RODOLFO A. VALDEZ, PHD. STEVEN M. HAFFNER, MD. BRAXTON D. MITCHELL, PHD. HELEN P. HAZUDA, PHD. OBJECTIVE â To examine the long-term ...
O R I G I N A L
A R T I C L E
B
efore the advent of NDDG (1) and WHO (2) criteria for diagnosing diabetes mellitus, there was much concern that many people diagnosed on the basis of earlier criteria ceased to be diabetic on follow-up (3-7). In the classic study by O'Sullivan and Mahan (4), using three then-current sets of criteria MICHAEL P. STERN, MD BRAXTON D. MITCHELL, PHD HELEN P. HAZUDA, PHD RODOLFO A . VALDEZ, PHD (U.S. Public Health Service, Mosenthal STEVEN M. HAFFNER, MD and Barry, and Fajans and Conn), 4 5 76% of diabetic patients reverted to norOBJECTIVE — To examine the long-term stability of the World Health Organization mal after 30 kg/m2 and age of onset >40yr. We identified 203 type II diabetic subjects and 301 IGT subjects at baseline. Of the 178 surviving diabetic subjects, 128 (71.9%) returned for the follow-up examination 8 yr later; of these, 117 completed an OGTT. Thus, we knew the diabetic status at follow-up for 142 cases originally diagnosed with type II diabetes (117 plus the 25 who had died; we assumed subjects who were diagnosed as diabetic at baseline and who died during the follow-up interval remained diabetic until their deaths). Similarly, of the 301 subjects diagnosed as having IGT at baseline, we knew the status of 226 (including 15 who had died) at follow-up. From 1984 to 1988, diabetic subjects were recalled for a diabetes complications examination, which took place from 2 to 6 yr after the original
1993
baseline examination. Of the 142 diabetic individuals who form the basis of this report, 105 attended the complications examination. Proteinuria was determined with an early morning spot urine with Albustix (Ames, Elkhart, IN) (16). Retinal photographs were obtained and graded for diabetic retinopathy by the University of Wisconsin Fundus Photographic Reading Center with previously described methods (17). Clinical proteinuria was considered present when the Albustix reading was > 1 +. Retinopathy was graded as none, background, preproliferative, or proliferative (17), based on the degree of retinopathy in the most severely affected eye. The stability of the diagnosis of diabetes was evaluated after 8 yr, based on whether, in addition to meeting the WHO plasma glucose criteria, other confirmatory evidence of diabetes existed. Specifically, we considered whether patients were on antidiabetic medications; or, if not, whether they had a prior diagnosis of diabetes or were newly diagnosed at the time of their survey visit. We computed 95% CI with the method described in Fleiss (18). Nonparametric ANOVA (Kruskal-Wallis test) was used to assess differences in selected characteristics between subjects who remained diabetic throughout the 8 yr and those who reverted to nondiabetic status (19). RESULTS — The prevalence at baseline of IGT was 186 of 1288 (14.4%) in Mexican Americans and 115 of 929 (12.4%) in non-Hispanic whites. The prevalence at baseline of previously diagnosed diabetes was 62 of 1288 (4.8%) in Mexican Americans and 25 of 929 (2.7%) in nonHispanic whites, and the prevalence of newly diagnosed diabetes was 81 of 1288 (6.3%) in Mexican Americans and 35 of 929 (3.8%) in non-Hispanic whites. These prevalences are based on the 301 patients with IGT and the 203 with diabetes identified at baseline, not the 226 and 142, respectively, whose follow-up status was known.
979
Stability of diagnostic criteria for type H diabetes
Table 1—Stability of diagnosis of diabetes according to WHO criteria with or without historical confirmation
amination. The remaining 16 who were nondiabetic at 8 yr showed a variety of patterns. To understand which factors contributed to the stability of the diagnostic criteria, we examined a number of characteristics of the diabetic subjects according to the various diagnostic subgroups defined in Table 1. Table 2 indicates that the sex ratio and the mean age and BMI were similar in all diagnostic subgroups. Of the 142 subjects originally diagnosed with diabetes, 98 (69%) were Mexican Americans. Although Mexican Americans with diabetes were less likely to revert to nondiabetic status at follow-up than were non-Hispanic white diabetic subjects (9 of 98[9.2%] vs. 11 of 44(25.0%]; P = 0.012), this does not appear to account for the differential reversion rates of the different diagnostic subgroups, because both the most stable diagnostic subgroup (on medication) and the least stable (newly diagnosed) contained similar percentages of Mexican Americans. Even though many cases were diagnosed solely on the basis of their medication history without regard to their plasma glucose values, the great majority of cases in all diagnostic categories did, in fact, meet WHO plasma glucose criteria, as shown in Table 2.
