Major decrements in glycated hemoglobin levels between 1978 and 1989 in patients with insulin-dependent diabetes mellitus

Major Decrements in Glycated Hemoglobin Levels Between 1978 and 1989 in Patients With Insulin-Dependent Diabetes Mellitus Cynthia L. Arfken, Lois E. Schmidt, Janet B. McGill, Neil H. White, and Julio V. Santiago

-.-

ABSTRACT The Diabetes Control and Complications Trial has shown that intensive treatment can deter the development and progression of diabetic complications. Integral to intensive treatment is improved glycemic control. To describe the trend in glycemic control for subjects with insulin-dependent diabetes mellitus, we examined the medical records of 662 subjects seen between 1978 and 1989 at the Model Demonstration Unit of the Diabetes Research and Training Center (Washington University School of Medicine). Mean value of glycated hemoglobin showed steady

INTRODUCTION

-1

‘$ he management of patients with insulindependent diabetes mellitus has changed substantially over the past decade.‘r2 Many changes have resulted from the availability of

Center for Health Behavior Research in Department of Medicine (C.L.A.), Division of Biostatistics (C.L.A.), and Departments of PediatricsandMedicine(L.E.S., J.B.M.,N.H.W., J.V.S.), Washington University School of Medicine, St. Louis, Missouri, USA. Send correspondence and reprint requests to be sent to: Dr. Cynthia L. Arfken, Center for Health Behavior Research, Washington University, Box 8504, 660 S. Euclid, St. Louis, MO 63110. Ms. Schmidt’s current address is Department of Medicine, University of Minnesota, Box 101, 420 Delaware Street, SE, Minneapolis, MN 55455. ]ournal of Diabetes and Its Complications 1996; IO:12 -17 0 Elsevier Science Inc., 1996 655 Avenue of the Americas, New York, NY 10010

decline from a peak of 11.5% in 1979 to 9.0% in 1989. This decline was observed both in subjects evaluated only once (annual rate of decline estimated from linear regression, - 0.17 f 0.03; p = 0.0001) and in subjects evaluated more than once (annual rate of decline estimated from growth curves, - 0.18 f 0.06; p = 0.0091). These results suggest that substantial lowering of glycated hemoglobin has occurred during the last decade. This reduction should result in a lowered risk of diabetic complications. (Journal of Diabetes and Its Complications 10;1:12-17, 1996.)

self-monitoring of blood glucose (SMBG), regimens for more physiologic insulin replacement, new dietary recommendations, extensive patient education, and laboratory tools, such as glycated hemoglobin, to assess long-term glycemic control. Concurrent with these changes has been a shift in many North American and European centers toward greater emphasis on improving long-term glycemic control, as measured by glycated hemoglobin.3 Results of the Diabetes Control and Complications Trial (DCCT) showing that intensive treatment with improved glvcemic control can deter the development and pro&&ion of diabetic complications* should accelerate changes toward emphasizing long-term glycemic control. Glycated hemoglobin has been shown to be declining over time. 5,6Klein et al. reported a significant de1056-8727/96/$15.00 SSDI 1056-8727(94)0004&U

DECREMENTS

crease in hemoglobin A1 (I-IbAl) measured at two points in time in a population-based sample of insulin-treated diabetic subjects with onset before age 30 years.5 Although nonsignificant, there was a trend towards lower HbAl measured at two points in time in a populationbased sample of subjects with insulin-treated diabetes and low stimulated l-h C-peptide levels from Michigan communities.6 It is not clear if these declines are continuing as clinicians and patients gain additional experience with the new treatment techniques. The purpose of this report is to describe the temporal trend in glycated hemoglobin in a large sample of subjects with insulin-dependent diabetes being followed at an university-based clinic with special interests in developing and translating new methods of care in patients with diabetes.

blood glucose. Use of self-monitoring of blood glucose (SMBG) increased progressively with time: 25% in 1981, 69% in 1983,89% in 1985 and 95% or more after 1987. Use of three daily insulin injections or insulin infusion pumps was 3% in 1981, 6% in 1983, 14% in 1985 and 17% in 1987 and 1989.

