Coronary heart disease (CHD) is the single largest cause. Epidemiology

Transcription

1 Epidemiology Twenty-Year Trends in Serum Cholesterol, Hypercholesterolemia, and Cholesterol Medication Use The Minnesota Heart Survey, to Donna K. Arnett, PhD; David R. Jacobs, Jr, PhD; Russell V. Luepker, MD; Henry Blackburn, MD; Christopher Armstrong, PhD; Steven A. Claas, MS Background Although US cholesterol concentrations have dropped, 50% of adults have total cholesterol concentrations 5.18 mmol/l, putting them at borderline-high risk for heart disease. Whether the decline has continued into the 21st century is unknown. We assessed 20-year trends in cholesterol, hypercholesterolemia, lipid-lowering drug use, and cholesterol awareness, treatment, and control from Minnesota Heart Survey (MHS) data. Methods and Results Five independent, cross-sectional, population-based surveys of 2500 to 5000 adults were conducted in the Minneapolis St. Paul, Minn, area from 1980 to Mean (nonfasting) total cholesterol concentrations have continued a 20-year decline, punctuated by an intervening lull. Age-adjusted mean total cholesterol concentrations in 2000 to 2002 were 5.16 and 5.09 mmol/l for men and women, respectively (in 1980 to 1982, 5.49 and 5.38 mmol/l for men and women, respectively) However, the decline has not been uniform across all age groups. Middle-aged to older people have shown substantial decreases, but younger people have shown little overall change and recently had increased total cholesterol values. The mean prevalence of hypercholesterolemia in 2000 to 2002 was 54.9% for men and 46.5% for women and has decreased significantly for both during the study. Age-adjusted mean high-density lipoprotein cholesterol concentrations in 2000 to 2002 were 1.09 and 1.40 mmol/l for men and women, respectively, and were not different from the prior survey. Lipid-lowering drug use rose significantly for both sexes aged 35 to 74 years. Awareness, treatment, and control of hypercholesterolemia have increased; however, more than half of those at borderline-high risk remain unaware of their condition. Conclusions Although hypercholesterolemia prevalence continued to fall, significant population segments still have cholesterol concentrations near or at the level of increased risk. (Circulation. 2005;112: ) Key Words: cholesterol hypercholesterolemia lipids epidemiology risk factors Coronary heart disease (CHD) is the single largest cause of death in the United States, leading to 1 in every 5 deaths in the United States. In 2002, roughly 13 million Americans had CHD; 2 million of them were hospitalized with the disease. In 2005, the estimated direct and indirect costs of CHD are in excess of $142 billion. 1 Clearly, CHD and its prevention remain important health concerns. Hypercholesterolemia is a major risk factor for CHD. Although mean serum total cholesterol concentrations have dropped in the United States in recent decades, 50% of adults ( 20 years old) have total cholesterol concentrations of at least 5.18 mmol/l (200 mg/dl [1 mmol/l mg/dl]), 2 the level that the National Cholesterol Education Program (NCEP) expert panel considers borderline-high risk. 3 This statistic Clinical Perspective p 3891 takes on special significance in light of recent observations that suggest that downward trends in total serum cholesterol may be flattening. 2,4 Whether this flattening has continued into the early 2000s and whether the observed trends are consistent across all age groups are unknown. Additionally, information about the awareness, treatment, and control of hypercholesterolemia is limited but critical for understanding and improving intervention and prevention. This report describes 20-year (1980 to 1982 through 2000 to 2002) trends in serum cholesterol; hypercholesterolemia; use of lipid-lowering drugs; and cholesterol awareness, treatment, and control with data from the Minnesota Heart Survey (MHS). Received March 17, 2005; revision received September 16, 2005; accepted October 5, From the Department of Epidemiology, University of Alabama at Birmingham (D.K.A.); the Division of Epidemiology and Community Health, University of Minnesota, Minneapolis (D.R.J., R.V.L., H.B., C.A., S.A.C.); and the Department of Nutrition, University of Oslo, Oslo, Norway (D.R.J.). The online-only Data Supplement, which contains Figure I, Figure II, and Figure III, can be found with this article at Correspondence to Donna K. Arnett, PhD, Department of Epidemiology, University of Alabama at Birmingham, 1665 University Blvd, Birmingham, AL. arnett@uab.edu American Heart Association, Inc. Circulation is available at DOI: /CIRCULATIONAHA

