As noted recently by Kevin Grumbach, in the past decade, primary care clinicians have experienced both the “best of times and the worst of times.”1 Multiple studies have been published suggesting that for selected conditions specialists may provide better care; one large study reported that a significant percentage of primary care physicians reported feeling pressured to provide care beyond the limits of their expertise, 2 while others have demonstrated the importance of primary care coordination to patients along with some concerns about ease of referrals to specialists. 3 In addition, market pressures have stimulated health care organizations to provide options for consumers to “opt out” of a restrictive relationship with a single physician through point of service and other options. Struggling with their role as gatekeepers while striving to provide patient-centered care, general internists have long been in the forefront of assessing what works in primary care practice with respect to the organization of practice, the impact of primary care on vulnerable subgroups, development and implementation of strategies to improve care in specific clinical domains, and identification of dimensions of patient-centered care. These efforts have been conducted at a time when interest in assessing and improving quality of care has intensified in response to public concerns about managed care and patient safety.
Despite these efforts, however, there is surprisingly little consensus regarding which aspects of primary care practice represent the critical elements of excellence in primary care and which strategies can assure consistent delivery of high quality care. The diversity of patients and problems, the small scale of many practices, and the low volume of any specific condition would appear to resist many straightforward approaches to quality measurement and improvement. While many clinical quality measures now used to accredit health plans are relevant to primary care, by definition measurement in any one clinical domain can only capture a small slice of any clinician's practice. Complementary strategies have included efforts to identify dimensions of primary care from the consumer's perspective, 4 assessment of patient satisfaction or experiences with care, 5 and evaluation of broad dimensions of primary care such as accessibility, continuity, coordination, and comprehensiveness. 6
Two articles in this issue of JGIM provide additional windows into evaluating important dimensions of high-quality primary care. Schillinger and colleagues 7 conducted a prospective randomized trial of gatekeeping in a public hospital. The team evaluated the impact of the intervention on utilization of primary care services, use of the Emergency Department (ED) for nonurgent or low acuity visits, specialty care, hospital admissions, and patient satisfaction. After 1 year, patients in the intervention group were found to have significantly fewer specialty visits and hospital admissions, a nonsignificant increase in primary care visits, and no differences in use of the ED for nonurgent care. Patient satisfaction at baseline and at 1 year was equivalent between the two groups. The researchers attempted to assess inappropriate reduction of specialty care by examining referrals to ophthalmology for diabetic patients, and found that utilization was comparably low (38% of eligible patients) in both groups. There were too few hospitalizations to assess whether the observed reduction in hospitalizations was attributable to those considered ambulatory-care sensitive, and the findings here are reported as differences in the number of hospitalizations rather than differences in length of stay or total hospital days.
These findings provide support for the premise that a gatekeeper model, albeit one not associated with financial penalties or incentives, can enhance outcomes for patients in a public hospital setting. The patients in this study were poorer and reported lower self-rated health than patients likely to be enrolled in commercial managed care plans. Of particular note, these findings suggest that coordination of care, a characteristic of primary care that has been maddeningly difficult to measure, can result in enhanced outcomes with no decrease in patient satisfaction. For general internists struggling to implement similar approaches in comparable settings these findings should be good news. The one specific clinical measure that was assessed, referrals for eye exams, however, while less than optimal, was close to the 50th percentile for commercial plans reporting from that region. 8
The study conducted by Hass and colleagues 9 focused on variations in quality for women with clinical breast complaints or abnormal mammograms, in an effort to identify which physician and patient factors predict variations in quality of care. In this descriptive study, quality of care was defined by compliance with a local guideline developed with the malpractice insurer for appropriate follow-up of women with an abnormal mammogram or a clinical breast complaint, time to resolution of presenting symptoms, and patient-reported quality of care. Overall, 69% of women enrolled in the study received care consistent with guideline recommendations. After controlling for family history of breast cancer, race, severity of mammographic abnormality, degree of worry about breast cancer, insurance status and site of care, women over the age of 50 years and those with an abnormal mammogram were more likely to receive recommended care than younger women or those with a clinical breast complaint. White women were more likely than nonwhite women to receive recommended care and to report higher satisfaction with care. Compliance with the guideline was not associated with patient-reported quality of care.
These results indicate no shortage of opportunities for improvement in the management of clinical breast complaints and follow-up of abnormal mammograms and provoke multiple questions for clarifying different dimensions of quality in primary care. While the authors acknowledge limited information about whether physicians agreed with or had even read the guideline, the implicit premise of the study is that physician knowledge and behavior is the ultimate focus of quality improvement efforts. Yet the findings invite additional inquiry unrelated to the motivation or knowledge of individual physicians. In view of multiple studies demonstrating the importance of the practice ecology or organizational context in facilitating requisite practice change, 10,11 it could be important to learn what is different about the site with significantly improved performance compared with other sites. Of equal importance is the question: what strategies can health care organizations with multiple practice sites use to discover and implement the best within their system? The lack of correlation between patient-reported quality and compliance with recommended practice also prompts the “why” question and invites further study. Understanding why there were observed differences in patient-reported and technical quality of care associated with patient race could inform strategies for understanding and eliminating disparities in health care delivery.
Both studies challenge researchers, clinicians, and those engaged in quality improvement activities to bring renewed emphasis to assessing and improving quality in primary care. Thanks to the efforts of SGIM members and others, we now have a growing collection of findings that suggest areas for improvement, new strategies for measurement, and possible strategies for translating what works in one setting to others. These accomplishments should not obscure how much we have yet to learn to understand the best of primary care. We have very little understanding of how assessment of clinical performance, patients' experiences with care, and organizational characteristics associated with improved outcomes complement one another, or how to match a particular strategy with the purpose. In addition, we still have much to learn from apparently discordant results, such as Haas's finding that women enrolled in managed care plans were more likely to receive recommended care but also were more dissatisfied.
Looking into the future, it is clear that the possibilities for expanding and refining the conceptual frameworks for identifying and replicating the best of primary care have only begun to emerge. The challenges of an aging and increasingly diverse patient population, the largely unexplored frontier of errors in primary care settings, and the possibilities for enhanced use of information technology in routine practice would suggest that the best of primary care practice is only beginning.—