Is race or ethnicity associated with under‐utilization of statins among women in the United States: The study of women's health across the nation

Abstract Background Rates of statin use among minority women are unclear. Hypothesis We hypothesized that statin use would vary by race/ethnicity with lower rates among minority women compared with Whites. Methods Data from the study of women's health across the nation, a multiethnic cohort of women collected between 2009 to 2011 were used to examine reported statin use by race/ethnicity and risk profile. Multivariable logistic modeling was performed to estimate the odds ratio (OR) of statin treatment. Results Of the 2399 women included, 234 had a diagnosis of atherosclerotic disease (ASCVD), 254 were diabetic (without ASCVD), 163 had an LDL ≥190 mg/dL, and 151 had a 10 year ASCVD pooled risk score ≥7.5%. Statins were used by 49.6% of women with CVD; 59.8% of women with diabetes without known ASCVD; 42.3% of women with an LDL ≥190 mg/dL; and 19.9% of women with an ASCVD risk ≥7.5%. Rates of statin use were 43.8% for women with ≥ two prior ASCVD events and 69.4% for women with ≥ one prior ASCVD event plus multiple high‐risk conditions. Among women eligible for statins, Black women had a significantly reduced adjusted odds of being on a statin (OR 0.53, 95% confidence interval [CI] 0.36‐0.78) compared with White women. Conclusions In this cohort of multiethnic women, rates of statin use among women who would benefit were low, with Black women having lower odds of statin use than White women.


| INTRODUCTION
For women, in particular, minority women, CVD is a leading cause of morbidity and mortality in the United States (US). [1][2][3] For over two decades, guidelines and associated clinical trials have recommended statin therapy for women with a history of CVD or CVD risk factors. [4][5][6][7] In 2013, the American Heart Association (AHA) and American College of Cardiology (ACC) guidelines emphasized statins use for four primary groups. 7 These groups include those with (a) atherosclerotic cardiovascular disease (ASCVD); (b) diabetes, age 40 to 75 years; (c) an LDL cholesterol ≥190 mg/dL; or (d) a 10 year ASCVD risk score ≥7.5%. 8,9 Prior studies report women are less likely to receive evidencebased therapies including statins compared to men. [10][11][12][13] However, data on rates of statin use among minority women for these statin eligible groups remains less studied. The study of women's health across the nation (SWAN) is a multiethnic prospective cohort study of midlife women, which includes data on CVD risk factors and medication use.
We sought to describe patterns of statin use among women by race/ ethnicity, using data from the 12th SWAN study visit in 2009 to 2011. We hypothesized that statin use would be lower for Black and Hispanic women with a history of ASCVD, diabetes, or elevated LDL cholesterol compared to White women. We also estimated the number of women by race/ethnicity, who would have been advised to initiate statin therapy secondary to having an ASCVD risk of 7.5% or higher.

