Lifetime Risk for Developing Dyslipidemia: The Framingham Offspring Study
Article Outline
Abstract
Background
High serum low-density lipoprotein
(LDL) cholesterol and low high-density lipoprotein (HDL) cholesterol are major vascular risk factors. National surveys indicate that 40% of individuals in the United States have borderline-high LDL cholesterol, and 13-34% have low HDL. The lifetime risk of developing dyslipidemia is unknown, however.
Methods
We estimated the 10- to 30-year long-term risks of developing “borderline-high” LDL cholesterol (≥130 mg/dL [3.4 mmol/L]), “high” LDL cholesterol (≥160 mg/dL [4.1 mmol/L]) and “low” HDL cholesterol (<40 mg/dL [1.0 mmol/L]) in 4701 Framingham Offspring Study participants (53% women) who attended at least 2 examinations between 1971 and 2000. We performed sex-specific analyses (for age groups 30-34, 40-44, 50-54 years), and estimated risks conditional on surviving without the lipid abnormality up to the baseline age. We also estimated risks accounting for baseline prevalence of dyslipidemia (elevated LDL, low HDL).
Results
Over a 30-year period, approximately 6 of 10 participants developed borderline-high LDL, 4 of 10 people developed high LDL, and 2 (women) to 4 (men) of 10 individuals developed low HDL levels; estimates were generally similar for different age groups. Adjustment for baseline prevalence of dyslipidemia increased these estimates: 30-year risks exceeded 80% for borderline-high LDL, 50% for high LDL, and 25% (women) to 65% (men) for low HDL; 20-50% had or developed a low HDL along with a high LDL level. The 30-year estimates approximate the lifetime risk in 50-year-olds.
Conclusions
The long term risks of developing dyslipidemia are substantial in both sexes, and considerably exceed prevalence estimates from cross-sectional surveys.
Keywords: Cholesterol, Cohort studies, Coronary heart disease prevention, Epidemiology, HDL, LDL, Lifetime risk
Elevated serum low-density lipoprotein (LDL) cholesterol and low high-density lipoprotein (HDL) cholesterol are major risk factors for coronary heart disease (CHD).1 The efficacy of lowering LDL has been demonstrated in clinical trials2, 3 (including those enrolling patients at moderate risk,4 or with established CHD).5 Treatments for low HDL levels also are evolving rapidly.6 Therefore, dyslipidemia is a key modifiable risk factor. Accordingly, National Cholesterol Education Program (NCEP) guidelines7 emphasize the identification and treatment of individuals with borderline-high (≥130 mg/dL) and high serum LDL (≥160 mg/dL) based on the concomitant presence of other risk factors and the absolute CHD risk. The NCEP guidelines define a HDL cholesterol <40 mg/dL as low overall; in the context of the metabolic syndrome, levels <40 mg/dL, in men and <50 mg/dL in women are considered ‘low.’
National cross-sectional surveys indicate that about 40% of individuals in the United States have a LDL cholesterol that is “borderline-high” (or greater),8 and 17-19% have “high” LDL cholesterol.9 The prevalence of low HDL cholesterol varies from 13% (women) to 34% (men) when the 40 mg/dL cut point is used.8 However, prevalence data do not indicate the true probability that an individual will develop dyslipidemia over a lifetime. The lifetime risk statistic is a better indicator of such probability. Previously the “lifetime risk” statistic has been used to calculate the long-term risk of CHD,10 stratification of long-term risk by risk factors11 and the lifetime risk of developing risk factors themselves, such as hypertension and obesity.12, 13 Currently, data on the long-term risks of developing elevated LDL or low HDL cholesterol levels in the general population are lacking.
Accordingly, we estimated the long-term risks of developing “borderline-high” and “high” serum LDL cholesterol, and the risks of developing low HDL cholesterol in a large community-based sample.
Methods
Study Sample
The Framingham Offspring Study is a prospective observational cohort study that began in 1971 with the enrollment of 5124 individuals (children of the original cohort14 and spouses of the children).15 These participants have been examined approximately every 4 years.
For the present investigation, we evaluated offspring cohort participants aged 30-59 years at the first examination (1971) and who attended subsequent Heart Study examinations up to and including the seventh examination (1998-2001). We excluded study participants who had developed a myocardial infarction during this time period because lipid-lowering treatment may be instituted at lower levels in people with CHD, and we used such treatment as an indicator of the presence of high LDL cholesterol (see below). We also excluded individuals with triglyceride levels ≥400 mg/dL (n=26) because calculated LDL values are valid only below this threshold. After exclusions, 4701 individuals remained eligible. All participants gave written informed consent and the study protocol was approved by the Institutional Review Board of the Boston Medical Center.
