- Email: [email protected]
Preventive care practices for diabetes management in two primary care samples
Preventive Care Practices for Diabetes Management in Two Primary Care Samples Russell E. Glasgow, PhD, Lisa A. Strycker, MA Purpose:
To assess the level of physician performance on American Diabetes Association Provider Recognition Program (PRP) measures in two samples of primary care patients, as well as to identify patient, physician, and office characteristics related to performance levels.
Methods:
In the two studies, we surveyed 435 Type 2 diabetes patients, cared for by 47 different physicians, on their receipt of PRP preventive care activities.
Results:
Overall, patients in the two samples reported receiving 74% and 64% of recommended services. In both samples, performance of microvascular/glycemic control activities and cardiovascular lab checks (84% and 74%) was significantly higher than behavioral self-management/patient-focused activities (61% and 48%) (p⬍0.001). From a set of patient, physician, and practice setting characteristics, only the use of community resources for chronic illness management support was associated with service performance.
Conclusions: We found considerable variability in the levels of performance in providing PRP-recommended activities. Greater attention should be focused on self-management and patient-focused activities, given that these are delivered less frequently than medical/laboratory checks. Medical Subject Headings (MeSH): diabetes mellitus, preventive medicine, primary health care, quality of health care, task performance and analysis (Am J Prev Med 2000;19(1): 9 –14) © 2000 American Journal of Preventive Medicine
Introduction
T
he emergence of clinical practice guidelines, disease management programs, evidence-based clinical pathways, “best practices,” and disease-specific performance measures is prominent among the many changes in health care during the past decade. Spurred by the confluence of evidence-based medicine1,2 and the desire of managed care and other organizations to reduce nonproductive variations in practice, clinical performance measures and guidelines are applied with increasing frequency.2– 4 Several factors have all contributed to the development of diabetes management recommendations, performance indicators, and guidelines: recent advances in diabetes self-management5–7; the emergence of conclusive data that demonstrate the efficacy of organized, comprehensive management in reducing diabetes complications and mortality8 –10; and recognition of the enormous health care costs of diabetes.11,12 Among these, the Provider Recognition Program (PRP) performance measures of the American Diabetes Association are promiFrom the AMC Cancer Research Center (Glasgow), Denver, Colorado; and Oregon Research Institute (Strycker), Eugene, Oregon Address correspondence and reprint requests to: Russell E. Glasgow, PhD, 11716 98th Place SW, Vashon, WA 98070. E-mail: [email protected].
nent. The related Diabetes Quality Improvement Program measures,13 essentially a subset of the PRP measures, have been adopted recently as Health Employer Data Info Survey criteria. InitialstudiesofthePRPmeasureshavefoundconsiderable variability in performance levels across various measures and providers.14 –16 Specifically, patient-focused and self-management practices appear to occur less often than laboratory/screening activities, and internists may differ from family physicians in their diabetes care practices.15 Much of the brunt of following these new recommendations and best practices falls on the primary care provider, as the great majority of diabetes patients are managed in primary care.2,17 Given that most of the recommended prevention activities must be accomplished or initiated during the ever more time-limited primary care visit, it is understandable that many of these activities are performed at substandard levels.2,10 Diabetes is a complex and challenging chronic illness that requires numerous lifestyle changes, including microvascular and macrovascular disease–prevention activities.6,11,12,18 Also, most diabetes patients have other comorbid chronic illnesses that further complicate their management. Little research exists on patient, provider, or office characteristics associated with either diabetes patient– physician interactions or with the level of recom-
Am J Prev Med 2000;19(1) 0749-3797/00/$–see front matter © 2000 American Journal of Preventive Medicine • Published by Elsevier Science Inc. PII S0749-3797(00)00157-4
9
Table 1. Characteristics of patients, providers, and office practices Mean (SD) or percent (for dichotomous variables)
Characteristic Patient characteristics Age % Female % on insulin % ⬎ High school education % ⬎ $30,000 income % Caucasian % Received diabetes education % Smoke CIRS total support score (scale⫽1–5) Years diagnosed # Other chronic illnesses (Range⫽0–12 in Study 1; 0–9 in Study 2)
Study 1 (n⫽275) 60 (9) 56% 16% 46% Not collected 92% 57% 13% 2.6 (.5) 6.4 (6.4) 2.7 (2.1)
Study 2 (n⫽160) 59 (9) 53% 36% 53% 52% 91% Not collected 13% 3.2 (.9) 11.3 (7.8) 2.7 (2.3)
Provider characteristics Years since degree Medical specialty % Family practice/GP % Endocrinologist/internal medicine Othera % Female % Caseload composed of diabetes patients ⬍5% 5%–10% 11%–25% 26%–50% 51%–75% ⬎76% Total # diabetes patients
Study 1 (n⫽33) 17 (9)
Study 2 (n⫽14) 23 (10)
32% 40% 24% 4% 0% 0% 418 (753)
20% 60% 10% 0% 0% 10% 1001 (792)
Office practice characteristics % Have diabetes registry % Use diabetes guidelines/flowsheets % Provide self-management support Barriers to providing self-management support (scale⫽1–4)
Study 1 (n⫽12) 50% 80% 100% 2.0 (.3)
Study 2 (n⫽5) 80% 80% 80% 2.3 (.7)
55% 33% 12% 45%
14% 86% 0% 21%
a
The “other” category includes physicians’ assistants and family nurse practitioners. SD, standard deviation.
mended preventive practices.15,19,20 In particular, more research has been recommended on contextual factors, such as the patient’s social environment and health care system characteristics,7,10 and on patient and physician factors, such as gender, that may influence performance of preventive activities.21 The purpose of this report is threefold: (1) to report on the overall levels of PRP preventive measures in two different primary care samples, as replication is important, especially in new areas of investigation; (2) to determine whether levels of performance for laboratory/ screening were higher than for patient-focused/selfmanagement activities; and (3) to determine whether we could replicate reported findings concerning patient and physician characteristics associated with performance of these prevention activities.
