Testing for COVID-19 during an outbreak within a large UK prison: an evaluation of mass testing to inform outbreak control

Objectives The aim of this paper was to describe the results of mass asymptomatic testing for COVID-19 in a male prison in England following the declaration of an outbreak. It provides novel data on the implementation of a mass testing regime within a prison during the pandemic. Methods The paper is an observational evaluation of the mass testing conducted for 6 months following the declaration of a COVID-19 outbreak within a prison. It investigated the incidence of positive cases in both staff and residents using polymerase chain reaction testing. Results Data from October 2020 until March 2021 was included. A total of 2170 tests were performed by 851 residents and 182 staff members; uptake was 48.3% for people living in prison and 30.4% for staff. Overall test positivity was 11.6% (14.3% for residents, 3.0% for staff), with around one-quarter of these reporting symptoms. The prison wing handling new admissions reported the second-lowest positivity rate (9.4%) of the eight wings. Conclusion Mass testing for COVID-19 over a short space of time can lead to rapid identification of additional cases, particularly asymptomatic cases. Testing that relies on residents and staff reporting symptoms will underestimate the true extent of transmission and will likely lead to a prolonged outbreak.


COVID-19 and prisons
Since the first cases were identified in 2019, COVID-19 has caused a global pandemic which has disrupted economies and caused significant morbidity and mortality. With COVID-19 predominantly spread through aerosols and those in close contact at the highest risk, concern was voiced that institutions such as prisons would be at risk of large outbreaks ( Burki, 2020 ), exacerbated by overcrowding and poor health of imprisoned people ( Davies et al . , 2020 ;Fazel et al . , 2001 ). Coupled with difficulties implementing basic infection prevention and control measures because of limited access to handwashing points, crowded conditions, little control over social distancing, and movement of both staff and people, it was expected that prisons would become a hotspot day, with limits placed on numbers unlocked and in exercise yards to enable social distancing. Mask-wearing was mandatory for staff and residents.
England introduced measures including specific cohort units for protective isolation, shielding units, and reverse cohorting ( Coleman et al . , 2022 ;O'Moore and Farrar, 2020 ), as well as compartmentalization: significantly reducing transfers between prisons ( Park et al . , 2021 ) and limiting to single-cell accommodation within the prison where possible, coupled with calls for early release or fewer custodial sentences ( Henry, 2021 ;Reinhart and Chen, 2020 ).
Outbreaks in prisons can be rapidly contained with effective outbreak management ( Wilburn et al . , 2021 ) and testing asymptomatic people to contain outbreaks and reduce further spread ( Lambert and Wilkinson, 2021 ;Malloy et al . , 2021 ). Mass asymptomatic testing has been implemented in particular contexts and settings, including in China ( Zhou et al . , 2021 ) and in establishments with a high risk of transmission, such as hospitals ( Reid et al . , 2020 ) and schools ( Torjesen, 2021 ). However, mass asymptomatic testing was not available until months after the start of the pandemic, and must consider frequency, the test used, and the accompanying public health measures, as well as balancing these with the disruption to the prison regime, the risk of harm from false negative and positive results, and the financial cost ( Lambert and Wilkinson, 2021 ).

Mass testing for COVID-19 in prisons in England
In outbreak situations from autumn 2020 until March 2022, testing guidance from the Ministry of Justice and Public Health England (now the UK Health Security Agency) was for both lateral flow tests and polymerase chain reaction (PCR) testing to be used ( UK Health Security Agency and Ministry of Justice, 2021 ), with all consenting residents tested regardless of whether or not symptomatic. PCR samples were processed by the local Lighthouse Laboratory using the ThermoFisher TaqPath TM COVID-19 test for the detection of SARS-CoV-2 ( Gravagnuolo et al . , 2021 ). The interval for this 'mass testing' was advised as follows: • At day 0 (the first day mass testing is available) • Between days 5 and 7 • On day 28, after the last confirmed or suspected case. This was changed to 14 days in January 2022.
Regular staff testing was also recommended to prevent the incursion of infection. Staff were encouraged to test twice weekly with lateral flow tests and once a week with a PCR test. This paper describes the results of a mass testing regime implemented in a male prison in the North West of England following the identification of a COVID-19 outbreak using the testing regime recommended above. It was implemented within a wider routine testing program for symptomatic residents, which is also reported here. This paper provides novel data on the operational implementation of a testing regime within an English prison during the COVID-19 pandemic.

