International Journal for Quality in Health Care Advance Access originally published online on March 9, 2006
International Journal for Quality in Health Care 2006 18(3):186-194; doi:10.1093/intqhc/mzi105
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Improving the quality of preventive cardiovascular care provided by primary care physicians: insights from a US Quality Improvement Organization
1 Qualidigm, Middletown, 2 Department of Medicine, Yale University School of Medicine, New Haven, 3 Department of Community Medicine, University of Connecticut Medical School, Farmington, CT, and 4 American Board of Internal Medicine, Philadelphia, PA, USA
Background. During 2000–03, Qualidigm, a US Quality Improvement Organization, conducted a project to improve the care received by elderly Medicare patients with coronary artery disease or cardiovascular risk factors.
Methods. We recruited primary care physicians in private practice in the state of Connecticut. Then, we identified approximately 30–50 patients per physician from the periods 1 January 2000 to 31 December 2000 and 1 November 2001 to 31 October 2002. We abstracted medical records to assess processes and outcomes of care, and we provided the physicians with performance data and a variety of practice-enhancing materials. The physicians utilized those materials that they perceived to be most helpful.
Results. We identified and recruited 974 primary care physicians to participate. Of these, 103 (10.6%) committed to participate, and 85 of the 103 completed the project. Among the intervention tools, physicians and their office personnel utilized personal digital assistants (PDAs) (36.5%) and patient education materials (34.1%) most commonly. Overall, quality of care improved for most physicians (mean quality score 62.0 to 67.8%, P < 0.001). However, not all improved, and most improvements were modest [mean absolute improvement in quality score 5.8%, standard deviation (SD) 6.8%].
Conclusions. Quality Improvement Organizations and others interested in improving outpatient quality of care face significant challenges in recruiting self-employed primary care physicians to quality improvement projects and in bringing about transformational change. Future primary care quality improvement projects should include careful assessments of practice-specific barriers, interventions that are linked to these barriers, and support of the practices on implementation.
Keywords: quality improvement, outpatient, primary care
Address reprint requests to Thomas P. Meehan, Qualidigm, 100 Roscommon Drive, Middletown, CT 06457, USA. E-mail: tmeehan{at}ctqio.sdps.org
Accepted for publication February 1, 2006.
The quality of outpatient healthcare provided to adults in the United States is variable and not consistent with recommended standards [1–4]. This is true for chronic, acute, and preventive care, women and men, elderly and non-elderly patients, and managed care and fee-for-service delivery systems. Because self-employed physicians in solo or small group practice arrangements deliver a majority of adult primary care [5], it is important that the care of these providers be assessed and improved.
Quality Improvement Organizations are contracted by the Centers for Medicare & Medicaid Services to engage local health care providers in collaborations to improve the quality of care provided to Medicare beneficiaries in all settings [6]. Thus, care delivered in the outpatient primary care setting is a target for such quality improvement initiatives. However, primary care physicians are under no contractual obligation to participate in Medicare-sponsored quality improvement projects, and Quality Improvement Organizations are challenged to engage these physicians. This is particularly true in solo or small group practices [7]. This article summarizes the results of an outpatient quality improvement project conducted in 2000–03 by Qualidigm, the Connecticut Quality Improvement Organization, and 85 primary care physicians in private practice. The article also presents practical insights that we gleaned in conducting the project and implications for future quality improvement in this setting.
 |
Methods |
|---|
 Top Methods ResultsDiscussionReferences |
|---|
Study design
We conducted a pre–post assessment of quality of preventive cardiovascular care provided to elderly Medicare beneficiaries by specific primary care physicians. We targeted physicians and their office staffs with a multi-faceted set of interventions to improve the care of patients with diagnosed coronary artery disease or with one or more cardiovascular risk factors. However, the physicians and their staffs determined which quality improvement (QI) interventions they would utilize, and to what degree. We abstracted process of care performance and intermediate outcomes from the medical records of independent samples of patients from each physician’s practice during one-year baseline and follow-up periods.
