Journal of Internal Medicine 1997; 242: 299–305

Social and biological predictors of early menopause: a model for premature aging P. NILSSON a , L. MÖLLER b , A. KÖSTER b & H. HOLLNAGEL c From the aDepartment of Community Health Sciences, University of Lund, Lund, Sweden; bCentre for Preventive Medicine, Glostrup University Hospital, Glostrup, and cDepartment of General Practice, Institute of Public Health, University of Copenhagen, Copenhagen, Denmark

Abstract. Nilsson P, Möller L, Köster A, Hollnagel H (University of Lund, Lund, Sweden; Glostrup University Hospital, Glostrup, and University of Copenhagen, Copenhagen; Denmark). Social and biological predictors of early menopause: a model for premature aging. J Intern Med 1997; 242: 299]305. Objectives. To investigate possible social, lifestylerelated and biological predictors of early menopause in middle-aged women, followed prospectively for 11 years. Design. A prospective, population-based, cohort follow-up, observational study. Setting. Glostrup Hospital, Copenhagen, Denmark. Subjects. A total of 493 female subjects, all aged 40 years at baseline, and divided into three groups according to self-reported menopausal age (40]45, 46]51, 511 years), after 12 months of amenorrhoea. Women having had medical or surgical interventions to influence menopausal state were excluded. Main outcome measures. Body mass index, glucose, insulin, lipids, creatinine, uric acid, thyroid-stimulating hormone (TSH), lung function tests (forced VC, FEV1, peak flow), blood pressure; a self-administered

Introduction A process of premature (neuroendocrine) ageing might represent the biological mechanism linking psychosocial stress, low social class, and an adverse lifestyle, with an increased overall morbidity and mortality [1]3]. A possible model for this concept of premature ageing is early menopause, but not that caused by specific medical conditions, e.g. genetic syndromes or surgical interventions such as salpingo-oophorectomy. © 1997 Blackwell Science Ltd

questionnaire with questions on psychosocial variables, lifestyle, and self-rated health. Results. An early menopausal age correlated in an univariate way with impaired lung function, increased smoking habits and low social class (in childhood or present), as well as with a feeling of tiredness, all measured at the baseline investigation. On the contrary, a later menopausal age correlated with higher serum insulin and uric acid levels. In multiple regression analysis, with menopausal age as the dependent variable, it was found that smoking habits (number of years smoking) was inversely (P , 0.001), and insulin as well as uric acid were positively (P , 0.05) correlated with menopausal age. Conclusions. Females who smoke run an increased risk of early menopause, whereas relative hyperinsulinaemia is independently associated with later menopause. At the age of 40 years, high insulin levels in females might be just a marker for normal female sex hormone physiology, not for insulin resistance, as seen in postmenopausal female subjects. Early menopause might be useful as a potential model of premature ageing. Keywords: ageing, insulin, menopause, smoking, social class, uric acid.

Based on epidemiological studies, early, or premature, menopause, has been associated with a rapid increase in cardiovascular risk factors [4], an increased risk of myocardial infarction [5], and an increased cardiovascular [6] as well as overall mortality [7, 8]. The mechanism of these outcomes is supposed to be the decline in female sex hormones seen in the postmenopausal state, so it is possible that postmenopausal hormonal replacement therapy (HRT) might show beneficial effects, especially in 299

