RSS Feeds
Improving Diabetes Care in the Primary Health Setting: The Indian Health Service Experience
- Dorothy Gohdes, MD;
- Steve Rith-Najarian, MD;
- Kelly Acton, MD; and
- Ray Shields, MD
- From the Indian Health Service Diabetes Program, Albuquerque, New Mexico; the Bemidji Area Indian Health Service, Bemidji, Minnesota; the Billings Area Diabetes Program, Tribal Health and Human Services, St. Ignatius, Montana; and the Portland Area Diabetes Program, Bellingham, Washington. Note: This article is one of a series of articles comprising an Annals of Internal Medicine supplement entitled “Risks and Benefits of Intensive Management in Non-Insulin-dependent Diabetes Mellitus: The Fifth Regenstrief Conference.” To view a complete list of the articles included in this supplement, please view its Table of Contents. Disclaimer: The opinions expressed here are those of the authors and do not necessarily reflect the views of the Indian Health Service. Acknowledgments: The authors thank the many dedicated Indian Health Services and tribal health providers who have contributed to this effort and Dr. Betty Skipper for her guidance. Requests for Reprints: Dorothy Gohdes, MD, Indian Health Service Diabetes Program, 5300 Homestead Road, NE, Albuquerque, NM 87110. Current Author Addresses: Dr. Gohdes: Indian Health Service Diabetes Program, 5300 Homestead Road, NE, Albuquerque, NM 87110.
Abstract
Purpose: To identify key systems issues from the Indian Health Service (IHS) experience that must be addressed to improve metabolic control among patients with noninsulin-dependent diabetes mellitus (NIDDM) who were followed in primary care settings.
Data Sources: Records of diabetic patients seen in IHS facilities in specific geographic regions from 1987 to 1994.
Study Selection: A representative sample of charts from each facility was reviewed yearly to measure key variables. The sampling frame was the number of diabetic patients currently active on the registry and the sample size calculated to measure a 10% change in selected practices at each facility.
Extraction: Regional diabetes coordinators reviewed charts or trained local providers to sample and extract data in a standard format.
Results: Regional data were examined to show trends in the performance of immunizations and foot examinations and in other variables such as hypertension and metabolic control. The percentage of diabetic patients who received a single dose of pneumococcal vaccine improved from 24% in 1987 to 1988 to 57% in 1994 (P < 0.01 for trend) among diabetic patients in Minnesota, Wisconsin, and Michigan. Rates of yearly comprehensive foot examination increased from 36% to 58% (P < 0.01 for trend) over the same period. In Montana and Wyoming, the percentage of diabetic patients with uncontrolled hypertension (defined as the mean of three systolic blood pressure measurements of more than equals 140 mm Hg or diastolic pressure measurements more than equals 90 mm Hg, or both, during the previous year) decreased from 36% in 1992 to 25% in 1993 after the regional diabetes coordinator emphasized hypertension control. In 1994, when less emphasis was placed on hypertension, 33% of the diabetic patients had uncontrolled hypertension. Estimates of metabolic control from records of diabetic patients in Washington, Oregon, and Idaho in 1994 showed that 29% of patients had excellent metabolic control (a hemoglobin A1c [HbA1c] level less than equals 7.5% or mean blood glucose level less than equals 9.2 mmol/L) within the past year; only 9% experienced poor control (a HbA1c level more than 12% or mean blood glucose level more than 18.9 mmol/L).
Conclusions: The IHS experience shows that standard, ongoing monitoring of key variables allows facilities to improve diabetes care. Simple, reliable methods of defining metabolic control combined with a feedback system in the primary care setting are needed to improve metabolic control in patients with NIDDM.
American Indians and Alaskan Natives, like many indigenous peoples, are experiencing an epidemic of non–insulin-dependent diabetes mellitus (NIDDM) [1-3]. To improve the long-term outcomes of diabetic patients, Indian Health Service (IHS) primary care physicians acting as regional diabetes coordinators identified preventive practices that could be incorporated into the care of diabetic patients who were followed in primary care facilities, along with key surrogate variables that could be measured to evaluate care and intermediate patient outcomes. Since 1986, 12 regional coordinators have promoted these variables as minimum standards of care and have revised them periodically to reflect clinically significant scientific advances. Efforts to encourage the 176 IHS, tribal, and urban facilities to incorporate these practices into their settings were collaborative and collegial and were not done as closely monitored research studies. Timely, affordable, and accurate measures of the implementation of these standards were essential to ensure that facilities in this geographically widespread, culturally diverse, and largely rural primary health system could assess their own practices and identify opportunities for improvement.
