Article Text
PDF
Abstract
Background Recent research indicates that sexual minority women are at increased risk for cardiovascular disease (CVD) compared with heterosexual women; however, few studies of CVD risk exist for sexual minority men (SMM). This study aimed to determine whether disparities in CVD risk exist for SMM and if CVD risk is consistent across subgroups of SMM.
Methods This study utilised publicly available data from the National Health and Nutrition Examination Survey (NHANES), pooled from 2001 to 2010. CVD risk was calculated using the Framingham General CVD Risk Score and operationalised as the ratio of a participant's vascular and chronological age. Differences in this ratio were examined between heterosexual and SMM as a whole, and within subgroups of SMM.
Results SMM had vascular systems that were, on average, 4% (95% CI −7.5% to −0.4%) younger than their heterosexual counterparts; however, adjustment for education and history of hard drug use rendered this difference statistically insignificant. Analysis of SMM subgroups revealed increased CVD risk for bisexual men and decreased CVD risk for both gay and homosexually experienced heterosexual men when compared with heterosexual men. Differences in CVD risk persisted for only bisexual and homosexually experienced heterosexual men after adjustment for education and history of hard drug use.
Conclusions Subgroups of SMM are at increased risk for CVD compared with heterosexual men, and this increased risk cannot be completely attributed to differences in demographic characteristics or negative health behaviours.
- Cardiovascular disease
- GENDER
- Health inequalities
- SOCIAL INEQUALITIES
Statistics from Altmetric.com
Introduction
Cardiovascular disease (CVD) is the leading cause of death in the USA, accounting for approximately one in every three deaths.1 The direct and indirect costs of CVD exceeded $297 billion in 2008, and it is expected that the direct costs of CVD will triple by 2030.1 While CVD affects both men and women, men accounted for over half of all US heart disease deaths in 2009 and comprised between 70% and 89% of the total sudden cardiac deaths.1 ,2
A subgroup of men who may be at increased risk for CVD, relative to all men, are sexual minority men (SMM, ie, men who identify as gay or bisexual, or who engage in same-sex sexual behaviour). Recent research has shown that SMM have disparately higher rates of smoking, alcohol and drug use, all of which may contribute to increased CVD risk.3–8 Moreover, SMM are more likely to have limited access to healthcare and to culturally competent care, which have been linked to increased rates of CVD among other stigmatised groups.9–20 Yet, despite this evidence suggesting that CVD disparities may exist for SMM, there is a dearth of empirical work examining CVD and CVD risk in this population.5 ,21 ,22 Of the work that has been performed, the majority has focused on the potential for increased CVD in HIV positive men.5
At present, only a few studies have examined differences in the prevalence of CVD diagnosis among SMM with population-based data.11 ,23 ,24 For example, using data from the California Quality of Life Survey, Cochran and Mays found that homosexually experienced heterosexual men were significantly more likely to report a diagnosis of heart disease compared with heterosexual men; however, there were no significant differences in self-reported heart disease for gay or bisexually identified men.23 Similarly, using aggregated data from the Massachusetts Behavioral Risk Factor Surveillance Surveys from 2001 to 2008, Conron et al11 found no difference in the report of CVD between heterosexual and gay or bisexual men; however, bisexual men and women were more likely to report multiple risks for CVD. These findings suggest that certain subgroups of SMM (ie, bisexual and homosexually experienced heterosexual men) may be at increased risk for CVD; however, they are difficult to generalise as each study utilised data collected from only one state.
In our recent work, we found differences in CVD risk, as measured by the Framingham General CVD risk score, for sexual minority women (SMW) who participated in the National Health and Nutrition Examination Survey (NHANES).25 In particular, these results indicated that SMW were at increased risk for CVD compared with heterosexual women, even after accounting for differences in smoking and alcohol use. At present, no such work exists for SMM. Consequently, this study aimed to determine whether similar disparities in CVD risk exist for SMM who participated in NHANES, and if these differences are consistent across subgroups of SMM. On the basis of our previous work with SMW, we hypothesised that SMM would be at greater risk for CVD than heterosexual men.
Methods
The present study utilised publicly available data from NHANES, pooled from 2001 to 2010. NHANES is a nationally representative cross-sectional survey of US adults and children that assesses health and nutritional status using in-home interviews and physical examinations. In 2001, sexual orientation questions were added to the sexual behaviour interview, and these data were included in the public use data set for participants aged 20–59 from 2001 to 2006 and for participants aged 20–69 in 2007–2010. The survey response rates for the 10 years included in our sample ranged from 75% to 80%.26 More detailed information regarding the NHANES design and sampling strategies are described elsewhere.27 A total of 7571 men aged 20–69 completed the sexual behaviour survey from 2001 to 2010. For the present study, 155 men (2.1%) were excluded because they either refused to answer the sexual orientation or same-sex behaviour question or they provided a response of ‘Something Else’, ‘Not Sure’ or ‘Don't Know’ to the sexual orientation question. Another 314 men (4.2%) were excluded due to the pre-existing CVD (self-reported congestive heart failure, coronary heart disease, angina, heart attack or stroke), resulting in a final analytic sample of 7078 men.
