Ethics statement

The study was conducted in accordance with the Declaration of Helsinki on ethical principles for medical research involving human subjects. The Ministry of Justice of the Republic of Chile approved of the study. The ethics committee and institutional review board of the University of Chile approved of the study (reference number 342/12). We used a standard written informed consent procedure approved by the ethics committee of the University of Chile. The inclusion of subjects was voluntary upon written informed consent. The capacity of giving informed consent was inferred from the capacity to understand the information about the study and explain the goals back to the interviewer. Age 15 or older and the capacity to give written informed consent were inclusion criteria. Minors of 15 to 17 years were subjected to the same informed consent procedure as adults of 18 years and older. Relatives or guardians were not involved in the written informed consent procedure. All potential participants who declined or otherwise did not participate were eligible for treatment as usual and were not disadvantaged in any other way by not participating in the study.

Design

We conducted a cross-sectional survey among 1008 of the 46,825 prisoners in closed correctional facilities in Chile. The sample was recruited from 7 penal institutions located in 6 different cities and 5 regions of the country. The sample was stratified for the 5 regions of the country (north, north-central, central, south-central, south), proportionately to the size of the prison population in each region.

Procedures

The participants were randomly selected from the 7 penal institutions throughout Chile (located in Arica, Valparaíso, 2 in Santiago, Rancagua, Concepción and Temuco). High security sectors within the prisons for inmates showing uncooperative, violent behaviours during detention or high numbers of daily insurgencies were excluded for safety reasons. Sectors of some of the penal institutions were excluded from the sampling due to the lack of basic infrastructure such as a separate room to conduct the interviews. Privacy of the information was assured to all possible participants. All interviewees provided written informed consent. The refusal rate was 1.0%. Two medical doctors and one information technologist were trained in using the Composite International Diagnostic Interview (CIDI) in the Institute of Psychiatry Ramón de Fuentes in Mexico, a WHO accredited centre. The team trained 28 interviewers in Chile for the fieldwork. The 28 trained fieldworkers conducted the interviews between August and October 2007. The field workers were selected on the basis of being professionals with complete tertiary level education related to the topic and setting of the study. They had different professions such as social scientists, sociologists, psychologists and social workers. All of the field workers were trained in the use of the CIDI in a two-day workshop with supervised practice interviews, role-plays and discussions of each question. The trainers evaluated the interviewers’ understanding of the purpose of the study, the purpose of the instrument and of the content of each question. Internal quality control was assured through retraining and mentoring in the field. Unannounced field visits supervising single interviews were carried out. Weekly scheduled meetings with the interviewers and field executives were held for problem identification and troubleshooting.

Instruments

The interview schedule included socio-demographic characteristics and a computerized version of the CIDI 3.0, a fully structured lay administered interview to establish DSM-IV diagnoses [22]. The instrument has a satisfactory concordance with professionally administered interview schedules [23]. It has been shown that non-clinicians can reliably use the CIDI after a relatively short training period [24]. The CIDI has shown good to excellent Kappa coefficients in test–retest and interrater reliability studies [24]. We opted for the 12-month prevalence rates, rather than 30-day or lifetime prevalence rates for its relevance regarding the need of treatment.

For the comparison of the prison and the general population, we compared prevalence rates separately for males and females. Data from the largest (n=2978), representative population based survey in Chile conducted with the CIDI (versions 1.0 and 1.1) in the 1990s was used for this comparison [21]. Both age adapted and non-adapted comparisons were made.

Analysis

Prevalence rates were calculated as per cent values with 95% confidence intervals with SPSS® statistical software version 19. For the comparison of frequencies a two-sample calculator for proportions with Stata® statistical software version 12.0 was used. Z scores > ±1.96 and p values <0.05 were considered statistically significant corresponding to the significance levels of the two-sided test.

Socio-demographic characteristics

The final sample included 1008 inmates from five different regions in the country (Table 1). The mean age of the sample was 32.6 ± 10.1 years; 84.8% were male and 15.2% female. There was information on the legal status of 877 subjects, of which a majority of 79.6% were convicts and 20.4% pre-trial or under trial detainees. About one fourth of the sample (25.8%) had a total time in detention of less than 12 months. The mean time spent in detention was 66.7 months. The mean number of people per cell was 5.1 for the included subjects (Table 1).

Prevalence rates of mental disorders in Chilean prisons

The 12-month prevalence rate for any mental disorder was 26.6% (95% CI, 24.0-29.3). The prevalence rate for any substance use disorder was 12.2% (95% CI, 10.2-14.1); within that group 6.6% (95% CI, 5.2-8.1) of the total sample had disorders associated with illicit drug use. The 12-month prevalence rates for the group of anxiety disorders was 8.3% (95% CI, 6.6-10.0), for the group of affective disorders 8.1% (95% CI, 6.5-9.8); within that group 6.9% (95% CI, 5.4-8.6) had major depressive disorders (Table 2).

