As healthcare costs for patients with non-communicable diseases such as diabetes have risen, governments and healthcare providers have sought creative measures to align financial incentives with better patient outcomes. One incentive payment system known as “pay-for-performance” (P4P), in which providers are beholden to metric-driven outcomes, represents a potential path forward for healthcare providers to improve healthcare processes, resulting in higher quality and better patient health outcomes. The evidence on the effectiveness of P4P programs, however, is mixed.

To address this uncertainty, a new study, published in The European Journal of Healthcare Economics, assesses the effectiveness, in monetary terms, of a P4P program for patients with diabetes at a hospital system in Taiwan. 

The study coauthors, including APARC’s Asia Health Policy Program Director and FSI Senior Fellow Karen Eggleston, employed new patient-level data on clinical indicators, utilization, and expenditures, combined with data from the national death registry, to better understand the costs and benefits of the P4P program. Their results show that Taiwan’s implementation of the P4P program for diabetic care yielded positive results in terms of net value, defined as the value of life years gained minus the cost of care.

Sign up for APARC newsletters to receive our experts' research updates.

Assessing Net Value of Taiwan’s P4P

Taiwan’s Bureau of National Health Insurance (now National Health Insurance Administration) introduced P4P in 2001 and enhanced the program in 2006 with an incentive for pay-for-reporting of outcomes. Financial incentives were used to encourage continuity of care with metrics such as new patient enrollment, follow-up visits, and annual reports, each tied to a specific monetary value. The program’s features are common across P4P, so the study’s findings have implications in other settings that incorporate similar designs in their P4P programs.

The study compares two different groups of patients at a large regional hospital in Taiwan, one consisting of newly enrolled P4P patients and another using P4P patients who have been enrolled since the beginning of the program. The researchers leverage detailed clinical data not used in previous assessments of the P4P program to better identify both costs and longer-term clinical outcomes based on measured biomarkers and predicted mortality.

Using an economic cost-benefit analysis conducted from a budgetary perspective, the study is the first analysis of any P4P program that estimates changes in the quality-adjusted price index relative to usual care. The authors consider health benefits in terms of survival and predicted survival and convert them into monetary terms. This net value approach is especially useful for policymakers and healthcare administrators who implement value-based purchasing and monitor outcomes for any service delivery innovation over time.

The study finds that Taiwan’s P4P program provided a positive net value for payers and patients, ranging from $40,084 USD to $348,717 USD. These positive net value results are primarily derived from health outcomes as measured by lower mortality rates in the P4P versus non-P4P cohorts, across both newly enrolled and continuously enrolled groups of patients. According to the authors, “these encouraging findings of the positive value of quality improvement net of expenditures adds evidence to the literature that has found mixed results of P4P programs.” 

 This study develops a new model for assessing the net value of service delivery innovations like P4P programs that can be applied in other contexts globally, providing healthcare systems researchers with new tools to better understand an emergent option for incentivized care. With a more economically-translatable understanding of P4P programs, this research helps build the bridge between the oft-disparate worlds of healthcare and policy.

This press release was originally published by the University of Tokyo. It discusses a new study, published by Lancet Public Health and co-authored by researchers from the University of Tokyo along with FSI Senior Fellow Karen Eggleston, director of the Asia Health Policy Program at APARC, and Jay Bhattacharya, Professor of Health Policy and director of the Center for Demography and Economics of Health and Aging at Stanford.

A new detailed microsimulation, using a database of 40 million people, has examined the future of Japan’s aging population up to 2043. It projects that more people will live longer, and that overall years spent living with dementia will decrease. However, the model highlighted the diversity of impacts on different segments of the population, as Japanese women with a less than high school education aged 75 and over may be disproportionately affected by both dementia and frailty. Better understanding where health gaps like this exist can help inform public health planning, to minimize future economic costs and support those most in need.

Taking care of the older members of society is a common concern around the world. Japan is famous for its long-lived residents, the number of which continues to rise. In 2020, almost 30% of the Japanese population was aged 65 years or older, and this age group is not projected to peak until 2034. Caring for people with age-related ailments, such as dementia and frailty, poses a challenge both to individuals and public health care systems.