300

P. NILSSON et al.

women experiencing early menopause. However, so far no randomized, controlled studies have reported on the role of HRT for preventing cardiovascular and overall mortality, so the possible benefits of HRT are judged solely from a large number of observational studies [9, 10]. Furthermore, women with an early menopause have also reported an increase, compared to women with a normal age distribution of menopause, in premenopausal episodes of clinical depression [11]. This means that neuroendocrine effects of depression, or other factors linked to depression such as lifestyle (e.g. smoking) or medication, may impair female hormonal regulation and decrease sex hormone levels, thus eventually triggering the onset of an early (premature) menopause, as well as bone loss [12]. Different stress factors might be relevant to this concept, and might be related to immunomodulatory processes associated with premature menopause [13]. The opposite chain of events might sometimes also be true in that a subclinical loss of sex hormones in the years prior to early menopause might manifest itself as depression. This is, however, less probable in the majority of women according to a recent review which stated that there is insufficient evidence at present to maintain that menopause causes depression [14]. Smoking is by far the most well-documented lifestyle factor associated with early menopause, but several other predictors have also been discussed [15]18]. Some studies, but not all, have shown an association between early menopause and low social class, not confounded by smoking habits [19, 20]. Certainly hereditary factors [21] and the number of lifetime ovulations [22, 23] are also of importance for the prediction of early menopause. Most studies in this area are, however, cross-sectional by design and only few provide prospective data. In the Scandinavian countries the median age for menopause is now approximately 51 years, and a considerable proportion of women are taking HRT in their first few postmenopausal years, and some for even longer periods. The prevalence of female smoking, however, differs amongst the Scandinavian countries, with the highest smoking rate recorded in Denmark [24]. The aim of this study was therefore to evaluate possible social, lifestyle-related and biological predictors of early menopause in a sample of Danish women, prospectively followed from the age of 40 to

the age of 51 years. Our hypothesis was that females experiencing worse psychosocial conditions and lifestyle habits would be prone to develop an early (premature) menopause compared to females with better lifestyle conditions.

Subjects and methods Subjects A representative cohort of 40-year-old women (n 5 630), born in 1936 and living in four municipalities served by Glostrup Hospital, in the county of Copenhagen, Denmark, has been followed since 1976 as part of an epidemiological survey of both males and females [25]. The original participation rate was 87%, corresponding to 548 women. In 1987]88 the health examination was repeated [26], with a net participation rate of 78% of the original cohort. In total, 493 women participated in both examinations. Of these women 271 were self-reported postmenopausal (amenorrhoea . 12 months) and 232 still menstruating at the age of 51 years. Of the postmenopausal women, 78 had experienced a surgical intervention causing menopause, and were thus excluded, as were menstruating females on HRT (n 5 62). Four women reported a very early menopause, before the age of 40 years, and were excluded. In the final analyses, therefore, 189 postmenopausal and 160 still premenopausal, naturally menstruating females were included. The women were divided for further comparisons into three groups: those menopausal at the age of (a) 40]45 years (n 5 23); (b) 46]51 years (n 5 166); and (c) 511 years (n 5 160). Methods All subjects were invited to come in the morning between 08.00 and 10.00 hours, after a fast for 12 h, and had two blood samples taken. The blood samples were immediately centrifuged and serum was frozen to ]208C. A physical examination was carried out with the measurement of body weight (kg) and height (m). The body mass index BMI (kg/m2) was calculated. Blood pressure (mmHg) was determined as the mean of two measurements after supine rest for 10 min, and a procedure standardized according to WHO criteria [26]. A 12 3 33 cm cuff was used, and all measurements were done by one and the same person. © 1997 Blackwell Science Ltd Journal of Internal Medicine 242: 299–305

PREDICTORS OF EARLY MENOPAUSE

The blood samples were analysed for lipoprotein lipids (total cholesterol, HDL cholesterol, triglycerides), insulin, glucose, creatinine, uric acid, and thyroid-stimulating hormone (TSH). Serum total and high density lipoprotein (HDL) cholesterol levels were determined enzymatically at the Glostrup Hospital laboratory (whose results are standardized against the Prague laboratory). Serum insulin concentrations were measured using radioimmunoassay by means of a double antibody technique [27]. The intra-assay and interassay variation was about 5 and 9%, respectively. All other laboratory variables were measured by routine methods at Glostrup Hospital. The recordings of lung function were made using a Godar spirometer, and peak expiratory flow rate was measured with a Wright McKerrow peak flowmeter. As a criterion for the correct performance, at least two measurements differing by , 5% from each other had to be produced. The largest volume was used in the analysis. The forced vital capacity (VC; L), the forced expiratory volume during one second (FEV1; L), and the peak expiratory flow (PEF; L min21) were registered. After the clinical procedures all participants were asked to complete a self-administered questionnaire with items on psychosocial and lifestyle variables. These questions covered family social background, present social class (stratas I]V), working conditions, cohabitation, consumption of tobacco (g day21) and alcohol (drinks per week; one drink 5 one glass of wine, 0.15 L, or beer, 0.33 L), physical activity, and personal-related physical problems (e.g. abdominal pain) and mental problems (e.g. insomnia and nervousness), as previously described in detail [26, 28, 29]. Subjects also reported their recalled weight and height at 30 years, so that BMI could be calculated based on these variables. The distribution of social class at 40 years was: social class I (highest; 2%), II (9%), III (28%), IV (26%), and V (35%). The overall proportion of smokers was 49%, ex-smokers 9%, and nonsmokers 42%. Statistics All continuous variables are presented as means (standard deviation; SD). A test for trend (Kruskal]Wallis one-way ANOVA) was used for comparing groups of women at different menopausal age intervals. Spearman’s linear correlation (r) was used for bivariate correlations between continuous vari© 1997 Blackwell Science Ltd Journal of Internal Medicine 242: 299–305