Differences of prevalence rates for short and long term prisoners

Subjects with a detention time ≤12 months had a higher rate (35.7%) for any mental disorder than subjects with >12 months in detention (23.1%; p<0.001; Table 3). Alcohol use disorders were significantly more common (12.6%; 95% CI 8.5-16.7) in inmates with a detention time ≤12 months as compared to inmates with >12 months in detention (2.1%; 95% CI 1.1-3.1) and illicit drug use disorders were more common for the prisoners ≤12 months in detention (8.5%; 95% CI 5.1-11.9) as compared to the ones >12 months in detention (3.5%; 95% CI 2.2-4.8). Since the reported data are 12-month prevalence rates the latter numbers correspond to the ones reporting continuous substance use during imprisonment. Nicotine dependence was also significantly more common among those with a detention time of less than 12 months. There was no statistically significant difference for any of the other disorders. Depression was more frequent in the ones with lower detention time, mania more frequent among the ones with longer detention time, but the differences were not statistically significant.

Comparison of prevalence rates in the Chilean prison population and general population

There were different patterns of mental disorders in the prison population and the general population (Table 4): Significantly higher prevalence rates among prisoners as compared to the general population in Chile were seen for major depression (6.1% vs. 3.7% males, Z=2.58, p<0.05) and illicit drug use (3.3% vs. 0.6% males with drug abuse, Z=2.04, p<0.05; 2.6% vs. 0.1% females with drug abuse, Z=5.36, p<0.001; 3.4% vs. 1.1% males with drug dependence, Z=3.70; p<0.001). To the contrary, dysthymia (6.5% vs. 15.6%, Z=-2.39, p<0.05), simple (3.3% vs. 11.5%, Z=-3.13, p<0.001) and social phobias (3.9% vs. 9.7%, Z=2.38, p<0.05) were significantly less frequent in the female prison populations than in the general population. One-year prevalence of alcohol abuse (2.3% vs. 3.9%; Z=-2.04; p<0.05) and dependence (2.7% vs. 8.2%; Z=-5.24; p<0.001) were less prevalent in the male prison population than in the general population. There was a trend for higher rates of non-affective psychosis in the prison population (p<0.1 for the two-sided z-test; if presuming higher rates in the prison population and performing the one-sided z-test, significance levels of <0.05 would have been reached). When the data was weighted for age, the differences remained significant for the same categories.

Co-morbidity

Co-morbidity rates are reported for the two most important diagnostic categories, major depression and illicit drug use disorders. Most frequent co-morbid diagnoses in people with illicit drug use (n= 70) were anxiety disorders and intermittent explosive disorder with 16.7% (95% CI 7.7-25.8) each. Affective disorders were co-morbid in 12.1% (95% CI 4.3-19.9) and ADHD in 10.6% (95% CI 3.2-18.0) of the people with illicit drug use. People with major depressive disorders (n= 67) had most commonly comorbid anxiety disorder (43.5%; 95% CI 31.9-55.8), substance use disorders (22.8%; 95% CI 13.3-33.1), intermittent explosive disorder (18.9%; 95% CI 9.7-28.1) and ADHD (7.2%; 95% CI 1.1-13.3).

Strengths and limitations

This is the largest and most systematic prevalence study of mental disorders in prison populations from a non-Western middle-income country using a standardized structured interview to generate diagnoses. It is the first study that compares nationwide data from prison populations with nationwide data from a population based study, which both used the same standardized interview schedule. A limitation for the comparison of the prevalence data between the prison population and the general population arises from a time lag of 8-15 years between the two studies. The prevalence data in the general population might have changed over time.

As in most prison mental health studies, there were specific challenges regarding the sampling. We had to exclude high security sectors and prisoners with violent or uncooperative behaviours to assure the safety of the fieldworkers. We also excluded sectors within the prisons that were so poorly resourced that they could not provide the infrastructure required to implement the study such as an interview room. Even though treatment capacities for prison inmates were still extremely low at the time of the study, we cannot exclude that few subjects were successfully treated because we did not collect information on current medication or treatment.

Comparison against the literature

There are several limitations for the comparison with studies from other countries: 1. The sampling, as explained above, may have been selective for prisoners with cooperative behaviours. It can be assumed most prison mental health studies are limited in reaching prisoners with violent, dangerous or uncooperative behaviours. 2. Other studies may refer to lifetime or current prevalence rates rather than the 12-month prevalence rates used in this study. 3. Different studies applied different structured psychometric instruments to generate the diagnoses. An alternative to the CIDI used in our study is the Structured Clinical Interview for DSM-axis I and II (SCID I and II) used by other groups for studying prevalence rates in prison populations [25]. The SCID has the advantage over the CIDI that it includes diagnoses of several personality disorders if both parts are applied. It may produce a higher level of diagnostic accuracy, because it is professionally administered. At the same time, this is a disadvantage for poorly resourced settings. Several recently published prison mental health studies used the Mini International Neuropsychiatric Interview [13,20,26], which is similarly to the CIDI a fully structured lay administered interview schedule. It has the advantage that its use is easier to train and teach, it is quicker to administer and it has a section for the detection of antisocial personality disorder, which is relevant for prison populations [27]. For international comparisons of prevalence data, consistent instruments and equivalent time spans should be used (e.g. point prevalence, one-year prevalence or life-time prevalence).