Measurement of Serum Lipid Cholesterol
At each Heart Study examination, blood samples were obtained in the morning (typically 7:30-9:00 am) from attendees after an overnight fast. At examinations 1 and 2, plasma total cholesterol was estimated with the Abell-Kendall technique.16 At examinations 3-7, plasma total cholesterol was measured using automated enzymatic assay.17 Cross-over validation methods were performed between manual and automated methods in 1983 (at start of examination 3). Plasma values were approximated to serum values by multiplying plasma values by 1·03. HDL cholesterol was measured using a precipitation method (manual method in examinations 1 and 2,18 and cross-validated automated methods19 thereafter). LDL cholesterol was calculated using the Friedewald formula.20
Definitions of Lipid Abnormalities (Outcomes)
We defined “borderline-high” and “high” LDL cholesterol as levels ≥130 mg/dL (or use of lipid-lowering medications), and ≥160 mg/dL (or use of lipid lowering medications), respectively. “Low” HDL cholesterol was defined as <40 mg/dL. These definitions are consistent with NCEP guidelines.7 Additional analyses evaluated the risks of developing HDL <50 mg/dL in women, since this cut point is used in the context of the metabolic syndrome.7 Because of known intra-individual variability in serum cholesterol,21 we performed secondary analyses evaluating risk of developing sustained elevations of LDL cholesterol at ≥2 consecutive examinations.
Risk of Developing High LDL and Low HDL Cholesterol in Individuals Without the Condition at Baseline
A primary objective of estimating long-term risk of developing dyslipidemia is to aid risk assessment in individuals free of the condition. We estimated the long-term (10-30 year) risks of developing borderline-high LDL, high LDL, low HDL, and the combined outcome of having both high LDL and low HDL (separate analyses for each outcome) for participants who attained the ages of 30, 40, and 50 years free of the lipid outcome of interest and attended at least 2 examinations between 1971 and 2001.
Participants were categorized into 5-year age groups: 30-34, 35-39, 40-44, 45-49, 50-54 and 55-59 years. Sex-specific analyses were performed separately for each baseline age group and for each lipid outcome. For simplicity, only estimates for baseline ages 30-34, 40-44, and 50-54 years are presented. Because we have approximately 30 years of follow-up on this cohort, for 30-year old subjects we estimated the risk of developing dyslipidemia through age 60, for 40- and 50-year old subjects we estimated the risk of developing lipid outcomes through ages 70 and 80, respectively. The 30-year estimate for baseline age 50 years approximates their residual lifetime risk.
Sex- and baseline age group-specific 10- to 30-year risk estimates of developing lipid outcomes were calculated using the Practical Incidence Estimators (PIE) macro.22 This modified form of survival analysis permits participants to contribute information at different ages and for varying durations of observation. It also yields cumulative incidence estimates adjusted for the competing risk of death (Appendix 1, available online).
In secondary analyses, we evaluated factors associated with the 10- and 30-year risk of developing an LDL cholesterol ≥160 mg/dL (or use of lipid lowering medications) in individuals free of the condition at baseline using multivariable logistic regression and modeling clinical correlates at baseline (age, sex, LDL level, body mass index, glucose level, smoking status, alcohol consumption and physical activity).
Risk of Developing High LDL and Low HDL Cholesterol Accounting for Lipid Abnormalities at Baseline
A second objective of estimating long-term risk estimates is to quantify the public health burden posed by dyslipidemia. The method of risk estimation described above is most valid for conditions where the baseline prevalence is minimal in younger adults (eg, Alzheimer’s disease). When estimating the lifetime risk of conditions that occur frequently at younger ages, it is important to consider both the baseline prevalence and future incidence.23 In our sample, a significant proportion of individuals already had developed borderline-high LDL, high LDL, and low HDL by the baseline age. Therefore, we calculated the cumulative incidence of developing dyslipidemia accounting for baseline prevalence.22, 23
Results
Prevalence of Borderline-high LDL, High LDL and Low HDL Serum Cholesterol Levels
The prevalence of borderline-high (≥130 mg/dL) and of high serum LDL (≥160 mg/dL) increased with age, exceeding 50% and approximating 25% at age 50 years, respectively (Table 1). The prevalence of low HDL cholesterol varied less with age, being about 40% in men and about 10% in women. About a third of women had HDL <50 mg/dL.
Table 1. Classification of Lipid Status at Baseline Age
| Men | Women | |||||
|---|---|---|---|---|---|---|
| Age in Years | Age in Years | |||||
| 30-34 | 40-44 | 50-54 | 30-34 | 40-44 | 50-54 | |
| (n=709) | (n=1215) | (n=1450) | (n=827) | (n=1413) | (n=1629) | |
| LDL ≥130 mg/dL (3.4 mmol/L) or use of lipid lowering medications, % | 44.4 | 56.9 | 60.5 | 23.7 | 35.5 | 51.4 |
| LDL ≥160 mg/dL (4.1 mmol/L) or use of lipid lowering medications, % | 16.4 | 24.1 | 28.6 | 6.1 | 11.6 | 23.5 |
| LDL ≥160 mg/dL (4.1 mmol/L) or use of lipid lowering medications at two consecutive exams, % | 10.6 | 17.4 | 22.1 | 3.1 | 7.1 | 17.7 |
| Low HDL (<40 mg/dL [1.0 mmol/L]), % | 35.7 | 36.5 | 40.3 | 11.1 | 10.4 | 10.1 |
| HDL <50 mg/dL (1.3 mmol/L) in women, % | 35.8 | 36.0 | 31.9 | |||
| LDL ≥160 mg/dL (4.1 mmol/L) and low HDL, % | 6.9 | 11.4 | 16.0 | 1.2 | 2.7 | 5.5 |
| LDL ≥160 mg/dL (4.1 mmol/L) and HDL <50 mg/dL (1.3 mmol/L) in women, % | 3.1 | 6.9 | 12.0 | |||
Risk of Ever Developing Lipid Abnormalities Conditional on Survival Free of Abnormality
Table 2 provides the estimated 10 to 30-year risks of developing borderline-high and high serum LDL cholesterol levels in participants free of the condition at a baseline age. Rates of developing elevated serum LDL cholesterol increased with follow-up in both sexes at all ages. Ten-year rates of developing elevated LDL cholesterol were generally lower in 30-34-year old women, relative to other age groups. The differences among the age groups narrowed with increasing duration of follow-up, however. Over a period of 30 years, approximately 55-70% of participants had developed borderline-high LDL cholesterol levels (or higher), and about 40 to 50% had developed elevated LDL cholesterol levels. Between a quarter and a third of individuals had sustained high LDL cholesterol levels at 2 or more consecutive examinations (with the exception of women aged 30-34 years).