301

ables, and a chi-squared test for comparison of proportions. Finally, a multiple regression analysis was applied, with age at menopause as the dependent variable and, in different models, physiological (insulin, uric acid, BMI, lung function), lifestyle (smoking, alcohol consumption), and social variables (social class in childhood and present) as the independent variables. A P-value below 0.05 was considered to be significant.

Results The subjects’ characteristics of biological variables, lifestyle factors and social background, at the age of 40 years, according to menopausal age are given in Tables 1 and 2. The calculated BMI at the age of 30 and 40 years showed no group differences. Women with a later menopause showed higher levels of insulin and uric acid, as well as a better expiratory lung function at the age of 40 years, than did women with an early menopause. No group differences were seen in lipid or glucose levels. A strong significant correlation (P , 0.001) was found between increasing smoking habits, as measured by years of smoking, mean daily tobacco consumption, or total lifetime tobacco consumption, and early menopause. No group differences were found for total alcohol intake or coffee. A higher intake of wine was correlated with higher social class (P , 0.001) in trend analysis, e.g. social classes I 1 II (2.7), III (2.5), IV (1.6), and V (1.3 glasses per week). A weaker correlation (P , 0.05) existed between early menopause and low social class as well as subjective feeling of tiredness, Table 3. In a set of multiple regression analyses, menopausal age was used as the dependent variable, and selected variables from various domains as independent variables, e.g. biological (insulin, uric acid, BMI, lung function), lifestyle (smoking, physical activity), and social class (in childhood and present). Finally, two regression models were tested with the inclusion number of years of smoking, insulin, uric acid, lung function, social class, and without (model 1), or with (model 2) BMI at 40 years as independent variables, Table 4. It was found that higher insulin and uric acid levels were independently correlated with a later menopausal age. The number of years of smoking was independently correlated with menopausal age in an inverse way. Social class (child-

302

P. NILSSON et al.

Table 1 Biological characteristics at baseline in 40-year-old females according to menopausal age intervals. Means (SD). P-value for trend (Kruskal-Wallis test) Menopausal age groups

Weight (kg) Height (m) BMI 40 years BMI 30 years SBP (mmHg) DBP (mmHg) Cholesterol (mmol L21) HDL-C (mmol L21) Triglycerdies (mmol L21) Glucose (mmol L21) Insulin (mU L21) Uric acid (mg L21) Creatinine (mg L21) TSH (mU L21) FEV1(L) Peak flow (L min21) Forced VC (L)

40–45 years (n 5 23)

46–51 years (n 5 166)

511 years (n 5 160)

P-value

058.6 (8.1) 163.6 (7.8) 022.0 (3.0) 021.0 (2.7) 114.7 (13.0) 065.6 (8.3)

062.1 (9.0) 164.3 (6.4) 023.0 (3.1) 021.8 (2.8) 116.8 (15.3) 067.3 (10.8)

063.0 (11.2) 164.4 (6.3) 023.3 (4.2) 022.2 (3.4) 116.5 (12.1) 067.7 (11.2)

0.25 0.88 0.71 0.43 0.10 0.75

5.82 (0.71) 0.91 (0.20) 0.98 (0.54) 4.87 (0.78) 8.3 (2.3) 40.7 (7.7) 80.4 (12.2) 2.4 (1.3)

6.02 (1.02) 0.90 (0.19) 0.95 (0.45) 4.85 (1.21) 9.6 (2.6) 44.1 (9.6) 80.9 (14.7) 3.0 (3.5)

