Epidemiology of Alcohol Misuse and Illicit Drug Use Among Young People Aged 15–24 Years in Fishing Communities in Uganda

Background: We determined the prevalence of and risk factors for alcohol misuse and illicit drug use among young Ugandans in fishing communities, a recognised “key population” for human immunodeficiency virus (HIV) infection. Methods: We conducted a cross-sectional survey among young people (15–24 years) in fishing communities in Koome, Uganda, in December 2017–July 2018. Using Audio-Assisted Self-Interviewing, we collected data on socio-demographic characteristics and alcohol use, including the Alcohol Use Disorders Identification Test (AUDIT) and timeline follow-back calendar (TLFB). Blood samples were analysed for HIV, herpes simplex virus 2 (HSV2), and Phosphatidyl ethanol (PEth 16:0/18:1). Urine samples were analysed for illicit drugs. Results: Among 1281 participants (52.7% male, mean age 20 years), 659 (51.4%) reported ever drinking alcohol, 248 (19.4%) had 12-month-AUDIT ≥ 8, and 261 (20.5%) had whole-blood PEth 16:0/18:1 concentration ≥ 20 ng/mL, indicating significant consumption. In multivariable analyses, PEth 16:0/18:1 ≥ 20ng/mL, AUDIT ≥ 8 and binge drinking (≥6 standard drinks per drinking occasion in the previous month from TLFB) were all strongly associated with older age, low education, smoking, and HSV2. Illicit drug use prevalence was 5.2% and was associated with older age, low education, being single, and smoking. Conclusion: Levels of alcohol misuse were high among young people in fishing communities and associated with HSV2, a proxy for risky sexual behaviour. Alcohol and illicit drug harm reduction services and HIV prevention programs in Uganda should prioritise young fisherfolk.


Introduction
Alcohol and illicit drug use are major global risk factors for disability and premature mortality and contribute to both communicable and non-communicable diseases [1,2]. This health burden is accompanied by significant economic costs, including expenditure on health care and law enforcement, loss of productivity and other direct and indirect costs including harm to others [1,2]. Adolescence is a vulnerability window for initiation and continuation of polysubstance use and associated harms [3]. For instance, globally, the prevalence of current drinking increases from 26.5% among [15][16][17][18][19] year olds to 40.7% among [20][21][22][23][24] year olds, while the prevalence of heavy episodic drinking increases from 13.6% to 21.8% [1]. An estimated 14% of all deaths among 20-39 year olds are attributable to alcohol [4].
The Sub-Saharan Africa (SSA) region has the highest estimated prevalence of heavy episodic drinking per drinker globally (25%), high levels of alcohol related disease and injury, large amounts of unrecorded consumption of illegally produced alcohol [1], and a growing market for alcohol sales [5]. An estimated 5.6% of young people aged 15-19 years (13.8 million) used cannabis in 2015 [4] and there are increasing reports of opioid and injection drug use [6]. Concurrent with the apparent increase in substance use in SSA, human immunodeficiency virus (HIV) remains a major public health problem, with young women aged 15-24 years and people who inject drugs disproportionately affected [7]. In a recent systematic review, we showed that in SSA, information on the burden of alcohol misuse and illicit drug use is limited, especially among young people belonging to certain occupational groups (i.e., sex workers, miners, truckers, fishing communities, uniformed personnel and motorcycle taxi riders) known to be at increased risk of HIV acquisition [8][9][10][11][12][13].
Fishing communities are a recognised key population group in Uganda, and some evidence suggests that as well as being at increased risk of HIV acquisition [10,14], they may also be at increased risk for substance use due to a combination of factors, including occupational norms, high mobility [15,16], geographical remoteness and inadequate substance use regulation [17,18]. However, studies on the burden of alcohol misuse remain few and diagnosis is challenging, as most studies have not used locally validated tools. The aim of this paper is to estimate the prevalence and correlates of alcohol misuse and illicit drug use, respectively, and their association with HIV risk behaviour, among young people in fishing communities on Lake Victoria in Uganda.

