Search results
The search from the five electronic databases and yielded 1535 results: 950 from PubMed, 173 from PsychINFO, 210 from web of science, 123 from CINAHL and 79 from Cochrane library. A further 13 studies were identified through a manual search of the AJADA journals (Volumes 1, 2 and 3). Studies were assessed for duplicates and 1154 articles remained after removal of duplicates. These underwent an initial screening based on abstracts and titles, and 946 articles were excluded. A second screen of full text articles was done for the 208 studies that were potentially eligible for the review. Twenty three studies were excluded as follows: 21 did not meet the eligibility criteria and 2 had duplicated results. A total of 185 studies were found to meet the inclusion criteria and were included in the review (Fig 1).
General description of studies
Of the 185 studies included in this review, 144 (77.8%) investigated the epidemiology of substance use/SUD (Additional file 3), 18 (9.7%) evaluated substance use/SUD interventions and programs (Additional file 4), and 23 (12.4%) were qualitative studies exploring perceptions on various substance use/SUD topics (Additional file 5). The studies were published between 1982 and 2020. The number of studies published has gradually increased in number over the years, particularly in the past decade. The graph below shows the trends in publications on substance use in Kenya.
Quality assessment:
The QATSDD scores ranged from 28.6% (21) to 92.9% (22). Only 14 studies (10,21,23–34) (all quantitative) had scores of less than 50%. Of these, the main items driving low quality were: no mention of user involvement in study design (n=14) (10,21,23–34) , no explicit mention of a theoretical framework (n=10) (10,21,23,25–30,32) and a lack of a statistical assessment of reliability and validity of measurement tools (n=10)(10,21,23,25–27,30,32–34) (Additional file 6).
Studies examining the epidemiology of substance use/SUD
Description of epidemiological studies: 144 studies examined the prevalence and or risk factors for various substances. The studies were published between 1982 and 2020. The 4 main study designs were cross-sectional (n=126), cohort (n=5), case-control (n=10), and mixed methods (n=2). One study used a combination of the ‘multiplier method’, ‘Wisdom of the Crowds’ (WOTC) method and a published literature review to document the size of key populations (35). The sample size for this category of studies ranged from 42 (36) to 72292 (37).
The studies were conducted in diverse settings including the community (n=72), hospitals (n=41), institutions of learning (n=24), streets (n=5), prisons and courts (n=3), in charitable institutions (n=1), and in needle-syringe program (NSP) sites (n=1). Of the studies conducted within the community, 12 were conducted in informal settlements. The study populations were similarly diverse as follows: general population adults & adolescents (n=39), persons with NCDs (n=11), primary and secondary school students (n=15), people with injecting drug use (PWID) (n=11), general patients (n=5), Men who have Sex with Men (MSM) (n=8), university and college students (n=9), commercial sex workers (n=7), psychiatric patients (n=6), orphans and street connected children and youth (n=6), people living with HIV (PLHIV) (n=6), healthcare workers (n=3), law offenders (n=3), military (n=1), and teachers (n=1). Only one study was conducted among pregnant women (38).
Sixty nine studies (47.6%) used a standardized diagnostic tool to assess for substance use. The AUDIT (n=21) and the ASSIST (n=10) were the most frequently used tools. Most papers assessed for alcohol (n = 109) and tobacco (n = 80). Other substances assessed included amphetamines (n = 41), opioids (21), sedatives (n=19), cocaine (n=19), inhalants (n=16), cannabis (n = 14), hallucinogens (n=7), prescription pills (n=2), emerging drugs (n=1) and ecstasy (n=1).
Key findings on prevalence rates: One studywith the largest sample size (n=72292) drawn from the community, reported the lifetime prevalence of tobacco smoking among adults as 11.2% and that of alcohol use as 20.7% (37). Using the Alcohol Use Disorder Identification Test (AUDIT), the 12 month prevalence of hazardous alcohol use ranged from 2.9% among adults drawn from the community (39), and 64.6% among female sex workers (FSW) (40). Based on the same tool, the lowest and highest rates of harmful alcohol use were both reported among FSWs i.e. 9.3% (41) and 64.0% (42) respectively, while the prevalence of alcohol dependence ranged from 8% among FSWs living with HIV (43) to 33% among MSM who sold sex (44).