Methods Recruitment of Practices, Providers, and Patients Within each of the two different health care systems in the Pacific Northwest, we made arrangements with influential
10
physician leaders to facilitate access to primary care providers within their systems. We approached 35 and 16 primary care physicians for Study 1 and Study 2, respectively. Of these, 33 (94%) and 14 (88%) participated. We summarize their characteristics in Table 1. Fifty-five percent and fourteen percent were family physicians, 45% and 21% were female, the average number of years of practice since receiving their degrees was 17 and 23, and 80% in both studies reported implementing some form of diabetes care guidelines in their practices. Participating physicians came from 17 different practices and almost all see a mixture of patients receiving managed care, fee-for-service, and government-provided health insurance. In the practices of the participating physicians, all Type 2 dependent diabetes patients aged 40 to 75 years received letters inviting them to participate in a randomized trial of interactive health communication to assist in diabetes selfmanagement (multimedia touch-screen and Internet-based intervention, in Study 1 and Study 2, respectively). As reported elsewhere22,23 76% and 60% of eligible patients participated in Study 1 and Study 2, respectively, and in both studies the characteristics of participants were similar to nonparticipants.
American Journal of Preventive Medicine, Volume 19, Number 1
Table 1 shows that patients in the two studies averaged age 60 and 59 years and had been diagnosed for an average of 6.4 and 11.3 years, 16% and 36% were taking insulin, 56% and 53% were women, and patients in both samples had an average of 2.7 comorbid chronic illnesses.
Table 2. Correspondence between patient survey and medical chart review on seven PRP items
Measure
Measures A patient version of the 11 performance measures recommended by the American Diabetes Association’s PRP program, used in previous research,15 was adapted for these studies. Each question asked, in lay terms, how long (e.g., 6 months, 1 year, 2 years, 5 years, never) it had been since the patient had received the service in question (e.g., “How long has it been since a health professional checked your feet for any sores or irritation?” and “About how long has it been since you last had your blood pressure taken by a doctor, nurse, or other health care professional?”). At the time this study was conducted, these 11 items assessed all PRP measures. Recently, lab values for HbA1C, lipids, and blood pressure have been added to the set. These data are not available for these samples. Study 1 measures were collected by paper and pencil survey, and Study 2 measures were collected using the Internet. Data for the present study were collected as part of the baseline measures for the two separate intervention studies. In addition to the patient and physician characteristics previously described, measures were also collected from patients by survey and for physicians and offices by surveying medical office managers. As shown in Table 1, other patient characteristics assessed were education, family, income, race, and smoking. Other physician and office characteristics were size of diabetes caseload, whether the practice had a diabetes registry, and whether it offered some form of self-management support. Finally, to provide a measure of the richness of the patients’ social environment regarding support for self-management activities, the Chronic Illness Resources Survey was administered. This measure and its psychometric characteristics are described in detail elsewhere,24 but the measure has been found to be a reliable and valid measure of the extent to which patients have used various levels of support resources throughout their community (e.g., family and friends, work, neighborhoods, the media). For a subsample of 27 patients in Study 2, whose physicians also participated in a separate quality improvement effort, we conducted a validational study using chart reviews to provide corroborative data on the seven preventive practices that we judged likely to be recorded in patient medical records. Some items, such as patient satisfaction, are by definition assessed only by patient survey and others, such as collaborative goal setting, are notoriously inconsistently recorded in patient charts.25,26 We report the results of this substudy in Table 2. As shown, patient reports were generally congruent with the charts reviewed. Patient surveys reported that an average 5.7 of the 7 practices were met, compared with an average of 5.1 recorded in the patients’ charts; we found relatively good correspondence between overall number of criteria met collected by the two methods (Kendall’s W⫽0.45, p⫽ 0.0005). The only real discrepancy between patient reports and medical records was on
HbA1C testing Dilated eye exam Foot exam Lipid testing Blood pressure testing Smoking-cessation counseling Nutrition counselinga Overall number of seven indices met
Patient survey
Chart review
92% 81.5% 93% 88.5% 100% 17%
96% 74% 96% 100% 96% 0%
54%
100%
Kendall’s coefficient of concordance (W) (significance)
5.7 (.67) 5.1 (1.1) .45 (p⫽0.0005)
Based on 27 patients from Study 2 for whom medical records were available. a Measure in chart review was referral to dietitian; item in patient survey was self-reported physician referral to dietitian.
referral for nutritional counseling (54% patient report vs 100% chart review).
Analyses We used descriptive statistics (means, standard deviations, and percentages) to describe the sample and level of performance of the individual performance measures. We used Kendall’s coefficient of concordance to evaluate the association between medical record and patient-reported overall measure of seven preventive actions. We used t tests to compare the numbers of laboratory/screening (0 to 6) and patient- focused/behavioral activities (0 to 5) performed across studies and across different provider characteristics. Bivariate associations between level of guidelines performance and physician, office, and patient characteristics (all conducted at the patient level) were evaluated either by Pearson product-moment or point-biserial correlations, as appropriate. Multiple regression analyses were used to evaluate the independent and combined effects of predictor variables and two specified a priori multiplicative two-way interaction terms, between selected provider and patient characteristics, to predict the level of preventive diabetes practices.
Results Table 3 summarizes the level of performance reported for each of the 11 PRP measures for Study 1 and Study 2. As shown, overall results were similar across the two studies. On average, 74% and 64% of the activities were reported as completed within the recommended time interval in Study 1 and Study 2, respectively (difference significant, t⫽5.15, p⬍0.001). Only 5% of patients in Study 1 and 3% of patients in Study 2 reported meeting all performance criteria. Patients in Study 1 reported somewhat higher levels of self-management/patientfocused activities (especially collaborative goal setting and nutritional counseling; difference significant at t⫽5.20, p⬍0.001). Am J Prev Med 2000;19(1)
11
Table 3. Percent of patients meeting each recommendation Performance measure I. Laboratory/screening measures A. Glycemic control/microvascular HbA1C (1/year) Dilated retinal exam (1/year) Foot exam (1/year) Albumin/microalbumin (1/year) B. Macrovascular Lipids (1/year) Blood pressure (1/year) II. Self-management/patient-focused Set collaborative self-management goal (1/year) Nutrition counseling (1/year) Self-monitoring blood glucose (yes/no) Smoking cessation (yes/no) Patient satisfaction (excellent)a III. Summary scores Total PRP summary (mean percent met) Lab/screening summary (mean percent met) Patient-focused/self-management (mean percent met)
Study 1
Study 2
88% 81% 73% 73%
85% 79% 71% 45%
91% 95%
81% 92%
70% 71% 87% 60% (n⫽35) 16%
46% 44% 78% 58% (n⫽19) 22%
74% 84% 61%
64% 74% 48%
a Excellent rating for all of five patient satisfaction items (diabetes questions answered, access during emergencies, explanation of lab results, courtesy/personal manner of provider, and diabetes care overall).
