International comparisons are an important tool for benchmarking health system performance, shedding light on health systems’ relative strengths and weaknesses. The present work examines how different groups of countries sharing similar health system characteristics perform relative to others. To make valid and useful comparisons, health systems may be grouped in ways that resonate with policy makers in countries and reflect the policy question at hand. The report specifically addresses three key policy areas: the influence of the overall design of health systems on performance, the role of financial incentives to providers and the role of a strong primary care system. The report shows that there is no indication that any one group of health systems would systematically outperform another. It further provides evidence that there is room for health systems sharing the same broad characteristics to improve performance by borrowing elements from other systems. Rather than engaging in large-scale system reforms, focusing on more targeted policy changes may be a better avenue for improving performance.
How Do Health System Features Influence Health System Performance?
Abstract
Executive Summary
International comparisons are an important tool for assessing health system performance and can raise awareness of health systems’ relative strengths and shortcomings, facilitating international learning and stimulating policy debates. Cross-country comparisons are widely used and valued by decision makers, but the most useful comparisons may not be the most obvious ones. The selection of the most policy relevant “comparator” countries should be determined by the policy question at hand. For example, if a Swiss policy maker were looking for examples on how to improve their primary care system, looking at the British experience may be less useful than looking at countries such as the Netherlands, where the healthcare structure has some similarities to their own. This may mean using different categorisations of health systems based on the characteristics that are most relevant for each policy question.
Clustering is a technique which can be used to form groups of similar health systems that share distinct properties. These shared characteristics might not be visible by simply exploring distributions and studying the effect of system features directly on the outcome of choice. To make valid and useful comparisons of performance, health systems can be clustered based on key qualitative features that underpin healthcare funding and delivery in a given country, for example arrangements to organise population coverage, the financing of healthcare insurance and delivery, the organisation of healthcare delivery, and key aspects of governance and resource allocation.
The analyses contained in this report considered whether different group of counties sharing similar health system characteristics appear to perform better relative to others. Specifically, a focus is whether health system efficiency – a measure that offers insights into performance – is influenced by the overall design of health systems. Health systems of different OECD countries have been clustered based on indicators constructed from countries’ responses to the OECD survey on health systems characteristics. Indicators considered in defining the clusters are the degree of user choice of basic health coverage, degree of public or private provision of primary care and outpatient specialist services, the degree of patient choice of providers, health insurance as a secondary source of coverage (“over the basic” coverage), and the role of primary care in the health system (gate‑keeping). Those indicators were selected as those that most differentiate health systems, and they are the same characteristics that were used to cluster OECD health systems in a previous analysis undertaken in 2008‑10. Nevertheless, there is an element of judgement that is involved in choosing these indicators, rather than others. An efficiency measure was then derived using health spending as a share of GDP as input and age standardised mortality rate as output.
The analysis confirms the results of previous OECD work suggesting that there is no single health system design that is most associated with higher efficiency. In other words, there is no indication that any one group of health systems would systematically outperform another.
The analysis suggests that there is room for health systems sharing the same broad characteristics to improve performance by adopting policy actions to improved efficiency that might borrow elements from other systems. Rather than engaging in wholesale reforms, improving health system efficiency requires the use of more targeted policies to respond to a specific policy question.
To this aim, two additional sets of analysis look at some actionable policy levers available to countries across clusters that seem to be particularly promising for countries to improve performance, regardless of their institutional set-up.
A second set of analyses looks at the role of provider payment mechanisms. The analysis considers whether differences in treatable mortality rates across countries can be explained by clustering health systems based on volume incentives embedded in physicians’ payment schemes and financial incentives for healthcare quality. A third set of analyses then looks at the role of primary healthcare. The analysis considers whether differences in avoidable hospital admissions for selected conditions – asthma, chronic obstructive pulmonary disease and congestive heart failure – can be linked to health systems being more primary care oriented.
Results show that health systems providing a higher degree of incentives for quality to providers may also achieve better access to high-quality care, helping to reduce treatable mortality rates, as compared to health systems relying on limited incentives for quality and more traditional fee‑for-service payments.
Results also indicate that primary care oriented health systems – defined by a stronger role of General Practitioners (GPs) as gatekeepers, better care continuity and stronger financial incentives directed to primary care physicians to improve quality – also display lower avoidable hospitalisations, a variable often used in the relevant literature as a marker of quality and access to primary care.
It is important to note that the relationship between health system clusters and outcome variables should not be interpreted as causal, as the methodology developed to answer the above questions is underpinned by several limitations. In reality, health systems are more nuanced than described in a set of health system indicators. Boundaries between different groups of health policies and institutions are rarely clear-cut. Thus, the panel dataset of health system features provides as complete a picture as possible of the contextual variation across OECD health systems, given data availability. Furthermore, findings are limited by the outcome variables available. In addition, cluster analysis algorithms will always produce a result (a set of clusters) whether there are true patterns in the data or not. When there are natural clusters in the data, there is no way of knowing whether the algorithm has clustered the data “correctly” as there are no “right answers” with which to compare. Finally, the effects of clusters on the outcome variables may be due to health systems features, lifestyle or socio‑economic factors that are not fully captured or controlled for in the analyses.
Despite these limitations, this type of analysis has a real value added for policy makers seeking to understand how changes in certain system features might influence performance. This analysis is also the best of the kind that could be done given the data available. Furthermore, while this report considered two set analyses around how actionable policy levers influence performance, the same analytical approach could be used to understand the impact of many other policy levers, using the rich data set of health systems characteristics.
