Our work aimed primarily to identify the prevalence of opioid dependence and OAT in the custodial setting in Berlin and to assess the actual OAT practice regarding substances used for OAT. Further, we aimed to identify factors that affect individuals’ access to OAT in prison.

We conducted the survey in the Berlin custodial setting, which comprises six prisons, the youth custody center, and the prison hospital of Berlin. On the date of data collection, 4,038 people were living in detention. Sentence types included penal incarceration, pre-trial detention, juvenile sentence, and compensation imprisonment, the last of which is a form of imprisonment assigned to individuals who are “unwilling or unable to pay a fine for committing a criminal offence” (37). In each prison, a physician is responsible for the entrance examination and primary healthcare services (2). OAT is executed by either general practitioners with additional qualifications in addiction medicine or by psychiatrists. All medical interventions performed during detention are to be documented in the medical section of the electronic documentation system called Basis-Web.

On March 25, 2019, we conducted a cross-sectional survey and extracted data from the routine records of the 4,038 people in detention recorded in the medical documentation system. Prior to analysis, all cases were assigned pseudonyms so that no connections could be made between cases and the individuals’ names.

We used the documentation system’s integrated, advanced search mode to extract for each detention facility separately all files marked with either the terms “BTM” (meaning Betäubungsmittel; an abbreviation for the German term for narcotics), substitution, detoxification, addiction disease, long-term substitution, drug addiction, tapered withdrawal, opioid dependence (corresponding with F11.2 in the International Classification of Diseases), or polyvalent substance use disorder (corresponding with F19.2 in the International Classification of Diseases) (38). Subsequently, we individually investigated the identified files for documented opioid dependence, as not every detention center used the same markers and not every marker exclusively referred to opioid dependence. Individuals were defined as opioid dependent if a medicinal prescription for OAT or withdrawal therapy was documented in their files. That means we focused on opioid dependence during imprisonment and not on a lifetime prevalence of opioid dependence. In our clinical routine, we experience that individuals directly mention substance use towards medical staff, which facilitates diagnosing substance dependence. We therefore relied on the detection of opioid dependent individuals through the documentation system, even if some individuals may have passed undiagnosed if they did not mention opioid dependence during diagnostic interviews. Duplications due to files marked with more than one term were eliminated. We included polydrug use in the search categories, since in the clinical routine, the diagnosis is also assigned to patients who mainly consume opioids alongside a varied co-usage of additional substances. We did not include dependence upon substances mainly used for pain management, such as fentanyl or tramadol, as this concerns only a minority of opioid users in Germany (39), probably due to restrictive prescription politics (40). We thus extracted 652 people living in detention with documented opioid dependence.

We obtained sociodemographic data for each individual with reported opioid dependence including age, sex, non-/German nationality. We extracted information about their OAT including the prescribed substance, if OAT was begun prior to imprisonment, and if OAT was begun or terminated during imprisonment. Tapered withdrawal with opioids was not considered an OAT. For information about the detention setting, we extracted the prison, the anticipated imprisonment duration, and the sentence type for each individual. Fifteen opioid-dependent individuals were in preventive detention or life imprisonment; in these cases, their estimated duration was not defined. For statistical reasons, we therefore labeled the detention length using the reports of the German Institute of Criminology (41, 42). Furthermore, we recorded each individual’s infection status for HIV and hepatitis C. “No infectious disease”/”no HIV/hepatitis C” noted in the entrance examination or documented negative test results were reported as no infection. Anamnestic infection or positive test results were reported as infection. If neither was documented, the status was reported as unknown. The comorbidity of schizophrenia was also extracted. We analyzed the schizophrenia diagnosis rather than psychosis because the latter was not documented consistently. Even if by this manner we could not verify the influence of psychosis on the access to OAT as assumed by Schulte et al. (27), we nevertheless included schizophrenia in the model as we assumed from our experience that schizophrenic individuals may experience barriers to access treatment due to their diagnosis. Finally, language-related communication barriers were analyzed. If the anamnesis contained the term “yes”, “good”, or “sufficient”, or if no annotation was made about an individual’s language skills, we recorded “no communication barriers”, which also signified that the physician and patient may have had another common language apart from German. Documentation of the term “no”, “some”, “language barrier”, “little”, or “broken” in reference to language skills was reported as a communication barrier.

Approval for the research was obtained from both the Criminological Service of the Law Enforcement Agency of Berlin (KrimD 45/3/009/19) and the local ethics committee at Charité-Universitätsmedizin (EA1/082/19).

