Associations Between Non-discrimination and Training Policies and Physicians' Attitudes and Knowledge About Sexual and Gender Minority Patients

A Comparison of Physicians From Two Hospitals

Jennifer M. Jabson; Jason W. Mitchell; S. Benjamin Doty


BMC Public Health. 2016;16(256) 

In This Article



HEI ratings are made annually and for this study, ratings from 2013 were used to identify hospitals from which to recruit physicians. This was a cross-sectional study involving two hospitals. Hospital A is a 581-bed private, academic, Level I trauma hospital located in an urban area of East Tennessee (2013 population = 450,000). Hospital B is an 834-bed private, academic, Level I trauma hospital located in an urban area of middle Tennessee (2013 population = 658,602).[17,18] Both hospitals have received high rankings for performance in U.S. News and World Reports.[19] According to the 2014 Urban–rural Classification scheme,[20] Hospital B is located in a medium metro metropolitan area compared to Hospital A that is located in a small metro metropolitan area.

To achieve a 31-57 % response rate, 1000 physicians were recruited via random sampling to obtain a sample size between 310 and 570 physicians. This sample size corresponded to greater than or equal to 80 % power in each dependent variable. A random sample of 1000 physicians was drawn from a complete list of physicians affiliated with one of two university Hospitals located in Tennessee; 500 from Hospital A (without commendation, meeting none of the four HEI criterion for non-discrimination policies and training) and 500 from Hospital B (with commendation for non-discrimination policies and training; achievement of all four HEI criterion). The names of all physicians affiliated with either of the two hospitals were entered into an excel spreadsheet. Then the excel function Rand() was used to assign a random number to all physicians in the list. The random numbers were then sorted numerically from lowest to highest and the first 500 physicians from each hospital were recruited to participate in the study. The 1000 randomly identified physicians were recruited to participate in this study via personalized invitation package delivered by mail. In an effort to guard against low response rate each invitation package included a personalized letter printed on university letterhead, addressed to the physician by name, and described the study and the study's purpose to better understand physicians' attitudes and beliefs about SGM patients. The personalized package also included a description of the compensation (i.e., random drawing for $200), a hard-copy version of the respondent-friendly survey, and the link to an online survey option.[21] The invitation encouraged participants to complete the survey either online or in hard copy and return using the enclosed self-addressed stamped envelope. This study was approved by the University of Tennessee Knoxville Institutional Review Board (protocol # 9395B).


General Attitudes Toward Lesbian, Gay, Bisexual, and Transgender Non-patients. Attitudes toward lesbians, gay men, bisexuals, (sexual minority) and transgender (gender minority) individuals (ATLG) were assessed with four separate measures. The ATLG and its individual subscales have shown consistent reliability. Cronbach's alpha is .90 for the full scale and subscales were as follows: lesbian α = .84 and gay α = .83. In the current study, for the full scale, Cronbach's alpha was .97. Cronbach's alpha for each subscale in our data was as follows: lesbian α = .78, gay α = .76, bisexual α = .78, and transgender α = .70. The ATLG has been consistently correlated with theoretical constructs, with higher scores correlating with more negative attitudes, such as adherence to traditional beliefs, lack of contact with sexual minority individuals, and dogmatism.[22] Herek's 6-item scale for "Attitudes toward Lesbians and Gays" was used and expanded to include bisexual and transgender individuals.[22] Individual items were set to a 5-point Likert scale (5 = strongly agree). Individual items included "Sex between two men is just plain wrong", "I think male homosexuals (gays) are disgusting", and "Male homosexuality is a natural expression of sexuality in men".[22] The language for each item was modified according to each of the three sexual orientation sub-groups and gender minorities. Positive items were reverse-scored and items were summed such that higher scores indicated having more negative attitudes toward SGM individuals.

