Diabetes autoantibodies do not predict progression to diabetes in adults: the Diabetes Prevention Program.

AIMS: To determine if the presence of diabetes autoantibodies predicts the development of diabetes among participants in the Diabetes Prevention Program. METHODS: A total of 3050 participants were randomized into three treatment groups: intensive lifestyle intervention, metformin and placebo. Glutamic acid decarboxylase (GAD) 65 autoantibodies and insulinoma-associated-2 autoantibodies were measured at baseline and participants were followed for 3.2 years for the development of diabetes. RESULTS: The overall prevalence of GAD autoantibodies was 4.0%, and it varied across racial/ethnic groups from 2.4% among Asian-Pacific Islanders to 7.0% among non-Hispanic black people. There were no significant differences in BMI or metabolic variables (glucose, insulin, HbA(1c), estimated insulin resistance, corrected insulin response) stratified by baseline GAD antibody status. GAD autoantibody positivity did not predict diabetes overall (adjusted hazard ratio 0.98; 95% CI 0.56-1.73) or in any of the three treatment groups. Insulinoma-associated-2 autoantibodies were positive in only one participant (0.033%). CONCLUSIONS: These data suggest that 'diabetes autoimmunity', as reflected by GAD antibodies and insulinoma-associated-2 autoantibodies, in middle-aged individuals at risk for diabetes is not a clinically relevant risk factor for progression to diabetes.

Long-term effects of the Diabetes Prevention Program interventions on cardiovascular risk factors: a report from the DPP Outcomes Study.

AIMS: Whether long-term cardiovascular risk is reduced by the Diabetes Prevention Program interventions is unknown. The aim of this study was to determine the long-term differences in cardiovascular disease risk factors and the use of lipid and blood pressure medications by the original Diabetes Prevention Program intervention group. METHODS: This long-term follow-up (median 10 years, interquartile range 9.0-10.5) of the three-arm Diabetes Prevention Program randomized controlled clinical trial (metformin, intensive lifestyle and placebo), performed on 2766 (88%) of the Diabetes Prevention Program participants (who originally had impaired glucose tolerance), comprised a mean of 3.2 years of randomized treatment, approximately 1-year transition (during which all participants were offered intensive lifestyle intervention) and 5 years follow-up (Diabetes Prevention Program Outcomes Study). During the study, participants were followed in their original groups with their clinical care being provided by practitioners outside the research setting. The study determined lipoprotein profiles and blood pressure and medication use annually. RESULTS: After 10 years' follow-up from Diabetes Prevention Program baseline, major reductions were seen for systolic (-2 to -3) and diastolic (-6 to -6.5 mmHg) blood pressure, and for LDL cholesterol (-0.51 to -0.6 mmol/l) and triglycerides (-0.23 to -0.25 mmol/l) in all groups, with no between-group differences. HDL cholesterol also rose significantly (0.14 to 0.15 mmol/l) in all groups. Lipid (P = 0.01) and blood pressure (P = 0.09) medication use, however, were lower for the lifestyle group during the Diabetes Prevention Program Outcomes Study. CONCLUSION: Overall, intensive lifestyle intervention achieved, with less medication, a comparable long-term effect on cardiovascular disease risk factors, to that seen in the metformin and placebo groups.

Plasma glucagon and insulin in rat pregnancy. Roles in glucose homeostasis.

