- Email: [email protected]
Glucose intolrrance and diabetic complications among Japanese-American women
Diabetes Research and CIinical Practice, I3 (199 1) 119-I 30 0 1991 Elsevier Science Publishers B.V. 016%8227/91/$03.50
119
DIABET 00505
Epidemiological
Note
Glucose intolerance and diabetic complications Japanese-American women W.Y. Fujimoto ‘, D.L. Leonetti2,
R.W. Bergstrom’, P.W. Wah15
J.L. Kinyoun3,
among
W.C. Stolov4 and
Departments of ‘Medicine, ‘Anthropology, ‘Ophthalmology, “Rehabilitation Medicine and ‘Biostatistics. Universit.v of Washington, Seattle, Washington, U.S.A. (Received 12 December 1990) (Revision received 4 February 1991) (Accepted 7 February 1991)
Summary The prevalence of glucose intolerance and diabetic complications was determined in second-generation Japanese-American (Nisei) women and compared to previously obtained results in Nisei men. A volunteer study sample of 191 Nisei women 45-74 years old was enrolled from a study population of 1489 Nisei women born 1913-1942, raised and educated in the U.S., and residing in King County, Washington. The enrolled sample included 72 with normal glucose tolerance, 67 with impaired glucose tolerance (IGT), and 52 with non-insulin-dependent diabetes. A random sample was also drawn from the study population to form a reference sample of 157 women. Based upon observations in the reference and enrolled samples, an estimated 16% of Nisei women in the study population have diabetes and 40% IGT. These rates compare to 20% diabetes and 36% IGT previously estimated for Nisei men 45-74 years old. The prevalence of cardiovascular disease (hypertension, peripheral vascular disease, and/or coronary heart disease) was highest among diabetic women, lowest in those with normal glucose tolerance, and intermediate in women with IGT. In comparison to diabetic men, there was a significantly lower frequency of neuropathy, peripheral vascular disease, and coronary heart disease in diabetic women. However, hypertension occurred equally often in both. Thus Japanese-American men and women 45-74 yr old have a similar prevalence of glucose intolerance, although less severe in women, and complications, except for hypertension, are reduced in women.
Key words: Coronary
heart disease; Complication; glucose tolerance; Japanese-American;
Correspondence to: W.Y. Fujimoto, University of Washington, Endocrinology and Nutrition, Seattle, WA 98195, U.S.A.
Diabetes; Peripheral
Department
Epidemiology; Hypertension; vascular disease; Women
of Medicine
RG-26, Division
Impaired
of Metabolism,
120
Introduction In an epidemiologic study of non-insulin-dependent diabetes conducted from 1983 to 1985 in a population of second-generation Japanese-American (Nisei) men who were between the ages 45 and 74 yr old and were residing in King County, Washington, an estimated 20% were diabetic while an additional 36% had impaired glucose tolerance (IGT) [ 11. The prevalence of cardiovascular disease (hypertension, coronary heart disease, and peripheral vascular disease) was highest in men with diabetes, lowest in those with normal glucose tolerance, and intermediate in those with IGT, while retinopathy and peripheral neuropathy were associated only with diabetes [2]. Since gender may modify the occurrence of glucose intolerance and its complicating conditions, we have extended the study to Nisei women 45 to 74 years old and compared the findings with those obtained in Nisei men.
Reference
sample
To provide sociodemographic information and an estimate of the prevalence of known diabetes in the study population, a reference sample was derived [ 11. A telephone survey was conducted in August 1983 of 405 households selected at random from a comprehensive directory of JapaneseAmerican households (6242 listings). There were 399 responding households containing 199 Nisei women. From this random telephone survey, a reference sample of 157 women was obtained by first identifying those Nisei women born in the U.S. between 1913 and 1942 (n = 181), followed by exclusion of Kibei (n = 24). Each respondent was asked whether she had ever been told by a physician that she had diabetes using the following wording: ‘diabetes/ blood too much sugar in the blood/high sugar/sugar in the urine’. The telephone survey also provided data on residential distribution and on family history of diabetes. Enrolled sample
Subjects and Methods Study population
Nisei include any Japanese person born in the U.S. of first-generation (Issei) immigrant parents who arrived in the U.S. before 1925, or of one Issei and one Nisei parent. We estimate that in King County, Washington, there were 1925 Nisei women born 1905-1944 among females of Japanese ancestry numbered in the 1980 census [3-61. A major portion (n = 1489) of this total population of Nisei women, born in the U.S. from 1913 through 1942 and educated in the U.S., comprised the study population. By selecting these birth years, the study population of Nisei women had the same age range as the earlier study sample of Nisei men born 1910-1939 [ 11. The study population excluded Nisei educated in Japan (Kibei) since education in the U.S. was an important aspect of Nisei life experience.
The enrolled sample is a volunteer group of 191 Nisei women (born 1913 through 1942 and excluding Kibei) who were examined from November 1986 to January 1988. Recruitment and screening was done as described previously for the Nisei men [ I]. Information about the project was disseminated throughout the Japanese-American community, followed by a call for volunteers. Using the same criteria as in the telephone survey quoted above, volunteers were asked to identify themselves as either diabetic or non-diabetic. Of 447 women who volunteered, 69 self-reported diabetes and 378 were self-reported non-diabetic. A short telephone interview was used to screen 306 of these volunteers (all 69 self-reported diabetic and 237 of the self-reported non-diabetic women) to determine their eligibility to participate. The self-reported diabetic women were screened first, and 46 (66.7 %) were enrolled. However, a final diagnosis of diabetes was not made until additional information was received, as described below. From the pool of 378 selfreported non-diabetic women, screening in-
121
terviews were carried out progressively by randomly selecting approximately 35 volunteers at a time until adequate sample sizes were achieved (approximately 70 subjects within the diagnostic categories of normal glucose tolerance and IGT). Ultimately 237 were screened, of whom 145 (61.2%) were enrolled. Excluded from participation in the study were women in whom any of the following were present: outside the age range (n = 23), Kibei (n = 20), raised outside the mainland U.S. (n = 9), not full J apanese (n = l), sister to a previously enrolled woman (n = 25), and potentially confounding medical conditions (n = 19). An additional 18 women who volunteered were not enrolled due to scheduling conflicts. The reasons for exclusion of volunteers were proportional between the two groups (self-reported diabetic or non-diabetic). Although sisters to women already enrolled were excluded in order to obtain independent observations on all female subjects, 26 Nisei women in the enrolled sample had a brother in the Nisei men’s enrolled study sample. For many of the comparisons between Nisei men and women, a second comparison was done that excluded men and women who were siblings. Class$cation ance status
of enrolled sample by glucose toler-
Classification of subjects was based upon a combination of medical information obtained from the subject and her physician, and a 75-g oral glucose tolerance test after a 10-h overnight fast, with findings assessed by World Health Organization [7] criteria for IGT (fasting serum glucose < 7.78 mmol/l and 2-h glucose 27.78 and < 11.11 mmol/l) and diabetes (fasting serum glucose 2 7.78 mmol/l or 2-h glucose 2 11.11 mmol/l) with normal glucose tolerance defined as absence of criteria for either IGT or diabetes (i.e., both fasting and 2-h serum glucose ~7.78 mmol/l). All subjects were on their usual diet before the glucose tolerance test. All medications usually taken in the morning, including oral hypoglycemic agents, were postponed until after the
test. Subjects receiving insulin reduced their morning dosage by approximately 33% on the day of the glucose tolerance test. Assessment of complicating conditions
An average blood pressure was calculated from three consecutive measurements by indirect auscultation and a mercury sphygmomanometer read to the nearest 2 mm Hg. Systolic blood pressure was determined by the first perception of sound and the diastolic (5th-phase) blood pressure by the disappearance of sound. All subjects underwent direct and indirect funduscopic examination of the eyes after mydriasis. In all subjects with diabetes and any subject with evidence of retinal pathology by ophthalmoscopy, standard photographs were taken of each fundus (7 standard stereoscopic fields) and retinopathy graded as P characteristics, modified from the Early Treatment Diabetic Retinopathy Study [ 81: mild non-proliferative without (PM) or with macular edema (P,), moderate non-proliferative (P,), severe non-proliferative (P2), and proliferative retinopathy (P3, P,, and Ps). A resting 12-lead electrocardiogram was done. Occurrence of Minnesota Code items 1. l-l .3 (abnormal Q and QS patterns), 5.1-5.3 (abnormalities of T-wave) and 7.1 (complete left bundle branch block) signified ischemic heart disease [9]. Code classes 1.1, 1.2, and 7.1 signified ‘probable myocardial infarction’, and code classes 1.3, 5.1, 5.2, and 5.3 signified ‘other suspect ischemia’ [lo]. Systolic blood pressure was measured by the Doppler method in both brachial arteries and in the anterior and posterior tibia1 arteries of both legs to calculate an ankle:arm index (AAI) for each ankle, with the higher of the two brachial pressures as the denominator and the lower of the two pressure determinations as the numerator for each ankle. AA1 < 0.95 signified occlusive arterial disease of the lower legs, and AA1 < 0.90 identified severe arteriosclerosis obliterans [ 111. Blood flow in the carotid arteries was assessed by ultrasonic duplex scanning of each carotid bifurcation, and the degree of carotid artery occlusive
122
disease was estimated from Doppler spectral wave forms [ 121. Subjects with 2 50% diameter reduction of the internal or common carotid artery were classified as having carotid artery disease. Inquiry was made about history of coronary heart disease (angina, myocardial infarction, coronary angioplasty, or coronary artery bypass graft) and peripheral vascular disease (stroke, transient ischemic attack, endarterectomy, intermittent claudication, aortic aneurysm, angioplasty, or peripheral arterial bypass surgery). This information was supplemented with the London School of Hygiene Cardiovascular Questionnaire [ 91. Nerve conduction studies of the right upper and lower extremities were done unless there was known to be a condition that might preferentially affect any of the nerves on that side, in which case studies were done on the left side. The following nerves and parameters were assessed: (1) lower extremity motor: peroneal nerve conduction velocity, distal latency, and evoked amplitude, and tibial nerve conduction velocity, distal latency, evoked amplitude, and H- and F-wave latencies; (2) lower extremity sensory: peroneai and sural nerve conduction velocity and evoked amplitude and sural nerve latency; (3) upper extremity motor: median and ulnar nerve forearm conduction velocity, distal latency, evoked amplitude, and F-wave latency; and (4) upper extremity sensory: median and ulnar nerve forearm conduction velocity, distal latency, and evoked amplitude. Neuropathy was present if conduction velocity or distal latency exceeded by > 2 SD the normal mean in at least three nerves. Fasting serum creatinine, 24-h urinary excretion of creatinine (creatinine clearance), and urinary protein were assessed. Creatinine was measured by the automated picric acid method and protein by trichloroacetic acid precipitation and photometry (University of Washington Medical Center, Department of Laboratory Medicine).