Methods
The actual strategy for measuring implementation of these standards and providing ongoing, appropriate feedback has evolved over several years, as discussed in detail in previous publications [4, 5]. Briefly, staff at each facility were encouraged to maintain a registry of active patients with diabetes who were living in the community; written guidance was provided for the definition of “active” patients. Selected clinical interventions and patient variables were measured yearly in a sample of charts drawn in a systematic, random fashion from the registry. The sample size was sufficient to yield estimates within approximately 10% of the true rate for that facility with a confidence of 90% or greater using a threshold estimate of 70% for most variables.
Chart review was done by the regional diabetes coordinators and other professional staff trained by the coordinators; a uniform set of definitions was used throughout the United States. The abstracted data were entered into a general-purpose microcomputer-based software program [6]. Summary reports were generated immediately to be available to the facility's staff for quality improvement activities and program planning. The data from participating sites were then combined regionally and finally aggregated to determine national rates.
This comprehensive, ongoing surveillance program provides timely, affordable, and accurate measures to identify and evaluate quality improvement strategies at the local, regional, and national levels. Selected trends from several geographic areas for recent years are reported here to document the utility of a simple feedback system and highlight challenges that must be addressed to improve care provided to diabetic patients in primary care settings.
Results and Discussion
Administering pneumococcal vaccine to protect diabetic patients from pneumococcal pneumonia is a standard recommendation from several national advisory bodies on immunization; it is particularly important to Native American communities, which have high death rates from pneumonia and a particular susceptibility to pneumococcal infection [7-10]. Table 1 shows a consistent increase from 1987 to 1994 in the percentage of diabetic patients in Minnesota, Wisconsin, and Michigan who received pneumococcal vaccination. After immunizations were promoted and immunization status was monitored over several years, the overall percentage of diabetic patients who had received the vaccine once increased from 24% to 59% (P < 0.001 for trend).
Over the same period, rates of yearly comprehensive foot examinations increased less consistently, as shown in Table 2. Accurate measures using monofilaments and clinical examinations to quantify the risk for lower-extremity problems were promoted as a preventive practice because many IHS primary care settings had a shortage of foot care resources and providers needed to identify patients who could benefit most from specialized intervention [11-14]. In a large study in 1993, targeting diabetic patients at high risk for foot problems was shown to reduce clinical abnormalities [15]. In Alaska, identifying high-risk patients and targeting services to these patients decreased rates of lower-extremity amputation [16]. From 1987 to 1994, primary care providers gradually became familiar with the techniques of neurologic examination with monofilaments and examining feet for peripheral pulses and deformities. However, primary care physicians often delegated the examination to other providers who were not consistently available, and the increase in examination rates was not sustained as consistently as the increase in immunization practices.
Surveillance of diabetes care can also measure other important, related parameters such as blood pressure. Hypertension control in diabetic patients has received widespread emphasis in recent years because of the substantial and growing evidence of the adverse effect of hypertension on many diabetic complications and the availability of effective, well-tolerated antihypertensive agents [17]. Because blood pressure is monitored regularly in most facilities, regional diabetes coordinators have been able to collect the last three blood pressure measurements recorded in the past year [5]. Mean systolic and mean diastolic pressures were calculated separately. Table 3 shows categories of blood pressure measurements taken in 1992, 1993, and 1994 in 10 IHS facilities in Montana and Wyoming. Uncontrolled hypertension—defined as a mean systolic pressure of 140 mm Hg or more or a mean diastolic pressure of 90 mm Hg or more, or both—was found in 36% of the patient charts in 1992. As a result of the data, quality improvement efforts were initiated. The regional diabetes coordinator visited each of the 10 facilities, reviewed the data with the providers to give feedback, and re-emphasized the importance of hypertension control. Despite staff turnover and the necessity for patients to return monthly for medication refills, the rate of uncontrolled hypertension decreased substantially to 25% during 1993, as shown in Table 3. During 1994, however, less emphasis was placed on hypertension, and the rate of uncontrolled hypertension increased to 33%. Staff members at each facility were quickly informed of the higher rate, which prompted redoubled efforts and new strategies to reduce it. Ongoing monitoring will permit an evaluation of these efforts. In summary, hypertension control in diabetic patients can be defined, measured, and monitored to create an evaluation system for interventions that should decrease morbidity from diabetic complications adversely influenced by hypertension [17].
