Factors associated with meningitis vaccine awareness and engagement among Latino men who have sex with men in South Florida

Abstract

Awareness and uptake of the meningitis vaccine remains low among marginalized groups, such as Latino men who have sex with men (LMSM), potentially due to structural and psychosocial barriers in accessing preventative healthcare. The current study explored awareness and uptake of meningitis vaccines among a group of LMSM (N = 99) living in South Florida. A three-pronged variable selection approach was utilized prior to conducting regression models (linear and logistic). Overall, 48.5% of the participants reported little to no knowledge about meningitis vaccines, and 20.2% reported being vaccinated. Living with HIV (OR = 10.48) and time since outbreak (OR = 1.03) were significant predictors of meningitis vaccine uptake. No significant correlates of meningitis vaccine awareness were identified. More research is needed to identify other important factors associated with meningitis vaccine awareness and uptake among LMSM, a multiple marginalized group.

Introduction

Meningitis and other meningococcal diseases are infectious diseases that cause significant inflammation of the meninges – the membrane tissue layers that cover the brain and spinal cord (Meningitis | CDC, 2023). Meningococcal diseases are transmitted via close person-to-person contact (e.g., kissing, sharing drinks) or through airborne transmission (e.g., coughing). People who contract meningococcal diseases most commonly experience high fevers, headaches, stiffness in their neck and shoulders, light sensitivity, nausea, and even delirium with the most severe cases leading to paralysis or even death. Rates of meningococcal diseases in the U.S. have declined from 2,505 cases in 1970 to 371 total cases in 2019 (Meningitis | CDC, 2023). Despite these low rates, outbreaks are unforeseeable, and consequences can be severe for affected individuals and the larger health systems supporting communities most susceptible to meningitis acquisition. Therefore, the Center for Disease Prevention and Control (CDC) recommends immunization to reduce the risk of contracting meningococcal diseases, particularly among subgroups who experience meningitis-related health disparities (CDC, 2023). Specifically, the CDC recommends children receive their first vaccine at age 11 or 12 years, followed by a booster shot between the ages of 16–18 years (CDC, 2023). Additionally, for anyone with increased risk of acquisition (e.g., being part of a specific community with an outbreak, first year college students living in dorms) CDC suggests further vaccination if an individual’s last booster was administered more than 5 years prior (Meningococcal Vaccine Recommendations | CDC, 2023).

Men who have sex with men (MSM) are at increased susceptibility for acquiring meningitis. Results from a surveillance report with data collected in 2015 indicated that meningitis incidence among MSM in the U.S. was 0.54 cases per 100,000 population, compared to an incidence of 0.10 per 100,000 among individuals who did not report being MSM (Bozio, 2018). Additionally, in jurisdictions experiencing a meningitis outbreak, cases among MSM hover around 3.27 cases per 100,000 population, significantly higher than among other subpopulations residing in the same area (Bozio, 2018).

MSM may be more likely than the general population to acquire meningitis due to several structural and behavioral factors. As mentioned above, meningitis is transmitted via close person-to-person contact. MSM may more frequently be in situations in which this close person-to-person contact with a greater number of other individuals and within a smaller network. For instance, MSM tend to report more sexual partners than their heterosexual peers (Glick et al., 2012) which can facilitate the transmission of meningitis through behaviors that involves close intimate contact (e.g., kissing, CDC, 2023). Furthermore, since HIV and STIs disproportionately affect MSM, MSM with HIV or an STI may have weakened immune systems that may facilitate a biological vulnerability for meningitis acquisition (Folaranmi et al., 2017). In fact, MSM with HIV had 10.1 times the risk of being diagnosed with meningococcal diseases compared to their HIV-negative peers (Folaranmi et al., 2017). Additionally, factors that have consistently demonstrated an increased burden of other infectious diseases (e.g., HIV, COVID-19) such as living in more crowded settings, having limited access to relevant healthcare services, and more direct-public interfacing professions (Chun et al., 2023; Earnshaw et al., 2013) may lead racial and or ethnic minority MSM to experience even greater meningococcal disease-related disparities compared with white MSM.

