Trends in Lipids and Lipoproteins in US Adults, 1988-2010 FREE

Epidemiologic studies have demonstrated that high concentrations of low-density lipoprotein cholesterol (LDL-C) and total cholesterol (TC) and low levels of high-density lipoprotein cholesterol (HDL-C) are major risk factors for coronary heart disease (CHD).^1- 6 Although triglyceride levels have not been shown to be an independent risk factor for CHD, there is increasing evidence of a strong association between elevated triglyceride levels and CHD risk.^7- 8 Very low-density lipoprotein cholesterol combined with LDL-C form non–HDL-C, which may enhance CHD risk prediction when triglyceride levels are high (200-499 mg/dL).⁹

Among adults in the United States, there have been consistent declines in mean TC from 1960 to 1994,^10- 11 with substantial decreases occurring between 1976-1980 and 1988-1991.¹¹ Between 1988-1994 and 1999-2002, age-adjusted mean TC and age-adjusted mean LDL-C levels continued to decline in all adults aged 20 years or older, and the decreases were substantial, in particular for men aged 60 years or older and women aged 50 years or older.¹²

In this study, trends in serum lipids and lipoproteins among adults aged 20 years or older including TC, HDL-C, non–HDL-C, LDL-C, and triglycerides over a 22-year period spanning 1988 to 2010 were analyzed. Trends in the percentage of adults receiving lipid-lowering medications, trends in these lipid concentrations for adults not receiving lipid-lowering medications, and trends in lipids for obese adults during this period are also presented.

Results are based on data from 3 National Health and Nutrition Examination Surveys (NHANES), including NHANES III (1988-1994),¹³ NHANES 1999-2002, and NHANES 2007-2010.¹⁴ Data from NHANES 2003-2006 are not presented to allow approximately equal intervals between the midpoints of the constructed survey periods and because of changes in laboratory methods during this period that most likely positively biased HDL-C values.¹⁵

The NHANES surveys are conducted by the National Center for Health Statistics/Centers for Disease Control and Prevention (CDC) and constitute a series of cross-sectional, nationally representative surveys of the US noninstitutionalized population. Beginning in 1999, NHANES became a continuous survey and data have been released in 2-year cycles.¹⁴ Each of the NHANES survey designs is a complex, multistage, area probability sample. Written informed consent was obtained from participants. The survey was approved by the National Center for Health Statistics ethics review board.

Each survey includes an interview and an examination. The interview consists of health-related questions including use of lipid-lowering medications. Persons participating in the interview are invited to participate in the examination carried out in a mobile examination center including phlebotomy and body measurements.^12- 13 Participants in the examination component of NHANES were randomly assigned to either a morning session (and asked to fast for at least 8 hours prior to the examination) or an afternoon/evening session. Each examined participant was eligible for TC and HDL-C measurements, whereas only participants in the morning session who fasted between 8.5 and 23 hours were eligible for triglyceride measurements.

Participants self-reported their race and ethnicity after being shown a list that included an open-ended response. Race/ethnicity was categorized as Mexican American, non-Hispanic white, and non-Hispanic black.

During the interview, adults were asked whether they had had their cholesterol checked. Those who responded affirmatively were asked whether they had been told by a physician that their cholesterol was high. Those who responded affirmatively were asked if a physician had told them to take prescribed medicine to lower their blood cholesterol. Those who responded affirmatively were then asked if they were currently taking lipid-lowering medications.

Body mass index (BMI) was defined as weight in kilograms divided by height in meters squared rounded to the nearest tenth. Adults were classified as underweight (BMI <18.5), healthy weight (BMI 18.5 to <25), overweight (BMI 25 to <30), and obese (BMI ≥30).

To convert TC, LDL-C, HDL-C, and non–HDL-C to millimoles per liter, multiply by 0.0259. To convert triglycerides to millimoles per liter, multiply by 0.0113.

All lipid analyses were conducted on venous samples collected according to a standardized protocol.^15- 17 There were changes in the laboratories, methods, and instruments used to measure lipid concentrations across survey periods.¹² In 2007-2010, similar to previous surveys, TC¹⁸ and triglycerides¹⁹ were measured using enzymatic reactions. High-density lipoprotein cholesterol was measured by the direct immunoassay method²⁰ during 2007-2010, whereas in 1988-1994 and 1999-2002, the heparin manganese precipitation method²¹ was primarily used. The Friedewald equation²² (LDL-C = TC − [HDL-C + triglycerides/5]) was used to calculate LDL-C for adults whose triglyceride level was less than or equal to 400 mg/dL. Non–HDL-C was calculated as TC minus HDL-C. Standardization of serum lipid measurements was performed according to the criteria of the CDC's Lipid Standardization Program.²³

