|
The Quality Of Chronic Disease Care In U.S. Community Health Centers
Community health centers (CHCs) are responsible for providing care for more than fifteen million Americans, many of whom are members of groups who have been documented to receive low-quality care. This study examines the quality of care for patients with chronic disease in a nationally representative sample of federally funded CHCs. Fewer than half of eligible patients received appropriate care for the majority of indicators measured, and uninsured patients received poorer care than insured patients. Although the quality of chronic disease care in CHCs compares favorably with that of care received in other settings, gaps in quality were observed for the uninsured.
A NUMBER OF STUDIES EXAMINING the quality of health care in the United States have documented major problems with quality and disparities according to patients’ race and socioeconomic status.1 These issues are of particular concern for publicly supported community health centers (CHCs), which are responsible for caring for more than fifteen million Americans, 23 percent of whom are uninsured and 64 percent of whom are members of immigrant or minority groups. Such groups have been previously documented to receive lower-quality care than insured, nonimmigrant/minority patients receive.2 The number of patients served by federally qualified CHCs increased by almost 50 percent during 1999–2004.3 Growth is likely to increase in the future because of anticipated changes in Medicaid eligibility rules, decreases in state-funded insurance programs, the rising cost of private insurance, and the projected expansion of the CHC sites available nationally as part of the 2002 Federal Health Center Growth Initiative.4 Although prior studies have examined the quality of care within CHCs, these studies have suffered from methodological limitations that limit their general-izability. For instance, some studies have examined a limited number of CHCs in one state or region.5 Others were nationally representative studies but limited their populations to children or the elderly, examined care only among the poor or Medicaid recipients, examined only preventive services or outcomes such as use of the emergency department (ED), or were unable to examine the relationship between patient or CHC characteristics and quality.6 To provide a better understanding of the quality of care delivered to a representative sample of patients in CHCs nationally, and of the patient and CHC characteristics that could be associated with quality, we examined the medical records of more than 5,600 patients receiving care for one of three chronic medical conditions in a national sample of sixty-four publicly funded CHCs throughout the United States.
The random sample of patients in our study were participating in an ongoing evaluation of quality improvement collaboratives sponsored by the Health Resources and Services Administration (HRSA). We performed a baseline assessment of quality for hypertension, diabetes, and asthma. Data from this assessment will be used to document any progress the collaboratives make in improving care for these chronic diseases. Study sites. Of 238 eligible centers identified by HRSA as participating in the Asthma, Diabetes, or Cardiovascular Collaboratives, 138 (58 percent) agreed to participate in our evaluation. From these, forty-eight centers were selected for participation in our study, including seventeen for diabetes, fifteen for hypertension, and sixteen for asthma. Selection was based on region, location (rural/urban/mixed), number of sites, and caseload. Each of these centers also provided baseline data for one of the other conditions under study. CHCs that had never participated in a collaborative were selected as potential control centers and matched with intervention centers using the same variables as above, yielding thirty-four potential control centers, of which twenty-two (65 percent) agreed to participate in our study. Four participating and two control centers dropped out of the evaluation because they were unable to identify appropriate patient lists, leaving a final study sample of forty-four intervention and twenty control centers. Five control centers provided data for only two of the three conditions, resulting in a total of forty-eight centers contributing records of diabetes patients, forty-two centers contributing records of hypertension patients, and forty-eight centers contributing records of asthma patients for medical record review. Some combination of clinician notes, medication lists, and laboratory tests were available in the form of an electronic medical record (EMR) for forty-one (64 percent) of the study’s CHCs. Patient sample. We selected random samples of patients with one of the diagnoses of interest during the one-year period prior to the beginning of the applicable quality improvement collaborative. Each CHC used clinic administrative data to generate lists of unique patients who had at least one visit to the center during the appropriate twelve-month period as well as in the year prior to the study period and who had received care for hypertension, asthma, or diabetes. The earliest one-year period of care started 1 January 1999, and the latest started 1 August 2000. Thus, all of the patient care examined occurred within a period of approximately two years. From each list, we randomly selected forty patients for each condition after excluding all patients with end-stage renal disease (ESRD), malignancy, or HIV infection. For diabetes and hypertension, we excluded patients under age eighteen and pregnant women; for asthma, we excluded patients under age two. Medical record review procedures. One to four staff members at each health center were trained to be medical record abstractors by a clinical consultant. Abstractors then completed two sample chart abstracts that were compared to the consultant’s "gold standard." Abstractors who scored at least 90 percent correct were certified to abstract charts at their centers; others were retrained as needed. Data abstracted from the medical records for the reporting period included sociodemographic information (such as age, sex, race, insurance status, and ZIP code), comorbid medical or psychiatric illnesses, and disease-specific quality indicators in the areas of preventative care and screening, disease treatment, and outcomes of care for each condition. Organizational survey. We asked the executive director or a designee from each participating CHC to complete a survey describing the CHC’s organizational structure. The survey asked about (1) governance and board of directors, (2) availability and type of physician and nursing staff, (3) financial information (such as sources of revenue and current financial situation), (4) patient care services, (5) availability of information systems, (6) patients’ demographic characteristics, and (7) ongoing quality improvement activities. Completed surveys were received from all sixty-four health centers.
