No previous researchers have sought to determine whether high-density lipoprotein (HDL) cholesterol levels are associated with survival to 85 years of age in a prospective cohort of aging men. We selected 652 men (mean age 65 years) enrolled in the VA Normative Aging Study who had ≥1 HDL cholesterol level documented during the study and who were old enough on the date of HDL cholesterol measurement to reach 85 years of age by the end of follow-up (July 1, 2008). We categorized initial HDL cholesterol into <40 mg/dl (reference group), 40 to 49 mg/dl, or ≥50 mg/dl. Information on co-morbidities, lifestyle factors, measured lipid parameters, and medications were collected during triennial visits. We used proportional hazards to determine hazard ratios (HRs) for mortality before age 85 years for each category of initial HDL cholesterol compared to the reference adjusting for co-morbidities, calculated low-density lipoprotein cholesterol, medications, smoking, body mass index, and alcohol consumption. Treating HDL cholesterol as a continuous predictor, we also determined the HR for each 10-mg/dl increment in HDL cholesterol. Fully adjusted HR (95% confidence interval) for survival to 85 years of age for participants with an initial HDL cholesterol level ≥50 mg/dl compared to the reference was 0.72 (0.53 to 0.98). Each 10-mg/dl increment in HDL cholesterol was associated with a 14% (HR 0.86, 0.78 to 0.96) decrease in risk of mortality before 85 years of age. In conclusion, after adjusting for other factors associated with longevity, higher HDL cholesterol levels were significantly associated with survival to 85 years of age.
Although several researchers have examined the relation between high-density lipoprotein (HDL) cholesterol and coronary heart disease (CHD) in subjects of exceptional longevity, few have examined the association between a baseline HDL cholesterol level and survival to old age in a prospective cohort. Prospective studies evaluating the relation between HDL cholesterol and survival to old age have included small numbers of participants or participants who had already survived to exceptionally old age at baseline and thus may not be generalizable to middle-aged populations with co-morbidities. In this study, we examined whether high initial HDL cholesterol levels measured in middle-aged men are associated with a lower risk of death before 85 years of age in a cohort of men followed prospectively in the VA Normative Aging Study (VA NAS). We hypothesized that men with higher HDL cholesterol levels in middle age would be less likely to die before 85 years of age than men with lower initial HDL cholesterol levels.
Methods
The VA NAS is a longitudinal study of aging in 2,280 men living in the greater Boston area who were 21 to 80 years old at the time of enrollment from 1961 through 1970. Men with a history of antihypertensive medication use, systolic blood pressure >140 mm Hg or diastolic blood pressure of >90 mm Hg, CHD, diabetes mellitus, or cancer were excluded from the study. Enrollees in the VA NAS attended medical examinations every 3 to 5 years, depending on their age, until 1984, after which all were seen every 3 years for the remainder of follow-up. During study visits, detailed information regarding interim development of diseases, clinical parameters such as height, weight, and blood pressure, lifestyle factors such as smoking, alcohol intake, and exercise, and medication use were collected. Blood samples were collected at study visits, from which lipid and other laboratory parameters were determined. Participants in the VA NAS provided written informed consent at the time of enrollment and at the time of each study examination. Further details regarding the design of the VA NAS have been published previously.
Beginning in 1979, HDL cholesterol and triglycerides were measured at each follow-up visit in surviving members of the VA NAS cohort. From these, 697 men were selected who had ≥1 HDL cholesterol level documented from 1979 through 1999 and who were old enough to potentially reach 85 years of age from that measurement through the end of follow-up (July 1, 2008). Men with missing information on the predictor of interest or any of the covariates or who were ≥85 years old at the time of their first HDL cholesterol measurement were excluded (n = 42). The sample examined in the analyses described below therefore consisted of 652 men with complete data who could have reached 85 years of age by July 1, 2008.
