Periodic Reporting for period 1 - PoS (The price of success: what is the impact of improving patients’ survival on utilisation of care?)
Periodo di rendicontazione: 2019-09-01 al 2021-08-31
Health care expenditure absorbs a large share of public resources and is growing fast in many European countries. The drivers of this growth go beyond demographic factors and have not been properly investigated until recently due to the lack of good quality data following the population of health service users for a sufficient number of years. The Price of Success (PoS) research project aims to fill this gap in the knowledge by applying state-of-the-art econometric models to a very rich dataset on the population of Danish patients. The overarching hypothesis investigated by PoS is that a severe health shock, such as a stroke, AMI, hip fracture, is likely to leave a permanent mark on the ability of the individual to take care of herself and overcome the next episode, including completely unrelated episodes and small episodes of illness that were previously resolved without the need of a hospital visit. The remarkable success of new technologies and medical practices in extending the life expectancy of high-risk patients over the past decades might have increased the population of frequent users of the health system who were unlikely to survive a health shock in the past. The main objective of the project was to investigate the impact of such a “survival effect” on the rapid growth of the demand of health care services. The main objective was articulated in three research questions:
R1: What is the impact of improving patient survival on subsequent costs of care?
R2: Is the impact of the survival effect different according to patient’s characteristics?
R3: To what extent can primary care reduce the use and costs of emergency secondary care?
R1: What is the impact of improving patient survival on subsequent costs of care?
R2: Is the impact of the survival effect different according to patient’s characteristics?
R3: To what extent can primary care reduce the use and costs of emergency secondary care?
PoS examined the demand for health care in a large population of individuals exposed to a health shock over a ten-year time window in Denmark. We found that 60% of the growth in the demand for health care remains unexplained after controlling for variation in health and sociodemographic characteristics of the individuals experiencing the health shock. Such a “residual increment” can be attributed to technological progress and change in medical practice occurred over the examined period. By using an indirect approach, we decomposed such a residual increment in the part that is due to postponing “time to death”, i.e. individuals being able to live longer thanks to new technologies and medical practices, and the part that is due to increasing intensity of resource use, i.e. resources becoming more expensive to allow for the cost of innovation. We found that postponing time to death can explain a quarter of the increment in the demand for health care in our study population.
Then, we examined the impact of variation in hospitals' ability to save the life of their patients on the demand for health care by using a direct approach. We found that the demand for health care is highly responsive to variation in hospitals’ ability to extend the life expectancy of their patients: a one percent variation results in a 6.3% increment in the demand for individuals with a heart attack and 5.3% for individuals with a stroke.
Our study suggests that the hospitals’ success in saving the life of individuals suffering a health shock can be an important driver of the demand for health care. This may have important implications for policy makers in planning the targets and resources for the national health system. In particular, health policies aiming at reducing the mortality of high-risk health conditions by improving the quality of care should be accompanied by additional resources to address the demand of care of an increasing number of individuals surviving such conditions.
Finally, we investigated the relationship between the demand for primary care and emergency secondary care and measure to what extent it is possible to reduce utilisation of the latter by increasing the utilisation of the former. Emergency care is associated with higher costs and poorer health outcomes than primary care for similar health conditions, thus resulting in inefficient allocation of resources and suboptimal health outcomes for the population. We proposed a new econometric model that allows for estimating the potential substitution effect between primary and emergency care. We provided an application of the new model to the study of the demand for primary and emergency care in individuals who are exposed to a health shock to their circulatory system. We found evidence of a substitution effect with an additional primary care visit reducing emergency care visits by 12.6 percentage points suggesting a fair scope for intervention to policies aiming at redirecting part of the demand of care from emergency to primary care.
Then, we examined the impact of variation in hospitals' ability to save the life of their patients on the demand for health care by using a direct approach. We found that the demand for health care is highly responsive to variation in hospitals’ ability to extend the life expectancy of their patients: a one percent variation results in a 6.3% increment in the demand for individuals with a heart attack and 5.3% for individuals with a stroke.
