A Mixed-Methods Comparison of Gender Differences in Alcohol Consumption and Drinking Characteristics among Patients in Moshi, Tanzania

1 1 A Mixed-Methods Comparison of Gender Differences in Alcohol Consumption and 2 Drinking Characteristics among Patients in Moshi, Tanzania 3 4 Authors and affiliations: 5 Alena Pauley MScGH1 – [email protected] (ORCID#: 0000-0002-7489-7469) 6 Mia Buono1 – [email protected] 7 Kirstin West1 – [email protected] 8 Madeline Metcalf1 – [email protected] 9 Sharla Rent, MD MScGH1,2 - [email protected] 10 Joseph Kilasara3,5 APRN MNS-FNP– [email protected] 11 Yvonne Sawe3 – [email protected] 12 Mariana Mikindo3 – [email protected] 13 Blandina T. Mmbaga, MD PhD3,4,5 – [email protected] 14 Judith Boshe MD MMED3,5 – [email protected] 15 João Ricardo Nickenig Vissoci MBA MSc PhD1,6 – [email protected] 16 Catherine A. Staton, MD MScGH1,6 – [email protected] 17 18 1 Duke Global Health Institute, Duke University, Durham, NC USA 19 2 Duke Department of Pediatrics, Duke University Medical Center, Durham, NC USA 20 3 Kilimanjaro Christian Medical Center, Moshi, Tanzania 21 4 Kilimanjaro Clinical Research Institute, Moshi, Tanzania 22 5 Kilimanjaro Christian Medical University College, Moshi, Tanzania 23 6 Duke Department of Surgery, Duke University Medical Center, Durham, NC USA . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. 2 24 25 Corresponding Author in event of publication: [email protected] (CAS) 26 ORCID#: 0000-0002-6468-2894 27 28 Funding: This project was funded by the Duke Global Health Institute Graduate Student funds, 29 and the Josiah Trent Foundation. Infrastructure built by NIH grant R01 AA027512 (PI Staton) 30 was used to support the data collection process for this grant to understand gender related aspects 31 of alcohol use at KCMC. 32 Competing interests: The authors declare no competing interests 33 34 Author contributions: 35 Conceptualization: AMP, CAS, BTM, JB 36 Methodology: AMP, CAS, JRNV 37 Formal analysis and investigation: CAS, BTM, JB 38 Data Collection: JK, YS, MMi, AMP 39 Writing - original draft preparation: AMP, MB, KW, MM 40 Writing - review and editing: SR, AMP, MB, KW, MM, CAS 41 Funding acquisition: AMP, CAS 42 Supervision: CAS, JRNV, BTM . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 3 43 ABSTRACT 44 Background: Excessive alcohol use stands as a serious threat to individual and community well- 45 being, having been linked to a wide array of physical, social, mental, and economic harms. 46 Alcohol consumption differs by gender, a trend seen both globally and in Moshi, Tanzania, a 47 region with especially high rates of intake and few resources for alcohol-related care. To develop 48 effective gender-appropriate treatment interventions, differences in drinking behaviors between 49 men and women must be better understood. Our study aims to identify and explore gender-based 50 discrepancies in alcohol consumption among Kilimanjaro Christian Medical Center (KCMC) 51 patients. 52 Methods: A systematic random sampling of adult patients presenting to KCMC’s Emergency 53 Department (ED) or Reproductive Health Center (RHC) was conducted from October 2020 until 54 May 2021. Patients answered demographic and alcohol use-related questions and completed 55 brief surveys including the Alcohol Use Disorder Identification Test (AUDIT). Through 56 purposeful sampling, 19 subjects also participated in in-depth interviews (IDIs) focused on 57 identifying gender differences in alcohol use. 58 Results: During the 8-month data collection timeline, 655 patients were enrolled. Men and 59 women patients at KCMC’s ED and RHC were found to have significant differences in their 60 alcohol use behaviors including lower rates of consumption among women, (average [SD] 61 AUDIT scores were 6.76 [8.16] among ED men, 3.07 [4.76] among ED women, and 1.86 [3.46] 62 among RHC women), greater social restrictions around women’s drinking, and more secretive 63 alcohol use behaviors for where and when women would drink. For men, excess drinking was 64 normalized within Moshi, tied to men’s social interactions with other men, and generally 65 motivated by stress, social pressure, and despair over lack of opportunity. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 4 66 Conclusion: Significant gender differences in drinking behaviors were found, primarily 67 influenced by sociocultural norms. These dissimilarities in alcohol use suggest that future 68 alcohol-related programs should incorporate gender in their conceptualization and 69 implementation. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 5 70 Introduction 71 Alcohol use is a leading risk factor for death and disability-adjusted life years (DALYs) 72 worldwide, accounting for over 3 million deaths each year (1,2). Globally, alcohol use and 73 alcohol-related harm are increasing, especially so within low-and middle-income countries(3). 74 For example, the World Health Organization (WHO) Africa region consumes on average 20% 75 more alcohol per day (40.0g/day) than the global average (32.8g/day), with Tanzania ingesting 76 especially large quantities. The rate of heavy episodic drinking in Tanzania (7.7% among women 77 and 33.4% among men 15 years or older), is almost twice that of neighboring countries (1). 78 Moshi, a popular tourist town located at the base of Mount Kilimanjaro in Northern Tanzania, 79 has particularly high rates of alcohol use, which have also been increasing in recent years 80 compared to nearby regions (4–6). This increase is influenced by a strong drinking culture and a 81 custom of early alcohol initiation in minors for members of the Chagga ethnic group who 82 constitute the majority of local inhabitants (7,8). Standing also as contributing factors is 83 alcohol’s ready availability mixed with its low cost and a recent increase in disposable income 84 among local inhabitants (4,7,9). 85 Both globally and within Tanzania, alcohol consumption patterns and alcohol-related 86 harms are dissimilar among men and women (10–15). On average, men consume alcohol in 87 higher quantities and more frequently than women, with a higher prevalence of Alcohol Use 88 Disorder (AUD) worldwide (11,16,17). Men are also more likely to engage in injurious 89 behaviors like drunk driving, violence, and crime while under the influence (18,19). While 90 women intake less alcohol, drinking can reduce inhibitions and awareness of risk, increasing the 91 likelihood of women being subject to unsafe scenarios like violence or sexual assault which in 92 turn increases the risk of acquiring sexually transmitted diseases (20–23). For women who are . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 6 93 pregnant, alcohol use also can introduce harm to themselves and their unborn child (24,25). 94 Sociocultural factors and norms are also known to impact alcohol consumption and use 95 behaviors. Among youth in both Tanzania and nearby Kenya, young people who spent time 96 around those already consuming alcohol or encouraging alcohol use were found to have higher 97 rates of intake themselves (7,26). 98 In Tanzania specifically, little is known on how or why alcohol use differs between 99 genders. The data that does exist has indicated that while intoxicated, men are more likely to 100 exhibit violent behaviors while women are more likely to be victims of violence. For example, 101 men exhibit a greater prevalence of alcohol use and abuse, a greater risk of injury after drinking, 102 are more likely to incur road-traffic injuries, and have more public displays of alcohol intake 103 compared to women (13,15,27). In contrast, alcohol consumption among women in the region 104 increases the risk of experiencing sexual violence and contracting sexually transmitted infections 105 (28). Estimates for alcohol abuse have ranged from 7% among women with partners and 22.8% 106 among men in the general population (4), to an AUD prevalence of 38.7% among men and 107 13.1% among women patients at a local outpatient clinic (29). 108 To reduce the burden associated with alcohol in Moshi, more research about how and 109 why alcohol intake varies between genders is needed. Given the known discrepancies in alcohol 110 use by men and women, this information can be used to design and implement more effective 111 alcohol-reduction programs that are both gender and culturally appropriate in this region. 112 Addressing this knowledge gap, this study aims to better understand the key gender differences 113 in alcohol consumption and use patterns among patients in Moshi, Tanzania. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 7 114 Methods 115 Overview 116 This was a sequential explanatory mixed-methods study that combined quantitative 117 survey score data and qualitative semi-structured in-depth interviews (IDIs). Quantitative data 118 was collected prior to IDIs to guide qualitative data sampling, with all data collection procedures 119 occurring from October 2021 until May of 2022 at the Kilimanjaro Christian Medical Centre 120 (KCMC) in Moshi, Tanzania (Figure 1). In total, 655 subjects participated in survey 121 questionnaires, and of those individuals, 19 were selected, via purposeful sampling, to participate 122 in IDIs addressing alcohol use. 123 Figure 1: Study Design Overview 124 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 8 125 Setting 126 This study was based within the Kilimanjaro Christian Medical Center (KCMC), a large 127 referral and teaching hospital that serves over 1.9 million people (6). KCMC is located in Moshi, 128 an urban town of over 200,000 residents situated in Northern Tanzania and bordering Kenya and 129 Kilimanjaro National Park. This study operated specifically within two clinical units at KCMC, 130 the Emergency Department (ED) and the Outpatient Unit for Gynecology, commonly referred to 131 as the Reproductive Health Centre (RHC). 132 For the Kilimanjaro region, KCMC’s ED serves as the referral unit for all injury patients. 133 Injuries have long been associated with excessive alcohol use (30–32). Thus, given the high 134 concentration of injury patients at KCMC’s ED, this clinical unit is ideal for studying risky 135 alcohol users and use behaviors. KCMC’s RHC serves as the referral unit for all women seeking 136 gynecological care within the Kilimanjaro region. The RHC’s large women patient population 137 was chosen to facilitate a deeper understanding of women’s drinking patterns while also 138 providing a safe, gynocentric environment for collecting the sensitive information needed for this 139 study. Data was collected in parallel at KCMC’s ED and RHC throughout the 8-month 140 enrollment timeline. Conducting this research at these locations allowed us to more accurately 141 compare alcohol use behaviors, risky drinking in particular, among both men and women 142 patients. 143 Participants 144 All enrolled participants met the following eligibility criteria: 1) were 18 years of age or 145 older, 2) had the capacity to give informed consent, 3) received initial care at KCMC’s 146 Reproductive Health Center or Emergency Department, 4) were conversant in KiSwahili, and 5) 147 were not prisoners. Capacity to provide informed consent was defined as being medically . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 9 148 stabilized, clinically sober, and well enough to complete the survey verbally on their own. For 149 those who were extremely ill or injured upon initial presentation, the research team re-evaluated 150 the patient within 24 hours of arriving at KCMC or before discharge, whichever came first. 