Statistical methods All group differences in measured characteristics were tested by parametric or non-parametric (Kruskal-Wallis test) analysis of variance procedures. Pairwise comparisons were evaluated by the two sample t-test, the Student-NewmanKeuls range test, or the non-parametric MannWhitney test for those characteristics having a non-normal distribution. Differences between diabetic men and women were also age-adjusted using analysis of covariance. Differences in categorical characteristics were evaluated by the Chisquare test.
Results Prevalence of known diabetes from the random telephone survey Prevalence of self-reported diabetes was 8.5 % for all 199 Nisei women respondents and 8.3% for the 157 reference sample women. These rates compared with 10.9% and 13.2% in similarly constructed samples of Nisei men [ 11. Volunteer response rates Based on the random survey, approximately 8.5 % or 164 of the total King County population of 1925 Nisei women would be expected to selfreport diabetes. The 69 volunteers with selfreported diabetes represented approximately 42% of this number. This response rate is less than the corresponding proportion for Nisei men (117 volunteered of an estimated 2 13 men, or a response rate of 55% ; x2 = 6.1, P < 0.02) [ 11. The 378 self-reported non-diabetic women volunteers represented approximately 21 y0 of the remaining estimated 1761 King County Nisei women expected to self-report absence of diabetes. This response rate was identical to the percentage who volunteered of the estimated 1738 Nisei men in King County expected to self-report the absence of diabetes [ 11.
123
Comparison of the enrolled sample of Nisei women to the reference sample of Nisei women and to the enrolled sample of Nisei men To ascertain that the enrolled sample was representative of the study population, several characteristics were compared between the enrolled and reference sample Nisei women, both of which exclude Kibei. Separate comparisons were made for self-reported diabetic and self-reported nondiabetic women. In addition, comparisons were also made between the enrolled sample of Nisei women and the enrolled sample of Nisei men examined earlier and found to be representative of the Nisei male study population [ 11. All of these comparisons were not significantly different, as summarized in Table 1. TABLE I Summary of X-square test statistics used in comparisons of characteristics of the enrolled sample of Nisei women to the reference sample of Nisei women and to the enrolled sample of Nisei men
Comparison of enrolled sample sample of Nisei women Age distribution Residential distribution” Positive family history of diabete? Comparison of enrolled sample sample of Nisei men’ Residential distribution Positive family history of diabetes Prefectural origin in Japan of parent?
Self-reported status
diagnostic
Diabetic
Non-diabetic
of Nisei women to reference 0.03 0.54
0.42 1.25
0.05 3.13 of Nisei women to enrolled 0.90 (0.40)
0
(0.03)
2.08 (1.54)
1.37 (0.30)
0
0.30
Corresponding P values were all not significant (P > 0.05). ” Ethnic areas of Seattle vs non-ethnic areas of Seattle and its suburbs. ’ History of diabetes in a parent or full sibling. c x7 value for comparison when siblings were removed is shown in parentheses. d All of Kyushu Island and the prefectures of Hiroshima and Yamaguchi vs all other areas of Japan. In the parallel study of Nisei men the geographic distribution of parental origins ofthe enrolled sample was not significantly different from that of the reference sample [I].
TABLE 2 Final diagnostic classification status of Nisei women compared with self-reported entry status Self-reported entry status
Final diagnostic classification status
Number (“,)
Non-diabetic (n = 145)
Normal IGT Diabetes Normal* IGT* Diabetes
69 61 15 3 6 37
Diabetic (w = 46)
* None were receiving treatment
(47.6) (42.1) (10.3) (6.5) (13.0) (80.4)
directed towards diabetes.
Diagnostic class@ation of enrolled sample The final diagnostic classification of self-reported nondiabetic and diabetic women in the enrolled sample is known in Table 2. The enrolled sample consisted of 72 women with normal glucose tolerance, 67 with IGT, and 52 with diabetes. Of the 52 diabetic women, 30 were receiving no medication for diabetes (14 of whom had entered the study as self-reported non-diabetic subjects); 20 were being treated with oral hypoglycemic agents and 2 with insulin. Two of the women on oral agents had been on insulin previously. Both of the women who were receiving insulin had a history consistent with non-insulin-dependent diabetes. Eighteen (34.5%) of the diabetic women had fasting serum glucose levels 27.78 mmol/l, one of whom was insulin-treated. All had 2-h serum glucose 2 11.11 mmol/l. Of the remaining diabetic women, 33 had 2-h serum glucose 2 11.11 mmol/l while one had a 2-h glucose of 6.89 mmol/l while receiving glyburide. Among diabetic women, there was a highly significant 2 7.78 association between fasting glucose mmol/l and treatment with either oral agents or insulin; 77.80,;, of the 18 women with fasting glucose 2 7.78 mmol/l were receiving medications for diabetes compared with only 23.5 Y0of the 34 women with fasting glucose < 7.78 mmol/l (x2 = 12.1, P = 0.0005).
124
Age of enrolled sample by diagnostic group The median age of all subjects was 63.3 years and the mean age ( i SD) was 62.7 k 6.2 years. Mean ages ( + SD) were similar among the three final diagnostic groups: 61.9 + 5.9 for normal women, 62.7 + 6.4 for IGT women, and 63.7 + 6.2 for diabetic women (F = 1.39, P = 0.25). For Nisei men, mean ages were 61.6 years for normal men, 61.3 years for IGT men, and 61.6 years for diabetic men [ 11.
diabetes in the study population was estimated to be 10.3x, based upon the proportion of selfreported non-diabetic enrolled sample women found to have diabetes. From these figures, the derived estimated prevalence rate of diabetes, both previously diagnosed and confirmed plus previously undiagnosed among Nisei women 45-74 years old (excluding Kibei) in King County is 16.1% (confidence interval 14.3%-17.9x), based upon the calculation: 0.067 + 0.103 x (1 - 0.083). This rate compared with 20.0% (confidence interval 18.0%-22.0%) for men 45-74 years old [ 11.
Estimated prevalence rate of diabetes in the study population Of the reference sample, 8.3 y0 self-reported previously diagnosed diabetes. However, 19.6% of self-reported diabetic women in the enrolled sample were not confirmed to have diabetes by oral glucose tolerance test. Therefore, the corrected prevalence of previously diagnosed diabetes in the reference sample was estimated to be 6.7% which was also assumed to be the prevalence of previously diagnosed diabetes in the study population. The prevalence of undiagnosed
Estimatedprevalence rate of IGT in the studypopulation In the enrolled sample, 13.0% of self-reported diabetic women and 42.1% of the self-reported non-diabetic women were found to have IGT. The derived estimated prevalence rate of IGT among Nisei women 45-74 years old (excluding Kibei) in King County, both undiagnosed plus self-reported as diabetes, is 39.7% (confidence
TABLE 3 Frequency
of complicating
conditions
in Nisei women by diagnostic Normal (n = 72)
Microangiopathy or neuropathy* Hypertension On antihypertensive medication and/or: Blood pressure > 140/90 Blood pressure > 160/95 On antihypertensive medication History of known hypertension Peripheral vascular diseaseb Coronary heart disease’ Cardiovascular diseased
category: number of subjects (%)
IGT (n = 67)
Diabetes (n = 52)
X2
P
5 (6.9)
3 (4.5)
15 (28.8)
19.2
0.0001
25 (34.7) 16 (22.2)
31 (46.3) 18 (26.9)
35 (67.3) 25 (48.1)
12.9 10.2
0.002 0.006
16 (22.2)
13 (19.4)
20 (38.5)
6.29
0.04
22 (30.6)
20 (29.9)
27 (51.9)
7.74
0.021
2 (2.8) 14 (19.4) 27 (37.5)
13 (19.4) 24 (35.8) 35 (52.2)
8 (15.4) 19 (36.5) 36 (69.2)
8.20 5.98 12.2
0.017 0.05 0.0022
a At least one of the following: serum creatinine 2 0.115 mmol/l (above the reference range for the laboratory), urinary protein excretion > 10 mg/dl (1 SD beyond the mean for normal subjects), presence of retinopathy, or electrophysiologic evidence for peripheral neuropathy. b History, AA1 < 0.95, or evidence of carotid artery disease by Doppler. c History or electrocardiographic evidence by Minnesota Code. d At least one ofthe following: on antihypertensive medication, blood pressure > 140/90, peripheral vascular disease, or coronary heart disease.