Study Setting, Design and Procedures
Between December 2017 and July 2018, we conducted a cross-sectional survey in fishing communities on the Koome Islands in Mukono District, Uganda, with an estimated population of 16,000. HIV prevalence in Koome was around 21% in adults aged 18+ in 2016 [19]. Based on data from a household survey conducted in 2013, as part of the LaVIISWA study [20], within the 15-25 year age group, reported alcohol misuse (defined as consuming at least five alcoholic drinks per day in the last 12 months on average), prevalence was 5%-6%.
The target population was young people aged 15-24 years and the primary outcome was prevalence of alcohol misuse. The target sample size of 1204 young people was calculated to allow us a precision of 2.5% with which to estimate a prevalence of alcohol misuse of up to 15% [19,21], assuming a design effect of 1.5. If the design effect was higher, at 2, then this sample size would allow us precision of 2.5% with which to estimate a prevalence of up to 11%. The target sample size would also provide 80% power to detect a difference in HIV/herpes simplex virus 2 (HSV2) prevalence of 20% versus 8% in young people who misuse alcohol compared to those who do not, assuming that prevalence of alcohol misuse is 12.5%, and the design effect is 2.
We used a two-stage sampling strategy, yielding a self-weighted sample. First, 20 villages were selected from the 27 villages on the islands using simple random sampling. Pre-existing village lists of households (defined as people who sleep in the same house and share meals) were available, but there was no list of individuals to use as a sampling frame. Therefore, we used the households as our sampling unit. We updated the village's lists of households, and then sampled households from each selected village. Each household had a probability of 0.69 of selection; this probability was calculated to achieve the required sample size based on prior knowledge of the approximate number of young people per household. This approach ensured that all households across the study setting had equal probability of selection.
We further assumed that the number of young people per household was consistent across the study setting. We included all young people in the selected households in our survey. This approach yielded a self-weighted sample. Young people aged 15-24 years and resident in the selected households were invited to participate and were enrolled after providing written informed consent. Non-emancipated minors gave assent and their parents/guardians provided parental consent. We excluded severely ill persons.

Data Collection
Survey data were collected by five trained interviewers (three psychiatry clinicians and two experienced non-clinical interviewers), privately, at the community hub or another agreed location. A questionnaire including items on socio-demographic characteristics of study participants, alcohol, tobacco and illicit drug use, depression, and sexual behaviour was developed. It was translated into the local language and back-translated into English separately by two trained translators; versions were compared, and necessary adjustments made to ensure accuracy. Most young people in Koome speak Luganda or English and the ability to do so was an inclusion criterion. All questions were then reviewed by community groups of adolescents from our target population for conceptual understanding and translation accuracy.
The questionnaire was administered using Audio Computer Assisted Self-Interviewing (ACASI). Respondents completed the questionnaire using a tablet and headphones, reading on their own and hearing a recorded voice speak the questions and response options. To build ACASI, both the English and Luganda versions of the questionnaire were audio recorded and programmed into the tablet by the research team, with skip patterns for questions that were not relevant. The questionnaire was pilot tested on 30 young people in a separate village in the same study area. Data collected from the pilot village were not included in this analysis; the results of the pilot were only used to modify survey collection instruments, field procedures, and data management systems, as necessary. During data collection, participants were provided with brief training on how to use the tablets and research staff were stationed nearby to answer questions and provide assistance, as necessary. Study participants were able to complete the interview in the language of their choice.

Measures
Demographic measures included age, religion, education, tribe, village, occupation and parity. Harmful alcohol use was assessed using the following measures. (i) The Alcohol Use Disorders Identification Test (AUDIT)-a 10-item screening tool developed by the World Health Organisation (WHO) to screen for harmful or hazardous drinking. An AUDIT score of ≥ 8 is considered to indicate hazardous or harmful alcohol use. The AUDIT has been validated across genders and in a wide range of racial/ethnic groups and is well-suited for use in primary care settings [22]. (ii) The timeline follow-back (TLFB) tool-a calendar method that can be used as a clinical and research tool to obtain a variety of quantitative estimates of alcohol or illicit drug use patterns over a specified time period that can vary from daily use up to 12 months from the interview date [23]. (iii) Phosphatidyl ethanol (PEth), a more sensitive and specific biomarker compared to traditional biomarkers (gamma-glutamyl transpeptidase (GGT), mean corpuscular volume (MCV) and carbohydrate-deficient transferrin (CDT)), for detecting chronic heavy drinking. PEth is detectable in whole blood for more than two weeks [24], and PEth ≥ 10 ng/mL has been shown to be highly sensitive (88%) and specific (89%) for any alcohol consumption among Ugandan adults living with HIV, in the prior 21 days [25]. Both AUDIT and TLFB tools were administered using ACASI while PEth was determined in dry blood spot (DBS) samples. Alcohol misuse based on AUDIT was defined as a score ≥ 8. Binge drinking based on TLFB was defined as ≥ 6 standard drinks per drinking occasion in the previous month. For this study, illicit drug use is defined as use of a substance where consumption has been prohibited by international drug control treaties except for medical purposes [26], and was assessed using urine tests. We also collected self-reported data on illicit drug use using the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST), a screening tool recommended by the WHO to detect and manage substance use and related problems in primary and general medical care settings; it has been found to be feasible and have adequate test-retest reliability among adult populations in several geographic locations [27][28][29]. Depression severity was measured using the Patient Health Questionnaire (PHQ-9) [30]. Sexual risk behaviour was assessed by HSV2 and HIV results.