2 Arnett et al Twenty-Year Trends in Serum Cholesterol 3885 Methods Details of the population and sampling methods have been published. 5 7 The MHS is an ongoing population-based surveillance of trends in cardiovascular disease risk factors, morbidity, and mortality in independent, cross-sectional, probability samples of adults in the Minneapolis St. Paul, Minn, metropolitan area. Surveys were conducted in 1980 to 1982, 1985 to 1987, 1990 to 1992, 1995 to 1997, and 2000 to 2002 according to consistent sampling protocols. Samples were selected by 2-stage cluster design. In the first sampling stage, MHS randomly selected 40 of the 704 clusters composing the 7-county metropolitan area, with each cluster containing 1000 households. Low housing growth during the 1980s allowed the initial 40 clusters to be used through 1990 to For the 1995 to 1997 survey, increased growth demanded the addition of 4 new randomly chosen clusters from among 10 high-growth clusters to maintain the population representation of the sample. Budget constraints during 1996 required early termination of the second year of sampling; therefore, bootstrap estimation was done for 26 unsampled clusters. Survey samples closely reflected the age and ethnic distributions of the 7-county metropolitan area according to 1980, 1990, and 2000 US census information. In 1980 to 1982, the survey sample was 95.6% white and 45.8% male, whereas the metropolitan population was 94.8% white and 48.6% male. In 1990 to 1992, the survey sample was 93% white and 46.5% male, whereas the metropolitan population was 92.3% white and 49.0% male. In 2000 to 2002, the survey sample was 85.9% white and 47.2% male, whereas the metropolitan population was 84.7% white and 49.3% male. In the second stage of each survey, a sample of households was randomly selected in each chosen cluster to generate a sample size of 5000 adults (aged 25 to 74 years for 1980 to 1982 and 1985 to 1087; aged 25 to 84 years thereafter). Only those households with at least 1 age-eligible person were included. After household enumeration, age-eligible individuals were interviewed in the home. For the 1980 to 1981, 1995 to 1997, and 2000 to 2002 surveys, all age-eligible individuals in the household were invited to participate, whereas a randomly selected individual was interviewed in other survey years. Because all analyses are sex specific and 97% of homes with 2 adults comprised opposite-sex couples, the sampling design that used all age-eligible individuals and 1 age-eligible individual per household yielded almost equivalent prevalence estimates. The home interview solicited information regarding sociodemographic characteristics; health attitudes; medical history with respect to hypertension, hypercholesterolemia, and diabetes; medications taken for these conditions; and behaviors. Subjects were then invited to a survey center where detailed questions were asked regarding medical history, smoking behavior, leisure-time physical activity during the past year, and 24-hour dietary recall. Blood pressure, height, weight, waist-to-hip ratio, and total cholesterol measures were taken. Response rates for the completed home interview, clinic visit, and overall (in that order) were 91%, 77%, and 69% (1980 to 1982); 88%, 71%, and 68% (1985 to 1987); 82%, 71%, and 68% (1990 to 1992); 83%, 68%, and 65% (1995 to 1997); and 75%, 64%, and 64% (2000 to 2002). Those with a completed survey were slightly more likely to be married, employed, nonsmoking individuals with more education than nonrespondents. This research was approved by the University of Minnesota Institutional Review Board, and informed consent was collected from all participants. Height was measured in stocking feet with a wooden triangle and a rigid ruler attached to a wall. Weight was measured without coat and shoes with a beam balance; the balance was calibrated daily with a kg (50-lb) weight. Body mass index was calculated as weight (kg) divided by the square of height (m 2 ). Current smoking was defined as having smoked 100 cigarettes in a lifetime and smoking currently. Leisure-time physical activity was measured with the Minnesota Leisure-Time Physical Activity Questionnaire 8 in a 50% systematic sample of participants (ie, every other participant listed on the scheduling log). In the 2000 to 2002 survey, physical activity measures were collected for all participants. The questionnaire queried participants on 63 leisure-time activities, the number of times the activity was performed per month, and the number of months the activity was performed during the past year. Alternate participants who did not complete the leisure-time questionnaire completed a 24-hour dietary recall administered by trained, certified interviewers (except during 2000 to 2002, when all participants completed the recall). The same food database was used for all surveys, although new foods were added so that the energy and nutrient composition of diets was accurately quantified according to the current food supply. 