Our study suggests that the hospitals’ success in saving the life of individuals suffering a health shock can be an important driver of the demand for health care. This may have important implications for policy makers in planning the targets and resources for the national health system. In particular, health policies aiming at reducing the mortality of high-risk health conditions by improving the quality of care should be accompanied by additional resources to address the demand of care of an increasing number of individuals surviving such conditions.
Finally, we investigated the relationship between the demand for primary care and emergency secondary care and measure to what extent it is possible to reduce utilisation of the latter by increasing the utilisation of the former. Emergency care is associated with higher costs and poorer health outcomes than primary care for similar health conditions, thus resulting in inefficient allocation of resources and suboptimal health outcomes for the population. We proposed a new econometric model that allows for estimating the potential substitution effect between primary and emergency care. We provided an application of the new model to the study of the demand for primary and emergency care in individuals who are exposed to a health shock to their circulatory system. We found evidence of a substitution effect with an additional primary care visit reducing emergency care visits by 12.6 percentage points suggesting a fair scope for intervention to policies aiming at redirecting part of the demand of care from emergency to primary care.
We proposed a new method for investigating the impact of technological progress on the demand for health care by applying a residual approach to microdata. This allow us to provide an accurate control for variation in the health care expenditure that is due to morbidity and to avoid assumptions over factors encouraging technological progress, such as income elasticity. Both are often problematic variables in macroeconomic models that traditionally investigated this topic as the former is often unavailable at the macro level and the latter influences the predicted effect of technological progress on health care expenditure.
We expanded the scope of existing literature investigating the impact of time to death on individual health care expenditure. Existing studies examine variation in time to death that steams from the characteristics of the demand for health care, e.g. variation in proximity to death due to age and morbidity. We showed that variation in time to death produced by variation in the supply of care has an important impact on health care expenditure that adds to demand-driven variation. Technological progress directed to extending life expectancy after a health shock may result in increasing health expenditure if individuals do not fully recover and continue to contribute to the demand for health care after the shock. Supply-driven variation in time to death may have an effect on the demand of care that may become increasingly relevant for an aging population as the risk of a health shock increases with age.
Finally, we formulated a new econometric model to study the potential substitution between primary care and emergency secondary care. The model extends the class of latent class models for longitudinal count data to the bivariate case by using a hidden Markov chain approach. The proposed new model extends the scope of previous studies by removing the need of an exogenous source of variation for identifying the model, thus allowing the researcher to test for a substitution effect in any segment of the demand of primary care and emergency secondary care. This could be particularly important from a policy perspective as it allows for exploring the potential scope of a new policy before the latter is implemented in a pilot or countrywide.
We expanded the scope of existing literature investigating the impact of time to death on individual health care expenditure. Existing studies examine variation in time to death that steams from the characteristics of the demand for health care, e.g. variation in proximity to death due to age and morbidity. We showed that variation in time to death produced by variation in the supply of care has an important impact on health care expenditure that adds to demand-driven variation. Technological progress directed to extending life expectancy after a health shock may result in increasing health expenditure if individuals do not fully recover and continue to contribute to the demand for health care after the shock. Supply-driven variation in time to death may have an effect on the demand of care that may become increasingly relevant for an aging population as the risk of a health shock increases with age.
Finally, we formulated a new econometric model to study the potential substitution between primary care and emergency secondary care. The model extends the class of latent class models for longitudinal count data to the bivariate case by using a hidden Markov chain approach. The proposed new model extends the scope of previous studies by removing the need of an exogenous source of variation for identifying the model, thus allowing the researcher to test for a substitution effect in any segment of the demand of primary care and emergency secondary care. This could be particularly important from a policy perspective as it allows for exploring the potential scope of a new policy before the latter is implemented in a pilot or countrywide.