Study participants
This study analyzed pooled data from the 2013, 2014, and 2015 waves of the National Health Interview Survey (NHIS), a series of cross-sectional household surveys conducted by the National Center for Health Statistics. A detailed description of the study design and procedures has been published elsewhere [15]. Briefly, the NHIS used a multistage probability design to allow for nationally representative sampling of the non-institutionalized U.S. civilian population. Trained U.S. Census Bureau interviewers conducted the surveys using computer assisted personal interviewing and collected data on a broad range of sociodemographic characteristics and health indicators. The response rate for sample adults was 81% (range: 80–82%). Study participants included in this study were non-Hispanic white, non-Hispanic black/African American, and Hispanic or Latino (henceforth, white, black/African American, and Hispanic/Latino) adults aged 18 to 85+ years (mean age: 47 ± 0.14 years). Participants were excluded if they had missing data (< 3%) on sexual orientation identity or race/ethnicity. Written informed consent was obtained from all study participants.
Measures
Sexual orientation identity
Regarding sexual orientation identity, participants were asked “Which of the following best represents how you think of yourself?” Response options included “gay” or “lesbian,” “straight, that is, not lesbian or gay,” “bisexual,” “something else,” and “I don’t know the answer.” Men were asked if they were “gay” or “straight, that is, not gay,” and women were asked if they were “lesbian or gay” or “straight, that is, not lesbian or gay.” Due to sample size constraints, participants identifying as “something else” or “I don’t know the answer” were not included in analysis. Participants identifying as gay, lesbian, or bisexual were also grouped into a single “sexual minority” category due to a limited sample size.
Health behaviors
Current smoking status and alcohol consumption were categorized as current, former, or never. Heavy drinking was defined as >2 drinks for men and >1 drink for women on 3–7 days per week in the past year. We also investigated participants who consumed ≥5 drinks on at least 2 days among men and women in 2013 and ≥5 drinks on at least 2 days among men and ≥4 drinks on at least 2 days among women for 2014. Leisure-time physical activity was classified as never/unable, low, or high. Participants reported how many hours they habitually slept per day, which was categorized as <7, 7–8, and >8 hours.
Health outcomes
Body mass index (BMI) was calculated using participants’ self-reported height and weight, with overweight classified as BMI ≥25 kg/m2 and obesity as BMI ≥30 kg/m2. Heart disease, hypertension, diabetes, cancer, or stroke was based on participants’ report of a health professional's diagnosis. Functional limitation was defined as having difficulties doing any of several specified activities because of a physical, mental, or emotional health problem other than pregnancy. At least one injury or poisoning episode serious enough to seek medical advice or treatment in the past 3 months was also measured. Participants responded to the question, “During the past 30 days, how often did you feel so sad that nothing could cheer you up?” with “none of the time,” “a little,” “some,” “most,” or “all the time.” As a measure of feeling depressed, participants responded to the question, “How often do you feel depressed?” with either the options of “daily,” “weekly,” “monthly,” “a few times a year,” or “never.”
Healthcare access and utilization indicators
Participants were asked if they had at least one place they usually went when they were sick or needed health advice. They also reported health insurance coverage or not, and if it was through Medicaid. Delayed medical care seeking, excluding dental care, because of worry about the cost during the past 12 months was measured. Those aged 64 years and under were asked if they ever received an HPV vaccine, and if they ever had an HIV test (excluding tests during blood donations). Women reported if they had a Pap smear or Pap test in the past 12 months, and women aged 30 years or older were asked if they had a mammogram in the past 12 months. Self-reported health status was categorized as excellent or very good, good, and fair or poor.
Race/ethnicity
As a potential modifier of associations between sexual orientation identity and health-related factors, participants self-identified their race/ethnicity upon being asked, “What race or races do you consider yourself to be? Please select 1 or more of these categories,” and “Do you consider yourself to be Hispanic or Latino?” Race/ethnicity was limited to White, Black, and Latino/Hispanic because other groups had an insufficient sample size.
Sociodemographic characteristics
Marital status was categorized as married (or living with a partner); divorced, separated, or widowed; and never married. Education was placed into four categories of < high school, high school (including general equivalency diploma), some college, and ≥ college-level education. Annual household income was dichotomized as <$35,000 and ≥$35,000, and poverty status was dichotomized based on the U.S. Census Bureau’s poverty threshold for total family or individual income. The 23 major occupation groups were grouped into professional/management, support services, and laborer occupations. Participants reported being US- or non-US-born, and region of residence included Northeast, Midwest, South, and West.
Statistical analysis
We estimated the prevalence of sociodemographic characteristics, health behaviors, health outcomes, and healthcare outcomes in relation to sexual orientation identity and race/ethnicity (stratified by sex/gender) using the direct standardization method for age and the 2010 U.S. Census as the standard population.
Multivariable Poisson regression with robust error variance was used to estimate prevalence ratios (PR) and 95% confidence intervals (CI) for the association between sexual orientation identity and health behaviors, health outcomes, and healthcare outcomes stratified by race/ethnicity for U.S. women and men, adjusting for potential confounders. PRs were used to compare both SM to heterosexual individuals in the same racial/ethnic group and SM to white heterosexual individuals among both women and men. An initial analysis examined gays/lesbians and bisexual individuals separately, but these data are not shown due to low sample size. Covariates selected a priori as potential confounders included age, educational attainment, annual household income, occupational class, self-reported health status, region of residence, marital/cohabiting status, and immigrant status.
Sampling weights, based on the NHIS multistage design with stratification, clustering, and oversampling of certain subpopulations (i.e. black, Hispanic, Asian, and those aged ≥65 years), were used for all estimates. The “subpop” command was used for variance estimation with Taylor series linearization in Stata, version 14 (Stata Corporation, College Station, Texas, USA). A two-sided p-value <0.05 was considered statistically significant.