Each participant’s first HDL cholesterol level was the exposure of interest. For the main analysis, HDL cholesterol level was categorized into <40 mg/dl (reference group), ≥40 to <50 mg/dl, or ≥50 mg/dl. Categories of HDL cholesterol change were chosen to reflect the spectrum of HDL cholesterol in the sample we studied and to be clinically meaningful. HDL cholesterol was also treated as a continuous variable to determine the effect of each 10-mg/dl increment of HDL cholesterol on risk for the outcome.
The primary outcome of interest was mortality before 85 years of age. Because the cohort was selected to include only men who would be ≥85 years old on July 1, 2008, the end of follow-up, all participants died during follow-up or survived to ≥85 years of age. No participant who was alive at the end of follow-up was <85 years of age. In the VA NAS, mortality of participants is tracked using periodic mailings. Once VA NAS staff is notified of the death of a participant, the death certificate is obtained to confirm death and determine cause of death.
Covariates other than body mass index (BMI) and alcohol consumption (see below) included in the analysis were determined at the time of the initial HDL cholesterol measurement. These included age and co-morbidities that had developed from VA NAS enrollment to the date of HDL cholesterol measurement including CHD, cerebrovascular disease, diabetes mellitus, hypertension, and cancer. Low-density lipoprotein cholesterol was calculated from each participant’s first HDL cholesterol, total cholesterol, and triglyceride measurements. Lifestyle factors were included because of known relations between HDL cholesterol and such factors. Smoking status, defined as currently smoking versus nonsmoking, was documented at the time of the first HDL cholesterol measurement. For each participant, the earliest documented information on alcohol consumption, defined as consuming ≥2 versus <2 beverages/day, and BMI (kilograms/meter squared) was used. For most participants included in the analyses, these lifestyle factors were documented on the date of HDL cholesterol measurement. Because a large number of participants had missing data on physical activity throughout the study and a very small number had information on physical activity at the time of their initial HDL cholesterol measurement, this was not included as a covariate in the analysis. Use of aspirin, β blockers, angiotensin-converting enzyme inhibitors, and cholesterol-modifying therapy (CMT) documented at the time of the initial HDL cholesterol were included in the models.
To determine whether a participant’s HDL cholesterol fluctuated substantially over time, we examined the correlation between the first and last HDL cholesterol levels for each participant using linear regression and adjusting for varying time between first and last HDL cholesterol measurements. We examined characteristics of survivors to 85 years of age versus nonsurvivors in co-morbidities, lipid parameters including the initial HDL cholesterol, and lifestyle factors by comparing means using t tests and proportions using Cochran-Mantel-Haenszel tests. We discuss the prevalence of death from various causes including CHD, stroke, and cancer because we expected that most men who died before 85 years of age died from CHD.
Using Cox proportional hazards, we determined the hazard ratio (HR) and 95% confidence interval (CI) for mortality before 85 years of age for each category of initial HDL cholesterol compared to the reference group (those with an initial HDL cholesterol level <40 mg/dl) and for each 10-mg/dl increment in initial HDL cholesterol. Age-adjusted and fully adjusted models including co-morbidities (CHD, diabetes mellitus, hypertension, cerebrovascular disease, and cancer), calculated low-density lipoprotein cholesterol, lifestyle factors (smoking status, BMI, and alcohol consumption), and medications (aspirin, β blockers, angiotensin-converting enzyme inhibitors, and CMT) were executed. Because of collinearity between increasing HDL cholesterol and decreasing triglycerides, the latter was excluded from the model as a potential confounder of the relation between the initial HDL cholesterol and longevity. Previous investigators have questioned the utility of considering triglycerides as an independent risk factor for CHD without incorporating the collinear effects of other lipids (including HDL cholesterol). To further illustrate the relation between initial HDL cholesterol category and survival in our study cohort, we plotted time to death for each of the 3 HDL cholesterol categories. As a sensitivity analysis, the association between categorical and continuous HDL cholesterol and mortality before 85 years of age was examined in a subgroup of 504 men who had information available on exercise any time during follow-up, defined as number of kilocalories expended per week. In addition, we reran our main analyses after excluding participants who were using aspirin, β blockers, angiotensin-converting enzyme inhibitors, or CMT at the initial examination.
