Testing for hepatitis C virus infection in UK prisons: What actually happens?

Prisons are a key demographic in the drive to eradicate hepatitis C virus (HCV) as a major public health threat. We have assessed the impact of the recently introduced national opt‐out policy on the current status of HCV testing in 14 prisons in the East Midlands (UK). We analysed testing rates pre‐ and post‐introduction of opt‐out testing, together with face‐to‐face interviews with prison healthcare and management staff in each prison. In the year pre‐opt‐out, 1972 people in prison (PIP) were tested, compared to 3440 in the year following opt‐out. From July 2016 to June 2017, 2706 people were tested, representing 13.5% of all prison entrants (median 16.6%, range 7.6%‐40.7%). Factors correlating with testing rates were as follows: pre‐admission location of the PIP (another prison or the community, OR 2.2, 95% CI 1.9‐2.3, P < 0.001); whether the PIP could access health care independently of prison officers (OR 1.7, 95% CI 1.5‐1.8, P < 0.001); the absence of out‐reach services for HCV treatment (OR 1.3, 95% CI 1.2‐1.5, P < 0.001), whether >50% of PIP reported ease of access to a nurse (OR 2.0, 95% CI 1.8‐2.2, P < 0.001), and whether prison health care was supplied by private or NHS providers (OR 1.3, 95% CI 1.2‐1.5, P < 0.001). Testing rates remained far below the minimum national opt‐out target of 50%. Inadequacy of healthcare facilities and constraints imposed by adherence to prison regimens were cited by healthcare and management staff at all prisons. Without radical change, the prison estate may be intrinsically incapable of supporting NHSE to deliver the HCV elimination strategy.

c Some prisons permit PIP to walk to the healthcare department unaccompanied, others require prison officers to collect PIP from the wings and take them. This can be dependent on the prisons security category or the PIPs behaviour.
TA B L E 2 Details of anti-HCV testing per prison during 12 months pre-and post-introduction of the opt-out policy  9 An opt-out approach requires testing to be embedded in routine care with the option to "opt-out", in contrast to an opt-in approach where patients are asked whether they would like to have a test. 10 The joint commissioning agreement stated that from April 2014 all PIP were to be tested near reception into prison or at other time points, unless they specifically declined, 9 with a target to test 50%-75% of those admitted. 11 In view of the major importance of achieving these targets for the ultimate success of the national HCV elimination strategy, and in responding to impending increases in NHSE national targets for HCV treatment, we set out to evaluate the impact of opt-out testing on the rates of test uptake in 14 prisons in the East Midlands geographical region, as defined by the NHS commissioning board 12 and to identify factors contributing to the success or failure of this policy.

| ME THODS
This report describes the first quantitative stage of a mixed methods sequential explanatory study conducted within a realist evaluation methodology. The qualitative interviews and the overall realist how PIP were able to access the prison healthcare facility; the date of commencement of DBS testing; and any perceived difficulties in achieving testing of PIP for BBV. Interviews were conducted with the head of health care (n = 4, one of whom managed two prisons), the deputy head of health care (n = 2), the primary healthcare matron (n = 2, one of whom was matron for three prisons), a senior healthcare nurse (n = 2) and a BBV lead nurse (n = 1). This contextualized the delivery of health care in which the opt-out approach to testing was situated.
Ethical approval was not believed to be required or sought for any of the above activities.

| Data analysis
Binomial logistic regression using SPSS V24 (IBM Corp., Armonk, NY, USA) was undertaken to describe the relationships between the percentages of people tested (dichotomous dependent variable) and the following explanatory categorical variables: the PIP's previous location prior to admission, the location of testing within the prison, the requirement for a prison officer escort to attend an appointment in health care, the availability of an out-reach HCV treatment service, the ease of seeing a prison nurse within each prison and whether the prison healthcare service was from private or NHS providers.

| Policy implementation
In the East Midlands, the opt-out policy was implemented at term injecting, 14 so this method was anticipated to encourage PIP to engage in the testing process. The date of opt-out policy commencement in each prison was therefore taken as dating from the first DBS sample referred for testing from that prison (see Table 1).

| Pre-and post-opt-out testing rates
The numbers of PIP undergoing testing for HCV in 14 prisons in the 12 months before and after the introduction of the opt-out policy in each prison is shown in YOI, young offender institute. a A prisoner may be admitted to prison custody on more than one occasion during a quarter, and a prisoner may be admitted on multiple occasions across different quarters. For example, a prisoner will be counted as being admitted to prison custody the first time they enter prison custody for an offence committed, further if the prisoner is transferred to another prison within the East Midlands, this will also be counted as an "admission". b Other locations include high security hospitals and approved premises.  TA B L E 5 Prison factors associated with venous and dried blood spot test uptake in other PIP being required to stay where they are because there are no security staff to oversee their movement from one department to another, for example to attend an appointment at the healthcare department.