Based on the Alcohol, Smoking & Substance Use Involvement Screening Test (ASSIST) questionnaire, the lifetime prevalence of tobacco use ranged from 23.5% among health care workers (HCWs) (45) to 95.7% among university students (46), that of khat use ranged from 11.5% among university students (47) to 55.2% among psychiatric inpatients (48), and that of cannabis use from 21.3% among persons with AUD (49) to 64.2% among psychiatric in-patients (48). The lifetime prevalence of opioid use ranged from 1.1% among PLHIV (50) to 8.2% among psychiatric in-patients as assessed using the ASSIST (48).
Key findings on associated/risk factors: Among children, youth and adolescents, substance use was associated with being male (32), engaging in sex (51), older age (52,53), being in a private school, living in an urban area (53), having a family member who uses substances (52), depression (38,54), suicidal behavior (55) and Human Immunodeficiency Virus (HIV) infection (56). Early substance use among this population was linked to ever engaging in sex, higher education, parental or guardian substance use, and suicidal ideation (57). Factors associated with multiple substance use were living in an urban area and being female (58).
Among adults, alcohol use was associated with several socio-demographic factors including being male (45,59–66), being unemployed (61), being self-employed (62), having a lower socio-economic status (SES) (37), being single or separated, living in large households (62), having a family member struggling with alcohol use and alcohol being brewed in the home (67). Alcohol use was linked to various health factors including glucose intolerance (68); poor cardiovascular risk factor control (69), having a diabetes mellitus diagnosis (70), hypertension (65,71), default from tuberculosis (TB) treatment (72), depression (49), psychological Intimate Partner Violence (IPV) (73), physical and sexual violence (74), tobacco use (66,73), and increased risk laryngeal (75) and esophageal cancer (76,77). Finally alcohol use was associated with involvement in Road Traffic Accidents (RTAs) (78,79), having injuries (78,80,81), and with having concurrent sexual relationships (66).
Tobacco use among adults was associated with being male (45,59,65). Several health factors were linked to tobacco use including hypertension (65), development of oral leukoplakia (34), pneumonia (82), increased odds esophageal cancer (77), ischemic stroke (83) and diabetes mellitus (70). In addition, tobacco use was associated with having had an injury in the last 12 months (81), emotional abuse (48) and psychological IPV (73). Longer duration of smoking was associated with a diagnosis of diabetes mellitus (84), lower SES (37), and hypertension (85,86).
Two case-control studies documented increased odds of reporting psychotic symptoms (87,88), and PTSD (Post-Traumatic Stress Disorder) symptoms (88) among khat users compared to non-users, while IDU was associated with depression, risky sexual behavior (89), Hepatitis-C Virus (HCV) infection (90), and HIV-HCV co-infection (91).
Other topics explored:In addition to prevalence and associated factors, these studies explored other topics including agreement between self-reported alcohol use and the biomarker phosphatidyl ethanol (92), primary health care workers self-efficacy for SUD management (93), reasons for substance use (27,30,94,95), and tobacco quit intentions(96). Papas et al. (92) reported a lack of agreement between self-reported alcohol use and the biomarker phosphatidyl ethanol among PLHIV with AUD. Among primary health care workers, self-efficacy for SUD management was lower in those practicing in public facilities and perceiving a need for AUD training. In that study, higher self-efficacy correlated with a higher proportion of patients with AUD in one’s setting, access to mental health worker support, HCW’s cannabis use at a moderate risk level, and belief that AUD is manageable in outpatient settings. Common reasons for substance use included leisure, stress and peer pressure among psychiatric in-patients (30), curiosity, fun, and peer influence among college students (94); peer influence, idleness, easy access, and curiosity among adults in the community (27); and peer pressure, to get drunk, to feel better and to feel warm among street children (95). Atwoli et al. 2011 (97) reported that most students were introduced to substances by friends while Astrom et al. 2004 (98) reported that HCWs, parents and school teachers were not discussing tobacco harms with youth. Kaai et al. 2019 (96) conducted a study regarding quit intentions for tobacco use and reported that 28% had tried to quit in past 12 months; 60.9% had never tried to quit, only 13.8% had ever heard of smoking cessation medication. Intention to quit smoking was associated with being younger, having tried to quit previously, perceiving that quitting smoking was beneficial to health, worrying about future health consequences of smoking, and being low in nicotine dependence. A complete description of the prevalence studies has been provided in additional file 3.