Results
In total 652 men were old enough at their first HDL cholesterol measurement to potentially reach 85 years of age by July 1, 2008 and had complete data. Mean ± SD follow-up time was 16.3 ± 6.3 years. Of the 627 who had >1 HDL cholesterol level measured during follow-up, mean ± SD change in HDL cholesterol between the first and last measurement was 1.66 ± 11.9 mg/dl. The correlation between the first and last HDL cholesterol measurements was 0.57, suggesting that there were not wide fluctuations in HDL cholesterol measurements during the course of follow-up. Table 1 lists characteristics of the cohort by initial HDL cholesterol category. Because only men with an opportunity to survive to 85 years of age were selected for the cohort, age did not vary across the 3 HDL cholesterol categories. Men with higher HDL cholesterol levels were more likely to have lower BMI and to consume ≥2 alcoholic beverages at the initial examination. They were less likely to have diabetes or hypertension at the initial examination. Table 2 presents a comparison of characteristics of men who did not survive to 85 years of age (282, 43.3%) to men who did survive to 85 years of age (370, 56.7%). Of those who did not survive to 85 years of age, most (105, 39.8%) died from cardiovascular causes including CHD (43.8%), cerebrovascular disease (19.1%), and other forms of heart disease (37.1%). The second most prevalent cause of death was cancer (94, 35.6%). Other causes of death included respiratory conditions (24, 9.1%), diseases of the nervous system (12, 4.6%), injury or poisoning (9, 3.4%), genitourinary conditions (7, 2.7%), mental disorders (7, 2.7%), digestive conditions (3, 1.1%), and endocrine causes (1, 0.4%). Of the 282 who died before 85 years of age, 2 (0.8%) died from unknown cause and 18 (6.4%) did not have a death certificate available.
| Characteristic | HDL Cholesterol (mg/dl) | p Value | ||
|---|---|---|---|---|
| <40 | 40–<50 | ≥50 | ||
| (n = 194) | (n = 228) | (n = 230) | ||
| Age (years), mean ± SD | 64.9 ± 5.3 | 65.7 ± 5.4 | 65.0 ± 4.8 | 0.97 |
| High-density lipoprotein cholesterol (mg/dl), mean ± SD | 33.3 ± 5.5 | 44.6 ± 2.8 | 59.8 ± 9.1 | N/A |
| Body mass index (kg/m 2 ), mean ± SD | 27.5 ± 3.3 | 26.6 ± 3.0 | 25.8 ± 2.7 | <0.0001 |
| Low-density lipoprotein cholesterol (mg/dl), mean ± SD | 161.2 ± 42.6 | 167.9 ± 39.1 | 165.3 ± 35.5 | 0.30 |
| Currently smoking | 25 (12.9%) | 18 (7.9%) | 18 (7.8%) | 0.08 |
| Alcohol ≥2 beverages/day | 27 (13.9%) | 42 (18.3%) | 87 (37.5%) | <0.0001 |
| Coronary artery disease | 41 (21.1%) | 33 (14.4%) | 30 (12.9%) | 0.03 |
| Cerebrovascular disease | 3 (1.6%) | 5 (2.2%) | 1 (0.43%) | 0.30 |
| Diabetes mellitus | 8 (4.1%) | 2 (0.90%) | 3 (1.3%) | 0.05 |
| Hypertension | 32 (16.5%) | 35 (15.3%) | 23 (9.9%) | 0.05 |
| Aspirin use | 4 (2.1%) | 8 (3.5%) | 8 (3.5%) | 0.41 |
| β-Blocker use | 37 (19.1%) | 14 (6.1%) | 27 (11.7%) | 0.03 |
| Angiotensin-converting enzyme inhibitor use | 0 | 1 (0.44%) | 0 | 0.95 |
| Cholesterol-modifying therapy | 4 (2.1%) | 3 (1.3%) | 0 | 0.04 |
| Survival to ≥85 years of age | 104 (53.6%) | 124 (54.4%) | 142 (61.7%) | 0.08 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