Several factors have been implicated as being directly related to meningitis vaccine uptake among MSM. Psychosocial factors like medical mistrust, fear of disclosing one’s sexual orientation, limited health literacy, and poor relationships with medical providers have been associated with reduced meningitis vaccination among MSM across the U.S. (Johnson et al., 2019). Accurate knowledge and meningitis risk perception appear to be predictors of meningitis vaccination among MSM, with Black MSM possessing less knowledge and risk perception compared to White MSM (Phillips et al., 2018). Naidu and Polonijo (2023) observed that the timing and delivery of meningitis vaccination messaging may be integral to MSM’s vaccine uptake. MSM enrolled in their study expressed that bundling vaccination information with rapid HIV testing could induce anxiety; thus, they preferred receiving information via pamphlets and resources that could be discretely accessed online instead. These psychosocial and structural barriers, as well as the fact that meningitis vaccination dissemination amongst primary care providers is inconsistent due to variable insurance reimbursement structures (Simon et al., 2013), may contribute in part to why meningitis vaccination rates among MSM remain suboptimal even after an outbreak (Holloway et al., 2018).

Although U.S. prevalence is typically low, outbreaks of meningococcal disease can be sudden and cause significant morbidity and mortality among subgroups with increased susceptibility for acquisition. Outbreaks have been documented in communities with high proportions of MSM across the globe since 2001 (Aubert et al., 2015; Hellenbrand et al., 2016; Kratz et al., 2015; Martín-Sánchez et al., 2020; Nanduri et al., 2016), with 7 outbreaks reported in the U.S. alone from 2003 to 2015 (Bozio, 2018). Notably, at the end of 2021 into 2022, one such outbreak occurred among MSM in South Florida, which resulted in 24 cases and 6 deaths; this was cited as “one of the worst outbreaks of meningococcal disease among sexual minority men in U.S. history” by the CDC (2023).

Given the high impact of meningitis among Latino MSM (LMSM) during this South Florida outbreak, there was an opportunity to fill a key gap in scientific knowledge of determinants associated specifically with LMSM’s meningitis vaccine interest and uptake, which could in turn inform efforts to equitably scale up and disseminate vaccines. Building on limited prior work examining meningitis vaccine determinants among MSM in general, this current study seeks to examine rates of meningitis vaccination among LMSM during an outbreak, and drivers of vaccination in the context of an outbreak.

Methods

Participants and procedures

This secondary analysis uses data from a larger cross-sectional study focused on exploring LMSM’s and HIV prevention/treatment service providers’ preferences for the content, design, and implementation of a new HIV-related services referral network. With the emergence of the meningitis outbreak in South Florida during data collection, we added questions about meningitis vaccination in collaboration with our LMSM Community Advisory Board (CAB).

All participants (N = 99) in this sub-sample (1) were 18 years or older, (2) identified as gay, bisexual, another non-heterosexual identity, or a MSM, (3) resided in Miami-Dade or Broward counties, and (4) spoke English or Spanish fluently at the time of survey completion. LMSM enrolled in this study completed a one-time online assessment (approximately 45–60 min) between March 2022 and September 2022 and were compensated $50 for their participation. Assessments were available in both English and Spanish depending on participants’ language preferences. All participants reviewed consent information and checked a box indicating their consent to participate prior to beginning the survey. Survey data was collected and stored through REDCap and all procedures were approved by the University of Miami Institutional Review Board prior to starting enrollment.

Measures

The quantitative assessment that was developed for the parent study was informed by both a preliminary review of qualitative interviews conducted in phase 1 of the study (Harkness et al., 2021), and a literature review on potential facilitators and barriers to LMSM’s engagement with HIV and sexual health services as well as providers’ referrals to services (Harkness et al., 2022a, b; Weinstein et al., 2022a). For a more in-depth account of all measures assessed in the larger parent study, please see (Harkness et al., 2022a). A review of all variables relevant to this secondary analysis is included below.

Demographics

Participants responded to a brief set of demographic questions including: race, HIV status, age, nativity, citizenship status, preferred language, education level, employment status, immigration history, health insurance status, and history of engagement with HIV prevention and treatment services. These questions were adapted from a larger set of demographic questions developed by the University of Miami and Florida International University joint Center for Latino Health Research Opportunities (Mitrani et al., 2017). The following subset of demographic questions were considered as potentially important predictors in analyses.

1. Race and ethnicity. Participants were asked to select their race and ethnicity from a series of checkbox options including (a) Hispanic, Latino/a/x, or Spanish, (b) Black or African American, (c) White, (d) American Indian or Alaska Native, d) Asian, (e) Middle Eastern or North African, (f) Native Hawaiian or Other Pacific Islander, or (g) other. Each variable was treated as a binary.

2. Education level. Participants were asked to select from a list of education levels the highest grade of school they completed. Participants could select from 1 of 6 options: 1st to 8th grade, 9th to 11th grade, GED (high school equivalent), high school diploma, some college/university, or college/university degree. This variable was analyzed continuously.

