Cultural adaptation and validation of a measure of prejudice against men who have sex with men among healthcare providers in western Kenya

Abstract

Sexual prejudice toward men who have sex with men (MSM) is a pressing concern in sub-Saharan Africa (SSA). Given the high HIV infection risk among this population, sexual prejudice perpetuated by healthcare providers, affects access to and willingness of MSM to seek HIV care services. However, data on healthcare providers’ attitudes towards MSM in SSA are limited, and there are no locally-adapted measures of sexual prejudice. We adapted a scale to measure sexual prejudice with a sample of 147 healthcare providers in western Kenya. Results from exploratory factor analysis revealed a single-factor structure. The scale demonstrated high internal consistency with Cronbach’s α=0.91. Healthcare providers who had prior interpersonal contact with MSM, had ever been trained on counselling MSM, and had higher knowledge about MSM health needs reported lower sexual prejudice scores, compared with peers who lacked these experiences (p<0.001). In contrast, healthcare providers who had experienced secondary stigma (negative judgments from peers and community) for providing care to MSM reported higher scores of sexual prejudice scale (p<0.001) compared with providers who had not experienced secondary stigma. The scale provides a contextualized tool to assess healthcare providers’ attitudes toward MSM in Kenya and countries in SSA with similar cultural norms.

Introduction

In sub-Saharan Africa, men who have sex with men (MSM) defined by the Joint United Nations Program on HIV and AIDS (UNAIDS) as “males who have sex with males, regardless of whether or not they also have sex with women or have a personal or social gay or bisexual identity” (UNAIDS, 2015) continue to experience higher rates of HIV compared to adults in the general population. In Kenya, HIV prevalence among MSM is almost three times (18.2%) that of the general population (5.9%) (National AIDS Control Council, 2016). Yet, many MSM in Kenya have limited access to relevant HIV prevention and treatment services. Consequently, they miss benefiting from ongoing advances in HIV prevention services such as pre-exposure prophylaxis (PrEP) and antiretroviral therapy (ART) treatment, impacting effectiveness of these services in stopping HIV incident cases. In 2014, UNAIDS set a goal to end AIDS epidemic by 2030 (UNAIDS, 2014), and in keeping with the UNAIDS goal, the government of Kenya set a target to reduce new HIV infections to zero by the year 2030 (National AIDS Control Council, 2014). Given the high burden of HIV among this population, targeted efforts to improve their uptake of and retention in HIV care is essential.

There is increased documentation that sexual prejudice impedes access to HIV prevention, testing, care, and treatment by MSM in SSA (Cele, Sibiya, & Sokhela, 2015; Pachankis et al., 2015; Ross et al., 2013a; Ross, Kajubi, Mandel, McFarland, & Raymond, 2013b; Ross et al., 2015). UNAIDS defines sexual prejudice as any negative attitude, belief, or action towards non-heterosexual people, relationships, and identities (UNAIDS, 2015). According to Herek (2004), sexual prejudice actions include discrimination, harassment, and acts of violence towards MSM. Laws criminalizing same sex sexual practices reinforce discrimination against MSM in these countries (Lamontagne et al., 2018). In Kenya, where same sex sexual practices remain illegal, there is evidence that the spread of HIV infection is linked to negative social attitudes towards the lesbian, gay, bisexual, and transgender (LGBT) community, especially MSM (Graham & Harper, 2017; Smith, Tapsoba, Peshu, Sanders, & Jaffe, 2009a). A further review of studies in African countries, found that LGBT populations have difficulties accessing health services as a result of heteronormative attitudes imposed by health care workers (Smith et al., 2009a; Sullivan et al., 2012) (Beyrer et al., 2012). Exposure to sexual prejudice and sexual stigma has severe implications for the well-being of MSM (Anderson, Ross, Nyoni, & McCurdy, 2015; Geibel, King’ola, Temmerman, & Luchters, 2012; Hladik et al., 2012).

Healthcare workers have an important role to play in addressing health inequities experienced by MSM (Beyrer et al., 2016; Mayer et al., 2012; E. M. van der Elst et al., 2013b). However, in Kenya, healthcare workers may experience secondary stigma in providing care to MSM, particularly if their personal cultural, moral or religious beliefs are conflicting with patients’ reported behaviours and preventative health needs. Secondary stigma, defined as the negative characteristics attributed to individuals who are in close contact with people who are stigmatized (Halter, 2008), may influence how healthcare providers care for MSM. Qualitative studies done in the coastal region of Kenya observed that healthcare providers were stigmatized by their colleagues for providing care to MSM (Micheni et al., 2017; E. M. van der Elst et al., 2015; Elise M van der Elst et al., 2013a). Findings from these studies further stress the need for culturally sensitive and competent health care for MSM.

