Recent Developments in Health Econometrics: Volume 297

We explore the effect of selective schooling, where students are assigned to different schools by ability, on adult health, well-being and labour market outcomes. We exploit the 1960s transition from a selective to a non-selective secondary schooling system in England and Wales. The introductio3n of mixed-ability schools decreased average school quality and peer ability for high-ability pupils, while it increased them for low-ability pupils. We therefore distinguish between two treatment effects: that of high-quality school attendance for high-ability pupils and that of lower-quality school attendance for low-ability pupils, with mixed-ability schools as the alternative. We address selection bias by balancing individual pre-treatment characteristics via entropy balancing, followed by ordinary least squares (OLS) regression. Selective schooling does not affect long-term health and well-being, while it marginally raises hourly wages, compared to a mixed-ability system, and school aspirations for high-ability pupils. Cognitive and non-cognitive abilities measured prior to secondary school are significantly and positively associated with all adult outcomes.

This chapter reviews the evidence on the role of physicians in shaping inequalities in access to and utilisation of healthcare. The authors examine three types of physician decisions that can influence inequalities in access and utilisation: location decisions, decisions to work in the public and/or private sector, and decisions or behaviours in the doctor–patient encounter. For each, the authors summarise the issues and empirical evidence on possible policies to help reduce inequalities in access. Future research to reduce inequalities should focus on changes to health systems that influence physician decisions, such as health insurance expansions, the public–private mix and financial incentives, as well as physician training and policies for a more diverse physician workforce.

Ensuring adequate access to healthcare services is a priority across European countries. The EU has developed performance indicators to compare access using self-reported unmet need. Cross-country comparisons require adjustment for factors outside the health systems' control. We address two research questions to improve the comparability of unmet need for medical and dental care across the EU and the comparability of socio-economic inequalities in unmet need across the EU. First, we explore the role of risk adjustment for demographic and socio-economic factors, which are outside health systems' control, for both overall unmet need and unmet need due to affordability, waiting lists and distance. Second, we compare differences in unmet need by socio-economic status, and investigate whether different forms of risk adjustment affect such comparison. We show that adjusting for age, gender and chronic conditions reduces dispersion of unmet need for medical care across the EU. Controlling for income further reduces the dispersion, mostly due to affordability. When comparing socio-economic inequalities across countries, risk adjustment for age, gender and chronic conditions play a limited role. Socio-economic inequalities by income and education vary by reason of unmet need: the income gradient, even controlling for education, is mostly due to affordability rather than waiting list or distance. For dental care, the main reason for unmet need is affordability. Risk adjustment for age, gender, chronic conditions and education plays a limited role. The income and education gradients are more pronounced for dental than medical care.

Economic evaluations often utilise individual-patient data (IPD) to calculate probabilities of events based on observed proportions. However, this approach is limited when interest is in the likelihood of extreme biomarker values that vary by observable characteristics such as blood glucose in gestational diabetes mellitus (GDM). Here, instead of directly calculating probabilities using the IPD, we utilised flexible parametric models that estimate the full conditional distribution, capturing the non-normal characteristics of biomarkers and enabling the derivation of tail probabilities for specific populations. In the case study, we used data from the Born in Bradford study (N = 10,353) to model two non-normally distributed GDM biomarkers (2-hours post-load and fasting glucose). First, we applied fully parametric maximum likelihood to estimate alternative flexible models and information criteria for model selection. We then integrated the chosen distributions in a probabilistic decision model that estimates the cost-effective diagnostic thresholds and the expected costs and quality-adjusted life years (QALYs) of the alternative strategies (‘Testing and Treating’, ‘Treat all’, ‘Do Nothing’). The model adopts the ‘payer’ perspective and expresses results in net monetary benefits (NMB). The log-logistic and Singh-Maddala distributions offered the optimal fit for the 2-hours post-load and fasting glucose biomarkers, respectively. At £13,000 per QALY, maximum NMB with ‘Test and Treat’ (−£330) was achieved for a diagnostic threshold of fasting glucose >6.6 mmol/L, 2-hours post-load glucose >9 mmol/L, identifying 2.9% of women as GDM positive. The case study demonstrated that fully parametric approaches can be implemented in healthcare modelling when interest lies in extreme biomarker values.

This chapter pays tribute to Andrew Jones' research in health programme evaluation, health-risky behaviour and income-related health inequalities by reviewing policy-relevant empirical studies in these domains using Italian data. In the first section, We discuss the impact of reimbursement systems on healthcare behaviour, particularly the transition from incurred-cost-based to prospective systems in hospitals. We explore incentive-driven practices like up-coding and cream skimming, while also considering the potential advantages of primary care incentives and the mixed outcomes associated with cost-sharing schemes. The second section delves into health-risk behaviours in Italy, encompassing substance use, preventive healthcare and responses to health information. The last section presents some evidence on socioeconomic status (SES)-related health disparities and discusses the necessity of accounting for these factors in the Italian National Health Service (NHS)'s resource allocation formula in line with British NHS experience.