3. Nativity. Participants were asked to select their country of birth from a list of countries. A binary variable was created to reflect U.S. born vs. foreign born, with U.S. born serving as the reference group.

4. HIV status. Participants were asked to report their current HIV status and whether they knew they had a detectable viral load. A binary variable was created to distinguish participants who reported being HIV-positive vs. HIV-negative or HIV status unknown. HIV-negative/unknown status served as the reference group.

5. Preferred language. Participants were asked to report their preferred language, whether it was English, Spanish, or something else. A binary variable was created such that Spanish- or another language as the preferred language served as the reference group.

6. Household income. Participants were given 5 income ranges in the past month (1 = less than $200, 2 = $200 - $499, 3 = $500 - $999, 4 = $1000 – $1999, and 5 = $2000 or more) and asked to indicate which range their monthly household income (including public assistance) fell into. This variable was analyzed continuously.

7. Health insurance status. Participants were given 9 insurance types and asked to indicate their current insurance plan. A binary variable was generated based on responses, with all options except for “uninsured” or “declined to answer” coded as 0, and “uninsured” coded as 1.

Sexual behavior

The current study assessed participants’ sexual activity levels based on questions developed by the CDC to assess PrEP eligibility. Participants were asked if they had any male sex partners in the past six months, and those who responded yes were coded as sexually active (reference group = not sexually active).

Substance use

Participants’ substance use was evaluated using the Single-Question Screening Test for Drug Use in Primary Care item (Smith et al., 2010). Participants were asked to report the number of times they used any substance in the past year via a 5-point Likert-type question (0 = never, 1 = less than once per month, 2 = 1–3 times per month, 3 = 1–2 times per week, 4 = more than 3 times per week). This variable was analyzed continuously such that higher scores reflected greater substance use frequency.

Binge drinking

A one-item measure was used to assess binge drinking behavior (Smith et al., 2009). Participants were asked to report the number of times in the past year they had four or more drinks in a day on a 5-point Likert-type scale (0 = never, 1 = less than once per month, 2 = 1–3 times per month, 3 = 1–2 times per week, 4 = more than 3 times per week). This variable was analyzed continuously such that higher scores indicated the frequency of binge drinking.

HPV vaccine history

Participants were asked if they had been vaccinated for HPV (yes/no) with non-vaccinated as reference group.

Time

A time variable (in days) was created by subtracting the date at which a participant completed the survey from the date in which the meningitis vaccination questions were added to the parent study questionnaire on April 27, 2022.¹

Meningitis vaccination: Awareness and uptake

With the emergence of the meningitis outbreak in South Florida during data collection, we added questions in collaboration with our CAB about meningitis vaccination. The CAB was comprised of five members (M_age = 37 years) who identify as LMSM (less than half born in the continental U.S.) and provide perspectives on LMSM’s experiences across South Florida. CAB members were chosen based on varied identifiers and experience with sexual health services (e.g., some actively working in the HIV services field and some not), to inform research.

The CAB agreed that the South Florida meningitis outbreak was an important opportunity to assess whether LMSM were aware of the outbreak and obtaining a vaccine. The CAB reviewed two one-item measures of vaccine awareness and uptake, respectively, described below that was written by the authors of this manuscript. The CAB found the items to be appropriate. However, they felt that given the ongoing outbreak, it was important to include information about the outbreak and vaccination, which we added after the questions. Following our discussion with the CAB in early April, we added the two items below on April 27, 2022, followed by links to information about the outbreak and vaccination, to the parent study.

Meningitis vaccine awareness

LMSM’s awareness of a meningitis vaccine was assessed using a one-item measure rated on a 5-point Likert scale. This measure was adapted from post-exposure prophylaxis and COVID-19 vaccine awareness measures we used in our prior work (Harkness et al., 2022a; Weinstein et al., 2022a). LMSM had the option to report the following regarding the meningitis vaccine: “I’ve never heard of it before today,” “I’ve heard about it, but I don’t understand it fully,” “I know a little bit about it,” “I know a fair amount about it,” “I know a lot about it.” Higher scores on this item reflected greater awareness of the meningitis vaccine.