| METHODS
SWAN is a community-based longitudinal study that includes a multiethnic population of women transitioning through menopause. The study design has been described in detail elsewhere. 14 A total of 3302 women were enrolled in seven sites within the US, used various sampling frameworks and recruitment strategies to enroll representative groups of women from the surrounding communities. All sites recruited non-Hispanic Whites. Specific racial/ethnic groups were recruited from different sites. Black women were enrolled in the Boston, Massachusetts, Chicago, Illinois, Pittsburgh, Pennsylvania, and southeast Michigan sites. Chinese and Japanese women were enrolled at the Oakland, California, and Los Angeles, California, sites. Hispanic women were enrolled at the Hudson County, New Jersey site.
All women enrolled were age 42 to 52 years upon entry and were perimenopausal. Women were ineligible if they reported no menses in the prior 3 months before enrollment. Women who had a hysterectomy or bilateral oophorectomy were also excluded, as were women taking oral contraceptives or hormonal therapy in the prior 3 months.
Additional exclusion criteria, including a history of CVD prior to enrollment. All participants provided written informed consent prior to enrollment. Each site obtained approval for the study from their respective Institutional Review Board.
To be eligible for this analysis, women had to attend visit 12 and have no missing data for ASCVD risk scoring. The women were categorized as statin eligible if they fit into one of the four groups including: (a) a history of ASCVD diagnosed during follow-up; (b) a diagnosis of diabetes and age 40 to 75 years; (c) an LDL-cholesterol ≥190 mg/dL; or (d) a 10 year ASCVD risk score ≥ 7.5%. To calculate the 10 years ASCVD score, we used the 2013 ACC/AHA guidelines for cardiovascular risk assessment. 8,9 Each woman could contribute to only one group; thus, a woman who had a diagnosis of CVD and had diabetes were grouped with those diagnosed with CVD. The ASCVD risk score was only calculated for women who were not included in the other three groups.
The 2013 ACC/AHA 10 year ASCVD risk score included blood pressure, a diagnosis of hypertension (HTN), a diagnosis of diabetes (defined as a history of self-report, fasting blood glucose >126 or use of antiglycemic medications), current smoking, total cholesterol, and high density lipoprotein cholesterol (HDL-C)levels, in addition to race and age. 8,9 Systolic and diastolic blood pressure (DBP) was measured using a standardized protocol, with two resting measurements averaged.
HTN was defined as systolic blood pressure (SBP) ≥140 mmHg, diastolic blood pressure ≥90 mmHg, or current use of antihypertensive medications (queried at each annual visit). Smoking status at visit 12 was categorized as never/past/current. Body mass index (BMI) (kg/m 2 ) was calculated using the height and weight collected at visit 12.
All lipid, lipoprotein, and apolipoprotein fractions were analyzed on ethylene diamine tetraacetic acid (EDTA)-treated plasma. 15,16 Total cholesterol and triglycerides were analyzed by enzymatic methods on a Hitachi 747 analyzer (Boehringer Mannheim Diagnostics, Indianapolis, Indiana) as previously described. 15 HDL-C was isolated using heparin-2 M manganese chloride. 16 When triglycerides were less than 400 mg/ dL, and the calculated LDL result was obtained by using the Friedewald formula (LDL = total cholesterol -HDL -triglycerides /five).
Women were considered at high-risk for future ASCVD events if they had two or more ASCVD events prior to visit 12 or had one ASCVD event and two or more high-risk conditions including age ≥ 65 years, and LDL ≥100 mg/dL, diabetes mellitus, HTN, current smoking. 17 ASCVD events were self-report. We did not have enough information to consider chronic kidney disease, congestive heart failure, or the guidelines' definition of persistently elevated LDL on maximally tolerated statin therapy and ezetimibe.
Participants were asked about medications at each study visit, and responses were verified by inspection of medication bottles. If the participant forgot to bring medication containers to the study visit, a review of medication lists was performed. Dosage information was not consistently provided and thus was not used for these analyses.
Demographic factors, including age, race/ethnicity, and socioeconomic (SES) characteristics (highest education level attained, income), were collected at baseline. Income was categorized as (< 20 K, 20-<35 K, 35-<50 K, 50-<75 K, and 75 K or more) and financial strain was assessed based on asking each participant was asked "how hard is it to pay for basics" with possible responses ranging from "very hard" to "somewhat hard" to "not hard at all." Medical history was collected via self-report at annual visits. ASCVD (including myocardial infarction, percutaneous coronary intervention or coronary artery bypass grafting, peripheral arterial disease, and stroke or transient ischemic attack), diabetes, and renal disease were asked at each visit.
A family history of premature coronary artery disease (CAD) was defined as CAD in first-degree relatives (male ≤55 years or female ≤65 years).
Menopausal status was defined as pre, peri (early, late), and postmenopausal according to standard definitions. 8 A woman who reported a menstrual period within the past 3 months with no change in regularity was considered premenopausal. Early perimenopause was defined as a menses within the past 3 months with a change in regularity. Women who reported no menses for ≥ three, but at least one period within the last 12 months was considered late perimenopausal. No menses for ≥ 12 consecutive months was considered as postmenopause if no other reason for the amenorrhea was evident.
For these analyses, we classified menopausal status as a surgical  with a 10 year ASCVD risk of 7.5% or higher. (Figure 1). A total of 1597 women either had a 10 year ASCVD risk score <7.5% or were missing data to calculate the score and were excluded from the analyses.
Women who met eligibility criteria for statins were slightly older and more likely to be Black or Hispanic compared to women who did not meet the criteria (Table 1). SES factors also differed between the two groups, with more women who were statins eligible reporting lower levels of education and income and higher rates of difficulty paying for basics. As expected, CVD risk factors were more prevalent among women who met the criteria for statins as compared to women who did not.