Laboratory Testing
To measure PEth, we used ethanol-free swabs for disinfection of the venous puncture site and collected capillary DBS samples. DBSs were dried at room temperature, without any direct sun, for three hours and stored in an airtight mini-grip bag with a drying agent in a fridge at <8 • C for up to 4 days. DBSs were then transported to the MRC/UVRI and LSHTM Uganda Research Unit reference laboratory in Entebbe and stored at −20 • C prior to express delivery on dry ice to the University of Bern, Switzerland, for analysis. DBSs were kept at −20 • C in the Bern laboratory and analysed within five to seven days of arrival to prevent in-vitro formation of PEth. Assays of PEth 16:0/18:1 and 16:0/18:2, the main PEth homologues in human blood [31] were conducted using high-performance liquid chromatography-tandem mass spectrometry (LC-MS/MS), using selected ion monitoring (SIM) in the negative mode of the deprotonated molecules. Here we report on PEth 16:0/18:1. Cut-offs for evaluating PEth 16:0/18:1 values were light or no consumption (< 20 ng/mL), significant consumption (20-209 ng/mL) and heavy consumption (≥ 210ng/mL) [32].
Additional laboratory tests included using a pre-defined nationally approved standard serial testing algorithm for HIV [33]: Alere Determine HIV 1/2 whole blood assay (Alere medical, Chiba, Japan) for screening followed by STAT-PAK rapid test HIV 1/2 (Chembio diagnostic systems New York, NY, USA) as confirmatory test and SD Bioline (Standard Diagnostics, Inc., Kyonggi-do, South Korea) as the tie breaker. For discordant results, confirmatory tests were conducted in an accredited laboratory using ELISA HIV (Murex, Diasorin: Ref 9E25-02, Saluggia VC, Italy) and Bio Elisa HIV1/2 Ag/Ab (BIOKIT, Barcelona, Spain). Newly diagnosed participants were referred for antiretroviral therapy initiation. HSV2 testing was performed at the Entebbe research laboratory using the Kalon ELISA (Kalon Biological, UK) with a cut-off optical density of 1.5. Fresh urine samples collected from all participants were tested at the study site for illicit drugs using a the (one-step rapid test, Protzek (Lörrach, Baden-Württemberg, Germany). This multiplate test is a sensitive immunochromatographic test based on a specific antigen-antibody reaction for the qualitative detection of amphetamines, methamphetamines, cocaine, ecstasy, opiates, and cannabinoids in human urine.

Data Management and Statistical Analysis
Data were captured electronically using an open data kit (ODK) and checked for missing fields and erroneous inputs. All statistical analyses were conducted using Stata, version 15 (Stata Corp, College Station, Texas, USA). We summarised participants' socio-demographic characteristics and conducted univariable and multivariable analyses to identify factors associated with alcohol misuse (four separate outcomes: AUDIT ≥ 8, PEth ≥ 20ng/mL, PEth ≥ 210ng/mL, binge drinking based on TLFB) and illicit drug use. Multivariable analysis to control for potential confounding was guided by a hierarchical conceptual framework (Figure 1), following the approach described by Victora et al. [34]. We classified each variable into one of three groups (levels) from most distal to most proximal to the outcomes of interest for this analysis (alcohol misuse and illicit drug use). Level 1 variables were those determined at birth or during childhood, level 2 variables comprised current socio-demographic variables, and level 3 variables comprised current behaviour factors and infections. Level 1 variables were thought likely to impact on the outcomes through their effects on level 2 and hence on level 3 variables. Therefore, when assessing the effect of level 1 variables on the outcomes, we did not adjust for level 2 or level 3 variables in order to avoid adjusting for factors on the causal pathway. Similarly, when assessing the effect of level 2 variables on the outcomes, we did not adjust for level 3 variables. In summary, variables at the first level were adjusted for each other, variables at the second level were adjusted for each other and for first level variables, and third level variables were adjusted for each other and for first and second level variables. The only exception to this was that results for HIV were not adjusted for HSV2 and vice versa, since the two were both considered proxies for sexual behaviour and are highly correlated. Variables were retained in models regardless of their p-value after adjustment following this strategy; they were not excluded on the basis of high p-values. This strategy allowed us to assess the effects of variables at each level of the conceptual framework, having adjusted for more distal variables [34]. We used a two-stage sampling strategy and the sample was self-weighted. We accounted for clustering at the village and household level: logistic regression with a robust variance implemented using the svyset command in Stata was used to assess risk factors for alcohol misuse and illicit drug use. . Any participants requiring medical treatment were referred to nearby government health centres where free treatment was offered.

