|
|
Background The Health Disparities Collaboratives of the Health Resources and Services Administration (HRSA) were designed to improve care in community health centers, where many patients from ethnic and racial minority groups and uninsured patients receive treatment.
Methods We performed a controlled preintervention and postintervention study of community health centers participating in quality-improvement collaboratives (the Health Disparities Collaboratives sponsored by the HRSA) for the care of patients with diabetes, asthma, or hypertension. We enrolled 9658 patients at 44 intervention centers that had participated in the collaboratives and 20 centers that had not participated (external control centers). Each intervention center also served as an internal control for another condition. Quality measures were abstracted from medical records at each health center. We created overall quality scores by standardizing and averaging the scores from all of the applicable measures. Changes in quality were evaluated with the use of hierarchical regression models that controlled for patient characteristics.
Results Overall, the intervention centers had considerably greater improvement than the external and internal control centers in the composite measures of quality for the care of patients with asthma and diabetes, but not for those with hypertension. As compared with the external control centers, the intervention centers had significant improvements in the measures of prevention and screening, including a 21% increase in foot examinations for patients with diabetes, and in disease treatment and monitoring, including a 14% increase in the use of antiinflammatory medication for asthma and a 16% increase in the assessment of glycated hemoglobin. There was no improvement, however, in any of the intermediate outcomes assessed (urgent care or hospitalization for asthma, control of glycated hemoglobin levels for diabetes, and control of blood pressure for hypertension).
Conclusions The Health Disparities Collaboratives significantly improved the processes of care for two of the three conditions studied. There was no improvement in the clinical outcomes studied.
Strategies to reduce disparities generally take one of two forms. In some cases, targeted programs focus on care for a particular class of patients. For instance, a quality-improvement program might be developed for a specific population, such as blacks with hypertension. A more common approach, however, is the development and implementation of programs to improve quality more broadly, with an expectation that across-the-board improvements in quality will narrow gaps in care.8,9 Moreover, these broader programs may target settings that care for large proportions of disadvantaged populations. One of the most important national initiatives of this type is the Health Disparities Collaboratives sponsored by the Health Resources and Services Administration (HRSA).
These collaboratives bring community health centers together to learn and disseminate quality-improvement techniques developed by the Institute for Healthcare Improvement.10,11,12 The rapid-cycle improvement method first requires the establishment of aims based on known deficiencies in quality. Then, each community health center implements and tests small-scale interventions at one or more practice sites. On the basis of these tests, new practices and procedures are adopted and refined. Successful interventions are then disseminated throughout the entire community health center. Since 1998, about two thirds of community health centers (645 centers to date) have voluntarily participated in collaboratives focusing on improving care for chronic medical conditions. To date, however, there has been no controlled evaluation of the effect of the Health Disparities Collaboratives on the quality of care. We report the results of a controlled, national evaluation of these collaboratives for the care of patients with three prevalent chronic medical conditions.
Methods
Community Health Centers
We enrolled community health centers participating in collaboratives to improve the care of patients with diabetes, asthma, or cardiovascular disease. For cardiovascular disease, we focused our assessment on the improvement of care for patients with hypertension. We compared each intervention center with a participating center in another collaborative and a center that had never participated in a collaborative, using a difference-in-differences design. The earliest collaborative started on January 1, 2000, and the latest collaborative started on August 1, 2001. Our study is based on data from care delivered during the period from January 1, 1999, to August 1, 2003.
Of the 238 eligible community health centers identified by HRSA as participating in an eligible collaborative, 138 (58%) agreed to participate in our independent evaluation. Of these centers, 44 intervention centers were selected on the basis of region, location (rural, urban, or mixed), the number of practice sites included in the center, and caseload. Each of these centers was also asked to serve as an internal control for one of the other conditions under study, and 40 were able to do so. Potential external control centers (that had never participated in a collaborative) were then matched with intervention centers according to the same variables that were used to select the intervention centers, yielding 34 potential external control centers of which 20 (59%) agreed to participate in the study. Three external control centers provided data for only one or two of the three conditions. For each center, we included one or two practice sites in the evaluation: the lead collaborative site, which was usually the largest site and included the team leader for the intervention, and one additional practice site that was randomly selected. Although most of the health centers in the collaboratives initially focused their efforts on a smaller population (e.g., at a specific practice site), successful interventions were designed to be spread to the entire community health center population.