STATUS AT FOLLOW- UP TOTAL
(%)
NORMAL
(%)
1GT (%)
DIABETIC
(%)
DECEASED
(%)
BASELINE STATUS
1GT DIABETIC O N MEDICATION
226 (100) 142 (100) 40 (100)
101 (44.7) 9 (6.3) 0(0)
65 (28.8) 11(7.7) 1 (2.5)
45 (19.9) 97 (68.3) 29 (72.5)
15 (6.6) 25 (17.6) 10 (25.0)
19 (100) 83 (100)
0(0) 9 (10.8)
1 (5.3) 9 (10.8)
15 (78.9) 53 (63.9)
3 (15.8) 12 (14.5)
NOT ON MEDICATION PRIOR DIAGNOSIS NEWLY DIAGNOSED
Table 1 summarizes the status at follow-up of the 226 individuals diagnosed with 1GT and the 142 diagnosed with diabetes at baseline and whose status was known 8 yr later. Of the 226 patients with IGT at baseline, 101 (44.7%; 95% CI 38.1-51.0%) were normal at follow-up. Only 45 (19.9%; 95% Cl 15.0-25.4%) progressed to diabetes according to WHO criteria. Of the 142 patients with diabetes at baseline, 20 (14.1%; 95% CI 9.0-20.5%) no longer had diabetes at follow-up; 9 reverted to normal, and 11 to IGT. Only 1 of 40 patients (2.5%; 95% CI 0.1-12.4%) receiving either insulin or oral antidiabetic medication and 1 of 19 (5.3%; 95% CI 0.3-23.7%) not on medication but with a prior history of diabetes reverted to nondiabetic status at follow-up. In contrast, 18 of 83 (21.6%; 95% Cl 13.7-31.2%) newly diagnosed cases reverted to nondiabetic status at
follow-up. In other words, virtually all patients who were no longer diabetic at follow-up (18 of 20) came from the group who at baseline were neither taking medication for nor had a prior diagnosis of diabetes. Moreover, two reverters who previously had been diagnosed at baseline reverted to IGT, not to normal. Although those taking antidiabetic medication at baseline had a higher mortality than those not on medication, the mortality among the latter was similar whether or not they had been diagnosed with diabetes before their survey visit. Of the 105 diabetic subjects who attended the baseline examination, the 8-yr follow-up examination, and the diabetes complications examination at 2 - 6 yr, 81 were diagnosed as diabetic all three times. An additional 8 subjects were diabetic at baseline and 8-yr follow-up but had IGT (7 case) or were normal (1 case) at the intermediate ex-
To understand further the sources of instability in the diagnostic criteria, we examined the newly diagnosed group in greater detail, because this group contributed 90% of those who reverted to nondiabetic status (Table 1).
Table 2—Selected baseline characteristics of diabetic subjects according to variations on WHO criteria MEETING
MEXICAN
WHO
BMI
PLASMA GLUCOSE
FASTING PLASMA
2 - H PLASMA
AGE (YR)
(KG/M 2 )
CRITERIA(%)
GLUCOSE (MG/DL)
GLUCOSE (MG/DL)
AMERICAN
n
(%)
40
75.0
47.5
52.5
30.7
77.5
196.7
353.5
19 83
47.4 71.1
42.1 49.4
55.4 51.8
29.2 29.7
94.7 100.0*
164.5 146.7
335.0 269.8
MALE
(%)
DIABETIC SUBJECTS O N MEDICATION NOT ON MEDICATION PRIOR DIAGNOSIS NEWLY DIAGNOSED
*By definition.