There are no known culturally appropriate instruments that have been validated for use to assess healthcare providers’ attitudes toward MSM in sub-Saharan African settings (Fitzgerald-Husek et al., 2017; Freeland et al., 2018). In order to design interventions to improve access and uptake of HIV and other health services among MSM in SSA, it is necessary to have validated measures of sexual-based stigma and sexual prejudices that are contextualized. The purposes of this study were (a) to describe the adaptation of a scale to measure sexual prejudice against MSM among healthcare providers (Wright et al., 1999), (b) to test the adapted scale’s reliability and validity with a sample of healthcare providers in western Kenya, and (c) to determine the correlates of sexual prejudice against MSM among healthcare providers in western Kenya.

Methods

Study Design

The study was carried out in two phases (Figure 1). Phase one involved cross cultural adaptation of an existing sexual prejudice scale by Wright, Adams, & Bernat (1999), which measures cognitive, affective, and behavioural components of sexual prejudice and which has been shown to be a reliable and valid indicator of sexual prejudice against MSM in the US context. Details of the original sexual prejudice scale are provided below. The second phase involved an evaluation of the scale’s psychometric properties using cross-sectional data from a sample of Kenyan healthcare providers (n=147. Brown University institutional review board (IRB) and Moi University institutional research and ethics committee (IREC) approved all phases of the study.

Figure 1.

Measure adaptation and validation process.

Study setting

All study procedures were conducted between June and August in 2017. The study was carried out in western Kenya, and based at the Academic Model for Providing Access to Healthcare (AMPATHplus) health care facilities. AMPATHplus is a partnership between Moi Teaching and Referral Hospital (MTRH), Moi University, and a consortium of North-American Universities focused on delivering HIV care in western Kenya (Einterz et al., 2007). AMPATHplus healthcare facilities operate in 10 counties. We recruited participants from six counties: Bungoma, Kisumu, Nandi, Uasin-Gishu, West Pokot, and Trans-Nzoia. Most participants were recruited from Uasin-Gishu County primarily from MTRH, the largest urban AMPATHplus clinic representing diverse healthcare providers.

Original Sexual Prejudice Scale

The original sexual prejudice scale consists of 25 items that form three subscales and an overall scale score (Wright Jr, Adams, & Bernat, 1999). The original scale was developed to measure cognitive-affective-behavioural factors comprising people’s homophobic attitudes towards gay and lesbian individuals. Exploratory factor analysis (EFA) of the original version resulted in three subscales: a subscale that assessed negative cognitions regarding homosexuality (5 items total; sample item, “homosexuality is acceptable to me”), a subscale that assessed negative affect and avoidance of homosexual individuals (10 items total; sample item, “gay people make me nervous”), and a subscale that assessed negative affect and aggression toward homosexual individuals (10 items total; sample item, “I avoid gay individuals”). The 25 items were rated by participants a 5-point Likert scale (Strongly agree, agree, neither agree nor disagree, disagree, and strongly disagree). The instrument was originally tested in a sample of 321 college students in US. Although the scale comprised of 3 subscales, it was also shown to be reliable and valid as a single scale comprising all 25 items with an overall internal consistency (α) of 0.94. We selected this scale as it was validated and no other scales were found in the literature assessing homophobic attitudes against MSM specifically assessing attitudes regarding homosexuality among healthcare providers or other similar populations.

Phase 1: Cross-cultural adaptation process

To adapt the scale items for use with healthcare providers in Kenya, we used the systematic method for cross-cultural adaptation of an existing self-report measure described by Beaton and colleagues (Beaton, Bombardier, Guillemin, & Ferraz, 2000). This process has been used in research to develop culturally adapted measures for stigma (de Oliveira Freitas et al., 2018) and health literacy (Reisi et al., 2017). The five steps include (1) forward translation; (2) synthesis of the translations; (3) backward translation; (4) use of expert committee; (5) pre-testing the translated instrument;

