82 Chapter 3 care [14, 42] We also find larger impacts of increased deductibles on low-income groups. While existing studies report mixed evidence of the effects of cost-sharing across income [24, 33, 34, 40, 42, 43], our results are in line with other studies in the Dutch context. Remmerswaal and colleagues, report that adults living in low-income areas reduce their consumption of “physical” health care more when facing a deductible versus a rebate, but no difference is seen in high-income areas [42]. Ravesteijn et al. studied the introduction of co-payments for specialist mental health treatment and finds a greater reduction in demand for Dutch adults living in low-income neighbourhoods [40]. Such differential impact was not identified by Lambregts and van Vliet when studying the same reform but measuring socioeconomic status with an aggregated score based on neighbourhood characteristics in terms of income, education and employment [33]. Our findings instead rely on individual-level measures of income, and are in line with two studies observing equivalent measures for Swedish young adults around the age of 20. These studies found that removing co-payments led to an increase in the number of outpatient medical appointments [34], while modest co-payments decreased the number of primary care visits [24]. In both cases, effects were substantially larger for low-income groups, and in the later also for females. Limitations The difference-in-discontinuity approach relies on the assumption that the effect of the deductible increase at age 18 can be isolated from other discontinuous changes at the same age, provided the latter are constant over the study period. While it is a limitation that this assumption is in principle untestable, several features of our modelling strategy and sensitivity analyses confirm the robustness of our findings. We found no indication of time-varying anticipation or time-varying confounding in terms of moving out of the parental house, personal income, or confounding trends in mental health care use during the low deductible period. Furthermore, the combination of year dummies with the difference-in-discontinuity design allows for changes in the mental health care sector that influence levels of care used by both adolescents and young adults (appendix A). Waiting times for specialist AMHS were also stable during our study period [44]. Secondly, we cannot distinguish between over- and underutilization of services since we lack a detailed measure of mental health need. If the increase of the deductible only reduced non-essential mental health care use, effects on mental health may have been limited. However, overuse of mental health services by youth is unlikely as there is often stigma around using mental health care [31], and since we find considerable discontinuation for those labelled as high-intensity users. Third, our data does not record information on support provided by GPs and mental health nurses in primary care. We cannot exclude increased substitution between mental health services and GPs support in high deductible years, especially given the increasing numbers of mental health nurses in primary care [45] and the deductible exemption for this type of care. Although substitution can be clinically appropriate in some cases, financially driven substitution is a concern given the limited scope of primary care to treat mental disorders in youth. Last, we mostly
RkJQdWJsaXNoZXIy MjY0ODMw