To determine if pancreatic glucoregulatory hormones can be implicated in the glucose fall of pregnancy, we have measured plasma immunoreactive insulin and glucagon (IRI and IRG) in rats. Fed rats in midgestation show a rise in IRI without a corresponding increase in IRG. In late gestation, IRG rises significantly, but only enough to keep pace with a further rise in IRI. On a molar basis, IRI remains the predominant hormone despite a marked fall in blood glucose. After a 48-h fast IRI falls to comparably low levels in pregnant and virgin rats. A small rise in IRG is seen in virgin but not in pregnant rats despite frank hypoglycemia in the latter. Thus, IRG secretion in pregnancy is diminished relative to IRI in the fed state and fails to increase in the fasted state despite the stimulus of a lower glucose in both instances. To evaluate IRG secretory reserve, the IRG response to i.v. alanine was assessed in late gestation. In fed rats a greater IRG increase is seen in pregnancy; after fasting no difference is seen between pregnant and virgin rats. These results preclude an absolute deficiency in glucagon secretion. Pancreas hormone stores were alos measured in an effort to explain the altered secretory state. We find reciprocal changes in IRI and IRG content favoring IRG in midgestation and IRI in late gestation. Thus, pancreas hormone storage is altered in pregnancy but does not account for the changes in hormone secretion. Rather, pregnancy exerts an effect on the islet secretory process itself. Release of IRI is enhanced relative to IRG regardless of the blood sugar level. These observations suggest that in the pregnant rat circulating levels of insulin and glucagon may act to limit hepatic glucose output. Available evidence from the literature supports the concept of restrained glucose production. It is proposed that a lower blood glucose production. It is proposed that a lower blood glucose in rat pregnancy may be a lesser liability teleologically than would be the obligate nitrogen wasting which accompanies gluconeogenesis.

Abnormal lipoprotein lipase in familial exogenous hypertriglyceridemia.

A 5-yr old male proband and his sister have had hypertriglyceridemia and hepatosplenomegaly since birth. When studied on a metabolic ward, they demonstrated rapid decreases in serum triglycerides on 3 g fat/day diets. Oral glucose tolerance tests were normal. Postheparin lipolytic activity (PHLA) against chylomicrons was virtually absent in both children whereas the mother and a normolipemic sister had levels approximately 50% normal. However, all four had a normal PHLA against commercial triglyceride emulsion (Intralipid). Two unrelated children from different kindreds of typical type 1 hyperlipoproteinemia and two patients with acquired type V hyperlipoproteinemia had deficient PHLA against both substrates. No inhibitors of PHLA could be demonstrated in the proband's plasma, and his own PHLA could not be enhanced by either normal concentrated plasma or pooled d > 1.063 lipoprotein fraction. The proband's postheparin plasma required almost 20 times the normal chylomicron-triglyceride concentration to reach one-half maximal lipase velocity. Both affected siblings showed heavy pre-beta lipoprotein electrophoretic bands plus chylomicrons in their fasting plasmas while ingesting a 33% carbohydrate, 30% fat diet. Incubation of their postheparin plasma with S(f) > 400 chylomicrons in vitro produced a smaller S(f) 20-400 "remnant" with pre-beta electrophoretic mobility that was not seen under the same conditions when normal postheparin plasma was used. Postheparin monoglyceridase and phospholipase activities were either normal or only moderately decreased when determined with appropriate artificial substrates. These data are consistent with either (a) a mutant gene producing a lipoprotein lipase with unusual substrate specificities or (b) an absolute deficiency of normal lipoprotein lipase with a compensatory increase in some other postheparin triglyceridase.

Human intestinal lipoproteins. Studies in chyluric subjects.