Steps 1 and 2 were performed by two independent persons (one with a background in public health and familiar with the subject matter (SS), and another local person who works in healthcare (a nurse) independently reviewed the original instrument) who reviewed the original measure, identified words or phrases that were possibly unclear in the Kenyan context, and revised them into the local context. The phrases such as “gay”, “faggot”, and “queer” that were used in the original scale were changed to equivalent words in the Kenyan context while maintaining its conceptual meaning. These words were replaced with “men who have sex with men”. Both translators synthesized their translated versions into the common Kenyan version of the instrument and noted issues that warranted consideration by the expert committee. These issues were: item 17 “I have damaged property of gay persons, such as keying their cars”; item 18 “I would feel comfortable having a gay roommate”‘ item 19 “I would hit a homosexual for coming on to me”; item 25 “I have rocky relationships with people I suspect are gay”. These items were flagged for further discussion with the expert committee. In the context of the Kenyan healthcare system, the primary language is English. Therefore we excluded backward translation (Step 3) of the Beaten et al. guidelines in this study, and followed the remaining steps as shown in Figure 1. An expert committee review (Step 4) comprised local professionals working in medical practice, social science fields, and translation (n=5) reviewed the adapted instrument for its simplicity, relevance, clarity and its applicability to a Kenyan healthcare setting, and made additional recommendations for adaptations. We purposively identified expert committee members familiar with the area of the study. They were contacted and invited to participate in the cross-cultural adaptation of the tool as recommended by Beaton et.al (2000). Items that were unclear were modified and translated as needed and a preliminary version of the instrument was produced and distributed among the review members. Then, the preliminary version of the instrument was pre-tested (Step 5) with two focus groups group 1, n=8; group 2, n=6) of healthcare providers-doctors, clinical officers, counsellors, and nurses. Focus groups were recruited using flyers that were posted on the healthcare facility noticeboard. We sought written informed consent from those who were interested in the study. Focus group discussions took place in a private room within the healthcare facility. Participants in the first focus group provided feedback on their understanding and acceptability of each item, and provided suggestions on how to modify items to enhance relevance in the Kenyan setting. The second focus group confirmed the final draft. Each focus group lasted about 1.5 hours. The scale was modified as needed at each step. The final version of the instrument was used in Phase 2 to evaluate psychometric properties (Step 6). Participants who participated in Phase 1 focus group discussions were excluded from taking the Phase 2 survey.

Phase 2: Evaluation of the psychometric properties of the scale

We conducted a cross-sectional survey to examine the validity and reliability of the adapted scale. Participants were recruited from their respective clinics in HIV care facilities in western Kenya. We posted flyers describing the study and eligibility criteria on the noticeboards of the study sites. Research Assistants visited the clinics during break periods to seek participants interested to participate in the study. A convenience sample of 151 healthcare providers that provided HIV care to patients was screened for enrolment into study. Healthcare providers were included if they met the following inclusion criteria: 1) medical officer, nurse, or clinical officer , 2) previously or at that moment were providing HIV care services, 3) voluntarily indicated their intention to participate. Participants provided written informed consent prior to participating in the survey and were given Kenyan shillings 500 ($5) as a lunch allowance for participating in the study. Four potential participants declined to participate in the study. The final sample therefore consisted of 147 participants. All study procedures were approved by the Moi University College of Health Sciences and Moi Teaching and Referral Hospital Institutional Research and Ethics Committee (IREC approval number 0001551). All participants provided written, informed consent to participate.

Measures

Demographics.

Demographic information collected included age, gender, years worked in healthcare, highest level of education, profession, and religion.

Adapted measure of prejudice against men who have sex with men.

Participants completed the adapted measure developed in Phase 1.

Secondary Stigma.

Participants answered 3 questions from HIV-related stigma tool that was previously validated among healthcare providers in Kenya, China, Puerto Rico, Dominica, and Egypt (Nyblade et al., 2013). These items were: (1) “In the past 12 months have you experienced people talking badly about you because you care for patients living with HIV?”, (2) “In the past 12 months, have you been avoided by friends and family because you care for men who have sex with men?”, and (3) “In the past 12 months, have you been avoided by colleagues because of your work caring for MSM?”. Responses were ‘never’, ‘rarely’, ‘sometimes’, ‘often’, and ‘always’.

Interpersonal contact.

Personal contact with MSM was operationalized as having prior clinical exposure to MSM at the healthcare facility. To measure personal contact, participants were asked “Whether they had ever counselled or provided care to MSM”. Responses were dichotomized as “yes” or “no”.

Knowledge on MSM health needs.