In contrast to hypertension control, levels of metabolic control are relatively difficult to measure, and no widely accepted classification of levels exists. A preliminary categorization scheme for monitoring glycemia, using 1994 data from charts of diabetic patients in Washington, Oregon, and Idaho, is shown in Table 4. In a sample of 568 charts taken from 1647 diabetic patient records, 29 charts contained insufficient data to categorize glycemia; 384 charts recorded HbA1c values from five laboratories using two commercial assays; and 155 charts recorded no HbA1c values but contained fasting or postprandial blood glucose levels, or both, from which the mean of the three most recent values was calculated. The mean blood glucose levels were then used to stratify the patients into categories of glycemic control, which were roughly equivalent to the categories defined using HbA1c values. These blood glucose level categories were constructed by interpolating the range of blood glucose levels equivalent to the HbA1c values in the published graph comparing the HbA1c assay used in the Diabetes Control and Complications Trial with the mean blood glucose levels of study participants [18]. Thirty percent of patients were in the acceptable range (HbA1c less than equals 7.5% or a mean blood glucose level less than equals 9.2 mmol/L), and 9% were in the poorest category (HbA1c more than 12% or a mean blood glucose level more than 18.9 mmol/L). Obviously, there are inherent methodologic issues in the measurement of glycosylated hemoglobin within and between laboratories, and the estimates of equivalent mean blood glucose levels are not precise [19-24]. Efforts to improve and standardize glycosylated hemoglobin measurements are under way and should greatly enhance their utility for both clinical and surveillance purposes. Nevertheless, the classification system presented is a preliminary framework and provides a useful overview.
A simple, reproducible, and affordable system to measure and track levels of glycemia in diabetic patient populations is essential for providing feedback to primary care systems. In Japan, the introduction of a simple categorization system using HbA1c was associated with improved levels of control [25]. Widespread use of such a standard system in the United States would allow investigators to examine variations in metabolic outcomes and could provide important information about the resources needed to achieve improved glycemic control in the primary care setting. Because diabetes educators and nutritionists are not always readily available, categorizing diabetic patients by level of glycemic control in managed care systems would help to identify patients who might benefit most from intensive educational interventions coordinated with appropriate pharmacologic therapy.
Conclusions
These examples show that simple, widely accepted changes in diabetes care can be implemented in a primary care setting if providers understand exactly what to do and have the resources to implement the recommendations. A surveillance system with timely feedback can help bring about desired changes in care practices under these circumstances. Complex systems behaviors are obviously more difficult to change, and this study did not review the extensive literature on the subject [26]. Further research is needed to outline the best ways to change these health systems behaviors to provide preventive care to diabetic patients.
Our experience in the IHS has shown that diabetes care can be improved by defining acceptable practices, measuring their implementation using accurate variables, and supplying feedback to care providers. Whether the resultant improvements in care in the IHS program are because of the effect of the data, the process of measurement itself, or the awareness of being observed (Hawthorne effect) is not known. Nonetheless, the IHS experience suggests that to improve metabolic control and minimize complications in NIDDM, consistent and explicit care practices must be defined for the primary health care setting, and the necessary resources to implement these practices must be made available. In the climate of decreasing health care resources, all primary health care systems such as the IHS must implement cost-effective feedback systems to monitor care practices, intermediate clinical variables, and, ultimately, long-term outcomes. The experience in the challenging setting of the IHS should provide encouragement and impetus for further research and surveillance of diabetes care and outcomes in other primary health care facilities.
Dr. Rith-Najarian: Bemidji Area Indian Health Service, 100 Federal Building, Bemidji, MN 56601.
Dr. Acton: Billings Area Diabetes Program, Tribal Health and Human Services, P.O. Box 280, St. Ignatius, MT 59865.
Dr. Shields: Portland Area Diabetes Program, 104 W. Magnolia, Room 306, Bellingham, WA 98225-8942.
- Copyright ©2004 by the American College of Physicians