Measures
Sexual minority status
NHANES contains the self-administered measures of both sexual orientation and sexual behaviour. Sexual orientation was assessed using the question: “Do you think of yourself as … Heterosexual or straight (attracted to women); homosexual or gay (attracted to men); bisexual (attracted to men and women); something else; or you’re not sure?” Sexual behaviour was assessed by asking participants to provide the total number of their past-year and lifetime same-sex and opposite-sex sexual partners. For the present study, SMM were defined as men who either self-identified as gay or bisexual, or who reported having had at least one lifetime same-sex sexual partner (homosexually experienced heterosexuals).
Cardiovascular risk
CVD risk was assessed using the Framingham General CVD risk score. The Framingham score is a sex-specific multivariable risk factor algorithm that utilises several established CVD risk factors to predict both the absolute 10-year likelihood of developing a first CVD event as well as an estimate of vascular age.28 ‘Vascular age’ is defined as the chronological age of a person with the same predicted CVD risk, given that he or she has risk factor levels in the normal range. As such, vascular and chronological age will be equal when a person has a normal risk factor profile, and the ratio of his or her vascular and chronological age will be equal to one. For example, the vascular age of a 35-year-old male smoker with untreated systolic blood pressure between 140 and 149 mm Hg is 42, and the ratio of his vascular to chronological age is 1.2. This ratio indicates that his vascular system is 20% older than would be expected, given his chronological age. Since age is a primary driver of CVD risk, the use of vascular age is a more appropriate measure of CVD risk in younger to middle-aged populations, as it is rare for persons in this age group to exhibit increased absolute risk whether or not they have multiple CVD risk factors.29
The risk factors included in the Framingham algorithm were age, sex, high-density lipoprotein (HDL) and total cholesterol level, systolic blood pressure, antihypertensive medication use, diabetes and current smoking status. A man was considered to have a normal risk factor profile if he did not currently smoke, was non-diabetic, had a total cholesterol level less than 160 mg/dl, had an HDL cholesterol level of 45 mg/dl or greater, and had an untreated systolic blood pressure of 129 mm Hg or less. Men were classified as current smokers if they answered either ‘Some days’ or ‘Every day’ to the question ‘Do you now smoke cigarettes’, and men were considered diabetic if they answered ‘Yes’ to the question ‘Have you ever been told by a doctor or health professional that you have diabetes or sugar diabetes?’.
Additional covariates
In addition to the variables that comprise the Framingham risk score, participants’ family history of early CVD, body mass index (BMI), education, annual household income, race/ethnicity, history of hard drug use and current alcohol use were assessed. Participants were considered to have a family history of premature CVD if they indicated having a first-degree relative who had a heart attack or angina before age 50. Participants’ current BMI (defined as weight in kg divided by height in m2) was recoded into three categories (normal/underweight, overweight, obese) based on NIH guidelines.30 Participants were considered to have a history of hard drug use if they indicated ever using illicit drugs (eg, cocaine, crack, heroin, methamphetamine) other than marijuana. Participants were classified into three categories of current alcohol users (risky drinkers, social drinkers, infrequent drinkers) based on the National Institute on Alcohol Abuse and Alcoholism criteria for risky drinking, with infrequent drinkers being those participants who indicated having fewer than 12 drinks in their lifetime.31 Lifelong abstention was not directly assessed in the survey.
Statistical analyses
Data were analysed using SAS V.9.3 software (SAS Institute Inc, Cary, North Carolina, USA), incorporating both the design information and weights as specified in the NHANES Analytic and Reporting Guidelines to account for the complex survey design.32 ,33 Demographic characteristics and individual CVD risk factors were compared by sexual minority status using the χ2 test for proportions and one-way analysis of variance for continuous variables. Linear regression was used to examine whether the ratio of vascular to chronological age varied by sexual minority status and multivariate linear regression was used to adjust for differences on demographic characteristics and other covariates that were not incorporated into the Framingham calculation. Variables were considered candidates for adjustment if they exhibited a statistically significant difference by sexual minority status at the α=0.20 level, and a 10% change in the β estimate for sexual minority status was used as the final criterion for determining which variables to retain as covariates in the final model.34 We conducted a sensitivity analysis to determine if restricting the definition of SMM to only those who identified as ‘Gay’ or ‘Bisexual’ affected the study's results, and we ran an analysis using a four-category variable for sexual minority status (ie, heterosexual, gay, bisexual, homosexually experienced heterosexuals) to determine if CVD risk differed among subgroups of SMM.
