Setting and participants

There are about 11 000 committals to seven prisons each year in the Republic of Ireland. We excluded two of these committal prisons from the survey because the numbers committed in preceding years were small (5% of annual committals).

We needed to recruit 534 participants in order to estimate the prevalence of antibodies to hepatitis C virus. There were 718 entrants to the five survey prisons during the survey period from 6 April to 1 May 1999; 85 individuals were released or transferred to another prison before they could be interviewed, and six individuals were unable to provide informed consent and were excluded, leaving 627 potential recruits.

Survey

Staff and prisoners were briefed in advance. We visited each prison daily and interviewed all those committed within the previous 48 hours. The list of entrants was obtained from the committal register maintained in each prison. The survey was anonymous and comprised a questionnaire and collection of an oral fluid sample.

Questionnaire—The self administered questionnaire, derived from questionnaires used in prison surveys in the United Kingdom2^–7 and Republic of Ireland,1 took five minutes to complete (see appendix on bmj.com).

Oral fluid tests—Oral fluid samples were collected with a proprietary device (EpiScreen, Epitope, Oregon, USA). Details of testing procedures and estimated sensitivity and specificity were reported previously.1

Statistical analysis

We used Pearson χ² test and Fisher's exact test to compare proportions in independent groups of categorical data and the χ² test for trend to identify linear trends. We developed multiple logistic regression models to identify factors associated with positive test results.

Results

All five prisons participated, and 607 of the 627 available prisoners took part (97%). This represents 85% of the total population committed to these prisons during the survey period and 6% of the roughly 11 000 committals to Irish prisons each year.

Our analyses refer to the 596 participants who provided analysable oral fluid samples or, for use of injected drugs, the 593 respondents who also declared their injector status. Denominators vary because not all respondents answered all questions.

Respondent characteristics

The median age (range) of respondents was 23 years (15–73). The age distribution of respondents was similar to that of the total population entering the study prisons (P=0.97). Forty one respondents (7%) were women. A third (197/591) had never previously been in prison.

Prevalence of viral antibodies

The overall prevalence of antibodies to hepatitis B core antigen was 6%, to hepatitis C virus 22%, and to HIV 2%. Prevalence was significantly lower in respondents who had never previously been in prison (table 1).

Twenty nine per cent of respondents (173/593) reported ever injecting drugs. Only 7% (14/197) of those entering prison for the first time had ever injected drugs compared with 40% (157/394) of those previously imprisoned (P<0.0001). Among injecting drug users the overall prevalence of antibodies to hepatitis B core antigen was 18%, to hepatitis C virus 72%, and to HIV 6%. The prevalence of each of the three infections was lower in the injecting drug users not previously imprisoned (table 1).

For those who had never used injected drugs, the prevalence of each marker of infection was low (table 1). Three of the five non-injectors who tested positive for hepatitis B had been in prison before, as had five of the six non-injectors positive for hepatitis C. Two non-injectors tested positive for HIV, one of whom had previously been imprisoned.

The prevalence of each of the three infections was significantly higher in women than men (table 2). The proportion of women prisoners reporting ever injecting drugs was also higher than in men (63% v 27%, P<0.0001).

Reported injecting practices

Over 70% (120/167) of injecting drug users stated that they had injected drugs in the month before the survey; 85 reported injecting more than 20 times. Almost three quarters of injectors previously imprisoned (110/155) started injecting three or more years ago, compared with 36% (5/14) of injectors among new entrants (P<0.01). Of the 156 injectors previously in prison, over half (85/156) reported sharing needles while incarcerated, although 35 of them stated they had not shared in the month before committal; almost a fifth (29/156) reported starting their injecting habit in prison.

Tattooing

Almost 60% of the respondents (352/596) reported having a tattoo. Injecting drug users were significantly more likely to have tattoos than non-injectors (137/172 (80%) v 215/420 (51%), P<0.0001). The proportion of prison entrants with tattoos also increased with increasing time spent in prison in the 10 years before the survey (χ² test for trend=76, P<0.0001). Thus, only 41% (81/197) of those who had not spent any time in prison were tattooed, compared with 45% (29/64) of those who had spent between one day and three months, 74% (127/170) of those who had spent three months to five years, and 89% (77/87) of those who had spent more than three years in prison. Eighty seven respondents were tattooed in prison.

Sexually transmitted infections

Forty four respondents (8%) reported that they had been treated for a sexually transmitted infection. Most of these were injecting drug users (27/44, 61%).

Hepatitis B vaccination

Of the respondents who had been in prison before, 29% (112/393) had received at least one dose of hepatitis B vaccine. Of these, 82% (89/108) had undergone their vaccination in prison.

Logistic regression

We constructed logistic regression models to clarify the associations between prisoners' characteristics and reported risk behaviours and their likelihood of testing positive for the three viral antibodies. Four groups of variables were considered for inclusion in each model: demographic and sentence characteristics, drug use and injecting practices, sexual history, and tattooing. Significant factors were retained in the models.

