From Belief to Behaviour: Why Condom Use Attitudes Alone Fail to Predict Use Among Men Who Have Sex with Men in Ghana's Western and Central Regions. A Cross-Sectional Study

Study Design and Study Period

During a consecutive three-month period after the study was approved and pretested, a cross-sectional snowball approach was used to select MSM using exponential sampling in Western and Central Regions. The comparison of condom use attitudes in the two Ghanaian regions was not only to allow for a comparison of condom use attitudes, but also to understand the barriers to condom use that MSM face in the two understudied regions of the country.

Study Setting

The study focused on the Western and Central Regions of Ghana. The Western and Central Regions were selected among the other regions of Ghana for several reasons. Based on the Ghana AIDS Commission's 2019 fact sheet estimates, ⁴ the Central and Western Regions are among the regions with high HIV infections among key populations, such as MSM. Moreover, the MSM population in the two Regions is a stigmatised population vulnerable to physical violence.^34,35 Stigmatisation and physical violence lead to the concealment of sexual practices and the avoidance of safe sexual services, such as condom use.^36,37 However, due to active LGBTQ + networks and non-governmental organisations in the two Regions, ³⁸ the Western and Central Regions provide access to the MSM subpopulation, who are otherwise a hidden population, enhancing data collection feasibility and MSM recruitment for the study.

Additionally, the increase in economic activities associated with Ghana's oil discovery and tourism in the Western and Central Regions, respectively, has increased migration, nightlife, and anonymity, which correlate with urbanisation.^39–41 Urbanisation not only increases MSM sexual networking but may also influence sexual behaviour patterns in the two study regions. The influx of foreign cultures in the Western and Central Regions, coupled with the conservative local culture and religious dynamics, may shape MSM condom use attitudes. Previous studies have primarily focused on Accra and Kumasi,^5,2542–46 creating a justified need for this study in the two Regions.

Study Population

The study population comprised MSM, aged 18 years and above.

Inclusion Criteria and Exclusion Criteria

Study Variables

The outcome variable was the condom use attitude score. A total condom use attitude was computed by averaging the scores of the 19 items from the MCAS (range: 1-7), with scores ⩾5.25 classified as “favourable attitude” and <5.25 as “poor attitude”. A composite mean score of 5.25 represents the empirical midpoint of the observed score distribution or the central tendency in this sample, thereby providing a balanced criterion for distinguishing relatively positive attitudes from relatively negative or ambivalent attitudes towards condom use. This mean-based threshold is a data-driven, methodologically sound approach that allows for clear categorisation and facilitates analysis in the absence of established normative cut-offs for the scale. ⁴⁷ Using the composite mean as a cut-off is a commonly accepted approach in behavioural and psychosocial research when established clinical or normative cut-off points are unavailable for a given population. ⁴⁸ Thus, a central tendency measure, such as the mean, was used in this study as it logically indicates a reference point on the attitudinal continuum.^48–50 The MCAS had a Cronbach's alpha score of 0.635, which suggested a moderate internal consistency. This Cronbach's alpha value reflects the reliability across the entire group, irrespective of the specific language used during data collection.

Explanatory variables. The explanatory variables examined in this study encompassed a range of sociodemographic characteristics as well as sexual behaviours such as sexual involvement with women in the past, frequency of sexual involvement with women, condom use during the last sexual involvement with a man or a woman and frequency of condom use with a man or woman. Sexual roles were categorised as insertive, receptive and versatile. Religious affiliation was categorised as Catholic/Orthodox, Pentecostal/Charismatic, Islamic, African Traditional Religion, and Non-religious. These variables were selected based on theoretical relevance and prior literature indicating their potential influence on condom use attitude. ^{47 51–53}

Ethical Approval and Consent to Participate

This study was ethically approved by the Committee on Human Research Publications and Ethics at the School of Medicine and Dentistry, Kwame Nkrumah University of Science and Technology (Ref. CHRPE/AP/778/25). The study was conducted in conformity with the Declaration of Helsinki. Administrative permissions were obtained from the Western Regional Directorate of Health of the Ghana Health Service and the Central Regional Health Directorate. Participants were provided with information on the purpose and nature of the study. Each participant signed an informed consent form before data were collected. The anonymity of respondents and confidentiality were ensured by keeping personal information safe on a computer secured with a password changed every three days. The principal investigator was the only researcher with access to respondents’ personal information. The STROBE guideline guided the manuscript formatting. ⁵¹ The completed STROBE checklist documenting adherence to the guideline is provided as Supplementary File 1.

Sample Size Determination

A cross-sectional design was employed to collect data. According to the Ghana AIDS Commission, as cited by GhanaWeb,^5,54 about 54,756 MSM are living in Ghana. Using Slovin's formula for population sampling, the sample size was determined as follows = N/1 + Ne²

where n = sample size, N = population size, e = margin of error (the level of precision or sampling error) at 5% margin of error → e = 0.05 at a 95% confidence interval. Using the formula, the sample size was arithmetically determined as = 54756/ 1 + 54756 (0.05)² = 397.099. The 10% adjustment for the non-response rate =441. Thus, the estimated sample size was = 441. The sample size was overestimated to 500 for each of the two regions to allow for greater diversity within the sample and non-responses, which would improve the generalisability of the findings.

Data Collection Tool

A modified version of the validated UCLA-MCAS, originally developed by Helweg-Larsen and Collins, ¹² was used as the data collection tool. Instead of the original 25-item UCLA-MACS, a 19-item instrument measuring five distinct factors was used due to their relevance to MSM relationships. However, the six items that were excluded were not relevant to the study population. The items were as follows: (a) 5-items measuring reliability and effectiveness of condoms, (b) 2-items measuring sexual pleasure associated with condom use, (c) 5-items measuring sexual health-related stigma (Identity Stigma), (d) 2-items measuring embarrassment about negotiation and use of condoms, and (e) 5-items measuring embarrassment about the purchase of condoms. The objective of the UCLA-MCAS was to assist researchers in evaluating condom use attitudes from a multidimensional perspective. However, this study may be the first of its kind to use the UCLA-MCAS to study condom use attitudes among MSM in a conservative context like Ghana. The study questionnaire consisted of three sections. The first section consisted of items that explored participants’ demographic and background characteristics, including age, sex, role, religion, sexual practices, and condom use during the last sexual experience. The second section explored participants’ knowledge of their HIV status and sexual practices. The third section explored participants’ condom use attitudes from the 19-item instrument of the UCLA-MCAS. Participants were asked to respond using a seven-point Likert scale that ranged from strongly disagree to strongly agree, where 1 = strongly disagree, 2 = slightly disagree, 3 = disagree, 4 = neutral, 5 = agree, 6 = slightly agree and 7 = strongly agree. Respondents were asked to choose the level of agreement with each statement that most represents their opinion. The original tool was written in English and translated into Fante and Twi, which are the local languages spoken in the Western and Central Regions by the research team and supported by a language expert with experience in the health sciences, who also carried out an independent second translation for comparison and transcribed back to English to check for consistency and correctness of the translation. This was done as an attempt to produce an instrument of the highest quality. To assess for potential mistranslations and language misunderstandings and subsequently monitor the process, the research team monitored the process and deliberated over the question items and data during regularly scheduled meeting sessions. Reliability analysis was conducted by computing Cronbach's α and composite reliability. Scales with Cronbach's α and composite reliability of 0.8 and 0.70, respectively, were considered to have met the minimum threshold. ⁵⁵ The modified 19-item questionnaire applicable to MSM, instead of the original 25-item questionnaire, was administered in the selected regions. After the validation process, the questionnaire was deployed in the study regions.

