Abstract: Leveraging administrative data on hospital and ambulance stations, we estimate the effect of sickness absences in ambulance services on patients’ mortality in emergency departments (ED). By exploiting quasi-random variation in sickness absences across ambulance stations serving the same ED, we find a significant negative effect of the ambulance sickness absence rate on the survival of patients who arrive at the ED by ambulance. An increase in the absence sickness rate by 1 standard deviation has an effect on the mortality comparable to an increase in the patient’s age from 50-64 to 65 and above. These results underscore the critical role of ambulance workers’ on emergency care quality and provide evidence on the damage caused by ambulance services understaffing.

We use monthly clinic-level panel data from 204 clinics in Rio de Janeiro between January 2009 and December 2016 to study the effects of violent events on healthcare access, utilization, and quality. Exploiting the proximity of clinics to geocoded episodes of violence related to police operations and drug gang battles, we find that exposure to violence leads to a significant reduction in healthcare utilization. Specifically, each additional police-related shooting is associated with a 12.3% reduction in primary care procedures, primarily driven by decreased access on the day of the event. These effects are more pronounced in neighborhoods with lower average income and socioeconomic status. We provide evidence that while violent events reduce healthcare utilization, they do not significantly impact quality indicators such as employee turnover or hospitalization. Thus, urban violence not only hampers access to health services but also exacerbates inequality in healthcare utilization in affected neighborhoods.

We investigate the impact of unexpected health shocks—defined as sudden diagnoses of cancer, stroke, or heart attack—on the probability of divorce among couples aged 50 and older, using longitudinal data from the Health and Retirement Study (HRS). Our objectives are threefold: to quantify the causal effect of health shocks on divorce probability, to examine the mechanisms through which health shocks influence marital stability—specifically mental health deterioration, cognitive decline, and financial strain—and to provide evidence-based recommendations for policies supporting affected couples and mitigating the risk of marital dissolution. Leveraging the longitudinal nature of the HRS, we employ a quasi-experimental design to construct counterfactual scenarios for affected households by comparing them to households poised to experience similar events in subsequent years. Using a stacked difference-in-differences estimator, as proposed by Wing et al. (2024), we find that health shocks significantly increase the probability of couple dissolution by approximately 21% of the mean divorce prevalence. This effect intensifies gradually over time rather than manifesting immediately after the health shock. Further analysis suggests that mental health, cognitive decline, and financial strain are key mechanisms mediating the relationship between health shocks and increased divorce probability. These findings highlight the long-term implications of adverse health events on marital stability and underscore the importance of targeted support for couples facing such challenges.

We investigate the mechanisms by which a student’s age relative to classmates (i.e., relative age) influences risky health behaviors among European adolescents. Using a two-stage least squares approach, we show that relatively young students are more prone to engage in risky behaviors. These results hold after controlling for absolute age, country fixed effects, and birth season effects. In the second part of the paper, we conduct two sets of analyses on possible mechanisms. First, causal mediation analyses reveal that students’ perceived academic performance is the primary mediator. Second, additional analyses suggest that perceptions of substance risks and peer usage prevalence may also play a significant role.

Ischaemic heart disease is the leading cause of death worldwide. Despite a lower (diagnosed) lifetime prevalence of heart disease, women are significantly more likely to die if hospitalized with Acute Myocardial Infarction (AMI). This is puzzling as one would expect those who are sicker and thus more likely to die from AMI to have a higher likelihood of being diagnosed in the first place. I show that healthcare providers are important drivers of this puzzle: While the mortality gap is substantially smaller for users of lipid-lowering drugs, women are typically less likely to use these drugs. Providers have substantial power over their patients' medication usage: Switching to a provider with a higher female usage rate of lipid-lowering drugs due to a clinic closure increases the uptake of these drugs and reduces mortality. Difficulties in interpreting female symptoms may drive the lower usage rate of women since providing cholesterol measurements outside the healthcare sector can reduce gender inequalities in the usage of lipid-lowering drugs - but the success of this depends on how information is transmitted. Gender differences only disappear when providers are directly informed about the results, and when it is not the patient's responsibility to report them. Lastly, I demonstrate that these gender differences also spill over to the labor market: women reduce their labor supply and increase disability benefit receipt much more than men - after AMI but also in anticipation.

