPs prescribing less antidepressant drugs, but hiring psychologists do not necessarily lead to reduction in antidepressant prescription across all levels of psychologists supply. The larger magnitude and the significance of the between estimate, compared with the within one, suggests a relevant role of unmeasured characteristics and processes that bring together GPs prescribing less and local groups with higher average number of psychologists. 5
RkJQdWJsaXNoZXIy MjY0ODMw