4,9 Nutrient intakes were estimated with the University of Minnesota Nutrition Coordinating Center Food and Nutrient Database. 10 Serum total cholesterol was measured in nonfasting samples with an AutoAnalyzer II (Technicon Corporation) with a nonenzymatic method between 1980 to 1982 and 1990 to 1992 and with an enzymatic method thereafter. The analytical laboratories participated successfully in the Centers for Disease Control and Prevention cholesterol standardization programs for all surveys. Between 1980 to 1982 and 1990 to 1992, laboratory drift assessment did not correctly adjust cholesterol levels. Therefore, we recalibrated the first 3 surveys against the reference Abell-Kendall method ( 0.17 mmol/l for 1980 to 1982; 0.22 mmol/l for 1985 to 1987; and 0.11 mmol/l for 1990 to 1992; P for all). 11 The Abell-Kendall method adjusted cholesterol concentrations for the first 3 surveys were used in this analysis. The enzymatic method produced results equivalent to the reference method. 4 Laboratory quality-control coefficient of variance estimates for total cholesterol were generally between 1.2% and 2.8% throughout the study, and 93% of all coefficient of variance estimates for total cholesterol were 3%. HDL (high-density lipoprotein) cholesterol was measured with an enzymatic method after precipitation of non-hdl cholesterol with heparin and Mn 2 (1980 to 1992) 12 or magnesium dextran sulfate (1995 to 2002). 13 Because of difficulties associated with the intercalibration of these 2 methods, HDL is reported here for 1995 to 2002 only. Hypercholesterolemia was defined as either a serum total cholesterol concentration 5.18 mmol/l or answering affirmatively to the question, Are you currently taking medication prescribed for you by a doctor for high blood fats? Patients with controlled hypercholesterolemia were defined as those taking lipid-lowering medication and having a serum total cholesterol concentration 5.18 mmol/l. For comparison with previous studies, the prevalence of hypercholesterolemia was also calculated from a 6.21 mmol/l cutpoint. Data were analyzed with SAS analytic software (SAS Institute, Inc). 14 All analyses were sex specific. The means and frequencies of study variables were calculated for each survey period and for 10-year age groups. Only participants aged 25 to 74 years were included in this analysis. To compute unbiased variances of the sample means, we took into account the neighborhood cluster design, adding the neighborhood cluster as a random-effect term. Two statistical tests were used to characterize temporal trends. Restricted maximum-likelihood regressions were used to test for overall linear trends through the duration of the study. ANCOVA was used to test for differences between temporally adjacent surveys. Results Descriptive characteristics of the study samples are shown in Table 1. We found no clustering within household for ancillary risk factors such as smoking and leisure-time physical activity. When adjusted for age, cholesterol intake, lipid-reducing drug use, and body mass index, serum total cholesterol estimates were virtually the same as those generated by the age-adjusted-only model; therefore, we report age-adjusted values here. Serum Total Cholesterol Total cholesterol concentrations for men and women are shown in Figure 1. Age-adjusted mean total cholesterol in 2000 to 2002 was 5.16 and 5.09 mmol/l for men and women,

3 3886 Circulation December 20/27, 2005 TABLE 1. Descriptive Characteristics of the MHS Study Population Men Women n Mean age, y * Education High school * 6.5* * * 4.7 High school * 30.3* 26.4* 22.9* Some college College graduate * 41.4* * 28.8* 36.3* 40.8* Ethnicity White * 86.0* * 85.8* Black * Other * * 8.5* Median LTPA, kcal/d Current smokers, * 25.3* * * 19.5 % Body mass index, * * * 26.6* 27.2* 28.4* kg/m 2 Fat intake, % * 35.0* 30.7* 32.7* * 34.3* 30.7* 32.2* P:S ratio 0.5o 0.56* 0.60* * * 0.67 Keys score * 42.0* 37.0* * 44.3* 41.6* 35.1* 36.7* Cholesterol intake, mg/1000 kcal 191.8* 164.2* 142.0* 113.1* 126.4* 186.0* 164.1* 134.8* 103.8* 116.8* LTPA indicates leisure-time physical activity; P:S ratio, polyunsaturated to saturated fat ratio. Other abbreviations are as defined in text. *P 0.05 when compared with the preceding survey. Diet and leisure-time physical activity information was gathered from an alternate, randomly selected 50% of the sample of each survey year except for survey, when both were collected in 100% of the sample. Mean dietary fat intake as a proportion of total calories. Keys score 1.35 (2S P) 1.5Z 2, where S saturated fat as a proportion of total calories, P polyunsaturated fat as a proportion of total calories; and Z cholesterol intake in mg/1000 kcal. respectively. Age-adjusted