6.12 (1.06) 0.92 (0.19) 0.91 (0.40) 4.85 (0.64) 10.0 (3.1) 45.2 (9.5) 81.1 (13.3) 2.8 (1.9)

0.70 0.94 0.65 0.79 0.05 0.04 0.82 0.78

2.10 (0.67) 376 (60) 2.84 (0.47)

2.35 (0.49) 399 (61) 3.04 (0.55)

2.45 (0.55) 410 (67) 3.06 (0.58)

0.008 0.06 0.20

BMI, body mass index (kg/m2), at 40 years and 30 years (recalled); SBP, systolic blood pressure (mmHg); DBP, diastolic blood pressure (mmHg).

hood or present) was not an independent predictor for menopausal age after controlling for the other variables in the model.

ables at the age of 40. smoking habits were an especially strong predictor of early menopause, as previously shown [15, 16]. Amongst the different smoking variables in our study, the total number of self-reported years of smoking was the strongest predictor. This might have some connection with the age at which smoking began. For many women this is in their early teens, which is a vulnerable period for possible adverse effects of nicotine on the ovarian function and other hormonal mechanisms. Smoking today is

Discussion The main finding of this study was that the transition into postmenopause, in a cohort of Danish women followed prospectively for 11 years was predicted by a number of biological, lifestyle-related and social vari-

Table 2 Lifestyle variables and psychosocial characteristics at baseline in 40-year-old females according to menopausal age groups. Means (SD). P-value for trend, using a Kruskal-Wallis test (for continuous variables), and chi-squared test (for proportions) Menopausal age groups 40–45 years (n 5 23)

46–51 years (n 5 166)

511 years (n 5 160)

P-value

Tobacco (g day21) Tobacco (kg/lifetime) Tobacco (years of smoking) Alcohol (drinks week21) Coffee (cups day21)

10.1 (7.9) 82.2 (63.4) 16.5 (9.4) 03.8 (3.7) 04.6 (3.4)

07.5 (8.2) 60.5 (67.9) 11.3 (10.4) 03.7 (5.1) 04.9 (3.6)

04.5 (6.9) 39.0 (60.0) 07.7 (9.6) 04.2 (5.0) 04.6 (3.5)

0.002 0.0004 0.001 0.66 0.21

Low social class, IV–V present (%) childhood (%) Non-cohabiting (%)

70 58 09

62 38 11

54 23 07

0.05 0.0008 0.19

© 1997 Blackwell Science Ltd Journal of Internal Medicine 242: 299–305

PREDICTORS OF EARLY MENOPAUSE

same finding for high fasting insulin levels might represent only a marker for increased weight, but on the other hand it has also been shown that hyperinsulinaemia might preserve the bone structure in females [30]. This finding might be linked to the influence of menopausal state and female sex hormones on insulin]glucose metabolism [31], as well as bone metabolism [32]. Laakso has previously reported that fasting hyperinsulinaemia is a good marker for increased insulin resistance in nondiabetic subjects [33], but this was based on findings in males and females of the age group 45]65 years, thus including mostly postmenopausal females. However, it is more probable that a relative hyperinsulinaemia in premenopausal females (of 40 years) represents a preserved normal beta-cell function, and might, therefore, be a marker for normal levels and normal physiology of female sex steroids, as extensively reviewed by Godsland [10]. This impression is also strengthened by the findings of a recent populationbased study in which it was shown that both absolute and age-adjusted plasma insulin levels were in fact higher in premenopausal than in postmenopausal women [34]. Furthermore, it has been well documented that premenopausal women are not insulinresistant [10], as compared to males or postmenopausal females.