METHODS Subjects and Protocol. The Model Demonstration Unit (MDU) of the Washington University Diabetes Research and Training Center, St. Louis, was the source of data for this study. Subjects participating in the MDU are referred from the private practices of family care physicians, internists, and pediatricians in the St. Louis metropolitan area and from the Washington University Clinics. These subjects, patients with either insulindependent or non-insulin-dependent diabetes, have diverse socioeconomic and educational backgrounds and varying degrees of diabetic control, complications, and regimen adherence. After giving informed consent, subjects undergo a detailed history and physical examination followed by testing designed to help characterize the status of diabetic complications. In addition, subjects undergo an assessment and an update of diabetes knowledge and skills from a Clinical Nurse Specialist and a Registered Dietitian. Results and recommended adjustments in the therapeutic regimen are then communicated to the patient and primary physician. This information included glycated hemoglobin results and what actions were expected if premeal blood glucoses were consistently high (i.e., over 180 mg/dL). Between 1978 and 1989 there were 3290 patient visits to the MDU. Of these, 3162 (96%) had a laboratory determination of glycemic control by glycated hemoglobin. Of these 3162 visits, 1782 visits were by 662 subjects who had been using insulin continuously for at least 2 years and had onset of diabetes before age 30 years. For purposes of this study, these subjects were considered to have insulin-dependent diabetes and are the subject of this report. Visits after 1989 are not included because there was a sharp decline in pediatric visits causing a major change in the average patient profile. In 1979, only 1% of the subjects were using three or more daily insulin injections or monitoring their

IN GLYCATED

HEMOGLOBIN

IN IDDM

13

] Diab Comp 1996; l&12-17

Glycemic Control. Glycemic control was assessed by measurement of glycated hemoglobin. From 1978 to mid-1981, glycated hemoglobin was measured as hemoglobin A1, by high-performance liquid chromatography (HPLC).7 From mid-1981 through late-1987, glycated hemoglobin was measured as hemoglobin A, (HbA1) by mini-column cation-exchange chromatography (Isolab, Akron, OH). Before changing to the minicolumn assay in mid-1981, glycated hemoglobin was determined simultaneously by the two procedures (HPLC and mini-column) in 121 diabetic subjects by our clinical laboratory, and the following relationship was found: HbAl (mini-column) = 0.786 HbA,, + 1.9 (Y = 0.92).s In late 1987, the assay for glycated hemoglobin was changed to measure total glycated hemoglobin (total GHb) by boronate affinity chromatography (GlycoTest, Pierce Chemical Co., Rockford, IL). Before changing assays in 1987, glycated hemoglobin was determined simultaneously by the two procedures (HbAI by mini-column and total GHb by affinity chromatography) in 56 diabetic subjects by our clinical laboratory, and the following relationship was found: glycated hemoglobin (mini-column) = 0.567 total GHb (affinity chromatography) + 2.15 (Y = 0.87).y,‘o Consistency of measurement over time in our laboratory has been accomplished by internal calibration and standardization against HPLC methods.” Conversion between assays has been rigorously assessed*O, advocated10,12and used13,14for consistency of measurement of long-term glycemic control over time. Such conversions are also currently being used to compare the DCCT sample and the population-based sample from the Wisconsin Epidemiologic Study of Diabetic Retinopathy (personal communication: R. Klein, M. Steffes, and J. Santiago). All measures of glycated hemoglobin in this study are expressed in units equivalent to hemoglobin A, by cation-exchange mini-column (HbA,) either as originally measured or as converted by the regression equations discussed above. The normal range of HbAl in our laboratory is from 4.6% to 5.7%.9,‘o The coefficient of variation for the three assays remained below 5%. Statistical Analysis. Levels of glycated hemoglobin are described for the years 1978 through 1989. Temporal trends for grouped data was tested using linear regression modeling with year as a continuous independent variable and grouped continuous data as the dependent variable in separate models. For categorical variables such as race, x2 test of trends was used. Such

14

ARFKEN ET AL.