125
interval 36.6% -42.9%), derived from the calculation: 0.13 x (0.083) + 0.421 x (1 - 0.083). This rate compared with 36.1 y0 (confidence interval 33.7”;-38.5%) for men 45-74 years old [ 11.
in 1 normal woman, 6 IGT women, and 3 diabetic women. Minnesota Code items 1.1, 1.2, or 7.1 indicating ‘probable myocardial infarction’ were found in none of the normal women, 4 IGT women and 2 diabetic women. Minnesota Code items 1.3, 5.1, 5.2 and 5.3 indicating ‘other suspect ischemia’ were found in 13 or 18.17; of normal women, 15 or 22.4% of IGT women, and 16 or 30.8% of diabetic women.
Frequency of complicating conditions in Nisei women
Complicating conditions showing significant differences by diagnostic category are shown in Table 3. Mean creatinine clearance, serum creatinine, and concentration of protein in a morning urine specimen were all not significantly different by diagnostic category. Retinopathy was observed in only 10 women with diabetes; 8 had mild and 1 moderate non-proliferative retinopathy and 1 had proliferative retinopathy. Electrophysiologic evidence of peripheral neuropathy was observed in 4 women with diabetes and 1 with IGT. A history of peripheral vascular disease was obtained from none of the normal women, 3 IGT women, and 2 diabetic women. AA1 < 0.95 was measured in 1 normal woman, 1 IGT woman, and 4 diabetic women (3 of whom had AA1 < 0.90). Evidence of carotid artery disease by Doppler was present in 1 normal woman, none of the IGT women, and 1 diabetic woman. A history of coronary heart disease was present
Comparison of diabetic Nisei women and men
Table 4 compares diabetic Nisei women in the present study and the previously reported diabetic Nisei men [l] for several characteristics relevant to their diabetes, most of which indicate less severe diabetes in women. Table 5 compares prevalence of complicating conditions associated with diabetes. Diabetic neuropathy was much less frequent in women. Prevalence of peripheral vascular disease (history or AA1 < 0.95) or coronary heart disease (history or electrocardiographic abnormalities limited to Minnesota Code item 1.1, 1.2 or 7.1 or ‘probable myocardial infarction’) was significantly less in diabetic women than in diabetic men. In the entire enrolled study sample of 19 1 Nisei women and 229 Nisei men, prevalence of arterio-
TABLE 1 Comparison
of diabetic Nisei men and women (mean + SD)
Variable
Age W-1
Age at diagnosis of diabetes (yr) Duration of diabetes (yr) Fasting glucose (mmol/l) 2-h glucose (mmol/l) 9,” Fasting glucose 2 7.78 mmol/l O0 Oral agents or insulin .I NS, not significant. ’ Mann-Whitney Z.
P”
Age-adjusted
Men (n = 78)
Women (n = 52)
rlx2
61.7 + 6.0
63.7
+ 6.2
- 1.85
0.066
55.2 f 7.5
59.6 + 8.9
- 2.97
0.004
0.043
6.5 f 6.7
4.7 + 6.7
2.07b
0.038
0.048
9.54 + 3.21 18.17 f 5.23
7.83 f 2.86 16.96 f 5.52
3.41b 1.31b
0.0007 NS
0.003 NS
60.3
34.5
7.2
0.007
48.7
42.3
0.29
NS
P”
126 TABLE 5 Comparison
of frequency of complications
Complication
Nephropathyb Retinopathy Neuropathy Any of the above Hypertension’ Blood pressure > 140/90 Blood pressure > 160/95 Peripheral vascular diseased e Coronary heart disease
between diabetic Nisei men and women ( %)
x2
P”
13.5 19.2 7.7 28.8
0.1 0.9 19.9 6.9
NS NS < 0.000 1 0.0084
67.3 48.1 7.7 7.7
0
NS
0.2 10.2 10.2
NS
Men (n = 78)
Women (n = 52)
(%)
(%)
16.7 11.5 46.2 53.8 67.9 53.8 33.3 33.3
0.01
0.01
a NS: not significant. b Serum creatinine 2 0.115 mmol/l or urinary protein > 10 mg/dl. ’ On antihypertensive agent or blood pressure exceeding 140/90 or 160/95 mm Hg. d History or AA1 < 0.95. e History or Minnesota Code items 1.l, 1.2, or 7.1 (probable myocardial infarction). TABLE 6 Comparison
of blood pressure
Normal glucose tolerance Systolic Diastolic IGT Systolic Diastolic Diabetes Systolic Diastolic
(mm Hg, mean + SD) between Nisei men and women Men
Women
(n = 78) 135.5 + 19.2 79.4 * 10.5 (n = 74) 140.3 * 18.8 81.7 _+ 8.9 (n = 78) 142.1 k 16.5 81.6 f 8.1
(n = 72) 130.1 f 19.3 74.9 + 9.7 (n = 67) 134.0 * 17.3 76.4 f 8.9 (n = 52) 146.3 + 20.5 80.4 f 8.6
t
P"
1.73 2.68
0.086 0.008
2.06 3.52
0.042 0.00 1
- 1.3 0.83
NS NS
’ NS: not significant. Differences among groups of men: systolic, F = 2.65, P = 0.073; diastolic, F = 1.6, P = NS. Differences among groups of women: systolic, F = 11.6, P < 0.0001; diastolic, F = 5.6, P = 0.0043.
sclerosis obliterans detected by the Doppler method (AA1 < 0.95) was much less in women (3.1% vs 16.6%, x2 = 18.7, P-c 0.0001). Prevalence of a history of peripheral vascular disease was also lower in women (2.6% vs 9.2%, x* = 6.6, P = 0.01). Hypertension seemed to occur equally often in diabetic men and women. Whereas in the normal and IGT categories women had lower blood pressures than men, this protective effect of gender was lost for the diabetic subjects for whom systolic and diastolic blood pressures were not signifi-
cantly different (Table 6).
