168 Chapter 5 Figure 2. Relationship between the share of depressed adults with antidepressants prescription and the average number of psychologists available at the local group level. Notes: The y axis distinguishes between high and low prescribing local groups (average and standard deviation sorted), while the number of psychologists is represented by the shape and colour of the marker (in quartiles). Within-between random effects model Table 2 reports results from the REWB model decomposing the relationship between antidepressant prescription and psychological therapy within and between variations. Different sets of covariates were sequentially added to the estimations, ranging from the unadjusted (1) to the fully adjusted (5) and our preferred specification. Both within and between coefficients in all 5 models suggest an association between higher psychotherapy supply and lower GP prescription of antidepressants. Nevertheless, coefficients for the within and between terms differ considerably in terms of magnitude. On the one hand, the within coefficients are small6 and mostly statistically insignificant at a 5% level, suggesting that GPs are not changing the prescription behaviour overtime by being exposed to one more psychologist/100,000 patients in their local group. On 6 The interpretation of the within coefficient for model 5 would be: the increase in one psychologist FTE per 100,000 patients in the local group where is associated with a decrease of 0.28 pp (CI 95% -0.699, 0.138) in the share of patients identified with depression that are prescribed antidepressant medication. When compared to the average of 26.88% of adults with depression prescribed antidepressants, this reduction translates into a 1.04% decrease.
RkJQdWJsaXNoZXIy MjY0ODMw