Meningitis vaccine uptake

To complement the meningitis vaccine awareness question, participants also indicated if they had received a meningitis vaccine on a 5-point Likert scale that mirrored the 5 stages of change ranging from 1 “precontemplation” to 5 “maintenance” (Prochaska & Velicer, 1997) and was modeled based on our prior work assessing COVID-19 vaccination with LMSM (Weinstein et al., 2022a) and PrEP (Harkness et al., 2022b). LMSM had the option to report the following: “I have not received or I don’t know if I have received a meningitis vaccine in the past 5 years and am not planning to get one,” “ I have not received or I don’t know if I have received a meningitis vaccine in the past 5 years but I am considering getting one,” “I have not received or I don’t know if I have received a meningitis vaccine in the past 5 years but I am actively planning to get one,” “I have already received a meningitis vaccine in the past 5 years,” “I have already received a meningitis vaccine in the past 5 years and I plan to continue to get boosters in the future when needed.” Vaccine uptake was then coded as a binary such that participants who had not been vaccinated scored a 0 compared to LMSM who have already received the meningitis vaccine who scored a 1.

Variable selection

Potential factors associated with our outcomes of interest – meningitis vaccine awareness and meningitis vaccine uptake among LMSM - were selected using a three-pronged approach prior to being modeled using traditional regression approaches. This approach has been utilized by authors in prior projects (e.g., Weinstein et al., 2022a, b) and is described in more detail below. The participant demographics table really needs to go first as Table 1.

Table 1.

Factors identified via a three-pronged variable selection approach and traditional linear regression model results for meningitis vaccine awareness (N = 99)

Predictors	SSVS Model R2 = 0.04	CAB Model R2 = 0.07	Prior Lit. Model R2 = 0.03	Final Model B (SE), Effect Size [semi-partial correlation] R2 = 0.10
Age	0.01 (0.01), [0.13]	--	--	0.03 (0.02), [0.22]
Substance use frequency	−0.12 (0.10), [−0.12]	--	--	−0.16 (0.14), [−0.12]
Education level	--	0.20 (0.13), [0.16]	--	0.25 (0.16), [0.15]
Born outside of the U.S.	--	−0.33 (0.33), [−0.10]	--	−0.45 (0.47), [−0.12]
Preferred language - English	--	0.37 (0.33), [−0.12]	--	−0.05 (0.55), [−0.01]
Being uninsured	--	−0.15, (0.37), [−0.04]	−0.36 (0.37), [−0.10]	−0.31 (0.49), [−0.01]
Sexually active in past 6-months	--	0.30 (0.35), [0.09]	0.49 (0.35), [0.15]	0.68 (0.60), [0.15]
History of HPV vaccination	--	--	−0.01 (0.31), [−0.002]	0.13 (0.38), [0.03]

Notes:

⁺p < 0.10

^*p < 0.05,

^**p < 0.01

Approach 1: Data driven via stochastic search variable selection (SSVS)

SSVS is an exploratory approach that allows researchers to reduce an extensive group of predictors into a select few that are most reliably associated with an outcome, among all potential predictors. Developed out of Bayesian statistics principles, SSVS estimates the probability that a given predictor is associated with an outcome of interest (i.e., has a non-zero regression coefficient) while concurrently, controlling for any statistical uncertainty that may be produced by other predictors simultaneously being assessed in the model (Bainter et al., 2020). In identifying the best fitting models, SSVS conducts thousands of simulations of samples of independent variable combinations and then ascertains the marginal inclusion probability, or the proportion of times each variable was selected in the sampled models, for each factor. This approach minimizes both Type I and Type II errors, by identifying factors that are reliably associated with the outcome of interest.

Thirty-one participants had at least some missing data on one of the correlates of interest. One of the assumptions of SSVS is that there is no missing data. Therefore, participants with missing data on the variables relevant to this secondary analysis were removed from SSVS analyses leaving a sample of 68 participants for the SSVS component of the variable selection process. SSVS models were run using the SSVSforPsych shiny app (Bainter et al., 2020) and SSVS specifications included a prior inclusion probability of 0.5, 1000 burn-in iterations, and 10000 total iterations. Variables that included a post inclusion probability of 0.2 were included in the final analytic models (Harkness et al., 2022a; Kwon et al., 2011; Weinstein et al., 2022a, b).

Approach 2: Literature driven via variables already identified as related to meningitis vaccine awareness and uptake in published research

Prior research has identified several variables associated with meningitis vaccine awareness and meningitis vaccine uptake among LMSM. Therefore, to be as comprehensive as possible in our variable selection approach, factors identified in the published literature that were not selected based on the SSVS and CAB feedback were added in the final traditional regression models.