Results
Table 1 provides a summary of demographic characteristics and individual CVD risk factors by sexual minority status. Of the 7078 men in the sample, 5.2% were classified as SMM and 94.8% were classified as heterosexual. Compared with heterosexual men, SMM were more likely to be chronologically older (p=0.007) and to have higher education levels (p<0.001). However, a greater percentage of SMM fell into the lowest income bracket compared with heterosexuals (21.3% vs 16.5%, p=0.02). In addition, SMM were more likely to have a history of hard drug use (39.4% vs 25.8%) than their heterosexual counterparts, but were less likely to be current risky drinkers (25.9% vs 39.1%) and had, on average, lower systolic blood pressure (119.1 vs 122.0). There were no statistically significant differences by sexual minority status in regard to race/ethnicity, smoking status, diabetes status, use of antihypertensive medication, family history of CVD, BMI categorisation, total cholesterol or HDL cholesterol.
Demographic characteristics and cardiovascular risk factors of male National Health and Nutrition Examination Survey participants from 2001 to 2010 by sexual minority status
The mean ratio of vascular to chronological age for the entire sample was 1.201, indicating that, on average, participants’ vascular age was 20.1% greater than their chronological age. To aid in interpretation, all subsequent ratios will be expressed as percentages. Table 2 provides a summary of the differences in CVD risk by sexual minority status. Regression analysis revealed a statistically significant difference in the ratio of vascular to chronological age by sexual minority status. SMM were, on average, 16.3% (95% CI 11.8% to 20.8%) older in vascular terms than their chronological age, which was 4% (95% CI −7.5% to −0.4%) less than their heterosexual counterparts. Five variables emerged from the bivariate analysis as potential candidates for adjustment: education, income, family history of premature CVD, history of hard drug use and alcohol use. In addition, year of data collection was also considered as a potential adjustment variable. Only education and history of hard drug use met the 10% change in β estimate for sexual minority status criteria (56.1% for education and 20.1% for history of hard drug use), and as such, they were retained as adjustment variables in all subsequent models. After adjustment for education and history of hard drug use, the effect of sexual minority status on CVD risk decreased from −4% to −2.4% (95% CI −5.9% to 0.1%) and was no longer statistically significant (p=0.16). When a sensitivity analysis was performed narrowing the definition of sexual minority status to only those who identified as ‘Gay’ or ‘Bisexual’, neither the unadjusted nor adjusted difference in the ratio of vascular to chronological age between sexual minority and heterosexual men was statistically significant (table 2).
Ratios of vascular to chronological age by sexual minority definition
A summary of demographic characteristics and individual CVD risk factors using the four-category variable for sexual minority status is presented in table 3. Among the SMM, 35.1% identified as gay, 27.9% identified as bisexual, and 37% were homosexually experienced heterosexuals. As in the original analysis, significant (p<0.01) differences were found in regard to education, income, systolic blood pressure and history of hard drug use. The revised categorisation revealed that income and education levels were the highest among gay men and the lowest among bisexual men, and that gay men were more likely to have a history of hard drug use than heterosexual men, but were less likely to have a history of hard drug use than bisexual or homosexually experienced heterosexual men. Systolic blood pressure was lower for both gay and homosexually experienced heterosexual men compared with heterosexuals, but there was no significant difference for bisexual men. In contrast to the original analysis, significant differences in diabetes status, antihypertensive medication use and total cholesterol emerged when using the revised categorisation; however, differences in alcohol use were not statistically significant.
Demographic characteristics and cardiovascular risk factors of male National Health and Nutrition Examination Survey participants from 2001 to 2010 by sexual minority status
Table 4 presents differences in overall CVD risk using the four-category variable for sexual minority status. Both gay and homosexually experienced heterosexual men had lower ratios of vascular to chronological age than their heterosexual counterparts (−9.1%; 95% CI −13.9% to −4.3% and −6.6%; 95% CI −11.5% to −1.7%, respectively). In contrast, the ratio of vascular to chronological age was significantly higher for bisexual men compared with heterosexuals (8.1%, 95% CI 1.3% to 15%), indicating that bisexual men were at increased risk of developing CVD. Adjustment for education and history of hard drug use slightly attenuated the estimate of increased risk for bisexuals, but did not appreciably change the difference estimate for homosexually experienced heterosexuals. For gay men, adjustment decreased the difference estimate by over half and rendered it statistically insignificant (p=0.07).
Ratios of vascular to chronological age by sexual minority category definition
Discussion
Our current findings did not mirror those of our previous work with women in NHANES where we found that SMW, as a whole, exhibited increased CVD risk compared with heterosexual women. Rather, we found that SMM, as a whole, exhibited no significant difference in CVD risk compared with heterosexual men after accounting for differences in education and history of hard drug use. One explanation for this finding may be that the proportion of bisexuals was greater among SMW compared with SMM, which could have accounted for the increased CVD risk among SMW as a whole. Another explanation may be that sexual minority status affects CVD risk differently for women and men.