Primary care physicians and patients
We identified 974 primary care physicians from Medicare Part B claims of general internists, family practitioners, and geriatricians who had provided care for at least 50 fee-for-service Medicare patients in 1999. Next, we mailed invitation letters and placed follow-up telephone calls to the physicians; 103 (10.6%) agreed to participate. For these, we identified approximately 40 patients each. All patients were 
65 years old and had existing coronary artery disease or one or more of the following risk factors: hypertension, diabetes, or dyslipidemia. We selected independent stratified random samples of patients from Medicare Part B claims submitted by the physicians. To assure that the physicians had provided longitudinal care, we selected claims from the calendar year (1999) before the baseline observation period (1 January 2000–31 December 2000) and from the calendar year (2001) before the follow-up observation period (1 November 2001–31 October 2002). We selected approximately 10 patients per target condition for each physician based on Part A or Part B Medicare claims-based diagnoses [International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM)] [8]: diabetes (250.00–250.93), dyslipidemia (272.0–272.4), hypertension (401.0–404.93), and coronary artery disease (411.0–411.89, 412, 413.0–413.9, 414.00–414.9). When medical records of selected patients were unavailable, Qualidigm staff identified additional patients from claims data or the physician office staffs identified convenience samples of patients with the target conditions.
Medical record abstraction
Qualidigm personnel provided lists of patients identified from claims data to each physician’s office to confirm that the patients were appropriately assigned to physicians. Patients who died, moved away, or changed primary care physicians during an observation year were replaced by other patients identified from claims data and confirmed by office personnel. Physician office personnel pulled medical records for on-site abstraction. Trained medical record abstractors from Qualidigm collected the data using a standardized instrument. We tested reliability of data abstraction by randomly selecting a subset of cases (n = 272 during the baseline and n = 233 during the follow-up observation periods) for independent data collection by a second abstractor. We calculated reliability rates and kappa statistics [9] for key variables and composite kappa statistics [10] for all performance measures. We converted continuous variables to categorical variables for these calculations. All variable-specific reliability rates were high (>90%), and kappa statistics indicated good to excellent inter-abstractor reliability (0.60–1.00). Composite kappa statistics also indicated good to excellent reliability, with two exceptions during the follow-up period. Reliability for the HgbA1C < 7.0 mg/dl and Smoking Cessation Counseling performance measures were in the fair range (0.40–0.60) during this time frame.
We abstracted demographic characteristics, chronic illnesses, tobacco use, physical examination findings, laboratory values, processes of care, and number of visits from each observation period. Demographic characteristics included age, gender, and race. Chronic illnesses included hypertension, dyslipidemia, diabetes mellitus, coronary artery disease, cerebrovascular disease, and peripheral vascular disease. We assessed current tobacco use and clinician counseling to quit. We captured blood pressure values from the most recent office visit during each observation period. Laboratory values included the most recent values for hemoglobin A1C and low-density lipoprotein (LDL) cholesterol. We recorded all medications in use at the last visit of each observation period as well as any documented allergies, side effects, contra-indications, or patient refusals relating to antithrombotic or anticoagulant therapies.
Interventions
We provided a multi-faceted set of quality improvement interventions to all participating physicians at introductory group dinner meetings (83.5%) or via the mail (16.5%). We provided the physicians the choice of four dates to attend a two-hour dinner meeting held at local restaurants. At the meetings, we provided an overview of the project and the intervention materials and distributed packets of the intervention materials. The packets included personal digital assistants (PDAs) and a variety of paper tools, including (i) patient education materials developed by the American Cancer Society, the American Diabetes Association, the American Heart Association, and the American Lung Association, (ii) patient reminders for recommended services, (iii) clinician education materials, (iv) clinician reminders for recommended services, e.g. chart stickers, (v) clinician documentation materials, e.g. flow sheets, (vi) patient documentation materials, and (vii) lists of community resources. We provided no specific training on how to utilize the paper tools as they were all familiar to the physicians. Subsequently, we invited the physicians to attend a two-hour evening training session on medical software for the PDAs. The content of the software programs covered medication prescribing and local Managed Care Organization formularies, recommended antibiotic treatment regimens, ICD-9-CM coding, a medical text reference, and a patient registry. We also mailed confidential physician performance reports for the baseline period to all participating physicians in October 2001, and we invited the physicians to attend another two-hour dinner meeting to discuss aggregate results. Finally, we mailed a follow-up performance report to the physicians at the completion of the project. We tracked provision of improvement interventions in an electronic database throughout the project.
Analysis
We obtained the following physician characteristics from the 2000 Folio’s Directory of Connecticut Physicians, 1999 Part B Medicare claims data, and 2000 census data: United States versus foreign medical school training, years since medical school graduation, family practice versus internal medicine specialty, solo versus group practice, number of Medicare patients, demographic characteristics of the physicians’ patients (age, race, gender), and community-level characteristics of patients (median family income, education level). We assessed and compared the characteristics of participating and non-participating physicians with descriptive statistics using chi-square tests for binary variables, t-tests for normally distributed continuous variables, and non-parametric tests for variables that were not normally distributed. Next, we constructed a multi-variable logistic regression model to examine factors associated with participation in the project.