Table 2. Long Term Risk of Elevated LDL Cholesterol in Men and Women Free of Condition at Baseline
| Years from Baseline | Men | Women | ||||
|---|---|---|---|---|---|---|
| Age at Baseline, Years | Age at Baseline, Years | |||||
| 30-34 | 40-44 | 50-54 | 30-34 | 40-44 | 50-54 | |
| Risk for developing LDL ≥130 mg/dL (3.4 mmol/L) or use of lipid lowering medications, % (95% CI) | ||||||
| 10 | 16.3(12.5-20.0) | 27.0(22.9-31.1) | 30.8(26.4-35.2) | 11.9(9.3-14.5) | 18.8(16.1-21.5) | 35.5(31.4-39.5) |
| 15 | 37.0(32.0-42.0) | 40.3(35.5-45.0) | 43.1(37.9-48.3) | 26.3(22.7-29.9) | 42.1(38.4-45.8) | 51.8(47.1-56.4) |
| 20 | 52.9(47.5-58.3) | 49.5(44.2-54.8) | 48.9(43.2-54.6) | 40.3(36.2-44.5) | 56.1(51.9-60.2) | 58.1(53.1-63.1) |
| 25 | 58.9(53.3-64.4) | 55.9(49.6-62.2) | 53.8(47.2-60.4) | 50.6(46.0-55.2) | 63.4(58.7-68.1) | 63.4(57.7-69.1) |
| 30 | 66.6(59.7-73.5) | 58.0(51.4-64.7) | 53.8(47.2-60.4) | 59.8(54.6-64.9) | 70.3(64.7-75.9) | 65.2(58.8-71.6) |
| Risk for developing LDL ≥160 mg/dL (4.1 mmol/L) or use of lipid lowering medications, % (95% CI) | ||||||
| 10 | 9.1(6.7-11.4) | 13.4(11.1-15.8) | 14.3(11.9-16.7) | 3.2(2.0-4.5) | 8.0(6.4-9.6) | 19.9(17.3-22.6) |
| 15 | 19.6(16.3-23.0) | 22.2(19.2-25.2) | 24.7(21.4-28.0) | 8.2(6.2-10.3) | 20.1(17.6-22.7) | 30.6(27.2-33.9) |
| 20 | 28.6(24.6-32.6) | 30.1(26.4-33.7) | 32.4(28.4-36.3) | 15.3(12.5-18.0) | 34.3(30.9-37.7) | 36.2(32.4-40.0) |
| 25 | 34.5(30.1-39.0) | 34.7(30.5-38.9) | 38.4(33.7-43.1) | 24.3(20.7-27.9) | 42.5(38.4-46.5) | 42.4(37.6-47.2) |
| 30 | 42.9(37.3-48.6) | 43.0(37.2-48.8) | 41.9(36.3-47.4) | 38.8(33.4-44.1) | 49.3(44.2-54.4) | 43.2(38.2-48.3) |
| Risk for developing LDL ≥160 mg/dL (4.1 mmol/L) or use of lipid lowering medications at two consecutive Framingham examinations, % (95% CI) | ||||||
| 10 | 4.7(2.7-6.6) | 8.8(6.7-10.9) | 8.1(5.9-10.3) | 1.5(0.5-2.5) | 4.4(3.1-5.8) | 11.8(9.3-14.2) |
| 15 | 12.6(9.4-15.8) | 16.8(13.7-19.9) | 15.5(12.2-18.8) | 4.0(2.3-5.6) | 13.9(11.3-16.5) | 17.8(14.7-21.0) |
| 20 | 21.8(17.5-26.1) | 21.1(17.4-24.9) | 23.2(18.6-27.8) | 8.7(6.1-11.3) | 22.4(18.7-26.1) | 25.8(21.4-30.3) |
| 25 | 28.6(23.3-33.9) | 25.6(20.9-30.2) | 26.8(21.2-32.3) | 16.6(12.6-20.6) | 27.8(23.3-32.4) | 33.8(27.4-40.1) |
Over the same time interval, nearly half of the men and about a fifth of the women had developed low serum HDL cholesterol levels (Table 3). About 40% of the women had developed HDL <50 mg/dL (Appendix Table 1). About a third of the men and 20-40% of women had developed a low HDL along with high LDL cholesterol (Table 3).
