Rationale for context informing comparative outcome research to advance improvements in health systems and primary care

Most health problems are managed in the community, where a large majority of the population are cared for, most of the time (Green et al., Reference Green, Fryer, Yawn, Lanier and Dovey2001). As a consequence, local circumstances define the care that can be provided. This applies to the health system itself and its resources, as well as to the socio-economic circumstances and determinants of health (CSDH, 2008). Primary health care with its community base exemplifies the requirement for considering the context of care delivery and the variability it creates in the health status of the population it serves. Every setting is unique and its attributes need to be understood as a precondition for implementing quality care (van Weel, Reference Weel2007). What is true for primary health care is true for health care in general: that it is shaped by the circumstances in which it is provided. A strategy for diagnosis, prevention or treatment that is effective in one setting is not necessarily effective elsewhere (Siregar et al., Reference Siregar, Komarudin, Wisaksana, Crevel and Baltussen2011).

Variation under which health care is provided is considered within some empirical paradigms as a research bias, which needs to be accounted for via study design and analysis. This paper argues the case to use this variation as an asset, to help to explain how seemingly similar interventions produce diverse outcomes. We argue that analysing and comparing outcomes across settings, that is comparative outcomes, allows us to understand the attributes of context, or real-life setting and conditions, that co-determine outcomes of care. Moreover, to advance scientific knowledge about the impact of context on variability in the quality and outcomes of care we specify a set of core variables that might be considered in conducting such research in primary care.

Understanding the attributes of the context and conditions in which health care is delivered, and the way this enhances or impedes its impact is important to further the effectiveness and quality of care. Comparisons of outcome of care between settings can be powerful in this regard. A recent example of the benefits of considering variability comes from a review of randomised trials conducted into health workers in sub-Saharan Africa, which identified nine contextual factors associated with performance (Blacklock et al., Reference Blacklock, Gonçalves Bradley, Mickan, Willcox, Roberts and Bergström2016). International collaboration can support health systems to innovate and change (van Weel et al., Reference Weel, Turnbull, Whitehead, Bazemore, Goodyear-Smith, Jackson, Lam, Linden, Meyers, Muijsenbergh, Phillips, Ramirez-Aranda, Tamblyn and Weel-Baumgarten2015). For example, pharma care coverage is a hot topic in many countries because all are challenged to produce optimal value for investment. The experience in Quebec was the first jurisdiction to institute mandatory universal prescription drug coverage through a public and private insurance approach, other regions and countries could learn from this experience through a comparison with their own performance: The Quebec policy reform showed a reversal of the international trend for poorer compliance in the most economically disadvantaged (Tamblyn et al., Reference Tamblyn, Eguale, Huang, Winslade and Doran2014). Advancing the capacity to conduct international comparisons provides unique opportunities to examine the role of different health systems and policies on health outcomes and equity.

It is important in this context to clarify and critically appraise the nature of outcomes that need to be considered in health systems and community-based primary care research. A good example is the ‘paradox of primary care’; while subspecialists may achieve better disease-specific results, in comparison to generalists – primary care results in enhanced population health (Stange and Ferrer, Reference Stange and Ferrer2009). Functional health status (Huber et al., Reference Huber, Knottnerus and Green2011) may be a more relevant outcome to inform primary health care and public health, than markers of disease while the latter will be pertinent to subspecialist interventions. Outcome selection will be justified by the study focus; specifically the impact of care on patients or populations, and/or the process or structure in which care is provided (Donabedian, Reference Donabedian1988; Reference Donabedian2005).

Implications and measuring quality

To compare outcomes of care asks for an analysis of the most important differences and similarities between settings. Understanding and interpreting variation are key features of comparative outcome research and this is at-odds with those paradigms of health research that aim to control the circumstances in which interventions are studied. Future research will continue to require the use of more sophisticated choice of designs that are appropriate for comparisons of complex interventions in different contexts, that will need to be informed by the views of multiple stakeholders including consumers and policy makers. Combining different domains and data sets (The National Centre for Geographic and Resource Analysis) is of great importance, while not all that matters may have been or can be, quantified. Capturing quantitative and qualitative data will enhance the understanding of differences between settings, requiring the use of use of matrices to combine mixed methods (Miles et al., Reference Miles and Huberman1984). A substantial literature exists on alternative research paradigms for dealing with local context, both how to characterise it and how to modify trial design to measure its impact, founded in principles of realistic evaluation first developed about 20 years ago (Tilley and Pawson, Reference Tilley and Pawson1997).

The quality of such research is marked by the richness of naturally available variation that can be included in the study of natural experiments, enriched by the availability of substantial advancements in methodologies to control for potential biases in observational research. This is essential for research to support practice and move from the creation of knowledge about what interventions can work to insight of their value of what works under different prevailing health care contexts that will allow us to optimise their effects. To achieve the expected advancement in the health of the population, we need to balance considerations and research investment to accelerate the transition from efficacy to effectiveness to implementation and to innovation in care (Box 1).

Box 1 From efficacy to system redesign

The issue: describing the context, especially in relation to primary care

Box 2 presents attributes that should be measured in assessing the context of care. To illustrate its application, a powerful example from two countries with highly comparable population health and socio-economic status, and also comparable level of medical education, is used as it shows the impact of an essentially different basis of financing general practice. This is especially notable in primary health care where professionals specialise in-depth in the patients’ context, so as to be able to address the full breadth of their health problems. Comparing outcomes between primary health care studies therefore would benefit from a standardisation of how to describe the setting and context. To support comparative outcome studies we propose the development of a core set of primary health care sensitive measures.

Box 2 Antibiotics prescribed for respiratory conditions: Belgium and the Netherlands

Measuring the context of health care

A number of domains can be distinguished in the real-life world in which health care is provided, and which can be included in the reporting of studies to support policy makers and may facilitate the generation of hypotheses of ‘context that matters’:

• ∙ Health system

Structure of health care towards access: navigated through primary health care versus patients’ freedom to access every physician; insurance and coverage (including that for mental health); availability of services; financial barriers for patients (co-payment, deductible); payment of provider (capitation, item for service, performance incentives); the contract relation between patient and provider: patients listing (rostered) with primary health care practices).

• ∙ Social welfare

  Pensions; unemployment benefits; sickness benefits; community support services for social needs.

• ∙ Population and society

  Population demographics (gender, age, social class, education and employment status, ethnicity, religious convictions, health status markers).

• ∙ Objectives of diagnosis and treatment

A diagnostic intervention can aim to rule-in or rule-out a health problem or specify individual risk status. Treatment can have the objective to prevent a health problem, cure it, provide palliation and symptom relief or improve functional capacity.

Consistent reporting of the study context will ensure rigour in comparative outcome analysis and inform professionals, policy makers and service users how the context of care may enhance or impede outcome. A proposal to standardise descriptions is given in Box 3.

Box 3 Presentation of information on context of care