Quality indicators.
We chose the quality-of-care indicators (Exhibit 1
Data analyses. For each patient, we created an overall quality score for the care received, by averaging the number of applicable indicators met for that patient. After averaging, we rescaled the scores to have the same mean and variance as the overall proportion of indicators met in the sample. Scores were also averaged across patients within a CHC to compute measures of quality at the center level. Higher scores represent higher quality of care.
We examined the overall association of patients’ demographic and clinical characteristics with quality of care and for each condition and estimated a series of models to examine the independent effect of each organizational characteristic on the percentage of applicable indicators met across participants after adjusting for patient characteristics, including presence of comorbid medical or psychiatric illnesses and the clustering of patients within centers. The final model retained each condition of interest and any patient or CHC characteristic that was predictive of quality at a significance level of p
CHC and patient characteristics. Because of our sampling methods, more study CHCs were located in the Midwest, and fewer were located in the Southeast, compared with the national sample (Exhibit 2  ). There were otherwise no differences between the study CHCs and CHCs nationally in location (urban versus rural), number of sites, number of users, or users’ demographic characteristics.
Overall, of a total of 5,690 patients in our sample, 2,002 had diabetes, 1,707 had hypertension, and 1,981 had asthma. In addition, 2,234 (39 percent) were white, 3,508 (62 percent) were women, and 1,105 (19 percent) were uninsured. Asthma patients were younger (mean age twenty-eight versus fifty-five for diabetes and fifty-six for hypertension) and less likely to be uninsured (12 percent versus 23 percent for hypertension and 24 percent for diabetes) than were patients with the other conditions (data not shown).
Comparisons of quality of care.
Fewer than half of eligible patients received appropriate care for fifteen of the twenty-two indicators examined (Exhibit 3
In bivariate analyses, quality differed significantly by patients’ race/ethnicity, insurance status, education, and income. White patients received recommended care for diabetes and asthma care more frequently than black and Hispanic patients did. Women received recommended care for asthma more frequently than men did, but sex was not related to care for the other conditions. The uninsured received recommended diabetes and asthma care less frequently than other insurance groups did (Exhibit 4  ).
Adjusted comparisons of quality.
In hierarchical models that adjusted for condition, patients’ demographic and clinical characteristics, patients’ comorbidities, and CHCs’ organizational characteristics, female patients received recommended care more frequently than male patients (p = .002), and uninsured patients continued to receive recommended care less frequently than patients in other insurance groups (p < .001) (Exhibit 5
CHCs that were newer centers (established less than thirty years ago) offered recommended care more frequently than did older CHCs (mean, 47.5 percent of applicable indicators met versus 43.0 percent, respectively), and CHCs with available computerized decision support continued to offer recommended care more frequently than did those without (mean 47.4 percent of applicable indicators met versus 43.1 percent) (both p = .07); however, these differences were of borderline statistical significance.
This study broadens the current understanding of the quality of care delivered in publicly funded CHCs to a broad cross-section of patients receiving care for chronic diseases. Our results identify the patient subgroups and CHC characteristics with the most room for quality improvement and could act as a guide in evaluating the effectiveness of the current HRSA health disparities collaboratives and the Federal Health Center Growth Initiative. Quality of CHC care. For the majority of indicators, we found that the quality of care delivered in CHCs is comparable to that delivered in other settings that provide care for underserved populations and to some national benchmark data from other sources. For example, Elizabeth McGlynn and colleagues, in a nationally representative sample of Community Tracking Survey (CTS) participants, observed that only 24 percent had three or more HbA1c levels checked within two years.9 CHCs’ rates of blood pressure control were better than the 37–45 percent documented in settings such as hospital-affiliated clinics or the Veterans Affairs (VA) health system or in commercial managed care populations.10 Prior examination of quality of asthma care in Medicaid populations demonstrated that influenza vaccination rates and anti-inflammatory medication use were comparable to or lower than rates we observed among CHCs, and recent examination of the National Ambulatory Care and National Hospital Ambulatory Care surveys documented rates of appropriate inhaled steroid use significantly lower than that observed in our sample.11 Our findings are consistent with studies reporting that CHCs provide better-quality care than other health care segments as measured by reduced hospitalizations and ED visits, higher rates of vaccination among children and the elderly, and higher rates of cancer screening among the poor and elderly.12 Although the quality of diabetes care delivered in CHCs is comparable to that in some national data sets, it is significantly lower than that documented in commercial managed care organizations and in the VA health care system. Eve Kerr and colleagues noted 20 percent higher rates of foot exams and 21 percent higher rates of glycosylated hemoglobin (HbA1c) control among diabetes patients receiving care in a nationally representative sample of managed care organizations.13 Even better performance was noted in all diabetes