Substance use disorder and associated factors among prisoners in a correctional institution in Jimma, Southwest Ethiopia: a cross-sectional study

The prison is managed by the Oromiya Regional Correcting Units Administrative Office. It was put into service after the expulsion of the occupying Italian forces in 1943 and serves the following regions: Oromiya, Southern nations and nationalities, and Gambella region. The facility was built to accommodate 450 prisoners but currently houses about 1460 (1418 male and 42 female). It is designated as a maximum-security facility. The prison population includes offenders on remand and people convicted to a limited or life-long sentence. The prison compound houses a medical clinic and sick bay, in addition to the usual prison facilities. The study was conducted from June 5 to July 5, 2017.

This was a cross-sectional study conducted in the correctional institution in Jimma, Southwest Ethiopia

The sample size (n) was calculated by the single population proportion formula, n = (Zalpha/2)² * P (1-P)/d² [13], assuming a prevalence (P) of 50%, i.e. 0.5 (because a priori prevalence of 0.5 yields a maximally conservative estimate for the required sample size and no similar study has been performed among a prison population in Ethiopia or any other African country), a 95% confidence interval (CI) of 1.96 (Zalpha/2 = 1.96), a 5% margin of error (d, 0.05), and a non-response rate of 10%.

We applied the single population proportion formula to give n = (1.96)² * [0.5 (1–0.5)]/0.05² = 384. Because the population size is < 10,000, we used the finite population correction formula with the calculated sample size of 384 and the total population of 1460:

Thus, assuming a 10% non-response rate the final sample size was 336.

Data were collected by five Master of Science in Psychiatry students. All data collectors were given two-day training on the study objectives, data collection methods, tools, methods for maintaining confidentiality, acquisition of informed consent, and handling of ethical issues. The five students were supervised by two Master of Science in Public Health students and the principal investigator. A pre-test was conducted on 5% of the sample in the Agaro prison, which is located 45 km from Jimma. On each day of data collection, the completed questionnaires were checked for completeness. The collected data were entered into a computer and processed in a timely fashion.

We assessed the presence of four substance use disorders: AUD, khat abuse, nicotine dependence, and cannabis use disorder. To identify AUD, we used the Alcohol Use Disorders Identification Test (AUDIT) [14]. An AUDIT score ≥ 8 is considered to indicate an AUD. The sensitivity and specificity of AUDIT for AUD are 0.90 and 0.80, respectively [14], and the reliability in this study was 0.87 (Cronbach’s α). We assessed nicotine dependence with the Fagerstrom Test for Nicotine Dependence (FTND), in which a score ≥ 1 indicates nicotine dependence [15]. The reliability of the FTND in this study was 0.80 (Cronbach’s α). To evaluate khat abuse, we used the Drug Abuse Screening Test (DAST); a score ≥ 3 indicates abuse [16]. The reliability of the DAST in this study was 0.88 (Cronbach’s α). Finally, we used the Cannabis Use Problems Identification Test (CUPIT) to assess cannabis use disorder; a score ≥ 12 indicates a cannabis use disorder [17]. The reliability of CUPIT in this study was 0.87 (Cronbach’s α). We chose those tools to assess substance use disorder because they were previously used in similar populations [18].

We used a questionnaire to assess the following potential explanatory variables for substance use disorder: socioeconomic factors (age, sex, marital status, ethnicity, religion, educational status, occupation, income); environmental factors (family history of substance use, social support, immigration history); behavioral and mental health factors (previous known mental illness, perception that substance use does not impair health, start of substance use at an early age, chronic physical illness, suicidal ideation and attempts); and criminal factors (previous arrest, previous substance-related offences, type of crime, committed a crime under the influence of a substance). We chose to explore these variables because an earlier study found that they are relevant for substance use disorder [7, 19]. We also assessed social support with the Oslo 3-item Social Support Scale [20], psychopathy with the Psychopathy Checklist: Screening Version (PCL: SV; cut-off score ≥ 13 [21]; sensitivity, 0.94; specificity, 0.85; reliability in this study: Cronbach’s α = 0.86), and adverse traumatic life events with the Life Events Checklist (positive if at least one traumatic event recorded).

Lastly, we used a 4-item questionnaire to gather data on other aspects of substance use: 1. which reason(s) explain(s) why you use the substance? (possible responses: [A] To relax; [B] To relieve stress; [C] To be accepted by peers (peer pressure); [D] To feel normal; [E] For confidence to commit an offence; [F] Others [specify]); 2. Have you been treated for a substance use disorder? ([A]Yes; [B] No); 3. Are you interested in receiving treatment for substance use disorder? ([A] Yes; [B] No); 4. Are you interested in continuing to take the substance after you are released from prison? ([A] Yes; [B] No).

EpiData Version 3.1 was used to enter the data; the mean was used in case of missing data. Then, the data were exported to Statistical Package for Social Science version 21.0 for further analysis. Descriptive statistics, such as the frequency and median, were computed, and bivariate and multivariable analyses were used to identify factors associated with the outcome variable. Factors associated with the outcome variables that had a P value < 0.25 in the bivariate analysis were included in the multivariable analysis. Statistical significance was set at P < 0.05. An odds ratio with a 95% CI was computed to assess the level of association and statistical significance.

The study protocol was approved by the Research Ethics and Approval Committee of the Jimma University Institute of Health, and the study was performed according to the Declaration of Helsinki. Verbal informed consent was obtained from prisoners. Participants were told that selection for participation in the study was random and that they had the right not to respond to questions they were not comfortable with and to ask questions. After data entry was complete, the questionnaires were kept securely locked away.