means for both men and women showed significant (P ) downward linear trends from 1980 to 1982 to 2000 to In both men and women, age-adjusted mean total cholesterol also decreased significantly (P 0.05) between each temporally adjacent survey except for 1990 to 1992 to 1995 to 1997, when there were no significant changes. In men, from 1980 to 1982 to 1995 to 1997, total cholesterol was highest in either the 45- to 54- or the 55- to 64-year-olds, but more recently (2000 to 2002), the 35- to 44-year-olds showed the highest concentration. In women, total cholesterol was highest in 65- to 74-year-olds in each survey. In both men and women, all age groups except 25- to 34-year-olds showed significant (P 0.05) 20-year linear decreases. The 25- to 34-year-old men and women showed no overall change in total cholesterol across the 5 survey periods. Figure I in the online-only Data Supplement shows a comparison of total cholesterol concentrations in the entire MHS population, by sex and age group, with total cholesterol of those not taking lipid-reducing drugs. Trends have generally been similar, except for the lower concentrations in the 2000 to 2002 survey for 45- to 74-year-old men and 65- to 74-year-old women in the unabridged sample. Serum HDL Cholesterol Serum HDL data for 1995 to 2002 are shown in Table 2. Age-adjusted mean ( SE) HDL cholesterol concentrations for men were in 1995 to 1997 and in 2000 to Age-adjusted mean HDL cholesterol concentrations for women were in 1995 to 1997 and in 2000 to Prevalence of Hypercholesterolemia Figure 2 shows sex-specific, age-adjusted mean and 10-yearage-group hypercholesterolemia trends. With the mmol/l definition, the age-adjusted mean hypercholesterolemia prevalence in 2000 to 2002 was 54.9% for men and 46.5% for women (23.9% and 17.3% for men and women, respectively, with the 6.21-mmol/L cutpoint). None of these values varied significantly (P 0.05) from the 1995 to 1997 survey. The age-adjusted mean prevalence of hypercholesterolemia defined by a 5.18-mmol/L threshold showed significant, downward, study-long linear trends for both men and women; with the 6.21-mmol/L cutpoint, neither men nor women showed a significant overall trend in age-adjusted means (see Figure II in the Data Supplement). The ageadjusted mean prevalence of hypercholesterolemia ( 5.18-

4 Arnett et al Twenty-Year Trends in Serum Cholesterol 3887 Figure 1. Age-adjusted (adj.) serum total cholesterol trends. Sig LT indicates significant linear trends throughout the study (P ; except for 35- to 44-year-old men, P , and 35- to 44-year-old women, P ). Encircled S indicates significant (P 0.05) difference between adjacent surveys. Error bars indicate 95% confidence intervals. Note difference in vertical axis scales. mmol/l cutpoint) decreased when those taking lipid-lowering drugs were omitted (see Figure III in the Data Supplement); this analysis may underestimate the true prevalence of hypercholesterolemia because most of those taking such drugs would have hypercholesterolemia if they were untreated. TABLE 2. Serum HDL Cholesterol Concentrations (mmol/l) Men Women years years years years years Age-adjusted mean There were no significant differences (P 0.5) between and surveys. Figure 2. Age-adjusted (adj.) prevalence of hypercholesterolemia ( 5.18 mmol/l or taking lipid-reducing drugs) trends. Sig LT indicates significant linear trends throughout the study (P 0.05;). Encircled S indicates significant (P 0.05) difference between adjacent surveys. Prevalence of Lipid-Reducing Drug Use The age-adjusted prevalence of lipid-reducing drug use is shown in Figure 3. In both men and women, drug use has shown a significant (P 0.002), linear 20-year increase in all age groups except 25- to 34-year-olds. Successively older age groups have generally had higher frequencies of lipidreducing drug use. Cholesterol Awareness, Treatment, and Control Sex-specific, age-adjusted trends in the fraction of hypercholesterolemic participants who where aware of their cholesterol level, were pharmacologically treating their hypercholesterolemia, and were controlling their cholesterol (ie, successfully reducing total cholesterol to 5.18 mmol/l) are shown in Figure 4. The fraction of both men and women who were unaware of their hypercholesterolemia continued to drop to 53% to 55% in 2000 to Of those individuals aware of their hypercholesterolemia, an increasing percentage in 2000 to 2002 were successfully treating their condition (13.1% of men, 6.0% of women). The percentage of both sexes who were aware of but unsuccessfully treating their cholesterol remained at 6% for the 2 most recent surveys. Those aware of their hypercholesterolemia but who were not undergoing treatment has been relatively stable in men (28%) and women (30% to 33%) since 1990 to 1992.