980
DIABETES CARE, VOLUME 16,
NUMBER 7, JULY
1993
Stern and Associates
Table 3—Baseline characteristics of newly diagnosed diabetic patients according to diabetic status at follow-up
n AGE (YR) DUNCAN SCORE BM1 (KG/M 2 ) CENTRALITY INDEX sBP (MMHG) dBP (MMHG) TOTAL CHOLESTEROL (MG/DL) TG (MG/DL) H D L CHOLESTEROL (MG/DL) FASTING GLUCOSE (MG/DL) 2-H GLUCOSE (MG/DL) FASTING INSULIN (JJLU/ML)
REVERTED TO
REMAINED
NONDIABETIC
DIABETIC
P
18 51.7 53.4 26.6 1.17 124.7 75.6 208 159 51.2 129 201 16.0
53 51.5 47.3 30.6 1.34 121.3 76.0 228 255 41.0 149 292 24.2
0.560 0.419 0.003 0.216 0.534 0.638 0.171 0.004 0.005 0.067 0.000 0.033
Table 3 shows the mean age was nearly identical in those w h o did and did not revert to nondiabetic status at follow-up. The former, however, were of higher s o cioeconomic status, were leaner, had less centrally located adipose tissue, lower TG and higher HDL cholesterol, and lower glucose and insulin concentrations, although not all of these differences were statistically significant. Still, the group that reverted clearly displayed substantially less of the metabolic derangements that characterize both the prediabetic (20) and the diabetic (21) state. The mean weight change among those who reverted to nondiabetic status was a gain of 0.4 kg (11.1 kg loss to 11.0 kg gain). Only 2 subjects lost > 5 kg and seven lost > 2 . 5 kg. Table 4 presents the prevalence of proteinuria and retinopathy at baseline among subjects who either remained diabetic or reverted to nondiabetic status at follow-up. The prevalence of proteinuria was substantially higher in those taking antidiabetic medications than in the other diagnostic subgroups. Similarly, of the 15 patients with more severe grades of retinopathy (preproliferative and proliferative), only 4 were in the newly diagnosed category. Of the patients w h o reverted to nondiabetic status
DIABETES CARE, VOLUME 16, NUMBER 7, JULY 1993
at follow-up, only 1 demonstrated proteinuria at baseline, and none demonstrated any grade of retinopathy, even background retinopathy. CONCLUSIONS— Only a relatively small number of studies have addressed the question of the stability of the diabetes diagnosis when made with modern NDDG or WHO criteria. Most studies have focused on the stability of the IGT category. Information on the stability of diabetes itself has usually been incidental to the main purpose of the study. Several studies have indicated that 26.7-84.8% of individuals originally classified as di-
abetic by WHO or NDDG criteria remained diabetic after 1 wk to 7 yr of follow-up (22-26). Many of these studies do not make clear whether they included previously diagnosed cases. The study with the highest reproducibility rate was a 7-yr follow-up study of a Japanese cohort (22). The percentage of originally diagnosed diabetic patients who were still diabetic after 7 yr was 84.8%, which agrees with our retention rate of 85.9%. The Japanese study, however, was based on only 33 cases. A study in Tanzania, in which subjects were recalled for a confirmatory OGTT 1 wk after the original test, had the lowest retention rate (26). Surprisingly, even after this short interval, only 26.7% of the 30 subjects originally classified as diabetic met the criteria on the second test. The authors raised the question of whether anxiety about an unfamiliar situation could have elevated the original test results and postulated a phenomenon similar to the acclimatizing reflex seen with BP measurements. None of the diabetic individuals in this study had been diagnosed previously. Thus, they are comparable with our newly diagnosed subgroup in whom the retention rate was considerably higher at 78.4%. Presumably, the prevalence of diabetes is quite low in rural Tanzanian villages, certainly lower than in urban Mexican Americans (10). The differences between our results and those from Tanzania are