Data Collection Procedure

This study adopted a cross-sectional design and utilised a snowball sampling approach to recruit eligible MSM in the Central and Western Regions of Ghana from May 2025 to July 2025. The snowball sampling approach was the most appropriate sampling approach for several reasons. Firstly, the snowball sampling method allowed the research team to connect with participants through existing contacts, facilitating access to participants with a more comprehensive understanding of their experiences.^56–59 Secondly, the snowballing method of sampling has been extensively used and proven to be effective for studying marginalised groups, such as MSM, who may not be readily identifiable through traditional sampling methods, ⁵⁶ especially in Ghana, where the country's laws criminalise LGBTQ + activities. ³⁵ Thirdly, due to the sensitive nature of the study topic, coupled with the isolation, social stigma, and physical violence experienced by the MSM community in Ghana, ⁵ snowball sampling reduces respondents’ being wary about external researchers, and it leverages trust networks for data collection. Lastly, evidence shows that not only is the MSM population in Ghana geographically dispersed, but they are also a minority group relative to the general population,^4,5 thus, snowball sampling is the most appropriate technique to reach such a hard-to-reach population with no sampling frame.

Eliciting participants from the community instead of MSM-friendly clinics is based on two methodological principles. Firstly, the community approach may minimise the social desirability bias that might occur in a healthcare environment as a result of perceived judgment from healthcare providers. In the presence of providers, participants may likely provide answers in a way they deem to be more socially acceptable than their “true” answers to project a favourable image of themselves to avoid judgment. The outcome of such bias may be either overreporting of socially desirable behaviours or underreporting of socially undesirable behaviours or attitudes. Secondly, community data collection may empower participants to engage in the research within a familiar and secure context, which may enhance trust and disclosure. The research team believed this sampling approach offered the best results for achieving the study objectives.

The research team employed different methods of advertising the study to MSM who were socially active and well-connected within the MSM communities in the study regions. The initial stage of the participants’ recruitment process commenced with a purposive pre-snowballing outreach in public MSM frequented spaces (hotspots) such as pubs, nightclubs, and cafes. A trained field officer distributed the study leaflets developed by the research team to potential participants, informing eligible MSM about the purpose of the study, eligibility criteria, and inviting voluntary participation.

To ensure adequate diversity and also provide access to the MSM communities in the study regions, initially five individuals who were sociodemographically diverse were purposively selected in each region. The purposive selection of five diverse individuals for a pre-snowballing sample could ensure that the overall sample is sociodemographically diverse in relation to factors such as age, location, and socioeconomic backgrounds.^60,61 Criteria for the purposeful selection of initial individuals included connection within the MSM communities, being trusted and respected by their peers, being willing and able to refer others, and the ability to reach a diverse subgroup of MSM who were different in age, socioeconomic status, geographical location, and extent of openness. Additionally, the initial five individuals who were denoted as seeds were recruited from leaders of MSM support groups, MSM social media influencers/event organisers, stylists, and fashion designers. Seeds refer to the initial subjects or participants selected to start the sampling process. ⁶¹ Seeds are crucial in snowball sampling as they help initiate the recruitment of additional participants through their connections within the network. ⁶¹ However, to minimise the potential for the seeds to influence the other participants, the seeds were only used to identify potential participants and provide them with the study information, including the initial screening form, but did not confirm the recruitment of participants. A field officer then vetted potential participants for diversity and other eligibility characteristics. Participants whose sociodemographic information represented adequate diversity were invited by the field officer or principal investigator to participate in the study. This process not only reduced the undue influences of participants by the seeds but also maintained confidentiality and reduced selection bias. ⁶² Additionally, the informed consent process clarified that participation did not require recruiting others, and researchers were not offering rewards for referrals.

The research team scheduled data collection appointments at a time that was convenient for the seed respondent so that the questionnaire completion would not disrupt their daily routines. Participants were allowed to decide on safe and private venues that would guarantee their comfort and confidentiality for questionnaire administration. The questionnaires were primarily self-administered. However, in cases where participants were unable to read or write, a member of the research team conducted interviews and recorded the responses on behalf of the participants. About 30 minutes were allotted for each data collection session. All the distributed questionnaires were returned immediately after completion. The field officer of the research team assessed each submitted questionnaire for completeness. Data collection commenced simultaneously in the Central and Western Regions, spanning approximately 16 weeks. In the Central Region, the principal investigator conducted the recruitment and data collection, while in the Western Region, that responsibility was delegated to a trained field officer. A hundred per cent response rate was attained.

The recruitment process proceeded in multiple waves as follows:

In the first wave, each seed making initial contact with the research team was provided five recruitment packs, one for use and four to be distributed to further participants (denoted as alters) within their networks. Each recruitment pack consisted of a fact sheet about the study, an initial screening form requesting information about participants’ sociodemographic details and condom use behaviour. The initial screening form was used to assess all potential participants for eligibility, reduce the non-response rate and facilitate the recruitment of a diverse group of participants to increase the representativeness of study outcomes.^63,64 Additionally, seeds were requested to define their sexual orientation as predominantly gay, bisexual, or heterosexual. All seeds in iteration 0 who returned the initial screening form were vetted for participation by the research team. Once eligibility was determined, they then consented to the study and were provided with the study questionnaire to complete and return immediately.

In the second wave, the alters, named by seeds constituting iteration 0, were invited for initial screening (iteration 1). The alters constituting iteration 1 were provided with five recruitment packs, one for use and four for onward invitation of alters who were named by the respondent(s) constituting iteration 1 (constitute iteration 2). Each of the four alters was screened using the initial screening questionnaire. Participants in iteration 2 who returned the initial screening forms, met the study inclusion criteria, and consented to participate were enrolled. This iteration recruitment process was repeated separately for each region until the predetermined sample size for the two study regions was attained. The stage of snowballing and the source of initial recruitment were identified using coded questionnaires. Written informed consent was provided, and participation was entirely voluntary. The principal investigator, along with a field officer with working knowledge of Twi and Fante, the two major local languages spoken in the study regions, collected the data.

Data Quality and Assurance

The questionnaire was meticulously designed and pretested to ensure data quality. Before the initiation of fieldwork, the lead investigator conducted training for the field officer, encompassing the study's objectives, checklist, sampling methodologies, and effective techniques for eliciting responses from participants. The questionnaire underwent pretesting on 10% of the total sample size in two other regions outside the designated regions for the study. The pre-testing was conducted in each of the three languages commonly spoken in the study regions, depending on which language the respondent understood and could communicate in effectively. This approach not only ensured that the questionnaire was comprehensible across the linguistically and sociodemographically diverse study population but also verified that the instrument accurately assessed linguistic ambiguities and other necessary information in line with the study objectives. Data were meticulously cleaned daily before inclusion in a spreadsheet.