Physicians' Attitudes Toward Lesbian, Gay, Bisexual, and Transgender Patients. Physicians' attitudes about treating lesbian, gay, bisexual, (sexual minority) and transgender (gender minority) (LGBT) patients (ATLGBTP) was measured with a modified version of Harris and colleagues' 6-item Questionnaire for Health Care Professionals.[23,24] Individual statements included: "I would prefer not to provide care for LGBT patients", "I would refuse care for a LGBT patient if I were aware that they identified as LGBT," "I feel competent to provide care for LGBT patients", "LGBT patients do not have any specific health needs," "I feel I would be able to talk to a patient who identifies as LGBT in a sensitive and appropriate manner," and "I believe my medical training adequately addressed the health needs of the LGBT population".[23,24] The ATLGBTP has shown modest reliability of .54 on Cronbach's alpha. In the current study Cronbach's alpha was .50. The number of scale items may have influenced its reliability. Results from the ATLGTP should be interpreted carefully; the minor modifications may have influenced the .04 difference in reliability between ours and the Cronbach's alpha published by others.[23] Each of the 6 items was set to a 5-point Likert scale (5 = strongly agree). Positive items were reverse scored and all items were summed such that higher scores indicated more negative attitudes about treating LGBT patients.

Knowledge of Lesbian, Gay, Bisexual, and Transgender Patients. This measure includes 13 true/false items modified from Strong and colleagues' (unpublished thesis) and Harris and colleagues' questionnaire.[24] The KLGBT scale a Cronbach's alpha level of .74 among health care professionals. The scale's construct validity was determined by having different types of health professionals (e.g., psychologists, social workers) take the test. Scores on the test were significantly higher for health professionals than for people with less education. The KLGBT has correlated with theoretical constructs, with higher scores being associated with significantly less prejudice. Correlations on various measures have ranged from -.41 to -.61.[25] In this sample, KR-20 was used as the reliability estimate (appropriate for dichotomous items); KR-20 = .59. Individual items included but were not limited to: "Sex and gender have the same meaning", "most homosexuals want to be members of the opposite sex", "Lesbian, gay, bisexual, and transgender patients do not seek medical treatment as early as heterosexuals because of fear of discrimination", and "most health care providers automatically make the assumption that their patient is heterosexual if they have not specifically addressed sexual orientation". Correct responses were scored a 1 and incorrect responses were scored a 0; total responses were then summed. High scores indicated more knowledge about SGM individuals.

Gender and Sexual Minority Affirmative Practice Scale. A modified version of the Gay Affirmative Practice (GAP) Scale, renamed for this study more inclusive (Gender and Sexual Minority Affirmative Practice Scale), was used to assess physicians' attitudes and beliefs about the treatment of SGM patients and physicians' behaviors in clinical setting with patients.[26] For this study this measure included 10 items set to a 5-point Likert scale (5 = strongly agree). The GAP scale has an overall Cronbach's alpha level of .95 for the full scale. The two subscales (beliefs and behaviors) have demonstrated Cronbach's alpha of .93 and .94, respectively, demonstrating high levels of reliability. Tests of the GAP's construct validity have shown that each item loads at a minimum of .60 in the intended domain.[26] Individual items included but were not limited to the following statements: "In their practice with lesbian, gay, bisexual and transgender clients, practitioners should support the diverse make up of their families", "Practitioners should verbalize respect for the lifestyles of lesbian, gay, bisexual and transgender clients" and "Practitioners should be knowledgeable about issues unique to lesbian, gay, bisexual and transgender individuals". Five statements present in the original GAP were removed from the current study because they did not apply to physicians.

Demographic and Medical Practice Characteristics. Physicians also answered questions regarding their age, gender, sexual orientation, and marital status. Physicians' medical practice-related characteristics were documented with the following questions: "Please define your specialization" (open ended response), "How many years have you been practicing medicine?" (open-ended response), "What is your role at the hospital where you are employed?" (open-ended response), "How often do you provide direct care to patients?" (daily/weekly/less than monthly/monthly) and "Have you ever received focused training regarding lesbian, gay, bisexual and transgender patient care?" (yes/no).


Descriptive and summary statistics were calculated to describe the sample and dependent variables. To test for mean differences between each hospital, non-normally distributed dependent variables were analyzed using the Mann–Whitney U Test whereas the t-test was used for normally distributed dependent variables. Independent samples t-tests and Mann–Whitney U tests, respectively, were calculated for each independent variable to determine differences by hospitals in physicians' attitudes, beliefs, and knowledge about SGM patients. Multivariate linear regression was used to test for associations between the Gender and sexual minority affirmative practice Scale and physicians' general attitudes about non-patient SGM individuals and attitudes about SGM patients, with adjustment for demographic and medical specialization. The frequency and percent missing data for each dependent variable was calculated. Associations between missingness and physicians demographic characteristics were tested. Analyses were calculated using IBM SPSS Statistics for Mac, v22.