quality-of-care measures among VA patients.14 Similarly, national HEDIS data during the same period demonstrated higher performance rates for dilated eye exams and lipid control compared to our sample.15 We found that disparities by race and ethnicity in quality of care for all three chronic conditions were eliminated after adjustment for insurance status and that limitations exist in CHCs’ capacity to provide equitable quality of care for the uninsured. These findings are supported by recent literature examining determinants of quality of chronic disease care across all health care segments nationally. Steven Asch and colleagues reported that while there were minimal differences in quality by race and ethnicity nationally, the uninsured received 5 percent and 8 percent fewer recommended disease treatment indicators than managed care and Medicare recipients, respectively.16 Improving quality and reducing disparities. The goals of the current HRSA health disparities collaboratives are to develop a number of programs aimed at improving quality and reducing disparities in CHCs through the use of quality improvement collaboratives. These approaches would allow organizations to share information about quality improvement techniques and provide technical assistance and information systems support to health center teams. Our findings identify a few potential targets that might prove sensitive to these interventions. For example, we found that CHCs with computerized decision support tended to provide better care than those without. Further research should be conducted to determine the extent to which the use of computerized decision support improves centers’ quality of care. The variability in the availability and use of computerized decision support among CHCs could be one reason for their poor performance relative to VA hospitals.17 Our findings also suggest that the disparities collaboratives might have a positive and disproportionate effect on the uninsured, because uninsured patients are the subgroup with the most room for improvement. CHCs are providing care for rapidly increasing numbers of this population, largely as a result of the effects of the Federal Health Center Growth Initiative, coupled with declines in grant funding and payments from Medicaid and other sources to CHCs nationally.18 Study limitations. Despite the strengths of our study design, it had several limitations. First, we did not use a pure random sample of CHCs. Instead, we sampled an equal number of CHCs from each census region, which resulted in a slight underrepresentation of CHCs from the Southeast. However, our study CHCs were similar to the remainder of CHCs nationwide in all other measured characteristics, which suggests that our study sample is representative of CHCs nationally.19 Furthermore, the fact that many of the study centers subsequently participated in a quality improvement collaborative does not suggest that our sample is biased toward higher-performing CHCs. To date, the vast majority of CHCs (approximately 70 percent) have participated in one of the HRSA collaboratives. Second, our reliance on medical charts to measure quality of care and limited documentation of some processes of care (for example, counseling) might have led to an underestimation of the actual level of care provided.20 In addition, there might have been unmeasured differences in the completeness of providers’ documentation related to CHCs’ resources or the presence of EMRs that were not accounted for. However, these limitations are common to all studies that rely on medical record abstractions and thus make our data more comparable to previous reports in the literature. Last, we report data obtained prior to national initiatives aimed at increasing CHC availability and improving the quality of chronic disease care. However, our findings provide new insight into baseline CHC performance, and this undertaking is an essential first step in evaluating the effects of these large-scale policy changes. Furthermore, our findings provide a clearer understanding of CHCs’ performance relative to other settings, and the time period of our data is similar that reported in the most recent literature about quality of care across settings nationally.21 OVER ALL, OUR FINDINGS SUGGEST that if policymakers plan to extend coverage for underserved populations through expanding the number of CHC sites, they might need to provide additional resources to meet the needs of uninsured patients at existing sites. Additional research is needed to examine the barriers that CHCs have in providing better-quality care for uninsured patients.
LeRoi Hicks (hicks{at}hcp.med.harvard.edu) is an instructor in health care policy at Harvard Medical School in Boston, Massachusetts. James O’Malley is an assistant professor of statistics in the Department of Health Care Policy at Harvard Medical School. Tracy Lieu is a professor at the Center for Child Health Care Studies, Department of Ambulatory Care and Prevention, at Harvard Pilgrim Health Care and Harvard Medical School. Thomas Keegan is project manager in the Department of Health Care Policy, Harvard Medical School. Nakela Cook is a fellow in cardiology at Massachusetts General Hospital and Harvard Medical School. Barbara McNeil is a professor and chair, Department of Health Care Policy, at Harvard Medical School. Bruce Landon is an associate professor in that department. Edward Guadagnoli is a professor of health care policy at Harvard Medical School. This project was supported by Grant no. 1 U01 HS13653 from the Agency for Healthcare Research and Quality and Grant no. 20030185 from the Commonwealth Fund. The authors thank Yang Xu for statistical programming, Rebecca Gregory for assistance with project management, Mary Ly for research assistance, and Laura Peterson for assistance with developing chart-abstraction instruments and training of abstractors.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Study Data And Methods
Study Results
.10. 