Closer inspection of Table 3 reveals that achievement of these objectives varied markedly across types of activities. We found substantial variability both in individual measures and in the three summary measures. In both Study 1 and Study 2, patients reported receiving significantly fewer self-management and patient-focused activities (61% and 48%, respectively) than laboratory/ screening measures (84% and 74%, respectively; t⫽14.38, p⬍0.001 and t⫽10.79, p⬍0.001, respectively). Individual preventive activities reported most often were blood pressure checks and lipid and HbA1C testing. Recommended practices received least often were smoking-cessation and nutrition counseling, and collaborative self-management goal setting. Clustering of patient results within physicians and medical offices was minimal and close to zero (intraclass correlations for physicians and offices ranged from 0.0003 to 0.001 and from 0.0005 to 0.008, respectively).
Characteristics Associated with Performance Table 4 presents the bivariate relationships of patient, provider, and office practice characteristics for the two summary variables of number of laboratory/screening activities and patient-focused/self-management activities. As shown, level of involvement in community resources supportive of self-management, as measured by our Chronic Illness Resources Survey,24 was the strongest patient characteristic associated with both laboratory and patient-focused activities, as well as the only variable significant in both studies. Greater use of such resources related to higher performance. In general, few patient demographic or medical status characteristics, or physician characteristics, were meaningfully related to either type of preventive activities. Age, 12
number of comorbid conditions, and years diagnosed with diabetes were also related to level of performance, but less strongly and in only one study. At the physician level, primary care providers who treated more diabetes patients showed higher levels of performance on patient-oriented measures, and patients of internists/endocrinologists had higher levels of laboratory screening measures (but not of patientoriented self-management activities). Because of the general lack of differences by physician type, we conducted a subanalysis to evaluate whether internists and endocrinologists, who were categorized together, differed in their diabetes care practices; there were not enough endocrinologists to present their data separately (n⫽2). We found no differences between the patients from endocrinologists and general internists on either laboratory or self-management measures, or on the overall percent of PRP measures completed. The data (available from authors) were virtually identical and in no case differed by more than 2%. At the office practice level, use of a diabetes registry and use of diabetes guidelines/flowsheets were related to better performance on patient-focused self-management activities. Reporting fewer barriers to self-management education and support was related to higher levels of performance. Variables significant in the bivariate analyses in Table 4 were entered into a multiple regressive equation to predict the level of (1) laboratory measures and (2) patient-focused preventive activities in each study. Greater use of community resources from the Chronic Illness Resources Survey was the only variable to remain significant in both studies, when controlling for the other factors. It significantly predicted the level of
American Journal of Preventive Medicine, Volume 19, Number 1
Table 4. Bivariate correlations between patient, provider, and office practice characteristics and percent of performance indices met
Patient characteristics (nⴝ272–275/155–160) Age Gender (1⫽male, 2⫽female) Take insulin (0⫽no, 1⫽yes) Education (0⫽ ⱕhigh school, 1⫽ ⬎high school) Race (0⫽other, 1⫽Caucasian) Smoking status (0⫽no, 1⫽yes) CIRS total support score Years diagnosed # Other chronic illnesses Physician characteristics (n⫽201–261/113–145) Years since degree Internist (endocrinologist) (0⫽no, 1⫽yes) Gender (1⫽male, 2⫽female) % Caseload composed of diabetes patients Total # diabetes patients Office practice characteristics (n⫽253) Have diabetes registry (0⫽no, 1⫽yes) Use diabetes guidelines/flowsheets (0⫽no, 1⫽yes) Provide self-management support (0⫽no, 1⫽yes) Barriers to providing self-management support (1–4)
Laboratory/ screening measures
Patient-focused self-management measures
.03/⫺.02 ⫺.08/.06 .07/.08 .03/.02 ⫺.07/.12 ⫺.04/.07 .19**/.10 .05/.12 .01/⫺.17*
⫺.10/⫺.16* ⫺.05/⫺.01 .07/.10 .01/.11 ⫺.06/.00 ⫺.06/.08 .41**/.25* ⫺.13*/.05 ⫺.08/⫺.15
.01/⫺.04 .13*/⫺.04 ⫺.07/⫺.04 .03/.10 .08/⫺.07
.06/.03 ⫺.11/⫺.01 .10/.03 .14*/⫺.03 ⫺.10/.06
.04/⫺.03 ⫺.02/.09 ⫺.11/.00 ⫺.08/.08
.13*/⫺.02 .14*/.11 .03/.09 ⫺.23**/.07
Values before the slash are correlations in Study 1; values following the slash are correlations in Study 2. *p⬍0.05 **p⬍0.001
laboratory measures conducted in Study 1 (p⬍0.001) and level of patient-focused activities in both studies (p⬍0.002). Neither of the a priori two-way interaction terms (gender of patient by gender of physician, or registry by guidelines) further enhanced prediction of the number of preventive services received.
Discussion Consistent with earlier reports of diabetes15,16,26 and other preventive practices,2,25,27 we observed both suboptimal and variable levels of preventive practices. We found some encouraging results pertaining to frequent testing of HbA1C (88% and 85%), blood pressure16 (95% and 92%), and lipids (91% and 81%). These findings suggest that the message from the Diabetes Control and Complications Trial8,28 and the United Kingdom Prospective Diabetes Study Group9 that “metabolic control and cardiovascular risk factors matter” has been heard. Unfortunately, other preventive practices— especially those involving patient counseling or behavioral activities as contrasted with laboratory assay measures—were performed significantly less often. We found this consistent with our earlier study on different samples of patients that used both chart reviews and the survey instrument used in this study. This suggests that quality improvement efforts may need to focus on collaborative behavior change activities.29,30 Such activities, including setting mutually agreed-upon goals, incorporating the patients’ social environment into the treatment
plan, and problem-solving to overcome anticipated barriers, characterize effective disease-management programs for diabetes and other chronic illnesses.10,21,29 The important new message to primary care practices is that good diabetes management involves working with patients, not just sending them to the lab. To our knowledge, this is the first study to comprehensively assess both patient and provider/medical office characteristics potentially associated with better performance of recommended preventive activities for diabetes. In contrast to previous research that found differences between various types of physicians, we detected few such relationships for either provision of laboratory or patientfocused services. The only variable consistently predictive and of clinically meaningful magnitude across studies was the amount of support from community resources assessed with the Chronic Illness Resources Survey.24 Although correlational, these results suggest the importance of assisting patients to take advantage of their choice of community resources, consistent with principles of community-oriented primary care.2,31 Strengths of this study include the generally consistent results in two different primary care patient samples and the inclusion of a variety of primary care providers. Another strength was the breadth of the measures of preventive diabetes management behaviors as well as potential patient, provider, and medical office characteristics associated with level of care. Although the primary measures were based on patient self-reports, the validaAm J Prev Med 2000;19(1)
13
tional substudy revealed good congruence between medical chart reviews and patient surveys for activities that could reasonably be expected to appear in both places.25 Limitations of the study include the cross-sectional nature of the predictive analyses and the relative lack of minority patients in the practices studied. Future research is indicated to identify practical methods of improving preventive care of diabetes, as only 3% to 5% of patients reported receipt of all 11 recommended services. Evaluations should be conducted of both provider/health care office or systemfocused interventions32 and patient-focused or “consumer demand” approaches. Need also exists to conduct both assessment and quality improvement research in settings providing service to minority, lower socio-economic status, and rural diabetes patients. Finally, it will be important to look at the provision of other preventive services, as well as diabetes-specific performance measures, because primary care providers often treat patients with multiple chronic illnesses, as well as general prevention needs.