We study how capacity constraints affect physicians’ willingness to gather additional information supporting their therapy decisions. Further, we examine how capacity constraints affect the utilization of additionally gathered information and the appropriateness of physicians’ therapy decisions. Using a controlled framed field experiment with German pediatricians ( n=247), we exogenously vary the extent to which physicians’ capacity is constrained. Our behavioral results evidence that physicians’ willingness to gather additional information that supports decision-making decreases as capacity constraints increase, but the utilization of the information is not affected. We also find that capacity constraints have a statistically significant and clinically relevant impact on the appropriateness of therapy decisions and, thus, on the quality of care. This is especially the case for physicians with little clinical experience. The probability of gathering additional information increases significantly with case severity. Additional information increases the appropriateness of therapy decisions in the case of undertreatment, while overtreatment persists even after the use of decision support. Behavioral results suggest that decreasing the extent to which capacity is constrained can be an effective way to enhance the utilization of decision support and thus help improve the appropriateness of therapy decisions. Implications of our findings for the management of healthcare organizations are discussed.

While evidence points at potentially strong effects of tightening eligibility criteria for Disability Insurance (DI), little is known about their targeting effects: are workers screened out also those with better future employment, health, and mortality outcomes? Knowing that DI receipt is mostly permanent for workers, it is key to understand the permanence of impairments and their implications for the ability to work. To shed light on this, we employ Regression-Discontinuity-in-Time regressions to compare the long-term health and mortality outcomes of DI application cohorts just before and after a reform in the Netherlands. This reform led to stricter screening in the sickness period before DI application and reduced the number of applicants by 33%. Up to 18 years after application, we find persistently lower survival rates, higher medical consumption, and lower employment rates of post-reform cohorts. Using detailed information on the future chronic diseases of these cohorts, we next construct a Chronic Disease Index (CDI) that explains survival rates and a Work Ability Index (WAI) that explains long-term employment rates. Inferences on changes in the CDI due to the reform show that compliers to the reform have health conditions with high expected survival rates (i.e., high CDI-scores). For employment (the WAI), however, screening effects were less substantial. This reflects the effect of reduced applications from screened-out workers with (mental) health conditions, having low mortality rates and relatively limited ability to work.

We provide evidence on the impact of firm productivity on the health maintenance of employees. Using linked employer-employee data from Hungary, we analyze the dynamics of healthcare use before and after moving to a new firm. We show that moving to a more productive firm leads to higher consumption of drugs for cardiovascular conditions and more physician visits, without evidence of deteriorating physical health, and, among older workers, to lower consumption of medications for mental health conditions. The results suggest that more productive firms have a beneficial effect on the detection of previously undiagnosed chronic illnesses and on mental health.

In this paper, I study how a policy aimed at not criminalizing illicit drug consumers, to recognize their health care needs, might negatively influence newborn’s health. In 2013, a policy that established maximum permissible amounts of possession of illicit substances for personal consumption was implemented in Ecuador. This regulatory change might be one of the drivers of increased availability of drugs, which in turn, may increase consumption. I exploit variation in drug exposure at the province level in a difference-in-differences framework, which compares Ecuadorian provinces with different intensities of drug exposure before and after 2013. I find that the share of low birth weight and very low birth weight newborns raise significantly by 16.26% and 82.22% in provinces initially more exposed to drugs. I also document a significant increase in both women and men’s substance use. I conclude that the increase in drug use is possibly one of the channels affecting newborn’s health living in more exposed provinces.

In this paper, we investigate the effect of delaying retirement age on late-career mental health. We contribute to the recent literature by examining how this effect varies depending on working conditions, utilizing exogenous data on job quality. Using pension reforms in several European countries, and repeated cross-sectional data from the Survey of Health, Ageing and Retirement in Europe and the European Working Conditions Survey, we find that extending work horizon generally has a negative impact on mental health of older workers, with significant heterogeneity depending on job characteristics. Specifically, both male and female workers in lower-skilled, insecure jobs with poor prospects experience a marked increase in depression when required to work longer. Moreover, the social dimension at work plays a critical role, with poor social conditions exacerbating the adverse effects of delayed retirement on mental health, and supportive working environments reversing them.