Approach 3: Community-based participatory driven variable selection via CAB discussions

To complement our data driven and literature generated approaches, authors leveraged the lived expertise of LMSM CAB members affiliated with the REACH Equity Team CAB to help select potential factors related to meningitis vaccine awareness and meningitis vaccine uptake among LMSM. A core component of community engaged research, CABs have played an integral role in promoting health equity among sexual minority communities, particularly in HIV prevention and treatment-related and adjacent research (Weinstein et al., 2023).

To facilitate this process, the senior author provided CAB members with both an overview of the parent study and secondary analysis project during the May 2023 CAB meeting. Next, CAB members were asked to identify up to five factors available in the parent dataset that they intuited as being most strongly related to meningitis vaccine awareness and meningitis vaccine uptake among LMSM in the greater Miami area. After CAB members independently identified potentially important factors, the senior and first author then facilitated a consensus discussion to determine the top 4 most likely correlates of meningitis vaccine awareness and meningitis vaccine uptake (found in Tables 2 and 3) respectively based on the lived experiences of CAB members. Furthermore, all CAB members were given the opportunity to review findings and offer feedback for interpretation of results.

Table 2.

Factors identified via a three-pronged variable selection approach and traditional logistic regression model results for meningitis vaccine uptake (N = 99)

Predictors	SSVS Model Exp(B), [95% CI: Exp(B)] R2 = 0.42	CAB Model Exp(B), [95% CI: Exp(B)] R2 = 0.12	Prior Lit. Model Exp(B), [95% CI: Exp(B)] R2 = 0.29	Logistic Regression Model Exp(B), [95% CI: Exp(B)] R2 = 0.49
Living with HIV	8.52, [1.52, 47.78]**	--	--	10.48, [1.49–73.59]**
Past year binge drinking frequency	0.51, [0.14, 1.92]	--	--	0.38, [0.09–1.63]
Education level	--	0.74, [0.47, 1.15]	--	0.83, [0.47–1.48]
Identify as Black/African-American	--	1.63, [0.27, 9.76]	--	7.39, [0.63–87.37]
Sexually active in past 6 months	--	0.72, [0.21, 2.42]	1.06, [0.30, 3.71]	1.47, [0.35–6.16]
Being uninsured	--	0.23, [0.03, 1.97]	0.53, [0.06, 4.76]	0.81, [0.07–9.94]
History of HPV vaccination	--	--		1.16, [0.31–4.32]
Past month income level	--	--	0.83, [0.52, 1.33]	1.10, [0.62–1.96]
Time (in days)	1.02, [1.00, 1.04]	--	1.03, [1.01, 1.05]**	1.02, [1.00–1.04]+

Notes:

⁺p < 0.10

^*p < 0.05,

^**p < 0.01

Table 3.

Descriptive statistics of latino sexual minority men (N = 99)

Variable	Frequency (%)/ mean (SD)
Age	39.59 (13.10)
Sexual Orientation
Gay	79 (79.8%)
Bisexual	17 (17.2%)
Another sexual minority orientation	3 (3.0%)
Race & Ethnicity 1
Hispanic, Latino/a/x, or Spanish2	99 (100.0%)
Black or African American	7 (7.1%)
White	17 (17.2%)
Middle Eastern or North African	1 (1.0%)
Nativity
U.S. born	39 (39.4%)
Foreign born	60 (60.6%)
Citizenship status
U.S. citizen	71 (71.7%)
Non U.S. citizen	27 (27.3%)
Missing/Declined to answer	1 (1.0%)
Preferred language
English only	59 (59.6%)
Spanish only	39 (39.4%)
Other	1 (1.0%)
Living with HIV
Yes	46 (46.5%)
No (HIV-negative or unknown HIV status)	51 (51.5%)
Missing/Declined to answer	2 (2.0%)
Health insurance status
Insured	79 (79.8%)
Uninsured	17 (17.2%)
Missing/Declined to answer	3 (3.0%)
Income past month
Less than $200	4 (4.0%)
$200 – $499	5 (5.1%)
$500 – $999	12 (12.1%)
$1000 – $1999	25 (25.3%)
$2000 or more	45 (45.5%)
Missing/Declined to answer	8 (8.1%)
Education level
1st to 8th grade	1 (1.0%)
9th to 11th grade	4 (4.0%)
GED (high school equivalent)	2 (2.0%)
High school diploma	11 (11.1%)
Some college/university	17 (17.2%)
College/university degree	64 (64.6%)
Alcohol use frequency
Never	32 (32.3%)
<1 per month	32 (32.3%)
1–3x per month	20 (20.2%)
1–2x per week	10 (10.1%)
>3 per week	5 (5.1%)
Substance use frequency
Never	59 (59.6%)
<1 per month	13 (13.1%)
1–3x per month	9 (9.1%)
1–2x per week	4 (4.0%)
>3 per week	12 (12.1%)
Missing/Declined to answer	2 (2.0%)
Sexual activity level
Sexually active in past 6 months	75 (75.8%)
Not sexually active in last 6 months	22 (24.2%)
Meningitis vaccine awareness
I’ve never heard about it before today	26 (26.3%)
I’ve heard about it, but I don’t understand it fully	22 (22.2%)
I know a little bit about it	25 (25.3%)
I know a fair amount about it	15 (15.2%)
I know a lot about it	10 (10.1%)
Missing/Declined to answer	1 (1.0%)
Meningitis vaccine uptake
Precontemplation: “I’ve not received or I don’t know if I have received a meningitis vaccine in the past 5 years and am not planning to get one.”	27 (27.3%)
Contemplation: “I’ve not received or I don’t know if I have received a meningitis vaccine in the past 5 years but I am considering getting one.”	33 (33.3%)
Preparation: “I’ve not received or I don’t know if I have received a meningitis vaccine in the past 5 years but I am actively planning to get one.”	18 (18.2%)
Action: “I’ve already received a meningitis vaccine in the past 5 years.”	9 (9.1%)
Maintenance: “I’ve already received a meningitis vaccine in the past 5 years and I plan to continue to get boosters in the future when needed.”	11 (11.1%)
Missing/Declined to answer	1 (1.0%)