Table 3. Long Term Risk of Low Serum HDL Cholesterol and Low HDL and High LDL Cholesterol Levels in Men and Women Free of Condition at Baseline
| Years from Baseline | Men | Women | ||||
|---|---|---|---|---|---|---|
| Age at Baseline, Years | Age at Baseline, Years | |||||
| 30-34 | 40-44 | 50-54 | 30-34 | 40-44 | 50-54 | |
| Risk for developing low HDL (<40 mg/dL [1.0 mmol/L]), % (95% CI) | ||||||
| 10 | 13.3(10.1-16.5) | 19.4(16.4-22.3) | 18.2(15.2-21.2) | 3.8(2.4-5.2) | 5.7(4.3-7.0) | 7.2(5.6-8.8) |
| 15 | 27.1(22.9-31.4) | 31.2(27.6-34.8) | 29.3(25.4-33.2) | 7.5(5.5-9.5) | 11.8(9.8-13.7) | 12.6(10.4-14.8) |
| 20 | 36.7(31.9-41.5) | 39.4(35.3-43.6) | 35.8(31.4-40.2) | 11.0(8.6-13.4) | 14.9(12.5-17.2) | 16.3(13.5-19.0) |
| 25 | 43.3(38.1-48.5) | 45.1(40.2-49.9) | 39.1(33.9-44.2) | 15.3(12.3-18.3) | 17.7(15.0-20.5) | 19.3(15.8-22.7) |
| 30 | 48.3(42.4-54.1) | 45.1(40.2-49.9) | 45.0(37.5-52.6) | 18.3(14.5-22.1) | 20.6(17.2-24.0) | 20.0(16.3-23.8) |
| Risk for developing LDL ≥160 mg/dL(4.1 mmol/L) and low HDL or use of lipid lowering medications and low HDL (<40 mg/dL [1.0 mmol/L]), % (95% CI) | ||||||
| 10 | 4.9(3.2-6.6) | 7.4(5.8-9.1) | 10.4(8.5-12.4) | 0.6(0.1-1.2) | 2.2(1.4-3.0) | 8.8(7.1-10.4) |
| 15 | 11.0(8.5-13.5) | 14.3(11.9-16.6) | 19.0(16.3-21.8) | 2.1(1.1-3.1) | 7.4(5.9-9.0) | 16.6(14.2-19.0) |
| 20 | 15.5(12.5-18.4) | 23.0(19.9-26.1) | 26.4(23.0-29.9) | 4.4(2.9-5.9) | 16.1(13.6-18.6) | 26.6(23.3-29.9) |
| 25 | 24.0(20.0-27.9) | 29.5(25.6-33.4) | 32.9(28.7-37.1) | 8.6(6.3-11.0) | 23.1(19.8-26.5) | 37.9(33.3-42.6) |
| 30 | 34.9(29.5-40.3) | 38.4(33.1-43.7) | 37.6(32.5-42.8) | 21.1(16.7-25.5) | 35.7(30.6-40.8) | 42.9(37.5-48.3) |
In multivariable analyses, age (hazard ratio [HR] per year increment, 1.02, 95% CI, 1.00-1.03, P <.03) and baseline LDL level was the most significant predictor of development of LDL ≥160 mg/dL on follow-up over 10 years in 3568 participants eligible for these analyses. Compared to individuals with a baseline LDL <100 mg/dL, those with baseline LDL 100-129 mg/dL experienced a 3-fold risk (HR 3.01, 95% CI, 1.93-4.69, P <.0001) and those with LDL 130-159 mg/dL had a 12-fold risk (HR 12.00, 95% CI, 7.88-18.25). In analyses of 30-year risk of developing dyslipidemia, baseline LDL remained the key risk factor for future development of a high LDL level.
Long-term and Lifetime Risk of Ever Developing Lipid Abnormalities Accounting for Baseline Prevalence at a Given Age
Table 4 presents the long-term risks of developing borderline-high and high LDL cholesterol at each baseline age, accounting for the baseline prevalence. The 30-year risk of having (at baseline) or developing (on follow-up) borderline-high LDL cholesterol levels varied from 70 to 80%, and from 43 to 60% (being similar in the 2 sexes generally) for high LDL. Approximately a third of participants had or developed sustained elevations of serum LDL cholesterol at ≥2 consecutive examinations.
