Total cholesterol concentration and mortality at a relatively young age: Do men and women differ?BMJ 1995; 311 doi: https://doi.org/10.1136/bmj.311.7008.779 (Published 23 September 1995) Cite this as: BMJ 1995;311:779
- a Department of Chronic Diseases and Environmental Epidemiology, National Institute of Public Health and Environmental Protection, PO Box 1, 3720 BA Bilthoven, Netherlands
- Correspondence to: Dr Verschuren.
- Accepted 14 July 1995
Objective: To investigate the relation between total cholesterol concentration and mortality from coronary heart disease, cardiovascular diseases, non-cardiovascular causes, and all causes.
Design: Population based cohort study.
Subjects: 23000 men and 26000 women aged 30-54 years examined between 1974 and 1980.
Main outcome measures : Mortality for the above mentioned end points for fifths of cholesterol distribution, and relative risks estimated by using Cox's proportional hazard (survival) analysis. Adjustment was made for age, smoking, systolic blood pressure, and body mass index.
Results: Mortality from coronary heart disease in men was five times higher than that in women. A strong positive association between total cholesterol concentration and mortality from coronary heart disease and cardiovascular diseases was observed in both men and women. The relative risk for the highest compared with the lowest fifth of the cholesterol distribution was for mortality from coronary heart disease (3.0 (95% confidence interval 1.8 to 5.1) in men and 3.8 (1.1 to 13.1) in women) and for mortality from cardiovascular disease (2.8 (1.8 to 4.2) in men and 2.9 (1.4 to 6.0) in women). No increase of non-cardiovascular mortality at low cholesterol concentration was observed. All cause mortality was significantly higher in the highest compared with the lowest fifth of the cholesterol distribution: relative risk 1.6 (1.3 to 2.0) in men and 1.5 (1.1 to 1.9) in women.
Conclusion: Total cholesterol concentration is a strong predictor of mortality from coronary heart disease, cardiovascular diseases, and all causes in women as well as in men. Low cholesterol concentrations are not associated with increased mortality from non-cardiovascular causes.
Although relative risks for mortality from coronary heart disease were similar in men and women, the absolute risk of death in women was about one fifth that in men
No significant excess mortality from non- cardiovascular causes was observed at low cholesterol concentrations
Total mortality in the highest fifth of the cholesterol distribution was 60% higher in men and 50% higher in women compared with the lowest fifth
There is growing awareness that cardiovascular diseases are not only an important public health problem in men but also in women.1 2 In the Netherlands about 40% of total mortality is caused by cardiovascular diseases, both in men and women.3
Because most research on cardiovascular disease has been carried out on middle aged men longitudinal data for women are still relatively scarce. In the Netherlands a screening project on risk factors for cardiovascular disease has been carried out between 1974 and 1980 in which 50000 men and women aged 30-54 years were examined.4 Mortality follow up of this cohort has recently been completed, and to our knowledge this is one of the largest cohorts that provides information on total cholesterol concentration in relation to mortality in women.
Although it has been shown in men that the relation between total cholesterol concentration and mortality from coronary heart disease is strong and graded,5 6 the risk of low concentrations is debatable.7 There are indications that the benefit with respect to cardiovascular mortality would be offset by an increase in mortality from cancer and other diseases.6 8 Therefore we also investigated the associations between total cholesterol concentration and non-cardiovascular mortality.
Subjects and methods
Population--The consultation bureau project on cardiovascular diseases was carried out from 1974 to 80 in five towns in the Netherlands: Amsterdam, Doetinchem, Maastricht, Leiden, and Tilburg.4 The response rate varied from 70 to 80%. The project was aimed primarily at the age group of around 40 years, but in some towns a wider age range was taken. The age range of the population examined was 30 to 54 years, with about 15% aged 30-34 years, 45% aged 35-39 years, 30% aged 40-44 years, and 10% aged 45-54 years.
Examination--Information about smoking habits and cardiovascular complaints and the respondents' history of myocardial infarction, stroke, hypertension, and diabetes mellitus was obtained from questionnaires. Blood pressure, weight, and height were measured. A non-fasting blood sample was taken in which total cholesterol concentration was measured (with an accuracy of 0.1 mmol/l) at the Central clinical chemistry laboratory of the University Hospital Dijkzigt in Rotterdam. This laboratory participated in the standardisation programme of the World Health Organisation (WHO). Total cholesterol concentration was determined according to a direct Liebermann-Burchard method.9 In more recent projects cholesterol concentration has been determined enzymatically, and cut off points for hypercholesterolaemia according to the Netherlands Cholesterol Consensus are based on enzymatic values.10 Therefore cholesterol values were converted based on a comparison study carried out in 1981 in which cholesterol was measured with both methods in over 5000 serum samples. The following equation for conversion from Huang to enzymatic values was obtained11: total cholesterolenzymatic=1.004* total cholesterolHuang-0.299 mmol/l.