To explore the role of the human intestine as a source of apolipoproteins, we have studied intestinal lipoproteins and apoprotein secretion in two subjects with chyluria (mesenteric lymphatic-urinary fistulae). After oral corn oil, apolipoprotein A-I (apoA-I) and apolipoprotein A-II (apoA-II) output in urine increased in parallel to urinary triglyceride. One subject, on two occasions, after 40 g of corn oil, excreted 8.4 and 8.6 g of triglyceride together with 196 and 199 mg apoA-I and on one occasion, 56 mg apoA-II. The other subject, after 40 g corn oil, excreted 0.3 g triglyceride and 17.5 mg apoA-I, and, after 100 g of corn oil, excreted 44.8 mg apoA-I and 5.8 mg apoA-II. 14.5+/-2.1% of apoA-I and 17.7+/-4.3% of apoA-II in chylous urine was in the d < 1.006 fraction (chylomicrons and very low density lipoprotein). Calculations based on the amount of apoA-I and apoA-II excreted on triglyceride-rich lipoproteins revealed that for these lipid loads, intestinal secretion could account for 50 and 33% of the calculated daily synthetic rate of apoA-I and apoA-II, respectively. Similarly, subject 2 excreted 48-70% and 14% of the calculated daily synthetic rate of apoA-I and apoA-II, respectively. Chylous urine contained chylomicrons, very low density lipoproteins and high density lipoproteins, all of which contained apoA-I. Chylomicrons and very low density lipoproteins contained a previously unreported human apoprotein of 46,000 mol wt. We have called this apoprotein apoA-IV because of the similarity of its molecular weight and amino acid composition to rat apoA-IV. In sodium dodecyl sulfate gels, chylomicron apoproteins consisted of apoB 3.4+/-0.7%, apoA-IV 10.0+/-3.3%, apoE 4.4+/-0.3%, apoA-I 15.0+/-1.8%, and apoC and apoA-II 43.3+/-11.3%. Very low density lipoprotein contained more apoB and apoA-IV and less apoC than chylomicrons. Ouchterlony immunodiffusion of chylomicron apoproteins revealed the presence of apoC-I, apoC-II, and apoC-III. In contrast, plasma chylomicrons isolated during a nonchyluric phase revealed a markedly altered chylomicron apoprotein pattern when compared with urinary chylomicrons. The major apoproteins in plasma chylomicrons were apoB, apoE, and the C peptides: no apoA-I or apoA-IV were present in sodium dodecyl sulfate gels indicating that major changes in chylomicron apoproteins occur during chylomicron metabolism. When incubated in vitro with plasma, urinary chylomicrons lost apoA-I and apoA-IV and gained apoE and apoC. Loss of apoA-I and apoA-IV was dependent upon the concentration of high density lipoproteins in the incubation mixture. These studies demonstrate that the human intestine secretes significant amounts of apoA-I and apoA-II during lipid absorption. Subsequent transfer of apoproteins from triglyceride-rich lipoproteins to other plasma lipoproteins may represent a mechanism whereby the intestine contributes to plasma apoprotein levels.

Stable insulin preparation for implanted insulin pumps. Laboratory and animal trials.

The stability and longevity of the polyethylene-polypropylene glycol-stabilized insulin have been tested in vitro and in vivo in an implanted insulin-infusion device, the programmable implantable medication system (PIMS). In vitro tests demonstrated long-term compatibility with refill cycles of up to 3 mo, with a preparation of 400 U/ml. Total test period in vitro was 3.2 device-yr (combined time of device use). Insulin retained 88-93% native structure. A major modification, which was biologically active and nonimmunogenic, was identified and partially characterized. Examination of one device by scanning electron microscopy and X-ray microanalysis after 1 yr of insulin infusion revealed surfaces clean of insulin precipitate or other material along the entire insulin-delivery pathway. Surface analysis of the silicone-lined polyethylene catheters after 6 mo of infusion also showed no evidence of major insulin precipitate. In vivo stability trials were accomplished with PIMS implanted in diabetic dogs with an intraperitoneal delivery site. There has been no insulin blockage of the catheter of active pumps after 5.1 dog-yr (combined time of trials) of trials (up to 5 mo between refills in a single dog). Structural stability of insulin was analyzed by high-performance liquid chromatography. On average, 90.8% of the insulin sampled from the reservoir in vivo was native insulin, with an average of 96.2% retention of active forms.

Cholesterol metabolism in diabetes. I. The effect of diabetic control on sterol balance.

To evaluate the effect of diabetic control on sterol balance, five subjects (one with fasting chylomicronemia) were studied when they had little control of blood glucose (period I) and, continuously, when increased insulin dosage improved diabetic control (period II). The four subjects with no chylomicronemia showed a diminished fecal bile acid (FBA) excretion. Although fecal neutral steroid excretion increased variably, the total steroid balance was unchanged. The results support the concept that good control of diabetes shifts the fecal steroid excretion towards diminished FBA, but steroid balance methods did not reveal an effect of insulin on total steroid balance.

Why intraperitoneal delivery of insulin with implantable pumps in NIDDM?