Notes:

¹Race and ethnicity options were “check all that apply” so that participants could select as many race and ethnicity options as they felt were appropriate

²For eligibility purposes, all participants in the study identified as Hispanic, Latino/a/x, or Spanish on the study screener. Therefore, data for this variable is composed of data collected from the screener or study survey

Data analytic plan

Prior to assessing study aims, participant demographics and key variables were explored via descriptive statistics. All categorical variables were first dummy coded. Next, the authors completed the three variable selection approaches described above and ran one regression model per variable selection approach, followed by a final regression model integrating all three variable selection approaches (four total regression models per outcome). Linear regression was used for the vaccine awareness outcome and logistic regression was used for the vaccine uptake outcome. R² values were obtained for all models to evaluate each model’s predictive power. All descriptive analyses were explored using SPSS version 27 (IBM Corp., 2021) and regression models were evaluated using Mplus version 8.7 (Muthén & Muthén, 2017).

Based on reviewer feedback, the time variable (in days) was created post-hoc. To determine if time should be included in regression models, simple regression models were conducted with time as the sole predictor for each outcome, vaccine awareness (linear model) and vaccine uptake (logistic model). Time was not a statistically significant predictor of vaccine awareness; thus, it was not retained for future linear regression models. Comparatively, since time was a statistically significant predictor of vaccine uptake, the time variable was included in all future models in which vaccine uptake was the outcome of interest.

Results

The final secondary data analytic sample included 99 LMSM living with and without HIV and residing in the South Florida area. On average, participants were 39.6 years old (SD = 13.1 years) and predominately identified as gay (79.8%). All participants identified as Hispanic, Latino/a/x, or Spanish and more than half (60.6%) were born outside of the U.S. (and U.S. territories). Among those who immigrated to the U.S., the mean number of years they had lived in the US was 16.6 years (SD = 13.2 years). Approximately half of participants were HIV-negative (51.5%). Just over half of participants (54.5%) were living off less than $2,000 a month and most LMSM had health insurance (79.8%). Substance use and binge drinking behavior were low among participants with LMSM often reporting no substance use in the past month (59.6%) and binge drinking less than one time on average in the past month (64.6%). Awareness of the meningitis vaccine was low amongst participants with 48.5% of LMSM reporting that they have “never heard about the vaccine before today” or “have heard about it, but don’t understand it fully” and only 20.2% of LMSM enrolled in the study had already received a meningitis vaccine at the time of survey completion. A more in-depth review of descriptive statistics can be found in Table 1.

Meningitis vaccine awareness

SSVS identified two predictors – age and past month substance use frequency - with markedly higher inclusion probabilities than the other potential predictors (i.e., above 0.20) which we selected for inclusion in the final linear regression model. In addition to age and substance use frequency, four predictors identified by the CAB: education level, insurance status (Simon et al., 2013), preferred language, and nativity were included in the final linear regression model. Furthermore, two additional factors gleaned from the larger literature on meningitis vaccine awareness, sexual activity in the past 6 months (binary variable; Cooper et al., 2018) and past vaccination history (i.e., HPV vaccination; Naidu & Polonijo, 2023), were also included in the final linear regression model (see Table 2 for all model predictors).