between
women
and
men
Discussion
The enrolled sample of Nisei women appears to be representative of the study population of Nisei women in King County, a conclusion based upon comparisons of the enrolled sample with the reference sample for age distribution, residential distribution, and family history diabetes. Further
127
comparisons of the sample of enrolled women with a similarly constructed sample of men, previously reported [ 11, for residential distribution, family history of diabetes, and geographic distribution of origin in Japan of parents also show no significant differences. As was the case for the Nisei men, the prevalence rates of diabetes and IGT for Nisei women were not directly ascertained in the study population, but rather were estimated from data obtained from the reference and enrolled study samples. This combination of a random (reference) sample and a volunteer (enrolled) sample allowed the use of a relatively small enrolled sample which could be intensively studied with comprehensive interviews and diagnostic tests. The data from these will be the subject of other reports. The prevalence of diabetes was estimated to be 16.1% for Nisei women 45-74 years, lower than the estimated prevalence of 20.0% for Nisei men [ 11. The estimated prevalence of IGT was 39.7;; for Nisei women compared to a prevalence of 36.1% for Nisei men [ 11. Although comparable data from Japan are limited, among 967 women in Tokyo > 40 yr old, the prevalence of diabetes (75-g OGTT, World Health Organization criteria) 4.5% compared to for was 2.5%, 3509 men > 40 years old [ 131. In the U.S. during the second National Health and Nutrition Examination Survey from 1976-1980 (NHANES II), the prevalence of diabetes in white women (physician-diagnosed medical history and results from a 75-g oral glucose tolerance test, World Health Organization criteria) was 8.7% at ages 45-54 years, 15.2% at 55-64years, and 17.0”; at 65-74years, compared to 7.8%, 9.40/,, and 19.1’; for white men in the same age groups [ 141. Thus diabetes prevalence rates for Nisei women seem to be similar to those for white women in the U.S., but much higher than those for Japanese women in Tokyo. Japanese women and JapaneseAmerican women have a lower prevalence of diabetes than men. U.S. white women have a higher prevalence of diabetes than men except in the oldest age group. In Tokyo, the prevalence of IGT among
women > 40 years old was 21.2”,, compared to 19.7% for men of the same age range [ 131. In NHANES II, the prevalence of IGT in white women in the U.S. was 14.5’?;, at ages 45-54 years, 13.7% at 55-64 years, and 23.0% at 65-74 years, compared to 12.6”;, 17.2%, and 22.8 % for white men in the same age groups [ 141. Thus, the prevalence rate of approximately 40% for IGT in Nisei women is much higher than among either Tokyo Japanese women or U.S. white women. Within each population, prevalence of IGT was similar for men and women. From these data we conclude that abnormal glucose tolerance (IGT and diabetes) is a frequent condition among older Japanese-American men and women. Over half of the self-reported nondiabetic women in the enrolled sample were subsequently diagnosed to have either IGT or diabetes, a finding similar to that previously made in Nisei men. For several of the complicating conditions associated with diabetes and IGT, prevalence was lower in Nisei women than found previously in Nisei men [ 21. The lower frequency of neuropathy may be due to the less severe hyperglycemia in diabetic Nisei women when compared to diabetic Nisei men. In addition, duration of diabetes was significantly shorter and age at onset of diabetes was significantly greater in women. Other studies have shown neuropathy to be associated with both the degree of hyperglycemia and the duration of diabetes [ 15-191. Our data suggest that diabetes may have a greater role as a risk factor for hypertension in Nisei women than men. Slightly higher systolic blood pressures in women than in men have been reported for diabetic Japanese women and men within the age range corresponding to the Nisei subjects of this study [20] as well as for diabetic Finnish women and men [21]. A recent analysis of data collected in the Framingham Heart Study has also shown that isolated systolic hypertension occurs more frequently in women than in men [22]. Arteriosclerosis obliterans was much less common in Nisei women than men. In Japan, the
128
prevalence of arteriosclerosis obliterans by a Doppler technique similar to that used in this study was about 2.5-fold higher in men than in women [23]. Beach and Strandness [24] have also reported an approximately 1.5-fold greater prevalence of arteriosclerosis obliterans in men than in women with non-insulin-dependent diabetes in a predominantly white population, using the identical procedure and criteria employed in our study. The observation of frequent abnormal electrocardiographic findings in the Nisei women was surprising and was quite discrepant from their medical history, in sharp contrast to the results in Nisei men [ 21. However, ischemic electrocardiographic changes have also been found more frequently in non-diabetic women than in nondiabetic men in Finland [21], while several other studies have suggested that ischemic electrocardiogram patterns in both resting and exercise electrocardiograms seem not to be well-correlated with the occurrence of clinical coronary heart disease in women [25-271. When medical history or electrocardiographic findings signifying ‘probable myocardial infarction’ are used to classify subjects, coronary heart disease is less frequent in diabetic Nisei women than men. Arteriosclerosis obliterans is also less frequent in Nisei women than men. There are several possible reasons for these gender-related differences which will be the subject of further investigation. In conclusion, Nisei men and women 45-74 yr old appear to have similar prevalence of glucose intolerance, which is less severe in women. Moreover, except for hypertension, complicating conditions are reduced in women.
Acknowledgements We thank our colleagues Drs L. Newell-Morris and W. Shuman for their support and assistance in the conduct of this research, and their helpful suggestions in the preparation of this report. We
gratefully acknowledge the skilled assistance provided by the nursing staff of the Clinical Research Center and the numerous staff members who performed the technical procedures. We are grateful to the King County Japanese American community for their support and cooperation and to the members of the Community Advisory Board for assistance. We are indebted to the many women who volunteered. This research was supported by National Institutes of Health Grant DK-31170 and by the facilities and services of the Diabetes Endocrinology Research Center (DK- 17047), the Clinical Nutrition Research Unit (DK-358 16) the University of Washington Medical Center General Clinical Research Center (RR-37), and the Northwest Lipid Research Center Lipoprotein Laboratory.
References Fujimoto, W.Y., Leonetti, D.L., Kinyoun, J.L., NewellMorris, L., Shuman, W.P., Stolov, W.C. and Wahl, P.W. (1987) Prevalence of diabetes mellitus and impaired glucose tolerance among second-generation Japanese-American men. Diabetes 36, 721-729. Fujimoto, W.Y., Leonetti, D.L., Kinyoun, J.L., Shuman, W.P., Stolov, W.C. and Wahl, P.W. (1987) Prevalence of complications among second-generation Japanese-American men with diabetes, impaired glucose tolerance, or normal glucose tolerance. Diabetes 36, 730-739. Bureau of the Census (1981) Census of the population: 1980. Races of the population by states, 1980. Supplementary report, July 1981. U.S. Govt. Printing Office, Washington, DC. Bureau of the Census (1973) Census of the population: 1970. Subject reports. National origin and language. PC(2)-IA. U.S. Govt. Printing Office, Washington, DC. Bureau of the Census (1973) Census of the population: 1970. Supplementary report 104. Races of the population by county: 1970. U.S. Govt. Printing Office, Washington, DC. Bureau of the Census (1973) Census of the population: 1970. Subject reports. Final report PC(Z)-1G. Japanese, Chinese, and Filipinos in the United States. U.S. Govt. Printing Office, Washington, DC. WHO Expert Committee on Diabetes Mellitus (1980) Second report. WHO tech. rep. ser. no. 646. WHO, Geneva. Early Treatment Diabetic Retinopathy Study (1980) Manual of operations. ETDRS Coordinating Center, Baltimore.
129 9 Rose, G.A. and Blackburn, H. (1968) Cardiovascular survey methods. WHO, Geneva. 10 Rose, G., Keen, H. and Jarrett, J. (1979) Epidemiologic methods in diabetic macrovascular disease. Diabetes Care 2, 91-97. I I Beach, K.W., Brunzell, J.D. and Strandness, D.E., Jr. (1982) Prevalence of severe arteriosclerosis obliterans in patients with diabetes mellitus. Relation to smoking and form of therapy. Arteriosclerosis 2, 275-280. 12 Chart, A., Beach, K.W., Martin, D.C. and Strandness, D.E., Jr. (1983) Carotid artery disease in NIDDM diabetes. Diabetes Care 6, 562-569. 13 Kitazawa, Y., Murakami, K., Goto, Y. and Hamazaki, S. (1983) Prevalence of diabetes mellitus detected by 75 g OGTT. Tohoku J. Exp. Med. 141 (Suppl.), 229-234. 14 Harris, ML, Hadden W.C., Knowler, W.C. and Bennett, P.H. (1987) Prevalence of diabetes and impaired glucose tolerance and plasma glucose levels in the U.S. population aged 20-74 yr. Diabetes 36, 523-534. 15 Tchobroutsky, G. (1978) Relation of diabetic control to development of microvascular complication. Diabetologia 15, 143-152. 16 Pirart, J. (1978) Diabetes mellitus and its degenerative complications: a prospective study of 4400 patients observed between 1947 and 1973. Diabetes Care 1, 168-188. 17 Pirart, J. (1978) Diabetes mellitus and its degenerative complications: a prospective study of 4400 patients observed between 1947 and 1973. Part 2. Diabetes Care 1, 252-263. 18 Skyler, J.S. (1979) Complications of diabetes melhtus: relationship to metabolic dysfunction. Diabetes Care 2, 499-509. 19 West, K.M., Ahuja, M.M.S., Bennett, P.H., Grab, B., Grabauskas, V.. Mateo-de-Acosta, O., Fuller. J.H., Jarrett, R.J., Keen, H., Kosaka, K., Krolewski, A.S., Miki, E., Schliack, V. and Teuscher, A. (1982) Interre-
20
21
22
23
24
25
26
27
lationships ofmicroangiopathy, plasma glucose and other risk factors in 3583 diabetic patients: a multinational study. Diabetologia 22, 412-420. Miki, E., Yoshizawa, K., Mizuno, Y.. Kosaka, K. and Kuzuya, N. (1983) On the characteristics of the Japanese in the WHO multinational study of vascular disease in diabetes (WHO MSVDD). Tohoku J. Exp. Med. I41 (Suppl.), 499-506. Uusitupa, M., Siitonen, O., Aro, A. and Pyorala, K. (1985) Prevalence ofcoronary heart disease, left ventricular failure and hypertension in middle-aged, newly diagnosed type 2 (non-insulin-dependent) diabetic subjects. Diabetologia 28, 22-27. Wjilking, S.V.B., Belanger, A., Kannel. W.B., D’Agostino, R.B. and Steel, K. (1988) Determinants of isolated systolic hypertension. J. Am. Med. Assoc. 260. 3451-3455. Seki, J., Ohashi, M., Sato, T., Yamamoto, M., Fujii, S. and Wada, M. (1983) Peripheral vascular disease in Japanese diabetics: screening by the Doppler ultrasonic technique. Tohoku J. Exp. Med. 141 (Suppl.), 499-506. Beach, K.W. and Strandness, D.E., Jr. (1980) Arteriosclerosis obliterans and associated risk factors in insulindependent and non-insulin-dependent diabetes. Diabetes 29. 882-888. Cumming, G.R., Dufresne, C., Kich. L. and Samm, J. (1973) Exercise electrocardiogram patterns in normal women. Br. Heart J. 35. 1055-1061. Linhart, J.W., Laws, J.G. and Satinsky, J.D. (1974) Maximum treadmill exercise electrocardiography in female patients. Circulation 50. 1173-l 178. Liao, Y., Liu, K., Dyer, A., Schoenberger, J.A., Shekelle. R.B., Colette, P. and Stamler, J. (1980) Major and minor electrocardiographic abnormalities and risk of death from coronary heart disease, cardiovascular diseases and all causes in men and women, J. Am. Coil. Cardiol. 12, 1494-1500.