Arithmetic means for TC, HDL-C, non–HDL-C, LDL-C, and the percentage of participants receiving lipid-lowering medications are presented. Because of the highly skewed distribution of triglycerides, the geometric mean, calculated by back-transforming the mean of the logarithm of the triglyceride values,²⁴ is presented. In estimating means, geometric means, and percentages, sample weights, which adjust for the unequal selection probabilities, nonresponse, and noncoverage, were incorporated. Standard errors used to construct confidence intervals were estimated using Taylor series linearization, a design-based method.²⁴ Confidence intervals for the percentage of adults taking lipid-lowering medications were constructed using the logit transformation²⁴ to avoid negative lower limits. Means, geometric means, and percentages for participants aged 20 years or older were age-adjusted by the direct method²⁵ to the projected 2000 Census population estimates using the age groups 20 to 39 years, 40 to 59 years, and 60 years or older.²⁶

Statistical hypotheses were tested using the t statistic and an α = .05 based on a 2-tailed test. The Bonferroni method was used to adjust for multiple comparisons by dividing the overall α level by the number of implied comparisons. For example, in testing for trends involving all adults, men, and women, there were 3 implied comparisons and an α = .017 was used. For trends involving the 6 sex and race/ethnicity groups, an overall α = .008 was used. Statistical hypotheses for triglycerides were tested on the log scale. Hypotheses of no survey trends in age-adjusted means and percentages over the 22-year period 1988 to 2010 (which involved 3 survey periods, 1988-1994, 1999-2002, and 2007-2010) were tested using orthogonal contrast matrices.²⁷ Rejection of this null hypothesis implied the existence of a trend. Both linear (ie, consistent increases or decreases between 1988 and 2010) and quadratic (ie, an increase from 1988-1994 to 1999-2002 followed by a decrease between 1999-2002 and 2007-2010) trends were tested. To control for the possible confounding effect of BMI and lipid-lowering medications as well as race/ethnicity and age, sex-specific multiple linear regression models were constructed with survey treated as a continuous variable. Age was also treated as continuous variable. The Satterthwaite-adjusted F test²⁸ was used to test for trends. All statistical analyses were carried out using SAS version 9.2.1 (SAS Institute Inc) and SUDAAN version 10.0 (RTI International).

There were 16 573 adults aged 20 years or older examined in NHANES III, 9471 in NHANES 1999-2002, and 11 766 in NHANES 2007-2010. Between 5% and 7%, depending on the survey year, were excluded from analyses because of missing data for TC, HDL-C, and non–HDL-C. Some participants were missing TC but not HDL-C data and vice versa; therefore, sample sizes differ slightly for each of these outcomes. Approximately 1% to 3% of the morning sample, depending on survey year, had missing triglyceride data and were therefore excluded from these analyses. For calculation of LDL-C by the Friedewald equation, approximately 3% to 5% of participants were excluded because of triglyceride levels greater than 400 mg/dL or missing TC or HDL-C data.

The age-adjusted mean TC level for adults declined linearly from 206 (95% CI, 205-207) mg/dL in 1988-1994 to 203 (95% CI, 201-205) mg/dL in 1999-2002 and to 196 (95% CI, 195-198) mg/dL in 2007-2010 (P < .001 for linear trend) (Table 1). Similar trends over this 22-year period were observed in age-adjusted mean TC levels for men (P < .001 for linear trend) and for women (P < .001 for linear trend).

Statistically significant declining trends in age-adjusted mean TC levels from 1988-1994 to 2007-2010 were observed in all sex and race/ethnicity subgroups except for Mexican American men (P = .03).

From 1988-1994 to 2007-2010, an increasing linear trend in age-adjusted mean HDL-C levels was observed for all adults (50.7 [95% CI, 50.0-51.0] mg/dL vs 52.5 [95% CI, 51.8-53.2] mg/dL; P = .001 for linear trend), for men (P = .007 for linear trend), and for women (P = .001 for linear trend) (Table 1). Age-adjusted mean HDL-C levels increased linearly in non-Hispanic whites of both sexes but not in non-Hispanic blacks or Mexican Americans.

Between 1988 and 2010, a linear decline in age-adjusted mean non–HDL-C level was observed for all adults (from 155 [95% CI,153-157] mg/dL in 1988-1994 to 152 [95% CI, 150-154] mg/dL in 1999-2002 and to 144 [95% CI, 143-145] mg/dL in 2007-2010; P < .001 for linear trend), for men (P < .001 for linear trend), and for women (P < .001 for linear trend) (Table 1).