In the normal state, pancreatic secretion of insulin results in a portal/peripheral gradient with the highest concentrations of insulin in the liver. In diabetic patients with absent or insufficient pancreatic insulin secretion who require exogenous insulin, this normal gradient is lost, resulting in numerous abnormalities. This consideration led to interest in the intraperitoneal delivery of insulin, hoping to produce a therapeutic state more closely resembling normal physiology. The development of implantable insulin pumps, which can deliver insulin intraperitoneally, led to numerous studies on insulin-dependent diabetes mellitus (IDDM) patients, demonstrating that insulin delivered intraperitoneally is rapidly and predictably absorbed with most of it going into the portal system, resulting in hepatic delivery of insulin. Studies in IDDM patients have demonstrated that good glucose control can be achieved with intraperitoneal delivery of insulin from implantable pumps with lesser glycemic fluctuations and, therefore, fewer episodes of hypoglycemia. Furthermore, intraperitoneal insulin results in carbohydrate and particularly lipid metabolism that more closely mimics the normal physiological state than produced by injections of insulin. Thus, implantable insulin pumps are being studied for use in IDDM. Many non-insulin-dependent diabetes mellitus (NIDDM) patients have insufficient pancreatic secretion and require exogenous insulin. Because of alterations in hepatic sensitivity to insulin, increments in insulin delivery to the liver may be even more important in NIDDM than IDDM. Furthermore, insulin resistance, which is an integral part of NIDDM, results in higher physiological levels of insulin, which are required for glucose control, and thus significant peripheral hyperinsulinemia occurs in patients receiving exogenous insulin.(ABSTRACT TRUNCATED AT 250 WORDS)

Correlation of serum triglyceride levels and hemoglobin AIc concentrations in diabetes mellitus.

Studies in 10 nonketotic diabetic subjects (five juvenile- and five adult-onset) before and after control of carbohydrate metabolism showed a high degree of correlation between hemoglobin AIc (HbAIc) concentrations and serum triglyceride levels. Serum triglyceride levels were found to correlate more closely with Hb AIc (r = 0.91, p less than 0.001) than did serum cholesterol (r = 0.47, p greater than 0.05), thus indicating a more direct relationship to carbohydrate metabolism.

Hyperphagia alters cholesterol dynamics in diabetic rats.

Rats with streptozotocin-induced diabetes stop growing and start eating more chow. These two events elicit an interacting series of changes in cholesterol dynamics. Hyperphagia increases dietary cholesterol intake and cholesterol synthesis by the small intestine. These increases are balanced by a decrease in cholesterol synthesis in the rest of the body so that total cholesterol input is normal. With growth failure, utilization of cholesterol for formation of new tissue ceases. This decrease is balanced by an increase in bile acid synthesis by the liver. The bile acid pool in the contents of the small intestine is enlarged by hyperphagia. Despite these changes, fecal sterol excretion and total utilization of cholesterol are normal. During the course of changes in growth and food intake and the attendant changes in cholesterol flux, the total tissue cholesterol pool does not change. Therefore influx equals efflux and the systems regulating cholesterol and bile acid synthesis are responding appropriately and are themselves unperturbed by insulin deficiency. However, plasma cholesterol level increases threefold. This elevation is due to increased influx of cholesterol from the small intestine and decreased synthesis in the rest of the body, so that a larger portion of total body cholesterol influx passes through the blood.

Salicylate-induced hypoglycemia and ketoacidosis in a nondiabetic adult.

A 78-year-old nondiabetic woman was admitted to the hospital with salicylate-induced hypoglycemia. Ketosis was present with a moderate metabolic acidosis and primary respiratory alkalosis. The patient's mental status improved immediately following intravenous administration of glucose. The case illustrates salicylate's hypoglycemic activity, and that the metabolic acidosis may have exacerbated symptoms of cerebral glucopenia.

Assessment of glycemia in diabetes mellitus: hemoglobin A1c.

The monitoring of glycemia is an essential component of diabetes care. It may be divided into self-monitoring of blood glucose (SMBG), which measures the immediate level of glycemia, and measurement of hemoglobin A1c (HbA1c), which reflects longer-term glycemia. SMBG was discussed in an earlier review. HbA1c is a measure of erythrocyte hemoglobin glycation, and since erythrocytes have about a 120 day life span, HbA1c reflects mean glycemia for the previous 3 months (weighted to the most recent month). There are several conditions that confound the HbA1c measurement such as hemolytic anaemia (lowers HbA1c) or aplastic anaemia (raises it), but in most circumstances HbA1c is a valid index of glycemia. The recommendation is to measure HbA1c every 3-6 months, and treat to a target level of < 7%. If these recommendations were successfully followed in most people with diabetes, long-term complications, especially microvascular complications, would be markedly reduced.