119
DIABET 00505
Epidemiological
Note
Glucose intolerance and diabetic complications Japanese-American women W.Y. Fujimoto ‘, D.L. Leonetti2,
R.W. Bergstrom’, P.W. Wah15
J.L. Kinyoun3,
among
W.C. Stolov4 and
Departments of ‘Medicine, ‘Anthropology, ‘Ophthalmology, “Rehabilitation Medicine and ‘Biostatistics. Universit.v of Washington, Seattle, Washington, U.S.A. (Received 12 December 1990) (Revision received 4 February 1991) (Accepted 7 February 1991)
Summary The prevalence of glucose intolerance and diabetic complications was determined in second-generation Japanese-American (Nisei) women and compared to previously obtained results in Nisei men. A volunteer study sample of 191 Nisei women 45-74 years old was enrolled from a study population of 1489 Nisei women born 1913-1942, raised and educated in the U.S., and residing in King County, Washington. The enrolled sample included 72 with normal glucose tolerance, 67 with impaired glucose tolerance (IGT), and 52 with non-insulin-dependent diabetes. A random sample was also drawn from the study population to form a reference sample of 157 women. Based upon observations in the reference and enrolled samples, an estimated 16% of Nisei women in the study population have diabetes and 40% IGT. These rates compare to 20% diabetes and 36% IGT previously estimated for Nisei men 45-74 years old. The prevalence of cardiovascular disease (hypertension, peripheral vascular disease, and/or coronary heart disease) was highest among diabetic women, lowest in those with normal glucose tolerance, and intermediate in women with IGT. In comparison to diabetic men, there was a significantly lower frequency of neuropathy, peripheral vascular disease, and coronary heart disease in diabetic women. However, hypertension occurred equally often in both. Thus Japanese-American men and women 45-74 yr old have a similar prevalence of glucose intolerance, although less severe in women, and complications, except for hypertension, are reduced in women.
Key words: Coronary
heart disease; Complication; glucose tolerance; Japanese-American;
Correspondence to: W.Y. Fujimoto, University of Washington, Endocrinology and Nutrition, Seattle, WA 98195, U.S.A.
Diabetes; Peripheral
Department
Epidemiology; Hypertension; vascular disease; Women
of Medicine
RG-26, Division
Impaired
of Metabolism,
120
Introduction In an epidemiologic study of non-insulin-dependent diabetes conducted from 1983 to 1985 in a population of second-generation Japanese-American (Nisei) men who were between the ages 45 and 74 yr old and were residing in King County, Washington, an estimated 20% were diabetic while an additional 36% had impaired glucose tolerance (IGT) [ 11. The prevalence of cardiovascular disease (hypertension, coronary heart disease, and peripheral vascular disease) was highest in men with diabetes, lowest in those with normal glucose tolerance, and intermediate in those with IGT, while retinopathy and peripheral neuropathy were associated only with diabetes [2]. Since gender may modify the occurrence of glucose intolerance and its complicating conditions, we have extended the study to Nisei women 45 to 74 years old and compared the findings with those obtained in Nisei men.
Reference
sample
To provide sociodemographic information and an estimate of the prevalence of known diabetes in the study population, a reference sample was derived [ 11. A telephone survey was conducted in August 1983 of 405 households selected at random from a comprehensive directory of JapaneseAmerican households (6242 listings). There were 399 responding households containing 199 Nisei women. From this random telephone survey, a reference sample of 157 women was obtained by first identifying those Nisei women born in the U.S. between 1913 and 1942 (n = 181), followed by exclusion of Kibei (n = 24). Each respondent was asked whether she had ever been told by a physician that she had diabetes using the following wording: ‘diabetes/ blood too much sugar in the blood/high sugar/sugar in the urine’. The telephone survey also provided data on residential distribution and on family history of diabetes. Enrolled sample
Subjects and Methods Study population
Nisei include any Japanese person born in the U.S. of first-generation (Issei) immigrant parents who arrived in the U.S. before 1925, or of one Issei and one Nisei parent. We estimate that in King County, Washington, there were 1925 Nisei women born 1905-1944 among females of Japanese ancestry numbered in the 1980 census [3-61. A major portion (n = 1489) of this total population of Nisei women, born in the U.S. from 1913 through 1942 and educated in the U.S., comprised the study population. By selecting these birth years, the study population of Nisei women had the same age range as the earlier study sample of Nisei men born 1910-1939 [ 11. The study population excluded Nisei educated in Japan (Kibei) since education in the U.S. was an important aspect of Nisei life experience.
The enrolled sample is a volunteer group of 191 Nisei women (born 1913 through 1942 and excluding Kibei) who were examined from November 1986 to January 1988. Recruitment and screening was done as described previously for the Nisei men [ I]. Information about the project was disseminated throughout the Japanese-American community, followed by a call for volunteers. Using the same criteria as in the telephone survey quoted above, volunteers were asked to identify themselves as either diabetic or non-diabetic. Of 447 women who volunteered, 69 self-reported diabetes and 378 were self-reported non-diabetic. A short telephone interview was used to screen 306 of these volunteers (all 69 self-reported diabetic and 237 of the self-reported non-diabetic women) to determine their eligibility to participate. The self-reported diabetic women were screened first, and 46 (66.7 %) were enrolled. However, a final diagnosis of diabetes was not made until additional information was received, as described below. From the pool of 378 selfreported non-diabetic women, screening in-
121
terviews were carried out progressively by randomly selecting approximately 35 volunteers at a time until adequate sample sizes were achieved (approximately 70 subjects within the diagnostic categories of normal glucose tolerance and IGT). Ultimately 237 were screened, of whom 145 (61.2%) were enrolled. Excluded from participation in the study were women in whom any of the following were present: outside the age range (n = 23), Kibei (n = 20), raised outside the mainland U.S. (n = 9), not full J apanese (n = l), sister to a previously enrolled woman (n = 25), and potentially confounding medical conditions (n = 19). An additional 18 women who volunteered were not enrolled due to scheduling conflicts. The reasons for exclusion of volunteers were proportional between the two groups (self-reported diabetic or non-diabetic). Although sisters to women already enrolled were excluded in order to obtain independent observations on all female subjects, 26 Nisei women in the enrolled sample had a brother in the Nisei men’s enrolled study sample. For many of the comparisons between Nisei men and women, a second comparison was done that excluded men and women who were siblings. Class$cation ance status
of enrolled sample by glucose toler-
Classification of subjects was based upon a combination of medical information obtained from the subject and her physician, and a 75-g oral glucose tolerance test after a 10-h overnight fast, with findings assessed by World Health Organization [7] criteria for IGT (fasting serum glucose < 7.78 mmol/l and 2-h glucose 27.78 and < 11.11 mmol/l) and diabetes (fasting serum glucose 2 7.78 mmol/l or 2-h glucose 2 11.11 mmol/l) with normal glucose tolerance defined as absence of criteria for either IGT or diabetes (i.e., both fasting and 2-h serum glucose ~7.78 mmol/l). All subjects were on their usual diet before the glucose tolerance test. All medications usually taken in the morning, including oral hypoglycemic agents, were postponed until after the
test. Subjects receiving insulin reduced their morning dosage by approximately 33% on the day of the glucose tolerance test. Assessment of complicating conditions
An average blood pressure was calculated from three consecutive measurements by indirect auscultation and a mercury sphygmomanometer read to the nearest 2 mm Hg. Systolic blood pressure was determined by the first perception of sound and the diastolic (5th-phase) blood pressure by the disappearance of sound. All subjects underwent direct and indirect funduscopic examination of the eyes after mydriasis. In all subjects with diabetes and any subject with evidence of retinal pathology by ophthalmoscopy, standard photographs were taken of each fundus (7 standard stereoscopic fields) and retinopathy graded as P characteristics, modified from the Early Treatment Diabetic Retinopathy Study [ 81: mild non-proliferative without (PM) or with macular edema (P,), moderate non-proliferative (P,), severe non-proliferative (P2), and proliferative retinopathy (P3, P,, and Ps). A resting 12-lead electrocardiogram was done. Occurrence of Minnesota Code items 1. l-l .3 (abnormal Q and QS patterns), 5.1-5.3 (abnormalities of T-wave) and 7.1 (complete left bundle branch block) signified ischemic heart disease [9]. Code classes 1.1, 1.2, and 7.1 signified ‘probable myocardial infarction’, and code classes 1.3, 5.1, 5.2, and 5.3 signified ‘other suspect ischemia’ [lo]. Systolic blood pressure was measured by the Doppler method in both brachial arteries and in the anterior and posterior tibia1 arteries of both legs to calculate an ankle:arm index (AAI) for each ankle, with the higher of the two brachial pressures as the denominator and the lower of the two pressure determinations as the numerator for each ankle. AA1 < 0.95 signified occlusive arterial disease of the lower legs, and AA1 < 0.90 identified severe arteriosclerosis obliterans [ 111. Blood flow in the carotid arteries was assessed by ultrasonic duplex scanning of each carotid bifurcation, and the degree of carotid artery occlusive
122
disease was estimated from Doppler spectral wave forms [ 121. Subjects with 2 50% diameter reduction of the internal or common carotid artery were classified as having carotid artery disease. Inquiry was made about history of coronary heart disease (angina, myocardial infarction, coronary angioplasty, or coronary artery bypass graft) and peripheral vascular disease (stroke, transient ischemic attack, endarterectomy, intermittent claudication, aortic aneurysm, angioplasty, or peripheral arterial bypass surgery). This information was supplemented with the London School of Hygiene Cardiovascular Questionnaire [ 91. Nerve conduction studies of the right upper and lower extremities were done unless there was known to be a condition that might preferentially affect any of the nerves on that side, in which case studies were done on the left side. The following nerves and parameters were assessed: (1) lower extremity motor: peroneal nerve conduction velocity, distal latency, and evoked amplitude, and tibial nerve conduction velocity, distal latency, evoked amplitude, and H- and F-wave latencies; (2) lower extremity sensory: peroneai and sural nerve conduction velocity and evoked amplitude and sural nerve latency; (3) upper extremity motor: median and ulnar nerve forearm conduction velocity, distal latency, evoked amplitude, and F-wave latency; and (4) upper extremity sensory: median and ulnar nerve forearm conduction velocity, distal latency, and evoked amplitude. Neuropathy was present if conduction velocity or distal latency exceeded by > 2 SD the normal mean in at least three nerves. Fasting serum creatinine, 24-h urinary excretion of creatinine (creatinine clearance), and urinary protein were assessed. Creatinine was measured by the automated picric acid method and protein by trichloroacetic acid precipitation and photometry (University of Washington Medical Center, Department of Laboratory Medicine).