Data Processing and Analysis

Descriptive analysis was performed on respondents’ demographics. A further analysis establishing an association between respondents’ sociodemographic factors and condom attitude was carried out. Data were analysed using SPSS version 27. Continuous data were summarised using mean, median, and standard deviation, and categorical data were summarised with frequencies and percentages. Reliability analysis was conducted by computing Cronbach's Alpha (α) and composite reliability. Scales with Cronbach's α and composite reliability of 0.8 and 0.70, respectively, were considered to have met the minimum threshold. In this study, the Cronbach α and composite reliability of construct items exceeded 0.7 except for the two-item pleasure factor, which had an alpha value (α)= 0.513. Nonetheless, the Spearman–Brown split-half method for the two items that measured pleasure loss with condom use indicated moderate reliability. ⁶⁵ The overall average reliability (α) score for the condom use attitude scale was = 0.635, indicating moderate internal consistency. The scale had items framed in both positive and negative sentences. To obtain a uniform direction of items where higher scores could represent favourable condom use attitudes, while lower scores could represent poor condom use attitudes, the negatively worded sentences were reverse-coded to allow for uniform direction of positive items. The reverse coding was done by transforming the original 7-point Likert scale, where 1 = strongly agree became 7 = strongly disagree in the reverse-coded version. A UCLA-MCAS self-assessment checklist consisting of the five psychometric properties was presented using mean, median, and standard deviation. The total average score for the scale was obtained, and the overall median was categorised as high and low to indicate respondents’ performance. A score above the overall median was categorised as high, and a score below or equal to the overall median was categorised as low.

The demographic and sexual behaviour variables were regressed for a final model selection. Subsequently, all the variables were included in an initial model. A backward stepwise approach was utilised to eliminate variables until the model Akaike Information Criterion (AIC) did not reduce significantly. Candidate models were developed, and the model with the smallest AIC was selected. Variables with p-values of <.05 were considered statistically significant. In the third phase of the analysis, a binary logistic and multivariate analysis was carried out to explore the relationship and the strength of the variables associated with women's intention to undergo screening. Variables that were significant at p < .05 in the bivariate and multivariate analyses were regressed on intention to screen.

The chi-square test, Mann–Whitney U test, and Kruskal–Wallis test were used to determine the relationship between the outcome variable and the independent variables. Normality tests (Shapiro–Wilk test and graphical representations) indicated that the data were not normally distributed, hence non-parametric tests Mann–Whitney U test and the Kruskal–Wallis test, were used to compute the associations. Inferential analysis was done using binary logistic regression to explore the relationship and the strength of the variables associated with condom use attitudes. Significant variables that were relevant to the study and independent variables that were significant at a p-value less than .05 in the bivariate analysis were included in a multivariable logistic regression model. In the regression analysis, all independent variables were included in the model. However, based on the outcome of Hosmer and Lemeshow goodness-of-fit test, variables that did not fit the data well, with p < .05, were excluded from the final regression outcome. In other words, all independent variables with p < .05 were excluded from the final regression model to obtain a more stable and explainable model. Variables with p > .05 (0.226), indicating the data fit well (no evidence of poor fit) were retained in the final regression model outcome as covariates.

Demographic Information of Participants, Sexual History, and Knowledge of HIV Status

Table 1 shows participants’ demographic information and sexual behaviour. The age of participants ranged from 18 to 49 years, with most of the participants (n = 646; 64.6%) aged 25–49 years. Most of the participants (n = 415; 41.5%) identified as insertive partners as their sex role. More than half of the participants (n = 851; 85.1%) have been sexually involved with women in the past. About 90.1% of participants used a condom during the last sexual encounter with either a man or a woman. Awareness of participants’ HIV status was almost universal (94.7%), with 99.2% of participants testing for HIV within the last 3 months prior to commencement of the study.

OPEN IN VIEWER

Table 1.

Demographics information and sexual history, and knowledge of HIV status (N = 1000).

Variables	Frequency (n)	Percentages (%)
Age group
18–19	84	8.4
20–24	270	27.0
25–49	646	64.6
Sexual roles
Versatile	277	27.7
Receptive	308	30.8
Insertive partner	415	41.5
Religion
Non-religious	246	24.6
Islam	185	18.5
Catholic/Orthodox	174	17.4
Pentecostal	347	34.7
African traditional religion	48	4.8
Previous sexual encounter with a woman
No	149	14.9
Yes	851	85.1
Last sexual encounter with a woman within the last 12 months
1–3 months	228	22.8
4–6 months	210	21.0
7–9 months	271	27.1
10–12 months	119	11.9
Over 12 months	23	2.3
Frequency of sexual involvement with women
Never, as I prefer people of my sex only	149	14.9
Always	108	10.8
Usually	94	9.4
Sometimes	569	56.9
Rarely	80	8.0
Condom use during the previous sexual encounter with a man or woman
No	99	9.9
Yes	901	90.1
Reason for no condom use
Not available	34	34.3
Not affordable	2	2.0
Partner refusal	40	40.4
Pleasure	20	20.2
Rape as my consent was not obtained	2	2.0
Others	1	1.0
Frequency of condom use with a man or a woman
Always	494	49.4
Usually	184	18.4
Sometimes	249	24.9
Rarely	42	4.2
Never as I prefer unprotected sex	31	3.1
Frequency of condom use in the future
Always	610	61.0
Usually	195	19.5
Sometimes	155	15.5
Rarely	40	4.0
Knowledge of HIV status
No	53	5.3
Yes	947	94.7
HIV test screening within the last 3 months
No	8	0.8
Yes	939	99.2
Reason for not screening for HIV
I am fearful of an HIV positive test result, as I associate a positive result with death, loss, or shame	16	15.2
I do not know where I can go for an HIV test	13	12.4
I do not consider myself at risk, and I often see no reason to test	15	14.3
The thought of testing regardless of the result can be a source of stress and uncertainty	5	4.8
I have concerns about how others will treat me if I test Positive, which deters me from testing	31	29.5
I fear my status might be disclosed to family, community, or workplace	12	11.4
I believe that HIV is a result of divine will, making testing unnecessary	12	11.4
Others	1	1.0

Percentages are based on the total sample size (N = 1000). Minor variations in total percentages are due to rounding. Data represent participants’ self-reported demographic characteristics, sexual history, and HIV-related knowledge. Condom use refers to participants’ most recent sexual encounter with either a male or female partner.

Descriptive Statistics for Items of the Condom Attitudes Score (UCLA-MCAS), Where Higher Scores Indicate More Favourable Condom Use Attitudes

Table 2 shows the minimum score, maximum score, median, mean, and standard deviation of items associated with the UCLA-MCAS self-assessment checklist. A high score indicates favourable condom use attitudes, and a low score represents poor condom use attitudes.

OPEN IN VIEWER

Table 2.

Means, standard deviations, and condom attitude scores.