Assessment of glycosylated hemoglobin measurement with sample collection papers.

It would be advantageous to be able to measure glycosylated hemoglobin (GHb) from a sample of blood dried on paper and sent to the physician by the patient. Indeed, such a technique has been introduced by Isolab (Akron, OH). We tested the validity of the method to determine whether applying the blood to paper and immediately assaying it correlates with the same sample assayed by the usual whole-blood technique, the size of sample applied to paper affects apparent GHb results, time and temperature of storage of the sample paper affect apparent GHb results, and plasma glucose concentration of the sample affects GHb results. The GHb of samples assayed immediately after application to sample paper versus those assayed as whole blood showed very good correlation (r = .93, P less than .001). Volume of the drop of blood applied to paper (25-100 microliters) did not affect results. However, there was a dramatic, temperature-dependent increase in apparent GHb when samples were stored on sample paper, averaging 12.5, 16.3, and 19.5% when stored at room temperature for 3, 6, and 9 days, respectively. Overall, apparent GHb rose from 1.3-fold in 3 days at 4 degrees C to 3.8-fold in 9 days at 37 degrees C. The rate of GHb formation was proportional to plasma glucose concentration, but removal of free glucose by ethanol or glucose oxidase did not yield consistent results for this method. We conclude that these sample papers are not useful as an approach to collecting blood samples for GHb measurement.

Normalization of composition of triglyceride-rich lipoprotein subfractions in diabetic subjects during insulin infusion with programmable implantable medication system.

OBJECTIVE: To investigate whether long-term improved glycemic control by intraperitoneal insulin infusion normalizes the compositional abnormalities of triglyceride (TG)-rich lipoproteins in insulin-dependent diabetes mellitus (IDDM). RESEARCH DESIGN AND METHODS: Seven subjects were studied before and 12-14 mo after initiation of treatment with the programmable implantable medication system (PIMS). Plasma TG levels were measured, and the composition of three TG-rich lipoprotein subfractions (Svedberg flotation [Sf] greater than 400, 100-400, and 20-100) were analyzed before and every 1.5 for 7.5 h after ingestion of corn oil. RESULTS: PIMS significantly improved glycemic control, as measured by mean blood glucose (P less than 0.02), and HbA1 (P less than 0.001, paired t test) levels. Weight loss was also observed during PIMS treatment. Significant changes occurred in the composition of TG-rich lipoprotein subfractions during PIMS treatment in both the fasting (P less than 0.002) and the postprandial (P less than 0.0001) state. Most changes were in the direction of nondiabetic values. PIMS treatment reduced the total cholesterol enrichment in IDDM subjects in all three subfractions in the postprandial state and the very-low-density lipoprotein subfractions (Sf 100-400 and 20-100) in the fasting state. Multivariate analysis showed that the compositional changes were affected by improved glycemic control, as assessed by both mean blood glucose and HbA1, whereas the very-low-density lipoprotein compositional changes were by both the improved glycemic control and body weight. CONCLUSIONS: In IDDM subjects during PIMS treatment, there was normalization of most abnormalities in the composition of fasting and postprandial TG-rich lipoproteins, including enrichment in total cholesterol, which is considered atherogenic.

Intraperitoneal insulin delivery decreases the levels of chylomicron remnants in patients with IDDM.