Participants completed 20 items from the “Most-at-risk populations (MARPS) in Africa” manual, a training guide for healthcare professionals working with HIV priority populations in Africa (MARPS., 2011). Sample items include: “All men who have sex with men identify as being gay or homosexual,” “Substance abuse is common among MSM and may lead to increased risk-taking behaviours.” Response options were “true”, “false” or “don’t know”. Questions answered correctly were coded as 1 and incorrect responses were coded as zero. All scores were summed up to produce a knowledge scale, with a potential range from 0-20. Higher scores indicated greater knowledge on MSM health needs.

Relevant Training.

Participants completed three items from the “MARPS in Africa” manual (MARPS, 2011): (1) “Have you ever been trained to counsel MSM?” Responses were dichotomized as “yes” vs “no”; (2) “Do you think providers need to be trained to provide care to MSM?” (3) “Have you ever been trained to counsel regarding anal sex?” Responses were “never/rarely” vs “sometimes/often/always”.

Skills to address MSM health needs.

Participants answered two questions regarding their ability to provide care to sexual minority patients. “Have you ever discussed anal sexual practices with any of your clients, male or female?” and “How often do you ask your patients whether they have sex with men, women, or both?” Responses were dichotomized to “yes” vs “no” and “never/rarely” vs “sometimes/often/always” respectively.

Data Analysis

We conducted descriptive analyses, including computing of means and standard deviations for continuous variables, as well as count and proportions for categorical variables. Exploratory factor analysis (EFA) using oblique rotation (promax) was performed to examine the factor structure of the sexual prejudice scale on the overall sample. The adequacy of the data for factor analysis was investigated with Kaiser-Meyer-Olkin measure of sampling adequacy (KMO) and Bartlett’s test of sphericity (Tabachnick & Fidell, 2007). The criteria used to determine the number of “meaningful” factors to retain were the eigenvalue-one criterion (Afifi, May, & Clark, 2011), the scree test (Kachigan, 1991), parallel analysis, and the proportion of variance extracted (Thompson & Daniel, 1996). For the first method, also known as the Kaiser criterion, factors with an eigenvalue greater than 1 were retained and interpreted. For the scree test, the eigenvalues associated with each factor were plotted and a “break” was determined between the factors with relatively large eigenvalues and those with small eigenvalues. In parallel analysis, we used the eigenvalue Monte Carlo simulation technique, in which the sizes of the observed eigenvalues are compared with those obtained from randomly generated data set of the same size. The factors that appeared before the break were assumed to be meaningful and were retained for rotation; those factors appearing after the break were assumed to be unimportant and were dropped. After the initial factors were extracted, a promax rotation was performed to determine what was measured by each of the retained factors. Factor loadings of 0.4 or higher were considered meaningful loadings. Scale reliability was assessed with Cronbach’s coefficient alpha. We performed multivariable linear regression to examine correlates of sexual prejudice. All statistical analyses were performed using Stata version 14.2 (StataCorp, College Station, TX).

Results

Phase 1: Cross cultural adaptation of sexual prejudice scale

Forward translation.

Both independent translators agreed that the term ‘gay’ or ‘homosexual’ as used in the original scale were terminologies not commonly used in the Kenyan clinical setting and changed these to MSM, which is a well-accepted term. Both translators agreed that four items from the original scale were less applicable to the Kenyan context: item 17 “I have damaged property of gay persons, such as keying their cars”; item 18 “I would feel comfortable having a gay roommate”‘ item 19 “I would hit a homosexual for coming on to me”; item 25 “I have rocky relationships with people I suspect are gay”. These items were flagged for further discussion with the expert committee.

Synthesis of translations.

SS prepared the synthesised version incorporating all changes noted in step 1. The phrases: “gay”, “faggot”, and “queer” were replaced with MSM. Also, items: 17 (I have damaged property of gay persons, such as keying their cars), 18 (I would feel comfortable having a gay roommate), 19 (I would hit a homosexual for coming on to me), and 25 (I would hit a homosexual for coming on to me) were flagged for the expert committee to discuss and provide feedback.

Expert committee.

Members of the expert committee (n=5) reviewed the preliminary revised scale item and provided feedback about item semantics, phenomena, and other issues raised during the forward translation process. Panel members recommended revising 4 items which used language or phenomena that were deemed inappropriate to the local context (see Table 3 for original and revised items).