Variables	Minimum	Maximum	Median	Mean	Std. deviation	Cronbach's alpha
Reliability and effectiveness	1.50	7.00	6.25	5.65	1.34	0.742
Condoms are an effective method of HIV control	1.00	7.00	7.00	5.78	1.79
The condom is a highly satisfactory form of protection against STIs	1.00	7.00	7.00	5.67	2.00
I think condoms are an excellent means of protection	1.00	7.00	7.00	5.66	1.83
Condoms are unreliableÞ	1.00	7.00	6.00	5.49	1.47
Pleasure	1.00	7.00	5.00	5.06	1.33	0.513
Condoms ruin the sex actÞ	1.00	7.00	5.00	5.16	1.58
The use of a condom is an interruption of foreplayÞ	1.00	7.00	5.00	4.97	1.65
Sexual health related stigma (identity stigma)	2.00	7.00	5.40	5.29	1.07	0.743
Men who suggest using a condom are not intelligentÞ	1.00	7.00	6.00	5.76	1.34
If a couple is about to have sex and the man suggests using a condom, it is less likely that they will have sexÞ	1.00	7.00	6.00	5.33	1.59
MSM partners think men who use condoms are inconsiderateÞ	1.00	7.00	6.00	5.29	1.45
MSM think partners who suggest using a condom do not trust their partnerÞ	1.00	7.00	5.00	4.98	1.58
MSM who suggest condom use are a little bit inconsiderateÞ	1.00	7.00	6.00	5.10	1.63
Embarrassment about negotiation and use	1.00	7.00	3.50	3.59	1.68	0.740
It is easy to suggest to my partner that we use a condom	1.00	7.00	3.00	3.48	1.82
I'm comfortable talking about condoms with my partner	1.00	7.00	4.00	3.70	1.94
Embarrassment about purchase	1.25	7.00	6.00	5.71	1.15	0.736
It is very embarrassing to buy condomsÞ	1.00	7.00	7.00	6.07	1.29
I often feel awful having to get condoms when I need themÞ.	1.00	7.00	6.00	5.71	1.51
It would be embarrassing to be seen buying condoms in a storeÞ	1.00	7.00	6.00	5.57	1.58
I always feel really uncomfortable when I buy condomsÞ	1.00	7.00	6.00	5.51	1.74
Total average condom use attitude	2.76	6.65	5.24	5.25	0.63	0.635

Scores were derived from the UCLA Multidimensional Condom Attitudes Scale (MCAS) using a 7-point Likert response format (1 = strongly disagree to 7 = strongly agree). Positive statements were scored directly, while negatively worded items (indicated by ^Þ) were reverse-coded prior to analysis so that higher scores reflected more favourable condom use attitudes. For factual items, correct responses indicated low risk and incorrect responses indicated high risk. Cronbach's alpha values represent the internal consistency reliability of each subscale.

Association Between Condom Use Attitude, Sexual History, Demographic Information, and Knowledge of HIV Status

Table 3 shows the association between condom use attitude and demographic information, as well as respondents’ sexual behaviour. Different levels of association were observed between demographics and the condom use attitude of respondents. A high mean score indicates favourable performance, and a low mean score represents poor performance. Significant differences were observed when the p-value was less than .05, indicating strong evidence of a statistically significant association.

OPEN IN VIEWER

Table 3.

Association between condom use attitude and demographics.

	Reliability and effectiveness		Pleasure		Sexual health-related stigma		Embarrassment about condom negotiation and use		Embarrassment about condom purchase
Variables	Mean ± SD	p-value	Mean ± SD	p-value	Mean ± SD	p-value	Mean ± SD	p-value	Mean ± SD	p-value
Age		.011*		<.001*		<.001*		<.001*		<.001*
18–19	5.90 ± 1.27		4.47 ± 1.12		4.72 ± 0.81		4.30 ± 1.41		5.16 ± 1.04
20–24	5.64 ± 1.47		4.60 ± 1.25		4.85 ± 1.00		4.04 ± 1.58		5.25 ± 1.07
25–49	5.62 ± 1.28		5.34 ± 1.31		5.56 ± 1.03		3.31 ± 1.68		5.98 ± 1.11
Subcategories of MSM		.465		<.001*		<.001*		<.001*		<.001*
Versatile	5.65 ± 1.37		4.84 ± 1.43		5.12 ± 1.07		3.69 ± 1.62		5.49 ± 1.07
Receptive partner (bottom)	5.55 ± 1.41		5.02 ± 1.27		5.24 ± 1.00		3.92 ± 1.66		5.63 ± 1.17
Insertive partner (top)	5.72 ± 1.26		5.25 ± 1.27		5.45 ± 1.10		3.28 ± 1.68		5.92 ± 1.14
Region*		<.001*		<.001*		<.001*		<.001*		<.001*
Central	5.96 ± 1.23		4.34 ± 1.16		4.66 ± 0.87		4.34 ± 1.35		5.11 ± 0.92
Western	5.34 ± 1.37		5.78 ± 1.07		5.93 ± 0.84		2.84 ± 1.64		6.32 ± 1.04
Religion		<.001*		<.001*		<.001*		<.001*		<.001*
Non-religious	5.28 ± 1.52		5.43 ± 1.30		5.68 ± 0.97		3.39 ± 1.76		5.92 ± 0.97
Islam	5.93 ± 1.22		5.03 ± 1.47		5.16 ± 1.08		3.61 ± 1.62		5.76 ± 1.10
Catholic/Orthodox	5.83 ± 1.30		4.53 ± 1.26		4.84 ± 1.01		4.28 ± 1.48		5.34 ± 1.14
Pentecostal	5.64 ± 1.21		5.18 ± 1.22		5.40 ± 1.04		3.32 ± 1.67		5.82 ± 1.26
Traditional	5.82 ± 1.40		4.41 ± 1.06		4.66 ± 1.03		3.99 ± 1.47		5.09 ± 0.83
Previous sexual encounter with women*	0.321		0.584		0.706		0.870		0.091
No	5.54 ± 1.43		5.12 ± 1.20		5.23 ± 1.21		3.62 ± 1.70		5.50 ± 1.35
Yes	5.67 ± 1.32		5.05 ± 1.35		5.30 ± 1.04		3.59 ± 1.68		5.75 ± 1.11
Frequency of sexual encounter with a woman	0.164		0.021*		0.001*		0.073		0.012*
Never, as I prefer people of my sex only	5.91 ± 1.14		5.03 ± 1.40		5.36 ± 1.07		3.85 ± 1.79		5.85 ± 1.09
Always	5.52 ± 1.45		4.68 ± 1.39		4.99 ± 1.00		3.83 ± 1.55		5.59 ± 1.02
Usually	5.59 ± 1.42		5.01 ± 1.28		5.21 ± 1.07		3.43 ± 1.69		5.54 ± 1.20
Sometimes	5.60 ± 1.35		5.17 ± 1.30		5.38 ± 1.07		3.50 ± 1.67		5.77 ± 1.16
Rarely	5.76 ± 1.32		4.94 ± 1.28		5.08 ± 1.09		3.63 ± 1.59		5.45 ± 1.19
Condom use during the last sexual encounter with a man or a woman*	0.044*		0.078		<0.001*		0.006*
No	5.89 ± 1.25		4.83 ± 1.34		4.87 ± 1.24		4.04 ± 1.72		5.28 ± 1.31	<0.001*
Yes	5.62 ± 1.35		5.09 ± 1.32		5.34 ± 1.04		3.54 ± 1.67		5.76 ± 1.12
Frequency of condom use with a man or a woman	0.144		0.002*		<0.001*		<0.001*		0.001*
Always	5.58 ± 1.38		5.18 ± 1.35		5.47 ± 1.01		3.30 ± 1.62		5.78 ± 1.04
Usually	5.63 ± 1.37		4.87 ± 1.41		5.14 ± 1.06		3.87 ± 1.75		5.65 ± 1.21
Sometimes	5.80 ± 1.22		5.06 ± 1.23		5.22 ± 1.13		3.93 ± 1.67		5.77 ± 1.21
Rarely	5.83 ± 1.32		4.43 ± 1.22		4.68 ± 1.13		3.95 ± 1.51		5.00 ± 1.17
Never, as I prefer unprotected sex	5.33 ± 1.30		5.24 ± 1.09		4.79 ± 0.88		3.34 ± 1.71		5.54 ± 1.51
Frequency of condom use in the future	0.009*		0.008*		<0.001*		<0.001*		<0.001*
Always	5.67 ± 1.34		5.17 ± 1.36		5.46 ± 0.98		3.43 ± 1.68		5.84 ± 1.07
Usually	5.80 ± 1.31		4.93 ± 1.24		5.11 ± 1.12		4.00 ± 1.74		5.69 ± 1.12
Sometimes	5.52 ± 1.25		4.90 ± 1.31		5.02 ± 1.22		3.65 ± 1.55		5.50 ± 1.33
Rarely	5.05 ± 1.67		4.78 ± 1.17		4.69 ± 0.88		3.80 ± 1.49		4.73 ± 1.16
Knowledge of HIV status*		.052		.653		.261		.542		.155
No	6.01 ± 1.18		5.16 ± 1.23		5.11 ± 1.17		3.49 ± 1.79		5.52 ± 1.10
Yes	5.63 ± 1.34		5.06 ± 1.33		5.30 ± 1.06		3.60 ± 1.67		5.72 ± 1.15