5 3888 Circulation December 20/27, 2005 Figure 3. Age-adjusted prevalence of the use of lipid-reducing drugs. Sig LT indicates those series that demonstrated significant linear trends throughout the course of the study (P ; except for women 35 to 44 years old, P ). Adjacent-survey significant differences are not shown. Discussion Serum Total Cholesterol The recent data suggest that the apparent flattening of the downward trend in total cholesterol (from 1990 to 1992 to 1995 to 1997) did not continue into 2000 to 2002; ageadjusted mean concentrations have again declined for both sexes. This extends the trend of generally declining serum total cholesterol concentrations observed nationally since ,15 21 Decreases in total cholesterol both 20-year trends and adjacent-survey differences are not, however, uniform across age groups (Figure 1). Most obviously, neither young (25 to 34 years) men nor young women have shown any significant trends in total cholesterol across the study, and in the past decade, both have posted at least 1 significant (P 0.05) adjacent-survey increase. Differences in trends among other age groups are more difficult to characterize. In women, the 3 oldest age groups (45 to 74 years) all showed significant declines in total cholesterol in the most recent survey, but the 2 youngest showed either no change (35 to 44 years) or an increase (25 to 34 years). The situation is similar, albeit not as simple, in men. The 3 oldest age groups (45 to 74 years) also showed significant recent declines. As in women, total cholesterol in 35- to 44-year-old men did not change significantly from 1995 to 1997 to 2000 to (This group showed nonsignificant increases during this period.) The 25- to 34-year-old men did show a recent significant adjacent-survey decline; this, however, came after a 1990 to 1992 to 1995 to 1997 increase. Taken together, Figure 4. Cholesterol awareness, treatment, and control for hypercholesterolemic ( 5.18 mmol/l or taking lipid-reducing drugs) men and women. Groups include those aware of, treating, and controlling their hypercholesterolemia; those aware, treating, but not controlling; those aware, but not treating and not controlling; and those unaware of their hypercholesterolemic status. these age-specific observations suggest that the rate of total cholesterol decline in the population may vary across ages, with older (and generally higher total cholesterol) groups showing a potentially more precipitous decrease. We compared the MHS results with the National Health and Nutrition Examination Survey (NHANES) values. Although the mean MHS total cholesterol estimates for 2000 to 2002 are lower than the NHANES values for 1999 to 2000 for both men (MHS, 5.16; NHANES, 5.25 mmol/l) and women (MHS, 5.09; NHANES, 5.28 mmol/l), 2 the trend in cholesterol in Minnesota has mirrored national trends. MHS cardiovascular risk factor trends have historically been harbingers of national trends. If these trends continue as they have in the past, 6,15 the recent flattening in national total cholesterol trends may also prove temporary. There are no other serial cross-sectional surveys in the United States that have collected both hypercholesterolemia and serum measures of cholesterol with which MHS data can be compared. Serum HDL Cholesterol Because of MHS changes in HDL measurement methods between the 1980 to 1992 and 1995 to 2002 survey periods, we were unable to comprehensively evaluate the long-term trends in HDL cholesterol. Unlike total cholesterol, there is

6 Arnett et al Twenty-Year Trends in Serum Cholesterol 3889 no gold standard measurement for HDL cholesterol that provides a benchmark against which to normalize periodspecific HDL values. Between 1995 to 1997 and 2000 to 2002, when HDL was measured with a magnesium dextran sulfate precipitate, there were no significant changes in HDL cholesterol in the population of Minneapolis St. Paul. The pattern of HDL showed that younger women tended to have lower concentrations of HDL; however, age-related trends in men were not as clear. Prevalence of Hypercholesterolemia The overall population decrease in serum cholesterol concentrations since 1980 is reflected in the small but significant declines in the prevalence of hypercholesterolemia, defined as total cholesterol 5.18 mmol/l or using lipid-reducing drugs. If the lipid-reducing drug criterion is removed from the definition of hypercholesterolemia (Figure III in the Data Supplement), the mean prevalence of hypercholesterolemia has shown steeper downward trends in both men and women during the course of the study, although this method of analysis may underestimate prevalence by omitting treated people who would have been hypercholesterolemic if they were untreated. The MHS 2000 to 2002 age-adjusted, mean, measured hypercholesterolemia prevalence figure for men is similar to the 1999 to 2000 national finding (MHS, 54.9%; NHANES, 55.7%; the MHS figure for women, however, is considerably lower than the national finding [MHS, 46.5%; NHANES, 53.5%]). 