Table 4—Prevalence of proteinuria and retinopathy at baseline among subjects who remained diabetic or who reverted to nondiabetic status at follow-up RETINOPATHY n (%) PROTEINURIA
(1+ OR >) n (%)
PREPRO-
PROLIFER-
BACKGROUND
LIFERATIVE
ATIVE
6 OF 28(21.4) 10 (35.7)
7 (25.0)
9(32.1)
2(7.1)
1 OF 13 (7.7) 7 (50.0) 3 OF 47 (6.4) 36 (76.6) 1 OF 16 (6.3) 15 (100.0)
7 (50.0) 7 (14.9) 0(0)
0(0) 4 (8.5) 0(0)
0(0) 0(0) 0(0)
NONE
DIABETIC SUBJECTS O N MEDICATION NOT ON MEDICATION PRIOR DIAGNOSIS NEWLY DIAGNOSED REVERTED TO NONDIABETIC STATUS AT FOLLOW-UP
981
Stability of diagnostic criteria for type H diabetes
predicted by Bayes Theorem, which states that for a test with a given sensitivity and specificity, fewer individuals who test positive truly will have the disease when the prior probability of disease is low, as is the case in Tanzania. As noted, in our study only 14.1% of subjects meeting the WHO criteria for diabetes were found to be nondiabetic at follow-up. Moreover, those who reverted to nondiabetic status displayed characteristics that might have provided a hint that they were false positives. Only 2 had been diagnosed previously by a physician, and of these, only 1 was on antidiabetic medication. Moreover, those who reverted to nondiabetic status had few of the associated findings common in diabetic patients. For example, they were leaner, had less centrally distributed adipose tissue, and had a more favorable lipid and lipoprotein profile than the true positives. Also, they had lower glucose and insulin values. Finally, they had little if any evidence of microvascular complications. Nevertheless, because these false positives are typically not excluded from epidemiological surveys, ours included, the implication is that such surveys may overestimate the prevalence and incidence of type II diabetes by as much as 16% (142/122 = 1.16). In fact, the actual overestimate could well be less than this. Because the sensitivity of the OGTT is presumably less than 100%, some genuine diabetic patients might have had false-negative tests at baseline and therefore would not be counted in the prevalence estimate. Also, some of the individuals we labeled as false positives could have been genuine diabetic individuals in remission at the time of their follow-up examination. None of these were on pharmacological treatment for diabetes, so remission could have resulted only from diet treatment, and, in particular, weight loss. Certainly, the 2 patients who lost >5.0 kg could have gone into remission, and this might also have occurred in some of
982
the patients with lesser degrees of weight loss. We conclude that the modern WHO and NDDG criteria for diabetes are substantially more stable over time than earlier criteria. Nevertheless, prevalence estimates in epidemiological studies still may contain false positives, particularly if only a single OGTT is used to detect newly diagnosed diabetes. Although the magnitude of the prevalence overestimate is modest in this study, under certain circumstances this might not be the case. For example, in populations with a low prior probability of diabetes, the positive predictive value of an OGTT would be lower than in populations with a higher prior probability of disease.
8.
9.
10.
11.
Acknowledgments—This work was supported by grants from the National Heart, Lung, and Blood Institute (R01-HL-24799 and R37-HL-36820) and by the ADA Mentor-based Postdoctoral Fellowship Program (R.A.V.).
12.
13. References 1. National Diabetes Data Group: Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 28:1039-57, 1979 2. World Health Organization: WHO Expert
3.
4.
5.
6.
7.