11.
12.
13.
14. 15.
16.
17. 18. 19.
20.
The research was supported by NIH grants RO1DK 51581 and RO1DK35524 –13. We express our appreciation to Shawn Boles, Jane Brown, Ed Feil, Lyn Foster, and Melda DeSalvo for their assistance in data collection.
21. 22.
23.
References 1. Brownson RC, Gurney JG, Land GH. Evidence-based decision making in public health. J Public Health Manage Pract 1999;5:86 –97. 2. Starfield B. Primary care: balancing health needs, services, and technology. New York: Oxford University Press, 1998. 3. U.S. Preventive Services Task Force. Guide to clinical preventive services, 2nd ed. Baltimore, MD: Williams & Wilkins, 1996. 4. National Committee for Quality Assurance. Health plan employer data and information set 3.0. Washington, DC: National Committee for Quality Assurance, 1996. 5. Clement S. Diabetes self-management education. Diabetes Care 1995;18: 1204 –14. 6. Glasgow RE, Eakin EG. Issues in diabetes self-management. In: Shumaker SA, Schron EB, Ockene JK, McBee WL, eds. The handbook of health behavior change. New York: Springer Publishing Company, 1998:435– 61. 7. Glasgow RE, Fisher EB, Anderson BJ, et al. Behavioral science in diabetes: contributions and opportunities. Diabetes Care 1999;22:832– 43. 8. DCCT Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 1993;329:977– 86. 9. UK Prospective Diabetes Study Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk complications in patients with type 2 diabetes (UKPDS 33). Lancet 1998;12:837–53. 10. Glasgow RE, Hiss RG, Anderson RM, et al. Behavioral research related to
14
24.
25.
26. 27.
28.
29. 30.
31.
32.
the establishment of a chronic disease model for diabetes care. National Institutes of Diabetes and Digestive and Kidney Diseases. Conference Proceedings. Bethesda, MD: 1999. National Diabetes Data Group, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. Diabetes in America (NIH Pub. No. 95–1468). Bethesda, MD: National Institutes of Health, 1995. Centers for Disease Control and Prevention. Diabetes: a serious public health problem. Atlanta, GA: U.S. Department of Health and Human Services, 1996. National Committee for Quality Assurance. Diabetes quality improvement project: Initial measures set. Washington, DC: National Committee for Quality Assurance, 1998. Joyner L, McNeeley S, Kahn R. ADA’s provider recognition program. HMO Pract 1997;11:168 –70. Glasgow RE, Boles SM, Calder D, Dreyer L, Bagdade J. Diabetes care practices in primary care. Results from two samples and three performance measure sets. Diabetes Educator 1999;25:755– 63. Kahn R, Greenfield S, Kaplan SH, Ninomiya J. Assessing quality of care in patients with diabetes. Are process or outcome measures adequate? Ann Intern Med. In press. Harris MI. Medical care for patients with diabetes: epidemiologic aspects. Ann Intern Med 1996;124:117–22. Brownson RC, Remington PL, Davis JR. Chronic disease epidemiology and control. Baltimore, MD: Port City Press, 1993. Khunti K, Baker R, Rumsey M, LaKhani M. Quality of care of patients with diabetes: collation of data from multi-practice audits of diabetes in primary care. J Fam Pract 1999;16:54 –9. Hiss RG, Anderson RM, Hess GE, Stepien CJ, Davis WK. Community diabetes care: a 10-year perspective. Diabetes Care 1994;17:1124 –34. Roter DL, Hall JA. Studies of doctor-patient interaction. Annu Rev Public Health 1989;10:163– 80. Glasgow RE, Toobert DJ. Brief, computer-assisted diabetes dietary selfmanagement counseling: effects on behavior physiologic outcomes, and quality of life. Medical Care. In press. Feil E, Glasgow RE, McKay HG, Boles SM, Barrera M Jr. Participation in Internet-based diabetes self-management. Diabetes Educator. In press. Glasgow RE, Strycker LA, Toobert DJ, Eakin EG. The Chronic Illness Resources Survey: a social-ecologic approach to assessing support for disease self-management. J Behav Med. In press. Stange KC, Zyzanski SJ, Smith TF, et al. How valid are medical records and patient questionnaires for physician profiling and health services research? Med Care 1998;36:851–7. PRO-West. Diabetes outcomes measurement project. Seattle, WA: 1999. Solberg LI, Kottke TE, Brekke ML, Conn SA, Magnan S, Amundson G. The case of the missing clinical preventive services systems. Effective Clin Pract 1998;1:33– 8. Fisher EB Jr, Heins JM, Hiss RG, et al. Metabolic control matters: nationwide translation of the Diabetes Control and Complications Trial: analysis and recommendations (NIH Pub. No. 94 –3773). Bethesda, MD: National Institute of Diabetes and Digestive and Kidney Diseases, 1994. Wagner EH, Austin B, Von Korff M. Improving outcomes in chronic illness. Managed Care Q 1996;4:12–25. Glasgow RE, Eakin EG. Medical office-based interventions. In: Snoek FJ, Skinner CS, eds. Psychological aspects of diabetes care. London: John Wiley. In press. Rhyne R, Bogue R, Kukulka G, Fulmer.H. Community-oriented primary care: health care for the 21st century. Washington, DC: American Public Health Association, 1998. Anderson LA, Janes GR, Jenkins C. Implementing preventive services: to what extent can we change provider performance in ambulatory care? A review of the screening, immunization, and counseling literature. Ann Behav Med 1998;20:161–7.