Quality-Improvement Intervention
The Health Disparities Collaboratives have been described in detail previously.13 Briefly, each collaborative generally includes 20 or more community health centers and consists of a prework period, a kickoff meeting, and subsequent 2-day learning sessions. During the learning sessions, improvement teams from each organization receive instruction on quality-improvement techniques, the use of a software registry program, and the Chronic Care Model, which identifies the essential elements of health care delivery systems that are necessary to provide high-quality care to patients with chronic diseases and thus helps centers identify target areas for interventions.14 Teams also have opportunities to share knowledge gained from their own experiences with rapid-cycle improvement techniques. Between the sessions, during "action periods," team members implement improvements based on the ideas discussed in the sessions. Each community health center also has access to a collaborative users group on the Internet, participates in monthly conference calls with the collaborative leaders, and submits monthly reports on its improvement interventions. The HRSA also developed a regional and state infrastructure that provides technical assistance and information systems support to community health centers participating in the collaboratives.
Study Population
We selected sequential, random samples of patients with each condition during the 1-year period before the beginning of the collaborative and the 1-year period beginning 1 year after the completion of the collaborative. These samples were selected with the use of electronic lists generated by each center on the basis of automated billing data and diagnostic codes. Each patient had to have been seen at the center at least once during the relevant measurement year and at least once before the measurement year. From each of these lists, we randomly selected 40 patients after excluding all patients with end-stage renal disease, cancer, or human immunodeficiency virus infection. For the group of patients with diabetes and hypertension, we excluded patients who were younger than 18 years of age or pregnant. For the group of patients with asthma, we excluded patients who were younger than 2 years of age. For the results presented here, we also excluded patients younger than 6 years of age because of the uncertainty of the asthma diagnosis in this population. Including the additional patients between 2 and 6 years of age did not substantively alter our results.
Review of Medical Records
One to four abstractors at each health center abstracted data from the medical records for the 1-year reporting periods. These data included sociodemographic information, coexisting medical or psychiatric illnesses, and disease-specific quality indicators for preventive care and screening, disease monitoring and treatment, and intermediate outcomes of care. The quality-of-care measures (Table 1) were selected to coincide with the required and optional quality-of-care measures identified by the collaborative faculty as areas for improvement. These measures were supplemented with existing validated measures. Because relatively few patients qualified for advice about smoking cessation, we created a composite measure that included both assessment of smoking status and advice about cessation.
|
We compared the characteristics of the study centers with those of community health centers nationally, the characteristics of the intervention centers with those of the control centers, and, for each condition, the characteristics of the patient populations at the intervention centers with those of the patient populations at the control centers. We created an overall quality score and composite scores for prevention and screening, disease monitoring and treatment, and outcomes for each condition and for all conditions by averaging the scores of all of the indicators applicable to the patient. Because the number of applicable measures varied among patients and the proportion met varied among measures, the component scores were standardized to a mean of 0 and a variance of 1 before averaging. After averaging, we scaled the scores to have the same mean and variance as the overall proportion of measures met in the sample. Mean changes from baseline to follow-up within each of the three groups (intervention centers, internal control centers, and external control centers) and the relative change over time between the groups were assessed for each clinical measure with the use of hierarchical linear and logistic-regression models that controlled for age, sex, race or ethnic group, insurance status, and an adapted version of the Charlson comorbidity index. We also performed a sensitivity analysis, including a term to test whether the intervention had a stronger effect in the lead practice site at the community health center than in the second selected site.
Finally, because community health centers might focus their improvement efforts on a particular condition, thereby diverting attention away from another condition (negative spillover effect), or because they might transfer the continuous quality-improvement techniques they learned as part of the collaborative to another condition (positive spillover effect), we also compared measures of improvement separately in the internal and external control centers. Greater improvement in the internal control centers than in the external control centers would be evidence of a positive spillover effect and vice versa.