Table 4. Long Term Risk of Elevated LDL Cholesterol in Men and Women Adjusted for Baseline Prevalence
| Years from Baseline | Men | Women | ||||
|---|---|---|---|---|---|---|
| Age at Baseline, Years | Age at Baseline, Years | |||||
| 30-34 | 40-44 | 50-54 | 30-34 | 40-44 | 50-54 | |
| Risk for Developing LDL ≥130 mg/dL (3.4 mmol/L) or use of lipid lowering medications, % (95% CI) | ||||||
| 10 | 53.5(49.8-57.2) | 68.5(65.8-71.2) | 72.7(70.2-75.1) | 32.8(29.6-36.0) | 47.6(44.9-50.3) | 68.7(66.2-71.2) |
| 15 | 65.0(61.4-68.6) | 74.2(71.6-76.9) | 77.5(75.0-80.0) | 43.8(40.3-47.2) | 62.6(59.8-65.4) | 76.6(74.0-79.1) |
| 20 | 73.8(70.3-77.3) | 78.2(75.5-80.9) | 79.8(77.2-82.4) | 54.5(50.8-58.1) | 71.6(68.7-74.5) | 79.7(77.0-82.3) |
| 25 | 77.1(73.7-80.6) | 81.0(78.0-84.0) | 81.7(78.9-84.6) | 62.3(58.5-66.1) | 76.4(73.2-79.5) | 82.2(79.3-85.1) |
| 30 | 81.4(77.4-85.5) | 81.9(78.8-85.0) | 81.7(78.9-84.6) | 69.3(65.2-73.4) | 80.9(77.2-84.5) | 83.1(79.9-86.3) |
| Risk for Developing LDL≥160 mg/dL(4.1 mmol/L) or use of lipid lowering medications, %(95% CI) | ||||||
| 10 | 23.9(20.8-27.1) | 34.3(31.6-37.0) | 38.8(36.2-41.5) | 9.1(7.1-11.1) | 18.7(16.6-20.8) | 38.8(36.1-41.4) |
| 15 | 32.8(29.2-36.4) | 41.0(38.0-43.9) | 46.2(43.3-49.2) | 13.8(11.4-16.2) | 29.4(26.8-32.0) | 46.9(44.0-49.8) |
| 20 | 40.3(36.4-44.2) | 46.9(43.7-50.1) | 51.7(48.5-55.0) | 20.4(17.5-23.3) | 41.9(38.7-45.2) | 51.2(48.0-54.4) |
| 25 | 45.3(41.1-49.4) | 50.4(46.9-54.0) | 56.0(52.4-59.7) | 28.9(25.3-32.5) | 49.2(45.4-52.9) | 55.9(52.0-59.8) |
| 30 | 52.3(47.3-57.2) | 56.8(52.1-61.4) | 58.5(54.3-62.7) | 42.5(37.3-47.6) | 55.2(50.6-59.8) | 56.6(52.6-60.6) |
| Risk for Developing LDL≥160 mg/dL (4.1 mmol/L) or use of lipid lowering medications at two consecutive Framingham examinations, % (95% CI) | ||||||
| 10 | 14.8(11.8-17.7) | 24.6(21.9-27.4) | 28.4(25.6-31.2) | 4.6(2.9-6.2) | 11.2(9.3-13.1) | 27.4(24.7-30.1) |
| 15 | 21.8(18.2-25.4) | 31.2(28.0-34.4) | 34.1(30.8-37.4) | 6.9(4.9-9.0) | 20.0(17.3-22.8) | 32.4(29.3-35.5) |
| 20 | 30.1(25.7-34.4) | 34.8(31.2-38.4) | 40.2(36.1-44.2) | 11.5(8.7-14.3) | 27.9(24.3-31.5) | 38.9(35.0-42.9) |
| 25 | 36.1(31.1-41.2) | 38.5(34.3-42.7) | 42.9(38.3-47.6) | 19.1(15.1-23.2) | 33.0(28.6-37.3) | 45.5(40.0-50.9) |
Over a 30-year period, over two-thirds of the men and a quarter of the women had or developed low HDL cholesterol (Table 5). Approximately 60% of women had or developed HDL <50 mg/dL (Appendix Table 1, available online). Between a third and a half of the participants had or developed low serum HDL along with high LDL cholesterol (with the exception of women aged 30-34 years).
Table 5. Long Term Risk of Low Serum HDL Cholesterol and Low HDL and High LDL Cholesterol Levels in Men and Women Adjusted for Baseline Prevalence
| Years from Baseline | Men | Women | ||||
|---|---|---|---|---|---|---|
| Age at Baseline, Years | Age at Baseline, Years | |||||
| 30-34 | 40-44 | 50-54 | 30-34 | 40-44 | 50-54 | |
| Risk for developing low HDL (<40 mg/dL [1.0 mmol/L]), % (95% CI) | ||||||
| 10 | 44.2(40.5-47.9) | 48.8(46.0-51.7) | 51.2(48.4-53.9) | 14.5(12.1-16.9) | 15.5(13.6-17.4) | 16.6(14.6-18.6) |
| 15 | 53.1(49.4-56.9) | 56.3(53.4-59.3) | 57.8(54.9-60.7) | 17.8(15.2-20.4) | 20.9(18.7-23.2) | 21.5(19.1-23.8) |
| 20 | 59.3(55.5-63.1) | 61.6(58.5-64.7) | 61.7(58.5-64.8) | 20.9(18.0-23.7) | 23.7(21.2-26.2) | 24.8(22.0-27.5) |
| 25 | 63.6(59.7-67.5) | 65.2(61.7-68.6) | 63.6(60.2-67.0) | 24.7(21.5-27.9) | 26.3(23.5-29.1) | 27.4(24.1-30.8) |
| 30 | 66.7(62.5-70.9) | 65.2(61.7-68.6) | 67.2(62.5-71.9) | 27.4(23.5-31.2) | 28.9(25.6-32.2) | 28.1(24.5-31.7) |
| Risk for developing LDL ≥160 mg/dL(4.1 mmol/L) and low HDL or use of lipid lowering medications and low HDL(<40 mg/dL [1.0 mmol/L]), %(95% CI) | ||||||
| 10 | 11.5(9.1-13.9) | 18.0(15.8-20.2) | 24.8(22.4-27.1) | 1.8(0.9-2.8) | 4.8(3.7-5.9) | 13.8(11.9-15.7) |
| 15 | 17.2(14.3-20.0) | 24.1(21.5-26.6) | 32.0(29.2-34.7) | 3.3(2.0-4.5) | 9.9(8.2-11.6) | 21.2(18.8-23.7) |
| 20 | 21.3(18.1-24.5) | 31.8(28.7-34.9) | 38.2(35.0-41.4) | 5.5(3.9-7.2) | 18.4(15.8-20.9) | 30.7(27.4-33.9) |
| 25 | 29.2(25.3-33.1) | 37.6(33.9-41.3) | 43.6(39.9-47.4) | 9.7(7.3-12.1) | 25.2(21.9-28.5) | 41.4(36.9-45.8) |
| 30 | 39.4(34.3-44.6) | 45.5(40.6-50.3) | 47.6(43.1-52.1) | 22.1(17.6-26.5) | 37.4(32.4-42.4) | 46.1(40.9-51.2) |
Discussion
Principal Findings
The present investigation provides estimates of the long-term risks of developing varying degrees of elevation of serum LDL cholesterol and low serum HDL-cholesterol in a community-based sample. It also extends our data on the lifetime risk of developing high total serum cholesterol levels (≥240 mg/dL).23 Our estimates of short- and long-term risks of developing dyslipidemia in people free of the condition at a given age can be useful for predicting future risk in individuals. The complementary estimates of long-term and lifetime risks of developing dyslipidemia accounting for baseline prevalence are provided to reflect the public health burden of dyslipidemia.