American Journal of Preventive Medicine, Volume 19, Number 1
To assess the level of physician performance on American Diabetes Association Provider Recognition Program (PRP) measures in two samples of primary care patients, as well as to identify patient, physician, and office characteristics related to performance levels.
Methods:
In the two studies, we surveyed 435 Type 2 diabetes patients, cared for by 47 different physicians, on their receipt of PRP preventive care activities.
Results:
Overall, patients in the two samples reported receiving 74% and 64% of recommended services. In both samples, performance of microvascular/glycemic control activities and cardiovascular lab checks (84% and 74%) was significantly higher than behavioral self-management/patient-focused activities (61% and 48%) (p⬍0.001). From a set of patient, physician, and practice setting characteristics, only the use of community resources for chronic illness management support was associated with service performance.
Conclusions: We found considerable variability in the levels of performance in providing PRP-recommended activities. Greater attention should be focused on self-management and patient-focused activities, given that these are delivered less frequently than medical/laboratory checks. Medical Subject Headings (MeSH): diabetes mellitus, preventive medicine, primary health care, quality of health care, task performance and analysis (Am J Prev Med 2000;19(1): 9 –14) © 2000 American Journal of Preventive Medicine
Introduction
T
he emergence of clinical practice guidelines, disease management programs, evidence-based clinical pathways, “best practices,” and disease-specific performance measures is prominent among the many changes in health care during the past decade. Spurred by the confluence of evidence-based medicine1,2 and the desire of managed care and other organizations to reduce nonproductive variations in practice, clinical performance measures and guidelines are applied with increasing frequency.2– 4 Several factors have all contributed to the development of diabetes management recommendations, performance indicators, and guidelines: recent advances in diabetes self-management5–7; the emergence of conclusive data that demonstrate the efficacy of organized, comprehensive management in reducing diabetes complications and mortality8 –10; and recognition of the enormous health care costs of diabetes.11,12 Among these, the Provider Recognition Program (PRP) performance measures of the American Diabetes Association are promiFrom the AMC Cancer Research Center (Glasgow), Denver, Colorado; and Oregon Research Institute (Strycker), Eugene, Oregon Address correspondence and reprint requests to: Russell E. Glasgow, PhD, 11716 98th Place SW, Vashon, WA 98070. E-mail: [email protected].
nent. The related Diabetes Quality Improvement Program measures,13 essentially a subset of the PRP measures, have been adopted recently as Health Employer Data Info Survey criteria. InitialstudiesofthePRPmeasureshavefoundconsiderable variability in performance levels across various measures and providers.14 –16 Specifically, patient-focused and self-management practices appear to occur less often than laboratory/screening activities, and internists may differ from family physicians in their diabetes care practices.15 Much of the brunt of following these new recommendations and best practices falls on the primary care provider, as the great majority of diabetes patients are managed in primary care.2,17 Given that most of the recommended prevention activities must be accomplished or initiated during the ever more time-limited primary care visit, it is understandable that many of these activities are performed at substandard levels.2,10 Diabetes is a complex and challenging chronic illness that requires numerous lifestyle changes, including microvascular and macrovascular disease–prevention activities.6,11,12,18 Also, most diabetes patients have other comorbid chronic illnesses that further complicate their management. Little research exists on patient, provider, or office characteristics associated with either diabetes patient– physician interactions or with the level of recom-
Am J Prev Med 2000;19(1) 0749-3797/00/$–see front matter © 2000 American Journal of Preventive Medicine • Published by Elsevier Science Inc. PII S0749-3797(00)00157-4
9
Table 1. Characteristics of patients, providers, and office practices Mean (SD) or percent (for dichotomous variables)
Characteristic Patient characteristics Age % Female % on insulin % ⬎ High school education % ⬎ $30,000 income % Caucasian % Received diabetes education % Smoke CIRS total support score (scale⫽1–5) Years diagnosed # Other chronic illnesses (Range⫽0–12 in Study 1; 0–9 in Study 2)
Study 1 (n⫽275) 60 (9) 56% 16% 46% Not collected 92% 57% 13% 2.6 (.5) 6.4 (6.4) 2.7 (2.1)
Study 2 (n⫽160) 59 (9) 53% 36% 53% 52% 91% Not collected 13% 3.2 (.9) 11.3 (7.8) 2.7 (2.3)
Provider characteristics Years since degree Medical specialty % Family practice/GP % Endocrinologist/internal medicine Othera % Female % Caseload composed of diabetes patients ⬍5% 5%–10% 11%–25% 26%–50% 51%–75% ⬎76% Total # diabetes patients
Study 1 (n⫽33) 17 (9)
Study 2 (n⫽14) 23 (10)
32% 40% 24% 4% 0% 0% 418 (753)
20% 60% 10% 0% 0% 10% 1001 (792)
Office practice characteristics % Have diabetes registry % Use diabetes guidelines/flowsheets % Provide self-management support Barriers to providing self-management support (scale⫽1–4)
Study 1 (n⫽12) 50% 80% 100% 2.0 (.3)
Study 2 (n⫽5) 80% 80% 80% 2.3 (.7)
55% 33% 12% 45%
14% 86% 0% 21%
a
The “other” category includes physicians’ assistants and family nurse practitioners. SD, standard deviation.
mended preventive practices.15,19,20 In particular, more research has been recommended on contextual factors, such as the patient’s social environment and health care system characteristics,7,10 and on patient and physician factors, such as gender, that may influence performance of preventive activities.21 The purpose of this report is threefold: (1) to report on the overall levels of PRP preventive measures in two different primary care samples, as replication is important, especially in new areas of investigation; (2) to determine whether levels of performance for laboratory/ screening were higher than for patient-focused/selfmanagement activities; and (3) to determine whether we could replicate reported findings concerning patient and physician characteristics associated with performance of these prevention activities.
Methods Recruitment of Practices, Providers, and Patients Within each of the two different health care systems in the Pacific Northwest, we made arrangements with influential
10
physician leaders to facilitate access to primary care providers within their systems. We approached 35 and 16 primary care physicians for Study 1 and Study 2, respectively. Of these, 33 (94%) and 14 (88%) participated. We summarize their characteristics in Table 1. Fifty-five percent and fourteen percent were family physicians, 45% and 21% were female, the average number of years of practice since receiving their degrees was 17 and 23, and 80% in both studies reported implementing some form of diabetes care guidelines in their practices. Participating physicians came from 17 different practices and almost all see a mixture of patients receiving managed care, fee-for-service, and government-provided health insurance. In the practices of the participating physicians, all Type 2 dependent diabetes patients aged 40 to 75 years received letters inviting them to participate in a randomized trial of interactive health communication to assist in diabetes selfmanagement (multimedia touch-screen and Internet-based intervention, in Study 1 and Study 2, respectively). As reported elsewhere22,23 76% and 60% of eligible patients participated in Study 1 and Study 2, respectively, and in both studies the characteristics of participants were similar to nonparticipants.