For the 652 diagnosed opioid users, we computed general descriptive statistics for their sociodemographic data and prevalence. The continuous parameters of age and estimated detention length are presented respectively as the arithmetic mean plus the standard deviation and the median plus the interquartile range. Categorical parameters are indicated as absolute frequencies and percentages. We formed sub-groups and compared central tendencies of the continuous variables using the Mann-Whitney test for the variable detention length (no normality assumption) and an independent t-test for the variable age (normality assumption) (43). Categorical parameters were compared using Pearson’s Chi-square test.

In order to identify the factors that statistically correlated with the provision of OAT, we calculated binary logistic regressions. The factors age, sex, non-/German nationality, language-related communication barriers, schizophrenia, hepatitis C, HIV, receipt of OAT prior to imprisonment, detention duration, prison, and sentence type were included as independent variables to assess their impact on the receipt of OAT during detention. As the youth custody center offered no OAT, it was excluded from the regression models; the women’s prison was additionally excluded because sex was a variable. Thus, a total of 641 people in detention were included in the models. The variables of hepatitis C or HIV infection and estimated detention length achieved missing values, and therefore we applied multiple imputation (m = 20 imputations). We included the previously defined independent and dependent variables as well as the respective outcomes in the imputation model.

For all analyses, p <.05 was considered significant. We performed the analyses using IBM SPSS Statistics, version 25 and DB-Browser for SQLite, version 3.11.2.

On March 25, 2019, 4,038 people were detained in the Berlin custodial setting. Of these individuals, 652 were documented as opioid dependent, thus representing 16% of the prison population (see Table 1). The prevalence of documented opioid dependence varied between 3% (n = 18/631) in a day-release prison and 25% (n = 211/857) in prison A.

Of the 652 opioid-dependent people in detention, 274 received OAT (42%; n = 274/652). All detention facilities except the youth custody center provided OAT and the substitution rate varied between 20% (n = 32/153) in prison B and 84% (n = 43/51) in the women’s prison (see Table 1).

A total of 202 individuals were already receiving OAT at the time of their imprisonment; of those treatments, 73% (n = 147/202) were continued without any interruption until the date of data collection (see Table 2), while 16% (n = 33/202) were ended at some point during detention and 11% (n = 22/202) at the beginning of imprisonment. As no OAT was provided in the youth custody center, one individual who received OAT at the time of imprisonment stopped receiving treatment. At the women’s prison, all OATs prior to imprisonment were continued until the date of data collection.

The people receiving OAT during detention and the individuals without OAT differed statistically significantly regarding age, nationality, and estimated detention length (see Table 3).

Three different substances were prescribed for OAT; 31% of individuals (n = 85/274) were treated with methadone, 55% (n = 151/274) with levomethadone, and 14% (n = 38/274) with buprenorphine (see Table 1). The number of prescribed substances varied widely among the different custodial facilities, ranging from 0–91% for methadone, 0–71% for levomethadone, and 4–30% for buprenorphine (see Figure 1).

The hepatitis C status was unknown for 27% (n = 176/652) and the HIV status for 43% (n = 280/652) of the opioid dependent people in detention. The infection status of both HIV and hepatitis C was known for 56% (n = 364/652) of the opioid-dependent individuals, while in 26% (n = 168/652) of the records, no information was entered for either. The documented infection status differed significantly between the subgroups (see Table 4).

We therefore calculated the prevalence of HIV and hepatitis C twice: once related to the overall number of opioid-dependent people in detention and once related to the number of people in detention for which the respective infection status was known. Thus, the overall HIV prevalence was 4% (n = 28/652) and 8% (n = 28/372; see Table 4), respectively, while the overall prevalence of hepatitis C was 31% (n = 199/652) and 42% (n = 199/476), respectively.

Thirty schizophrenia cases were reported, which amounted to an overall prevalence of 5% (n = 30/652) among opioid-dependent people in detention (see Table 4).

In the binominal logistic regression model, the most significant predictors of OAT provision were female sex [adjusted odds ratio: 28.575, (95%, CI: 9.057–90.156), p <.000], German nationality [2.170, (1.330–3.539), p = .002], receipt of OAT at the time of imprisonment [12.071, (7.175–20.308), p <.000], estimated detention duration [1.012, (1.005–1.018), p = .001], compensation imprisonment status [3.383, (1.675–6.833), p = .001], as well as detention in prison A [6.285, (2.728–14.478), p <.000] and prison C [3.840, (1.611–9.153), p = .002; see Table 5]. Hepatitis C infection had a statistically significant impact in the model only preceding imputation [1.977, (1.069–.657), p = .030].