Over a 30-year period, approximately 6 in 10 participants developed borderline or greater serum LDL cholesterol, 4 of 10 people developed high LDL, and 2 (women) to 4 (men) of 10 individuals developed low HDL cholesterol; estimates were similar for the 2 sexes and for different age groups. About 20-40% of the participants had developed a low serum HDL along with high LDL cholesterol. These 30-year estimates correspond to the lifetime risks in 50-year olds. Adjustment for baseline prevalence of dyslipidemia considerably increased these estimates: 30-year risks exceeded 80%, 50%, and 25 (women) to 65% (men) for already having or developing borderline-high LDL, high LDL, and low HDL cholesterol levels, respectively. Between 20-50% had or developed a low serum HDL level along with a high LDL cholesterol level.
Comparison With National Cross-sectional Data on the Prevalence of Dyslipidemia
National surveys indicate that 40% of the people in the United States between ages 20-74 years have borderline or greater LDL cholesterol,8 and 17-19% of individuals have high LDL cholesterol.9 The baseline prevalence in our sample resembles these data. However, our long-term risk estimates of developing elevated serum LDL cholesterol are 50-100% higher than these prevalence estimates, and 100-150% higher when baseline prevalence is accounted for.
The prevalence of low HDL (using the 40 mg/dL threshold) in national surveys ranges from 13 (women) to 34% (men),8 similar to the baseline prevalence in our sample. Our estimates of long-term risk of having or developing low HDL cholesterol were 50 (men) to 100% (women) higher when compared to these prevalence estimates. Additionally, we observed that about 60% of women had or developed an HDL <50 mg/dL on long-term follow-up.
Strengths and Limitations
The strengths of our analyses are the well-characterized cohort with long-term data during a contemporary period. We used lifetime risk estimates that adjust for the impact of death and variations in person-years contributed by participants.
Several limitations merit attention. First, Offspring Study participants are predominantly white, limiting the generalizability of our estimates. Second, we used a single lipid measurement obtained at each examination to categorize individuals. Since intra-individual variation of lipids is random, we are equally likely to have overestimated or underestimated the burden of dyslipidemia. Nonetheless, supplementary analyses indicated that the lifetime risk for having or developing sustained elevation of LDL exceeded 33% (with the exception of women aged 30-34 years), consistent with our overall observations. Third, our estimates reflect the overall experience of our cohort, and must be applied cautiously to individual patients. The risk of developing dyslipidemia varies according to the presence of obesity, sedentary lifestyle,24, 25, 26 genetic27 and dietary factors.28
Implications for Public Health Policy and for Individuals
The NCEP7 recommends that all adults over age 20 years should be screened for serum cholesterol levels every 5 years. The “Healthy People 2010” initiative proposes to reduce the proportion of adults with high serum cholesterol in the United States to 17% by the year 2010.29 Recent national trends, based on the National Health and Nutrition Examination Survey (NHANES) 1999-2002,30 suggest that this target has been attained along with parallel achievements in mean lipid levels; mean LDL declined to 123 mg/dL from 126 mg/dL in NHANES III (1988-1994), although no statistically significant change in mean HDL levels were reported.
Our data suggest that prevalence estimates from surveys may seriously underestimate the burden of elevated LDL and low HDL cholesterol. Our baseline prevalence data and the 10-30 year rates of developing elevated LDL cholesterol levels also demonstrate an interesting pattern: a higher proportion of men developed elevated LDL cholesterol at a younger age compared to women. However, the steep increases in the incidence of borderline-high and high serum LDL cholesterol levels between the ages of 40 and 60 in women (ie, 10-20 year follow-up of 30-year old women) translate into an overall lifetime risk that is equivalent to that observed in men. These observations are consistent with the known increase in serum LDL cholesterol during peri-menopausal years.31 The lower baseline risk of developing dyslipidemia in premenopausal women suggests a potential window of opportunity for focusing preventive efforts.
