206 Chapter 7 leading to inequalities in health. This aspect is relevant in Chapter 3, where findings would ideally have gone beyond decreased access to care into studying treatment outcomes. In Chapter 2, we measured inequalities in mental health treatment outcomes and established that these did not result from inequalities in treatment minutes. Further evaluating the link between treatment and outcomes would require more extensive documentation of the quality of the care provided and how patients adhered to treatment. Last, Chapter 4 is the strongest in terms of outcomes measurement, including non-health domains that should be routinely considered when evaluating policy or programmes (e.g., labour and income). In this case, the instrumental variable approach prevents us from identifying the exact groups that benefited from supported housing eligibility. In conclusion, inequalities in health care remain a significant concern and future research should focus on measuring need and evaluating health outcomes more. Findings about whom and for whom? Generalizability considerations of policy evaluation The design and implementation of policies and programmes are usually shaped by the local context, making it challenging to disentangle findings of their evaluation from the setting in which they were implemented. The generalizability of findings to other settings (e.g. countries) would often require a detailed transferability exercise. This also applies to this thesis’s studies, even when similar programmes those studied have been implemented internationally (e.g. supported housing, Chapter 4). Still, causal findings remain particularly interesting for international audiences, including country-specific exposures such as the increase in the Dutch deductible paid from 18 years onwards. While a similar rise in deductibles at the transition to adulthood might not apply in other countries, this study should caution policymakers of using cost-sharing as a blunt tool during developmentally-sensitive phases of the life-course, when it is likely to widen inequalities in care. It also shows that financial barriers compound to the transitional gap in mental health care, warranting the investigation of similar obstacles elsewhere, particularly in countries with higher out-of-pocket costs. Analogously, the conclusions of Chapter 2 carry important learnings at the international level. Despite described for the Dutch mental health care system (with some specificities such as the way first treatment course was defined), the magnitude and persistence of these disparities across different diagnoses/therapies suggest that they result from general processes of care provision. These processes, such as the interaction between the patient and practitioner or the influence of the patient’s socioeconomic environment, will likely prevail in other countries. As for chapter 3, the magnitude of the disparities identified in chapter 2 might well be more substantial in countries with weaker welfare provisions than the Netherlands. Essential considerations on generalizability also arise from the nature of the quasiexperimental designs. One aspect common to studies with discontinuity-based methods (Chapter 3) and instrumental variables (Chapter 4) is that their findings capture LATEs. In the case of Chapter 3, the conclusions only apply to those turning 18. In the case of
RkJQdWJsaXNoZXIy MjY0ODMw