Table 3 Bivariate correlations between age at menopause and selfreported symptoms as well as selected psychosocial variables in 40-year-old women (Spearman’s test) Variable

P-value

Subjective tiredness Feeling of stress Abdominal pain Low social class present in childhood Non-cohabitation Leisure time physical activity Use of contraceptive pills Ever pregnant

0.04 0.64 0.56

303

0.04 0.02 0.20 0.82 0.84 0.46

unequally distributed in the female population and is more prevalent in lower social classes, as well as in young women with a less well established social network, e.g. lonely mothers. It is therefore possible that such women will be more prone to experience early menopause in the future if they tend to start smoking early and continue for long periods of their life. Poor lung function and an increased subjective feeling of tiredness were predictors of early menopause, and both may be influenced by smoking habits. Conversely it was found that good lung function was associated with a later menopausal age. The

Table 4 Multiple regression analyses with age at menopause as the dependent variable, and selected biological, lifestyle, and psychosocial factors as independent variables Model 1 R2 5 0.11, d.f. 5 5 Variable

B

SE B

Beta

T

P-value

No. of years of smoking Insulin Uric acid FEV1 Social class Constant

20.01 20.03 20.01 20.01 20.06 22.83

0.003 0.01 0.003 0.01 0.04 0.28

20.19 20.11 20.12 20.09 20.11 2 10.1

23.6 22.1 22.3 21.7 21.5 20.0001

0.004 0.039 0.022 0.099 0.14

Variable

B

SE B

Beta

T

P-value

No. of years of smoking Insulin Uric acid FEV1 Social class BMI (40 years) Constant

20.01 20.03 20.01 20.01 20.06 20.007 22.84

0.003 0.01 0.004 0.01 0.04 0.01 0.32

20.19 20.11 20.12 20.09 20.11 20.004 28.8

23.6 22.0 22.2 21.6 21.5 20.07 20.0001

0.004 0.05 0.03 0.10 0.14 0.94

Model 2 R2 5 0.11, d.f. 5 6

© 1997 Blackwell Science Ltd Journal of Internal Medicine 242: 299–305

304

P. NILSSON et al.

The association between elevated levels of uric acid and later menopause is unexplained. Uric acid metabolism is linked to diet, obesity, insulin and various disease states [35]. The way in which it might interact with the transition into postmenopause should be the aim for other future studies. It has been suggested that premature ageing is the common biological pathway for the influence of adverse psychosocial factors on morbidity and mortality [1, 36], and thus might influence longevity [37]. This has been supported by the finding that poor lung function, independent of smoking habits, is linked to low social class and a strong predictor of morbidity and mortality due to many different causes, not just pulmonary or cardiovascular diseases [38]40]. In our previous cross-sectional study in middle-aged Danish males from the same area, poor lung function was associated with a sex hormone profile resembling male ageing independent of smoking and alcohol intake [3]. We therefore conclude that in both males and females, representative of an urban Danish geographical area (Glostrup), subjects experiencing the interaction of worse psychosocial and lifestyle factors might be prone to a process of premature ageing. This is noted in the impact on biological variables, e.g. lung function. In the future, follow-up studies might shed light on the consequences of this process on overall morbidity and mortality. Such studies are currently under way as the subjects are going to be reinvestigated at the age of 60 years. Possible limitations of the present study include the limited number of women and the fact that they only represent a North European, Caucasian, urban population. Therefore, predictors of early menopause might prove to be different in other ethnic populations where neither reproductive life patterns and lifestyle, nor the perception of menopausal symptoms, are the same [41]. Malnutrition and a high number of pregnancies may negatively influence the health of many women in developing countries, possibly causing premature menopause, but no systematic studies in this area have been found by us. A considerable methodological problem is how to delineate social and lifestyle factors from biological effects in a causal chain of events. For example, in this study social class was not independently associated with early menopause, but smoking is such a strong marker for social class, and measured with stronger precision, that ‘smoking’ will often tend to

remain in statistical models when ‘social class’ is eliminated. That social class and psychosocial stress really are important predictors of an early menopause, at least in some females, was recently shown in a group of 185 healthy US women followed for 9 years [42]. In conclusion, females who smoke run an increased risk of early menopause, whereas females with good lung function tend to have a later menopause. A relative hyperinsulinaemia is independently associated with later menopause. At the age of 40 years, high insulin levels in females might just be a marker for normal female sex hormone physiology, however, not for insulin resistance, as seen in postmenopausal females. Taken together, these findings might have important implications for differences in the current and future cardiovascular risk of individuals. Early menopause might thus be seen as an example of premature ageing linked to the possible interaction of adverse biological, lifestyle-related and psychosocial background factors, e.g. poor lung function, smoking and low social class. Hormonal replacement therapy and smoking cessation may prove to be of significant clinical importance in reducing the cardiovascular risk in this group of women.