Committee on Diabetes Mellitus, Second Re14. port. Geneva, World Health Org., 1980 (Tech. Rep. Ser., no. 646) McDonald GW, Fisher GF, Bumham C: Reproducibility of the oral glucose tolerance test. Diabetes 14:473-80, 1965 15. O'Sullivan JB, Mahan CM: Prospective study of 352 young patients with chemical diabetes. N Engl J Med 278:103841, 1968 16. Fitzgerald MG, Malins JM: Ten-year follow-up report on the Birmingham Diabetes Survey of 1961. Br MedJ 2:35-37, 1976 Jarrett RJ, Keen H, Fuller JH, McCartney M: Worsening to diabetes in men with 17. impaired glucose tolerance ("borderline diabetes"). Diabetologia 16:25-30, 1979 Keen H, Jarrett RJ, McCartney P: The ten-year follow-up of the Bedford Survey
(1962-1972): glucose tolerance and diabetes. Diabetologia 22:73-78, 1982 Stem MP: Type II diabetes mellitus: Interface between clinical and epidemiological investigation. Diabetes Care 11: 119-26, 1988 Stern MP, Rosenthal M, Haffner SM: A new concept of impaired glucose tolerance: relation to cardiovascular risk. Arteriosclerosis 5:311-4, 1985 Stem MP, Rosenthal M, Haffner SM, Hazuda HP, Franco LJ: Sex difference in the effects of sociocultural status on diabetes and cardiovascular risk factors in Mexican-Americans: The San Antonio Heart Study. Am J Epidemiol 120:834-51, 1984 Haffner SM, Hazuda HP, Mitchell BD, Patterson JK, Stem MP: Increased incidence of type II diabetes mellitus in Mexican-Americans. Diabetes Care 14:102108, 1991 Haffner SM, Ferrannini E, Hazuda HP, Stern MP: Clustering of cardiovascular risk factors in confirmed prehypertensive individuals. Hypertension 20:38-45, 1992 Haffner SM, Stern MP, Hazuda HP, Pugh JA, Patterson JK: Do upper body and centralized adiposity measure different aspects of regional body fat distribution? Relationship to non-insulin-dependent diabetes mellitus, lipids, and lipoproteins. Diabetes 36:43-51, 1987 Haffner SM, Stern MP, Hazuda HP, Pugh JA, Patterson JK: Hyperinsulinemia in a population at high risk for non-insulindependent diabetes mellitus. New Engl J Med 315:220-24, 1986 Duncan OD: A socioeconomic index for all occupations. In Occupations and Social Status. Reiss AJ Jr, Ed. New York, Free Press, 1961, p. 109-258 Haffner SM, Mitchell BD, Pugh JA, Stern MP, Kozlowski MK, Hazuda HP, Patterson JK, Klein R: Proteinuria in MexicanAmericans and non- Hispanic whites with N1DDM. Diabetes Care 12:530-36, 1989 Haffner SM, Fong D, Stern MP, Pugh JA, Hazuda HP, Patterson JK, Van Heuven WAJ, Klein R: Diabetic retinopathy in Mexican-Americans and non-Hispanic whites. Diabetes 37:878-84, 1988
DIABETES CARE, VOLUME 16,
NUMBER 7, JULY
1993
Stern and Associates
18. Fleiss JL: Statistical Methods for Rates and Proportions. Second Edition. New York, Wiley, 1981, p. 14 19. SAS Institute: SAS/STAT User's Guide, 22. Version 6, Fourth Edition, Volumes 1-2, Cary, NC, SAS Institute, 1989 20. Haffner SM, Stern MP, Hazuda HP, Mitchell BD, Patterson JK: Cardiovascular risk factors in confirmed prediabetic 23. individuals: Does the clock for coronary heart disease start ticking before the onset of clinical diabetes? JAMA 263: 2893-98, 1990 24. 21. Stern MP, Haffner SM: Dyslipidemia in
DIABETES CARE, VOLUME 16,
NUMBER 7, JULY
1993
type II diabetes. Implications for theraban population: A new approach for peutic intervention. Diabetes Care 14: identifying high-risk cases. Diabetologia 1144-59, 1991 33:526-31, 1990 Sasaki A, Suzuki T, Horiuchi N: Devel- 25. Balkau B, Eschwege E: Repeatability of opment of diabetes in Japanese subjects the oral glucose tolerance test for the with impaired glucose tolerance: A sevdiagnosis of impaired glucose tolerance en-year follow-up study. Diabetologia 22: and diabetes mellitus (Letter). Diabetolo154-57, 1982 gia 34:201-203, 1991 Forrest RD, Jackson CA, YudkinJS: The 26. Swai ABM, McLarty DG, Kitange HM, abbreviated glucose tolerance test in Kilima PM, Masuki G, Mitinangi BI, screening for diabetes: The Islington DiChuwa L, Alberti GMM: Study in Tanzaabetes Survey. Diab Med 5:557-61, 1988 nia of impaired glucose tolerance: methEriksson K-F, Lindgarde F: Impaired gluodological myth? Diabetes 40:516-20, cose tolerance in a middle-aged male ur1991
983
