The clarion call to focus on quality of health care has been heeded by many and has led to a body of research aimed at identifying factors associated with improved quality of care.1–6 In this issue, Friedberg et al.7 explore the association between quality of care—as measured by Health Plan Employer Data and Information Set (HEDIS) indicators—and two characteristics of physician group practices. Specifically, Friedberg et al.7 examined whether HEDIS measures varied by the size of the physician group and by whether or not the physician group was a member of a “physician network”. They reviewed data from five managed care organizations that encompassed contracts with more than 4,300 primary care physicians in group practices in Massachusetts. Friedberg et al.7 found that in multivariate models, physician group size was associated with only 3 of 12 HEDIS measures. Of note, whereas for two diabetes-related HEDIS measures medium-sized groups had more favorable performance scores, for another diabetes-related HEDIS measure, small physician groups had the more favorable performance scores.
The data regarding an association between network affiliation and HEDIS performance yielded more consistent results. Specifically, Freidberg et al. found that in multivariate analyses, network affiliation was associated with higher HEDIS performance scores in 8 (of 12) measures. These findings were fairly robust, overall holding up well under various conditions such as excluding the two highest performing networks from one analysis and excluding the two largest networks from another analysis.
How does the examination of these organizational constructs (group size and network affiliation) contribute to our understanding of quality care and what are the policy implications of the data presented by the findings by Friedberg et al.?7 From a policy perspective, it is particularly important to consider interventions that are not just effective but are also feasible, efficient, relatively generalizable to a variety of healthcare settings. Furthermore, interventions that can be applied at the system level are attractive, as they do not rely solely on the vagaries of individual physician behavior.
The analyses on group size failed to reveal a pattern that could be used to develop and test an intervention that has high potential for meeting even the fundamental policy criteria of being effective. Why? Because in their analysis, there was no preponderance of evidence that any one group size had a distinct advantage over another vis-à-vis the selected HEDIS performance measures. This is not necessarily a disappointing result. Indeed, as the ability of primary care physicians to organize themselves into medium or large group practices is likely constrained by a number of factors (geographic location, population density, etc.), the data by Friedberg et al.7 offer some comforting assurance that there are not huge missed opportunities to improve quality of care imposed by the constraints on altering group practice size.
The analyses on network affiliation yielded more promising policy implications. Network affiliation was associated with improved HEDIS performance scores across a majority of measures. The policy implications are intriguing because if network affiliation is truly an important player in the causal pathway toward quality, then developing broad-based strategies to enhance physician group enrollment into networks would be feasible relatively efficiently and potentially generalizable to a variety of practice settings. Indeed, during the 1990s, there was rapid expansion of the number of physicians affiliated with networks.8 At the end of the day, however, the findings presented by Friedberg et al.7 are intriguing and provocative, but insufficient in and of themselves to be the sole impetus for sound policy development. Their findings, however, when taken into account with studies looking at the impact of other additional organizational strategies such as pay for performance9 should provide healthcare markets with sufficient information to begin to implement and test the impact of specific interventions on quality measures in a prospective manner.