Acknowledgements This study was supported by grants from The Danish Heart Foundation and from The Danish Health Insurance Foundation.

References 1 Nilsson PM. Premature ageing: the link between psychosocial risk factors and disease. Med Hypothes 1996; 47: 39]42. 2 Björntorp P. Neuroendocrine ageing. J Intern Med 1995; 238: 401]4. 3 Nilsson P, Möller L, Solstad K. Adverse effects of psychosocial stress on gonadal function and insulin levels in middle-aged males. J Intern Med 1995; 237: 479]86. 4 Matthews KA, Meilahn E, Kuller LH, Kelsey SF, Caggiula AW, Wing RW. Menopause and risk factors for coronary heart disease. N Engl J Med 1989; 321: 641]6. 5 Palmer JR, Rosenberg L, Shapiro S. Reproductive factors and risk of myocardial infarction. Am J Epidemiol 1992; 136: 408]16. 6 van der Schouw YT, van der Graaf Y, Steyerberg EW, Eijkemans MJ, Banga JD. Age at menopause as a risk factor for cardiovascular mortality. Lancet 1996; 347: 714]8. 7 Snowdon DA, Kane RL, Beeson WL, Burke GL, Sprafka JM, Potter J, et al. Is early menopause a biological marker for © 1997 Blackwell Science Ltd Journal of Internal Medicine 242: 299–305

PREDICTORS OF EARLY MENOPAUSE

health and ageing? Am J Public Health 1989; 79: 709]14. 8 Sowers MR, LaPietra MT. Menopause: its epidemiology and potential association with chronic diseases. Epidem Rev 1995; 17: 287]302. 9 Johannes CB, Crawford SL, Posner JG, McKinlay SM. Longitudinal patterns and correlation of hormone replacement therapy use in middle-aged women. Am J Epidemiol 1994; 140: 439]52. 10 Godsland IF. The influence of female sex steroids on glucose metabolism and insulin action. J Intern Med 1996; 240 (Suppl. 738): 1]60. 11 Harlow BL, Cramer DW, Annis KM. Association of medically treated depression and age at natural menopause. Am J Epidemiol 1995; 141: 1170]76. 12 Michelson D, Stratakis C, Hill L, Reynolds J, Galliven E, Chrousos G, et al. Bone mineral density in women with depression. N Engl J Med 1996; 335: 1176]81. 13 Letur-Kornish H, Raoul-Duval A, Cabau A, Bomsel-Helmreich O, al-Mufti W, Homo-Delarche F. Stress and premature menopause. Can R Acad Sci 1995; 318: 691]8. 14 Nicol-Smith L. Causality, menopause, and depression: a critical review of the literature. Br Med J 1996; 313: 1229]32. 15 Jick H, Porter J. Relation between smoking and age at natural menopause. Lancet 1977; i: 1354]5. 16 Cramer DW, Harlow BL, Xu H, Fraer C, Barbieri B. Cross-sectional and case]controlled analyses of the association between smoking and early menopause. Maturitas 1995; 22: 79]87. 17 Cramer DW, Xu H. Predicting age at menopause. Maturitas 1996; 23: 319]26. 18 Vermeulen A. Environment, human reproduction, menopause, and andropause. Envir Health Perspectives 1993; 101 (Suppl. 2):91]100. 19 Brambilla DJ, McKinlay SM. A prospective study of factors affecting age at the menopause. J Clin Epidemiol 1989; 42: 1031]9. 20 Luoto R, Kaprio J, Uutela A. Age at natural menopause and sociodemographic status in Finland. Am J Epidemiol 1994; 139: 64]76. 21 Cramer DW, Xu H, Harlow BL. Family history as a predictor of early menopause. Fertil Steril 1995; 64: 740]45. 22 Ginsburg J. What determines the age at the menopause? The number of ovarian follicles seems the most important factor (Editorial). Br Med J 1991; 302: 1288]9. 23 Cramer DW, Xu H, Harlow BL. Does ‘incessant’ ovulation increase risk for early menopause? Am J Obstet Gynecol 1995; 172: 568]73. 24 Chener L, Osler M, McKee M, Krasnik A. Changing life expectancy in the 1980s: why was Denmark different from Sweden? J Epidemiol Community Health 1996; 50: 404]7. 25 Hagerup L, Eriksen M, Schroll M, Hollnagel H, Agner E, Larsen S. The Glostrup Population Studies collection of epidemiological tables. Scand J Soc Med 1981; Suppl. 20. 26 Moller L, Kristensen TS, Hollnagel H. Social class and cardiovascular risk factors in Danish men. Scand J Soc Med 1991; 19: 116]26. 27 Hales CN, Randle PJ. Immunoassay of insulin with insuli-anti-

