Current Topics in Health Economics

Einleitung / Introduction

Protective vaccinations represent a preventive tool to fight infectious diseases and are therefore not only relevant for the general population but especially for patients with malignant diseases. Even though there are both, effective vaccines and specific recommendations of the Standing Commission on Vaccination (STIKO), e.g. for vaccinations against respiratory diseases, the vaccination coverage of these high-risk patients, including those treated in hospitals, is insufficient. Neither the process of checking vaccination status nor administering vaccines are currently firmly embedded in the inpatient and outpatient care provision of hospital clinical routines. The aim of this analysis is to evaluate structural and economic barriers and facilitators regarding the implementation of protective vaccination processes of respiratory diseases in oncological patients focussing on the comprehensive hospital setting.

Methode / Method

A cause-and-effect analysis using the Ishikawa diagram will be conducted to detect structural impacts regarding processes and reimbursement that may be responsible for insufficient vaccination coverage in hospitals. A qualitative assessment with medical experts of the University Hospital Cologne (UHC) will be carried out. UHC Standard Operating Procedures for different tumor entities will jointly be analysed by the medical experts to identify potential barriers and facilitators along the patient journey contributing to an insufficient vaccination coverage in hospitals. Thereby, both the inpatient sector and its ambulatory supply units will be considered. In addition, it is intended to identify points in time suitable for low-threshold opportunities to vaccinate with regard to e.g. differences between types of tumors, tumor stages, and urgency to initiate therapy.

Ergebnisse / Results

The analysis will provide insights on potential hospital-internal barriers and facilitators regarding the integration, administration, and reimbursement of vaccines in hospitals. Results will be summarized and illustrated in the Ishikawa diagram. Findings may provide a framework to develop a best practice vaccination model at the UHC.

Zusammenfassung / Conclusion

Results will contribute to the public debate as it will present barriers and facilitators of vaccination processes in the current clinical routine. The analysis may also promote the interface management between already existing expertise, infrastructure acquired during the COVID-19 pandemic, and hospital processes of regular protective vaccinations. The best practice vaccination model of the UHC may serve as ideal for both, other hospitals, and vaccines against other than respiratory diseases. In the long run, this may strengthen the general impact of preventive measures and disburden healthcare providers by avoiding preventable diseases at the same time.

AutorInnen

Florian Jakobs

Julia Jeck

Udo Holtick

Oliver A. Cornely

Alexander Völker

Johannes N. Urban

Maria Wohlleben

Florian Kron

Einleitung / Introduction

Severe health shocks may affect one’s objective health and subjective health perceptions, but potentially in different ways. Specifically, self-perceived health might revert to pre-shock levels as individuals adapt to their new health state over time, even when the effects on objective health remain persistent. This difference is important given that individuals make decisions based on these biased subjective health perceptions. We explore how health perceptions, lifestyle behavior, and medication use change after experiencing a negative health shock that persistently affects objective health: an ischemic stroke or an acute myocardial infarction. We shed new light on post-health shock recovery and on the role of health perceptions and choices using a novel combination of detailed administrative and survey data.

Methode / Method

We combined two large Dutch health surveys from 2012 and 2016 with administrative data on hospital admissions, healthcare demand and death records from 1995 to 2018. We identified a sample of 13,000 heart attack and 9,000 stroke patients providing survey responses on subjective outcome measures at different relative time points to their respective health shocks. The resulting repeated cross- section of heart attack/stroke patients is interviewed between 6 years prior and 7 years after their event. We use a doubly robust event-study approach that exploits the exogenous timing of the occurrence of these shocks to explore the causal effects of heart attacks and strokes on subjective outcome measures, risky health behaviors, and medication use over time.

Ergebnisse / Results

A heart attack or stroke has large immediate negative effects on subjective health perceptions as self-assessed health decreases substantially by one-third of a standard deviation in the year after a heart attack or stroke occurs. Despite these substantial initial differences, the effects attenuate quickly. While individuals experience an increased objective burden of disease over time, reflected in an increasing prevalence of long-term physical disability, their self-assessed health reverses towards pre- shock levels. Further we observe heterogenous impacts on a range of (risky) health behaviors with both health shocks leading to long-term decreased smoking prevalence of around 10-15 percentage points but only temporary decreases in alcohol consumption and no change in overweight rates or physical activity.