The age-adjusted geometric mean triglyceride level for all adults increased from 118 (95% CI, 114-121) mg/dL in 1988-1994 to 123 (95% CI, 119-127) mg/dL in 1999-2002 and then declined in 2007-2010 to 110 (95% CI, 107-113) mg/dL (P < .001 for quadratic trend) (Table 2). Similar quadratic trends over this 22-year period were observed for men (P = .01 for quadratic trend) and for women (P < .001 for quadratic trend). Decreases in the age-adjusted geometric mean triglyceride level occurred from 1999-2002 to 2007-2010 for non-Hispanic white men (P < .001) and Mexican American women (P = .001). For non-Hispanic white women, there was a significant quadratic trend over this 22-year period (P < .001) (Table 2).

From 1988 to 2010, there was a decreasing linear trend in age-adjusted mean LDL-C levels for all adults, from 129 (95% CI, 127-130) mg/dL in 1988-1994 to 123 (95% CI, 121-125) mg/dL in 1999-2002 and to 116 (95% CI, 114-117) mg/dL during 2007-2010 (P < .001 for linear trend) (Table 2). Age-adjusted mean LDL-C levels for both men and women also decreased linearly (P < .001 for linear trend for both men and women) and converged over this 22-year period. Although men had a higher age-adjusted mean LDL-C level than women during 1988-1994 (P < .001) and 1999-2002 (P < .001), during 2007-2010 there was no longer a sex difference (P = .37).

Between 1988 and 2010, significant declining trends in mean TC, non–HDL-C, and LDL-C levels of up to 32 mg/dL were generally observed in adults of both sexes aged 50 years or older (P < .001 except for LDL-C in women aged 50-59 years) (eTable 1 and eTable 2). For both men and women aged 50 years or older, decreasing trends in geometric mean triglyceride levels of up to 23 mg/dL were observed during this 22-year period, which were significant only for men aged 50 to 59 years and men aged 70 years or older (eTable 2). From 1988-1994 to 2007-2010, increases in mean HDL-C levels of approximately 3 mg/dL in men aged 50 years or older and ranging from 2 to 4 mg/dL in women aged 40 years or older were observed but were significant only for women aged 40 to 49 years (P < .001) (eTable 1).

From 1988 to 2010, there was an increasing trend in the age-adjusted percentage of adults taking lipid-lowering medications (from 3.4% [95% CI, 2.9%-3.9%] in 1988-1994 to 9.3% [95% CI, 8.7%-10.0%] in 1999-2002 and to 15.5% [95% CI, 14.7%-16.3%] in 2007-2010; P < .001 for linear trend) (Table 3). The increase in lipid medication use was significant for all sex and race/ethnicity subgroups. Among men and women aged 50 years or older, increases in lipid-lowering medications of up to 35% were observed.

Declining trends in age-adjusted mean TC and non–HDL-C levels and increasing trends in age-adjusted mean HDL-C levels similar to those observed in adults aged 20 years or older in the US population between 1988 and 2010 were also seen in adults not receiving lipid-lowering medications (Table 4). For example, the age-adjusted mean TC level for adults not taking lipid-lowering medications decreased from 206 (95% CI, 204-207) mg/dL in 1988-1994 to 204 (95% CI, 202-205) mg/dL in 1999-2002 and to 199 (95% CI, 198-201) mg/dL in 2007-2010 (P < .001 for linear trend) (Table 4). Quadratic trends in age-adjusted geometric mean triglyceride levels and declining linear trends in age-adjusted mean LDL-C levels similar to those observed in the US population over this 22-year period were also observed in adults not taking lipid-lowering medications. For example, the age-adjusted mean LDL-C level for adults not receiving lipid-lowering medications declined from 128 (95% CI, 127-130) mg/dL in 1988-1994 to 124 (95% CI, 122-126) mg/dL in 1999-2002 and to 119 (95% CI, 117-121) mg/dL in 2007-2010 (P < .001 for linear trend) (Table 5).

Statistically significant declines in age-adjusted mean TC, non–HDL-C, and LDL-C levels between 1988 and 2010, similar to those observed for US adults and adults not taking lipid-lowering medications, were seen in obese adults, obese men, and obese women (P < .001 for linear trend). In contrast to adults in the US population and to adults not taking lipid-lowering medications, there was a significant linear decrease in age-adjusted geometric mean triglyceride levels between 1988-1994 and 2007-2010 for obese adults, obese men, and obese women (P < .001 for linear trend) but not a significant quadratic trend. Also, the age-adjusted mean HDL-C level for all obese adults and for obese men did not change significantly, but there was an increasing linear trend in age-adjusted mean HDL-C level for obese women (P = .008 for linear trend) (eTable 3).