Results
Study Centers
The study centers were representative of community health centers nationally. Approximately half of the study centers were located in urban areas, and the centers were well distributed throughout the country. The average study center had approximately seven distinct sites of care (Table 2).
|
|
We studied 9658 patients with one of the three target conditions (3392 with asthma, 2904 with diabetes, and 3362 with hypertension) (Table 4). The average number of patients per center in the preintervention and postintervention periods was 41 (range, 16 to 84; median, 40). For illustrative purposes, we describe the patients with diabetes, but patients in the three groups were all similar, with the exception that patients with asthma were younger. Approximately 60% of the participants with diabetes were women, and the average age was approximately 55 years. About 25% of these patients were covered by Medicaid, and 20 to 26% had no insurance.
|
The percentage of patients who received treatment with an antiinflammatory medication for persistent asthma increased by 14 more percentage points in the intervention centers than in the external control centers (P<0.01) and by 8 more percentage points than in the internal control centers (P>0.05) (Table 5). The percentage of patients with an asthma management plan increased by 17 more percentage points and by 16 more percentage points in the intervention centers than in the two control groups, respectively (P<0.001 for both comparisons). For diabetes, the percentage of patients with two or more assessments of glycated hemoglobin levels increased by 16 more percentage points in the intervention centers than in the external controls (P<0.001) and by 12 more percentage points for the comparison with internal controls (P<0.01). The only hypertension measure that showed significantly greater improvement in the collaborative centers than in the control centers was the assessment of smoking status and advice about cessation, if appropriate (an increase of 12 percentage points as compared with internal controls, P<0.05).
|
For all three conditions, the intervention centers improved care by an additional 4.9 percentage points as compared with internal controls and by an additional 4.5 percentage points as compared with external controls (P<0.001 for both comparisons) (Table 6). The intervention centers also had significant improvement in the composite indicators for prevention and screening, with an increase of 6.2 percentage points as compared with internal controls (P<0.001), and an increase of 4.5 percentage points as compared with external controls (P<0.01). The composite indicators for disease monitoring and treatment also improved significantly (5.5% for internal controls and 5.9% for external controls, P<0.001). However, there was no significant improvement in intermediate outcomes.
|
Overall, there was no significant difference in improvements between the internal and external controls for any of the composite measures we examined. For the individual diseases, the pattern was mixed and did not show consistent evidence of negative or positive spillover effects. (See the Supplementary Appendix, available with the full text of this article at www.nejm.org.)
Discussion
We conducted a large-scale evaluation of an intervention to improve the care at community health centers for patients who have diabetes, asthma, or hypertension, three prevalent chronic diseases that in aggregate affect more than 25% of the adult population in the United States.23 A quality-improvement intervention based in community health centers is particularly relevant today because of their substantial growth in both the number of sites and numbers of patients served over the past decade, their prominent role in providing care for members of minority groups and other disadvantaged populations, and President George W. Bush's stated goal to expand their number in order to increase access to care for the uninsured.
Our study showed that participating in a collaborative improved the processes of care related to prevention and screening and disease monitoring and treatment for diabetes and asthma, but participating in a collaborative did not improve these processes for hypertension. The study did not show improvements in intermediate outcomes of care. Whether these differences in diseases and outcomes are related to the nature of each targeted condition, aspects of the participating organizations, ease of patient adherence, or the specific collaborative under study is not known. Finally, improvements over time were seen in both intervention and control centers for almost every individual measure. Whether this broad improvement reflects general secular trends20 or the general quality-improvement environment at community health centers is not known.
There are several potential explanations for the lack of effect with respect to intermediate outcomes. First, many of the processes of care that we studied (e.g., retinal examinations for patients with diabetes) are linked to longer-term outcomes but are not directly linked to the intermediate outcomes that we examined. For those that are related to intermediate outcomes, the magnitude or importance of the linkage might not be strong. For instance, even though glycated hemoglobin must be monitored in order to achieve and maintain optimal levels, the documentation of this monitoring does not necessarily mean that important changes in management by the clinician or by the patient have been undertaken in response to suboptimal results. In addition, to the extent that the processes of care we examined are linked to these intermediate outcomes, it would be difficult to detect a difference in outcomes solely related to the modest improvements in the processes that we observed. Finally, achieving improvements in both longer-term and intermediate outcomes may require more intensive interventions in order to overcome environmental factors that pose particular challenges for patients treated at community health centers.
A focus on intermediate outcomes may also detract attention from some of the most important evidence-based processes that can be improved. For instance, smoking cessation and daily aspirin for men at high risk for coronary artery disease are among the most powerful clinical interventions we measured in this study. Yet the associated outcomes occur in the longer term, and a longer study period and larger sample would be required to document a meaningful effect. This focus on outcomes that can be measured in the short term to the exclusion of important longer-term outcomes may understate the true effect of quality-improvement collaboratives. Moreover, this issue has implications not only for these types of programs but also for other programs that are designed to improve quality, including pay-for-performance and public reporting programs that may divert attention from longer-term improvement.