*Associations were determined using the Mann–Whitney U test at a significance level (α= 0.05). All other variables and their associations were determined using the Kruskal–Wallis test at the significance level (α= 0.05). RE = reliability and effectiveness; P = pleasure; SHS = sexual health-related stigma; ENU = embarrassment about condom negotiation and use; EAP = embarrassment about condom purchase.

Participants who anticipated using condoms “rarely” in the future had a higher perceived loss of sexual pleasure associated with condom use (4.78 ± 1.17), with a p-value <.008 compared to respondents who anticipated using condoms “sometimes” (4.90 ± 1.31), “usually” (4.93 ± 1.24), and “always” (5.17 ± 1.36) in the future.

Insertive partners had the lowest perceived sexual health-related stigma (5.45 ± 1.10) compared to receptive (5.24 ± 1.00) and versatile (5.12 ± 1.07) partners, with p-value <.001.

The Western Region had higher perceived embarrassment associated with condom negotiation and use (2.84 ± 1.64) compared to the neutral response (4.34 ± 1.35) in the Central Region.

Predictive Correlates of Favourable Condom Use Attitudes

Table 4 presents the binary logistic regression analysis, which examined the number of determinant variables that established participants’ condom use attitudes. Seven variables were significantly associated with high condom use attitudes, including region of residence, being a Pentecostal Christian, being a receptive partner, reporting sexual involvement with women “always”, “sometimes”, “rarely”, as well as anticipating “rarely” using a condom in the future.

OPEN IN VIEWER

Table 4.

Predictive correlates of favourable condom use attitude in MSM.

Variables	Unadjusted OR (95% CI)	p-Value	Adjusted OR (95% CI)	p-Value
Region
Central	Reference		Reference
Western	3.81 (2.93–4.95)	<.001	3.54 (2.57–4.89)	<.001
Religion
Non-religious	Reference		Reference
Islam	0.98 (0.67–1.43)	.912	1.40 (0.90–2.20)	.139
Catholic/Orthodox	0.73 (0.49–1.08)	.114	1.32 (0.82–2.11)	.250
Pentecostal	1.28 (0.92–1.78)	.14	1.45 (1.00–2.11)	.048
Traditional	0.34 (0.17–0.69)	.003	0.78 (0.36–1.66)	.514
Subcategories of MSM
Versatile	Reference		Reference
Receptive partner (bottom)	1.49 (1.07–2.07)	.018	1.54 (1.08–2.22)	.019
Insertive partner (top)	2.00 (1.46–2.72)	<.001	1.40 (0.98–2.01)	.064
Frequency of sexual involvement with women
Never, as I prefer people of my sex only	Reference		Reference
Always	0.93 (0.56–1.56)	.789	0.44 (0.25–0.76)	.003
Usually	0.88 (0.51–1.53)	.645	0.65 (0.37–1.14)	.135
Sometimes	2.08 (1.43–3.04)	<.001	0.67 (0.45–0.99)	.044
Rarely	0.47 (0.25–0.87)	.017	0.56 (0.31–1.01)	.055
Condom usage during last sexual involvement with a man or woman
No	Reference		Reference
Yes	1.32 (0.87–2.02)	.189	1.11 (0.68–1.82)	.681
Frequency of condom use in the future
Always	Reference		Reference
Usually	1.04 (0.75–1.43)	.817	1.35 (0.95–1.92)	.100
Sometimes	0.88 (0.62–1.25)	.473	0.87 (0.58–1.30)	.485
Rarely	0.21 (0.09–0.48)	<.001	0.35 (0.14–0.86)	.023
Knowledge of HIV status
No	Reference		Reference
Yes	0.99 (0.57–1.71)	.957	0.75 (0.40–1.39)	.355

Binary logistic regression was conducted to identify determinants of favourable condom use attitudes among men who have sex with men (MSM). Odds ratios (OR) and 95% confidence intervals (CI) are presented. Adjusted odds ratios (AOR) control for all variables included in the model. Statistical significance was set at p < .05. Reference categories represent baseline comparison groups. Hosmer–Lemeshow goodness-of-fit test indicated an adequate model fit (p > .05).

The fitness of the model was assessed using the Hosmer–Lemeshow goodness-of-fit test. Variables such as age and participants’ frequency of condom use with a man or woman were eliminated from the final model to improve model stability, accuracy, and interpretability. The inclusion of age and participants’ frequency of condom use with a man or woman can distort regression estimates, inflate standard errors, and reduce the interpretability of the model.

The final model demonstrated a good fit as indicated by the Hosmer–Lemeshow goodness-of-fit test (p-value .226).

Participants in the Western Region of residence had 3.54 times higher odds of having a favourable condom use attitude as compared to participants in the Central Region of Ghana (95% CI: 2.57–4.89). Participants who were affiliated with Pentecostal Christianity had 1.45 times higher odds of having a favourable condom use attitude compared to their non-religious counterparts (95%CI: 1.00–2.11). Receptive partners had 1.54 times higher odds of having a favourable condom use attitude compared to their versatile counterparts (95% CI: 1.08–2.22). Respondents who reported sexual involvement with women “always” had 56% lower odds of having a favourable condom use attitude compared to counterparts whose sexual preferences were men only (0.44, 95% CI: 0.25–0.76). Respondents who reported sexual involvement with women “sometimes” had 33% lower odds of having a favourable condom use attitude compared to counterparts whose sexual preferences were men only (0.67, 95% CI = 0.45–0.99). Respondents who reported sexual involvement with women “rarely” had 44% lower odds of having a favourable condom use attitude compared to counterparts whose sexual preferences were only men (0.56, 95% CI = 0.31–1.01). Participants who anticipated sexual involvement with women “rarely” in the future had 65% lower odds of having a favourable condom use attitude compared to counterparts whose sexual preferences were men only (0.35, 95% CI: 0.14–0.86).