TABLE

] Uiab

1. DEMOGRAPHIC

AND CLINICAL

CHARACTERISTICS

OF SUBJECTS,

C”Vlp

BY YEAR Duration

Percentage

Year

n

Female %

1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

128 141 154 158 185 171 154 141 143 140 136 128

61 56 57 61 65 59 64 56 64 59 60 55

n For females,

White % 74 72 75 73 74 73 67 70 74 74 73 75

Ageb

<18 yP

mean (SD)

%

21.2 21.6 22.7 24.0 24.6 25.0 24.2 24.8 25.6 24.4 27.8 28.2

55 53 45 41 32 30 34 35 32 36 27 20

(12.6) (12.1) (12.3) (11.8) (11.3) (11.1) (11.2) (11.8) (11.4) (10.8) (12.4) (10.7)

desirable weight“ b mean (SD) 102 105 107 106 111 109 110 109 111 115 119 117

1996; IO: 12. 1;

of

diabetesb mean (SD) 10.4 10.1 11.3 11.8 11.9 12.6 12.4 13.3 13.8 13.3 15.6 16.1

(28) (29) (28) (25) (29) (26) (27) (29) (29) (31) (32) (30)

(8.6) (7.8) (8.7) (8.8) (8.0) (8.1) (7.8) (8.2) (8.4) (7.5) (9.2) (8.3)

kg/m2 x 4.76; for males, kg/m’ x 4.39.

h p < 0.01 by linear regression model or x2 test of trends on grouped data.

analysis does not include information at the individual level. To test the significance of the observed temporal changes and the influence of demographic and clinical factors at the individual level, two approaches were used. First, for those subjects who only had one visit during this study period, stepwise multiple linear regression analysis, with glycated hemoglobin as the dependent variable and year and demographic and clinical factors as the independent variables, was conducted (n = 272). To allow for assay changes, two dummy variables corresponding to the specific assays used were included in the models. Residuals were then inspected for violations of assumptions. Similar analyses were also conducted for the pediatric and adult subsets as well as for the subset with glycated hemoglobin measured only using the cation-exchange mini-column assay. Second, for those subjects who had more than one visit during this study period (390 subjects with a median of six visits), growth curves were constructed.‘j For this, we computed the slope of each subject’s glycated hemoglobin measurements over time using linear regression models. The resulting slopes for all these subjects were then averaged to obtain an estimate of the annual rate of decline. To test if the average slope differed from zero, we used a t test after verifying the assumptions were met. This approach does not require that visits be equally spaced or similar across subjects. Differences in initial glycated hemoglobin, demographic, and clinical factors were assessed using t tests, Mann-Whitney U tests and x2 tests of homogeneity. Annual declines in glycated hemoglobin calculated from the linear regression model and from the growth curves are expressed as mean -t standard error.

RESULTS Table 1 presents the demographic and clinical characteristics of the subjects by year of visit. As a group, the proportions of females (p = 0.80) and whites (p = 0.97) remained stable throughout the 1Zyear study period but age, duration of diabetes, proportion of adults, and percentage desirable weight increased (all p < 0.01). During this study period, the mean value of glycated hemoglobin showed a decline (Figure 1) over time from a peak of 11.5% in 1979 to 9.0% in 1989 (p = 0.001, R* = 0.67). This 2.5% fall in glycated hemoglobin represents a 22% decline and a drop of approximately 75 mg/dL in the average blood glucose level between the peak in 1979 and 1989. Such a decline

12 -

11 -

z ;

lo-

9-

a

I

I

i 978

I

i 980

I

0

i 982

1984

1

i 986

I

i 9aa

I

1990

Year

FIGURE 1. Mean glycated hemoglobin Bars represent standard errors.

by year (p = 0.001).