2 The difference in women in Minnesota from the national sample may be due to the higher level of education in Minnesota women and their access to medical care relative to the general US population. This is supported by Behavioral Risk Factor Surveillance System (BRFSS) data (from 1995 to 2002 only) that suggest that improvements in education for Minnesota men is generally closer to the national trend for men than Minnesota women are to the national trend for women. The same is true for access to health care: Minnesota women are generally outpacing Minnesota men in gains on the national average for healthcare access (BFRSS data at Factors Affecting Serum Cholesterol Trends The large increase in the use of lipid-reducing drugs observed in this study is comparable to the pattern described for US retail pharmacies: The number of lipid-reducing medications dispensed nearly doubled between 1991 and The dramatic increase in the use of lipid-reducing drugs since 1980 to 2002 (Figure 3) invites an analysis of the effects of this trend on total cholesterol and the prevalence of hypercholesterolemia. A comparison of total cholesterol concentrations in the entire MHS population with total cholesterol of those not taking lipid-reducing drugs (Figure I in the Data Supplement) shows that trends have generally been quite similar in the drug-taking versus entire MHS sex and age groups; however, recent data suggest potentially important differences. In the 2000 to 2002 survey, 45- to 74-year-old men and 65- to 74-year-old women in the total MHS population (ie, including those taking lipid-reducing drugs) showed a steeper downward trend in total cholesterol concentrations than those in the nonmedicated population. These trends suggest that the higher prevalence of lipid-reducing drug use in older age groups may be partially responsible for these groups continued reduction of total cholesterol. In the absence of adoption of lipid-lowering therapy by a large segment of the population, the downward trends in total cholesterol would be less dramatic. This observation highlights the challenge of the use of a composite hypercholesterolemia definition in assessing population trends, because the definition reflects the mixture of prescription drug use and the laboratory cholesterol value. Therefore, the trend in the true level of cholesterol in the population unconfounded by treatment cannot be ascertained. Cholesterol Awareness, Treatment, and Control Increases in the awareness, treatment, and control of hypercholesterolemia are encouraging; however, more than half of those at or above borderline-high risk remain unaware of their condition (Figure 4). In 2000 to 2002, 46% of hypercholesterolemic ( 5.18-mmol/L cutpoint) individuals claimed to be aware of their condition; this is somewhat higher than the 35.0% reported by NHANES for 1999 to However, recently released NHANES data (for the 6.21-mmol/L cutpoint only) for 1999 to 2002 suggest that the national rate for awareness may be approaching the MHS rate (50.9% for NHANES 1999 to 2000; 63.3% for NHANES 1999 to 2002; 69% for MHS). 2,23 The percentage of those who are aware of their hypercholesterolemia but who are not treating it has fallen throughout the study but remains high (58.5% for men, 73.7% for women); these figures are comparable to the 1999 to 2000 national estimates for men (60.0%) and women (71.2%). 2 The percentage of those treating their hypercholesterolemia who have successfully controlled it has increased throughout the study and in 2000 to 2002 was 66.9% for men and 50.0% for women; NHANES reported 53.6% for men and 36.3% for women for 1999 to As emphasized elsewhere, 24,25 MHS data suggest that bridging the gap between screening and control remains an important public health goal. In 2000 to 2002, awareness, treatment, and control rates were all lower in women than in men; in fact, the percentage of hypercholesterolemic women successfully controlling their cholesterol (6.0%) was less than half that of men (13.1%). Although control rates have increased through the MHS study, this ratio has been observed since 1990 to 1992 and mirrors the difference observed in the most recent NHANES sample (7.5% for men, 3.7% for women). Sex bias in lipid management has been documented in other studies 26 28; however, the causal factors leading to this bias remain unclear. 29,30 Study Limitations Our definition of hypercholesterolemia based on total cholesterol rather than low-density lipoprotein cholesterol may be viewed as a limitation of the study. However, the differential risk levels associated with total cholesterol s constituent components were not as firmly established when MHS was begun as they are today, and validated analytic methods for low-density lipoprotein were not developed until recently. These historical constraints, combined with the that fact total cholesterol can be reliably measured in fasting or nonfasting