Table 3:

Item factor loading patterns and means for all items of sexual prejudice scale

Factors	Factor 1	Mean ±SD
1. MSM make me nervous	0.609	1.86 ± .12
2. Homosexuality is acceptable to me * reverse coded	0.608	2.82 ± .11
3. If I discovered a friend was a MSM, I would end the friendship	0.738	1.42 ± .14
4. I think homosexual people should not work with children	0.457	1.99 ± .11
5. I enjoy the company of MSM * reverse coded	0.578	2.86 ± .10
6. Marriage between homosexual individuals is acceptable* reverse coded	0.347	3.01 ± .11
7. I make remarks like “shoga” or “queer” to people who I suspect are MSM	0.295	0.95 ± .10
8. It does not matter to me whether my friends are MSM or heterosexual	0.612	2.07 ± .12
9. It would upset me if I learned that a close friend was homosexual	0.752	2.07 ± .12
10. Homosexuality is immoral	0.688	2.47 ± .12
11. I tease and make jokes about MSM	0.255	0.86 ± .09
12. I feel that you cannot trust a person who is homosexual	0.799	1.48 ± .12
13. I fear homosexual persons will make sexual advances towards me	0.476	1.50 ± .14
14. Organizations which promote gay rights are not necessary	0.638	1.45 ± .13
15. Homosexuality should be treated as an illness (original item: I have damaged property of gay persons, such as keying their cars)	0.586	1.92 ± .12
16. Homosexuality is un-African/is something brought by foreigners (original item: I would feel comfortable having a gay roommate)	0.561	2.30 ± .12
17. I would cause physical harm to homosexual who makes sexual advances on me (original item: I would hit a homosexual for coming on to me)	0.567	2.82 ± .11
18. Homosexual behavior should not be against the law	0.473	2.44 ± .12
19. I avoid MSM individuals	0.728	1.61 ± .11
20. It bothers me to see two homosexual people together in public	0.756	1.94 ± .12
21. I make derogatory remarks about MSM people	0.673	1.88 ± .11
22. When I meet someone, I try to find out if he is a MSM	0.179	0.93 ± .08
23. Homosexual as heterosexual men have the same rights to public/tax-funded services (original item: I have rocky relationships with people I suspect are gay)	0.296	1.11 ± .11
Eigenvalue	7.667
Percentage variance (%)	33.33

Note. Bold items( factors loadings <0.4) were dropped based on factor loadings

Pre-testing.

We convened two focus groups comprising local healthcare providers (group 1, n=8; group 2, n=6) to complete the full draft questionnaire, remark on words or sentences that were difficult to understand, and suggest words or phrases that could be used to avoid misunderstanding or confusion. Participants in the first focus group found it difficult to understand two items that used Western colloquial expressions (“MSM people deserve what they get” and “When I see an MSM I think, ‘what a waste’”). Based on discussion with the expert panel, the items were dropped from the measure. The pre-final scale consisting of 23 items was presented to the second focus group. This group reviewed all items in the measure and agreed that the adapted scale captured key indicators of prejudice toward MSM patients among Kenyan healthcare providers (Table 3).

Phase 2: Validation of psychometric properties of sexual prejudice measure

Table 1 presents the characteristics of the Phase 2 sample (n=147) that completed the full survey. More than half were female (55%), the majority had a high school diploma, and ages ranged from 21 to 58 years (M=33.5, SD=7.5). Almost all participants identified as Christian (97%). Over one-third (37%) were nurses, 42% were clinical officers, 11% were clinical counselors, and 9% were medical doctors. Mean length of experience working in the healthcare profession was 8.4 years (SD=7.1), with 41.2% reporting that they had ever provided healthcare to MSM. Regarding knowledge on MSM health needs, 44% reported having low knowledge of MSM health needs, with more than half reporting that they were comfortable to provide care to MSM.

Table 1:

Participant characteristics, Phase 2 (N=147)