We assessed quality improvement interventions, patient demographic and clinical characteristics, and processes and outcomes of care with descriptive statistics, excluding patients with missing values from calculations for the relevant variables. We used clinical data abstracted from the medical records to confirm the claims-based diagnoses before processes of care were assessed. Confirmation criteria for diabetes and hypertension included documentation of (i) a clinician’s diagnosis in a problem list or progress note, and, (ii) a diagnostic laboratory value (fasting blood sugar >126 mg/dl or non-fasting >200 mg/dl) [11], physical finding (systolic blood pressure 140 mm of Hg or diastolic blood pressure 90 mm of Hg) [12], or medication used to treat the condition. Confirmation criteria for coronary artery disease included documentation of a clinician’s diagnosis in a problem list or progress note. We assessed systolic and diastolic blood pressure, hemoglobin A1C, and LDL cholesterol levels categorically and as means with 95% confidence intervals (95% CIs). We assessed use of antithrombotic or anticoagulant therapy in patients who did not refuse or have contra-indications to the therapy and who were at moderate or high risk according to the American Heart Association’s guidelines for primary and secondary prevention of coronary artery disease [13,14] and secondary prevention of stroke [15]. We considered the following as contra-indications to antithrombotic or anticoagulant therapy: asthma, peptic ulcer disease, chronic liver disease, chronic renal disease, or creatinine >3 mg/dl, positive fecal occult blood test, platelet count <100 x 109/L, anemia, or (hemoglobin <10 g/dl or hematocrit <30%) any clinician notation that the patient had increased risk of bleeding, age >80 years, dementia, allergy, or previous adverse reaction, terminal illness, or patient refusal. Subgroup analyses were conducted for blood pressure control and LDL cholesterol levels among patients with hypertension or hyperlipidemia and indications for tighter control versus patients without such indications.
We constructed hierarchical random effects logistic regression models to assess relationships between patient characteristics, physician volume of patients with target conditions, between-physician variance and changes in dichotomous outcomes over time. Similarly, we constructed hierarchical random effects linear regression models to adjust for the same factors when assessing changes in continuous performance measures over time. We assessed unadjusted associations between each independent variable and change in performance measure first. Then, we created the multi-variable models. We determined odds ratios (logistic regression models) or reduction ratios (linear regression models) and 95% CIs for changes in performance measures over time. We included the following patient characteristics in the models: age, gender, race, number of physician visits, and number of risk factors. For each multivariable model, we evaluated assumptions with partial residual plots [16], fit with the chi-square goodness-of-fit statistic [17], and model discrimination with the receiver-operating characteristic curve [18].
Because denominators for individual physician performance measures were necessarily small, we calculated summary quality of care scores across all performance measures for each physician during each observation period. We constructed denominators for these scores by summing the number of each physician’s patients that were eligible for each performance measure. We constructed numerators by summing the number of the physician’s patients who received the targeted processes of care or achieved the targeted outcomes. We graphed distributions of baseline and follow-up quality scores and pre–post differences using the normal probability plots’ method [19]. We performed all calculations using software for SAS, version 8.12 (SAS Institute, Inc., Cary, NC), STATA, version 4.0 (STATA Corp, College Station, Texas), and HLM, version 5.0 (Scientific Software International, Inc., Lincolnwood, IL).
|
Results |
|---|
|
TopMethodsResultsDiscussionReferences |
|---|
Primary care physicians
Participating and non-participating physicians were similar in all measured characteristics except one, i.e. physicians who graduated from medical school before 1980 were less likely to agree to participate in the project than physicians who graduated at a later date [odds ratio (OR) = 0.96, 95% CI = 0.94–0.99]. Of the 103 physicians who agreed to participate, 85 (82.5%) completed the project. Sixty-five (76.5%) of these were internists, and 20 (23.5%) were family practitioners. Sixty-eight (85.0%) were board certified. Fifty-four (63.5%) graduated from a US medical school, and 31 (36.5%) were non-US medical school graduates. Thirty-eight (44.7%) graduated before 1980. Thirty-five (41.2%) were solo practitioners; 25 (29.4%) had one partner; and 25 (29.4%) had
2 partners. There were no significant differences in physician-training characteristics, size of physician group, size of Medicare panel, physician demographic characteristics, or community-level income/educational level of patients between the 85 physicians who completed the study and the 18 who did not.