© 1997 Blackwell Science Ltd Journal of Internal Medicine 242: 299–305

305

body precipitate. Biochem J 1963; 88: 137]9. 28 Moller L, Kristensen TS. Plasma fibrinogen and ischemic heart disease risk factors. Arteriosclerosis and Thrombosis 1991; 11: 344]50. 29 Moller L, Kristensen TS, Hollnagel H. Self rated health as a predictor of coronary heart disease in Copenhagen, Denmark. J Epidemiol Community Health 1996; 50: 1]6. 30 Barrett-Connor E, Kritz-Silverstein D. Does hyperinsulinemia preserve bone? Diab Care 1996; 19: 1388]92. 31 Walton C, Godsland I, Proudler A, Wynn V, Stevenson J. The effects of the menopause on insulin sensitivity, secretion and elimination in nonobese, healthy women. Eur J Clin Invest 1993; 23: 466]73. 32 Vico L, Prallet B, Chappard D, Pallot-Prades B, Pupier R, Alexandre C. Contributions of chronological age, age at menarche and menopause and of anthropometric parameters to axial and peripheral bone densities. Osteoporosis Int 1992; 2: 153]8. 33 Laakso M. How good a marker is insulin levels for insulin resistance? Am J Epidemiol 1993; 137: 959]65. 34 Pasquali R, Vicennati V, Bertazzo D, Casimirri F, Pascal G, Tortelli O, et al. Determinants of sex hormone-binding globulin blood concentrations in premenopausal and postmenopausal women with different estrogen status. Metabolism 1997; 46: 5]9. 35 Sagginai F, Pilati S, Targher G, Branzi P, Muggeo M, Bonora E. Serum uric acid and related factors in 500 hospitalized subjects. Metabolism 1996; 45: 1557]61. 36 Ballinger S. Stress as a factor in lowered estrogen levels in the early postmenopause. Ann NY Acad Sci 1990; 592: 95]113. 37 Christensen K, Vaupel JW. Determinants of longevity: genetic, environmental and medical factors. J Intern Med 1996; 240: 331]41. 38 Persson C, Bengtsson C, Lapidus L, Rybo E, Thiringer G, Wedel H. Peak expiratory flow and risk of cardiovascular disease and death: a 12-year follow-up of participants in the population study of women in Gothenburg, Sweden. Am J Epidemiol 1986; 124: 942]8. 39 Burchfiel CM, Marcus EB, Sharp DS, Enright PL, Rodriguez BL, Masaki KH, et al. Characteristics associated with rapid decline in forced expiratory volume. Ann Epidemiol 1996; 6: 217]27. 40 Hole DJ, Watt GC, Davey-Smith G, Hart CL, Gillis CR, Hawthorne VM. Impaired lung function and mortality risk in men and women: findings from the Renfrew and Paisley population study. Br Med J 1996; 313: 711]15. 41 Maoz B, Antonovsky A, Apter A, Wijesenbeck H, Datan N. The perception of menopause in five ethnic groups in Israel. Acta Obstet Gynecol Scand 1977; Suppl. 65:69]76. 42 Bromberger JT, Matthews KA, Kuller LH, Wing RR, Meilahn EN, Plantinga P. Prospective study of the determinants of age at menopause. Am J Epidemiol 1997; 145: 124]33. Received 17 February 1997; accepted 21 May 1997. Correspondence: P. Nilsson, MD, PhD, Department of Community Health Sciences, Helgeandsgatan 16, S-223 54 Lund, SWEDEN (fax: 146 4617 6428).