CLINICAL DIABETES
VOL. 14 NO. 3 MAY/JUNE 1996

These pages are best viewed with Netscape version 3.0 or higher or Internet Explorer version 3.0 or higher. When viewed with other browsers, some characters or attributes may not be rendered correctly.

Commentary

Treatment of NIDDM by Endocrinologists
and Generalists: A Critique of the
Medical Outcomes Study Report

Matthew C. Riddle, MD and Diane M. Karl, MD

A recent article in the Journal of the American Medical Association (JAMA)¹ addresses the factors affecting medical outcomes of patients with diabetes and hypertension. Because it deals with two important disorders and is part of the Medical Outcomes Study, a large ongoing analysis of medical practice, the article has attracted considerable attention.

The authors started from the premise that fee-for-service practice and specialist physicians use more resources than managed systems and generalists.² They asked whether long-term medical outcomes differ between systems, and between generalists and specialists, independent of short-term costs. Their general conclusion was that “although prepaid medicine relies more heavily on generalist physicians than does fee-for-service, there is no evidence from these analyses that the quality of care of moderately ill patients with these two common diseases was adversely affected.” A sympathetic editorial in the same issue endorsed this view, stating that “despite some limitations, this study does suggest that a diabetic or hypertensive patient is apt to fare equally well under the care of a competent generalist or specialist or in a managed care or fee-for-service setting.”³

Readers are likely to conclude the study proves generalists treat such patients as well as and less expensively than specialists, both initially and in the long term. This message seems to justify efforts to prevent patients with chronic disorders like diabetes and hypertension from seeing physicians with special training in their management.

The implications of these views are important, so the data on which they are based deserve close inspection. This critique focuses on one part of the study, that which compares patients of endocrinologists and generalists. At least for this aspect, the authors’ conclusions are not well supported.

This was a very complex study, and only part of the methods and findings were presented in the published report. Data were collected from many offices and practice systems in three cities. A total of 1,425 patients were identified as having non-insulin-dependent diabetes mellitus (NIDDM). Three kinds of outcomes for subgroups of these patients were examined: 1) mortality over a 5-year period; 2) patient-reported functional status after 4 years; and 3) historical, physical, and laboratory findings at entry and after 2 years.

Different numbers of patients were used for each kind of assessment. In each case there were far fewer patients than in the group originally identified. Patients were excluded for various reasons, including chronic illness, recent acute illness, primary care from other kinds of providers, and lack of cooperation, insurance, or follow up. After this attrition, data were available for mortality analysis from 30% of the original group (424 of 1,425 patients), for functional status from 22% (317 of 1,425), and for exams by the study personnel for 12% (167 of 1,425).

Mortality figures were not given in detail. The authors predicted from baseline demographic and clinical data that the mortality rate for family physicians’ patients should be just 27% of that for patients of internists, and 17% of that for patients of endocrinologists. This projection argues that the endocrinologists cared for more complex patients than the generalists, especially the family physicians. Correction for differences at baseline is theoretically possible, but the authors concluded that, with small numbers of patients and a modest mortality rate (about 3% yearly), definitive analysis was not feasible.

The functional outcomes were also not clearly defined. The findings for the group of 317 patients followed for 4 years were not given separately. Instead, a table presenting data for the 167 patients who were studied more intensively included values for physical function, role function, general health perceptions, and energy level. These showed no differences between patients of general internists, family physicians, or endocrinologists, and none between baseline and follow up.

The lack of differences was interpreted as reflecting equivalent effectiveness of care by different kinds of physicians. However, assessment of wellness is difficult. The sameness of these results could, alternatively, reflect insensitivity of the methods used for concerns specific to diabetes, and in a population from which the most healthy and most ill were excluded.

The firmest data were those directly collected from 167 patients on two occasions, 2 years apart. The patients’ average age was about 60 years, and half were female. Their mean total glycosylated hemoglobin at the first examination was about 9.5%, which is equivalent to 8.3% HbA1c. One hundred and nine of these subjects were patients of general internists, 27 were patients of family physicians, and 31 were patients of endocrinologists.

There was no large between-group difference in age, gender, glycemic control at baseline, or prevalence of retinopathy. The groups did differ at baseline in other respects. Forty-eight percent of the patients of endocrinologists used insulin, compared to 33% and 22% of the patients of internists and family physicians. Thirteen percent of the endocrinologists’ patients had foot ulcers, compared to 3% and 7% of the patients of internists and family physicians. Seventeen percent of the endocrinologists’ patients had foot infections, compared to 9% and 0% of the patients of internists and family physicians. As mentioned above, the endocrinologists’ patients had higher predicted mortality rates, presumably related to clinical and laboratory abnormalities at baseline that were not included in this article.

In the 2 years of follow up, the patients were seen about monthly by all three groups of physicians. Three objective measures were compared between baseline and the 2-year follow up: glycosylated hemoglobin, foot ulcers, and foot infections. Of these measures, the only one that showed changes that were statistically different between groups was the frequency of foot ulcers, which improved among the patients of endocrinologists.

Several limitations of the study are addressed in the authors’ discussion and the accompanying editorial. These include the selection process that excluded most of the potential subjects and the short follow-up interval. However, the two most crucial problems require more comment. These are the differences between groups at baseline and the small numbers of patients studied.

Consider the baseline differences between the groups of patients first. Patients do not go to endocrinologists randomly. Most often, someone believes the patient faces a greater or different challenge than the average patient with diabetes, and thus may benefit from seeing a specialist. This selection of more complex patients into the practices of endocrinologists is evident in the study. The endocrinologists’ patients needed insulin more often, had more foot complications, and were predicted from baseline findings to have a higher mortality rate.