OBJECTIVE: To investigate whether intraperitoneal insulin (IPII) decreases the levels of circulating chylomicron remnants in patients with insulin-dependent diabetes mellitus (IDDM). RESEARCH DESIGN AND METHODS: Eight nonobese, normolipidemic IDDM patients were studied twice: before (while on subcutaneous insulin) and 6 months after initiation of IPII by a programmable implanted medication system. Fasting and mean blood glucose, HbA1, and lipid values were determined. Blood samples were also drawn before and every 2 h for 10 h after ingestion of a fat meal (corn oil + Vitamin A). Triglycerides (TGs), apolipoprotein B (apoB), and retinyl esters were determined over time in two TG-rich lipoprotein subfractions (Sf > 100 and Sf20-100) isolated from plasma by density-gradient ultracentrifugation. RESULTS: IPII slightly decreased the mean blood glucose from 7.8 +/- 1.1 to 7.4 +/- 1.1 mmol/l (mean +/- SD, P = 0.027, paired Student's t test) and the HbA1 from 9.4 +/- 1.5 to 8.7 +/- 1.2 (NS). TG and apoB levels in postprandial Sf > 100 and Sf20-100 were not changed by IPII. On IPII, however, retinyl ester levels in Sf > 100 decreased (P = 0.05, analysis of variance [ANOVA]) and tended to be lower in Sf20-100 (P = 0.075). In addition, following IPII, the retinyl ester:apoB ratio was lower in Sf > 100 (P = 0.0002) and marginally lower (P = 0.06) in Sf 20-100. CONCLUSIONS: IPII decreased chylomicron remnant levels, which might decrease the atherosclerotic risk in IDDM. Since glycemic control was only slightly improved, the effect was most likely due to the intraperitoneal route of delivery.

Effect of insulin therapy on macrovascular risk factors in type 2 diabetes.

Many patients with type 2 diabetes require insulin therapy for improved glycemic control after beta-cell failure. However, many physicians are reluctant to institute insulin therapy in type 2 diabetes for fear of accelerating atherosclerosis. The epidemiological evidence is reasonably sound that hyperinsulinism correlates with increased cardiovascular disease in nondiabetic people and those with early type 2 diabetes. It is much less clear, however, that insulin concentration plays a negative role when less well controlled diabetes is considered. The data are more consistent, in fact, with the glucose hypothesis, i.e., that hyperglycemia is a risk factor, although the magnitude of the glucose effect is not well defined. Certainly, the dysmetabolism associated with poor glycemic control could increase the risk of macrovascular events through well-known mechanisms. There is direct evidence that insulin therapy can reduce the risk of macrovascular events by improving glycemic control and diabetes-associated dyslipidemias, although the beneficial effects may be significantly compromised by excessive weight gain. Insulin therapy does not appear to induce hypertension independent of changes in body weight. It is concluded that optimal glycemic control confers a known benefit and can only be achieved with insulin therapy in some people with type 2 diabetes. In these circumstances, the use of insulin has a net benefit on cardiovascular risk, mediated primarily through improvement in dyslipidemia and glycemia itself.

Effect of diabetes education on self-care, metabolic control, and emotional well-being.

Participants (n=165) entering a week-long outpatient education program completed a protocol measuring self-care patterns, glycosylated hemoglobin levels, and emotional well-being. Emotional well-being was reassessed at the end of the program, and the entire protocol was completed again at 6 mo (n=124). At the program's end, participants improved on all measures of emotional well-being (P less than .01). Self-esteem and diabetes self-efficacy rose, whereas anxiety and depression fell. At 6 mo, improvement in emotional well-being continued, and important self-care behaviors improved from preprogram levels. Self-monitoring of blood glucose and exercise rose (both P less than .001), and bringing (P less than .01) and glycosylated hemoglobin levels (P less than .001) fell. Program effects were unrelated to demographic or disease characteristics but strongly related to initial status. Participants who entered the program with high levels of emotional well-being or good self-care patterns or glycemic control tended to change little, if at all, at later measurements. On the other hand, people who entered the program with low levels of emotional well-being or with poor self-care patterns or glycemic control improved substantially. Our findings suggest that diabetes education can promote long-term benefits in self-care, metabolic control, and emotional status if the program is specifically designed to provide these benefits. Aspects of the program that contribute to its efficacy are discussed.

Differential effect of diabetes education on self-regulation and life-style behaviors.

OBJECTIVE: To examine the effect of diabetes education on self-regulation and life-style behaviors. RESEARCH DESIGN AND METHODS: Participants in an outpatient diabetes education program completed a protocol measuring several self-care behaviors and glycemic control at entry (n = 165) and 6 (n = 124) and 12 (n = 89) mo after the program. RESULTS: Improvement was noted at 6 mo for most self-care behaviors and glycemic control. At 12 mo, lower glycosylated hemoglobin levels were maintained (P less than 0.001) without increases in perceived hypoglycemia. Improvement was not maintained for those self-care behaviors that require change in life-style, i.e., diet and exercise. However, self-care behaviors that allow patients to self-regulate their glycemic control--self-monitoring of blood glucose and insulin dose self-adjustment--were improved at 12 mo over preprogram levels (P less than 0.001). Frequency of insulin self-adjustment continued to increase during the period between follow-ups. CONCLUSIONS: The findings suggest that diabetes education is effective in promoting self-regulation behaviors, although it has less effect on traditional regimen behaviors such as diet and exercise.