Several other studies have explored the value of collaboratives for the care of patients with chronic disease.24,25,26,27,28 The results of these studies vary. For example, for the care of patients with asthma, one study showed that a quality-improvement collaborative made no difference as compared with a control group, whereas another study showed improvement in self-care but not in other measures.25,28 For patients with diabetes, there was no effect on quality of care when six intervention centers were compared with control groups affiliated with the same organizations.24 However, a preintervention and postintervention analysis of a quality-improvement collaborative in community health centers in the Midwest showed substantial improvement in the care of patients with diabetes, as compared with national benchmarks.29
Although we evaluated quality-improvement collaboratives based on the Institute for Healthcare Improvement model — the most prevalent and reproducible type of quality-improvement program — many variations on this model and other approaches to quality improvement have been tried.30,31,32 There is still much to learn about the tools and methods for quality improvement and their potential effectiveness.33,34 More research would be helpful regarding the operations and effectiveness of individual quality-improvement collaboratives as well as the broad range of organizational factors that can affect their success.31,32,33,35 In addition, although our study did not show evidence of spillover effects, we need a better understanding of the effects of focused quality-improvement interventions on other aspects of care.
Our study was subject to several limitations. First, we were not able to perform a pure randomized trial of the intervention. Instead, we relied on matching and statistical models to adjust for potential confounding variables. Second, although we assessed important markers of the quality of care that were the main focus of the collaboratives, some clinics might have improved areas of care (e.g., patients' experiences) that we did not measure. Third, the collaboratives for the care of patients with cardiovascular disease were focused on a set of goals broader than just control of hypertension, so there might have been improvement in other areas of care or for other populations with cardiovascular disease that we did not study. Fourth, we included in the study the lead collaborative site at each community mental health center, as well as one additional site. Thus, if the intervention was not fully implemented throughout the center, our study may have overestimated the effect of the intervention. Stratified analyses according to the type of intervention site showed findings that were largely consistent with the primary analyses presented here. Finally, some of the improvements we observed might have resulted from improved documentation, rather than improved care. However, measures that might be more sensitive to the effects of documentation (e.g., smoking-related measures) did not show more improvement than measures that required an action (e.g., the assessment of glycated hemoglobin levels).
The HRSA Health Disparities Collaboratives are an important national initiative to improve the quality of care for underserved populations at community health centers. Our study showed that these collaboratives significantly improved several processes of care without any observed improvement in intermediate outcomes. The substantial room for improvement in the postintervention period suggests the need for continued refinement of these methods.
Supported by grants from the Agency for Healthcare Research and Quality (1U01 HS13653), the HRSA, and the Commonwealth Fund (20030185).
No potential conflict of interest relevant to this article was reported.
We thank Yang Xu for statistical programming; Mary Ly, Lynn Huynh, and Adam Lessler for research assistance; and Laura Peterson for assistance with developing chart-abstraction instruments and the training of abstractors. We also thank our colleagues at the HRSA and at the participating community health centers without whom this research could not have been completed.
Source Information
From the Department of Health Care Policy, Harvard Medical School (B.E.L., L.S.H., A.J.O., T.K., B.J.M., E.G.); the Division of General Medicine and Primary Care, Beth Israel Deaconess Medical Center (B.E.L.); the Division of General Internal Medicine (L.S.H.) and the Department of Radiology (B.J.M.), Brigham and Women's Hospital; and the Department of Ambulatory Care and Prevention, Harvard Pilgrim Health Care and Harvard Medical School, and the Division of General Pediatrics, Children's Hospital (T.A.L.) — all in Boston.
Address reprint requests to Dr. Landon at the Department of Health Care Policy, Harvard Medical School, 180 Longwood Ave., Boston, MA 02115, or at landon{at}hcp.med.harvard.edu.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
Improving the Management of Chronic Disease
Sadof M. D., Rosenbaum S., Smolkin M. T., Selby J. V., Mangione C. M., Gerzoff R. B., Landon B. E., Hicks L. S., Guadagnoli E.
Extract |
Full Text |
PDF
N Engl J Med 2007;
356:2422-2424, Jun 7, 2007.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2010 Massachusetts Medical Society. All rights reserved. |
Previous