Patients like this are likely to have worse outcomes and higher medical expenses than well people, no matter what physicians they see. Unless patients are either randomized to different providers, or groups of patients are stratified by complexity and severity of illness, meaningful comparison of outcome is impossible. The authors of the JAMA article have discussed this issue previously.⁴ Even so, the lack of effective stratification in this study undermines the analysis.

Interpretation of the study is further limited by the small numbers of subjects. Only 167 subjects were directly examined, and of these, only 31 were patients of endocrinologists. From a simplistic viewpoint, a sample of 31 seems entirely too small to generate evidence for a generalized conclusion that specialists offer no added value. From a statistical viewpoint, the power of a given design to answer questions can be accurately defined for either observational or interventional studies. It is routine to do a power calculation to determine the size of groups necessary to show, with reasonable likelihood, that clinically important differences between groups are not present. No such calculation was done for the clinical measures of outcome in this study.

One result of this decision is that mortality rates could not be compared. Another is the dilemma posed by the glycosylated hemoglobin findings. The patients of family physicians and endocrinologists, whose values were somewhat lower than those of the general internists’ patients initially, showed 0.6% and 1.1% increases over 2 years, respectively. The Wisconsin Study of Diabetic Retinopathy5 argues these changes may be clinically significant. But with only 29 and 31 patients in the two groups, they cannot be evaluated statistically. Thus, the design of the study precludes use of two of the more important increases of outcome.

Examination of the findings presented in the JAMA article supports neither the authors’ conclusions nor their implied relevance to health policy. The mortality data are not interpretable. The self-reported functional status data are incompletely presented and not yet validated for this kind of population. The clinical data suggest the endocrinologists’ patients were less healthy at baseline and, unlike those of generalists, had improved at follow up. Power calculations were not done for mortality or glycosylated hemoglobin, which are the most concrete measures of outcome. If they were, they would likely show that a sample of 31 patients of endocrinologists is too small to support negative conclusions.

What can be made of all this? This study does provide useful information on important matters. However, it should be considered a limited and preliminary survey of medical experience. It is not a definitive study that justifies any specific change of tactics or policy.

Perhaps its clearest finding is that, even with ongoing medical care by experienced physicians, glycosylated hemoglobin values are unacceptably high. More than half the subjects had baseline glycosylated hemoglobin values above the 8% A_1c level that demands corrective action, according to American Diabetes Association standards.⁵ Poor glycemic control predicts both mortality and severe disabilities.^5,7,8 Neither the human costs nor the economic costs of our presently ineffective management of NIDDM should be allowed to continue.

But how can we do better? Unfortunately, common sense and objectivity are at risk amid the stress of health-care reform. We are distracted by theoretical and political arguments over whether generalists or endocrinologists are the right physicians to care for patients with diabetes. These arguments will not improve treatment of NIDDM.

From this viewpoint, the JAMA article poses the wrong question. A better one would be: how can we best deploy the resources we have? Specifically, how can we match different categories of diabetic patients with the medical resources they need? Which patients can be treated cost-effectively by generalists, which respond better to individualized problem-solving by endocrinologists, and how can dietitians, educators, and other professionals contribute most efficiently? Studies of these practical issues, perhaps including future analyses from the Medical Outcomes Study, should lead to more helpful conclusions.

REFERENCES

 ¹Greenfield S, Rogers W, Mangotich M, Carney MF, Tarlov AR: Outcomes of patients with hypertension and non-insulin-dependent diabetes mellitus treated by different systems and specialities. Results from the Medical Outcomes Study. JAMA 274: 1436-44, 1995.

²Greenfield S, Nelson EC, Zubkoff M, Manning W, Rogers W, Kravitz RL, Keller A, Tarlov, Ware JE: Variations in resource utilization among medical specialties and systems of care. Results from the Medical Outcomes Study. JAMA 267: 1624-30, 1992.

³Fihn SD: Physician specialty, systems of health care, and patient outcomes. JAMA 274: 1473-74, 1995.

⁴Kravitz RL, Greenfield S, Rogers W, Manning WG, Zubkoff M, Nelson EC, Tarlov AR, Ware JE: Differences in the mix of patients among medical specialties and systems of care. Results from the Medical Outcomes Study. JAMA 267; 1617-23, 1992.

⁵Klein R: Hyperglycemia and microvascular and macrovascular disease in diabetes. Diabetes Care 18: 258-68, 1995.

⁶American Diabetes Association: Standards of care for patients with diabetes mellitus.Diabetes Care 17: 616-23, 1994.

⁷Kuusisto J, Mykkanen L, Pyorala K, Laakso M: NIDDM and its metabolic control predict coronary heart disease in elderly subjects. Diabetes 43: 960-67, 1994.

⁸Andersson DKG, Svardsudd K: Long-term glycemic control relates to mortality in Type 2 diabetes. Diabetes Care 18: 1534-43, 1995.

Matthew C. Riddle, MD, is head of the Section of Diabetes at the Oregon Health Sciences University, Portland, and Diane M. Karl, MD, is co-director of the Diabetes Treatment Center at Providence Medical Center, Portland, Oreg.

 Note of Disclosure: Dr. Riddle serves on an advisory board for Bayer Pharmaceuticals, a subsidiary of which (Miles Diagnostics) has developed an instrument for measuring hemoglobin A1c levels.

Return To American Diabetes Association Home Page

Copyright © 1996 American Diabetes Association

Last updated: 6/3/97
For ADA Related Issues contact CustomerService@diabetes.org

For Technical Issues contact webmaster@diabetes.org