Zusammenfassung / Conclusion

Our findings suggest that even after a severe health shock like a heart attack or a stroke, individuals health perceptions return to pre-shock levels within a short period of time, despite an increasing burden of disease observed. Our results on health behaviors confirm previous studies that find smoking behavior to be consistently affected by a health shock. In line with the temporary effect on subjective health, we find that most effects on lifestyle behaviors are short-lasting. Only for smoking we observe a permanent decrease. In ongoing work, we aim to extend our analyses to the health perceptions and behaviors of cohabiting family members and to explore whether the observed pattern of adapted health perceptions also influence economic behaviors such as the decision to retire or rejoin the labor force, individuals’ choice of health insurance deductibles or adherence to prescription medicine.

AutorInnen

Jannis Stöckel, Erasmus Schoolf of Health Policy & Management, Erasmus University Rotterdam

Pieter Bakx, Erasmus Schoolf of Health Policy & Management, Erasmus University Rotterdam

Bram Wouterse, Erasmus Schoolf of Health Policy & Management, Erasmus University Rotterdam

Einleitung / Introduction

Health data often include either diagnostic data or cost data but not both. This poses a challenge for studies like cost-of-illness studies. Information on health care utilization in insurance claims data can be used to make informed guesses about diseases. However, this requires detailed knowledge about treatment options and billing. In this study, we use a data-driven approach to identify diseases in a unique data set combining diagnostic information from inpatient stays with claims data from outpatient care in Switzerland.

Methode / Method

We combine insurance claims data of the hospitalized population of a large Swiss insurance company with the Swiss hospital inpatient registry (Medizinische Statistik der Krankenhäuser) from 2017 at the patient level. This unique data set contains outpatient health care utilization and detailed diagnostic coding that allows for the identification of utilization patterns associated with 42 major diseases. We use various machine learning algorithms such as Decision Trees, Random Forests and Boosted Decision Trees to predict the presence of diseases based on claims data alone. The numerous available features include drugs (four-digit ATC codes), spending by service provider, and spending by subchapter of national fee-for-service catalogues (physician services, laboratory tests, medical devices). We evaluate the performance of the algorithms based on prediction accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve.

Ergebnisse / Results

Our preliminary results show that only some diseases can be predicted with a high accuracy. These are diseases with very specific treatment options, such as drugs for diabetes. For many diseases, drug utilization by ATC chapter is the most important feature in the prediction. We further assess prediction accuracy using data from individuals hospitalized in 2016, but not in 2017. We find that the algorithms perform well on these unseen observations from a different underlying population. This suggests that we could generalize our results to the non-hospitalized population.

Zusammenfassung / Conclusion

Information on health care utilization from claims data can be a useful predictor of the presence of diseases. The models we trained for the hospitalized patients could be used to estimate prevalence rates in the non-hospitalized population, but also associated costs of diseases. This is especially interesting when no diagnostic information is available in outpatient care data.

AutorInnen

Michael Stucki, Zurich University of Applied Sciences

Andreas Kohler, Zurich University of Applied Sciences

Einleitung / Introduction

A large literature aims to establish a causal link between education and health using changes in compulsory schooling laws. It is however unclear how well more education is operationalized by marginal increases in school years. We shed a new light on this discussion by analyzing the health effects of a reform in Germany where total years of schooling for students in the academic track were reduced from nine to eight while keeping cumulative teaching hours constant by increasing instruction intensity. These reforms are also of special interest, because most studies are focussing on the lower end of the ability distribution by looking the effect of compulsory schooling laws, whereas we are drawing insights from the German academic track, which encompasses roughly the upper 50% of the ability distribution.

Methode / Method

We use two different approaches: First, we use the reform as instrument for total years of schooling. Second, the sequential introduction of the reform across federal states, allows us to implement recent state of the art difference-in-difference strategies accounting for potentially heterogeneous treatment effects as suggested by de Chaisemartin and D'Haultfoeuille (2020).

Ergebnisse / Results

Using data from the German Socio-Economic Panel (2002-2018), we find that increased weekly instruction time has, if at all, only small health effects for subgroups.

AutorInnen

Johanna Sophie Quis, IHE, Leibniz Universität Hannover

Simon Reif, ZEW