Use of quality improvement interventions
All 85 primary care physicians received a PDA as an incentive to participate in the project and as a quality improvement intervention (Table 1). Thirteen (15.3%) physicians participated in a subsequent group training session on the five medical software programs for the PDA, and 31 reported using the PDA on a regular basis throughout the project. In addition, all 85 primary care physicians were mailed the confidential written report summarizing their performance during the baseline observation period and comparing it with aggregate performance and the range of individual physicians’ performance. All of the physicians were invited to attend a group session to discuss the baseline findings; twenty (23.5%) attended the session. Seventy-one (83.5%) physicians received the entire package of additional QI interventions. Over the duration of the project, physicians and their staffs re-ordered patient education materials (34.1%), clinician education materials (18.8%), and clinician reminders (15.3%) most commonly. Fewer than 10% re-ordered patient reminders, patient documentation materials, or community resource lists.
|
Patients
There were 3322 patients who we identified as receiving care from the participating physicians for coronary artery disease or one or more of the targeted risk factors and whose charts we abstracted for the baseline period; there were 3788 patients for the follow-up period (Table 2). The median number of patients whose care we assessed was 44 (range 11–55) per physician during the baseline period and 46 (range 19–53) during the follow-up period.
|
Patient demographic and clinical characteristics
In general, patients were elderly, had multiple vascular disease risk factors, and saw their primary care physicians frequently in both time periods. In aggregate, there were no statistically significant differences in age, race, or gender between the baseline and follow-up cohorts, although patients had slightly more risk factors (P < 0.001) and existing vascular disease in the baseline period (55.6 versus 49.0%, P < 0.001).
Processes and intermediate outcomes of care
Aggregate performance rates for most measures improved from baseline to follow-up (Table 3). Overall performance ranged from 43.6% for blood pressure control <140/90 mm Hg to 86.8% for biennial lipid testing during the baseline period, and from 50.3% for blood pressure control <140/90 mm Hg to 89.5% for biennial lipid testing during the follow-up period. Patients with indications for more aggressive treatment, i.e. hypertension and diabetes or heart failure, or coronary artery disease and hyperlipidemia, had higher rates of control in both time frames: high-risk hypertension 45.9 to 52.9% and existing coronary artery disease 82.3 to 86.0%. In aggregate, five of eight performance measures demonstrated statistically significant improvement from baseline to follow-up. When examined as continuous measures, aggregate mean values decreased for systolic blood pressure (141 to 137 mm of Hg, P < 0.001), diastolic blood pressure (77 to 75 mm of Hg, P < 0.001), and LDL cholesterol levels (117 to 109 mg/dl, P < 0.001), but not hemoglobin A1C levels (7.3 to 7.2 mg/dl, P = 0.31). The same patterns of improvement were seen after adjustment for patient characteristics and physician volume. As expected, individual physician performance rates varied widely, in part because of the small denominators at the level of individual physicians.
|
Summary physician quality of care scores
Summary quality of care scores for individual physicians were normally distributed in both observation periods. Overall scores increased from baseline to follow-up (baseline : median 60.6%, range 38.5–81.6%, 25th–75th percentile 56.3–69.0%; follow-up: median 67.7%, range 48.5–82.3%, 25th–75th percentile 62.3–74.0%) (Figure 1a). Changes in individual physician summary quality of care scores were also normally distributed and variable (Figure 1b). Baseline to follow-up changes ranged from –14 to +22% among all physicians (Figure 2a). However, less overall improvement occurred in the physicians whose baseline performance was high (Figure 2b).
|
|
|
Discussion |
|---|
|
TopMethodsResultsDiscussionReferences |
|---|
The project is an example of a voluntary quality improvement collaboration between a Medicare-sponsored Quality Improvement Organization and primary care physicians in private practice. The project demonstrated aggregate improvements in evidence-based processes of care and intermediate outcomes relating to patients with coronary artery disease and cardiovascular risk factors. Although the improvements were generally modest and did not occur in every practice, they were associated with simple interventions on the part of the Quality Improvement Organization, i.e. provision of performance data, PDAs, and paper-based tools to the physicians and their staffs. The project also highlighted important challenges to Quality Improvement Organizations and other external groups interested in improving quality of care in physician offices, i.e. the difficulty of recruiting self-employed physicians to participate, and the need for more effective intervention approaches.