With a 16% prevalence of opioid dependence, our results reveal a lower rate than do previously conducted studies, which estimated the prevalence of current or former IDU at 21.9–29.7% among people living in German prisons (27, 28). This discrepancy may have been influenced by our definition of opioid dependence, which contrasts with other studies in that it focuses on actual opioid consumption at the time of imprisonment rather than a lifetime incidence of substance injection. Further, it could reflect the overall decrease of IDU observed in European countries (44) as the data from the previously cited studies is more than ten years old. More generally, this result fits within the estimated range of prevalence at 2–38% for IDU in European prisons (21). The relation of non-German to German opioid-dependent people in detention corresponds with those in the overall Berlin custodial setting, as about half of the people living in Berlin prisons are not of German nationality (45). As a whole, the results again point out that opioid dependence is more frequent in the custodial setting than in the community, where it is estimated at around 3.1/1000 in Berlin as well as in Germany (46).

Our observed OAT rate of 42% is significantly higher than the estimated 10% of dependent individuals in detention receiving OAT on the national level (47). The OAT rate in the community can be estimated at 48% in Berlin (46, 48) and in 2012, the European Monitoring Center for Drugs and Drug Addiction estimated that at least one in two of the estimated population of problem opioid users in Europe receive OAT (49). These statistics suggest that an OAT rate of 42% in the Berlin custodial setting, especially with a variability between 0–84%, is still rather low compared to the extramural practice. One may assert that not all people in detention eligible for OAT are willing to begin an agonist treatment. Even if this were true, the results of a study conducted in a remand prison in Switzerland suggest that opioid-dependent individuals who are entering detention are highly willing to begin OAT (50). This implication suggests that an OAT rate of 42% is not necessarily due to an individuals’ lack of interest, but rather may reflect the ineffective implementation of OAT in the custodial setting.

Contrarily to the extramural setting, only three substances were administered alongside individuals’ OATs in the Berlin prisons. Compared to the extramural practice based on the statistics of the annual German Report on Drugs and Addiction (48), we observed that methadone and buprenorphine were used less often and levomethadone more often. Despite being used in the extramural setting (48), codeine, dihydrocodeine, diamorphine, and retarded morphine are not offered in Berlin prisons. Though, different treatment substances create the opportunity to more efficiently address individual physical or mental adverse effects and differences in metabolization (51, 52). As Kourounis et al. have determined, a lack of pluralism in medication options creates a treatment design barrier that makes it more difficult for patients receiving OAT to remain in treatment (53). Thus, the use of only two different substances in some prisons may reflect a restricted prescription practice in prison that may present a significant treatment design barrier.

The fact that existing OAT at the time of imprisonment had a major impact on the access to OAT during detention aligns with findings from the German and European level (27, 30, 31). However, a study conducted at the German national level found that 70% of individuals undergoing OAT at the time of imprisonment were required to end their treatment upon incarceration (54) while our results show that 73% of treatments that began in the extramural system were continued until the date of data collection. This discrepancy may reflect the differences in the OAT practices at the transition from the extramural to the intramural sector across various federal states. For instance, in Saarland, no prisons are supplied with OAT medicines, and in Lower Saxony, all prisons are supplied with such substances (24).

Our research demonstrated lower rates of opioid dependence among women than previous studies, which estimated dependence on opioids between 27–50% in the women’s custodial setting (55, 56). However, we found similar to higher rates of OAT provision which varied between 13–84% in other works (2, 55). It is to note that the rate of opioid dependence in the women’s prison was still among the highest in our study. Further a study previously conducted in the custodial setting in Berlin pointed out that among women with addiction living in prison, 90% had at least one other mental disorder (55). This shows that opioid dependent women remain a small, but vulnerable group in the prison setting, which needs to be addressed by prison health services.

Furthermore, we were surprised by the impact of the German nationality on treatment access. In the extramural system, access to OAT is essentially dependent upon individuals’ access to healthcare, which is closely associated with nationality and legal residency status. Assuming that German nationality is an indicator for individuals’ health insurance status, the extramural health insurance situation still seems to influence their intramural access to treatment. This is even more striking considering that healthcare costs during detention are covered by the federal states (57).

We observed a significantly higher share of people in detention with language barriers among those without OAT than among those who received treatment. This finding suggests that communication abilities still have a practical impact on individuals’ receipt of OAT.

