Discussion

This study found extremely high levels of drug resistance among tuberculosis cases diagnosed while in prison in the UK. Prisoners were also more likely to have infectious forms of disease and not complete treatment, with management complicated by transfer, release and social factors.

The number of prisoners identified with tuberculosis in this study is likely to be a minimum estimate. Only cases in prison at the time of diagnosis were included, so those with a history of imprisonment were excluded. The study also could not include undiagnosed or unreported cases among prisoners, and those reported but unidentifiable as prisoners. As imprisonment was not routinely collected through national surveillance, this relied on searching occupation and address fields. By using multiple data sources, a number of cases were identified that had not been reported to national surveillance: in particular, 12 culture confirmed cases had not been reported.

Tuberculosis incidence cannot be reliably estimated among prisoners in England and Wales from these data. As stated, our numerator is likely to be a minimum estimate. Determining an accurate denominator for the prison population in the study period was also problematic due to the high turnover of prisoners, many having received short remand sentences, and difficulties in obtaining duration of imprisonment required to calculate person time at risk. Previous estimates of incidence in similar settings have ranged from 215 per 100 000 prisoners in Paris6 to 2775 per 100 000 prisoners in Barcelona.7 High prevalence of tuberculosis among prisoners assessed through cross-sectional surveys has also been reported from London (208 per 100 000)2 and Italy (290 per 100 000).8

The high levels of drug resistance and low proportion of prisoners completing treatment within 1 year is concerning. Although more drug resistance leads to a greater number still on treatment at 1 year, many prisoners were lost to follow-up, a potentially avoidable outcome that further increases the risk of onward transmission and drug resistance. This supports previous findings of high rates of drug-resistant disease and poor treatment adherence among prisoners in London.2

The fact that cases who were part of the isonaizid-resistant outbreak were less likely to be lost to follow-up than other prisoners probably reflects increased use of directly observed therapy (DOT) on the advice of an Outbreak Control Committee.12 DOT has been found to improve treatment completion compared with self administration among tuberculosis patients released from prison, particularly if administered at a field site.15 High levels of treatment adherence among prisoners is achievable, even after release, with adequate coordination between programmes both within and external to the prison (such as methadone maintenance programmes, and housing or other social support) as demonstrated in previous studies.16

The large proportion of clustered cases should be interpreted with caution, as this only represents 40% of culture confirmed cases. As routine prospective typing was not done for all cases, those suspected as being part of an outbreak (due to isoniazid resistance or other factors) were more likely to be typed. Nevertheless, the identification of patients with indistinguishable strain types reflects appreciable levels of transmission.

Attributing transmission to imprisonment, however, is not possible in this study. Over-representation of high-risk groups among prisoners is a known factor in explaining higher incidence, but the congregate settings of prisons may contribute to transmission. Delay between infection and onset and the high turnover of prison populations also mean difficulties identifying the source and site of infection. An Australian study found that increased incidence among prisoners was not attributable to prison environment:17 in Spain, however, recent transmission was associated with lengthy incarceration, overcrowding and delayed diagnosis within prisons.7

High rates of tuberculosis, especially drug-resistant tuberculosis, among prisoners along with low levels of successful treatment mean prisoners present a particular challenge for tuberculosis control. These findings support national guidance recommending prisoners to be screened on entry, all those on treatment to receive DOT, and continuity of care ensured for transferred and released cases.18 The importance of this is highlighted by the situation in Russia, where failure to treat prisoners successfully is a significant driver of the tuberculosis epidemic and high rates of drug resistance.5

Initiatives are under way to strengthen tuberculosis control in UK prisons. National tuberculosis surveillance now includes current and previous imprisonment, and other social risk factors. This, combined with universal prospective strain typing, will allow a more comprehensive assessment of the contribution of imprisonment to tuberculosis in the UK, and aid evaluation of interventions. A digital teleradiology network is being installed across London prisons and key regional prisons to screen new prisoners and transfers on arrival. Screening prisoners for tuberculosis with chest radiography has been shown to be a cost-effective intervention to strengthen tuberculosis control among high-risk populations that may be reached through screening in prisons.19 Improving treatment outcomes among these complex cases through increased use of DOT, social and housing support is an essential step to control tuberculosis successfully in the UK.