We analyze the impact of son preferences on individual and household-level outcomes. First, we show how the presence of a second-born female in a household affects the health and nutritional outcomes of first-born children. Using nationally representative survey data from rural Bangladesh and utilizing the randomness of the second child’s gender, we find that first-born children’s height-for-age z-score increases significantly by 0.10 standard deviation if the second-born is a female. This is due to the higher proportion of household calories allocated to the eldest child in this sibling gender combination. Second, using two-period panel data on the same households and employing a triple difference regression framework, we show that removed financial constraints through the Feed the Future program, a US government’s initiative addressing global hunger and food security, do not encourage families with second-born female children to diversify food consumption. Our findings thus imply that despite the efforts of the Feed the Future program, being a daughter is still a distress in rural areas of Bangladesh.

Objectives: We test the effectiveness in promoting flu vaccination uptake in a video informational campaign of Motivational Interviewing (MI), a gentle communication technique, compared to traditional unidirectional communication (UD) methods. While MI has proven effective in direct doctor-patient communications and is currently recommended by the WHO, applying it to video campaigns would address the standing issue of resource and time constraints associated with training health professionals.

Methods: We conducted a two-wave survey experiment on a representative sample of 12000 Italians aged between 40 and 90. Participants were randomly assigned to watch a 100-second informational video about the flu vaccine presented either in an MI or UD format. The experiment also varied the informant's identity (medical doctor vs. layperson) and gender concordance with the receiver, in a 2^3 factorial design. The primary outcomes were self-reported flu vaccination status and intention to vaccinate. Secondary outcomes included perceptions of the informant and attention scores. Causal forest analyses were conducted to explore heterogeneity in treatment effects.

Results: As predicted by theory, MI significantly improved perceptions of the informant as more collaborative, considerate, sincere, and trustworthy compared to UD (5% improvement). It also outperformed the informant’s identity nudge. However, MI decreased participants' intention to vaccinate by 2-3 percentage points (-7%) and had no significant effect on actual vaccination behavior. Causal forest analyses revealed that while the overall behavioral effect of MI was null, it varied among subgroups: MI increased vaccination uptake among older individuals in poor health and decreased uptake among those with higher baseline vaccine compliance. Moreover, individuals who improved their perceptions never experienced significant behavioral effects.

Discussion: While MI can enhance perceptions of health communicators, it may not effectively translate these positive perceptions into increased vaccination uptake. In fact, MI can potentially backfire when used in large-scale video campaigns, particularly among those already inclined to vaccinate. These findings suggest that MI should be applied cautiously and perhaps targeted towards specific groups, as done in direct doctor-patient interactions, rather than the general population. In that case, it might contribute to cutting implementation costs. The findings highlight the need for tailored communication strategies to address vaccine hesitancy effectively.

Objective: Including productivity in economic evaluation is debated, with arguments against that it disadvantages some groups, and arguments in favour that higher productivity increases society's wealth and therefore the resources available for healthcare. This study examines whether increased productivity from life-extending and/or quality of life-improving treatments is likely to be sufficient to improve everyone's health due to the additional resources available to society.

Methods: The UK Household Longitudinal Survey has collected data annually from around 80,000 people since 2009. Individuals complete the SF-6D, from which utility on a full-health=1, dead=0 scale was estimated. People's financial outcomes and whether they die in-between waves is also recorded. Random effects models were used to estimate individuals' quality-adjusted life expectancy, the total surplus they generate over their working life, as well as their net contribution to public funds. The effects of marginal increases in the probability of surviving until the next survey wave and utility were used to approximate the effects of life-extending and quality of life-improving treatments.

Results: Life-extending treatments resulted in lifetime surplus generated increasing by up to £284,000 per quality adjusted life-year (QALY) for those aged around 30 who were already in good health. However, life-extending treatments to anyone aged over 45 implied net costs to the social decision-maker, which could be as high as £19,600/QALY. Quality of life improving treatments always resulted in improvements both in lifetime surplus generated and net contributions to public funds, but these increases were modest and never exceeded £5,000/QALY.

Discussion: Whether or not to include productivity costs in health technology assessment is an ongoing debate, and practice varies from country to country. This study contributes to the debate by showing that, while life-extending treatments can improve individuals' productivity, the largest gains are seen for those who already have a good level of health, and treating anyone middle-aged or older is likely to reduce the budget available for healthcare. However, low cost quality of life-improving treatments have the potential to pay for themselves.

Objectives: The translation of diagnostic biomarker tests into clinical practice demands significant investment in time and resource, and thus early economic evaluation is useful to prioritize tests with the greatest potential for successful adoption in clinical settings. We conduct an early economic analysis to identify the necessary characteristics of novel biomarker tests for GCA to be cost effective.