151 Those who remained unable to consent within this time frame were excluded from study 152 participation. As this study was also conducted relatively early in the COVID-19 pandemic, for 153 the safety of the data collection team, patients who tested positive for COVID-19 were also not 154 approached. No women participants presented to both the ED and RHC. 155 Procedures 156 All data were collected in the local language of KiSwahili by a team of three Tanzanian 157 research assistants (two women and one man) who had been hired specifically for work on this 158 study. The final team was chosen based on their expertise in research and strong interpersonal 159 and leadership skills. 160 For data collection, the male research assistant surveyed and interviewed all men 161 participants, and the two female research assistants surveyed and interviewed all women 162 participants. This gender-matching between research assistants and interviewees was done to 163 encourage open and honest reporting of patients’ experiences with or opinions on alcohol based 164 on local culture and research team experience. (33) Prior to data collection, thorough instruction 165 on Good Clinical Practices and extensive study-specific training was provided to the Tanzanian 166 team, including an overview of qualitative data collection methods. All members of the 167 Tanzanian research team were fluent in both English and KiSwahili and held a college degree or 168 higher. The Tanzanian research team held a diverse set of competencies that promoted the 169 collection of robust and reliable data, such as clinical nursing skills and numerous years of 170 previous research experience, including within the field of alcohol-related research specifically. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 10 171 Quantitative Data 172 Sample Size Calculations 173 During the initial study design, it was hypothesized that a final sample size of 587 174 subjects would be required to determine if a significant difference in the prevalence of risky 175 drinking, defined as a score of 8 or greater on the Alcohol Use Disorder Identification Test 176 (AUDIT) scores exists between a) men and women KCMC Emergency Department (ED) 177 patients and b) women patients at the KCMC ED and KCMC Reproductive Health Center (RHC) 178 ( Table 1). From the Emergency Department, 94 participants (47 men and 47 women) were 179 surveyed to compare across genders the proportion of those who screen positively for risky 180 drinking behavior with 80% power and 90% confidence. Likewise, 540 female participants (270 181 from the ED and 270 from the RHC) were surveyed to compare the prevalence of those who 182 were positive for risky drinking behavior between these two units with 80% power and 90% 183 confidence. The final sample size was 587, as survey score data collected from female ED 184 patients were used in both analyses. 185 Table 1: Initial Sample Size Calculations . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 11 186 However, as of early December 2021, two months into the eight-month data collection 187 timeline, the proportion of AUDIT scores ≥ 8 among ED patients was approximately 40% for 188 women and 45% for men, a difference of only 5% as compared to 15% difference upon which 189 the original sample size calculations were based. This meant that to maintain 90% confidence 190 and 80% power, 1,200 patients would need to be enrolled to determine the difference in 191 proportions. While study funds and the data collection timeline limited this doubling in sample 192 size, with IRB amendment approval, the study’s targeted enrollment goals were increased based 193 on the re-estimated prevalence. Thus, enrolling as many patients as was feasible within the 194 original logistical study bounds resulted in a total of 655 total participants by the conclusion of 195 this study timeline. 196 Procedures 197 To collect as representative of a sample as possible in the ED and RHC, a systematic 198 sampling method was employed. Patients seeking care at KCMC’s EMD or RHC were enrolled 199 Monday through Friday from 10:00 am until 6:00 pm local time, with the exception of Tanzanian 200 holidays. While enrollment of women patients was consistent throughout the entire data 201 collection period, enrollment of men patients was paused from January 1st, 2021, until March 202 31st, 2022, pending expanded sample size regulatory approvals. 203 Within the RHC, which sees a large daily volume of women patients, every third patient 204 listed on the general intake registry log was approached and offered study participation. Given 205 the nature of RHC care which often requires follow-up visits, a month into data collection, 206 approximately a fifth of the daily RHC patient population had already been approached for study 207 participation. As such, when sampling individuals from the intake registry, if a woman had . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 12 208 previously been asked about study participation, her name was skipped and every third patient 209 starting from the following individual was approached. 210 At the ED, which sees significantly fewer women patients than men patients, every 211 woman, but every 3rd man on the triage registry was approached. This was done to maintain 212 planned enrollment goals and a representative, systematic random sampling of patients. Each of 213 the three research assistants primarily enrolled one patient population (ED men, ED women, and 214 RHC women). Halfway through data collection (once 135 women patients from both the ED and 215 RHC had been collected), the two women research assistants switched clinical units to minimize 216 any bias that may have arisen as a result of differences in their style of patient interaction or 217 information extraction. 218 Patients were only approached once, and all were given the option to decline participation 219 or terminate their participation early if they chose. All patients were approached in a quiet, 220 private location only once medically stabilized. Here, an overview of the study, including the 221 study goals, procedures, potential risks, and benefits was explained. If, after this discussion, the 222 patient was willing to participate, written consent was obtained. Surveys were administered 223 orally by the same-gender research assistants so that patients of all literacy levels were able to 224 participate and the responses were recorded into a secure Research Electronic Data Capture 225 (REDCap) database. As surveys were collected at a single time point, there were no patients lost 226 to follow-up. In rare instances where survey collection was interrupted, some surveys were left 227 incomplete (n = 23). Incomplete surveys were included in analyses but were not counted as part 228 of the final sample needed for determining differences in prevalence. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 13 229 Instruments 230 Quantitative surveys consisted of five main components: (1) basic demographic data, (2) 231 self-reported alcohol use data, (3) the Alcohol Use Disorder Identification Test (AUDIT), (4) the 232 Drinkers’ Inventory of Consequences (DrInC), and (5) the depression module of the Patient 233 Health Questionnaire (PHQ-9). AUDIT, which ranges from 0 to 40, is a commonly used survey 234 tool for measuring alcohol consumption and alcohol-related problems (34,35). Both locally and 235 globally, patients scoring greater than or equal to 8 are earmarked as clinically significant for 236 harmful or hazardous drinking (HHD) (35–39). Patients with HHD represent individuals whose 237 alcohol intake is detrimental to their physical well-being and require further alcohol-related 238 clinical care and support (40). As such, the prevalence of HHD (defined as AUDIT ≥ 8) was a 239 primary cut-off point in this analysis. DrInC is a 50-question survey (with possible scores 240 ranging from 0 to 50) that measures alcohol-related consequences: in particular, interpersonal, 241 intrapersonal, social responsibility, impulse control, and physical consequences (41). While there 242 is no clinically significant cut-off score, higher scores indicate greater consequences for an 243 individual (42). Finally, PHQ-9 is a diagnostic tool used to identify the existence and severity of 244 depression (43). This scale ranges from 0 to 27, with higher values suggesting increasingly 245 severe depressive symptoms. In the KiSwahili-translated version of the PHQ-9, scores of 9 or 246 greater were found to be the optimal cut-off score for identifying clinical depression (44) and 247 thus was the cut-off point used in this analysis. All three scales had previously been cross- 248 culturally adapted, psychometrically validated, and clinically tested in the local context 249 (39,44,45). 250 All non-survey tool questions were reviewed, revised, translated, and pilot tested by the 251 Tanzanian research team prior to data collection. As this analysis focuses specifically on . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 14 252 patients’ alcohol consumption, DrInC and PHQ-9 scores were not included in this manuscript but 253 will be analyzed in later work. Of note, while this study focuses on gender differences, patients 254 self-identified according to their biological sex. Given that there is little reported gender diversity 255 in Tanzania, for the purposes of this analysis, those identifying as men were categorized as male, 256 and those identifying as women were categorized as female. 257 Analysis 258 Gender differences in alcohol consumption and alcohol-related problems were assessed 259 quantitatively through an exploratory analysis of AUDIT scores, the prevalence of HHD, and 260 demographic and self-reported alcohol consumption data. Consumption was also measured 261 through self-reported alcohol consumption questions that asked participants how much and how 262 often they consumed alcohol, what types of alcohol they preferred, and how much money they 263 typically spent on alcohol per week. All data were analyzed using descriptive frequencies and 264 proportions, and all variables were categorical with the exception of AUDIT scores and HHD 265 status. As alluded to above, AUDIT scores were dichotomized according to HHD status; scores 266 of 8 or greater were classified as ‘HHD,’ while scores less than 8 were ‘not HHD’. Except for 267 age, measures of income, and educational attainment, missing data were minimal for all variables 268 analyzed. The age question was mistakenly omitted in the first several surveys and was added a 269 week into data collection, however for income and educational attainment, participants were 270 more hesitant to disclose this information to research staff. 271 Unlike the qualitative data, all quantitative data were compared across three groups: ED 272 women, ED men, and RHC women rather than by gender alone. This was done to a) identify 273 which clinical unit had the highest incidence of unhealthy alcohol users and b) provide more 274 accurate descriptions of the two women patient populations as the RHC and ED women . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 15 275 populations held significantly distinct demographic and alcohol use-related characteristics. All 276 statistical analyses were conducted in RStudio (version 1.4) using user-created and validated R- 277 Packages. 