After controlling for the possible confounding effects of race/ethnicity, age, BMI, and lipid-lowering medications, a significant decrease in TC of approximately 4 mg/dL per year was observed in both men and women between 1988 and 2010 (P < .001 by Satterthwaite-adjusted F test for trend). Decreases in LDL-C of approximately 6 mg/dL per year for men and 5 mg/dL per year for women (P < .001 by Satterthwaite-adjusted F test for trend) and decreases in non–HDL-C of approximately 6 mg/dL per year were also observed over this 22-year period (P < .001 by Satterthwaite-adjusted F test for trend). High-density lipoprotein cholesterol increased by approximately 2 mg/dL per year in both men and women (P < .001 by Satterthwaite-adjusted F test for trend) (eTable 4).

Between 1988 and 2010, there were declining trends in age-adjusted mean TC, non–HDL-C, and LDL-C levels for adults overall, for men, and for women aged 20 years or older. The declines observed between 1988-1994 and 1999-2002 have continued to 2007-2010. The Healthy People 2010 guideline²⁹ of an age-adjusted mean TC level of 200 mg/dL or less has been achieved in adults, in men, in women, and in all race/ethnicity and sex subgroups. However, the age-adjusted mean LDL-C level in adults of 116 mg/dL is higher than the optimal range of below 100 mg/dL associated with a lower risk of CHD.⁹ Age-adjusted mean HDL-C levels have increased in men and women between 1988 and 2010. Although the age-adjusted geometric mean triglyceride level increased in adults overall, in men, and in women between 1988-1994 and 1999-2002, it decreased thereafter. These same patterns in HDL-C and triglycerides were also observed in adults not receiving lipid-lowering medications.

Declining trends in mean TC, non–HDL-C, and LDL-C levels over this 22-year period have also been observed for all adults, for men, and for women not taking lipid-lowering medications and for obese adults of both sexes; these trends persisted after controlling for the possible confounding effects of race/ethnicity, age, BMI, and use of lipid-lowering medications.

Declines in TC and LDL-C have been reported elsewhere. A community-based study conducted in 3 geographically different parts of France suggested significant declines in TC and LDL-C from 1997 to 2007.³⁰ A report on trends in TC in 199 countries and territories indicated that TC declined in high-income regions of the world (Australasia, North America, and Western Europe).³¹

The favorable trends in TC, non–HDL-C, and LDL-C may be due in part to a decrease in consumption of trans -fatty acids or other healthy lifestyle changes,³² in addition to an increase in the percentage of adults taking lipid-lowering medications. They are unlikely to be the result of changes in physical activity, obesity, or intake of saturated fat. The intake of saturated fat as a percentage of calories did not decrease between 1999 and 2008.³³ Little progress was made from 1998 to 2008 in increasing leisure-time physical activity levels of adults as measured by the criteria defined in the 2008 Physical Activity Guidelines.^34- 36 Also, the prevalence of obesity among adults remains high at more than one-third of the population.^37- 38 Although the percentage of adults receiving lipid-lowering medications continued to increase between 1999-2002 and 2007-2010, declining trends in TC, non–HDL-C, and LDL-C also occurred for adults not taking lipid-lowering medications. The effect of intake of trans -fatty acids on TC, non–HDL-C, and LDL-C, controlling for the possible confounding effects of healthy lifestyles including physical activity, weight loss, and diet, still needs to be investigated.

There is some evidence that the increase in HDL-C and the decrease in triglycerides between 1999 and 2010 are due to changes in cigarette smoking and carbohydrate intake. The age-adjusted percentage of current smokers aged 18 years or older declined from 1999 to 2010.³⁹ Also, there was a decreasing trend in the age-adjusted mean carbohydrate intake of both men and women from 1999 to 2008.³³ Increases in use of specific types of lipid-lowering medications that may also increase HDL-C slightly and decrease triglycerides might also contribute to these changes.

This study had some limitations. Our analysis of trends in lipids among adults is based on 3 NHANES survey periods. Data from other future surveys are needed to confirm the favorable trends. Also, the use of lipid-lowering medications is based on self-report of any lipid-lowering medications rather than on specific types of lipid-lowering medications. The change in the method of measuring HDL-C could have affected the HDL-C results. However, during all 3 survey periods, the values were standardized according to the CDC Lipid Standardization Program to minimize method effects.

In summary, between 1988 and 2010, there was a favorable trend in serum lipid levels among US adults. Further work is needed to assess simultaneously the effects of trans -fatty acids, lipid-lowering medications, and healthy lifestyle factors on TC, HDL-C, non–HDL-C, LDL-C, and triglycerides.