Despite these limitations and the heterogeneous findings of the previous 7 studies from LMICs, our results should be seen against the background of other studies: In comparison, the prevalence rates of mental disorders reported here are lower [13,15]. A study in the closed correctional system of Brazil reported rates of 17.6% for lifetime depression, 5.2% for current depression, 6.6% for lifetime bipolar disorder, 26.6% for lifetime alcohol addiction, 27.9% for lifetime drug addiction, and 1.4% for psychotic disorders [13]. Data reported from India were part of an intervention study including mainly pre-trial detainees. Prior to the intervention, prisoners in two jails were screened for severe mental disorders (psychosis and mood disorders) and a 10% prevalence rate was found [14]. A study from Iran reported 3.1% current and 3.9% lifetime prevalence for psychotic disorders, 30.6% current and 48.7% lifetime prevalence for mood disorders, 11.2% current and 78.0% lifetime prevalence for substance use disorders. Different low- and middle-income countries present different legal and cultural frameworks. Iran employs, as other Middle-Eastern countries, strict laws on substance use, which may explain in part that very high rates of substance use problems were found in the prison population. Rates of current psychotic and mood disorders were also higher than the 12-month prevalence rates found in our study [15]. A prison mental health survey conducted in Malaysia primarily targeted HIV positive prisoners and the comorbidity of HIV, mental and substance use disorders [16]. A study from Mexico on mental disorders of female prisoners showed a prevalence rate of 62% for depression in a selective sample with a history of a substance use disorder [17]. The study from Nigeria in 100 inmates reported a lifetime prevalence of 25% for substance use disorders. Two cases of schizophrenia, two cases of major depression and twenty-one cases of recurrent mild depression were in the sample [18]. The most recently published prison mental health survey from South Africa reported that 42% had a substance use disorder, 10.4% current major depression, 24.9% lifetime depression, 4.7% current psychosis and 7.3% lifetime prevalence of psychosis [20]. Especially current non-affective psychotic episodes may have been overestimated by previous research. Apart from a general difficulty to some diagnoses, especially current active schizophrenia, with diagnostic structured interviews, the different interview schedules may have different thresholds for producing positive scores [28].

Even as compared to high-income countries the prevalence rates found in our study appear to be low [2]. In a sample of long-term prisoners in Sweden, 40% of the participants were estimated to have ADHD as compared to 2.2% in our sample [29]. There are several possible explanations for the moderate prevalence rates: 1. Chile has 305 prisoners per 100,000 population, one of the highest prison population rates in Latin America [1], more than double of the worldwide and South American averages. As for many other countries, the prison population in Chile has more than doubled since 1992 [30]. When prison population rates are very high, the prison population may be more similar to the general population regarding their mental health than in countries with lower prison population rates.

2. There was a majority of long term convicts in our sample who show lower rates especially of substance use disorders than recently admitted prisoners. Levels of distress may be highest shortly after admission because of adaptation to the adverse conditions of imprisonment over time. Reduced rates of overall mental disorders were found in a longitudinal study of the effects of long-term imprisonment in a Western setting including a stabilization of pathological personality traits [31]. Imprisonment may hinder use of drugs and alcohol, but does not preclude it since these substances are still available. However, the type of abuse may be different in and outside of prisons. Our data indicate that it is more common to maintain illicit drug use than alcohol use. That is in line with reports from prison inmates about illicit drug dependence inside the institution, whereas it is more unusual to maintain alcohol dependence [32]. There may also be a reporting bias in the sense of underreporting of active substance use inside the prison. Our data underline the weight the penal institutions have regarding the institutionalization of people with substance use disorders. This window of opportunity should be used for the care of those disorders. Even though our study was not longitudinal the results give indirect evidence that prison inmates reduce active substance use to a certain degree during imprisonment. This may neither be sustainable beyond the sentence nor a cost effective way to deal with addiction nor humane but it constitutes a reality for many people affected by substance use disorders. Inmates show high rates of relapse and death from overdose after release from prison [33].

There are two possible explanations for the lower rates of simple and social phobia in the female prison population as compared to the general population: 1. Simple and social phobia may be associated with traits that are protective against engaging in criminal behaviour. 2. Imprisonment may constitute a protective environment characterized by monotony, predictability and avoidance of possible trigger situations for simple and social phobia.