Sociodemographic correlates of substance use and psychosis in a large-scale african cohort

Among people with psychotic disorders, substance use (SU) and substance use disorders (SUD) are highly prevalent. Studies suggest that the high rates of substance use are moderated by demographic characteristics such as sex, age, marital status and location; however, there is limited data on substance use among people with psychosis in Africa. We described patterns of substance use across South Africa, Ethiopia, Kenya and Uganda for four major classes of drugs (alcohol, tobacco, cannabis and khat) across three primary diagnostic categories: bipolar disorder (BD), schizophrenia (SCZ) and schizoaffective disorder (SAD). We utilized data from the Neuropsychiatric Genetics of African Populations-Psychosis project, a large case-control study which will soon have genetic data on over 42,000 participants, half with psychosis. Substance abuse information was collected using the Alcohol, Smoking and Substance Involvement Screening Test v3 (ASSIST). Lifetime and frequency of substance use for alcohol, tobacco, cannabis and khat (a stimulant used primarily in East Africa) and outcomes were defined as regular (daily or weekly use) or irregular (monthly or less, including never) use of an individual substance. Each substance was modeled individually as a dependent variable in a logistic regression model adjusting for demographic factors of sex, education, and country. Stratified analyses were performed to assess country and sex-specific effects. We included 16,508 individuals (42% female) with primary diagnoses of BD, SCZ or SAD, and age ranging from 18 to 83 (mean: 36). We found evidence to suggest differences in regular substance use across the diagnostic categories. Individuals with BD show a greater odds of regular consumption of alcohol and cannabis than those with SCZ; (OR=1.2; p=0.02, and OR=1.7, p=0.001, respectively). Men showed a higher consumption of all substances as compared to women, all but khat reaching statistical significance (alcohol: OR=1.89; p=1.7 × 10-11, tobacco: OR= 2.26, p= 1.6 × 10-13, cannabis: OR=1.38, p=1.9ⅹ10-9, khat: OR = 4.60, p=0.36). Country specific differences were found across all substances, with alcohol being the predominant in Uganda (6%) and Kenya (6%) and cannabis (11%) in South Africa and khat in Ethiopia (14%). We found no relationship between level of education and frequency of substance consumption. Country and sex specific differences in substance use behaviors exist in a large-scale African cohort of people with psychosis. Our findings are in line with previous work regarding sex and regional differences; however, we differ from studies conducted in US populations in that we found no relationship between level of education and frequency of substance use for any of the substances studied. This suggests that there may be distinct sociodemographic correlates of substance use in Africa and highlights the critical need to consider individuals of diverse ancestry in large-scale studies - taking into account regional differences when considering genetics of substance abuse.