Characteristics	N (%), Mean ±SD (Range)
Age (years)	33.5 ±7.5 (21-58)
Female	81 (54.7)
Healthcare provider type
Clinical Officer/Medical Officer	72 (49.0)
Nursing officer/Clinical counselor	75 (51.0)
Education
Diploma	92 (63)
University	54 (37)
Years worked in healthcare	8.4 ± 7.1 (1-32)
Religion affiliation
Christian	142 (96.6)
Muslim	5 (3.4)
Experienced people talking badly because you care for MSM
Never/rarely	101 (69.2)
Sometimes/often/always	45 (30.8)
Been avoided by friends and family because you care for MSM
Never/rarely	138 (94.5)
Sometimes/often/always	8 (5.5)
Been avoided by colleagues because of your work caring for MSM
Never/rarely	128 (88.3)
Often/sometimes	17 (11.7)
Ever provided care to MSM
Yes	61 (41.2)
Ever been trained to counsel MSM
Yes	41 (27.7
Ever been trained to counsel regarding anal sex
Yes	38 (25.9)
Skills to serve MSM health needs
How often do you ask your patients whether they have sex with men, women or both?
Never/rarely	99 (67.3)
Sometimes/often/always	48 (32.7)
Have you ever discussed anal sexual practices with any of your clients, male or female
Yes	72 (49.0)

Factor analysis

The dataset was suitable for exploratory factor analysis with KMO of 0.85 and statistical significance for Bartlett’s test of sphericity (p<0.001). Principal axis factoring via the promax rotation method was performed to obtain the best fitting structure and the correct number of factors. Using the four criteria for factor extraction, a one-factor solution was retained for exploratory factor analyses. Using the eigenvalue-one criterion, the first three factors were retained as their eigen values were >1 (Table 2); however, with the scree test, the eigenvalues associated with the factors were plotted (Figure 2) and there was a relatively large break between factors 1 and 2. These results suggested that large variance was explained by only one factor; these results were supported by parallel analysis which yielded a plot that indicated clear presence of a single-factor solution (Figure 3). The last criterion in deciding the number of factors was proportion of variance in the data set. Table 2 lists the variance extracted by the three factors. The cumulative variance explained by these three factors was 48.8%, with factor 1 explaining a larger proportion (34.3%) of variance than factors 2 and 3 (7.7% and 6.8%, respectively). Thus, this analysis supported a single-factor structure as the most appropriate fit of the data.

Table 2:

Principal Component Analysis for sexual prejudice scale- Eigenvalues and Variance explained by factors as the original scale

Factors (23 items)	Eigen value	Percentiles of Variance	Cumulative Percentiles
Factor 1	7.89	34.30	34.3
Factor 2	1.77	7.71	42.02
Factor 3	1.56	6.80	48.81

Note. Original scale had 3 subscales.

Figure 2.

A scree plot of eigenvalues after exploratory factor analysis

Figure 3:

Parallel analysis of eigenvalues Monte Carlo simulation technique.

Table 3 presents each of the items representing the single-factor structure and their corresponding loadings. We retained items with loadings > 0.4. Five items had loadings <0.4 and were not included in further analysis (see Table 3). The final scale thus included 18-items mapping onto one dimension which reflected sexual prejudice towards MSM.

Sexual Prejudice Scale

The internal consistency of the final scale (Cronbach’s alpha) was 0.91. The mean score on the sexual prejudice scale was 35.22 (SD=15.9).

Correlates of sexual prejudice

We examined characteristics associated with the validated sexual prejudice scale among healthcare providers (Table 4). Social demographic variables (age, sex, work experience, education, and religion) and variables that were associated with sexual prejudice against MSM in bivariate analyses at the P < 0.05 level were included in multivariable analysis. In adjusted analysis, participants who reported ever being avoided by colleagues because of providing care for MSM (secondary stigma) reported higher sexual prejudice scale scores (β=8.22; 95% CI=0.55, 15.89; p<0.05). In contrast, lower sexual prejudice scale scores were reported among healthcare providers who had prior interpersonal contact with MSM (β=−.7.28; 95% CI=−12.67, −1.89; p<0.01), had higher knowledge scores on MSM health needs (β=−1.40; 95% CI=−2.31, −0.48; p<0.001), had ever been trained to provide health care counselling to MSM patients (β=−6.18; 95% CI=−12.01 −0.35; p<0.01), and endorsed the need for further training on MSM health (β=−9.60; 95% CI=−18.27, −0.92; p<0.05).