| Operational features of prison health care
In half of the prisons, the PIP were permitted to walk to the healthcare department unaccompanied by a prison officer at des-  (see Table 4). 1643 of those were tested using DBS and 1063 via standard venepuncture. The overall rate of anti-HCV-positive results could not be calculated as this information from venous blood samples was not available from all participating laboratories, but of the DBS tests, the anti-HCV positivity rate was 9.3% (152/1643), with a between-prison range of 0% to 23% and a median of 3.5%.

| Other factors
Five factors were shown to correlate with prison testing rates ( was fourfold greater in prisons where the largest intake was from the community, rather than from another prison (OR 3.9, 95% CI 2.7-5.6, P < .001).

| D ISCUSS I ON
The  (Table 4), giving a crude rate of 13.5%.
This outcome is very far below the Health and Justice Indicators of Performance lower testing threshold of 50% and entirely inconsistent with achieving WHO targets. 11 There are a number of unavoidable constraints on our data. Most importantly, there are systematic difficulties in defining both the denominator (the total population that ought to have been tested) and the numerator (number of PIP tested) used for calculating testing rates. Whilst this is the first study, to the authors' knowledge, to incorporate novel Ministry of Justice data detailing the numbers of people entering prisons over time, some individuals may have entered more than one prison in the study period, leading to an overestimate of the denominator. For the numerator, it cannot be assumed that the tests recorded were, in fact, limited to those who entered the prison estate, as PIP already within the estate who did not move prisons within the study timeframe may also have contributed to the test numbers. It is also likely that some of the DBS samples were duplicates from the same individual, especially for those PIP moving between prisons, and similarly, that some venous samples were sent from PIP already tested by DBS, in order to confirm positive DBS test results. We are unable to quantify this without reference to individualized patient data, but given that the HCV positivity rates from venous samples in 5 prisons were in excess of 20% (see Table 4), we believe the combined total of DBS plus venous samples will, in fact, be greater than the true number of PIP tested. Our figure of 13.5% PIP tested (2706 by either DBS or VBT of an intake of 20 075 entrants) is therefore the best calculation we can perform on the available data, but we believe this likely to be an overestimate. There is an evident need for more accurate data relating to the numbers of individuals tested (as opposed to the numbers of tests performed) and the true denominator against which to compare the numbers tested. ing structural barriers to accessing care in other settings. 18,19 Whilst prison regimens will have to take precedence, improving access to health care may nonetheless be an achievable target.
The observations that out-reach services impacted negatively on testing rates (which is somewhat counterintuitive), and that undertaking HCV testing on the prison wings may increase uptake (this did not quite achieve statistical significance, P = 0.063) will be explored in a future qualitative publication following interviews with PIP. The former is likely to reflect the established importance of confidentiality within the prison system. 6 from 20% to 53% in a male prison 26 and from 70% to 98% in a mixed prison 27 when tested the day after arrival using rapid point of care oral tests. These studies suggest that moving towards a test which does not require a blood sample and/or provides a same day result may impact on uptake of testing offered as part of an opt-out policy.
In summary, we have conducted an evaluation of HCV testing in a large UK Prison Estate. Whilst this evaluation was conducted in a single geographical region, the 14 prisons in the East Midlands vary in security category, size and purpose, and are therefore likely to provide a representative sample of the challenges of conducting efficient testing for HCV in UK prisons. Further, the different data types gathered in our study collectively strengthen the evaluation by highlighting factors which were predictive of uptake and supplying a broader insight into the issues at stake beyond simply the test uptake figures. Prisons present an apparently opportune context for the delivery of health care, and a more ordered environment for the delivery of treatment than many community settings. They are, however, principally establishments that remove individuals' liberty both as punishment and for public protection, so the maintenance of prison security regimens will always be the priority. The evidence presented here clearly illustrates that rates of anti-HCV testing in a representative prison estate in the East Midlands following the introduction of opt-out testing remain far below national targets. Our study indicates that, despite models of successful interventions in individual prisons (see Table 2 and reference 21 ), urgent systematic change is required to create simple and transferable models of care and normalize the concept of HCV testing in prisons. Further, our multiple data sources and contextual information from stakeholders lead us to conclude that, whilst an increased application of the opt-out policy may make an incremental change, the major factors operating in this failure relate principally to the infrastructure within prison healthcare facilities, which in most prisons are not equipped in terms of staff or space to deal with the increased workload that a 50% testing target would engender, and the low priority of BBV testing within the overall prison regimen. As the prison estate is a critical demographic in the UK drive to eliminate HCV as a major public health threat by 2025, 2 our findings have major implications for the success of this policy. We therefore suggest that further dialogue between the National Offender Management Service, NHS England and PHE will be required to take account of our findings and ensure the ultimate success of strategies to increase rates of testing for HCV in UK prisons.