A total of 329 prisoners participated in the study. The response rate was 97.9%: of the 336 prisoners approached to participate in the study, n = 7 (2.1%) declined to participate because they were unwilling to be interviewed about their substance use histories. The median age of the participants was 26 years (inter-quartile range [IQR] 14). The majorities of participants had been residing in urban areas before imprisonment and were unmarried. The most common ethnicity was Oromo, and the most common religion, Muslim. Detailed information on socio-demographic characteristics is given in Table 1.

A total of n = 227 (69.0%) of the participants had a history of substance use. More than half of all participants (n = 184/329, 55.9%) reported a substance use disorder within the 12 months before imprisonment, and half (n = 165/329, 50.2%) had used a substance within the 30 days before imprisonment.

Khat was the most commonly used substance in the 12 months before imprisonment, (n = 138/329, 41.9%). Among the participants with khat abuse, n = 81(58.7%) had harmful use, and n = 57 (41.3%), khat dependence. The prevalence of AUD within the 12 months before imprisonment was 36.2% (n = 119/329). Among the participants with an AUD, n = 57 (47.9%) had hazardous drinking; n = 20 (16.8%), harmful drinking; and n = 42 (35.2%), alcohol dependence. The next most commonly used substance was nicotine: n = 65/329 (19.8%) of participants had a history of nicotine dependence within the 12 months before imprisonment (low dependence: n = 28, 43.1%; low to moderate dependence: n = 10, 15.4%; moderate dependence: n = 22 (33.8%); and high dependence: n = 5 (7.7%). The prevalence of cannabis use disorder within the 12 months before imprisonment was 3.6% (n = 13/329).

A total of n = 105 participants (31.9%) had a history of two or more substance use disorders. Among the prisoners with an AUD, n = 87 (73.1%) had a history of khat abuse; n = 41 (34.5%), a history of nicotine dependence; and n = 13 (10.9%), a history of cannabis use disorder. Among the prisoners with khat abuse, n = 87 (63%) had a history of AUD; n = 43 (31.2%), a history of nicotine dependence; and n = 13 (9.4%), a history of cannabis use disorder. Among the prisoners with nicotine dependence, n = 43 (66.2%) had a history of khat abuse; n = 41 (63.1), a history of AUD; and n = 12 (18.5%), a history of cannabis use disorder.

The median age at the first use of a substance was 16 years (IQR 5), and 44% of the participants started using the substance before the age of 15 years. The median duration of substance use was 6 years (IQR 7).

The main reasons for starting substance use reported by participants were peer pressure (n = 77, 41.8%), recreational reasons (n = 63, 34.2%), and stress relief (n = 54, 29.3%). Almost all of the prisoners with a substance use disorder had not received treatment prior to imprisonment (n = 178, 96.7%), but n = 122 (66.3%) of them were interested in receiving treatment. A third (n = 62, 33.7%), however, wanted to continue using the substance after they were released from prison.

Among the prisoners with a substance use disorder, the most common reasons for imprisonment were assault (n = 65, 35.3%) and theft (n = 47, 25.5%). The most common reasons for imprisonment among participants with an AUD, nicotine dependence, or cannabis use disorder were assault (n = 42, 35.3%; n = 24, 36.9%; and n = 5, 38.5%, respectively) and murder (n = 28, 23.5%; n = 12, 26.2%; and n = 4, 30.8%, respectively), and among prisoners with khat abuse, assault (n = 48, 34.8%) and theft (n = 38, 27.5%).

The median age of participants at the first imprisonment was 24 years (IQR 13). Nearly all of the participants were convicted prisoners. At the time of the study, the median duration of incarceration was 48 months (IQR 90) and the median duration of time spent in prison was 9 months (IQR 20).

Assault and murder were the most common causes of imprisonment. Most of the participants had no previous history of imprisonment, and most of the crimes were not committed under the influence of a substance. Among the crimes that were committed under the influence of a substance (n = 42/329), n = 24 (57.1%) were committed under the influence of alcohol; n = 14 (33.3%), under the influence of khat; and n = 4 (9.6%), under the influence of cannabis. Table 2 shows the study participants’ forensic history. See (Table 2)

Various behavioral, mental health, environmental, and criminal factors were found to be associated with substance use disorder (see Table 3).

Multivariable logistic regression showed that four variables were significantly associated with substance use disorder: poor social support (adjusted odds ratio [AOR]: 4.41; 95% CI, 2.22–8.77), living in an urban setting (AOR: 2.42; 95% CI, 1.33–4.40), psychopathy (AOR: 4.68; 95% CI, 1.71–12.78), and a family history of substance use (AOR: 4.39; 95% CI, 2.49–7.79). Prisoners with poor social support were more than four times more likely to develop a substance use disorder than prisoners with good social support (AOR:4.41; 95% CI, 2.22–8.77). Also, prisoners living in an urban setting were more than two times more likely to have a substance use disorder than prisoners living in a rural setting (AOR:2.42; 95% CI, 1.33–4.40). Prisoners with psychopathy were nearly five times more likely to develop a substance use disorder than those without psychopathy (AOR: 4.68; 95% CI, 1.71–12.78). Prisoners with a family history of substance use were more than four times more likely to have a substance use disorder than prisoners with no family history of substance use (AOR: 4.39; 95% CI, 2.49–7.79) (see Table 4).

This study may have a social desirability bias, i.e. prisoners may have underestimated, underreported or denied their substance use. Also, data on the previous 12 months were collected by interview, which has a risk of recall bias. DAST was not validated in our population, even though it has been shown to be useful in screening for khat abuse across cultures. Another limitation is that prisoners’ reports of past or present physical illness and mental illness were not clinically confirmed.