DECREMENTS IN GLYCATED HEMOGLOBIN

] Diab Camp 2996; 20:22-27

FIGURE 2. Percentage of subjects with glycated hemoglobin below 9% by year (p = 0.001).

cannot be attributed solely to lowering extreme values; the percentage of subjects with an acceptable value of glycated hemoglobin (arbitrarily set at less than 9.0%) showed a concurrent steady increase over time (p = 0.001, Figure 2). By 1989, 60% of subjects had a glycated hemoglobin below 9.0%, whereas in 1978, only 20% had values below 9.0%. To test the significance of the observed temporal changes and the influence of demographic and clinical factors at the individual level, two approaches were used. First, for those subjects who only had one visit during this study period, stepwise multiple linear regression analysis, with glycated hemoglobin as the dependent variable and year, demographic, and clinical factors as the independent variable, was conducted. Those subjects with only one visit had longer duration of diabetes (p = 0.02), a visit later in the study period (p = O.OOOl), and trends toward lower glycated hemoglobin (p = O.lO), and less likelihood of being female TABLE

2. DEMOGRAPHIC

AND CLINICAL

CHARACTERISTICS

Female, % White, % Adults (18 or older), % Age at visit, yr

Percentage desirable weightb Duration of diabetes, yf Initial glycated hemoglobin” Calendar yea?

n = 272 51% 80% 47% 21.1 f 103 + 10.5 + 10.3 + 1983.7 +

’ Mean of: standard deviation. b For females, kg/m2 x 4.76; for males, kg/m2 x 4.39.

10.9” 23” 8.0” 2.3” 3.9”

15

(p = 0.08) when compared to those with subjects with more than one visit (Table 2). After stepwise elimination of nonsignificant factors (p > 0.05), the only predictor of glycated hemoglobin was year. The annual rate of decline in glycated hemoglobin was -0.17 of:0.03 (p = 0.0001, R2 = 0.08). Similar results were found in the subsets of adults 18 years or older (annual rate of decline was -0.13 + 0.05, p = 0.01, R2 = 0.05), children (annual rate of decline was -0.20 f 0.05, p = 0.0001, R2 = 0.12), and subjects with glycated hemoglobin measured only using cation-exchange mini-column (annual rate of decline was -0.15 f 0.06, p = 0.02, R2 = 0.02). Incontrast, amodel using age as the sole independent predictor was not significant (p = 0.14, R2 = 0.008). The second approach was to examine the subset of subjects returning for at least one visit (n = 390) using growth curves. This analysis utilizes all measurements of glycated hemoglobin for each subject who had more than one visit. Consistent with the results of the subset with only one visit, this analysis found an annual rate of decline of -0.18 + 0.06 (p = 0.0001). No demographic or clinical factors from the initial visits predicted the individual slopes (all p’s > 0.05). Hemoglobin Ai, measured by HPLC and hemoglobin A1 as measured by cation-exchange mini-column may be inaccurate in subjects with hemoglobin variants (e.g., hemoglobin S, hemoglobin C, hemoglobin F). To exclude this possible source of error and because hemoglobin electrophoresis was not performed in all subjects, the analyses were repeated after excluding African-American subjects, who are more likely to have hemoglobinopathies. The same trends were observed in this subset as for the entire sample. For those subjects with only one visit (n = 219), the annual rate of decline was -0.21 f 0.04 (p = 0.0001, R2 = 0.13). For those subjects with more than one visit (n = 297), the annual rate of decline was -0.25 k 0.07 (p = 0.0001).

Subjects with one visit only Characteristic

IN IDDM

OF SUBJECTS AT FIRST VISIT

Subjects with more than one visit n = 390 58% 76% 47% 20.4 f 103 + 9.2 f 10.6 + 1981 *