Statistical methods All group differences in measured characteristics were tested by parametric or non-parametric (Kruskal-Wallis test) analysis of variance procedures. Pairwise comparisons were evaluated by the two sample t-test, the Student-NewmanKeuls range test, or the non-parametric MannWhitney test for those characteristics having a non-normal distribution. Differences between diabetic men and women were also age-adjusted using analysis of covariance. Differences in categorical characteristics were evaluated by the Chisquare test.
Results Prevalence of known diabetes from the random telephone survey Prevalence of self-reported diabetes was 8.5 % for all 199 Nisei women respondents and 8.3% for the 157 reference sample women. These rates compared with 10.9% and 13.2% in similarly constructed samples of Nisei men [ 11. Volunteer response rates Based on the random survey, approximately 8.5 % or 164 of the total King County population of 1925 Nisei women would be expected to selfreport diabetes. The 69 volunteers with selfreported diabetes represented approximately 42% of this number. This response rate is less than the corresponding proportion for Nisei men (117 volunteered of an estimated 2 13 men, or a response rate of 55% ; x2 = 6.1, P < 0.02) [ 11. The 378 self-reported non-diabetic women volunteers represented approximately 21 y0 of the remaining estimated 1761 King County Nisei women expected to self-report absence of diabetes. This response rate was identical to the percentage who volunteered of the estimated 1738 Nisei men in King County expected to self-report the absence of diabetes [ 11.
123
Comparison of the enrolled sample of Nisei women to the reference sample of Nisei women and to the enrolled sample of Nisei men To ascertain that the enrolled sample was representative of the study population, several characteristics were compared between the enrolled and reference sample Nisei women, both of which exclude Kibei. Separate comparisons were made for self-reported diabetic and self-reported nondiabetic women. In addition, comparisons were also made between the enrolled sample of Nisei women and the enrolled sample of Nisei men examined earlier and found to be representative of the Nisei male study population [ 11. All of these comparisons were not significantly different, as summarized in Table 1. TABLE I Summary of X-square test statistics used in comparisons of characteristics of the enrolled sample of Nisei women to the reference sample of Nisei women and to the enrolled sample of Nisei men
Comparison of enrolled sample sample of Nisei women Age distribution Residential distribution” Positive family history of diabete? Comparison of enrolled sample sample of Nisei men’ Residential distribution Positive family history of diabetes Prefectural origin in Japan of parent?
Self-reported status
diagnostic
Diabetic
Non-diabetic
of Nisei women to reference 0.03 0.54
0.42 1.25
0.05 3.13 of Nisei women to enrolled 0.90 (0.40)
0
(0.03)
2.08 (1.54)
1.37 (0.30)
0
0.30
Corresponding P values were all not significant (P > 0.05). ” Ethnic areas of Seattle vs non-ethnic areas of Seattle and its suburbs. ’ History of diabetes in a parent or full sibling. c x7 value for comparison when siblings were removed is shown in parentheses. d All of Kyushu Island and the prefectures of Hiroshima and Yamaguchi vs all other areas of Japan. In the parallel study of Nisei men the geographic distribution of parental origins ofthe enrolled sample was not significantly different from that of the reference sample [I].
TABLE 2 Final diagnostic classification status of Nisei women compared with self-reported entry status Self-reported entry status
Final diagnostic classification status
Number (“,)
Non-diabetic (n = 145)
Normal IGT Diabetes Normal* IGT* Diabetes
69 61 15 3 6 37
Diabetic (w = 46)
* None were receiving treatment
(47.6) (42.1) (10.3) (6.5) (13.0) (80.4)
directed towards diabetes.
Diagnostic class@ation of enrolled sample The final diagnostic classification of self-reported nondiabetic and diabetic women in the enrolled sample is known in Table 2. The enrolled sample consisted of 72 women with normal glucose tolerance, 67 with IGT, and 52 with diabetes. Of the 52 diabetic women, 30 were receiving no medication for diabetes (14 of whom had entered the study as self-reported non-diabetic subjects); 20 were being treated with oral hypoglycemic agents and 2 with insulin. Two of the women on oral agents had been on insulin previously. Both of the women who were receiving insulin had a history consistent with non-insulin-dependent diabetes. Eighteen (34.5%) of the diabetic women had fasting serum glucose levels 27.78 mmol/l, one of whom was insulin-treated. All had 2-h serum glucose 2 11.11 mmol/l. Of the remaining diabetic women, 33 had 2-h serum glucose 2 11.11 mmol/l while one had a 2-h glucose of 6.89 mmol/l while receiving glyburide. Among diabetic women, there was a highly significant 2 7.78 association between fasting glucose mmol/l and treatment with either oral agents or insulin; 77.80,;, of the 18 women with fasting glucose 2 7.78 mmol/l were receiving medications for diabetes compared with only 23.5 Y0of the 34 women with fasting glucose < 7.78 mmol/l (x2 = 12.1, P = 0.0005).
124
Age of enrolled sample by diagnostic group The median age of all subjects was 63.3 years and the mean age ( i SD) was 62.7 k 6.2 years. Mean ages ( + SD) were similar among the three final diagnostic groups: 61.9 + 5.9 for normal women, 62.7 + 6.4 for IGT women, and 63.7 + 6.2 for diabetic women (F = 1.39, P = 0.25). For Nisei men, mean ages were 61.6 years for normal men, 61.3 years for IGT men, and 61.6 years for diabetic men [ 11.
diabetes in the study population was estimated to be 10.3x, based upon the proportion of selfreported non-diabetic enrolled sample women found to have diabetes. From these figures, the derived estimated prevalence rate of diabetes, both previously diagnosed and confirmed plus previously undiagnosed among Nisei women 45-74 years old (excluding Kibei) in King County is 16.1% (confidence interval 14.3%-17.9x), based upon the calculation: 0.067 + 0.103 x (1 - 0.083). This rate compared with 20.0% (confidence interval 18.0%-22.0%) for men 45-74 years old [ 11.
Estimated prevalence rate of diabetes in the study population Of the reference sample, 8.3 y0 self-reported previously diagnosed diabetes. However, 19.6% of self-reported diabetic women in the enrolled sample were not confirmed to have diabetes by oral glucose tolerance test. Therefore, the corrected prevalence of previously diagnosed diabetes in the reference sample was estimated to be 6.7% which was also assumed to be the prevalence of previously diagnosed diabetes in the study population. The prevalence of undiagnosed
Estimatedprevalence rate of IGT in the studypopulation In the enrolled sample, 13.0% of self-reported diabetic women and 42.1% of the self-reported non-diabetic women were found to have IGT. The derived estimated prevalence rate of IGT among Nisei women 45-74 years old (excluding Kibei) in King County, both undiagnosed plus self-reported as diabetes, is 39.7% (confidence
TABLE 3 Frequency
of complicating
conditions
in Nisei women by diagnostic Normal (n = 72)
Microangiopathy or neuropathy* Hypertension On antihypertensive medication and/or: Blood pressure > 140/90 Blood pressure > 160/95 On antihypertensive medication History of known hypertension Peripheral vascular diseaseb Coronary heart disease’ Cardiovascular diseased
category: number of subjects (%)
IGT (n = 67)
Diabetes (n = 52)
X2
P
5 (6.9)
3 (4.5)
15 (28.8)
19.2
0.0001
25 (34.7) 16 (22.2)
31 (46.3) 18 (26.9)
35 (67.3) 25 (48.1)
12.9 10.2
0.002 0.006
16 (22.2)
13 (19.4)
20 (38.5)
6.29
0.04
22 (30.6)
20 (29.9)
27 (51.9)
7.74
0.021
2 (2.8) 14 (19.4) 27 (37.5)
13 (19.4) 24 (35.8) 35 (52.2)
8 (15.4) 19 (36.5) 36 (69.2)
8.20 5.98 12.2
0.017 0.05 0.0022
a At least one of the following: serum creatinine 2 0.115 mmol/l (above the reference range for the laboratory), urinary protein excretion > 10 mg/dl (1 SD beyond the mean for normal subjects), presence of retinopathy, or electrophysiologic evidence for peripheral neuropathy. b History, AA1 < 0.95, or evidence of carotid artery disease by Doppler. c History or electrocardiographic evidence by Minnesota Code. d At least one ofthe following: on antihypertensive medication, blood pressure > 140/90, peripheral vascular disease, or coronary heart disease.