278 Qualitative Data 279 Sample Size Calculations 280 Of all participants who completed the quantitative survey, a small subset was selected for 281 participation in semi-structured IDIs. 20 (or until saturation was reached) IDIs were originally 282 anticipated to be collected, 10 from ED men patients, 5 from ED women patients, and 5 from 283 RHC women patients to facilitate gender-balanced perspectives on alcohol use. The men ED 284 population reached saturation (defined as the absence of new themes and information following 285 three consecutive interviews) in 9 interviews, thus, 19 IDIs in total were completed. 286 Procedures 287 IDIs were used in this study given the highly sensitive and stigmatizing nature of the 288 study topic, especially for women participants. The one-on-one interview structure encouraged 289 women in particular to share their thoughts more freely and helped to ensure their privacy and 290 confidentiality. IDI procedures were first initiated during survey collection. At the time of the 291 quantitative survey, if a research assistant identified an individual whom they thought would be 292 an excellent candidate for an IDI, they asked if the subject would be willing to participate in an 293 IDI. Those who were invited were purposefully chosen to encompass a diverse range of 294 demographic backgrounds (including age, marital status, education level and occupation, tribe, 295 and religion), perspectives on, and personal experiences with alcohol. IDI participants were also 296 selected to speak to trends related to risky drinking that arose from preliminary quantitative 297 findings. One example of this is that women who were either divorced or widowed were . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 16 298 associated with above-average alcohol intake early on in the data collection period. 299 Subsequently, a woman who had been recently divorced and had a high alcohol intake was 300 purposefully asked to participate in an IDI. To ensure diverse representation and minimize any 301 unintentional bias in sampling, the characteristics of IDI participants were reviewed monthly by 302 the study lead, and any needed changes in the subject sampling were implemented at the next IDI 303 selection. 304 If agreeing, the research assistant obtained the patients’ phone numbers with their consent 305 and scheduled a later time to meet. All interviews were held in private rooms within KCMC and 306 were conducted by a same-gender interviewer who had an established relationship with the 307 patient as they had previously spoken with the patient in-depth during survey collection. The 308 goals of the research study were communicated again before interviews commenced, and a small 309 fund of 5,000 TSH (~2 USD) was given to participants as a transportation reimbursement. All 310 interviews were audio-recorded and generally lasted from 60 to 100 minutes, with a break and 311 snacks offered midway through. 312 Instruments 313 As with the surveys, all interview questions were created in English and then translated 314 into KiSwahili. These translations were then reviewed for appropriate phrasing and syntax, 315 subsequently revised, and then pilot tested by the Tanzanian research team to ensure cultural 316 appropriateness, relevancy, and retention of the original meaning. The interview guide consisted 317 of open-ended questions with built-in probes. Additional probing questions were added on a 318 case-by-case basis by the Tanzanian research team if a participant said something unclear, 319 contradictory, or warranted further explanation. The guide was developed using a team-based 320 approach and was structured and organized across the 6 following domains: (1) effect on and . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 17 321 expectations of the community, (2) men’s use, (3) women’s use, (4) gender differences in use, 322 (5) use during pregnancy, and (6) recommendations for future interventions. The qualitative data 323 in this manuscript pulls primarily from domains 2 through 4, to specifically explore how gender 324 impacts alcohol consumption and use behaviors. Important themes regarding alcohol use and 325 depression arose from the first seven interviews, so several questions related to these concepts 326 were added at this point and were included in all following interviews. With the exception of this 327 addition, the guide remained the same for all interviews. 328 Analysis 329 IDIs were analyzed using an applied thematic, grounded theory approach. (46) As female 330 drinking behaviors in this region have received little prior research attention, a grounded theory 331 approach was best suited for this study as it allowed new themes to arise and be sufficiently 332 explored. A codebook, which was only accessible to the qualitative research team, was 333 developed by the main data analyst based on the first four 4 interviews following a mix of 334 deductive and inductive coding schemes. The initial codebook was discussed with all members 335 of the Tanzanian research team, and changes were made based on the received feedback, 336 ensuring content validity and cultural accuracy. The codebook was used as a dynamic document 337 and was updated as new themes emerged from the data. After revisions to the codebook were 338 made, previous transcripts were revisited and recoded if necessary to encompass newly identified 339 themes that emerged. 340 In partnership with the main analyst, the Tanzanian research team was trained on 341 qualitative analysis and interview coding using NVivo 12. The initial interviews were 342 independently coded in four separate documents by the main analyst and the three members of 343 the Tanzanian research team. These documents were then compared to establish an agreement on . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 18 344 the coding strategy and codebook development. When disagreement arose between researchers, 345 the research team discussed the codes in question until a consensus was reached. This process 346 was repeated until 80% agreement was obtained (47) among the four analysts which occurred 347 after three interviews were coded and reviewed. After a high rate of internal consistency in 348 coding was obtained amongst the 4 initial coders, the primary analyst used the final codebook to 349 code the remaining 16 interviews. The final coding was approved by the analysis team. Content 350 memos were created per each emerging theme and code, summarizing the findings in an ongoing 351 fashion. The content memos served as a basis for discussion and feedback to the entire research 352 team. 353 Research Ethics 354 Prior to data collection, ethical approval was obtained from the Duke University 355 Institutional Review Board, the Kilimanjaro Christian Medical University College Ethical 356 Review Board, and the Tanzanian National Institute of Medical Research. As much as possible, 357 data was maintained in a de-identified manner and shared by data share agreement. Personal 358 health information was used for screening and enrollment, but data were collected, stored, and 359 analyzed in a de-identified manner. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 19 360 Results 361 Quantitative 362 Between October 11th, 2021, and May 31st, 2022, all eligible patients present during 363 study hours were approached, with 655 patients (Table 2) completing the surveys. Few 364 individuals declined study participation; however, as reported by the Tanzanian research team, 365 women were more likely than men to do so, with the primary reasons being that (a) they did not 366 wish to discuss their alcohol use and (b) concern for their privacy. 367 Following the enrollment goals for each of the three patient populations, this study 368 sample was composed primarily of women (82.6%). Most participants were also Christian 369 (80%), employed (57%), and living with a partner either in a registered (50%) or unregistered 370 (12%) marriage. RHC women had the greatest proportion of young patients, with 37% being 371 between the ages of 25 and 34. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 20 372 Table 2: Study Population Demographics Demographics by Population Type Overall, N = 6551 ED Women, N = 2711 RHC Women, N = 2701 ED Men, N = 1141 Age Category, missing: 49 18 to 24 110 / 606 (18%) 47 / 252 (19%) 53 / 249 (21%) 10 / 105 (9.5%) 25 to 34 159 / 606 (26%) 41 / 252 (16%) 92 / 249 (37%) 26 / 105 (25%) 35 to 44 106 / 606 (17%) 42 / 252 (17%) 45 / 249 (18%) 19 / 105 (18%) 44 to 54 94 / 606 (16%) 47 / 252 (19%) 32 / 249 (13%) 15 / 105 (14%) Over 55 137 / 606 (23%) 75 / 252 (30%) 27 / 249 (11%) 35 / 105 (33%) Personal Income Category (TZS per month), missing: 11 0 to 50,000 205 / 644 (32%) 110 / 270 (41%) 68 / 268 (25%) 27 / 106 (25%) 50,001 to 100,000 44 / 644 (6.8% 17 / 270 (6.3%) 15 / 268 (5.6%) 12 / 106 (11%) 100,001 to 150,000 56 / 644 (8.7%) 17 / 270 (6.3%) 29 / 268 (11%) 10 / 106 (9.4%) 150,001 to 200,000 91 / 644 (14%) 34 / 270 (13%) 37 / 268 (14%) 20 / 106 (19%) > 200,000 248 / 644 (39%) 92 / 270 (34%) 119 / 268 (44%) 37 / 106 (35%) Household Income Category (TZS per month), missing: 11 0 to 50,000 62 / 644 (9.6%) 33 / 271 (12%) 16 / 267 (6.0%) 13 / 106 (12%) 50,001 to 100,000 47 / 644 (7.3%) 17 / 271 (6.3%) 14 / 267 (5.2%) 16 / 106 (15%) 100,001 to 150,000 74 / 644 (11%) 32 / 271 (12%) 31 / 267 (12%) 11 / 106 (10%) 150,001 to 200,000 72 / 644 (11%) 37 / 271 (14%) 20 / 267 (7.5%) 15 / 106 (14%) > 200,000 389 / 644 (60%) 152 / 271 (56%) 186 / 267 (70%) 51 / 106 (48%) Religion Christian 522 / 655 (80%) 218 / 271 (80%) 222 / 270 (82%) 82 / 114 (72%) Muslim 121 / 655 (18%) 47 / 271 (17%) 45 / 270 (17%) 29 / 114 (25%) None 11 / 655 (1.7%) 5 / 271 (1.8%) 3 / 270 (1.1%) 3 / 114 (2.6%) Other 1 / 655 (0.2%) 1 / 271 (0.4%) 0 / 270 (0%) 0 / 114 (0%) Highest Educational Attainment, missing: 60 College 170 / 595 (29%) 65 / 242 (27%) 87 / 241 (36%) 18 / 112 (16%) Graduate 13 / 595 (2.2%) 4 / 242 (1.7%) 2 / 241 (0.8%) 7 / 112 (6.2%) None 52 / 595 (8.7%) 30 / 242 (12%) 7 / 241 (2.9%) 15 / 112 (13%) Primary 182 / 595 (31%) 82 / 242 (34%) 61 / 241 (25%) 39 / 112 (35%) Secondary 133 / 595 (22%) 48 / 242 (20%) 64 / 241 (27%) 21 / 112 (19%) Vocational 45 / 595 (7.6%) 13 / 242 (5.4%) 20 / 241 (8.3%) 12 / 112 (11%) . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 21 Demographics by Population Type Overall, N = 6551 ED Women, N = 2711 RHC Women, N = 2701 ED Men, N = 1141 Marital Status, missing: 1 Divorced or Separated 31 / 654 (4.7%) 16 / 271 (5.9%) 6 / 270 (2.2%) 9 / 113 (8.0%) Living with a partner, not in a registered marriage 79 / 654 (12%) 26 / 271 (9.6%) 41 / 270 (15%) 12 / 113 (11%) Living with a partner in a registered marriage 327 / 654 (50%) 128 / 271 (47%) 140 / 270 (52%) 59 / 113 (52%) Never Married or Single 135 / 654 (21%) 55 / 271 (20%) 56 / 270 (21%) 24 / 113 (21%) Refused/Do not know 1 / 654 (0.2%) 0 / 271 (0%) 0 / 270 (0%) 1 / 113 (0.9%) Widowed 81 / 654 (12%) 46 / 271 (17%) 27 / 270 (10%) 8 / 113 (7.1%) Employment Status Employed 371 / 655 (57%) 127 / 271 (47%) 192 / 270 (71%) 52 / 114 (46%) Unemployed 215 / 655 (33%) 111 / 271 (41%) 50 / 270 (19%) 54 / 114 (47%) Student 69 / 655 (11%) 33 / 271 (12%) 28 / 270 (10%) 8 / 114 (7.0%) Tribe Chagga 329 / 655 (50%) 126 / 271 (46%) 146 / 270 (54%) 57 / 114 (50%) Iraq 21 / 655 (3.2%) 9 / 271 (3.3%) 8 / 270 (3.0%) 4 / 114 (3.5%) Maasai 25 / 655 (3.8%) 13 / 271 (4.8%) 7 / 270 (2.6%) 5 / 114 (4.4%) Mmeru 29 / 655 (4.4%) 14 / 271 (5.2%) 11 / 270 (4.1%) 4 / 114 (3.5%) Muha or Non-African 6 / 655 (1.0%) 3 / 271 (1.1%) 3 / 270 (1.1%) 0 / 114 (0%) Nyaturu 17 / 655 (2.6%) 7 / 271 (2.6%) 10 / 270 (3.7%) 0 / 114 (0%) Other African 98 / 655 (15%) 55 / 271 (20%) 34 / 270 (13%) 9 / 114 (7.9%) Pare 70 / 655 (11%) 27 / 271 (10.0%) 25 / 270 (9.3%) 18 / 114 (16%) Sambaa 26 / 655 (4.0%) 7 / 271 (2.6%) 10 / 270 (3.7%) 9 / 114 (7.9%) Sukuma 34 / 655 (5.2%) 10 / 271 (3.7%) 16 / 270 (5.9%) 8 / 114 (7.0%) 1n / N (%); Mean (SD) . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 22 374 Across the three patient populations, the highest average [SD] AUDIT scores belonged to 375 ED men (6.76 [8.16]), followed by ED women (3.07 [4.76]), and RHC women (1.86 [3.46]) 376 (Table 3). ED men also had the highest prevalence of HHD (38%) across all patients, although a 377 significant percentage of ED women (17%) still had AUDIT scores ≥ 8 (Table 3; Figure 3). RHC 378 women had the lowest percentage of individuals with HHD (7.4%) across the three groups 379 (Table 3; Figure 3). 380 In other markers of alcohol use, men continued to score above both women populations. 381 ED men spent the most money on alcohol per week (4.4% of ED men, 0.7% of ED women, and 382 0.4% of RHC women spent between 50,001 to 1000,000 TZS per week), drank in the largest 383 quantities (4.4% of ED men, 0.4% of ED women, and 0% of RHC women drank more than 6 384 standard drinks per sitting), and drank the most frequently (3.5% of ED men drank multiple 385 times per day, but neither ED nor RHC women reported drinking more than daily). Interestingly, 386 while men consumed the most, ED men and ED women answered affirmatively that they have 387 attempted to quit drinking previously in roughly equal (51% for women and 52% for men) 388 proportions, and men were also the most likely (89%) to believe that alcohol use was unhealthy 389 (compared to 71% among ED women and 65% among RHC women). 390 While men had the highest rates of consumption, unhealthy alcohol users were present 391 among both women patient populations. For example, 3.0% and 1.1%, of ED and RHC women, 392 respectively, consumed 5 or more standard drinks in a sitting, and 3.7% of ED women and 1.1% 393 of RHC women reported drinking alcohol every day. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 23 394 Table 3: Alcohol Use Characteristics Alcohol Use Characteristics by Population Type Overall, N = 6551 ED Women, N = 2711 RHC Women, N = 2701 ED Men, N = 1141 Alcohol Preferences Beer 136 / 655 (21%) 66 / 271 (24%) 41 / 270 (15%) 29 / 114 (25%) Changaa, Dadii, Gongo, or Piwa 8 / 655 (1.2%) 5 / 271 (1.8%) 0 / 270 (0%) 3 / 114 (2.6%) Light Beer 60 / 655 (9.2%) 20 / 271 (7.4%) 21 / 270 (7.8%) 19 / 114 (17%) Liquor/Spirits 12 / 655 (1.8%) 3 / 271 (1.1%) 1 / 270 (0.4%) 8 / 114 (7.0%) Mbege (banana-based beer) 83 / 655 (13%) 43 / 271 (16%) 24 / 270 (8.9%) 16 / 114 (14%) Ulanzi (bamboo-based liquor) 5 / 655 (0.8%) 2 / 271 (0.7%) 1 / 270 (0.4%) 2 / 114 (1.8%) Wine 85 / 655 (13%) 30 / 271 (11%) 38 / 270 (14%) 17 / 114 (15%) None 258 / 655 (39%) 100 / 271 (37%) 143 / 270 (53%) 15 / 114 (13%) Other/Refused/Do not know 8 / 655 (1.2%) 2 / 271 (0.7%) 1 / 270 (0.4%) 5 / 114 (4.4%) Drinking Frequency, missing: 2 0 times/week 269 / 653 (41%) 101 / 271 (37%) 146 / 269 (54%) 22 / 113 (19%) 1-2 times/week 289 / 653 (44%) 130 / 271 (48%) 100 / 269 (37%) 59 / 113 (52%) 3-4 times/week 66 / 653 (10%) 28 / 271 (10%) 17 / 269 (6.3%) 21 / 113 (19%) 5-6 times/week 6 / 653 (0.9%) 2 / 271 (0.7%) 2 / 269 (0.7%) 2 / 113 (1.8%) Every day 17 / 653 (2.6%) 10 / 271 (3.7%) 3 / 269 (1.1%) 4 / 113 (3.5%) Multiple times a day 4 / 653 (0.6%) 0 / 271 (0%) 0 / 269 (0%) 4 / 113 (3.5%) Refused/Do not know 2 / 653 (0.3%) 0 / 271 (0%) 1 / 269 (0.4%) 1 / 113 (0.9%) Drinking Quantity, missing: 1 0 drinks 268 / 654 (41%) 102 / 271 (38%) 145 / 270 (54%) 21 / 113 (19%) 1-2 bottles 264 / 654 (40%) 119 / 271 (44%) 87 / 270 (32%) 58 / 113 (51%) 3-4 bottles 97 / 654 (15%) 42 / 271 (15%) 34 / 270 (13%) 21 / 113 (19%) 5-6 bottles 16 / 654 (2.4%) 7 / 271 (2.6%) 3 / 270 (1.1%) 6 / 113 (5.3%) >6 bottles 6 / 654 (0.9%) 1 / 271 (0.4%) 0 / 270 (0%) 5 / 113 (4.4%) Refused/Do not know 3 / 654 (0.5%) 0 / 271 (0%) 1 / 270 (0.4%) 2 / 113 (1.8%) Weekly Alcohol Expenses (TZS), missing: 1 0-10000 531 / 654 (81%) 228 / 271 (84%) 238 / 270 (88%) 65 / 113 (58%) 10001-50000 110 / 654 (17%) 41 / 271 (15%) 29 / 270 (11%) 40 / 113 (35%) 50001-100000 8 / 654 (1.2%) 2 / 271 (0.7%) 1 / 270 (0.4%) 5 / 113 (4.4%) Refused/Do not know 5 / 654 (0.8%) 0 / 271 (0%) 2 / 270 (0.7%) 3 / 113 (2.7%) . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 24 Alcohol Use Characteristics by Population Type Overall, N = 6551 ED Women, N = 2711 RHC Women, N = 2701 ED Men, N = 1141 Attempted Quitting, missing: 1 No 340 / 654 (52%) 132 / 270 (49%) 166 / 270 (61%) 42 / 114 (37%) Refused/Do not know 16 / 654 (2.4%) 1 / 270 (0.4%) 2 / 270 (0.7%) 13 / 114 (11%) Yes 298 / 654 (46%) 137 / 270 (51%) 102 / 270 (38%) 59 / 114 (52%) Reason for Quitting for the 298 patients who Attempted: Family 17 / 298 (5.7%) 9 / 137 (6.6%) 3 / 102 (2.9%) 5 / 59 (8.5%) Financial 24 / 298 (8.1%) 7 / 137 (5.1%) 4 / 102 (3.9%) 13 / 59 (22%) Health 105 / 298 (35%) 52 / 137 (38%) 37 / 102 (36%) 16 / 59 (27%) Other 1 / 298 (0.3%) 1 / 137 (0.7%) 0 / 102 (0%) 0 / 59 (0%) Personal 128 / 298 (43%) 60 / 137 (44%) 49 / 102 (48%) 19 / 59 (32%) Spiritual 23 / 298 (7.7%) 8 / 137 (5.8%) 9 / 102 (8.8%) 6 / 59 (10%) Alcohol Use perceived as Unhealthy, missing: 1 No 179 / 654 (27%) 74 / 270 (27%) 92 / 270 (34%) 13 / 114 (11%) Refused/Do not know 7 / 654 (1.1%) 4 / 270 (1.5%) 3 / 270 (1.1%) 0 / 114 (0%) Yes 468 / 654 (72%) 192 / 270 (71%) 175 / 270 (65%) 101 / 114 (89%) Sought Treatment for Alcohol Use, missing: 3 No 606 / 652 (93%) 251 / 269 (93%) 260 / 269 (97%) 95 / 114 (83%) Refused/Do not know 1 / 652 (0.2%) 1 / 269 (0.4%) 0 / 269 (0%) 0 / 114 (0%) Yes 45 / 652 (6.9%) 17 / 269 (6.3%) 9 / 269 (3.3%) 19 / 114 (17%) Sought Psychiatric Treatment, missing: 2 No 592 / 653 (91%) 245 / 270 (91%) 242 / 270 (90%) 105 / 113 (93%) Refused/Do not know 2 / 653 (0.3%) 1 / 270 (0.4%) 1 / 270 (0.4%) 0 / 113 (0%) Yes 59 / 653 (9.0%) 24 / 270 (8.9%) 27 / 270 (10%) 8 / 113 (7.1%) AUDIT Score 3.22 (5.36) 3.07 (4.76) 1.86 (3.46) 6.76 (8.16) HHD Status (AUDIT >= 8) 109 / 655 (17%) 46 / 271 (17%) 20 / 270 (7.4%) 43 / 114 (38%) 1n / N (%); Mean (SD) 396 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 25 397 Figure 2: Comparison of AUDIT Score Distributions across Patient Populations 398 399 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 26 400 Figure 3: Prevalence of Harmful/Hazardous Drinking (HHD) Across Patient Populations 401 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 27 402 Qualitative 403 Of the 655 individuals surveyed, 19 individuals (RHC women, n = 5; ED women, n = 5; 404 ED men, n = 9) participated in an IDI. Most IDI participants were Christian (78.9%), their ages 405 ranged from 20 to 70 years, and they held a variety of education levels that stretched from 406 primary education to a college degree. Almost half (9 out of 19) of the respondents were living 407 with a partner in a registered marriage at the time of the interview, five were never married, two 408 were divorced or separated, two were living with a partner in an unregistered marriage, and one 409 was widowed. Only one participant was pregnant at the time of the interview, and five 410 participants did not consume alcohol regularly. The intake of the other respondents ranged from 411 1 to 2 bottles 1 to 2 times per week to 3 to 4 bottles 5 to 6 times per week. 412 Alcohol use behaviors differed between men and women across all major themes (how, 413 why, when, where, and what) assessed (Table 4). In general, respondents reported that men had 414 greater agency around their drinking, meaning greater ability and access to control their own 415 drinking behaviors. They also drank more publicly, and in greater quantities with higher alcohol- 416 content drinks. For men, alcohol use was encouraged in social situations and was influenced by 417 the potential for gaining social power or despair over lack of life opportunity. In comparison, IDI 418 participants remarked that most women drank less, consumed lighter beers and wines, and were 419 more restricted as to where and when it was acceptable for them to drink. This restriction was in 420 part because of traditional gender roles keeping women more closely tied to the home and family 421 life, and also the concern for them to incur physical harm while drinking. For women, motives to 422 drink were most closely tied to relationship stress and social pressure. 423 424 Table 4: IDI Alcohol Consumption Themes and Subthemes . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 28 Themes Sub-themes How Consumption Differs Men consume larger quantities of alcohol Traditional gender roles impact alcohol consumption Male drinking is more culturally normalized Motivations to Drink Relationship stress Lack of opportunity Social pressure Social power When Alcohol Intake Occurs The best time to drink is after completing daily tasks Women with families are too busy with household tasks to drink Women with families are encouraged to drink after her kids are asleep Where Alcohol Intake Occurs Men drink in public locations to socialize with others Public drinking for women creates a risk physical harm Women drink privately to avoid social stigma What Types of Alcohol is Consumed Ability of Alcohol to Intoxicate the Drinker Social clout Wealth and socioeconomic status 425 426 How Use Differs by Gender 427 When asked how alcohol use differs between men and women, most IDI participants 428 reported that men drank more than women, a trend that was facilitated by men having greater 429 agency to drink, and men’s drinking being more culturally normalized. 430 Overwhelmingly, respondents agreed that “alcohol use among men is high compared to 431 women” (IDI #12, Male), but with the caveat that especially in recent years, this has been 432 changing; “sometimes women drink more alcohol than men. This is because women have [more] 433 economical power than men nowadays, so are the ones that ruling the world.” (IDI #5, Male) 434 Even still, for most, pre-existing community and familial norms appeared to limit 435 women’s alcohol consumption in comparison to men. Traditional gender roles meant that men 436 had greater freedom to drink alcohol, as explained in IDI #16. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 29 437 438 “Women are obliged to stay at home for a long time for the nurturing and taking care of 439 children and family at large, so it’s not easy for a woman to find time every day to go out 440 with friends to drink alcohol…After [men] come out of job they don’t have a lot to do at 441 home like taking care of kids, they leave that all to women and mostly they instead go out 442 with friends to drink…when a man comes from work he can pass by home and see the 443 family leave some money and off he goes to have a drink outside…men have an ample 444 time to relax, enjoy and that’s when they get to drink alcohol.” – IDI #16, Female 445 446 This sentiment was taken one step further by several male respondents who suggested 447 that because of these responsibilities, some women are denied the option to drink at all; “Woman 448 are not even allowed to drink …because they get drunk easily and, once drunk, they will not be 449 able to do home chores” while “men drink more than women because they are free and have 450 money, so they can buy alcohol whenever they want” (IDI #2, Male). As illustrated by these 451 quotes, in comparison to men, women appeared to lack control over their ability to drink. This 452 lack of agency is also confounded by having fewer funds, less free time, and more intensive 453 home responsibilities. 