Given that the lifetime risk of having or developing borderline-high serum LDL cholesterol exceeds 80% and that for high LDL cholesterol crosses 50%, our data emphasize the public health burden posed by dyslipidemia. Recent studies4 have demonstrated that significant reductions in serum LDL cholesterol and in CHD incidence can be achieved with statins in patients with “borderline-high” LDL cholesterol per current guidelines.7 It is tempting to speculate that over 70% of the United States population may qualify for such treatment at some stage in their lives based on our estimates, with substantial health and economic consequences. Furthermore, our data indicate that between a third and a half of the people have or are likely to develop a low HDL along with a high LDL cholesterol, thereby underscoring the importance of screening programs. The high lifetime risk for developing dyslipidemia suggests far wider use of pharmacological strategies in the future will become the norm, unless such incidence can be stemmed via population-wide dietary and lifestyle measures.
The high lifetime risk of developing dyslipidemia in those initially free of the condition suggests that it is important to consider a strategy of cautious optimism when counseling individuals with LDL cholesterol <130 mg/dL. It is noteworthy that informing patients that they have ‘normal’ serum LDL cholesterol during screening has been associated with a subsequent increase in saturated fat intake and serum cholesterol levels.32
Conclusions
Dyslipidemia is one of the leading treatable causes of lost healthy life world-wide.33 Our data further highlight the public health burden of dyslipidemia. Additional research is warranted to investigate the biological and social underpinnings of dyslipidemia to prevent the inexorable rise of this condition in acculturated societies.
Acknowledgments
This work was supported through National Institute of Health/ National Heart, Lung & Blood Institute (NHLBI) Contract N01-HC-25195, and 2K24HL04334 (Dr Vasan).
Appendix 1.
The Practical Incidence Estimators (PIE) macro uses age as a time scale and calculates the cumulative incidence of a given event as a sum of the probabilities of failure at each age within the range of interest. For example, to estimate the cumulative incidence of developing the lipid abnormality of interest by age 60 for participants entering the study free of the condition at age 40 we would add the probabilities of developing the lipid abnormality between ages 40 and 41 years, 41 and 42 years, and so on up to ages 59 and 60 years. The way in which these 1-year probabilities are calculated differentiates the lifetime risk approach from the standard Kaplan-Meier analyses. The standard Kaplan-Meier method calculates the probability of failure as a product of the hazard of developing the event at age ‘A’ and the probability of surviving past age ‘A-1.’ The estimate of the probability of survival up to age ‘A-1’ is based on an analysis in which deaths during follow-up are censored; thus, it is an estimate of the probability of survival event-free but not necessarily being alive. The treatment of death as censored observations inflates the probabilities of failure and the estimated cumulative incidence because people who die are not excluded from the pool of those who can still develop the event (the ‘at-risk’ group). The PIE macro approach builds on a modification suggested by Gaynor et al34 that computes the probabilities of failure using the probability of surviving past age ‘A-1’ being both event-free and alive: deaths are counted as events in the estimation of the probability of survival past age ‘A-1.’
Appendix Table 1. Risk of Developing HDL <50 mg/dL (1.3 mmol/L) in Women
| Years from Baseline | Age, years | |||||
|---|---|---|---|---|---|---|
| 30-34 | 40-44 | 50-54 | 30-34 | 40-44 | 50-54 | |
| Risk for developing HDL <50 mg/dL (1.3 mmol/L) in women free of condition at baseline, % (95% CI) | Risk for developing HDL <50 mg/dL (1.3 mmol/L) in women baseline adjusted for baseline prevalence, % (95% CI) | |||||
| 10 | 12.9(10.0-15.8) | 13.6(11.2-15.9) | 17.9(15.1-20.6) | 44.1(40.7-47.5) | 44.6(42.0-47.3) | 44.1(41.5-46.7) |
| 15 | 26.3(22.4-30.2) | 22.3(19.4-25.3) | 26.9(23.5-30.3) | 52.7(49.2-56.1) | 50.3(47.5-53.0) | 50.2(47.4-53.1) |
| 20 | 31.6(27.4-35.8) | 31.2(27.6-34.9) | 34.4(30.3-38.5) | 56.1(52.6-59.6) | 56.0(53.0-58.9) | 55.3(52.2-58.5) |
| 25 | 37.5(32.9-42.0) | 38.2(33.8-42.6) | 38.5(33.7-43.2) | 59.8(56.3-63.4) | 60.4(57.2-63.7) | 58.1(54.6-61.6) |
| 30 | 41.9(36.8-47.1) | 43.1(37.8-48.4) | 39.5(34.4-44.6) | 62.7(58.9-66.5) | 63.6(59.9-67.2) | 58.8(55.1-62.5) |
| Risk for developing LDL ≥160 mg/dL (4.1 mmol/L) or use of lipid lowering medications and HDL <50 mg/dL (1.3 mmol/L) in women free of condition at baseline, % (95% CI) | Risk for developing LDL ≥160 mg/dL (4.1 mmol/L) or use of lipid lowering medications and low HDL <50 mg/dL (1.3 mmol/L) in women adjusted for baseline prevalence, % (95% CI) | |||||