American Journal of Preventive Medicine, Volume 19, Number 1
Table 1 shows that patients in the two studies averaged age 60 and 59 years and had been diagnosed for an average of 6.4 and 11.3 years, 16% and 36% were taking insulin, 56% and 53% were women, and patients in both samples had an average of 2.7 comorbid chronic illnesses.
Table 2. Correspondence between patient survey and medical chart review on seven PRP items
Measure
Measures A patient version of the 11 performance measures recommended by the American Diabetes Association’s PRP program, used in previous research,15 was adapted for these studies. Each question asked, in lay terms, how long (e.g., 6 months, 1 year, 2 years, 5 years, never) it had been since the patient had received the service in question (e.g., “How long has it been since a health professional checked your feet for any sores or irritation?” and “About how long has it been since you last had your blood pressure taken by a doctor, nurse, or other health care professional?”). At the time this study was conducted, these 11 items assessed all PRP measures. Recently, lab values for HbA1C, lipids, and blood pressure have been added to the set. These data are not available for these samples. Study 1 measures were collected by paper and pencil survey, and Study 2 measures were collected using the Internet. Data for the present study were collected as part of the baseline measures for the two separate intervention studies. In addition to the patient and physician characteristics previously described, measures were also collected from patients by survey and for physicians and offices by surveying medical office managers. As shown in Table 1, other patient characteristics assessed were education, family, income, race, and smoking. Other physician and office characteristics were size of diabetes caseload, whether the practice had a diabetes registry, and whether it offered some form of self-management support. Finally, to provide a measure of the richness of the patients’ social environment regarding support for self-management activities, the Chronic Illness Resources Survey was administered. This measure and its psychometric characteristics are described in detail elsewhere,24 but the measure has been found to be a reliable and valid measure of the extent to which patients have used various levels of support resources throughout their community (e.g., family and friends, work, neighborhoods, the media). For a subsample of 27 patients in Study 2, whose physicians also participated in a separate quality improvement effort, we conducted a validational study using chart reviews to provide corroborative data on the seven preventive practices that we judged likely to be recorded in patient medical records. Some items, such as patient satisfaction, are by definition assessed only by patient survey and others, such as collaborative goal setting, are notoriously inconsistently recorded in patient charts.25,26 We report the results of this substudy in Table 2. As shown, patient reports were generally congruent with the charts reviewed. Patient surveys reported that an average 5.7 of the 7 practices were met, compared with an average of 5.1 recorded in the patients’ charts; we found relatively good correspondence between overall number of criteria met collected by the two methods (Kendall’s W⫽0.45, p⫽ 0.0005). The only real discrepancy between patient reports and medical records was on
HbA1C testing Dilated eye exam Foot exam Lipid testing Blood pressure testing Smoking-cessation counseling Nutrition counselinga Overall number of seven indices met
Patient survey
Chart review
92% 81.5% 93% 88.5% 100% 17%
96% 74% 96% 100% 96% 0%
54%
100%
Kendall’s coefficient of concordance (W) (significance)
5.7 (.67) 5.1 (1.1) .45 (p⫽0.0005)
Based on 27 patients from Study 2 for whom medical records were available. a Measure in chart review was referral to dietitian; item in patient survey was self-reported physician referral to dietitian.
referral for nutritional counseling (54% patient report vs 100% chart review).
Analyses We used descriptive statistics (means, standard deviations, and percentages) to describe the sample and level of performance of the individual performance measures. We used Kendall’s coefficient of concordance to evaluate the association between medical record and patient-reported overall measure of seven preventive actions. We used t tests to compare the numbers of laboratory/screening (0 to 6) and patient- focused/behavioral activities (0 to 5) performed across studies and across different provider characteristics. Bivariate associations between level of guidelines performance and physician, office, and patient characteristics (all conducted at the patient level) were evaluated either by Pearson product-moment or point-biserial correlations, as appropriate. Multiple regression analyses were used to evaluate the independent and combined effects of predictor variables and two specified a priori multiplicative two-way interaction terms, between selected provider and patient characteristics, to predict the level of preventive diabetes practices.
Results Table 3 summarizes the level of performance reported for each of the 11 PRP measures for Study 1 and Study 2. As shown, overall results were similar across the two studies. On average, 74% and 64% of the activities were reported as completed within the recommended time interval in Study 1 and Study 2, respectively (difference significant, t⫽5.15, p⬍0.001). Only 5% of patients in Study 1 and 3% of patients in Study 2 reported meeting all performance criteria. Patients in Study 1 reported somewhat higher levels of self-management/patientfocused activities (especially collaborative goal setting and nutritional counseling; difference significant at t⫽5.20, p⬍0.001). Am J Prev Med 2000;19(1)
11
Table 3. Percent of patients meeting each recommendation Performance measure I. Laboratory/screening measures A. Glycemic control/microvascular HbA1C (1/year) Dilated retinal exam (1/year) Foot exam (1/year) Albumin/microalbumin (1/year) B. Macrovascular Lipids (1/year) Blood pressure (1/year) II. Self-management/patient-focused Set collaborative self-management goal (1/year) Nutrition counseling (1/year) Self-monitoring blood glucose (yes/no) Smoking cessation (yes/no) Patient satisfaction (excellent)a III. Summary scores Total PRP summary (mean percent met) Lab/screening summary (mean percent met) Patient-focused/self-management (mean percent met)
Study 1
Study 2
88% 81% 73% 73%
85% 79% 71% 45%
91% 95%
81% 92%
70% 71% 87% 60% (n⫽35) 16%
46% 44% 78% 58% (n⫽19) 22%
74% 84% 61%
64% 74% 48%
a Excellent rating for all of five patient satisfaction items (diabetes questions answered, access during emergencies, explanation of lab results, courtesy/personal manner of provider, and diabetes care overall).