7 3890 Circulation December 20/27, 2005 samples 31,32 and that NCEP sanctions the use of nonfasting total cholesterol measurements for making initial health management decisions, 3 explain and legitimize our choice of using total cholesterol throughout the study. The overall response rate for the MHS survey declined slightly during the 20-year period. It should be noted that we calculated a true response rate, wherein the denominator was inflated to account for the number of households that we were unable to contact or that had members who refused to participate, multiplied by the average number of eligible individuals in a household. It is unlikely that the 5% drop in participation resulted in a significant bias in the observed cholesterol trends, because the sample survey closely matched the age and ethnic distributions of the Minneapolis St. Paul population at large at each survey period. Because of the number of statistical tests performed in these analyses and because we made no adjustments for multiple testing, some caution must be taken when interpreting our reported probability values. Because Minnesota ranks lowest among US states in age-adjusted death rates for total cardiovascular disease, 1 MHS data may serve better as a point of comparison rather than a basis for generalization for cardiovascular risk factor trends in the wider population. Public Health Implications Mean serum total cholesterol concentrations have continued a 20-year decline in the Minneapolis St. Paul population, punctuated by an intervening lull. However, this trend has not been uniform across all age groups. Although middle-aged to older people (45 to 74 years) have shown substantial decreases, younger (25 to 44 years) people have shown little overall change and have, in fact, recently experienced increased total cholesterol concentrations. Although the prevalence of untreated hypercholesterolemia has continued to fall, significant segments of the population still have serum cholesterol concentrations near or at the NCEP level of borderline-high risk. The greater than one half of the population who remain unaware of their hypercholesterolemic status points to the continued need for improved screening. The MHS data reinforce that improved screening alone, however, is insufficient; we must continue to encourage behavioral, dietary, and pharmacological control of cholesterol in those individuals who are aware of their cholesterol risk. Acknowledgment This research was supported by a grant from the National Institutes of Health, National Heart, Lung, and Blood Institute, 5 R01 HL23727, Community Surveillance of Cardiovascular Disease Risk Factor Survey (MHS). None. Disclosures References 1. American Heart Association. Heart Disease and Stroke Statistics 2005 Update. Dallas, Tex: American Heart Association; Ford ES, Mokdad AH, Giles WH, Mensah GA. Serum total cholesterol concentrations and awareness, treatment, and control of hypercholesterolemia among US adults: findings from the National Health and Nutrition Examination Survey, 1999 to Circulation. 2003;107: 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). Bethesda, Md: National Heart, Lung, and Blood Institute; Arnett DK, McGovern PG, Jacobs DR Jr, Shahar E, Duval S, Blackburn H, Luepker RV. Fifteen-year trends in cardiovascular risk factors ( through ): the Minnesota Heart Survey. Am J Epidemiol. 2002;156: Sprafka JM, Burke GL, Folsom AR, Luepker RV, Blackburn H. Continued decline in cardiovascular disease risk factors: results of the Minnesota Heart Survey, and Am J Epidemiol. 1990; 132: McGovern PG, Pankow JS, Shahar E, Doliszny KM, Folsom AR, Blackburn H, Luepker RV. Recent trends in acute coronary heart disease: mortality, morbidity, medical care, and risk factors; the Minnesota Heart Survey Investigators. N Engl J Med. 1996;334: Luepker RV, Jacobs DR, Gillum RF, Folsom AR, Prineas RJ, Blackburn H. Population risk of cardiovascular disease: the Minnesota Heart Survey. J Chronic Dis. 1985;38: Taylor HL, Jacobs DR Jr, Schucker B, Knudsen J, Leon AS, Debacker G. A questionnaire for the assessment of leisure time physical activities. J Chronic Dis. 1978;31: Harnack L, Lee S, Schakel SF, Duval S, Luepker RV, Arnett DK. Trends in the trans-fatty acid composition of the diet in a metropolitan area: the Minnesota Heart Survey. J Am Diet Assoc. 2003;103: Schakel SF, Sievert YA, Buzzard IM. Sources of data for developing and maintaining a nutrient database. J Am Diet Assoc. 1988;88: Myers GL, Cooper GR, Winn CL, Smith SJ. The Centers for Disease Control-National Heart, Lung, and Blood Institute Lipid Standardization Program. An approach to accurate and precise lipid measurements. Clin Lab Med. 1989;9: National Heart, Lung, and Blood Institute. Lipid and lipoprotein analysis. In: Manual of Operations, Lipid Research Clinics Program. Bethesda, Md: National Heart, Lung, and Blood Institute; Warnick GR, Benderson J, Albers JJ. Dextran sulfate-mg 2 precipitation procedure for quantitation of high-density-lipoprotein cholesterol. Clin Chem. 1982;28: SAS/STAT Guide. Cary, NC: Statistical Analysis Systems Institute, Inc; Burke GL, Sprafka JM, Folsom AR, Luepker RV, Norsted SW, Blackburn H. Trends in CHD mortality, morbidity and risk factor levels from 1960 to 1986: the Minnesota Heart Survey. Int J Epidemiol. 