35 Income inequalities beyond access to mental health care Finally, the likelihood of having an additional treatment record was higher for lower income groups, although the differences were of a small magnitude. Compared with the 35.7% of patients in the lowest income quintile who had an additional treatment record, 1.3 pp (95% CI 1.0–1.5) fewer patients had it in the second quintile, 2.2 pp (1.9–2.4) in the third quintile, 2.9 pp (2.6–3.2) in the fourth quintile, and 3.0 pp (2.7–3.3) in the fifth quintile (table 2). Again, the income gradient did not change when adjusting for need and treatment minutes. Table 3 shows how income gradients in functional improvement and additional treatment records differ by reason of closure. Functional improvement was most common when termination occurred by mutual agreement (lowest income quintile mean: 41.1%), followed by patient-side reasons (22.7%), provider-side reasons (20.0%), administrative reasons (15.9%), and finally records terminated after intake, diagnosis, or a crisis episode (8.7%). For all reasons of closure, we observed that patients with higher income were more likely to have a functional improvement. Additional treatment records often existed when records were closed administratively (lowest income quintile mean: 74.7%) but much less frequently when treatment was terminated by all other reasons (6.4–16.3%). Income gradients were small in relative terms and their direction differed depending on the reason of closure. The likelihood of having an additional record decreased with income for records closed by mutual agreement or administratively (small magnitude) but increased with income when records were closed after intake, diagnosis, or crisis episodes. Notably, there were also disparities in the reason of record closure as an outcome (Table A7). Those living in households with higher incomes were more likely to terminate the treatment record by mutual agreement, whereas patients with lower income were more likely to terminate for all other reasons. Figure 1 presents results by diagnosis for depressive disorder, anxiety disorder, and all other disorders. The gradients were overall similar to the main results. An exception is the gradient of treatment minutes for other disorders: treatment minutes were the highest for patients in the lowest income quintile and similar for the other four quintiles. Individuals with depressive and anxiety disorders had higher levels of functional improvement than those with other disorders, across all income quintiles (Table A8). Gender-stratified analyses showed that income gradients were overall similar for women and men, despite women having slightly lower disease severity at baseline and higher functional improvement across all income quintiles (Table A6). Our findings were robust to the inclusion of different sets of demographic covariates, higher levels of primary diagnosis grouping, a proxy for addiction, different types of therapies, and to the modelling of the treatment outcomes stratified by the initial level of GAF and with logistic regression to model binary outcomes (Tables A9 to A15). 2
RkJQdWJsaXNoZXIy MjY0ODMw