Closer inspection of Table 3 reveals that achievement of these objectives varied markedly across types of activities. We found substantial variability both in individual measures and in the three summary measures. In both Study 1 and Study 2, patients reported receiving significantly fewer self-management and patient-focused activities (61% and 48%, respectively) than laboratory/ screening measures (84% and 74%, respectively; t⫽14.38, p⬍0.001 and t⫽10.79, p⬍0.001, respectively). Individual preventive activities reported most often were blood pressure checks and lipid and HbA1C testing. Recommended practices received least often were smoking-cessation and nutrition counseling, and collaborative self-management goal setting. Clustering of patient results within physicians and medical offices was minimal and close to zero (intraclass correlations for physicians and offices ranged from 0.0003 to 0.001 and from 0.0005 to 0.008, respectively).
Characteristics Associated with Performance Table 4 presents the bivariate relationships of patient, provider, and office practice characteristics for the two summary variables of number of laboratory/screening activities and patient-focused/self-management activities. As shown, level of involvement in community resources supportive of self-management, as measured by our Chronic Illness Resources Survey,24 was the strongest patient characteristic associated with both laboratory and patient-focused activities, as well as the only variable significant in both studies. Greater use of such resources related to higher performance. In general, few patient demographic or medical status characteristics, or physician characteristics, were meaningfully related to either type of preventive activities. Age, 12
number of comorbid conditions, and years diagnosed with diabetes were also related to level of performance, but less strongly and in only one study. At the physician level, primary care providers who treated more diabetes patients showed higher levels of performance on patient-oriented measures, and patients of internists/endocrinologists had higher levels of laboratory screening measures (but not of patientoriented self-management activities). Because of the general lack of differences by physician type, we conducted a subanalysis to evaluate whether internists and endocrinologists, who were categorized together, differed in their diabetes care practices; there were not enough endocrinologists to present their data separately (n⫽2). We found no differences between the patients from endocrinologists and general internists on either laboratory or self-management measures, or on the overall percent of PRP measures completed. The data (available from authors) were virtually identical and in no case differed by more than 2%. At the office practice level, use of a diabetes registry and use of diabetes guidelines/flowsheets were related to better performance on patient-focused self-management activities. Reporting fewer barriers to self-management education and support was related to higher levels of performance. Variables significant in the bivariate analyses in Table 4 were entered into a multiple regressive equation to predict the level of (1) laboratory measures and (2) patient-focused preventive activities in each study. Greater use of community resources from the Chronic Illness Resources Survey was the only variable to remain significant in both studies, when controlling for the other factors. It significantly predicted the level of
American Journal of Preventive Medicine, Volume 19, Number 1
Table 4. Bivariate correlations between patient, provider, and office practice characteristics and percent of performance indices met
Patient characteristics (nⴝ272–275/155–160) Age Gender (1⫽male, 2⫽female) Take insulin (0⫽no, 1⫽yes) Education (0⫽ ⱕhigh school, 1⫽ ⬎high school) Race (0⫽other, 1⫽Caucasian) Smoking status (0⫽no, 1⫽yes) CIRS total support score Years diagnosed # Other chronic illnesses Physician characteristics (n⫽201–261/113–145) Years since degree Internist (endocrinologist) (0⫽no, 1⫽yes) Gender (1⫽male, 2⫽female) % Caseload composed of diabetes patients Total # diabetes patients Office practice characteristics (n⫽253) Have diabetes registry (0⫽no, 1⫽yes) Use diabetes guidelines/flowsheets (0⫽no, 1⫽yes) Provide self-management support (0⫽no, 1⫽yes) Barriers to providing self-management support (1–4)
Laboratory/ screening measures
Patient-focused self-management measures
.03/⫺.02 ⫺.08/.06 .07/.08 .03/.02 ⫺.07/.12 ⫺.04/.07 .19**/.10 .05/.12 .01/⫺.17*
⫺.10/⫺.16* ⫺.05/⫺.01 .07/.10 .01/.11 ⫺.06/.00 ⫺.06/.08 .41**/.25* ⫺.13*/.05 ⫺.08/⫺.15
.01/⫺.04 .13*/⫺.04 ⫺.07/⫺.04 .03/.10 .08/⫺.07
.06/.03 ⫺.11/⫺.01 .10/.03 .14*/⫺.03 ⫺.10/.06
.04/⫺.03 ⫺.02/.09 ⫺.11/.00 ⫺.08/.08
.13*/⫺.02 .14*/.11 .03/.09 ⫺.23**/.07
Values before the slash are correlations in Study 1; values following the slash are correlations in Study 2. *p⬍0.05 **p⬍0.001
laboratory measures conducted in Study 1 (p⬍0.001) and level of patient-focused activities in both studies (p⬍0.002). Neither of the a priori two-way interaction terms (gender of patient by gender of physician, or registry by guidelines) further enhanced prediction of the number of preventive services received.
Discussion Consistent with earlier reports of diabetes15,16,26 and other preventive practices,2,25,27 we observed both suboptimal and variable levels of preventive practices. We found some encouraging results pertaining to frequent testing of HbA1C (88% and 85%), blood pressure16 (95% and 92%), and lipids (91% and 81%). These findings suggest that the message from the Diabetes Control and Complications Trial8,28 and the United Kingdom Prospective Diabetes Study Group9 that “metabolic control and cardiovascular risk factors matter” has been heard. Unfortunately, other preventive practices— especially those involving patient counseling or behavioral activities as contrasted with laboratory assay measures—were performed significantly less often. We found this consistent with our earlier study on different samples of patients that used both chart reviews and the survey instrument used in this study. This suggests that quality improvement efforts may need to focus on collaborative behavior change activities.29,30 Such activities, including setting mutually agreed-upon goals, incorporating the patients’ social environment into the treatment
plan, and problem-solving to overcome anticipated barriers, characterize effective disease-management programs for diabetes and other chronic illnesses.10,21,29 The important new message to primary care practices is that good diabetes management involves working with patients, not just sending them to the lab. To our knowledge, this is the first study to comprehensively assess both patient and provider/medical office characteristics potentially associated with better performance of recommended preventive activities for diabetes. In contrast to previous research that found differences between various types of physicians, we detected few such relationships for either provision of laboratory or patientfocused services. The only variable consistently predictive and of clinically meaningful magnitude across studies was the amount of support from community resources assessed with the Chronic Illness Resources Survey.24 Although correlational, these results suggest the importance of assisting patients to take advantage of their choice of community resources, consistent with principles of community-oriented primary care.2,31 Strengths of this study include the generally consistent results in two different primary care patient samples and the inclusion of a variety of primary care providers. Another strength was the breadth of the measures of preventive diabetes management behaviors as well as potential patient, provider, and medical office characteristics associated with level of care. Although the primary measures were based on patient self-reports, the validaAm J Prev Med 2000;19(1)
13
tional substudy revealed good congruence between medical chart reviews and patient surveys for activities that could reasonably be expected to appear in both places.25 Limitations of the study include the cross-sectional nature of the predictive analyses and the relative lack of minority patients in the practices studied. Future research is indicated to identify practical methods of improving preventive care of diabetes, as only 3% to 5% of patients reported receipt of all 11 recommended services. Evaluations should be conducted of both provider/health care office or systemfocused interventions32 and patient-focused or “consumer demand” approaches. Need also exists to conduct both assessment and quality improvement research in settings providing service to minority, lower socio-economic status, and rural diabetes patients. Finally, it will be important to look at the provision of other preventive services, as well as diabetes-specific performance measures, because primary care providers often treat patients with multiple chronic illnesses, as well as general prevention needs.