Regional Variations in Condom Use Attitude

Table 5 shows the association between region and condom use attitude. Most of the participants from the Western Region (66.5%) had a favourable condom use attitude as compared to those from the Central Region. From the results, the Region of participants’ residence was found to be statistically significant with condom use (p-value = < .001), indicating strong evidence of association.

OPEN IN VIEWER

Table 5.

Association between regions and condom use attitudes.

Region	Poor attitude(n = 513)	Favourable attitude(n = 487)	p-Value
			<0.001
Central	337 (65.7)	163 (33.5)
Western	176 (34.3)	324 (66.5)

p-Value was calculated using the Pearson chi-square test at a significance level of α = 0.05. The result (p < .001) indicates a statistically significant association between region and condom use attitude, with participants from the Western Region showing a favourable condom use attitudes compared to those from the Central Region.

Frequency of Condom Use with a Man or a Woman and Condom Use Attitude

A greater proportion (n = 252; 51%) of respondents who used condoms consistently had a poor condom use attitude. Even among regular condom users, the situation was the same, as a greater proportion of participants had poor condom use attitudes. There was a statistical difference between the frequency of condom use with a man or woman and condom use attitudes.

Discussion

Ghana is a multicultural and multiethnic country ⁶⁶ with different cultural norms. From the results, the Western and Central Regions differ in sociocultural norms about condoms. Each of the two regions contributed approximately 500 participants to the study. Many of the MSM sampled in this study disclosed that they were either married to women or in some form of stable relationship with women to prevent public scrutiny. The predictive correlates of condom use attitude are being a resident in the Central Region of Ghana, affiliated with Pentecostal Christianity, being a Receptive MSM partner, and engaging “always”, “sometimes, or “rarely” in sexual activity with women. The Central Region had a poor condom use attitude due to higher perceived loss of sexual pleasure with condom use, higher perceived sexual health-related stigma and higher perceived embarrassment associated with condom purchase. In contrast, counterparts in the Western Region had a favourable condom use attitudes due to higher perceived condom reliability and effectiveness, lower perceived loss of sexual pleasure associated with condom use, lower perceived sexual health-related stigma, and lack of perceived embarrassment associated with condom purchase. Based on the comparative analysis of condom use attitudes in both regions, participants in the Western Region were more likely to implement consistent condom use than their counterparts in the Central Region. Practically, situational factors of the sexual encounter, such as perceived embarrassment associated with condom negotiation and use in both regions, could hinder consistent condom use in both Regions.

There was a high perceived reliability and effectiveness of condoms in the Central and Western Regions. The high perceived condom reliability and effectiveness could have translated to the high, almost three-fold (90.1%) increase in actual condom use during the last sexual encounter observed in this study, compared to the 38.9% reported among MSM in a previous Ghanaian study. ³³ Again, 90.1% of perceived condom reliability and effectiveness in this study exceeded the 51.8% reported in a study in Nigeria. ⁶⁷ This favourable perception of condom reliability and effectiveness among participants is attributed to several factors. First, the almost universal condom awareness campaigns in Ghana ⁶⁸ and in the sub-Saharan African regions on both social and traditional media could have fostered MSM confidence in condom reliability and effectiveness. ⁶⁹ Second, Ghana is a country with an elevated HIV prevalence in the population.^4,70 This elevated risk perception can enhance both the motivation for persistent condom usage and the importance attributed to condom effectiveness and reliability as a preventive measure. ⁷¹ MSM residing in a HIV high-prevalence country may be more inclined to perceive condoms as a reliable and effective means of HIV prevention. The perceived slightly higher condom reliability and effectiveness in the Central Regions relative to those in Western Region may be due to the benefits of sustained reproductive health education and HIV prevention outreach programmes in the Central Region, reinforcing education on condom reliability and effectiveness. ⁷² Despite the high perceived condom reliability and effectiveness as shown in Table 3, a greater proportion of respondents had poor condom use attitudes (Table 6).

OPEN IN VIEWER

Table 6.

Association between frequency of condom use with a man or a woman and condom use attitude.

Variables	Poor attitude (n = 513)	Favourable attitude (n = 487)	p-Value
Frequency of condom use with a man or a woman			.010
Always	252 (51.0)	242 (49.0)
Usually	96 (52.2)	88 (47.8)
Sometimes	114 (45.8)	135 (54.2)
Rarely	29 (69.0)	13 (31.0)
Never, as I prefer unprotected sex	22 (71.0)	9 (29.0)

p-Values were calculated using the Pearson chi-square test at a significance level of α = 0.05. The observed p-value (0.010) indicates a statistically significant association between the frequency of condom use with a man or woman and participants’ condom use attitudes.

Participants in the Central Region expressed a negative perception that condom use diminishes sexual pleasure. A previous study in Ghana associates the negative beliefs about condom protected sex to negative cultural norms that associate condom use with reduced sexual pleasure. ⁷³ The sociocultural beliefs that condoms interfered with the pleasure of sexual activity made condom use unappealing to sexually active individuals. ¹² In sub-Saharan Africa, including Ghana, condom-protected sex is often likened to “eating a candy with the wrapper on”, reflecting a view that sex with a condom is inferior or that condoms interfere with the desired feelings of sexual intercourse. ⁷⁴ This metaphor illustrates the cultural perception that “real sex” requires skin-to-skin contact, which is thought to enhance pleasure and meaning. ⁷⁴ Such negative sentiments contribute to the stigma surrounding condom use, as individuals associate condom use with a loss of intimacy and satisfaction in sexual experiences. ⁷⁴ Participants who did not use a condom in a previous sexual encounter with either a man or a woman had a higher perceived sexual pleasure loss compared to participants who used condoms. It can be deduced from Table 3 that the more frequently the use of condoms, the greater the perceived loss of sexual pleasure. For instance, respondents who reported using condoms “always” during sexual encounters with a woman had the greatest perceived sexual pleasure loss with condom use. Perceived pleasure loss during condom-protected sex is linked to lower sexual satisfaction, inconsistent use and a poor attitude towards its use.^75,76 This perceived altered sexual sensation associated with condom use may contribute to poor use. ⁷⁴ For instance, participants who reported “never” using condoms with either a man or a woman had the lowest perception that condoms diminished sexual pleasure, as shown in Table 3. Conversely, respondents in the Western Region maintained that condom use did not detract from sexual pleasure. It is more logical to predict that respondents in the Western Region would have a favourable attitude towards condom use, as shown in Table 6.

There was higher perceived embarrassment associated with condom negotiation and use in the Western Region relative to the Central Region, where participants were undecided. Despite evidence that both regions benefit from the Ghana AIDS Commission's HIV education programmes and NGO-led initiatives that target key populations, including MSM, ⁷⁷ there are regional differences in the level of perceived embarrassment about condom negotiation and use. These programmes often promote open discussion, provide peer support, and create safer spaces for sexual minorities. ⁷⁷ The difference in stigma related to condom negotiation and use reflects the varied sociocultural attitudes and exposure to sexual health education among sexual minorities. Each region's unique cultural norms could promote or limit participants’ level of perceived embarrassment associated with condom negotiation and use. It appears that participants in the Central Region may have developed more resilience, confidence, and negotiation skills when discussing condom use with partners rather than negative perceptions of embarrassment. However, respondents in the Western Region experienced greater embarrassment associated with condom negotiation and use (Table 3). Sexual minority groups may have less access to supportive networks or targeted education about open discussion about condom use with sexual partners, intensifying feelings of shame or awkwardness about condom negotiation. This difference underscores the contextual nature of condom use attitudes and the influence of regional sociocultural environments. Thus, condom use campaigns targeting MSM should be regionally tailored rather than the one-size-fits-all approach presently being implemented by Ghana's National AIDS Control Program.