Mortality follow up--Mortality follow up was initiated in 1986 and completed in 1993. The administration cards of the project were retrieved from the archives and entered on to a computer. The vital status was checked through the municipal registries. Census date was the date on which the information was obtained from the municipal registry (for the living) or date of death. The minimum length of follow up was therefore six years (person examined in 1980 and checked for vital status in 1986) and the maximum 19 years (person examined in 1974 and checked for vital status in 1993). The total number of person years was over 275000 in men and over 300000 in women; mean (SD) duration of follow up was 11.8 (2.3) years for men and 12.0 (2.2) years for women. If a person had moved to an unknown destination the date on which the person was dropped from the municipal registry was used as census date. For 49202 of the 50887 people examined the administration card was still present in the archive. For 49018 of these mortality follow up was successfully completed, resulting in a loss to follow up of only 3.7%. A total of 1319 people had died. For 1289 subjects the cause of death was obtained from the Central Bureau of Statistics, but for 30 people who died outside the Netherlands such information could not be obtained. Causes of death were coded according to the ninth revision of the International Classification of Diseases (ICD-9). For deaths that had occurred before 1 January 1979 (n=89) the eighth revision (ICD-8) was used. For the end points used in the present analyses all codes from ICD-8 remained in the same category when defined according to ICD-9. Cause specific end points used in the analyses were coronary heart disease (ICD 410-414), cardiovascular disease (ICD 390-459), cancer (ICD 140-239), and non-cardiovascular and non-cancer (ICD 0-139, 240-389, 460-999).
Statistical methods--SAS statistical programs were used for the analyses.12 Cholesterol concentration was divided into sex specific fifths and into categories of 0.5 mmol/l width. Cox's proportional hazards (survival) analysis was used to calculated hazard ratios. These hazard ratios provide a good estimate of the more commonly used relative risk and will be referred to as such in this paper. Adjustment for smoking was made based on a dichotomous variable “current smoking: yes or no” whereas age, systolic blood pressure, and body mass index (weight (kg)/height (m)2 were entered into the model as continuous variables. Trend tests were performed for both linear and non-linear (J shaped or U shaped) associations, but there were no significant differences in fit between linear and non-linear models. In the tables the P values for the linear trend tests are presented. For cause specific analyses subjects with an unknown cause of death were excluded. Exclusion of prevalent cases of angina, stroke, and myocardial infarction (n=1349) did not substantially alter the results, and therefore they were included in the analyses.
The overall mean age was 39 years: the total cholesterol concentration was 5.55 mmol/l in men and 5.19 mmol/l in women. For all cardiovascular risk factors levels were higher in men than in women (table I). In men cardiovascular diseases, cancer, and noncancer or non-cardiovascular diseases each contributed about a third to total mortality. In women cancer was the predominant cause of death (table II).
Mean total cholesterol concentration per fifth of the distribution ranged from 4.18 mmol/l in the lowest to 7.14 mmol/l in the highest fifth in men and from 3.94 mmol/l to 6.69 mmol/l in women (table III).
For coronary heart disease, mortality in men was five times that in women. The highest compared with the lowest fifth showed a threefold higher risk in men and an almost fourfold higher risk in women after adjustment for age, smoking, blood pressure, and body mass index (table IV). For cardiovascular disease the highest compared with the lowest fifth showed an almost threefold higher risk in both men and women.
In both men and women no associations between fifths of the cholesterol distribution and mortality from cancer or non-cardiovascular or non-cancer causes were observed (table IV). Also no associations with site specific cancers nor with subgroups of non-cardiovascular or non-cancer causes mentioned in table II were observed (results not shown).
All cause mortality was significantly higher in the highest fifth of the cholesterol distribution by about 60% in men and 50% in women (table IV). To explore the shape of the relation between total cholesterol concentration and all cause mortality in more detail cholesterol concentration was divided into nine categories with 0.5 mmol/l steps (figure). The middle category (5.5-5.99 mmol/l) was used as the reference category. A total cholesterol concentration of 7.5 mmol/l or more gave a relative risk of 2.3 (95% confidence interval 1.8 to 3.0) in men and a relative risk of 1.8 (1.2 to 2.8) in women (table V). Risk increased continuously with increasing cholesterol concentration in women. In men the lowest all cause mortality was observed in the category 4.5-4.99 mmol/l. The relative risk for a total cholesterol concentration below 4.0 mmol/l compared with a concentration of 5.5-5.99 mmol/l in men was 1.18 (0.81 to 1.71).