We noted statistically and clinically significant improvements in aggregate blood pressure control, lipid testing and control, diabetes control, and smoking assessments. These improvements in care persisted after adjustment for differences among patients and physicians. Although the absolute changes in these process and outcome measures were modest, their expected impact on long-term clinical outcomes is important. Extrapolating from the Framingham Heart Study, improvements in these measures would be expected to translate to long-term decreases in coronary artery disease incidence and mortality among the target population [20]. In addition, mortality and adverse events from cerebrovascular and peripheral vascular disease would also be expected to decrease based on improvements in the targeted process and outcome measures [21,22].
What practical insights did we take away from this project that can inform future outpatient QI projects? The first is that recruiting primary care physicians to quality improvement projects is difficult. Despite free PDAs, performance data reports, and a variety of practice-enhancing tools, only 10.5% of the state’s primary care physicians were willing to participate in this project, and 17.5% of these subsequently dropped out. This finding is consistent with other work showing that strong incentives are necessary to engage physicians in quality improvement [23] and that physicians from small practices are least likely to participate [7]. At the national level, pay-for-performance [24], public recognition [25,26], and revised re-certification criteria requiring demonstrated QI activity [27] are increasingly utilized and promising incentive strategies but their full effectiveness is as yet unknown.
A second lesson relates to choice and implementation of quality improvement interventions. We provided the practices with a variety of interventions that have been shown to be at least modestly effective when used individually and more effective when used in combination [28]. Further, we allowed physicians and their staffs to choose from among these, i.e. to customize their strategies for improvement based on knowledge of their own practice. This seemed to be a reasonable approach to maximize acceptance and utilization of the interventions by the practices. However, we did not have the resources to do practice-by-practice needs and barrier assessments, or staff training on use of the intervention tools. Also, we did not systematically assess how the physicians and their staffs used the materials or what barriers they encountered. Although our data indicate that the interventions may have been under-utilized, we do not have a deep understanding of what actually occurred in the physician offices. The results of this project indicate that providing primary care physicians with a variety of ‘easy to use’ QI tools is not likely to lead to dramatic improvement. In a companion qualitative study done with a subset of these physicians, we learned that the physicians were largely unfamiliar with the science of quality improvement and the concept of systems-based thinking [29]. Thus, they may not have chosen the best interventions for their practices or known how to optimally implement them. Future quality improvement work with primary care providers should include practice-specific assessments of goals and barriers, training and support of practice personnel on implementation of interventions, and real-time collection of formative data to improve the intervention tools and implementation approaches.
Some limitations of our QI project should be mentioned. We had a relatively small number of self-selected physician participants, and there was some self-selection of patient records by the physicians. These issues undoubtedly introduced selection bias. However, the impact of these issues on the performance measures could have been positive or negative. A second issue is that our pre–post study design without a comparison group did not allow us to tease out the impact of secular trends. In addition, our data source, i.e. medical records, could have yielded incomplete or inaccurate data if documentation was imperfect. However, our data abstraction methods were rigorous and the reliability of our abstractors was high. A final limitation relates to the interventions, i.e. utilization of the materials provided to the practices by the Quality Improvement Organization appeared to be low, and no details relating to in-office use of the materials are known. Thus, the interventions were not as strong as they might have been had there been practice-specific needs and barrier assessments, and training and support on use of the tools.
Our study has demonstrated that a Medicare-sponsored Quality Improvement Organization can successfully collaborate with primary care physicians in private practice to improve the quality of preventive cardiovascular care. However, Quality Improvement Organizations and others interested in improving outpatient quality of care face significant challenges in recruiting self-employed primary care physicians to quality improvement projects and in bringing about dramatic change quickly. Future primary care quality improvement projects should include careful assessments of practice-specific barriers, interventions that are linked to these barriers, and support of the practices on implementation.
|
References |
|---|
|
TopMethodsResultsDiscussionReferences |
|---|
- McGlynn EA, Asch SM, Adams J et al. The quality of health care delivered to adults in the United States. N Engl J Med 2003; 348: 2635–2645.
[Abstract/Free Full Text] - Ma J, Stafford S. Quality of US outpatient care. Temporal changes and racial/ethnic disparities. Arch Intern Med 2005; 165: 1354–1361.
[Abstract/Free Full Text] - Pham HH, Schrag D, Hargraves JL, Bach PB. Delivery of preventive services to older adults by primary care physicians. JAMA 2005; 294: 473–481.