Participants in the Western Region had 3.54 times higher odds of having a favourable condom use attitude compared to participants in the Central Region (95% CI: 2.57–4.89). This difference is plausibly explained by regional sociocultural and programmatic factors. National data and programmes describe the Western Region among Ghana's regions with relatively higher HIV prevalence. The Region thus attracts concentrated HIV prevention activities (condom distribution, behaviour-change campaigns, peer outreach, facility-based counselling). ⁷⁸ More exposure to targeted HIV prevention tends to increase positive attitudes towards condoms. A study in the Central Region shows stronger resistance to open condom and sexuality education, sustaining embarrassment around condom negotiation. ⁷⁹ Such tension plausibly produced less favourable condom use attitudes. ⁷⁹ This outcome aligns with the argument that attitudes are shaped by perceived norms. ⁸⁰ While sexual health programmatic intervention may have produced progressive attitudes towards sexual health in the Western Region, traditional beliefs and customs still have a strong influence on how participants in the Central Region perceive sex and behave sexually. Thus, cultural norms in the Western Region favoured positive condom use attitudes, while those in the Central Region remain repressive. Except for perceived embarrassment associated with condom negotiation and use, all other psychometric properties in the Western Region were better than those in the Central Region, as shown in Table 6.

Participants who were affiliated with Pentecostal Christianity had 1.45 times higher odds of having a favourable condom use attitude compared to their non-religious counterparts (95% CI: 1.00–2.11). Generally, regular participation in religious activities is linked to lower sexual risk behaviours, as religious involvement correlates with reduced odds of unprotected sex. ⁸¹ Specifically, research in Zimbabwe has reported a similar favourable attitude towards condom use among Pentecostal Christians.^82,83 Some Pentecostal teachings and community norms may emphasise personal responsibility, health, and moral accountability, which could influence individual attitudes towards safer sexual practices, including condom use. ⁸² Perhaps these teachings and community norms could have influenced the higher odds of favourable condom use attitude among Pentecostal Christians in this study.

Receptive partners had 1.54 times higher odds of having a favourable condom use attitude compared to their versatile counterparts (95% CI: 1.08–2.22). Despite the secrecy about same sex activities and homonegative attitudes towards sexual minorities, ⁸⁴ evidence shows that peer social networks that promote safer sexual practices, especially among receptive partners, ultimately contribute to HIV prevention efforts among MSM. ⁷¹ The higher odds of favourable condom use attitude among receptive partners could be due to the increased condom use awareness campaigns, as well as emphasising knowledge of the increased risk susceptibility of receptive partners to HIV/STI in condomless sex situations. The increased risk perception among receptive partners could have fostered these favourable attitudes towards condom use to minimise their sexual risk and improve their overall health.

More than two-thirds of MSM (n = 851; 85.1%) previously had sexual engagement with women, as shown in Table 1. This high sexual engagement of MSM with women may be a consequence of Ghana's punitive laws ⁸⁵ and homonegative attitudes against same-sex relationships. ⁸⁴ Due to the negative attitudes towards MSM in Ghana, ⁸⁴ participants reported consistent sexual relations with their wives or regular partners as a form of social cover. Nonetheless, they also disclosed sexual engagement with men, which they described as their preferred choice. It is plausible that participants were succumbing to Ghana's societal expectations of marriage and fatherhood, leading respondents to engage sexually with women despite their same-sex attractions. This concealment produces a “double life” effect, with decreased confidence to negotiate for condom use, enhanced shame, and negative attitudes towards condom use, due to fear of being outed. ⁸⁶ The high discordance between sexual identity and sexual behaviour may be necessary to prevent social scrutiny and adhere to hetero-normative patriarchal expectations. However, sexual identity-behaviour incongruence has been linked to negative sexual health behaviours, such as poor condom use, attitudes and inconsistent condom use. ⁸⁷ Sexual identity-behaviour discordance can be a form of denial for a homonegative attitude towards MSM. ⁸⁷ This explains why about one-tenth of participants reported sexual involvement with women “always” had poor condom use attitudes, as shown in Table 4.

The lower odds of having a favourable condom use attitudes among participants who were sexually involved with women previously, as shown in Table 4, could have negative implications for Ghana's HIV/AIDS control. For instance, respondents who reported sexual involvement with women “always” had 56% lower odds of having a favourable condom use attitude compared to those who engaged exclusively with men (0.44, 95% CI: 0.25–0.76). Respondents who reported sexual involvement with women “sometimes” had 33% lower odds of having a favourable condom use attitude compared to those who engaged exclusively with men (0.67, 95% CI = 0.45–0.99). Respondents who reported sexual involvement with women “rarely” had 44% lower odds of having a favourable condom use attitude compared to those who engaged exclusively with men (0.56, 95% CI = 0.31–1.01). Research in Costa Rica ⁸⁸ and South Korea ⁸⁹ attributes the lower odds of having a favourable condom use attitudes among MSMW to their decreased risk perception in the situation of vaginal sex, underestimating their vulnerability to sexually transmitted infections. It is logical to assume that the poor condom use among MSMW in this study is because vaginal sex is perceived as “relatively safe”, coupled with the lack of confidence to negotiate for their use, as shown in Table 3, suppressing favourable condom attitudes during sexual encounters with female partners. The poor condom use attitudes among MSMW are concerning, as this behaviour has implications for public health. MSM who also have sexual engagements with women (MSMW) are described as a “bridging group” ⁹⁰ and can increase cross-transmission of STIs between MSM and the heterosexual population in Ghana, given MSM's disproportionately high HIV prevalence in Ghana could with their limited access to preventive services due to stigmatising and discriminatory care attitudes.^91,92 The high number of respondents with MSMW characteristics, coupled with their increased poor beliefs about condoms and a lack of culturally sensitive sexual health care services^91,92 in a population subtype with disproportionately high HIV prevalence, requires attention. Apart from the concealment of their sexual identity producing a “double life” effect, with the lack of confidence to negotiate for condom use, enhancing shame, and poor attitudes towards condom use, due to fear of being outed, ⁸⁶ Nakamura et al ⁹³ associated the poor condom use attitudes among MSMW to their lack of confidence to negotiate their use, as using condoms with a female partner is associated with infidelity, disease prevention and the intention to keep emotional distance. To avoid relationship discord, MSMS develop poor condom use attitudes due to their lack of confidence to safely negotiate their use. ⁹⁴ This can make condom use with a female partner awkward, especially in a long-term relationship.