Table 4:

Bivariable and multivariable analyses of factors associated with sexual prejudice

Characteristics	Bivariable, B (95% CI)	Multivariable, B (95% CI)
Age (years)
>35 years	−2.57 (−8-08, 2.97)	0.14 (−6.48, 6.76)
≤35 years (ref)
Provider type
Clinical officer/Medical officer	−0.24 (−4.95, 5.43)	−2.44 (−7.86, 2.96)
Nursing officer/HIV counsellors (ref)
Male (Ref=female)	0.23 (−4.95, 5.42)	1.41 (−3.90, 6.73)
University education (ref=diploma)	−0.65 (−6.03, 4.72)	2.82 (−2.20, 7.86)
Years worked in healthcare	−0.23 (−60, 0.14)	−0.20 (−0.65, 0.25)
Stigma
aExperienced people talking badly because you care for MSM
Sometimes/often/always	−1.80 (−7.40, 3.79)
Never/rarely (ref)
aBeen avoided by friends and family because you care for MSM
Sometimes/often/always	−3.15 (−14.50, 8.20)
Never/rarely (ref)
Been avoided by colleagues because of your work caring for MSM
Often/sometimes	8.25 (0.26, 16.22)*	8.22 (0.55, 15.89)*
Never/rarely (ref)
Ever provided care to MSM (Yes; No=ref)	−11.52 (−16.42, −6.62)***	−7.28 (−12.67, −1.89)**
aEver been trained to counsel MSM (Yes; No=ref)	−8.46 (−14.07, −2.86)***	−6.18 (−12.01, −0.35)*
Ever been trained to counsel regarding anal sex (Yes; No=ref)	−13.30 (−18.79, −7.80)***
Do you think providers need to be trained to provide care to MSM? (Yes; No=ref)	−15.38 (−0.24.90, −5.86)**	−13.83(−23.66, −4.01)**
Knowledge on MSM needs	−1.59 (−2.51, −0.67)**	−1.40 (−2.31, −0.48)**
Skills to serve MSM health needs
aHow often do you ask your patients whether they have sex with men, women or both?
Sometimes/often/always	−4.15 (−9.63, 1.32)
Never/rarely (ref)
Have you ever discussed anal sexual practices with any of your clients, male or female (Yes; No=ref))	−7.44 (−12.47, −2.41)***	−1.81 (−7.34, 3.72)

Note. CI=confidence interval,

^aNot included in the multivariable model,

^*p<0.05;

^**P<0.01;

^***P<0.001.

Discussion

This is the first known study to culturally adapt and evaluate the psychometric properties of a scale to measure sexual prejudice against MSM among healthcare providers in Kenya. The final scale consisted of 18 items. Factor analysis revealed a single-factor structure for this sexual prejudice scale. The scale demonstrated strong internal consistency. The scale demonstrated appropriate convergent and divergent validity. In adjusted multivariable analyses, the scale was positively correlated with a measure of secondary stigma, and it was negatively correlated with four measures of knowledge and sensitivity toward MSM (personal contact with MSM, knowledge about MSM health heeds, previous training on counselling MSM patients, and desire for further training on MSM health).

The single-factor structure observed in our study differs with the three dimensions supported in the original scale validation (Wright Jr et al., 1999), corresponding to negative affect, behavioral aggression, and cognitive negativism, respectively. Variation in the factor structure of the new versus original scales could be attributed to differences in the nature of the participants recruited in these studies, as well as cultural differences in the underlying nature of MSM prejudice between the two settings. Whereas the original scale was validated in a sample of U.S. university students, the current study intentionally sought to adapt and validate the scale among Kenyan healthcare workers. In this cultural context, the distinctions between affective, cognitive, and behavioural indications of MSM prejudice may be less salient compared with the U.S context where the three-factor structure received support. Indeed, the scree plot and parallel analysis in our model showed that most items were retained in one-factor, and this factor was meaningfully correlated with other indicators of MSM prejudice and MSM sensitivity.

We propose this 18-item, single-structure as a psychometrically and culturally appropriate measure of sexual prejudice against MSM among Kenyan healthcare providers. This measure fills an important gap in research. In a recent comprehensive literature review, Freeland et al. (2018) noted a lack of culturally tailored measures of sexuality-based stigma and discrimination for gay, bisexual, and MSM in resource-limited settings. This review found that only 7 out of 56 published reports included measures validated within country of study. A similar review by Fitzgerald-Husek et al. (2017) noted that there were limited cross-culturally tailored measures of stigma and discrimination. The current study addresses this gap in literature by providing a locally validated tool for use with healthcare provider in Kenya, which can also serve as a basis for assessing healthcare providers’ sexual prejudice against MSM in other countries in east and southern Africa.