Microsimulation models, which are computer models that can provide detailed analysis on an individual basis, are currently used to project future population health in some countries, such as the U.K. and the U.S. Professor Hideki Hashimoto and researchers at the University of Tokyo, along with researchers from Stanford University in the U.S., wanted to create a new microsimulation model for Japan, which would take into account more diverse conditions than had been considered before.

“We developed a new Japanese microsimulation model that accounts for 13 chronic conditions (including heart disease, stroke, diabetes, depression and dependency), as well as frailty and dementia,” explained Hashimoto. “Using an ultralarge data system, we were able to ‘follow’ a virtual cohort of more than 40 million people aged 60 and over from 2016 to 2043.”

According to Hashimoto, projections of aging in Japan usually rely on the “average” status of older people and so don’t consider the diversity of the population. “I believe that problems of aging are a matter of health gaps over the course of people’s lives,” he said. “Our projection brings attention to a widening health gap among older people. It highlighted that women with a less than high school education aged 75 or over are more likely to be affected.”

Identifying where health gaps like this exist could be used to better inform public policy, not only about health care but other influential aspects of life. “Japan’s case may suggest that improvement in educational attainment, as well as population health, could be a key to making a healthier and more manageable aging society,” said Hashimoto.

Positively, this study shows hope for a future where many people live longer and more healthy lives. “People might believe that an increase in cases of dementia is inescapable, given population aging. However, in this study we found that in Japan, despite an aging population, the number of people with dementia is expected to decrease over the next two decades,” said Hashimoto. “Population aging does not necessarily mean an increase of social burden for care, but it does bring a diversity of problems that requires careful study and science-based policy attention, to close the health gap.”

Modeling Health Outcomes Among East Asian Populations

The chronic progression to diabetes-related complications is apt for computer simulation modeling due to the long-term nature of health outcomes and the time lag for interventions to impact patient outcomes. It is problematic, however, to estimate the impacts of health interventions on East Asian populations with diabetes using existing models, which were developed and validated in European and North American populations with different epidemiology and outcomes.

To fill in this gap, Eggleston and her colleagues set out to develop and validate an outcomes model for the progression of diabetes and related complications in Chinese populations. They compared this new model, called the Chinese Hong Kong Integrated Modeling and Evaluation (CHIME), to two widely used existing models developed and validated in the United Kingdom (known as the United Kingdom Prospective Diabetes Study Outcomes Model 2, or UKPDS-OM2) and in the United States/Canada (called Risk Equations for Complications of type 2 Diabetes, or RECODe). Despite the continuum of risk across the spectrum of risk factor values, these two existing models ignore the progression from prediabetes to diabetes.

The CHIME integrates prediabetes and diabetes into a comprehensive model comprising 13 outcomes. These include mortality, micro- and macrovascular complications, and the development of diabetes. The researchers developed the CHIME simulation model using data from a population-based cohort of 97,628 participants in Hong Kong with type 2 diabetes (43.5%) or prediabetes (56.5%) from 2006 to 2017. Known as the Hong Kong Clinical Management System (CMS), this cohort makes one of the largest Chinese electronic health informatics systems with detailed clinical records. 

The next step was to externally validate the CHIME model against individual-level data from the China Health and Retirement Longitudinal Study (CHARLS) cohort (2011-2018), a nationally representative longitudinal cohort of middle-aged and elderly Chinese residents age 45 and older. The researchers validated the CHIME model against six outcomes measures recorded in the CHARLS data and an additional 80 endpoints from nine published trials of diabetes patients using simulated cohorts of 100,000 individuals.

Towards Reducing the Disease Burden of Diabetes

The researchers found that the CHIME model outperformed the widely used UKPDS-OM2 and RECODe models on the data used, meaning that the validation of the CHIME model was more accurate for trials with mainly Asian participants than trials with mostly non-Asian participants. The results indicate that the CHIME model is a validated tool for predicting the progression of diabetes and its outcomes, particularly among Chinese and East Asian populations, for which the existing models have been unsuitable.

With the new model, clinicians and health economists can evaluate population health status for prediabetes and diabetes using routinely recorded data and therapies related to the long-term management of diabetes. In particular, the CHIME outcomes model enables them to assess patients' quality of life and measure cost per quality-adjusted life-years over the long-time horizon of chronic disease conditions. The new model thus supports the economic evaluation of policy guidelines and clinical treatment pathways to tackle diabetes and prediabetes, address micro- and macrovascular complications associated with these conditions, and improve life expectancy.