Patterns of expenditures and use of services among older adults with diabetes. Implications for the transition to capitated managed care.

OBJECTIVE: To examine health care use and expenditures among older adults with diabetes, investigate factors that are associated with higher expenditures, and describe the policy implications of caring for this population under managed care. RESEARCH DESIGN AND METHODS: A cross-sectional analysis of expenditures for individuals with diabetes over age 65 years from a nationwide 5% random sample of Medicare beneficiaries was conducted during 1992. All components of medical care covered under Medicare were examined. Multivariate analysis was used to assess the contribution of age, race, sex, number of diabetic complications, and comorbidity (Charlson Index) on total expenditures. RESULTS: On average, individuals with diabetes (n = 188,470) were 1.5 times (P < 0.0001) as expensive as all Medicare beneficiaries (n = 1,371,960). However, there were wide variations, with the most expensive 10% of beneficiaries with diabetes accounting for 56% of expenditures for individuals with diabetes and the least expensive 50% accounting for 4%. Acute care hospitalizations accounted for the majority (60%) of total expenditures, whereas outpatient and physician services accounted for 7 and 33%, respectively. There were no differences in the number of complications for all older adults with diabetes compared with those with the highest expenditures. However, the average number of hospitalizations was 1.6 times (0.53 vs. 0.34; P < 0.0001) higher, and the average length of stay was 2 days longer, among older adults with diabetes (P < 0.0001). In the regression model, age and male sex (factors currently used to set payment rates for Medicare managed care enrollees), and number of diabetic complications, but not race, were positively related to expenditures, yet had minimal predictive power (R2 = 0.0006). The addition of the Charlson Index, also positively related to expenditures, was able to explain up to 20% of the variation in total expenditures (R2 = 0.196). CONCLUSIONS: There are large variations in expenditures among older adults with diabetes. Because elderly beneficiaries with diabetes are more expensive than the average older adult, current Medicare capitation rates may be inadequate. To avoid selection bias and under-treatment of this vulnerable population under managed care, methods to construct fair payment rates and safeguard quality of care are desirable.

Predicting expenditures for Medicare beneficiaries with diabetes. A prospective cohort study from 1994 to 1996.

OBJECTIVE: To describe health care expenditures and utilization patterns among older adults with diabetes and to examine factors associated with expenditures over a 3-year period. RESEARCH DESIGN AND METHODS: We conducted a prospective cohort study of health care expenditures and utilization by diabetic patients from a random nationwide sample of aged Medicare beneficiaries from 1994 to 1996. All services covered by the Medicare program were examined. Multivariate regression was used to assess the contribution of patient characteristics in 1994 on Part B, inpatient, and total expenditures in 1995 and 1996. RESULTS: Per capita expenditures for beneficiaries with diabetes (n = 169,613) were 1.7 times greater than those for those beneficiaries without diabetes (n = 968,832) in 1994. This ratio remained fairly constant over the 2 years of follow-up. Expenditures for beneficiaries with diabetes were highly skewed. However, few of these individuals remained in the highest expenditure quintile over the 2 years of follow-up. Using multiple regression analysis to adjust for demographic and clinical characteristics, we were able to explain 7% of the variation in total expenditures in 1995 and 6% of the variation in 1996. Using the same model, we were able to explain 10.7% of the variation in Part B expenditures in 1995 and 8% in 1996. CONCLUSIONS: Beneficiaries with diabetes are consistently more expensive than beneficiaries without diabetes. Demographic and clinical factors at baseline are able to predict only a small portion of future expenditures among this population, and the most expensive patients in one year were often not the most expensive in subsequent years. More work is necessary to assure equitable risk adjustment in the calculation of capitation rates for health plans and practitioners who specialize in the care of individuals with diabetes.