454 In alignment with men’s alcohol intake being higher, there were more examples of men 455 who displayed dependent or problematic drinking behaviors – men who “can’t be okay without 456 drinking” (IDI #1, Female). Moreso, these examples were largely normalized – drinking “too 457 much… seems normal in the community” (IDI #1, Female). One woman said: 458 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 30 459 “[Men] drink alcohol as routine and it has now become compulsory for some of them to 460 take a drink everyday of their life to complete their daily activities. For example, the men 461 with heavy physical jobs but even doctors and businessmen also take alcohol every 462 morning before going to work. I have witnessed a man who wouldn’t go to his workplace 463 before drinking because otherwise his hands would be shaking and he cannot touch 464 anything and that’s his routine...He was a doctor back in the village I was living.” – IDI 465 #3, Female 466 Motivations to Drink 467 Among men and women, stress was the leading factor in alcohol initiation and 468 progression into unhealthy drinking habits. This was followed by social pressure, and for men, 469 social power. It is important to note that these factors are often intertwined within interviews, for 470 example, social pressures could be seen as a form of relationship stress, and relationship stress is 471 a form of social pressure. 472 Speaking first to stress, many respondents agreed that “most people drink alcohol in the 473 community due to stress…alcohol is no longer considered as a refreshing drink, rather than a 474 stress remover substance!” (IDI #5, Male). The causes of stress usually stemmed from conflicts 475 in relationships – “in town people drink too much because of stress related to love affairs, you 476 find that there is conflict between them, especially cheating; this lead to excessive alcohol use to 477 relieve the stress” (IDI #4, Male) – or financial strain – “life has become so difficult, and people 478 cannot afford life expenses” (IDI #5, Male). 479 For women especially, conflicts in relationships arose as a significant stressor and reason 480 for alcohol initiation. 481 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 31 482 “Where I live so many women live with stress because of being abandoned by men or men 483 are there but they don't take care of the family…something which leads a lot of women to 484 drinking too much alcohol every day…myself I have lived well with my husband but after 485 a long illness he has run away from home and he has left me no money. If I didn't know 486 God and hold onto him, I could end up drinking too much alcohol because of stress until 487 I die” (IDI #13, Female) 488 489 Another individual reported, 490 491 “women drink alcohol to reduce stress when she has arguments with her husband, she 492 decides to drink too much alcohol so that when she comes home, she will not talk to her 493 husband anymore she will only fall to sleep.” (IDI #6, Female) 494 495 The theme of relationship stress was present for males as well, but was not as strong of a 496 factor as it was for females. IDI #7, Female, said “when one has misunderstanding with his wife, 497 he can get drunk to relieve the anger.” 498 Limited career opportunities that led to financial stress and subsequent poor coping 499 strategies, on the other hand, facilitated problematic drinking behaviors primarily among men. 500 This disproportionate effect is in part because men are traditionally seen as the economic 501 providers in Tanzanian culture. 502 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 32 503 “Most young men drink too much alcohol…because they don’t have any future about 504 their life. This is associated with lack of employment, people finishing school but there is 505 not a job to do which makes them frustrated.” (IDI #8, Male) 506 507 A lack of professional opportunity seemed to impact young and older men alike; another 508 participant, IDI #2, a middle-aged male, noted that the previous year he was “preoccupied with a 509 lot of stress” as his tourist company was “not doing well” and his “mom was sick, and there was 510 no money to take care of her.” These stressors led him and other men to drink–what he described 511 as “a poor coping strategy” (IDI #2, Male) – “men drink more than women because they have 512 stress and depression about life challenges” (IDI #8, Male). One participant postulated that men 513 relied on alcohol to relieve stress more than women did because 514 515 “They don’t want to speak about things which troubles them, they take out their 516 problems through drinking alcohol...Men think when they talk about their problems it is 517 a sign of weakness” (IDI #15, Female) 518 519 Social pressure was another key factor driving alcohol use. For women, social pressures 520 included peer pressure from social circles, engaging in drinking behaviors with their husbands, 521 and using alcohol to obtain confidence in social situations. For example, one respondent said “it 522 is [unusual] to find a women drink when the husband doesn’t” (IDI #12, Male). This was 523 expounded upon by another participant, “women drinks when their men drink as well, or 524 sometimes they drink because of peer pressure or company…[they] drink because they want to 525 relax, to gain confidence in social situations” (IDI #11, Male). Further, many women . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 33 526 interviewed expressed sentiments of alcohol allowing them to participate in activities “they can’t 527 do while sober” and “to get rid of shyness” (IDI #3, Female). 528 Beyond stress and social pressure, some men consumed alcohol because it was both a 529 tool for connecting with other men and a symbol of social power, IDI #10, Female, for example, 530 called alcohol use among men “prestigious.” 531 532 “Men drink in order to please their friends or the people around them. If the surrounding 533 people drink he too can drink to feel the sense of oneness with them and to be in one 534 accord with the rest of the men around him.” (IDI #3, Male) 535 536 IDI #15, Female, elaborates on this, saying “there are some girls who prefer to be with a man 537 who drinks alcohol. As I told you a man who drinks alcohol, he is working and can provide for 538 the family…he is husband material.” 539 Alcohol intake helped facilitate social connections for men in part because buying and 540 consuming expensive alcohol was seen as a metric of financial power – 541 542 “Men tend to drink much for showing off to his colleagues and gaining considerable 543 prestige that he is wealthy” as alcohol could cost “up to three hundred thousand shilling 544 per day” (IDI #7, Female) (the equivalent of $130 USD). 545 546 Importantly, this was not the same for women; while women may have felt social 547 pressure to drink, respondents remarked specifically that alcohol did not enable social 548 interactions with other women in the same way that it did for men. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 34 549 550 “Men use alcohol as a social catalyst to help people meet and discuss issues…alcohol is 551 not used among women as a source of them to meet and discuss. Is not as important as it 552 is for men.” (IDI #4, Male) 553 When Alcohol Intake Occurs 554 Overall, respondents reported that the times it was most acceptable to drink were after 555 their daily tasks were completed, which in some cases differed between men and women. 556 Because of the longer working hours of women with families, these individuals faced more 557 restrictions as to when their drinking was seen as appropriate. 558 In general, respondents noted that men prefer to drink in the “evening times to night 559 hours after the work hours are over” (IDI #3, Female), but for those who had heavier alcohol use 560 or are “addicted” (IDI #17, Male) to alcohol,“even in the morning you may find a man is already 561 drunk” (IDI #13, Female). Drinking in the early hours of the day was linked with being 562 unemployed in IDIs for men, for example, “for those who do not have permanent work they 563 drink alcohol from morning they spend all day drinking alcohol” (IDI #6, Female). 564 In contrast, most respondents agreed that women, especially those with families, faced 565 more restrictions on when they should drink. This stemmed from their role as the primary family 566 caretaker, and the reasons underlying this were two-fold: constraints on time, and avoiding 567 drinking in front of children. Many noted that the full-time responsibility of taking care of the 568 house and children left women little time to drink – “for women, this is a challenge to get time 569 free to drink because there are family duties waiting for her to handle” (IDI #3, Female). These 570 restrictions occasionally were in some instances enforced by male respondents, for example, IDI . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 35 571 #12, a male, remarked “I told [my wife] that she should drink at the night when there are no 572 activities at home.” 573 Beyond the time restraint, women were also perceived as drinking alcohol when it was 574 less visible by their family and community. Several remarked that mothers should not drink 575 alcohol in front of their children – “women drink at night when kids are asleep…kids are not 576 supposed to see the mother drunk, it’s a shame for a mother” (IDI #4, Male) – also because “they 577 don’t want kids to develop drinking habits” (IDI #2, Male). One went so far as to say – 578 “[women] are scared to be seen drinking alcohol during the daylight so they use the darkness to 579 their advantage” (IDI #15, Female). Importantly, even for men who had children, these same 580 restrictions were not mentioned by any IDI respondent regardless of gender. 581 These viewpoints on women’s drinking were not held by all, however, one saying that the 582 time when women drink “don’t differ much with men” (IDI #18, Male) and another noting that 583 compared to men “time frame are just the same, from evening to night hours after they are done 584 with work and business” (IDI #16, Female). 585 Where Alcohol Intake Occurs 586 Even more than restrictions on appropriate times, strict rules for where a woman could 587 drink arose among most IDIs. While men could drink anywhere but preferred bars and clubs, 588 because of the potential for stigma and physical harm surrounding women’s public drinking, 589 women predominantly drank in the home due to their household duties. 590 There were few restrictions on where men could drink noted in IDIs – “men are allowed 591 to drink everywhere” (IDI #12, Male), but the “majority…like to go out to places like pubs, bars, 592 and nightclubs” (IDI #3, Female). Almost all participants reported that of all places for men to 593 drink, “most dislike to drink at their homes” (IDI #7, Female), partially to form social . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 36 594 connections but also as an escape from their home life – “especially in marriages, when there’s 595 no peace at home it’s easy for a man to drink too much…he spends so much time in bars to avoid 596 staying at home” (IDI #10, Female). . One participant elaborated on this trend – 597 598 “Men prefer to drinking out of their home environment because the main reason is 599 meetup with friends and discuss business ideas. Just imagine myself I’m always in the 600 office, and when you go back home you meet a very same person that you always stay 601 with, hahaha so it’s better to drink in bars or hotel or another place that sells alcohol 602 where you meet different people and exchange ideas.” (IDI #4, Male). 603 604 Women respondents noted that bars and clubs were also desirable for men as a way to avoid 605 interactions with their wife and meet other women – “if someone has misunderstanding with his 606 wife, he chooses to go bar and drink with other side women in order to get relief” (IDI #7, 607 Female). 608 For women though, the home was mentioned by the vast majority of respondents as the 609 most appropriate place for women to drink – “women drink at home, it’s very rare to find them in 610 bars or hotels drinking beers, morally is not allowed unless they go with her husband or 611 boyfriend” (IDI #4, Male). The remaining few either said that men and women drink in the same 612 places as men (“[women] now days drinks in the bars and pub because they go with their 613 husbands. But in other settings women stays at home and drink alcohol” (IDI #11, Male) or that 614 women should not drink in the home “because kids will be watching you” (IDI #17, Male). . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 37 615 Respondents also said that women drank at home so that “the community will not 616 perceive them badly” (IDI #6, Female), alluding to a greater stigma surrounding women’s 617 alcohol use. 618 619 “Finding a woman in a bar drinking alcohol in the early hours of the morning is not quite 620 appropriate. A man can drink alcohol anywhere but a woman who drinks alcohol where 621 many people can see her, they will look at her as a drunkard.” (IDI #6, Female) 622 623 Rather, in stark contrast to men, the home was the best location for women’s drinking 624 because it is a private, “secret” setting where a woman can be with people “she trusts,” (IDI #3, 625 Female) and likewise, where she can’t be seen: 626 627 “Mostly women drink in hiding areas, [they] tend to hide themselves from either the 628 husband or their relatives, they don’t want others to know if they drink that much. A 629 woman can leave like she is going to the shop to deceive her relative, while she is 630 actually going to drink alcohol in the hiding where no one will notice her.” (IDI #7, 631 Female) 632 633 Another underlying factor as to why women did not drink at bars was the risk incurred on 634 one’s physical safety – “liquor clubs it is not safe for a woman to drink alcohol because when 635 they drink and get drunk, it often happens violence action against women” (IDI #13, Female). 636 Echoing this statement, a female college student reported: 637 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 38 638 “It is a risk for a woman to go to bars and clubs because of her safety, she may fail to get 639 back home, get kidnapped, raped and sexually assaulted and lose her belongings… I 640 have a female friend who drinks so much and that day she went to a club in town and on 641 her way back she was robbed and beaten, and her belongings were taken. So, it is not a 642 safe environment for women.” (IDI #3, Female) 643 What Men and Women Drink and Why 644 Men and women respondents both agreed that women typically prefer “light alcohols 645 with smaller percentages of alcohol” whereas men liked “to drink strong liquors alcohol such as 646 spirits” (IDI #6, Female). These preferences were primarily influenced by alcohol’s ability to 647 intoxicate the drinker, social clout, and cost. 648 An overarching theme impacting men and women’s choice of drink was that “men drink 649 for the aim of becoming drunk” – IDI #12, Male), and since men have a higher alcohol tolerance, 650 men preferred strong spirits so they could “get drunk faster.” One participant elaborated “Most 651 men drinks alcohol in Tanzania, whether they drink on public or secretly, but men drinks…Those 652 who drinks strong beers the main reason is enjoyment and those who drinks local or least costive 653 beers is because of stress or life hardship!” (IDI #4, Male). In contrast, participants responded 654 oppositely for women, noting that women “don’t want to get drunk ” (IDI #14, Male), so they 655 widely gravitated towards lower-percentage and more mild alcohols. These choices for wines 656 and “light beers” helped women to relax while also allowing them to complete their “daily 657 activities at home” (IDI #14, Male; IDI #11, Male). 658 When asked more about why men and women prefer certain beverages, respondents 659 explained the impact drinks had on interpersonal perceptions and social clout. For men, one 660 factor contributing to the preference of stronger alcohols was to show their dominance and . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 39 661 importance within the community – “men drink strong alcohol because want to prove to women 662 that they are superior and above everything (high self-esteem)” (IDI #14, Male). Another echoed 663 this statement, saying “once he uses strong alcohol...people will view him as a civilized and rich 664 person instead for those who drinks mbege and other local beers” (IDI #18, Male). These 665 sentiments on certain alcohol’s social prestige were not mentioned in relation to women. 666 For men too, echoing the motivations to drink, social clout was heavily tied with wealth 667 and socioeconomic status. This in turn impacted the type of alcohol they had the ability to buy – 668 “people with high income drink alcohol of high standards and are expensive but those with low- 669 income drink local brew” (IDI #14, Male). Men living in more rural areas and of lower 670 socioeconomic status for example were often marked as drinking the “local brew” that was 671 widely available, affordable, and lacked testing for alcohol content levels before being sold. One 672 male participant described: 673 674 “Most men drink beer but there are those who drink local brew, so it depends with the 675 income of that particular person. But sometimes even those who drink beers can fall 676 under local brew once they are bankrupt! In the villages the story is different; there 677 they prefer strong alcohol, but locally made because it’s cheap and people have no 678 work to do, you may find them drinking from morning until late nights.” (IDI #5, Male) . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 40 679 Discussion 680 This study aimed to explore gender-based differences in alcohol use characteristics 681 between patients at KCMC's ED and RHC. To the best of our knowledge, this study is the first of 682 its kind to specifically explore alcohol consumption behaviors stratified by gender, both 683 quantitatively and qualitatively, in Moshi, Tanzania. While current literature has investigated 684 alcohol consumption among injury patients in Moshi and has demonstrated gender-specific 685 differences in alcohol use (4,29) as well as the potential influence of gender stigma on risky 686 alcohol behaviors (48–51), our study goes one step further to provide an in-depth mixed-methods 687 analysis on how alcohol consumption and use behaviors differ by gender. Our results indicate 688 that for men only, alcohol use can facilitate social interactions and act as a boon to their overall 689 social status. In contrast, women face greater socially sanctioned restrictions surrounding their 690 drinking and can have less agency than men in their personal alcohol use behaviors. Finally, in 691 light of these social factors, rates of alcohol use among women patients at KCMC EMD were 692 significantly higher than at the RHC, with male patients likewise exhibiting especially high use 693 compared to the surrounding Moshi community. For populations in this region experiencing 694 alcohol-related harm, understanding key sociocultural gender differences in alcohol use can more 695 appropriately inform, target, and tailor future alcohol-related interventions. 696 We found that for men, alcohol was viewed both as a tool for connecting with other men 697 and as a symbolic representation of their social power. This connotation encouraged men to 698 drink, to do so in social settings like bars, and to consume expensive liquors as a marker of their 699 economic status. Two recent studies in Moshi and one in neighboring Uganda (52) have 700 similarly described alcohol as a form of “social currency” and “an important part of social 701 celebrations” (53), however, without delineating whether this association is skewed by gender. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 41 702 This lack of delineation may stem from the aim of these studies which focus on overall drinking 703 culture as opposed to gender differences in drinking. Even still, a finding in Osaki’s work that 704 young men feel greater pressure to initiate alcohol intake from their peers than young women 705 (53) parallels ours. This disproportionate peer pressure on men likely stems from alcohol use is 706 more interlinked to men’s social lives than women’s. 707 While specific data on male drinking and social power in the African context is lacking, 708 this association may find its roots in traditional African cultural practices, where alcohol has 709 been traditionally reserved for high-class men (54). Among the Nyakyusa people of Southern 710 Tanzania (55), for example, men who were older and more socially respected were the primary 711 consumers of beer. The reliance on alcohol as a social tool and symbol among men in Moshi 712 likely contributes to higher continued rates of alcohol use. Concerningly, the fact that alcohol is 713 interwoven into the fabric of male social life may likely make it even more difficult for male 714 problematic drinkers to reduce or abstain from intake in this region. 715 In stark contrast to alcohol’s interconnectedness with male social life, we found that 716 women experience significant social restrictions around drinking and tend to have less agency 717 than men in their personal alcohol use behaviors. Previous studies in Tanzania and neighboring 718 regions have found social influences (12,13) affect alcohol use, but there is a significant dearth 719 of data on how women’s alcohol use is impacted by sociocultural-based gender roles and 720 responsibilities. Our findings did have strong parallels to what relevant research was available, 721 specifically two studies based in Tanzania. Work by Griffin et al found that women who drink 722 experience disproportionate social stigma (49) and literature by Meier et al found that drinking 723 publicly is considered culturally inappropriate for women but not so for men. In addition to 724 social stigma and sanctions, this finding by Meier et al is likely a result of the potential physical . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 42 725 harm women stand the risk of incurring while drinking in public spaces as the link between 726 alcohol use and gender-based violence has been well-established (56,57). Interestingly, our 727 finding stood opposite to Dumbili et al’s work in Nigeria where young women consumed 728 excessive quantities of alcohol to gain social capital (12). This may be because, while on the 729 same continent, Nigeria and Tanzania have distinct cultural differences that social capital is 730 achieved in different manners. 731 An important consequence of the heightened restrictions around women’s drinking is that 732 it leads to secretive alcohol use behaviors, a finding that has a unique clinical implication for 733 women’s healthcare. As we found in this analysis, some women did not want others to see them 734 consuming alcohol which led them to drink at night, in private locations, or only around certain 735 people. This alludes to a wider stigma around women’s alcohol use, which is concerning as 736 stigma has been associated globally (51,58–60) and within Africa (61) as a barrier to effective 737 healthcare service delivery. Because women who drink in Tanzania have been shown to face 738 greater stigma than men, this obstacle in alcohol use-related treatment delivery may be especially 739 prominent for them (49). Thus, women’s secretive alcohol consumption, a behavior born out of 740 stigma and social restrictions around their drinking, will likely make it difficult for healthcare 741 workers to identify, diagnose, and treat women with unhealthy alcohol use. 742 In addition to the sociocultural and economic factors that serve as determinants for 743 alcohol consumption, we found that the highest incidence of alcohol misuse was present within 744 the ED at KCMC. While potentially harmful alcohol consumption was the most pronounced 745 among male ED patients specifically (38%), female ED patients also exhibited higher rates of 746 concerning alcohol use compared to their RHC counterparts (17% ED and 7.4% RHC). 747 Comparing our findings with local estimates, Mitsunga and Larsen found that 7% of women with . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 43 748 partners, 9.3% of women without partners, and 22.8% of men tested positive for alcohol abuse 749 (alcohol abuse here defined as a CAGE score of 2-4) (4). Mushi et al. observed the prevalence of 750 AUDIT ≥ 8 (HHD) to be 23.9% for outpatient primary healthcare patients in Moshi, and when 751 stratifying for gender, 38.7% of males and 13.1% of females tested positive for HHD (29) 752 compared to our findings of HHD among 38% of ED men patients and 17% of ED women. Rates 753 of HHD were comparable among men in these two studies, but for women, a significant 754 difference could be seen. This suggests that KCMC’s ED has a particularly high concentration of 755 women with high-risk alcohol use compared to other populations in Moshi. 756 Our finding of higher rates of alcohol use among men versus women echoes current 757 global trends (13,62,63). The global literature also supports the increased rates of alcohol use and 758 high-risk drinking behaviors present within EDs compared to other hospital departments or 759 wards. This is because of alcohol’s close association with trauma and injuries potentially 760 requiring more immediate medical care (32,64). For those with limited access to care because of 761 financial constraints, EDs may serve as one of the only options to obtain health services for 762 minor alcohol-related consequences. Within the east Africa and LMIC context, previous research 763 has also found elevated rates of alcohol use among ED patients in Tanzania, most often 764 presenting with injuries resulting from road traffic incidents (65). The high rates of harmful 765 alcohol consumption among both men and women ED patients in Tanzania suggest that the ED 766 may be an optimal site for alcohol-related interventions. 767 While alcohol-related harm was not central to this analysis, it is important to emphasize 768 not only the negative impact excessive alcohol use can cause on the Moshi community but also 769 the opportunities for reducing this burden moving forwards. The findings discussed in this 770 manuscript – the high proportion of unhealthy alcohol users in KCMC’s ED, the restrictions . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 44 771 imposed on women’s alcohol use and resulting secretive drinking behaviors, and the association 772 between social life and alcohol use for men – all provide valuable insight that can be used to help 773 shape future alcohol-reduction interventions more effectively. 774 First, our finding that the ED had a significantly high proportion of unhealthy alcohol 775 users, especially women users, suggests that this clinical unit may be a good location in which to 776 base alcohol-related interventions. This is especially important when considering our finding of 777 women’s limited agency and secretive behaviors around their alcohol use. The lack of agency 778 women have around drinking may impede the delivery of alcohol-related treatments. However, 779 an ED-based intervention could help to efficiently target this high-risk group and do so as part of 780 their usual care. Thus, an ED-targeted intervention may provide additional benefits to high-risk 781 women within the community who may face greater social restrictions in accessing alcohol- 782 related health services. Second, the significant dissimilarities in men’s and women’s alcohol use 783 behaviors – from where, when, what, and why they drink – point to the need for creating alcohol- 784 related interventions that are differentiated by gender. For example, as shown in our analysis, the 785 reasons for drinking varied by gender – for men, they were more closely tied to social pressure 786 and lack of opportunity whereas for women it was mostly related to relationship stress. In 787 reducing alcohol intake, the underlying causes of drinking must be addressed, but as this 788 comparison shows, different sociocultural and environmental factors are important for men 789 versus women in their alcohol use. Alcohol-reduction-related interventions and programs that are 790 conscientious of gender differences in alcohol use behaviors may be more able to bring about a 791 lasting change in consumption patterns. 792 Several limitations were present in this study. Despite extensive measures at ensuring 793 confidentiality, patient recall of alcohol use characteristics may have been implicated by recall . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 45 794 bias, whereby subjects could not accurately recount drinking behaviors, or subjects were cautious 795 of exposing their drinking habits to investigators. Moreover, women were more likely to decline 796 study participation, with primary reasons being (a) they did not wish to discuss their alcohol use, 797 and (b) concern for privacy. Consequently, it is important to consider the possibility that our 798 study did not capture the full extent of alcohol consumption among women given the 799 implications of gender stigma present in this context. That is, it may be that women are more 800 reluctant to share the extensiveness of their drinking behaviors in fear of being stigmatized by 801 community, and family members. This may have influenced the accuracy and validity of these 802 results. Additionally, survey and IDI data was obtained from three different clinical settings, 803 warranting replication and external validation. Although missing data was not significant, we 804 must also be weary of the external validity of our results. Subsequently, further studies should be 805 conducted, and replicated at additional clinics, with limited missing data and increased assurance 806 of confidentiality and comfortability. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 46 807 Conclusion 808 Men and women patients at KCMC’s ED and RHC were found to have significant 809 differences in their alcohol use behaviors, including lower rates of consumption among women, 810 greater social restrictions around women’s drinking, and more secretive alcohol use behaviors 811 including where and when women could drink. Conversely, for men, excess drinking was 812 normalized within the Moshi community, tied to men’s social interactions with other men, and 813 generally motivated by stress, social pressure, and despair over lack of opportunity. These 814 dissimilarities point to the need for future alcohol-related programs to incorporate gender in their 815 design and implementation. Our findings can be used to make these programs more effective at 816 reducing alcohol-related harm in the Kilimanjaro region. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 47 817 Bibliography 818 1. WHO. Global status report on alcohol and health 2018. World Health Organization. 2018; 819 2. Shield K, Manthey J, Rylett M, Probst C, Wettlaufer A, Parry CDH, et al. National, regional, and 820 global burdens of disease from 2000 to 2016 attributable to alcohol use: a comparative risk 821 assessment study. Lancet Public Health. 2020 Jan;5(1):e51–61. 822 3. Nadkarni A, Bhatia U, Bedendo A, de Paula TCS, de Andrade Tostes JG, Segura-Garcia L, et al. 823 Brief interventions for alcohol use disorders in low- and middle-income countries: barriers and 824 potential solutions. Int J Ment Health Syst. 2022 Aug 8;16(1):36. 825 4. Mitsunaga T, Larsen U. Prevalence of and risk factors associated with alcohol abuse in Moshi, 826 northern Tanzania. J Biosoc Sci. 2008 May;40(3):379–99. 827 5. Casey ER, Muro F, Thielman NM, Maya E, Ossmann EW, Hocker MB, et al. Analysis of 828 traumatic injuries presenting to a referral hospital emergency department in Moshi, Tanzania. Int J 829 Emerg Med. 2012 Jun 8;5(1):28. 830 6. Staton CA, Msilanga D, Kiwango G, Vissoci JR, de Andrade L, Lester R, et al. A prospective 831 registry evaluating the epidemiology and clinical care of traumatic brain injury patients presenting 832 to a regional referral hospital in Moshi, Tanzania: challenges and the way forward. Int J Inj Contr 833 Saf Promot. 2017 Mar;24(1):69–77. 834 7. Staton CA, Zhao D, Ginalis EE, Hirshon JM, Sakita F, Swahn MH, et al. Alcohol Availability, 835 Cost, Age of First Drink, and Its Association with At-Risk Alcohol Use in Moshi, Tanzania. 836 Alcohol Clin Exp Res. 2020 Nov 10;44(11):2266–74. 837 8. Castens V, Luginga F, Shayo B, Tolias C. Alcohol Abuse in Urban Moshi, Tanzania. 2012; 838 9. Francis JM, Weiss HA, Mshana G, Baisley K, Grosskurth H, Kapiga SH. The Epidemiology of 839 Alcohol Use and Alcohol Use Disorders among Young People in Northern Tanzania. PLoS ONE. 840 2015 Oct 7;10(10):e0140041. 841 10. Wilsnack RW, Vogeltanz ND, Wilsnack SC, Harris TR, Ahlström S, Bondy S, et al. Gender 842 differences in alcohol consumption and adverse drinking consequences: cross-cultural patterns. 843 Addiction. 2000 Feb;95(2):251–65. 844 11. White AM. Gender differences in the epidemiology of alcohol use and related harms in the united 845 states. Alcohol Res. 2020 Oct 29;40(2):01. 846 12. Dumbili EW. “What a man can do, a woman can do better”: gendered alcohol consumption and 847 (de)construction of social identity among young Nigerians. BMC Public Health. 2015 Feb 848 21;15:167. 849 13. Erol A, Karpyak VM. Sex and gender-related differences in alcohol use and its consequences: 850 Contemporary knowledge and future research considerations. Drug Alcohol Depend. 2015 Nov 851 1;156:1–13. 852 14. Nolen-Hoeksema S. Gender differences in risk factors and consequences for alcohol use and 853 problems. Clin Psychol Rev. 2004 Dec;24(8):981–1010. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 48 854 15. Nolen-Hoeksema S, Hilt L. Possible contributors to the gender differences in alcohol use and 855 problems. J Gen Psychol. 2006 Oct;133(4):357–74. 856 16. Agabio R, Pisanu C, Gessa GL, Franconi F. Sex differences in alcohol use disorder. Curr Med 857 Chem. 2017;24(24):2661–70. 858 17. Wilsnack RW, Wilsnack SC, Kristjanson AF, Vogeltanz-Holm ND, Gmel G. Gender and alcohol 859 consumption: patterns from the multinational GENACIS project. Addiction. 2009 860 Sep;104(9):1487–500. 861 18. Cherpitel CJ, Ye Y, Bond J. Attributable risk of injury associated with alcohol use: cross-national 862 data from the emergency room collaborative alcohol analysis project. Am J Public Health. 2005 863 Feb;95(2):266–72. 864 19. Korlakunta A, Reddy CMP. High-risk behavior in patients with alcohol dependence. Indian J 865 Psychiatry. 2019;61(2):125–30. 866 20. Medley A, Seth P, Pathak S, Howard AA, DeLuca N, Matiko E, et al. Alcohol use and its 867 association with HIV risk behaviors among a cohort of patients attending HIV clinical care in 868 Tanzania, Kenya, and Namibia. AIDS Care. 2014 Apr 29;26(10):1288–97. 869 21. Leddy AM, Kerrigan D, Kennedy CE, Mbwambo J, Likindikoki S, Underwood CR. “You already 870 drank my beer, I can decide anything”: using structuration theory to explore the dynamics of 871 alcohol use, gender-based violence and HIV risk among female sex workers in Tanzania. Cult 872 Health Sex. 2018 Dec;20(12):1409–23. 873 22. Chersich MF, Luchters SMF, Malonza IM, Mwarogo P, King’ola N, Temmerman M. Heavy 874 episodic drinking among Kenyan female sex workers is associated with unsafe sex, sexual violence 875 and sexually transmitted infections. Int J STD AIDS. 2007 Nov;18(11):764–9. 876 23. Bonner CP, Browne FA, Ndirangu JW, Howard B, Zule WA, Speizer IS, et al. Exploring the 877 associations between physical and sexual gender-based violence and HIV among women who use 878 substances in South Africa: the role of agency and alcohol. AIDS Care. 2019 Nov;31(11):1369–75. 879 24. Addila AE, Bisetegn TA, Gete YK, Mengistu MY, Beyene GM. Alcohol consumption and its 880 associated factors among pregnant women in Sub-Saharan Africa: a systematic review and meta- 881 analysis’ as given in the submission system. Subst Abuse Treat Prev Policy. 2020 Apr 15;15(1):29. 882 25. Popova S, Lange S, Probst C, Gmel G, Rehm J. Estimation of national, regional, and global 883 prevalence of alcohol use during pregnancy and fetal alcohol syndrome: a systematic review and 884 meta-analysis. Lancet Glob Health. 2017 Mar;5(3):e290–9. 885 26. Matara CM, Winga MA. Sociocultural Determinants of Alcohol Abuse by Youths in Luanda Sub- 886 county, Vihiga County, Kenya. ESJ. 2021 Apr 30;17(14). 887 27. Rehm J, Room R, Monteiro M, Gmel G, Graham K, Rehn N, et al. Alcohol as a risk factor for 888 global burden of disease. Eur Addict Res. 2003 Oct;9(4):157–64. 889 28. Ghebremichael M, Paintsil E, Larsen U. Alcohol abuse, sexual risk behaviors, and sexually 890 transmitted infections in women in Moshi urban district, northern Tanzania. Sex Transm Dis. 2009 891 Feb;36(2):102–7. 892 29. Mushi D, Moshiro C, Hanlon C, Francis JM, Teferra S. Missed opportunity for alcohol use . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 49 893 disorder screening and management in primary health care facilities in northern rural Tanzania: a 894 cross-sectional survey. Subst Abuse Treat Prev Policy. 2022 Jul 6;17(1):50. 895 30. Staton CA, Vissoci JRN, Toomey N, Abdelgadir J, Chou P, Haglund M, et al. The impact of 896 alcohol among injury patients in Moshi, Tanzania: a nested case-crossover study. BMC Public 897 Health. 2018 Feb 21;18(1):275. 898 31. Cherpitel CJ, Bond J, Ye Y, Borges G, Macdonald S, Giesbrecht N. A cross-national meta-analysis 899 of alcohol and injury: data from the Emergency Room Collaborative Alcohol Analysis Project 900 (ERCAAP). Addiction. 2003 Sep;98(9):1277–86. 901 32. WHO. Alcohol and Injury in Emergency Departments. World Health Organization. 2007; 902 33. Meier BJ, El-Gabri D, Friedman K, Mvungi M, Mmbaga BT, Nickenig Vissoci JR, et al. 903 Perceptions of alcohol use among injury patients and their family members in Tanzanian society. 904 Alcohol. 2020 Mar;83:9–15. 905 34. Skipsey K, Burleson JA, Kranzler HR. Utility of the AUDIT for identification of hazardous or 906 harmful drinking in drug-dependent patients. Drug Alcohol Depend. 1997 May 2;45(3):157–63. 907 35. Saunders JB, Aasland OG, Babor TF, de la Fuente JR, Grant M. Development of the Alcohol Use 908 Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons 909 with Harmful Alcohol Consumption--II. Addiction. 1993 Jun;88(6):791–804. 910 36. Piccinelli M, Tessari E, Bortolomasi M, Piasere O, Semenzin M, Garzotto N, et al. Efficacy of the 911 alcohol use disorders identification test as a screening tool for hazardous alcohol intake and related 912 disorders in primary care: a validity study. BMJ. 1997 Feb 8;314(7078):420–4. 913 37. Lundin A, Hallgren M, Balliu N, Forsell Y. The use of alcohol use disorders identification test 914 (AUDIT) in detecting alcohol use disorder and risk drinking in the general population: validation 915 of AUDIT using schedules for clinical assessment in neuropsychiatry. Alcohol Clin Exp Res. 2015 916 Jan;39(1):158–65. 917 38. Conigrave KM, Hall WD, Saunders JB. The AUDIT questionnaire: choosing a cut-off score. 918 Alcohol Use Disorder Identification Test. Addiction. 1995 Oct;90(10):1349–56. 919 39. Vissoci JRN, Hertz J, El-Gabri D, Andrade Do Nascimento JR, Pestillo De Oliveira L, Mmbaga 920 BT, et al. Cross-Cultural Adaptation and Psychometric Properties of the AUDIT and CAGE 921 Questionnaires in Tanzanian Swahili for a Traumatic Brain Injury Population. Alcohol Alcohol. 922 2018 Jan 1;53(1):112–20. 923 40. Reid MC, Fiellin DA, O’Connor PG. Hazardous and harmful alcohol consumption in primary care. 924 Arch Intern Med. 1999 Aug 23;159(15):1681–9. 925 41. Miller WR, Tonigan JS, Longabaugh R. The Drinker Inventory of Consequences (DrInC). 926 American Psychological Association (APA). 1995; 927 42. Forcehimes AA, Tonigan JS, Miller WR, Kenna GA, Baer JS. Psychometrics of the drinker 928 inventory of consequences (drinc). Addict Behav. 2007 Aug;32(8):1699–704. 929 43. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J 930 Gen Intern Med. 2001 Sep;16(9):606–13. 931 44. Smith Fawzi MC, Ngakongwa F, Liu Y, Rutayuga T, Siril H, Somba M, et al. Validating the . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 50 932 Patient Health Questionnaire-9 (PHQ-9) for screening of depression in Tanzania. Neurol 933 Psychiatry Brain Res. 2019 Feb;31:9–14. 934 45. Zhao D, Staton CA, He Q, Mmbaga BT, Vissoci JRN. Cross-Culture Adaptation and Psychometric 935 Properties of the DrInC Questionnaire in Tanzanian Swahili. Front Public Health. 2018 Nov 936 20;6:330. 937 46. McQueen K, Namey E, Guest G. Applied Thematic Analysis. 2014, editor. SAGE Publications 938 Inc; 2014. 939 47. Kurasaki KS. Intercoder Reliability for Validating Conclusions Drawn from Open-Ended Interview 940 Data. Field methods. 2000 Aug 1;12(3):179–94. 941 48. El-Gabri D, Nickenig Vissoci JR, Meier BJ, Mvungi M, Haglund M, Swahn M, et al. Alcohol 942 stigma as it relates to drinking behaviors and perceptions of drink drivers: A mixed method study 943 in Moshi, Tanzania. Alcohol. 2020 Nov;88:73–81. 944 49. Griffin SM, Karia FP, Zimmerman A, Minnig MCC, Swahn M, Makelarski J, et al. A Mixed- 945 Methods Study: Sex Differences in Experiences of Stigma Associated With Alcoholism and 946 Alcohol Use Disorders Among Injury Patients in Tanzania. Alcohol Clin Exp Res. 2020 Aug 947 9;44(8):1700–7. 948 50. Zhao D, Nickenig Vissoci JR, Mmbaga BT, Abdullah AS, Staton CA. Alcohol-Related Risky 949 Behavior Patterns and Their Association With Alcohol Use and Perceived Alcohol Stigma in 950 Moshi, Tanzania. J Stud Alcohol Drugs. 2020 Sep;81(5):614–23. 951 51. van Boekel LC, Brouwers EPM, van Weeghel J, Garretsen HFL. Stigma among health 952 professionals towards patients with substance use disorders and its consequences for healthcare 953 delivery: systematic review. Drug Alcohol Depend. 2013 Jul 1;131(1–2):23–35. 954 52. Ssebunnya J, Kituyi C, Nabanoba J, Nakku J, Bhana A, Kigozi F. Social acceptance of alcohol use 955 in Uganda. BMC Psychiatry. 2020 Feb 7;20(1):52. 956 53. Osaki H, Mshana G, Mbata D, Kapiga S, Changalucha J. Social space and alcohol use initiation 957 among youth in northern Tanzania. PLoS ONE. 2018 Sep 7;13(9):e0202200. 958 54. Heath D. Oshodin OG. Nigeria. In: Heath D, editor. International Handbook on Alcohol and 959 Culture. Westport, Conneticut: Greenwood Press; 1995. p. 213–23. 960 55. Willis J. ‘beer used to belong to older men’: drink and authority among the nyakyusa of tanzania. 961 Africa. 2001 Aug;71(3):373–90. 962 56. Pitpitan EV, Kalichman SC, Eaton LA, Cain D, Sikkema KJ, Skinner D, et al. Gender-based 963 violence, alcohol use, and sexual risk among female patrons of drinking venues in Cape Town, 964 South Africa. J Behav Med. 2013 Jun;36(3):295–304. 965 57. Schulkind J, Mbonye M, Watts C, Seeley J. The social context of gender-based violence, alcohol 966 use and HIV risk among women involved in high-risk sexual behaviour and their intimate partners 967 in Kampala, Uganda. Cult Health Sex. 2016 Jul;18(7):770–84. 968 58. Latoo J, Mistry M, Wadoo O, Islam SMS, Jan F, Iqbal Y, et al. Why mental health service delivery 969 needs to align alongside mainstream medical services. Asian J Psychiatr. 2022 May;71:103053. 970 59. Adams JM, Volkow ND. Ethical imperatives to overcome stigma against people with substance . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 51 971 use disorders. AMA J Ethics. 2020 Aug 1;22(1):E702-708. 972 60. Agterberg S, Schubert N, Overington L, Corace K. Treatment barriers among individuals with co- 973 occurring substance use and mental health problems: Examining gender differences. J Subst Abuse 974 Treat. 2020 May;112:29–35. 975 61. Zewdu S, Hanlon C, Fekadu A, Medhin G, Teferra S. Treatment gap, help-seeking, stigma and 976 magnitude of alcohol use disorder in rural Ethiopia. Subst Abuse Treat Prev Policy. 2019 Jan 977 18;14(1):4. 978 62. World Health Organization (WHO). Global status report on alcohol and health 2018 [Internet]. 979 World Health Organization (WHO); 2018 [cited 2021 Nov 21]. Available from: 980 https://apps.who.int/iris/bitstream/handle/10665/274603/9789241565639-eng.pdf 981 63. GBD 2016 Alcohol Collaborators. Alcohol use and burden for 195 countries and territories, 1990- 982 2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet. 2018 Sep 983 22;392(10152):1015–35. 984 64. Alcohol and Injury in Emergency Departments: Summary of the Report from the WHO 985 Collaborative Study on Alcohol and Injuries. Geneva, Switzerland: World Health Organization; 986 2007. 987 65. Mundenga MM, Sawe HR, Runyon MS, Mwafongo VG, Mfinanga JA, Murray BL. The 988 prevalence of alcohol and illicit drug use among injured patients presenting to the emergency 989 department of a national hospital in Tanzania: a prospective cohort study. BMC Emerg Med. 2019 990 Jan 24;19(1):15. 991 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint 52 992 Supporting Information 993 1. S1 – IRB Approval – Duke University 994 2. S2 – IRB Approval – KCMC 995 3. S3 – IRB Approval – NIMR 996 4. S4 – STROBE Checklist 997 5. S5 – PLOS Checklist . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.12.23289897doi: medRxiv preprint