1989; 18:S73 S Erlinger TP, Pollack H, Appel LJ. Nutrition-related cardiovascular risk factors in older people: results from the Third National Health and Nutrition Examination Survey. J Am Geriatr Soc. 2000;48: Szklo M, Chambless LE, Folsom AR, Gotto A Jr, Nieto FJ, Patsch W, Shimakawa T, Sorlie P, Wijnberg L. Trends in plasma cholesterol levels in the Atherosclerosis Risk in Communities (ARIC) study. Prev Med. 2000;30: Johnson CL, Rifkind BM, Sempos CT, Carroll MD, Bachorik PS, Briefel RR, Gordon DJ, Burt VL, Brown CD, Lippel K, Cleeman JI. Declining serum total cholesterol levels among US adults. the National Health and Nutrition Examination Surveys. JAMA. 1993;269: Sytkowski PA, Kannel WB, D Agostino RB. Changes in risk factors and the decline in mortality from cardiovascular disease: the Framingham Heart Study. N Engl J Med. 1990;322: Burke GL, Sprafka JM, Folsom AR, Hahn LP, Luepker RV, Blackburn H. Trends in serum cholesterol levels from 1980 to 1987: the Minnesota Heart Survey. N Engl J Med. 1991;324: Sempos CT, Cleeman JI, Carroll MD, Johnson CL, Bachorik PS, Gordon DJ, Burt VL, Briefel RR, Brown CD, Lippel K, Rifkind BM. Prevalence of high blood cholesterol among US adults: an update based on guidelines from the second report of the National Cholesterol Education Program Adult Treatment Panel. JAMA. 1993;269: Siegel D, Lopez J, Meier J. Use of cholesterol-lowering medications in the United States from 1991 to Am J Med. 2000;108: Disparities in screening for and awareness of high blood cholesterol United States, MMWR Morb Mortal Wkly Rep. 2005;54: Ruof J, Klein G, Marz W, Wollschlager H, Neiss A, Wehling M. Lipidlowering medication for secondary prevention of coronary heart disease in a German outpatient population: the gap between treatment guidelines and real life treatment patterns. Prev Med. 2002;35:48 53.

8 Arnett et al Twenty-Year Trends in Serum Cholesterol Natarajan S, Nietert PJ. National trends in screening, prevalence, and treatment of cardiovascular risk factors. Prev Med. 2003;36: Miller M, Byington R, Hunninghake D, Pitt B, Furberg CD. Sex bias and underutilization of lipid-lowering therapy in patients with coronary artery disease at academic medical centers in the United States and Canada; Prospective Randomized Evaluation of the Vascular Effects of Norvasc Trial (PREVENT) Investigators. Arch Intern Med. 2000;160: Nau DP, Mallya U. Sex disparity in the management of dyslipidemia among patients with type 2 diabetes mellitus in a managed care organization. Am J Manag Care. 2005;11: O Meara JG, Kardia SL, Armon JJ, Brown CA, Boerwinkle E, Turner ST. Ethnic and sex differences in the prevalence, treatment, and control of dyslipidemia among hypertensive adults in the GENOA study. Arch Intern Med. 2004;164: Kim C, Hofer TP, Kerr EA. Review of evidence and explanations for suboptimal screening and treatment of dyslipidemia in women: a conceptual model. J Gen Intern Med. 2003;18: Berra K. Women, coronary heart disease, and dyslipidemia: does gender alter detection, evaluation, or therapy? J Cardiovasc Nurs. 2000;14: Rifai N, Merrill JR, Holly RG. Postprandial effect of a high fat meal on plasma lipid, lipoprotein cholesterol and apolipoprotein measurements. Ann Clin Biochem. 1990;27(pt 5): Bachorik PS, Cloey TA, Finney CA, Lowry DR, Becker DM. Lipoprotein-cholesterol analysis during screening: accuracy and reliability. Ann Intern Med. 1991;114: CLINICAL PERSPECTIVE Mean serum total cholesterol concentrations have continued a 20-year decline in the Minneapolis St. Paul Minnesota Heart Survey (MHS) population, punctuated by an intervening lull. However, this trend has not been uniform across all age groups. Although middle-aged to older people (45 to 74 years) have shown substantial decreases, younger (25 to 44 years) people have shown little overall change and have, in fact, recently experienced increased total cholesterol concentrations. Although the prevalence of untreated hypercholesterolemia has continued to decline, significant segments of the population still have serum cholesterol concentrations near or at the National Cholesterol Education Panel level of borderline-high risk. Practicing clinicians should be aware that more than one half of the Minneapolis St. Paul population remains unaware of their hypercholesterolemic status. This corroborates national findings and points to the continued need for improved screening. The percentage of those who are aware of their hypercholesterolemia but who are not treating it remains high (58.5% for men, 73.7% for women in MHS 2000 to 2002), as does the percentage of those unsuccessfully treating their hypercholesterolemia (33.1% for men and 50.0% for women in MHS 2000 to 2002). By guiding patients in their behavioral, dietary, and pharmacological interventions, physicians can play an important role in helping narrow the significant gap between cholesterol screening and control. Finally, although the causal factors contributing to sex bias in lipid management (13.1% of hypercholesterolemic men controlled their cholesterol, but only 6.0% of women in MHS 2000 to 2002 were doing so) remain unknown, clinicians should carefully assess their screening and intervention practices with respect to sex bias.