None of the potentially influential variables selected via our three-pronged selection approach were significantly associated with meningitis awareness among LMSM in this study when employing a traditional linear regression model. Additionally, R² values were low for the four models: 0.04 for the SSVS model, 0.07 for the CAB model, 0.03 for the prior literature model, and 0.10 for the final full model. This demonstrates that less than 10% of the variance in our outcome of interest, meningitis vaccine awareness can be explained by the predictors assessed.

Meningitis vaccine uptake

For our second outcome of interest, SSVS identified three predictors with markedly higher inclusion probabilities (i.e., above 0.20) – HIV status, identifying as Black, and frequency of binge drinking in the past month - that were then entered into a traditional logistic regression model. To complement these three data driven selected variables, additional variables identified as important in the literature were also added including insurance status (Johnson et al., 2019; Simon et al., 2013), being sexually active in the past month (Mann-Barnes et al., 2023), income (Johnson et al., 2019), and past vaccination history (i.e., HPV vaccination; Naidu & Polonijo, 2023). In addition to insurance status and being sexually active in the past 6 months, the CAB also suggested that education level would be an influential variable for the final model. As described above, we added time as a predictor post-hoc (see Table 3 for all model predictors).

A traditional logistic regression model was employed to assess our second outcome of interest, meningitis vaccine uptake. Of the nine potential factors associated with meningitis uptake among LMSM, only HIV status was significantly associated with meningitis vaccine uptake. LMSM living with HIV had 10.48 times the odds of being vaccinated for meningitis (exp(B) = 10.48, CI = 1.49–73.59) compared to LMSM without HIV at the time of survey completion. Additionally, time (in days) trended towards significance in its relation to meningitis vaccine uptake (exp(B) = 1.02, CI = 1.00–1.04). This suggests about a 2% increase in likelihood for LMSM to be vaccinated against meningitis for each additional day.

There was significant variability in R² values across the three variable selection approaches. The CAB model yielded the lowest R² value (0.12) followed by the prior literature review (0.29). The SSVS model had the highest R² value (0.42) suggesting that this approach had the greatest explanatory power. The final full model R² was high (0.49), indicating that 49% of the variability in our outcome of interest, meningitis vaccine uptake can be explained by the predictors assessed.

Discussion

Sparked by a recent meningococcal virus outbreak in Florida starting in December 2021 and lasting through December 2022, this cross-sectional study assessed rates of meningitis vaccine awareness and uptake and explored determinants of vaccination among a group of LMSM living in South Florida. Overall, findings indicate low meningitis vaccine awareness (50.6% reported knowing a lot or a fair amount about the meningitis vaccine) and limited uptake (20.2%) among LMSM enrolled in this study. Further, the findings revealed that HIV status and time since outbreak were significant predictors of meningitis vaccine uptake such that LMSM living with HIV were far more likely to be vaccinated compared to their HIV-negative peers. No significant correlates of meningitis vaccine awareness were observed.

Overall, meningitis vaccine awareness and uptake were low among this group of LMSM living in South Florida. This survey was administered to LMSM living in South Florida, a population that was disproportionately affected by a large meningitis outbreak from December 2021 to December 2022; therefore, these low rates of vaccine awareness and uptake are concerning from a public and personal health perspective. Compared to other vaccines that are either annual (e.g., influenza), or more widely discussed in popular media (e.g., COVID-19), the meningitis vaccine is less commonly known among MSM (Naidu & Polonijo, 2023) and has a more challenging vaccination regimen (CDC, 2023). These two factors may complicate consistent vaccination coverage among this marginalized group.

Time since the start of the outbreak and HIV status were associated with meningitis vaccine uptake within this sample. Our time variable, measured in days, showed that each additional day was associated with a 2% greater likelihood of vaccine uptake. Although this finding only trended towards significance, it is consistent with published research on patterns of vaccination which suggest that the longer it is since the development of a vaccine and or the initial date of an outbreak, the more likely it is for members of the public to be vaccinated (Koskan et al., 2023; Weinstein et al., 2022a). Additionally, LMSM with HIV were more than 10 times more likely to have been vaccinated against meningitis compared to HIV-negative LMSM in this study. Similar trends have been observed in the larger vaccine literature suggesting that systems of HIV care often do well in delivering both HIV care and ancillary healthcare services to ensure the holistic health of people living with HIV (Bulstra et al., 2021; Monroe et al., 2019).