Risk Factors for Substance Use
Supplementary Table S1 shows that factors associated with hazardous drinking (PEth 16:0/18:1 ≥210ng/mL) at multivariable analysis included male gender, older age group 20-24 and low education. Table 4 shows the relationship between the characteristics and binge drinking in the past month ascertained using TLFB. In crude analysis, binge drinking was associated with all characteristics except tribe, occupation, duration in community, exposure to school curriculum on alcohol, and depressive symptoms. After adjusting for potential confounders, binge drinking was associated with older age, low education, being single, parity, exposure to alcohol adverts, and smoking, illicit drug use and HSV2, with weak evidence for association with mild depressive symptoms. Table 5 shows the relationship between the examined characteristics and illicit drug use (positive urine test) in adolescents. In univariable analysis, the prevalence of illicit drug use was associated with gender, age, education, income, occupation, smoking and depressive symptoms. After adjustment for potential confounders, illicit drug use remained positively associated with male gender, older age group, lower education, being single, smoking, and moderate to severe depressive symptoms.

Discussion
Alcohol misuse was common among young people aged 15-24 years in fishing communities in Uganda, and was consistently associated with older age, male gender (except for AUDIT ≥ 8), lower education, smoking, HSV2 and depressive symptoms, regardless of the alcohol use assessment tool used. The prevalence of illicit drug use was lower than that of alcohol misuse. The only drugs detected were cannabinoids and amphetamines. Illicit drug use was associated with male gender, older age, lower education, being single, smoking, and depressive symptoms.
There are limited data on alcohol misuse and illicit drug use from similar settings [8]. The prevalence of alcohol misuse in our study was comparable to findings in a recent study among young people in a general population setting in Northern Tanzania, with 11%-28% of males screening positive for alcohol use disorder (AUD) [35], and findings among young people in Goa, India (17.7%) [36]. However, the prevalence of AUD in our study was much lower than the prevalence (53%, measured by CAGE) among 14-24 year old female sex workers in Kampala [37], as well as in another study of a key population where, among 1476 young men who reported sex with men in Kenya, 44% had reported AUD via the AUDIT questionnaire [38]. The prevalence of binge drinking in our study was slightly lower than that reported among young people in the WHO Africa region comprising 47 countries [1]. Young men were more likely to screen positive for both AUD based on PEth and binge drinking, and this is consistent with estimates from Africa and other WHO regions [1], although interestingly, AUDIT ≥ 8 was not associated with gender in this setting. Taken together, these studies further highlight the need for integration and prioritisation of alcohol reduction in HIV prevention interventions among key populations.
The prevalence of illicit drug use was 4.4%, similar to a large population-based survey in South Africa among people aged 15 and older [39]. The most commonly detected drug was cannabis, and this might be attributed to better availability and social acceptability, or lower cost compared to other illicit drugs.
Globally, predictors of harmful alcohol use and illicit drug use among young people include male gender, peer pressure, family experience, psychological factors/mental health, positive expectations regarding alcohol use, violence, socio-economic factors, and increased alcohol and illicit drug availability and marketing [35,40,41]. Furthermore, young people are potentially more likely to engage in reckless behaviour while drunk [42] or whilst on drugs than older adults. In our study, socio-demographic characteristics associated with AUD were similar to those found in studies conducted in more general population settings elsewhere in East Africa and Europe/US. They included male gender, older age group (20-24 years), using illicit drugs and HSV2 infection [37]. However, other studies among female sex workers reported contrary risk factors such as younger age [37]. The findings emphasize the need for integration of alcohol and illicit drug use prevention interventions, and for interventions that will be able to reach males.
Extensive evidence shows that alcohol misuse and illicit drug use are associated with higher HIV risk at societal, community and individual levels , stemming mainly from associations with high risk sexual behaviour, increased HIV shedding and inflammation at mucosal sites [43] and needle sharing for injection drug users [44]. Among people living with HIV, alcohol and illicit drug use may adversely affect disease progression through several mechanisms, including reduced uptake of HIV prevention and care services [45], and through impacts on viral replication, host immunity and treatment compliance and efficacy [43]. Our cross-sectional study did not show a significant association between HIV and AUD or illicit drug use after adjustment for potential confounders, although adjusted odds ratios were greater than 1 for all measures of AUD, and lower limits of 95% confidence intervals only marginally overlapped 1. The lack of significant relationships between these variables may be in part due to the relatively small number of HIV events and consequent limited power to detect an association. However, they should not be interpreted as evidence for the absence of such relationships. There was a strong association between substance use and HSV2. Evidence suggests a direct and reciprocal biological interaction between HIV and HSV2 [46]. Additionally, HIV and HSV2 are associated with similar risk factors [age, sex, partner change, condom use and male circumcision) [46]. Thus, HSV2 may be considered a proxy measure for risky sexual behaviour. Therefore, prioritizing alcohol interventions and harm reduction among young people in key population settings is critical not only to reducing substance-use-related harms but also to reducing sexually transmitted infections and could accelerate progress towards the UNAIDS 95-95-95 HIV epidemic control targets. Our study documents both under-age alcohol use and misuse (the legal age for drinking in Uganda is 18). This implies that additional research to understand the drivers of substance use among minors and interventions to address them is critical.