Methods: A combined decision tree and Markov model was developed to assess the potential cost-utility of biomarker tests for patients with symptoms suggestive of GCA at secondary care. Input parameters were based on secondary data sources and expert opinions. A lifetime horizon and the NHS and Personal Social Services perspective were adopted for the analysis. Sensitivity analyses were undertaken to figure out the maximum price at each combination of sensitivity and specificity for them to be cost effective.

Results: At a willingness-to-pay threshold of £20,000 per QALY, a hypothetical biomarker test with perfect accuracy (100% sensitivity and specificity) could be priced up to £6,559 compared to the standard pathway of biopsy and clinical judgment, and up to £7,956 compared to the standard pathway including ultrasound. The price levels decreased with shorter time horizons and increased when a longer duration of steroid-related effects was assumed. Probabilistic sensitivity analysis over 5,000 simulations produced similar results as in the base-case scenario, with standard deviations around 20% of the mean values, suggesting that a 20% uncertainty range should be considered around the maximum cost of biomarker tests.

Discussion: The analysis shows the potential for the hypothetical biomarker tests to improve diagnosis of GCA and reduce risks of glucocorticoid toxicity, along with the market potential of developing the tests for clinical use in the NHS. Our study shows that economic models can provide vital information for developers at an early stage, although with limited evidence available.

Objectives: The launch of lecanemab and donanemab marks a new era in the treatment of Alzheimer’s Disease (AD). However, their value, affordability, and global accessibility remain uncertain. This study aims to propose value-based pricing for lecanemab and donanemab that aligns with national cost-effectiveness thresholds across low-, middle-, and high-income nations.

Methods: We built a Markov model to estimate cost and effectiveness outcomes for individuals with early AD from a healthcare perspective. We conducted a value-based analysis to propose prices for lecanemab and donanemab that yield an acceptable cost per quality-adjusted life year (QALY) gained based on published estimates of cost-effectiveness thresholds in 174 countries. Mean age of the modeled cohort was 70; 55% had mild cognitive impairment due to AD and 45% mild AD dementia; 92% lived in the community and 8% in facilities. We performed sensitivity and scenario analyses to examine uncertainty.

Results: In high-income countries, the median value-based prices of lecanemab and donanemab were $1,745 and $2,574, respectively. Switzerland had the highest prices, with lecanemab at $8,860, and donanemab at $12,864. For upper-middle-income countries, prices were $333 and $486. In lower-middle-income countries, prices were $42 and $66, and in low-income countries, they were $6 and $12. Adopting a societal perspective increased value-based prices up to two-fold in some nations. Additionally, prices were affected by treatment duration, efficacy, and the use of PET scans for monitoring patients on treatments.

Discussion: At current prices, neither lecanemab nor donanemab would be cost-effective for early AD in 174 countries. Substantial price reductions are necessary to improve global affordability and accessibility.

The objective of this paper is to explore the impact of two extreme cases: solitude and primitive privacy deprivation, on young people’s mental health during the COVID-19 pandemic. Using longitudinal survey data from four developing countries: Ethiopia, India, Peru, and Vietnam, I apply a quasi-experimental approach and use Difference-in-Difference-in-Difference method as an empirical identification strategy. Firstly, I categorise the four countries into two groups based on stricness of COVID policies: highly affected category and marginally affected category. This categorisation is based on the individual exposure intensity measures I constructed. Intuitively, how exposed an individual is to the COVID policies are determined by two factors: (i ). the strictness of policy responses to COVID-19; and (ii ). the timing and duration of an individual’s compliance with these restrictive measures. To define individual-level variation in primitive privacy, I use living conditions as a proxy based on which, I then define two treatment groups and one comparison groups: (i ). Solo dwellers as solitude group (treatment group 1), (ii ). sharing a bedroom with other members in a multi-person household as primitive privacy-deprived group (treatment group 2), and (iii ). those who live together with other family members but have their own independent bedrooms are considered the comparison group. Solitude and primitive privacy deprivation have both temporal and spatial dimensions. However during lockdowns, these two dimensions are interwined to a certain extent. Preliminary results show that solitude contributes negatively to subjective well-being, compared to the well-being of those living in multi-person households during the pandemic. Neither solitude nor privacy deprivation significantly affect symptoms of anxiety and depression. Anxiety and depression show a high correlation, suggesting comorbidity. Such findings are also policy relevant and suggesting policy interventions in groups with comorbidity.