Compared with men, women were almost three times more likely to test positive for hepatitis B core antibodies, seven times more likely to test positive for hepatitis C antibodies, and almost 10 times more likely to test positive for HIV antibodies (table 2). The most important predictor of hepatitis antibodies was a history of injecting drugs. Those who reported injecting drugs were 89 times more likely to have hepatitis C antibodies and 16 times more likely to have hepatitis B core antibodies than non-injectors. The likelihood of testing positive for hepatitis C antibodies increased with increasing time spent in prison in the preceding 10 years. Although inferences from the HIV regression model are limited by small numbers, those who had spent more than three of the preceding 10 years in prison were significantly more likely to test positive for HIV antibodies.

We constructed separate models for respondents with and without a history of injecting drugs (tables 3 and 4). Among injecting drug users, hepatitis B core antibodies were more common in older respondents (≥30 years old) and in those who reported having more than 10 sexual partners in the year before the survey, while hepatitis C antibodies were 3.5 times more likely in women and six times more likely in those who reported frequent current injecting or sharing needles in prison, and HIV antibodies were more common in older respondents and in those who shared needles in the month before imprisonment (table 3).

No independent risk factors were identified for the five non-injectors who tested positive for hepatitis B core antibodies (two had no reported risk factors, two had tattoos, and one reported smoking heroin and having been treated for a sexually transmitted infection). Of the six non-injectors who tested positive for hepatitis C antibodies, five had spent time in prison and four had had a tattoo done in prison (table 4). The model indicated that non-injectors who were tattooed inside prison were more likely to test positive for this virus than those who had tattoos done outside prison. Only two non-injectors tested positive for HIV antibodies.

Discussion

This survey showed that, of the third of prison entrants being imprisoned for the first time, only 7% reported injecting drugs, compared with 40% of those who had been imprisoned previously. Bloodborne infections among drug injectors who had previously been in prison were higher than among injectors who had not previously been in prison.

Limitations of study

Conclusions from cross sectional surveys are limited. It is therefore not possible to deduce from this survey whether the higher infection rates in recidivist prisoners are because of their more chaotic drug use patterns (for example, a higher proportion of injectors previously imprisoned had started injecting more than three years earlier) or because of the previous exposure to prison. Increased risk associated with exposure to prison is probably because of the high risk injecting practices adopted in prison (such as sharing a small number of needles with a large and varied cohort of inmates) rather than spending time in prison in itself.

The validity of oral fluid assays is high except for the 80% sensitivity of the hepatitis C antibody test.1 The prevalence of hepatitis C antibodies reported in this survey is therefore likely to be an underestimate of the true prevalence, which could be as high as 90% in injecting drug users entering Irish prisons. This is substantially higher than the prevalence reported in entrants to Australian prisons (64% and 66%). 8 9

Comparison with other studies

This is the first time that the same methods have been used in both a national survey of prison inmates and a national committal survey, enabling direct comparisons. Although the overall prevalence of hepatitis antibodies was lower in prison entrants, the prevalence of these antibodies in entrants previously in prison was similar to that reported in the prison inmates, as was the prevalence in recidivist drug injectors.1 In both surveys injecting drug use was by far the most important risk factor for hepatitis C, with injectors who reported sharing needles in prison or frequent current injecting being more likely to test positive. In both surveys about a fifth of injectors reported that they had started injecting in prison. Surveys in some Scottish prisons have reported similarly high initiation figures. 4 6 10

The prevalence of hepatitis B core antibodies (18%) in drug injectors entering Irish prisons was lower than the 52% and 43% reported in drug injectors entering Australian prisons, 8 9 and also lower than in drug injectors entering French prisons (37%).11 Ireland has a programme of proactive hepatitis B vaccination in prisons, and the vaccination coverage is higher than reported in UK prisons.7 This may contribute to the lower than expected prevalence of hepatitis B in Irish prisoners. Offering the vaccine to all prisoners during committal procedures could further reduce the transmission of hepatitis B virus in Irish prisons.

Tattooing in prison was the only independent risk factor identified for the presence of hepatitis C antibodies in respondents who had never used injected drugs. Abildgaard and Peterslund reported the presence of hepatitis C antibodies in an individual with a tattoo but no other risk factors,12 and Turnbull et al reported that 6% of prisoners interviewed had a tattoo done on their last occasion in prison and that half of these had shared tattooing equipment.13 Taken together, these findings suggest that tattooing may be responsible for transmission of hepatitis C in prison. It may be advisable to include a question on tattooing in future studies of viral prevalence.

Conclusions

Research questions raised by this study are whether the high prevalence of bloodborne infections in recidivist prisoners derives from chaotic drug behaviour outside prison that leads to repeat imprisonment or from the risk behaviours adopted within prison; whether programmes of proactive hepatitis B vaccination in prisons will lower the prevalence of hepatitis B; and the role of tattooing in transmitting hepatitis C virus and whether provision of sterile tattoo kits would help reduce infection rates.

It is clear that both use of injected drugs and infection with hepatitis C virus are endemic in Irish prisons. Only a small number of the new entrants committed during the survey period were infected with one or more of the viruses. As imprisonment leads to high risk practices, this survey points to the need for increased infection control and harm reduction measures in Irish prisons.