During the follow up period only 38 women died of coronary heart disease compared with 194 men, showing that in middle age coronary heart disease is already a prominent cause of death in men but not yet in women. The association between total cholesterol concentration and mortality from coronary heart disease was comparable in men and women: risk was higher above the fourth quintiles, although in women risk above the fifth quintile only was significant. The relative risk for the highest compared with the lowest fifth was 3.0 in men and 3.8 in women. A slightly higher relative risk for women compared with men has been observed in other studies as well.1 13 In all studies, however, absolute risk was lower in women than in men.
In the present study a single measurement of total cholesterol concentration was obtained, leading to underestimation of the relative risk, a phenomenon often referred to as regression dilution bias.14 Law et al estimated the magnitude of the dilution factor to be 1.4.15 When we apply this dilution factor to the relative risk for the highest fifth in the present study the observed
The relative risks for coronary heart disease in women in the present study were high compared with those reported in other studies, even after adjustment for age, smoking, blood pressure, and body mass index.1 13 This might be due to the fact that the respondents were relatively young at the time of examination: three quarters were aged 35-45 years. Several studies have shown that the relative risk is higher the younger the age at examination.16 17 18 Law et al estimated that after 13 years of follow up a 0.6 mmol/l decrease in total cholesterol concentration was associated with a 31% decrease in mortality from coronary heart disease in men aged 35-44 at baseline, whereas in men aged 55-64 this decrease was only 12%.15
The Dutch cholesterol consensus applies the same cut off point for hypercholesterolaemia (total cholesterol >/=6.5 mmol/l) to men and women.10 Because concentrations of high density lipoprotein cholesterol are on average 0.25 mmol/l higher in women than in men,19 this cut off point is in fact more strict for women. In the present study the risk of coronary heart disease in women with cholesterol concentrations above 5.9 mmol/l was almost four times that in women with concentrations below 4.3 mmol/l. Therefore, the consensus criteria seem justified. As absolute risk is much lower in women compared with men, however, the decision to treat hypercholesterolaemia in women has to take the absolute risk (and therefore also other risk factors for coronary heart disease besides total cholesterol) into account.
Recently, concern has been raised about an increased risk of mortality at the lower end of the cholesterol distribution.7 Several studies have reported an increased risk for non-cardiovascular mortality at a low cholesterol concentration or even inverse associations over the entire cholesterol range.8 20 21 22 23 24 25 In many studies these associations remained after exclusion of the first few years of follow up, contradicting the hypothesis that the associations were due to preexisting diseases at baseline.8 20 23 24 25 In most studies in which these observations were found study subjects were middle aged men. It has been suggested that in “healthy” cohorts consisting of working people27 or relatively young people,27 these associations are not present. In the present study of relatively young people no association was observed between total cholesterol concentration and mortality from cancer and non-cardiovascular or non-cancer causes. Also for site specific cancers (digestive, respiratory, and breast) and for subgroups of non-cardiovascular or non-cancer mortality (respiratory diseases, digestive diseases, and accidents) no association with total cholesterol concentration was observed. Our data thus support the healthy cohort hypothesis.
In men and women all cause mortality was increased by about 60% and 50%, respectively, in the highest compared with the lowest fifth of the cholesterol distribution. To examine the possibility of an increase in all cause mortality at the lower end of the cholesterol distribution in more detail smaller categories than fifths were used. In women all cause mortality increased continuously with increasing cholesterol concentration. In men all cause mortality was lowest at a cholesterol concentration of 4.5-4.9 mmol/l, but the difference with mortality at cholesterol concentrations below 4.5 mmol/l was not significant. In the multiple risk factor intervention trial all cause mortality at cholesterol concentrations below 4.1 mmol/l was increased in men.8 In a recent meta-analysis all cause mortality was increased at low cholesterol concentrations in men but not in women.7 In the national health and nutrition examination survey (NHANES I) epidemiological follow up study all cause mortality at low cholesterol concentrations was increased in both men and women, but this increase was significant only in those aged over 60 at baseline.27 Our results support the hypothesis that the relation between total cholesterol concentration and mortality is dependent on age at baseline.
The present study shows that total cholesterol concentration is a strong risk factor for coronary heart disease and all cause mortality in men as well as in women. No significant increase in all cause mortality at low concentrations was observed. Our results add to the evidence that increased non-cardiovascular mortality at low cholesterol concentrations is limited to subjects who are middle aged or older at baseline.
We thank Mrs Rontgen-Pieper for her invaluable secretarial help with the mortality follow up and Statistics Netherlands for providing data on causes of death.
Funding Former Dutch Ministry of Public Health and Environmental Protection.
Conflict of interest None.