Objectives: The loss of a spouse can lead to a radical break in individual life courses, yet, the literature on the effects of spousal loss is very limited.This paper estimates the causal effect of spousal bereavement on health.We identify causal effects using event study regressions, by focusing on older populations where bereavement is not uncommon, by providing estimates for men losing their wives, and by considering a wide range of health outcomes.

Methods: We use panel data from the SHARE study, which focuses on individuals aged 50 and above from 28 European countries. We use event study regressions to examine how individuals’ health changes in the years before, during and after the loss of a spouse.Under the assumption that the timing of spousal loss is random, the impact of a spousal loss is identified by comparing the health status of individuals that experienced spousal bereavement recently to individuals that are still married but will transition into widowhood later in the panel. For our event study we use both classical two-way fixed effects estimator  as well as new estimators of Callaway and Sant’Anna, 2020 and Sun and Abraham, 2021. We also consider an instrumental variable strategy, which allows us to use data from individuals that remain married (as a control group).We propose the age difference between partners as an instrument for spousal loss.Intuitively, we build on the idea that individuals with an older spouse are at a higher risk of experiencing bereavement than individuals with a younger spouse.

Results: We estimate three different event study models for the effect of widowhood on the depression for women.We see a stable pre-trend in the years before the death.In the year of the death of their husband, the number of depressive symptoms increases significantly by more than one symptom.This effect persists in the following year, and then depression slowly reverts back to baseline levels.Similar patterns appear for men too.Considering other health outcomes, we find less precisely estimated increases in the risk of hypertension and heart disease for women, and corresponding increases in the consumption of drugs for anxiety and heart disease.For men, we only observe a significant increase in the consumption of drugs for anxiety.Our IV regressions show that an age difference above 3 years between partners increases the risk of spousal bereavement by between 5 and 6 percentage points.The estimated health effects largely confirm the conclusions from our event study regressions.

Discussion: Following the loss of a spouse, older Europeans report significantly more depressive symptoms.Reassuringly, our results suggest that depressive symptoms revert back to baseline levels within four years, which suggests that there are no scarring effects.Similarly, we find very limited evidence for effects on physical health, although the effects on heart disease and hypertension for widows are concerning.

Objectives: There is wide consensus about the importance of early life investments in improving health and economic outcomes later in life. Encouraging women to give birth in health facilities rather than at home therefore remains a key policy objective in low- and middle-income countries where less than 75 percent of births occur in health facilities. Despite this focus, there is little unconfounded evidence on whether facility-based delivery improves child health and economic outcomes relative to delivering at home. This paper uses a rigorous instrumental variables approach with a machine learning component to examine the impact of facility-based delivery on child outcomes in Africa, including spillovers between siblings.

Methods: I use rich data from the Demographic and Health Surveys that includes women's detailed birth histories, household characteristics and geographic data. I link births with high-resolution, high-frequency weather data from the ERA5-Land database. My identification strategy uses exogenous weather shocks such as rain and high temperatures in the week before birth as instruments for facility-based delivery since they may hinder travel and induce women who considered delivering in a health facility to instead deliver at home. I exclude births occurring in the days following extreme weather shocks like flooding – these likely violate the exclusion restriction since they may influence birth and maternal outcomes in ways other than via the likelihood of facility-based delivery. I use machine learning methods (lasso and elastic net) to select the strongest instruments from a large set of 400 potential instruments for facility-based delivery.

Results: Results from the elastic net instrumental variables estimation show that facility-based delivery is statistically significantly associated with a 5.4 percentage point increase in under-five survival, an 8.4 point increase in the height-for-age percentile, an 8.3 point increase in the weight-for-age percentile and an 11 percentage point increase in the likelihood of the child receiving any vaccinations relative to children who were delivered at home. There is a positive spillover on these four outcomes from older siblings who were delivered in a health facility to younger siblings. Having a younger sibling who was delivered in a health facility increased the likelihood that older siblings received any vaccination but did not affect anthropometrics.

Discussion: The instrumental variables approach identifies large benefits of facility-based delivery that accrue to women who are dissuaded from traveling to a health facility by relatively small, potentially surmountable barriers. There is therefore a need for better policies to encourage, incentivize and enable facility-based delivery for this group.