| 10 | 1.2(0.4-2.0) | 4.6(3.4-5.8) | 12.5(10.5-14.6) | 4.3(2.9-5.7) | 11.3(9.6-12.9) | 23.1(20.8-25.3) |
| 15 | 4.3(2.8-5.8) | 12.2(10.2-14.2) | 21.0(18.3-23.7) | 7.3(5.5-9.1) | 18.3(16.1-20.4) | 30.5(27.8-33.2) |
| 20 | 9.8(7.5-12.0) | 22.9(20.0-25.8) | 29.0(25.6-32.4) | 12.6(10.2-15.0) | 28.2(25.3-31.1) | 37.6(34.3-40.8) |
| 25 | 15.6(12.6-18.6) | 29.9(26.3-33.6) | 37.0(32.5-41.6) | 18.3(15.2-21.4) | 34.8(31.3-38.3) | 44.6(40.5-48.7) |
| 30 | 27.1(22.4-31.8) | 38.7(33.6-43.7) | 39.7(34.7-44.7) | 29.4(24.7-34.0) | 42.9(38.2-47.7) | 46.9(42.4-51.4) |
References
- . High-density lipoprotein, low-density lipoprotein and coronary artery disease. Am J Cardiol. 1990;66:7A–10A
- Primary Prevention of Acute Coronary Events With Lovastatin in Men and Women With Average Cholesterol Levels: Results of AFCAPS/TexCAPS. JAMA. 1998;279:1615–1622
- Prevention of Coronary Heart Disease with Pravastatin in Men with Hypercholesterolemia. N Engl J Med. 1995;333:1301–1308
- Prevention of coronary and stroke events with atorvastatin in hypertensive patients who have average or lower-than-average cholesterol concentrations, in the Anglo-Scandinavian Cardiac Outcomes Trial--Lipid Lowering Arm (ASCOT-LLA): a multicentre randomised controlled trial. Lancet. 2003;361:1149–1158
- Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet. 1994;344:1383–1389
- . High-density lipoproteins as therapeutic targets. Curr Opin Lipidol. 2005;16:345–349
- Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III). JAMA. 2001;285:2486–2497
- Heart Disease and Stroke Statistics--2006 Update: A Report From the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation. 2006;113:e85–e151
- . Heart Disease and Stroke Statistics - 2005 Update. 35. Dallas, Texas: American Heart Association; 2005;
- . Lifetime risk of developing coronary heart disease. Lancet. 1999;353:89–92
- Prediction of Lifetime Risk for Cardiovascular Disease by Risk Factor Burden at 50 Years of Age. Circulation. 2006;113:791–798
- Residual lifetime risk for developing hypertension in middle-aged women and men: The Framingham Heart Study. JAMA. 2002;287:1003–1010
- Estimated Risks for Developing Obesity in the Framingham Heart Study. Ann Intern Med. 2005;143:473–480
- . Epidemiological approaches to heart disease: the Framingham Study. Am J Public Health. 1951;41:279–281
- An investigation of coronary heart disease in families (The Framingham offspring study). Am J Epidemiol. 1979;110:281–290
- . A simplified method for the estimation of total cholesterol in serum and demonstration of its specificity. J Biol Chem. 1952;195:357–366
- . Automated enzymatic standardized lipid analyses for plasma and lipoprotein fractions. Clin Chim Acta. 1987;166:1–8
- Lipid Research Clinics Program (Manual of Laboratory Operation). Bethesda, MD: National Institutes of Health; 1974;75-628
- . Dextran sulfate-Mg2+ precipitation procedure for quantitation of high-density-lipoprotein cholesterol. Clin Chem. 1982;28:1379–1388
- . Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972;18:499–502
- . Intra-individual variation of serum cholesterol, triglycerides and high density lipoprotein cholesterol in normal humans. Atherosclerosis. 1982;45:259–266
- Computing estimates of incidence, including lifetime risk: Alzheimer’s disease in the Framingham Study (The Practical Incidence Estimators (PIE) macro). Stat Med. 2000;19:1495–1522
- Estimating lifetime risk of developing high serum total cholesterol: Adjustment for baseline prevalence and single occasion measurements. Am J Epidemiol. 2007;165:464–472
- . Exercise, lipids, and lipoproteins in older adults: a meta-analysis. Prev Cardiol. 2005;8:206–214
- Diet and plasma lipids in women (I. Macronutrients and plasma total and low-density lipoprotein cholesterol in women: the Framingham nutrition studies). J Clin Epidemiol. 1996;49:657–663
- Effects of weight loss in overweight/obese individuals and long-term lipid outcomes--a systematic review. Obes Rev. 2004;5:43–50
- Apolipoprotein E alleles, dyslipidemia, and coronary heart disease (The Framingham Offspring Study). JAMA. 1994;272:1666–1671
- Effect of apolipoprotein E phenotype on diet-induced lowering of plasma low density lipoprotein cholesterol. J Lipid Res. 1994;35:1965–1975
- . Healthy People 2010. 2nd ed.. Washington, DC: U.S. Government Printing Office; 2000;
- Trends in Serum Lipids and Lipoproteins of Adults, 1960-2002. JAMA. 2005;294:1773–1781
- Effects of age, gender, and menopausal status on plasma low density lipoprotein cholesterol and apolipoprotein B levels in the Framingham Offspring Study. J Lipid Res. 1994;35:779–792
- Impact of receiving blood cholesterol test results on dietary change. Am J Prev Med. 1998;14:103–110
- Selected major risk factors and global and regional burden of disease. Lancet. 2002;360:1347–1360
- On the use of cause-specific failure and conditional failure probabilities: examples from clinical oncology data. J Am Stat Assoc. 1993;88:400–409
PII: S0002-9343(07)00044-7
doi:10.1016/j.amjmed.2006.12.015
© 2007 Elsevier Inc. All rights reserved.