125
interval 36.6% -42.9%), derived from the calculation: 0.13 x (0.083) + 0.421 x (1 - 0.083). This rate compared with 36.1 y0 (confidence interval 33.7”;-38.5%) for men 45-74 years old [ 11.
in 1 normal woman, 6 IGT women, and 3 diabetic women. Minnesota Code items 1.1, 1.2, or 7.1 indicating ‘probable myocardial infarction’ were found in none of the normal women, 4 IGT women and 2 diabetic women. Minnesota Code items 1.3, 5.1, 5.2 and 5.3 indicating ‘other suspect ischemia’ were found in 13 or 18.17; of normal women, 15 or 22.4% of IGT women, and 16 or 30.8% of diabetic women.
Frequency of complicating conditions in Nisei women
Complicating conditions showing significant differences by diagnostic category are shown in Table 3. Mean creatinine clearance, serum creatinine, and concentration of protein in a morning urine specimen were all not significantly different by diagnostic category. Retinopathy was observed in only 10 women with diabetes; 8 had mild and 1 moderate non-proliferative retinopathy and 1 had proliferative retinopathy. Electrophysiologic evidence of peripheral neuropathy was observed in 4 women with diabetes and 1 with IGT. A history of peripheral vascular disease was obtained from none of the normal women, 3 IGT women, and 2 diabetic women. AA1 < 0.95 was measured in 1 normal woman, 1 IGT woman, and 4 diabetic women (3 of whom had AA1 < 0.90). Evidence of carotid artery disease by Doppler was present in 1 normal woman, none of the IGT women, and 1 diabetic woman. A history of coronary heart disease was present
Comparison of diabetic Nisei women and men
Table 4 compares diabetic Nisei women in the present study and the previously reported diabetic Nisei men [l] for several characteristics relevant to their diabetes, most of which indicate less severe diabetes in women. Table 5 compares prevalence of complicating conditions associated with diabetes. Diabetic neuropathy was much less frequent in women. Prevalence of peripheral vascular disease (history or AA1 < 0.95) or coronary heart disease (history or electrocardiographic abnormalities limited to Minnesota Code item 1.1, 1.2 or 7.1 or ‘probable myocardial infarction’) was significantly less in diabetic women than in diabetic men. In the entire enrolled study sample of 19 1 Nisei women and 229 Nisei men, prevalence of arterio-
TABLE 1 Comparison
of diabetic Nisei men and women (mean + SD)
Variable
Age W-1
Age at diagnosis of diabetes (yr) Duration of diabetes (yr) Fasting glucose (mmol/l) 2-h glucose (mmol/l) 9,” Fasting glucose 2 7.78 mmol/l O0 Oral agents or insulin .I NS, not significant. ’ Mann-Whitney Z.
P”
Age-adjusted
Men (n = 78)
Women (n = 52)
rlx2
61.7 + 6.0
63.7
+ 6.2
- 1.85
0.066
55.2 f 7.5
59.6 + 8.9
- 2.97
0.004
0.043
6.5 f 6.7
4.7 + 6.7
2.07b
0.038
0.048
9.54 + 3.21 18.17 f 5.23
7.83 f 2.86 16.96 f 5.52
3.41b 1.31b
0.0007 NS
0.003 NS
60.3
34.5
7.2
0.007
48.7
42.3
0.29
NS
P”
126 TABLE 5 Comparison
of frequency of complications
Complication
Nephropathyb Retinopathy Neuropathy Any of the above Hypertension’ Blood pressure > 140/90 Blood pressure > 160/95 Peripheral vascular diseased e Coronary heart disease
between diabetic Nisei men and women ( %)
x2
P”
13.5 19.2 7.7 28.8
0.1 0.9 19.9 6.9
NS NS < 0.000 1 0.0084
67.3 48.1 7.7 7.7
0
NS
0.2 10.2 10.2
NS
Men (n = 78)
Women (n = 52)
(%)
(%)
16.7 11.5 46.2 53.8 67.9 53.8 33.3 33.3
0.01
0.01
a NS: not significant. b Serum creatinine 2 0.115 mmol/l or urinary protein > 10 mg/dl. ’ On antihypertensive agent or blood pressure exceeding 140/90 or 160/95 mm Hg. d History or AA1 < 0.95. e History or Minnesota Code items 1.l, 1.2, or 7.1 (probable myocardial infarction). TABLE 6 Comparison
of blood pressure
Normal glucose tolerance Systolic Diastolic IGT Systolic Diastolic Diabetes Systolic Diastolic
(mm Hg, mean + SD) between Nisei men and women Men
Women
(n = 78) 135.5 + 19.2 79.4 * 10.5 (n = 74) 140.3 * 18.8 81.7 _+ 8.9 (n = 78) 142.1 k 16.5 81.6 f 8.1
(n = 72) 130.1 f 19.3 74.9 + 9.7 (n = 67) 134.0 * 17.3 76.4 f 8.9 (n = 52) 146.3 + 20.5 80.4 f 8.6
t
P"
1.73 2.68
0.086 0.008
2.06 3.52
0.042 0.00 1
- 1.3 0.83
NS NS
’ NS: not significant. Differences among groups of men: systolic, F = 2.65, P = 0.073; diastolic, F = 1.6, P = NS. Differences among groups of women: systolic, F = 11.6, P < 0.0001; diastolic, F = 5.6, P = 0.0043.
sclerosis obliterans detected by the Doppler method (AA1 < 0.95) was much less in women (3.1% vs 16.6%, x2 = 18.7, P-c 0.0001). Prevalence of a history of peripheral vascular disease was also lower in women (2.6% vs 9.2%, x* = 6.6, P = 0.01). Hypertension seemed to occur equally often in diabetic men and women. Whereas in the normal and IGT categories women had lower blood pressures than men, this protective effect of gender was lost for the diabetic subjects for whom systolic and diastolic blood pressures were not signifi-
cantly different (Table 6).
between
women
and
men
Discussion
The enrolled sample of Nisei women appears to be representative of the study population of Nisei women in King County, a conclusion based upon comparisons of the enrolled sample with the reference sample for age distribution, residential distribution, and family history diabetes. Further
127
comparisons of the sample of enrolled women with a similarly constructed sample of men, previously reported [ 11, for residential distribution, family history of diabetes, and geographic distribution of origin in Japan of parents also show no significant differences. As was the case for the Nisei men, the prevalence rates of diabetes and IGT for Nisei women were not directly ascertained in the study population, but rather were estimated from data obtained from the reference and enrolled study samples. This combination of a random (reference) sample and a volunteer (enrolled) sample allowed the use of a relatively small enrolled sample which could be intensively studied with comprehensive interviews and diagnostic tests. The data from these will be the subject of other reports. The prevalence of diabetes was estimated to be 16.1% for Nisei women 45-74 years, lower than the estimated prevalence of 20.0% for Nisei men [ 11. The estimated prevalence of IGT was 39.7;; for Nisei women compared to a prevalence of 36.1% for Nisei men [ 11. Although comparable data from Japan are limited, among 967 women in Tokyo > 40 yr old, the prevalence of diabetes (75-g OGTT, World Health Organization criteria) 4.5% compared to for was 2.5%, 3509 men > 40 years old [ 131. In the U.S. during the second National Health and Nutrition Examination Survey from 1976-1980 (NHANES II), the prevalence of diabetes in white women (physician-diagnosed medical history and results from a 75-g oral glucose tolerance test, World Health Organization criteria) was 8.7% at ages 45-54 years, 15.2% at 55-64years, and 17.0”; at 65-74years, compared to 7.8%, 9.40/,, and 19.1’; for white men in the same age groups [ 141. Thus diabetes prevalence rates for Nisei women seem to be similar to those for white women in the U.S., but much higher than those for Japanese women in Tokyo. Japanese women and JapaneseAmerican women have a lower prevalence of diabetes than men. U.S. white women have a higher prevalence of diabetes than men except in the oldest age group. In Tokyo, the prevalence of IGT among
women > 40 years old was 21.2”,, compared to 19.7% for men of the same age range [ 131. In NHANES II, the prevalence of IGT in white women in the U.S. was 14.5’?;, at ages 45-54 years, 13.7% at 55-64 years, and 23.0% at 65-74 years, compared to 12.6”;, 17.2%, and 22.8 % for white men in the same age groups [ 141. Thus, the prevalence rate of approximately 40% for IGT in Nisei women is much higher than among either Tokyo Japanese women or U.S. white women. Within each population, prevalence of IGT was similar for men and women. From these data we conclude that abnormal glucose tolerance (IGT and diabetes) is a frequent condition among older Japanese-American men and women. Over half of the self-reported nondiabetic women in the enrolled sample were subsequently diagnosed to have either IGT or diabetes, a finding similar to that previously made in Nisei men. For several of the complicating conditions associated with diabetes and IGT, prevalence was lower in Nisei women than found previously in Nisei men [ 21. The lower frequency of neuropathy may be due to the less severe hyperglycemia in diabetic Nisei women when compared to diabetic Nisei men. In addition, duration of diabetes was significantly shorter and age at onset of diabetes was significantly greater in women. Other studies have shown neuropathy to be associated with both the degree of hyperglycemia and the duration of diabetes [ 15-191. Our data suggest that diabetes may have a greater role as a risk factor for hypertension in Nisei women than men. Slightly higher systolic blood pressures in women than in men have been reported for diabetic Japanese women and men within the age range corresponding to the Nisei subjects of this study [20] as well as for diabetic Finnish women and men [21]. A recent analysis of data collected in the Framingham Heart Study has also shown that isolated systolic hypertension occurs more frequently in women than in men [22]. Arteriosclerosis obliterans was much less common in Nisei women than men. In Japan, the
128
prevalence of arteriosclerosis obliterans by a Doppler technique similar to that used in this study was about 2.5-fold higher in men than in women [23]. Beach and Strandness [24] have also reported an approximately 1.5-fold greater prevalence of arteriosclerosis obliterans in men than in women with non-insulin-dependent diabetes in a predominantly white population, using the identical procedure and criteria employed in our study. The observation of frequent abnormal electrocardiographic findings in the Nisei women was surprising and was quite discrepant from their medical history, in sharp contrast to the results in Nisei men [ 21. However, ischemic electrocardiographic changes have also been found more frequently in non-diabetic women than in nondiabetic men in Finland [21], while several other studies have suggested that ischemic electrocardiogram patterns in both resting and exercise electrocardiograms seem not to be well-correlated with the occurrence of clinical coronary heart disease in women [25-271. When medical history or electrocardiographic findings signifying ‘probable myocardial infarction’ are used to classify subjects, coronary heart disease is less frequent in diabetic Nisei women than men. Arteriosclerosis obliterans is also less frequent in Nisei women than men. There are several possible reasons for these gender-related differences which will be the subject of further investigation. In conclusion, Nisei men and women 45-74 yr old appear to have similar prevalence of glucose intolerance, which is less severe in women. Moreover, except for hypertension, complicating conditions are reduced in women.