| DISCUSSION
In this large, multiethnic cohort of midlife women, we observed the majority of women eligible for statins did not report statins use, with low rates across all racial/ethnic groups and clinical risk groups. Of note, Black women represented the group with the greatest percent of individuals eligible for statins; however, they were least likely to report taking statins.

| Statin use in participants of SWAN with ASCVD is reduced
Statin therapy for adults with CVD has been supported by guidelines and trial evidence for several decades. 6 (58.1% vs 49.6%, respectively). 24 However, the authors also noted a 19% lower rate of statin use among women compared to men. The MEPS data also demonstrated that only one-third of adults with ASCVD were taking a high-intensity statin, which has been observed in other studies. 24,26 Among Marketscan and Medical data, low rates of high-intensity statin use among women with CVD have also been noted. 26 It is likely that among SWAN women, rates of high-intensity statin use would have been lower than the rate of reported statin use as well. Furthermore, Marketscan/Medicare data did not demonstrate an increase in statin use over the 2 years after the 2013 guideline publication. 26 Unfortunately, data on statin dose was not available in SWAN. In a recently published study, which used Medicare and Mar-ketScan data, statin rates were noted to be 52% for those with heart disease and 43% for those with cerebrovascular disease. 29 These rates are minimally higher than we observed in SWAN; however, rates were not reported separately by gender. In a large meta-analysis of 43 studies with over two million participants, rates of statin use were higher than we observed, with a pooled prevalence of statin use among 60% of women, which was significantly lower than the rates noted for men. 13 It should also be noted that MEPS, Medicare, and MarketScan differ from women compared to White or Asian women. However, in multivariable models, these factors did not appear to be associated with receipt of statins. Data from Finland suggests lower SES was associated with lower statin adherence among men but not women. 35 However, Finland has universal health coverage, which limits generalizability to the United States. In the MEPS data, investigators observed no difference in statin use for when the family income was less than 100% of the federal poverty Level; however, being uninsured was associated with lower rates of statin use compared to having public or private insurance. 24

| Study limitations
Our study has several limitations. First, we did not have information on the dose of statins for the majority of women. Therefore, we were unable to determine if women were on recommended doses. We also do not have information on the provider-measured lipid profiles, which may have triggered a statin prescription; however, this would not be relevant for recommendations to use statin in women with CVD or diabetes. Second, SWAN data was collected shortly before the publication of the 2013 guidelines, which outlined specific definitions for the four groups recommended for statin therapy. However, three of these groups (those with CVD, diabetes, or elevated LDL cholesterol) would have been recommended for statins therapy based on earlier guidelines 6 ; thus majority of SWAN women met these earlier recommendations. Among women who did not meet the definitions for one of the four benefit groups, some women reported being on statins. These women may have had an ASCVD risk score, which was higher in the past, but subsequent statin use reduced their risk score.
This limitation would not influence the observed rates of statin use for women with CVD, diabetes or a history of an elevated LDL of 190 mg/dL or greater, but could mean the rate of statin use among women with an ASCVD risk score of 7.5% was higher than we observed. We also could not ascertain if racial differences in statin use reflect differences in adherence or provider recommendations. Lastly, ASCVD events were self-reported, which could lead to misclassification.

| CONCLUSION
This study suggests that women who are eligible for statin therapy are commonly not taking statins, and in particular Black women report low rates of statin use. Strategies to increase statin use among women at risk for future CVD, including minority women, should be considered.