Strengths and Limitations
A strength of the study was the large sample size, which meant that we had good power to detect associations with alcohol misuse for most covariates of interest. We selected a population-based sample and the response rate was high; thus, our findings may be generalisable to similar fishing community settings in the Eastern and Southern Africa regions, although we did not collect detailed information on those who did not participate to further assess this. Another strength is that we used a range of standard tools to assess alcohol use, encompassing the different types of alcohol consumption measures, i.e., tools that report alcohol consumption status, average volume alcohol consumption, and frequency and volume of binge drinking. We identified several key covariates that were strongly associated with alcohol use across all three measures used. These standard tools are essential for monitoring public health and evaluating alcohol control policies and other interventions.
Our results might be less prone to social desirability bias/misreporting, which could bias our estimates of association either towards or away from the null, for two reasons: (i) use of objective measures to supplement self-reported measures of alcohol and illicit drug use and sexual risk behaviour; (ii) administration of ACASI, which was acceptable and feasible. Previous literature suggests that ACASI allows more accurate reporting of sensitive behaviours including alcohol (mis)use. However, we did not have an interviewer-administered comparison group to test this hypothesis in our study. Therefore, our findings might not be generalizable to AUDIT being administered face-to-face. We further minimised reporting bias by using young gender-matched interviewers, and by using a customised TLFB calendar and pictorial displays of standard alcoholic drinks [35]. Our survey had a response rate (based on residents listed in a previous census) of approximately 80%, therefore we cannot exclude the possibility that our results are subject to selection bias. Non-response, especially in a study of a controversial topic such as substance abuse or sexual behaviour, could potentially introduce selection bias and consequently lead to underestimation of outcome and risk factor prevalence [47], and may affect the internal validity of the study [48]. Interpretation of the effect sizes found in this study should be undertaken with caution, as these may depend on the specific characteristics of the samples included in our study. A further limitation of our study was the cross-sectional design, which precludes us from demonstrating causality between independent variables and AUD and or illicit drug use, and between AUD and HIV or HSV2. We chose to model HIV and HSV2 as risk factors for alcohol misuse and illicit drug use since alcohol misuse and illicit drug use were our primary outcomes of interest, but reverse causation is a plausible explanation for the associations seen.

Conclusions
Alcohol misuse and illicit drug use are common among young people in fishing communities in Uganda and associated with sexual risk behaviour and smoking. Uganda is an HIV-hyperendemic country, with one of the youngest and fastest-growing populations and one of the highest estimated alcohol per capita consumption rates globally [1]. There is potential for additional benefit of combined prevention and intervention activities targeting alcohol and illicit drug use, and HIV particularly, in key population settings.