75 Patient cost-sharing, mental health care and inequalities cessation vs. treatment initiation, distinguishing the two effects captured in main results: patients that stop care from individuals that do not start treatment but would do so in absence of the deductible increase. Our main objective with this analysis is to understand who are the patients that stop care the most, by studying different subgroups of previous users. It uses the special case of model (1) where all individuals to the left of the age 18 discontinuity are service users (i.e. essentially only using variation in mental health care use from month 228 onwards). In addition to estimates of treatment cessation for all mental health patients at the age of 17, we evaluate subgroup effects of treatment cessation by treatment intensity and medicines used at the same age. For this, we interact age polynomials, intercepts and discontinuities with (a) mental health care intensity – quartiles of expenditure, (b) number of psychotropic drug classes used, and (c) type of psychotropic drug class for those using only one class, all at age 17. Finally, we conduct an additional series of sensitivity analyses to further verify the robustness of our results. We use alternative bandwidths for age, perform a placebo difference-in-discontinuity at age 16, use non-linear models, include postcodeyear fixed effects to allow for geographical patterns changing over the study years, and exclude all controls. All analyses are performed using STATA (StataCorp 2017) and STROBE guidelines for reporting observational studies are followed. RESULTS Demographics Our main study population consists of 1,541,210 individuals. Table 2 shows that the study population is characterised by slightly more males (51.0%) and one fifth of first- or secondgeneration migrants. The proportion of adolescents that used mental health services at 17 is higher for females than males (8.8% vs 7.3%). Specialist care was more frequently received than basic care, and the latter was more frequent for females (2.8% comparing to 1.4%). Gender differences are also observed in the proportion of adolescents using any psychotropic medicine (3.4% for females and 5.3% for males) and the type of drug classes used. From those using only one class of psychotropic drugs, males were almost three times more likely to use ADHD agents, while anxiolytics and antidepressants are twice more common among females. The characterization of the subpopulation of mental health care users at 17 is provided in Table A2 in Appendix E. 3
RkJQdWJsaXNoZXIy MjY0ODMw