In contrast with the findings of other studies (27, 32) and our expectations, we found that HIV and/or hepatitis C infection did not seem to be a predictor for the provision of OAT during detention. This result may be explained by the fact that both the HIV and hepatitis C infection statuses were exclusively known for 56% of people in detention, which thus renders rather unlikely a systematic decision regarding whether or not individuals should begin OAT depending on their infection status. Furthermore, the fact that infection status was documented significantly more often among individuals who received OAT during detention suggests that an individual’s receipt of OAT is associated with a higher rate of proposed testing for HIV and hepatitis C. Meanwhile, among people in detention worldwide, HIV and hepatitis C prevalence is estimated at 3.8 and 15.1%, respectively (58). A recent German study found that 66% of individuals who inject drugs are infected with hepatitis C and 4.9% with HIV (39). We determined similar results with an estimated prevalence of 4–8% for HIV and 31–42% for hepatitis C among the opioid dependent individuals, showing that both HIV and hepatitis C still present a major health issue in the custodial setting. Infection status seems to be less often known among male opioid dependent individuals without OAT and among women, while in our results these even presented higher infection rates than men of both HIV and hepatitis C. The controlled structure of imprisonment should be used to systematically propose testing, counselling and treatment of infectious diseases (39). The supply of OAT should be intensified as a strategy of harm reduction among others, in order to prevent new infections among people in detention (17, 19, 20, 39).

Contrary to our expectations, diagnosed schizophrenia had no statistical impact on provision of OAT. However, we identified a 5% prevalence for schizophrenia among opioid-dependent people in detention. As such, schizophrenia remains an important comorbidity, as its prevalence is higher herein than in the overall population, where it is estimated at 3.1% (59).

The fact that each month of detention increased a person’s likelihood to receive OAT may reflect the attitudes of physicians who prefer to administer OAT to individuals with longer sentences in order to assure the treatment’s stability and durability. Meanwhile, the WHO recommends to propose OAT to people in detention who are not yet receiving such treatment even if the remainder of their sentence is short, as OAT reduces both the risk of overdose after release and reincarceration rates (3) and this further could reduce infection rates with hepatitis C (17).

Eventually, we observed that the access to OAT seemed to depend on the prison in which individuals were detained. It is noteworthy that prison D unites two different custodial facilities with two different medical entities, one of which primarily detains individuals under compensation imprisonment. Thus, compensation imprisonment represented a predictor for OAT most likely due to factors related to this sub-prison. Several authors discuss differences in attitudes held toward liberal and harm-reduction drug politics—which translate into different institutional policies and regulations—as a main reason for the general hesitation to use OAT in prisons and its high implementation variability between different federal states (24, 30, 34, 60). Yet the variability of implementation seems not only to be limited to the national level (24), but also to apply to the federal state level. This variability in treatment implementation at every institutional level—within countries and federal states—has been observed in other European countries (33). It suggests that different prisons have different OAT practices and that indications for agonist treatment do not follow a common approach. Another reason for the variability of implementation may be differences in the respective prison physicians’ qualifications in addiction medicine. This observation is even more striking considering that Berlin represents a federal state and a city at the same time. It could have been assumed that the geographical closeness and the institutional frame would lead to a consistent treatment approach. It is important to consider that, even if OAT implementation in prison presents certain limitations due to institutional implications, it remains a setting that may theoretically offer a low-threshold service for drug users regarding accessibility barriers (53). The German federal state of North Rhine-Westphalia has recently shown that the amount of people in detention receiving OAT could significantly be increased by a clear statement of the Ministry of Justice that OAT has to be implemented in prisons as well as treatment recommendations developed by the medical profession defining a standard of care, medical education of prison doctors and a monitoring system (61). Systematically offering OAT through primary healthcare, based on existing international and national treatment guidelines for opioid dependence (3, 4), would reduce selective intake criteria and consequently improve accessibility. Other European countries have demonstrated that this approach is both possible and generally well-accepted by people living in detention (50, 62).

When interpreting our findings, it must be noted that all results are as valid as the documentation provided for analysis. While the people in detention receiving OAT were quite thoroughly documented, individuals who did not mention opioid dependence to medical staff during the entrance examination or during detention did not appear in our analysis. It is likely that we overestimated the rate of individuals receiving agonist treatment.

Further, it must be considered that the documentation system used in the Berlin prison setting was not designed for epidemiological analysis; this fact most importantly relates to the estimated prevalence of HIV and hepatitis C. As mentioned above, the infection statuses were not consistently documented in the electronic system and we had to rely on documented test results as well as on anamnestic information. Though, documented positive or negative test results are more useful than anamnestic information of “no infectious disease”, as this statement may imply a summary of negative results but could also be a simple re-statement of unconfirmed medical history. Such, each calculated prevalence and its impact on individuals’ access to OAT merely present an approximation.

Eventually, we conducted quantitative, cross-sectional research that cannot explain all of our findings and does not display long-term outcomes. Individuals’ perspectives of substitution treatment remain unknown and are necessary to consider if we are accurately to assess their needs and experienced barriers to accessing OAT in prison. Similarly, institutional factors of the prison setting that affect OAT implementation are only marginally represented in our study. Our results can therefore be considered a first quantifying step that necessitates further qualitative research.