10.8 28” 7.5” 2.2” 2.9”

p values 0.08 0.18 0.97 0.42 0.88 0.02 0.10 0.0001

DISCUSSION Our data show that there has been improvement over time in glycemic control, as measured by glycated hemoglobin, in a sample of younger-onset insulin-using subjects with diabetes mellitus. This downward trend over time in glycated hemoglobin, from 11.5% to 9.0%, confirms and extends previous reports using population-based samples.. 5,6As those previous reports were limited to two points in time (early and mid 198Os), we were able to extend their conclusions by showing a steady continuing decline over more than 10 years. Although selection bias cannot be ruled out as explaining the decline in glycated hemoglobin, there were no institutional factors or systematic attempts to encourage subjects with better or worse glycemic control to participate in the MDU. Subjects with more than one visit did have marginally lower mean glycated hemoglobin values but they also initiated their visits later during the study period. It would be hard to argue that there was a systematic effort to include them as their rate of decline, as a group, was similar to those subjects with only one visit. Considerable effort over the last decade has been placed on developing technology and implementing clinical interventions to lower glycated hemoglobin values. These interventions include SMBG, regimens for more physiologic insulin replacement, new dietary recommendations, extensive patient and professional education, and better laboratory determinations to document glycemic control and to prompt more aggressive treatment effort among patients with high values of glycated hemoglobin. Each of these factors has been shown to lower glycated hemoglobin. The frequency of SMBG (as reported in a research setting or assessed by finger prick marks) has been associated with lower glycated hemoglobin values.“,lh Adherence to newer dietary recommendations has been observed to be associated with lower glycated hemoglobin values.17,1* Enhanced patient education has been the focus of many case studies and controlled studies and has been shown overall to result in l%-2% lower glycated hemoglobin values, at least over the short term.19rz0Regimens for more physiologic insulin replacement (using either multiple daily insulin injections or continuous subcutaneous insulin infusion) have been shown to lower glycated hemoglobin in selected patients.4,2’ Moreover, determination of glycated hemoglobin along with feedback about these results to the patient has been shown to be associated with the beneficial effect of lowering glycated hemoglobin.22,23 It is likely that the decline in glycated hemoglobin observed in our center and by others is the result of a cumulative effect of many of these changes. The low fit of the models highlights the considerable room for many interventions to have impacted glycated hemo-

globin. The relative impact on glycemic control of specific aspects of diabetes treatment remains to be determined. Neither the use of insulin-infusion pumps nor three or more daily insulin injections, however, could account for all the observed changes because less than 10% of the patients were using either of these treatment regimes in 1984.? Furthermore, although prescriptions for SMBG may be estimated fairly reliably, how this information was used or how often monitoring was actually done by individual subjects was not assessed in this study. Our finding that subjects seen only once in the MDU also demonstrate a decline over time in glycated hemoglobin may suggest successful changes in diabetes care have reached beyond the university based research center setting. This conclusion, however, needs to be tempered by the observation that many of the subjects received treatment from physicians aware of these changes. Advances in treatment, which are often pioneered in a MDU setting, have been transferred to local and referring physicians as well as being implemented within our clinics. For example, when physicians are made aware of high glycated hemoglobin values, many may intensify or revise treatment. Data from communitybased programs have also shown that advances in diabetes management (i.e., number of insulin injections per day and frequency of SMBG) are being incorporated into practices outside the university setting and beyond the diabetologists’ offices.s,” The findings reported here represent a continuing step in understanding the trends in glycemic control and management of younger-onset insulin-using subjects with diabetes. From our data, it would appear that glycemic control, as measured by glycated hemoglobin, continues to improve over time at least through 1989. To what extent specific interventions contributed to this trend cannot be accurately determined from our observational data. However, the decline probably represents a cumulative effect of multiple changes in modern diabetes care and will hopefully reflect a lower burden of diabetic complications in the future. ACKNOWLEDGMENT This work was supported in part by National Institutes of Health grants P60-DK20579,MOl-RROO036,and MOl-RRO6021. REFERENCES 1. Spratt IL: Reflections on diabetes care, 1957-1992. Clin Diabetes 10:66, 1992. 2. Santiago, JV: Insulin therapy in the last decade. Diabetes Care 16(suppl3):143-154, 1993. 3. Nathan DM: Hemoglobin &,-infatuation or the real thing? N En@ J Med 323:X)62-1064, 1990. 4. The Diabetes Control and Complications Trial Research Group: The effect of intensive treatment of diabetes on the development and progression of long-term compli-

/ Diab Camp 1996; 10:12-17

cations in insulin-dependent 1 Med 329:977-986, 1993.