[Abstract/Free Full Text] - Ornstein SM, Jenkins RG. Quality of care for chronic illness in primary care: opportunity for improvement in process and outcome measures. Am J Manag Care 1999; 5: 621–627.[Web of Science][Medline]
- Institute for the Future. Health and Health Care 2010: the Forecast, the Challenge, second edition. San Francisco, CA: Jossey-Bass, 2003.
- Centers for Medicare & Medicaid Services. Quality Improvement Organizations (QIOs): www.cms.hhs.gov/qio/ Accessed 19 August 2004.
- Audet AJ, Doty MM, Shamasdin J, Schoenbaum SC. Measure, learn, and improve: physicians’ involvement in quality improvement. Health Aff (Millwood) 2005; 24 (3): 843–853.
[Abstract/Free Full Text] - National Center for Health Statistics. International Classification of Disease, Ninth Revision, Clinical Modification. Washington, DC: US Department of Health and Human Services, US Public Health Services; 1989.
- Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas 1960; 20: 37–46.[CrossRef][Web of Science]
- Huff ED. Comprehensive reliability assessment and comparison of quality indicators and their components. J Clin Epidemiol 1997; 50: 1395–1404.[CrossRef][Web of Science][Medline]
- American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2005; 28S: S37–S42.
[Free Full Text] - Joint National Committee on Prevention, Evaluation and Treatment of High Blood Pressure. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. JAMA 2003; 289: 2560–2572.
[Abstract/Free Full Text] - American Heart Association. Consensus Panel Statement. Guide to primary prevention of cardiovascular disease. Circulation 1997; 95: 2330.
- American Heart Association. Consensus Panel Statement. Comprehensive risk reduction for patients with coronary and other vascular disease. Circulation 1995; 92: 2–4.
- American Heart Association. Consensus Panel Statement. Preventing ischemic stroke in patients with prior stroke and transient ischemic attach. Stroke 1999; 30: 1991–1994.
[Free Full Text] - Landwehr J, Pregibon D, Shoemaker A. Graphical methods for assessing logistic regression models. J Am Stat Assoc 1984; 79: 61–63.
- Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley and Sons, Inc. 1989.
- Hanley JA, McNeil BJ. The meaning and use of the area under the receiver operating characteristic (ROC) curve. Radiology 1992; 143: 29–36.
- Wilk MB, Gnandesikan R. Probability plotting methods for the analysis of data. Biometrika 1968; 55: 1–17.
[Abstract/Free Full Text] - Grundy SM, Pasternak R, Greenland P, Smith S, Fuster V. Assessment of cardiovascular risk by use of multiple-risk-factor assessment equations: a statement for health care professionals from the American Heart Association and the American College of Cardiology. Circulation 1999; 100: 1481–1492.
[Free Full Text] - Straus SE, Majumdar SR, McAlister FA. New evidence for stroke prevention. Scientific review. JAMA 2002; 288: 1388–1395.
[Abstract/Free Full Text] - Kannel WB. Risk factors for atherosclerotic outcomes in different arterial territories. J Cardiovasc Risk 1994; 1: 333–339.[Medline]
- Casalino L, Gilles RR, Shortell SM et al. External incentives, information technology, and organized processes to improve health care quality for patients with chronic illnesses. JAMA 2003; 289: 434–441.
[Abstract/Free Full Text] - Epstein AM, Lee TH, Hamel MB. Paying physicians for high-quality care. N Engl J Med 2004; 350: 406–410.
[Free Full Text] - National Committee for Quality Assurance. Diabetes Physician Recognition Program (DPRP). Available at http://www.ncqa.org/sohc2002/sohc_2002_DPRP_TEXT.html. Accessed 7 January 2005.
- National Committee for Quality Assurance. AHA-ASA-NCQA Heart Stroke Recognition Program. Available at http://www.ncqa.org/hsrp/HSRPBrochure.pdf. Accessed 7 January 2005.
- Brennan TA, Horwitz RI, Duffy FD, Cassel CK, Goode LD, Lipner RS. The role of physician specialty board certification status in the quality movement. JAMA 2004; 292: 1038–1043.
[Abstract/Free Full Text] - Grimshaw JM, Shirran L, Thomas R et al. Changing provider behavior. An overview of systematic reviews of interventions. Med Care 2001; 39 (8) (suppl. 2): II2–II45.[Web of Science][Medline]
- Holmboe E, Kim N, Cohen S et al. Primary care physicians, office-based practice, and the meaning of quality improvement. Am J Med 2005; 118 (8): 917–922.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||