Our results indicate that certain subgroups of SMM are at increased risk for CVD compared with heterosexual men and that this increased risk cannot be completely attributed to differences in demographic characteristics or negative health behaviours. Specifically, we found that bisexual men were at significantly increased risk for CVD and homosexually experienced heterosexual men were at significantly decreased risk for CVD compared with heterosexual men. Gay men did not exhibit significantly increased CVD risk compared with heterosexual men after accounting for differences in education and history of hard drug use. These findings indicate that it is critically important to look at differences within sexual minority categories, as combining gay, bisexual and homosexually experienced heterosexual men into one broad category may mask important differences.
Our findings were consistent with those of Conron and colleagues that bisexuals were at increased risk for CVD compared with heterosexuals, but were inconsistent with Cochran and Mays's finding that homosexually experienced heterosexual men were significantly more likely to report a diagnosis of heart disease than heterosexual men. Among our sample, homosexually experienced heterosexuals had significantly lower CVD risk than heterosexual men, which persisted even after adjustment for possible confounding factors. Potential explanations for this inconsistency include both the use of different sampling frames, (ie, state vs nation) and the use of different outcome measures (ie, self-reported CVD vs CVD risk).
Consistent with previous research, we found that SMM were more likely to have a history of hard drug use than heterosexual men; however, we did not find significantly increased rates of current risky drinking or smoking for SMM. One explanation for the lack of any difference in these variables is that the prevalence of both current risky drinking and smoking among heterosexual men in our sample is higher than would be expected based on other national estimates (39.1% vs 23.2% for binge drinking; 29.1% vs 22.3% for smoking).35 ,36 When compared with these national estimates, the prevalence of both current risky drinking and smoking is higher among all groups of SMM in our sample, which is consistent with previous research.
The strengths of this study include the use of a national population-based sample, the use of both identity and behaviour measures to operationalise sexual minority status, and the use of a validated multirisk algorithm to measure CVD risk. Limitations include a lack of sexual behaviour data for older participants, a relatively small sample of SMM, and the exclusion of men who identified their sexual orientation as ‘Something Else’, ‘Not Sure’ or ‘Don't Know’ because we could not definitively infer the meaning of their sexual orientation. As such, our sample may not be representative of all SMM. Moreover, the use of self-reported measures of substance use, sexual identity and sexual behaviour could have introduced misclassification bias due to the under-reporting of these behaviours. Finally, although we assessed and controlled for confounding factors using a well established and validated approach, there exists a potential for uncontrolled and residual confounding.
The increased CVD risk we found for bisexual men and the increased prevalence of substance use among all SMM reinforce the need for culturally competent interventions to reduce substance use and CVD risk in this population. Future work is also needed to elucidate the mechanisms by which sexual minority status confers increased CVD risk and, in particular, how these mechanisms may differ among sexual minorities and by gender. Critical in such research is the need to utilise assessments of sexual minority status that accurately capture variation within sexual minority categories and allow sexual minorities to be disaggregated into subgroups. In addition, longitudinal studies are needed to determine whether disparities in CVD risk for sexual minorities ultimately result in increased CVD for this population.
What is already known on this subject
-
Sexual minorities may be at increased risk for cardiovascular disease (CVD), in part due to the increased prevalence of tobacco, alcohol and drug use among this population. Previous research using the Framingham General CVD risk score found that sexual minority women were at increased CVD risk compared with heterosexual women. No such work currently exists for sexual minority men (SMM).
What this study adds
-
We found that CVD risk varies among subgroups of SMM, with bisexual men having increased CVD risk and homosexually experienced heterosexual men having decreased CVD risk compared with heterosexual men. Our findings illustrate the importance of examining differences within subgroups of sexual minorities as well as between sexual minorities and heterosexuals.
Acknowledgments
This research was supported in part by the National Institute on Drug Abuse (grants F31DA032220, R01DA014363) and the National Institute on Alcohol Abuse and Alcoholism (grant R01AA012640). The authors would like to acknowledge the Mentoring Program of The Center for Population Research in LGBT Health, supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant R21HD051178). The authors would also like to express their appreciation to Vanessa Xanthakis, PhD, of Boston University for her statistical guidance. Note. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
References
Footnotes
-
http://group.bmj.com/products/journals/instructions-for-authors/licence-forms
-
Contributors GWF originated the study idea, completed the analyses and wrote the initial draft of the article. DJB and KKB assisted in developing the study design and supervised the analyses. DJB, KKB, LHF and TEB assisted in the interpretation of the findings and contributed to the writing of the final draft of the article. All authors edited drafts of the article. GWF had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
-
Funding This work was supported by the National Institute on Drug Abuse grant number F31DA032220.
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.