Acknowledgements We thank our colleagues Drs L. Newell-Morris and W. Shuman for their support and assistance in the conduct of this research, and their helpful suggestions in the preparation of this report. We
gratefully acknowledge the skilled assistance provided by the nursing staff of the Clinical Research Center and the numerous staff members who performed the technical procedures. We are grateful to the King County Japanese American community for their support and cooperation and to the members of the Community Advisory Board for assistance. We are indebted to the many women who volunteered. This research was supported by National Institutes of Health Grant DK-31170 and by the facilities and services of the Diabetes Endocrinology Research Center (DK- 17047), the Clinical Nutrition Research Unit (DK-358 16) the University of Washington Medical Center General Clinical Research Center (RR-37), and the Northwest Lipid Research Center Lipoprotein Laboratory.
References Fujimoto, W.Y., Leonetti, D.L., Kinyoun, J.L., NewellMorris, L., Shuman, W.P., Stolov, W.C. and Wahl, P.W. (1987) Prevalence of diabetes mellitus and impaired glucose tolerance among second-generation Japanese-American men. Diabetes 36, 721-729. Fujimoto, W.Y., Leonetti, D.L., Kinyoun, J.L., Shuman, W.P., Stolov, W.C. and Wahl, P.W. (1987) Prevalence of complications among second-generation Japanese-American men with diabetes, impaired glucose tolerance, or normal glucose tolerance. Diabetes 36, 730-739. Bureau of the Census (1981) Census of the population: 1980. Races of the population by states, 1980. Supplementary report, July 1981. U.S. Govt. Printing Office, Washington, DC. Bureau of the Census (1973) Census of the population: 1970. Subject reports. National origin and language. PC(2)-IA. U.S. Govt. Printing Office, Washington, DC. Bureau of the Census (1973) Census of the population: 1970. Supplementary report 104. Races of the population by county: 1970. U.S. Govt. Printing Office, Washington, DC. Bureau of the Census (1973) Census of the population: 1970. Subject reports. Final report PC(Z)-1G. Japanese, Chinese, and Filipinos in the United States. U.S. Govt. Printing Office, Washington, DC. WHO Expert Committee on Diabetes Mellitus (1980) Second report. WHO tech. rep. ser. no. 646. WHO, Geneva. Early Treatment Diabetic Retinopathy Study (1980) Manual of operations. ETDRS Coordinating Center, Baltimore.
129 9 Rose, G.A. and Blackburn, H. (1968) Cardiovascular survey methods. WHO, Geneva. 10 Rose, G., Keen, H. and Jarrett, J. (1979) Epidemiologic methods in diabetic macrovascular disease. Diabetes Care 2, 91-97. I I Beach, K.W., Brunzell, J.D. and Strandness, D.E., Jr. (1982) Prevalence of severe arteriosclerosis obliterans in patients with diabetes mellitus. Relation to smoking and form of therapy. Arteriosclerosis 2, 275-280. 12 Chart, A., Beach, K.W., Martin, D.C. and Strandness, D.E., Jr. (1983) Carotid artery disease in NIDDM diabetes. Diabetes Care 6, 562-569. 13 Kitazawa, Y., Murakami, K., Goto, Y. and Hamazaki, S. (1983) Prevalence of diabetes mellitus detected by 75 g OGTT. Tohoku J. Exp. Med. 141 (Suppl.), 229-234. 14 Harris, ML, Hadden W.C., Knowler, W.C. and Bennett, P.H. (1987) Prevalence of diabetes and impaired glucose tolerance and plasma glucose levels in the U.S. population aged 20-74 yr. Diabetes 36, 523-534. 15 Tchobroutsky, G. (1978) Relation of diabetic control to development of microvascular complication. Diabetologia 15, 143-152. 16 Pirart, J. (1978) Diabetes mellitus and its degenerative complications: a prospective study of 4400 patients observed between 1947 and 1973. Diabetes Care 1, 168-188. 17 Pirart, J. (1978) Diabetes mellitus and its degenerative complications: a prospective study of 4400 patients observed between 1947 and 1973. Part 2. Diabetes Care 1, 252-263. 18 Skyler, J.S. (1979) Complications of diabetes melhtus: relationship to metabolic dysfunction. Diabetes Care 2, 499-509. 19 West, K.M., Ahuja, M.M.S., Bennett, P.H., Grab, B., Grabauskas, V.. Mateo-de-Acosta, O., Fuller. J.H., Jarrett, R.J., Keen, H., Kosaka, K., Krolewski, A.S., Miki, E., Schliack, V. and Teuscher, A. (1982) Interre-
20
21
22
23
24
25
26
27
lationships ofmicroangiopathy, plasma glucose and other risk factors in 3583 diabetic patients: a multinational study. Diabetologia 22, 412-420. Miki, E., Yoshizawa, K., Mizuno, Y.. Kosaka, K. and Kuzuya, N. (1983) On the characteristics of the Japanese in the WHO multinational study of vascular disease in diabetes (WHO MSVDD). Tohoku J. Exp. Med. I41 (Suppl.), 499-506. Uusitupa, M., Siitonen, O., Aro, A. and Pyorala, K. (1985) Prevalence ofcoronary heart disease, left ventricular failure and hypertension in middle-aged, newly diagnosed type 2 (non-insulin-dependent) diabetic subjects. Diabetologia 28, 22-27. Wjilking, S.V.B., Belanger, A., Kannel. W.B., D’Agostino, R.B. and Steel, K. (1988) Determinants of isolated systolic hypertension. J. Am. Med. Assoc. 260. 3451-3455. Seki, J., Ohashi, M., Sato, T., Yamamoto, M., Fujii, S. and Wada, M. (1983) Peripheral vascular disease in Japanese diabetics: screening by the Doppler ultrasonic technique. Tohoku J. Exp. Med. 141 (Suppl.), 499-506. Beach, K.W. and Strandness, D.E., Jr. (1980) Arteriosclerosis obliterans and associated risk factors in insulindependent and non-insulin-dependent diabetes. Diabetes 29. 882-888. Cumming, G.R., Dufresne, C., Kich. L. and Samm, J. (1973) Exercise electrocardiogram patterns in normal women. Br. Heart J. 35. 1055-1061. Linhart, J.W., Laws, J.G. and Satinsky, J.D. (1974) Maximum treadmill exercise electrocardiography in female patients. Circulation 50. 1173-l 178. Liao, Y., Liu, K., Dyer, A., Schoenberger, J.A., Shekelle. R.B., Colette, P. and Stamler, J. (1980) Major and minor electrocardiographic abnormalities and risk of death from coronary heart disease, cardiovascular diseases and all causes in men and women, J. Am. Coil. Cardiol. 12, 1494-1500.