Participants who anticipated sexual involvement with women “rarely” in the future had 65% lower odds of having a favourable condom use attitude compared to those who engaged exclusively with men (0.35, 95% CI: 0.14–0.86). Anticipated future heterosexual contact among MSM is correlated with less favourable attitudes towards condoms, mirroring behavioural research with MSM who actively or sporadically have female partners. This is not limited to actual behaviours but extends to anticipated future behaviours, suggesting these individuals may experience enduring attitudinal barriers regardless of current sexual activity with women. ⁹³ Those anticipating heterosexual activity, even rarely, may have less interaction with gay-identified health services or social venues where condom use is routinely promoted. ⁹⁵ This siloing can further erode favourable condom use attitudes, as prevention messaging often misses subpopulations not closely tied to gay community networks. ⁹⁵ Additionally, MSM anticipating rare sexual encounters with women often face unique internalised stigma, pressure to maintain secrecy, and conflicts between heterosexual and homosexual identities. ⁹⁵ These factors reduce their comfort and confidence with condom negotiation and use, particularly in circumstances involving female partners. ⁹⁶ This effect is seen both in group comparisons and in qualitative analyses of disclosure comfort and anticipated relationship tensions. ⁹⁵ Studies in India ⁹⁷ and Britain ⁹⁸ confirm that rare expected contact with women does not eliminate, and may even exacerbate, poor condom attitudes among MSMW, partly due to fragmentation in sexual identity and community affiliation. Even when MSM have only minimal anticipation of heterosexual involvement, the internalised stigma, reduced community engagement, and anticipated relationship complications result in significantly poor condom attitudes compared to MSM who envision relationships only with men.^93,99

More than half of the participants (n = 513; 51.3%) had a poor condom use attitude (Table 5). The results show that neither the decision to use nor not to use condoms during the last sexual encounter with a man or woman was significantly impacted by the participants’ condom use attitudes. Even among respondents who reported using condoms regularly, over half of the participants had poor condom use attitudes. Participants’ attitudes or perceptions about condoms may not have as much of an impact on condom use in this population subtype as the sexual context, such as an MSM's ability to negotiate condom use. It also draws attention to a potential discrepancy between condom use attitudes and actual use, where participants may continue to use condoms despite negative beliefs, perhaps due to sexual health concerns. These findings implied that while efforts to improve attitudes towards condoms remain important, programmes should simultaneously prioritise partners’ ability to negotiate condom use, as this may have a stronger influence on actual condom use behaviour. These results have important public health implications for interventions aimed at increasing condom use for MSM in Ghana. This study shows that if Ghana were to understand condom use behaviour among MSM fully, the country cannot rely solely on the condom use attitudes among MSM, but on the characteristics of interpersonal situations, including the confidence to negotiate condom use. This finding replicates the results of previous condom use attitude studies.^100–102 According to Visser and Smith, ¹⁰³ having an agreement to use condoms during sexual encounters is more predictive of actual use than individual condom use attitudes. Evidence suggests that participants’ poor attitudes towards condom use may not significantly affect their actual use, suggesting that contextual factors, such as partner negotiation, play a more pivotal role in influencing condom use behaviour.^101,102 Thus, authors emphasised the need for improved partner negotiation alongside improving attitudes towards condom use.

Limitations

Questionnaires were the only instrument used to gather data; it is therefore recommended that future studies include in-depth interviews, focus group discussions or a stratified sampling technique to obtain the diverse views of respondents. Because the study involved two of Ghana's 16 regions, the findings are not nationally generalisable or representative. It is suggested that future studies cover all 16 regions of Ghana.

The Cronbach's alpha reported reflects the reliability across the entire sample, irrespective of the specific language used during data collection. Assessing the UCLA-MCAS reliability separately for Twi, Fante, and English would have strengthened the study's psychometric rigour by ensuring that the scale functioned consistently across linguistic groups. This is critical in multilingual settings, as translation nuances, idiomatic expressions, or cultural connotations may affect how items are interpreted, even when content is conceptually equivalent. Separate analysis based on language could have identified additional translation-related issues, reduced language-induced measurement bias, and improved internal validity. Language-specific reliability estimates would also support more accurate comparisons across linguistic groups, justify pooled analyses where appropriate, and enhance the cultural and contextual relevance of the findings in a multilingual setting like Ghana.

The study did not assess the context or location in which sexual encounters occurred. The location of sexual activity, such as a private home, bar or club, or public space, may influence respondents’ ability to either use condoms or negotiate for their use, particularly among MSM who may avoid private settings due to fear of stigma, violence, or hostility. Sexual encounters occurring in public or semi-public spaces may present unique challenges that complicate condom negotiation and consistent use among participants. The absence of data on the location of sexual encounters limits the study's ability to fully explain the contextual and structural factors shaping embarrassment associated with condom negotiation and use among MSM in the study regions.

The lack of stratification of condom use by partner type is a study limitation. For instance, given the sexual characteristics of the sample, a greater proportion (85%) of them were MSMW compared to 15% of participants having sexual engagements exclusively with men. About one-seventh of the participants who sexually engaged exclusively with men is a limitation of the study. However, the study did not disaggregate condom use by partner sex; specifically, condom use with women was not reported separately from condom use with men. As a result, the analysis was unable to capture potential differences in condom use patterns across partner types. Disaggregating condom use by male and female partners could have provided a more nuanced understanding of sexual risk behaviours among MSM, particularly for individuals who engage in sexual activity with both men and women. Inclusion of this information would have allowed for the identification of distinct condom use patterns that may differ by partner gender, thereby strengthening the interpretation of the findings and informing more targeted and context-specific HIV prevention interventions. Thus, a greater representation of MSM with sexual engagements exclusively with men in the sample could produce different themes that can allow for targeted intervention. Future studies should therefore incorporate partner-specific condom use measures to better support the design of tailored prevention strategies for MSM populations.

Additionally, a stratified sampling design drawing comparison of participants using sex roles can improve granularity and allow for targeted condom use attitudes interventions, as well as comparison of condom use attitudes in relation to sex roles of MSM in Ghana.

Policy Recommendations

This study demonstrates that promoting actual condom use requires more than changing population-level attitudes towards condom use, as additional psychological and relational barriers may constrain consistent use. While fostering positive condom use attitudes among individuals remains important, attitude alone may not be sufficient to determine use. A lack of confidence in condom negotiation and use underscores the need for sexual health education programmes that go beyond information provision. Teaching condom negotiation skills in isolation may be insufficient. This is because in relationships characterised by power imbalances, a dominant partner may assume control over sexual decision-making, making it difficult for the non-dominant partner to assert sexual health preferences, including requesting condom use. At the same time, norms of dominance and entitlement to sexual fulfilment may inhibit open discussion of condom use by dominant partners, as such discussions could be perceived as challenging their authority. These dynamics may result in situations where non-dominant partners experience powerlessness to request condom use, while dominant partners are similarly constrained in their ability to engage in mutual negotiation. In other words, promoting condom use among MSM may also require challenging broader societal norms and relationship-dominant discourses that shape power relations within same-sex relationships. Consequently, it is recommended that the Ghana AIDS Control Program and the Ghana AIDS Commission incorporate practical strategies for partner negotiation, alongside culturally sensitive messaging that addresses entrenched stereotypes and negative perceptions surrounding condom use in Ghana's sexual health education programmes. The agencies’ programmes of work should prioritise contextual and relational factors, such as practical skills in condom use negotiations and use between partners during sexual encounters, rather than focusing exclusively on individual-level attitudes. In light of these findings, future versions of the Ghana National Condom Strategy and the National HIV and AIDS Strategic Plan should incorporate targeted message strategies aimed at strengthening MSM condom use negotiation skills within their relational and sociocultural contexts.