Depression imposes substantial individual and societal economic costs, including lower productivity and higher healthcare use. However, while the relationship between employment and mental health has been explored, less is known about the potentially countervailing effects of different types of economic inactivity on depression among older individuals. The authors employ a series of models, including fixed effects panel data models and matching on rich data from the English Longitudinal Study of Ageing (ELSA) to investigate whether different types of inactivity might have heterogenous effects on depression. The authors find that whereas transitions to involuntary inactivity (unemployment) do not appear to have a perceivable effect on depression, transitions to voluntary inactivity (retirement) seem to decrease it.

This chapter presents a mediation model that aims to disentangle the indirect from the direct effects of retirement on health, considering the mediating role of lifestyles. The model is applied to the risk of depression, and physical inactivity is assumed to potentially mediate the effect of retirement. The results indicate that there is a significant indirect effect via the mediator, albeit relatively small in comparison to the direct effect. The analysis highlights the importance of further exploring the influence of lifestyle factors in the relationship between retirement and health, in order to gain a better understanding of the potential pathways through which retirement impacts health.

A growing body of research finds a consistently negative relationship between medical cannabis access and aggregate measures of opioid use. Nothing is currently known about the types of opioids that are being most impacted by cannabis access. Using the Callaway and Sant’Anna (2021) difference-in-differences (DID) estimator for the main analysis and data on all opioid shipments to every United States (US) pharmacy from 2006 to 2014, the authors found no evidence of overall change in the total number of morphine milligram equivalent (MME) units of opioids shipped to pharmacies, following the opening of medical cannabis dispensaries. However, across all opioids, the authors found a reduction in the highest MME dosage strengths (8.8% decrease in 50–89 MME doses and 11.3% decrease in 90+ MME doses). This decrease appears to be driven predominantly by commonly diverted opioids, where the authors found a reduction in the highest MME dosage strengths (12.2% in 50–89 MME doses and 13.8% in 90+ MME doses). Further, the authors see a 6.0% increase in low-to-moderate dose opioids (0–49 MMEs). This is consistent with patients using cannabis concomitantly with opioids in order to achieve a lower opioids dose.

This chapter shows the prevalence, trends and heterogeneity in maternal smoking around birth in the United Kingdom (UK), focussing on the war and post-war reconstruction period in which there exists surprisingly little systematic data on (maternal) smoking behaviours. Within this context, the authors highlight relevant events, the release of new information about the harms of smoking and changes in (government) policy aimed at reducing smoking prevalence. The authors show stark changes in smoking prevalence over a 30-year period, highlight the onset of the social gradient in smoking as well as genetic heterogeneities in smoking trends.

There is widespread concern about low adherence to clinical practice guidelines (CPGs) and the low adoption of new medical technologies. To assist the regulatory response, we propose benchmarking clinical practice on the lower bound on the probability that a recommended treatment/new technology achieves a better outcome. This inequality–probability bound can be estimated from marginal outcome distributions. We illustrate the approach by comparing Swedish cardiologists' adoption of drug-eluting stents (DESs) with the inequality–probability bound on this technology improving outcomes. A substantial fraction of cardiologists are below the benchmark.

This chapter reviews the econometric approaches typically used to deal with the spike of zeros when modelling non-negative outcomes such as expenditures, income, or consumption. Relying on the assumptions of selection on observables for evaluating a policy or treatment, this chapter discusses other issues that arise with spikes of zeros in the data, including the analyst's choice between full information versus quasi-likelihood methods, considering whether observed zeros are true or masking more complex behavioural decisions, and dealing with zeros that arise due to self-selection. This chapter ends with discussions of empirical strategies to deal with these behavioural assumptions and a brief review of the literature where such strategies were employed.

Multiple chronic conditions (MCCs) have attracted significant public policy and clinical attention. Whether MCCs determine other important outcomes, or are themselves the outcomes of health-producing activities or interventions, metrics based thereon have potential to be useful indicators of the health of populations and of differences between and among the health of subpopulations. While the attention MCCs are attracting in various policy circles is impressive, MCCs' potential roles as indicators of population health and of how health determinants influence population–health outcomes have received less attention. The purpose of this chapter is to direct attention towards questions that involve considerations of chronic condition (CC) patterns as health outcomes; specifically, this paper hopes to advance the consideration of patterns of MCCs as indicators of individual and population health. Using data from the United States (US) Behavioural Risk Factor Surveillance System (BRFSS), the chapter explores whether both the ‘intensity’ (i.e. the number or count) of CCs as well as their ‘composition’ (i.e. the patterns of particular CCs) might be jointly of interest when considering the prevalence of MCCs in populations and how the nature of MCCs may vary across subpopulations of interest. It is seen that information about intensity tells an incomplete story about MCC health outcomes.