DECREMENTS

diabetes mellitus.

N Engl

5. Klein R, Moss SE, Klein BEK: Change in glycemia in a four-year interval in younger-onset insulin-dependent diabetes. Ann Epidemiol 2:283-294, 1992. 6. Hiss RG (ed): Diabetes in Communities II. Ann Arbor, MI, University of Michigan, 1992, pp. 1-136. 7. Davis JE, McDonald JM, Jarrett L: A high-performance liquid chromatography method for hemoglobin A,,. Diabetes 27:102-107, 1978. 8. Hammons GT, Junger K, McDonald JM, Ladenson JH: Evaluation of three mini-column procedures for measuring hemoglobin A,. Clin Chem. 28:1775-1778, 1982. 9.

10.

Nahm MH, Cryer I’, Clutter W, Santiago JV: New assay method for glycated hemoglobin. Barnes Hospital Division of Laboratory Medicine Newsletter lO:l-5, August 1987. Little RR, Wiedmeyer HM, England JD, Wilke AL, Rohling CL, Wians Jr FH, Jacobson JM, Zellmer V, Goldstein DE: Interlaboratory standardization of measurements of glycohemoglobins. Clin Chem 3812472-2478, 1992.

11. Bodor GS, Little RR, Garrett M, Brown W, Goldstein DE, Nahm MH: Standardization of glycohemoglobin determinations in the clinical laboratory: three years of experience. C/in Chem 38:2414-2418, 1992. 12. Santiago JV: Lessons from the Diabetes Control and Complications Trial. Diabetes 42:1549-1554, 1993. 13. Semenkovich CF, Ostlund RE Jr, Schechtman KB: Plasma lipids in patients with type 1 diabetes mellitus. Influence of race, gender, and plasma glucose control: Lipids do not correlate with glucose control in black women. Arch Intern Med 149:51-56, 1989. 14. Ostlund RE Jr, Semenkovich CF, Schechtman KB: Quan-

IN GLYCATED

HEMOGLOBIN

IN IDDM

17

titative relationship between plasma lipids and glycohemoglobin in type 1 patients. Longitudinal Study of 212 patients. Diabetes Care 12:332-336, 1989. 15. Matthews JNS, Altman DG, Campbell MJ, Royston P: Analysis of serial measurements in medical research. BMJ 300:230-235, 1990. 16. Belmonte MM, Schiffrin A, Dufresne J, Suissa S, Goldman H, Polychronakos C: Impact of SMBG on control of diabetes as measured by HbAi. 3-yr survey of a juvenile IDDM clinic. Diabetes Care 11:484-488, 1988. 17. Delamater AM, Smith JA, Kurtz SM, White NH: Dietary skills and adherence in children with type I diabetes mellitus. The Diabetes Educator 14:33-36, 1988. 18. Delahanty L, Halford B: Diet behavior of patients with IDDM who have achieved near normal HbA,, in the DCCT. Diabetes 37(suppl l):lOSA, 1988. 19. Padgett D, Mumford E, Hynes M, Carter E: Meta-analysis of the effects of educational and psychosocial interventions on management of diabetes mellitus. ] Clin Epidemiol 41:1007-1030, 1988. 20. Delamater AM, Bubb J, Davis SG, Smith JA, Schmidt L, White NH, Santiago JV: Randomized prospective study of self-management training with newly diagnosed diabetic children. Diabetes Care 13:492-498, 1990. 21. The Diabetes Control and Complications Trial Research Group: Diabetes Control and Complications Trial (DCCT): Results of Feasibility Study. Diabetes Care lO:l-19,1987. 22. Larsen ML, Horder M, Mogensen EF: Effect of longterm monitoring of glycosylated hemoglobin levels in insulin-dependent diabetes mellitus. N Engl J Med 323: 1021-1025, 1990. 23. Bacon GE, Ladu C, Shein HE, Rucknagel DL: Evaluation of glycosylated hemoglobin in the management of young patients with insulin-dependent diabetes mellitus. J Adolesc Health Care 7:187-190, 1986.