There are several reasons why HIV status may have been significantly associated with meningitis uptake in this study. First, people living with HIV are immunocompromised, and as such, may be more cognizant of vaccination efforts to keep themselves safe from other infectious diseases. This is consistent with past research suggesting that individuals living with HIV were more likely to be COVID-19 vaccinated compared to their HIV-negative peers (Campbell et al., 2023). Alternatively, it is possible that HIV status may have served as a proxy for engagement with healthcare services and that increased uptake of the meningitis vaccine among participants who were HIV-positive was related to these individuals being more regularly engaged with comprehensive health services. Future research should elucidate the mechanisms by which HIV status influences LMSM’s vaccine uptake.

Prior health-related literature suggests that social determinants of health and engagement with certain health promotion behaviors may be associated with meningitis vaccine uptake among groups with increased susceptibility (Nadarzynski et al., 2021; Naidu & Polonijo, 2023). Therefore, given the published literature, the absence of significant findings in our study was unexpected, but not entirely surprising due to several limitations. First, due to the small sample size (N = 99), our analyses may be underpowered which probably affected our ability to detect significant effects. Similarly, low variability across variables, including vaccine uptake, may have further impeded model estimation and interpretation in reducing the amount of unique information that could be analyzed within the traditional regression models. Furthermore, vaccine awareness is harder to operationalize than vaccine uptake due to its subjectivity. It is possible that enrolled LMSM interpreted awareness differently within the sample, making it more challenging to systematically identify significant findings. Future studies could explore vaccine awareness with objective knowledge measures (e.g., true/false knowledge questions). Lastly, given that this was a secondary analysis, we were limited in the predictors we could include in our models. Future studies should employ mixed-methods approaches that utilize triangulation (i.e., the integration of multiple data sources or methods) to reduce the likelihood of missing important determinants (e.g., perceived susceptibility for acquisition, health literacy, attitudes towards vaccination in general, number of sexual partners).

Another scientific contribution of this study is our comparison of different variable selection approaches and their explanatory power. Our data driven method, SSVS, is designed to empirically identify predictors that maximize the explanatory power of a predictive model. As such, it is not necessarily surprising that this model explained more of the variance in our outcomes of interest, compared to the CAB and prior literature models. Although the CAB and prior literature models did not explain as much of the variance in the outcomes as the SSVS model did, the predictors selected via these two approaches did aid in improving the overall predictive power of the full model, which included predictors across all three approaches. Our findings suggest the utility of further exploring different variable selection approaches, and their combinations, in future research.

Despite its limitations, this study possessed strengths that enhance the credibility of the observed findings. Primarily, our innovative three-pronged variable selection method that synthesized participatory data science (i.e., CAB recommendations) and data driven (i.e., literature review, stochastic search variable selection) approaches should increase both replicability and validity of findings. Additionally, the study’s emphasis on LMSM, a multiple marginalized group with increased susceptibility for meningitis and other infectious disease acquisition helps to expand the vaccine-related literature among this under-researched population. Furthermore, the fact that this study was conducted in South Florida, a region with significant health inequities, a recent history of a meningitis outbreak (CDC, 2023), and persistent infectious disease epidemics (e.g., HIV, COVID-19; CDC, 2020; Witkowski et al., 2023) increases the generalizability of these findings to other U.S. locales with similar infectious disease disparities.

Conclusion

As infectious disease outbreaks and global pandemics become more frequent, research examining how to support marginalized groups with health-related inequities is urgently needed. Even though this study did not yield an extensive list of new correlates for meningitis vaccine awareness and uptake among LMSM, results indicate that HIV status and time since the start of an outbreak are important predictors of vaccination among this group. Additionally, findings from this study highlight the utility of a three-pronged variable selection approach that utilizes data (i.e., SSVS) and theoretically driven (i.e., CAB, prior literature) approaches to create inclusive models with strong predictive power in regression models. Although future studies should aim to corroborate and extend these findings among other populations of LMSM using a similar, inclusive three-pronged variable selection approach, current results can contribute to meta-analyses centered on meningitis vaccination among both the general population more broadly, and LMSM specifically.

Data availability

The data that support the findings of this study are available upon reasonable request (e.g., methodologically sound proposal and signed data use agreement) to the corresponding author (A.H.), following publication.