Kenyan healthcare providers who reported ever interacting with MSM had lower sexual prejudice scale scores compared to those who had not interacted with MSM, suggesting prior contact or interaction with MSM can contribute to lower sexual prejudice. These findings are consistent with research by Earnshaw et al. (2016) showing that medical and dental students who had interpersonal contact with MSM in Malaysia were less prejudiced and less likely to discriminate against MSM compared to those who had not interacted with MSM (Earnshaw et al., 2016). Meta-analytic evidence on intergroup contact and sexual prejudice found a significant negative relationship between previous contact and sexual prejudice (Pettigrew & Tropp, 2006; S. J. Smith, Axelton, & Saucier, 2009b); however, studies included in these meta-analyses were primarily conducted in North America and did not focus on healthcare providers.

We also found that prior training and increased knowledge on MSM health needs were associated with lower sexual prejudice toward MSM among providers. These findings are consistent with research demonstrating that Kenyan MSM reported being more likely to seek care from providers that they perceived to be friendly, non-judgmental, and confidential (Micheni et al., 2017; Okall et al., 2014; Elise M van der Elst et al., 2013a). Training healthcare providers in providing inclusive and non-stigmatizing services to MSM can provide a foundation for achieving the HIV prevention and treatment goals in Kenya and other settings in southern Africa where anti-MSM stigma is pervasive (Scheibe, Duby, Brown, Sanders, & Bekker, 2017; E. M. van der Elst et al., 2013b). Our culturally adapted measure of sexual prejudice toward MSM can provide a key tool to assess training for healthcare providers in this context.

We found that healthcare providers who had previously experienced secondary stigma (specifically, being ridiculed by colleagues for providing care to MSM) had higher scores on the sexual prejudice measure. This finding aligns with research by Van der Elst et al. (2013), which found that healthcare providers who experienced secondary stigma would avoid providing care to MSM or spent the shortest time with MSM patients. This highlights the need to address stigma in health facilities.

There are limitations to this research. First, the sample involved a non-representative sample of targeted HIV clinics in western Kenya, thus findings might not be generalizable to healthcare providers working in non-HIV care settings. Additionally, the limited geographic region for recruitment may also limit generalizability. Second, the majority of our sample were nurses and clinical officers, and fewer were medical doctors. However, it is important to note that in Kenya, nurses and clinical officers are the primary deliverers of patient care and, thus, these are crucial perspectives for improving MSM services in the Kenyan context. Third, our study selected one of several previously validated instruments for assessing sexual prejudice toward MSM. Other measures include the Attitudes Toward Lesbians and Gay Men (ATLG) scale (Herek, 1984), the Modern Homonegativity scale (Raja & Stokes, 1998), and the Index of Homophobia scale (Ricketts & Hudson, 1980). The first two measures (ATLG and Modern Homonegativity) were developed to measure heterosexual peoples’ attitudes toward lesbians and gay men, and thus was deemed not as suitable for our study, and the Index of Homophobia scale was considered outdated.

In conclusion, the sexual prejudice scale reported here is a psychometrically valid measure that can be relevant to research and evaluation on Kenyan healthcare providers’ ability to serve MSM patients. Our study findings offer a validated tool for assisting and determining areas to intervene among health care providers in terms of prejudice, stereotyping, and discrimination. Additionally, the results inform areas for improvement, including targeted training, which will assist in the goal of decreasing sexual-stigma and increasing the quality of health care access among MSM and other stigmatized populations in Kenya and sub-Saharan Africa. In light of the growing burden of HIV among MSM in sub-Saharan Africa, further trainings and interventions will necessary to improve healthcare providers’ ability to serve this population effectively. The scale can be useful for other contexts with similar cultural values, beliefs, and norms; however, the need for additional adaptation and consideration of psychometric properties should be considered when using this scale in different geographic settings.

Appendix 1:

One structure validated final sexual prejudice scale

Items
1. MSM make me nervous
2. Homosexuality is acceptable to me
3. If I discovered a friend was a MSM, I would end the friendship
4. I think homosexual people should not work with children
5. I enjoy the company of MSM
6. It does not matter to me whether my friends are MSM or heterosexual
7. It would upset me if I learned that a close friend was homosexual
8. Homosexuality is immoral
9. I feel that you cannot trust a person who is homosexual
10. I fear homosexual persons will make sexual advances towards me
11. Organizations which promote gay rights are not necessary
12. Homosexuality should be treated as an illness
13. Homosexuality is un-African/is something brought by foreigners
14. I would cause physical harm to homosexual who makes sexual advances on me
15. Homosexual behavior should not be against the law
16. I avoid MSM individuals
17. It bothers me to see two homosexual people together in public
18. I make derogatory remarks about MSM people