11.
12.
13.
14. 15.
16.
17. 18. 19.
20.
The research was supported by NIH grants RO1DK 51581 and RO1DK35524 –13. We express our appreciation to Shawn Boles, Jane Brown, Ed Feil, Lyn Foster, and Melda DeSalvo for their assistance in data collection.
21. 22.
23.
References 1. Brownson RC, Gurney JG, Land GH. Evidence-based decision making in public health. J Public Health Manage Pract 1999;5:86 –97. 2. Starfield B. Primary care: balancing health needs, services, and technology. New York: Oxford University Press, 1998. 3. U.S. Preventive Services Task Force. Guide to clinical preventive services, 2nd ed. Baltimore, MD: Williams & Wilkins, 1996. 4. National Committee for Quality Assurance. Health plan employer data and information set 3.0. Washington, DC: National Committee for Quality Assurance, 1996. 5. Clement S. Diabetes self-management education. Diabetes Care 1995;18: 1204 –14. 6. Glasgow RE, Eakin EG. Issues in diabetes self-management. In: Shumaker SA, Schron EB, Ockene JK, McBee WL, eds. The handbook of health behavior change. New York: Springer Publishing Company, 1998:435– 61. 7. Glasgow RE, Fisher EB, Anderson BJ, et al. Behavioral science in diabetes: contributions and opportunities. Diabetes Care 1999;22:832– 43. 8. DCCT Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 1993;329:977– 86. 9. UK Prospective Diabetes Study Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk complications in patients with type 2 diabetes (UKPDS 33). Lancet 1998;12:837–53. 10. Glasgow RE, Hiss RG, Anderson RM, et al. Behavioral research related to
14
24.
25.
26. 27.
28.
29. 30.
31.
32.
the establishment of a chronic disease model for diabetes care. National Institutes of Diabetes and Digestive and Kidney Diseases. Conference Proceedings. Bethesda, MD: 1999. National Diabetes Data Group, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. Diabetes in America (NIH Pub. No. 95–1468). Bethesda, MD: National Institutes of Health, 1995. Centers for Disease Control and Prevention. Diabetes: a serious public health problem. Atlanta, GA: U.S. Department of Health and Human Services, 1996. National Committee for Quality Assurance. Diabetes quality improvement project: Initial measures set. Washington, DC: National Committee for Quality Assurance, 1998. Joyner L, McNeeley S, Kahn R. ADA’s provider recognition program. HMO Pract 1997;11:168 –70. Glasgow RE, Boles SM, Calder D, Dreyer L, Bagdade J. Diabetes care practices in primary care. Results from two samples and three performance measure sets. Diabetes Educator 1999;25:755– 63. Kahn R, Greenfield S, Kaplan SH, Ninomiya J. Assessing quality of care in patients with diabetes. Are process or outcome measures adequate? Ann Intern Med. In press. Harris MI. Medical care for patients with diabetes: epidemiologic aspects. Ann Intern Med 1996;124:117–22. Brownson RC, Remington PL, Davis JR. Chronic disease epidemiology and control. Baltimore, MD: Port City Press, 1993. Khunti K, Baker R, Rumsey M, LaKhani M. Quality of care of patients with diabetes: collation of data from multi-practice audits of diabetes in primary care. J Fam Pract 1999;16:54 –9. Hiss RG, Anderson RM, Hess GE, Stepien CJ, Davis WK. Community diabetes care: a 10-year perspective. Diabetes Care 1994;17:1124 –34. Roter DL, Hall JA. Studies of doctor-patient interaction. Annu Rev Public Health 1989;10:163– 80. Glasgow RE, Toobert DJ. Brief, computer-assisted diabetes dietary selfmanagement counseling: effects on behavior physiologic outcomes, and quality of life. Medical Care. In press. Feil E, Glasgow RE, McKay HG, Boles SM, Barrera M Jr. Participation in Internet-based diabetes self-management. Diabetes Educator. In press. Glasgow RE, Strycker LA, Toobert DJ, Eakin EG. The Chronic Illness Resources Survey: a social-ecologic approach to assessing support for disease self-management. J Behav Med. In press. Stange KC, Zyzanski SJ, Smith TF, et al. How valid are medical records and patient questionnaires for physician profiling and health services research? Med Care 1998;36:851–7. PRO-West. Diabetes outcomes measurement project. Seattle, WA: 1999. Solberg LI, Kottke TE, Brekke ML, Conn SA, Magnan S, Amundson G. The case of the missing clinical preventive services systems. Effective Clin Pract 1998;1:33– 8. Fisher EB Jr, Heins JM, Hiss RG, et al. Metabolic control matters: nationwide translation of the Diabetes Control and Complications Trial: analysis and recommendations (NIH Pub. No. 94 –3773). Bethesda, MD: National Institute of Diabetes and Digestive and Kidney Diseases, 1994. Wagner EH, Austin B, Von Korff M. Improving outcomes in chronic illness. Managed Care Q 1996;4:12–25. Glasgow RE, Eakin EG. Medical office-based interventions. In: Snoek FJ, Skinner CS, eds. Psychological aspects of diabetes care. London: John Wiley. In press. Rhyne R, Bogue R, Kukulka G, Fulmer.H. Community-oriented primary care: health care for the 21st century. Washington, DC: American Public Health Association, 1998. Anderson LA, Janes GR, Jenkins C. Implementing preventive services: to what extent can we change provider performance in ambulatory care? A review of the screening, immunization, and counseling literature. Ann Behav Med 1998;20:161–7.
American Journal of Preventive Medicine, Volume 19, Number 1