Studies evaluating substance use/ SUD programs and interventions
A total of eighteen studies evaluated specific interventions or programs for the treatment and prevention of substance use (see Additional file 4). These were carried out between 2009 and 2020. The studies used various approaches including randomized control trials (RCT) (99–105) mixed methods (106–108), non-concurrent multiple baseline design (109), quasi experimental (110), cross-sectional (111,112), and qualitative (113–115). One study employed a combination of qualitative methods and mathematical modeling (116).
11 studies investigated feasibility, acceptability or efficacy of individual-level treatment or prevention interventions using various outcomes. The interventions evaluated included cognitive behavioral therapy (CBT) (101,106), motivational interviewing (MI) (99), a combination of behavioral activation, MI and gender norm transformative strategies (109), World Health Organization (WHO) brief intervention (100,102,103,107,110), psychoeducation sessions (105), and contingency management (104). Only one study evaluated family outcomes (109). All interventions were tested among adult populations. These included persons attending a Voluntary Counseling & Testing (VCT) center (107), PLHV (101,106), college students (105), FSWs (100,102), and adult males and females (103,109,110) drawn from the community.
The interventions were delivered using various approaches including trained lay providers (101,106,109,110), digital health means (99), and trained primary care workers such as nurses (100,102), and VCT service providers (Mackenzie 2009). The number of sessions ranged from one (103,107) to six (100–102,106). Most of the individual level interventions targeted harmful alcohol use (n=9) (99–102,104,106,107,109,110). One study targeted khat use among men (103), while another the use of multiple substances (105). All interventions had a positive impact on substance use except the study that used the contingency management approach (104). The interventions were delivered in various settings as follows: community settings (n=6) (99,102–104,109,110) of which one was delivered in a HIV prevention drop in center (102), HIV treatment or testing out-patient clinics (n=3) (101,106,107), alcohol and drug abuse rehabilitation facility (n=1) (100), and within a college (n=1) (105). One study described the process of culturally adapting a CBT intervention for alcohol use, for use among a population of PLHIV (106).
Five studies evaluated various aspects of substance use treatment programs. The studies evaluated perceptions on benefits of methadone programs (113,116) and NSPs (115), healthcare workers knowledge and practices on tobacco cessation (112), and utilization of community based outpatient SUD treatment services (111). The methadone and NSP programs were perceived as beneficial and impactful by stakeholders and service users (113,115) and knowledge and practice on tobacco cessation as inadequate (112). Deveau et al., 2010 (111) reported a 42% abstinence rate 0-36 months post-treatment termination.
Two studies evaluated population-level interventions. One evaluated the appropriateness and effectiveness of HIC anti-tobacco adverts in the African context and found the adverts to be effective and appropriate (108). Another study that examined community member’s perspectives on the impact of the government’s public education messages on alcohol abuse reported the messages as ineffective and unpersuasive (114). A complete description of included intervention studies is in additional file 4.
Studies qualitatively exploring various substance use/SUD topics (other than interventions)
There were 23 qualitative studies included in our review. The studies were conducted between 2004 and 2020. Data was collected using several approaches including in-depth interviews (IDIs) only (n=6) (117–121), focus group discussions (FGDs) only (n=2) (122,123), a combination of FGDs and IDIs (n=10) (114,124–132), a combination of observation and individual IDIs (120,133), a combination of observation, IDIs and FGDs (134), a combination of literature review, observation, IDIs and FGDs (132). One study utilized the participatory research and action approach (135).
The study populations for the qualitative studies included persons using heroin (129,133), males and females with IDU (116,119,124,125,127,136), community based organizations (114,123), youth (120,131,134), FSWs (118,121), refugees and Internally Displaced Persons (IDPs) (132), and PLHIV (122,135).
Various themes were explored in these qualitative studies including risk and protective factors for substance use (120,127,131,134), health and or socio-economic effects of substance use (119,122,123,130,132), perceptions on heroin use (133), transitions from heroin smoking to injection (117), and stages of change in participants enrolled in an intervention (121). Substance use was perceived as having a negative socio-economic and health impact (122,132). Specifically, substance use was perceived to have a negative impact on contraceptive use (128), on utilization of antenatal and maternal & child health services (137), as well as on sexual and reproductive health (130). In addition, substance use emerged as a driver of risky sexual behavior and HIV among both adults (118,138) and adolescents (120,134). Several factors were perceived to contribute to substance use including gender inequality, influence of intimate partners and the need to cope with stress among women (119,124,125), and stigma and perceived medicinal value among PLHIV (135). Finally, access to care for substance use was reported as limited (119,132,135).