Setting
This paper describes a COVID-19 outbreak in a local Category B (second-highest level of security) closed male prison in the North West of England with a capacity of 750 residents, holding people on remand and sentenced who have been admitted from the community via the courts or transferred in from other prisons. It is a Victorian radial design prison, organized into seven accommodation wings subdivided into spurs.

Participants
Eligible participants were all residents within the establishment and all staff members working at the facility. In addition to residents already held within the establishment at the start of the outbreak, this also included new arrivals from court and those transferred from other prisons.
Staff members included prison officers and those who worked in allied areas such as education, temporary staff employed by agencies, and those employed by external healthcare providers. Symptomatic staff were advised not to come to work and had the opportunity to undergo testing in the community outside of prison testing.
Those who had previously tested positive for COVID-19 in the 90 days before testing did not receive a further PCR test, in line with national guidance ( UK Health Security Agency and Ministry of Justice, 2021 ).

Design
This service evaluation investigated the incidence of positive cases during a COVID-19 outbreak in an English prison, following the recommended asymptomatic national testing regime for prisons at that time ( UK Health Security Agency and Ministry of Justice, 2021 ).

Materials
Following the declaration of an outbreak by the outbreak control team on October 21, 2020, mass testing capacity was mobilized as soon as was feasible and began on November 7, 2020. PCR testing was conducted using a nose and throat swab test on asymptomatic staff and residents.
PCR test results were linked to existing datasets: Prison National Offender Management Information System (p-NOMIS) for residents and human resources records for staff. These datasets held information on demographics and roles within the prison (for staff). Presence of COVID-19 symptoms was also recorded. The national COVID-19 case definition was used ( UK Health Security Agency, 2020 ).

Data analysis
In addition to the mass testing of residents, data were also collected from the wider, routine symptomatic testing program occurring concurrently and the asymptomatic staff testing program conducted from November 7. Symptomatology and laboratory results were reported into a central system linked to demographic data and exported to Excel. Data on residents were held within the p-NOMIS records system, and data on staff were held by the prison management system. Data were shared for analysis using a secure, encrypted method and stored on a secure server at Public Health England. Descriptive epidemiology and statistical tests were conducted using Excel. Chi-square test was used as independence of participants at each round of testing was assumed due to the rapid turnover of people living in the prison. Clopper-Pearson exact test was used to calculate 95% confidence intervals of testing positivity. Uptake of testing used the number of residents and the total number of staff in the establishment at the midpoint of the mass testing period as the denominators.

Ethics
Participation in the testing program was voluntary, and there were no repercussions for those who opted not to be tested. No ethical approval was required as this was a service evaluation which collected data to monitor the outbreak.

Results
The first positive test was recorded on October 13, 2020, with analyses including data from October 12, 2020, until March 20, 2021. This includes routine testing conducted on symptomatic prison residents and tests on asymptomatic prison staff who were participating in the routine staff testing program, in addition to the mass testing implemented following the outbreak declaration. A total of 2170 unique tests were performed, consisting of 1639 tests performed by 851 residents and 531 by 182 staff members. Results from the wider routine testing program for symptomatic residents are reported first, including the scheduled mass testing program, followed by the results of the mass testing program alone.
The number of tests conducted, response rate and test positivity by resident or staff is listed in Table 1 . The total number of positive tests was 252, giving an overall test positivity of 11.6%. Response rates are calculated from the average number of tests conducted monthly and use the monthly population statistics for residents (679) and staff members (349) from the midpoint of evaluation.
Positive tests performed by residents and staff are shown in Figure 1 . Gray arrows indicate the dates of the start of the scheduled mass testing as per the testing schedule and denote when testing was scheduled to take place.

Resident testing
Location details were available for 1205 (97.0%) tests carried out by residents ( Table 2 ). The highest test positivity was seen on G wing (24.6%), followed by B wing (20.9%) and D wing (16.9%). No positive tests were returned by residents on H wing.
Positive tests performed by residents, by location and over time are shown in Figure 2 . Although the majority of positive tests were initially from different wings, there was a clear peak on D wing around day 26 (the first day of mass testing as per the schedule), followed by a large number of positive tests from A wing on day 27. After this, the majority of infections appeared to be contained within B wing until the end of the asymptomatic mass testing, whereby it remained at a low level on C wing until the end of the period of evaluation. Two spurs of C wing were used as the reverse cohorting unit. Additional information on infection rates within spurs are shown in Table 1 , Supplementary Materials.

Staff testing
Data from staff testing by role ( Table 3 ) shows the highest test positivity was in prison officers (6.3%), followed by staff working in healthcare (2.7%). No positive tests were returned by those working for external agencies. Symptomatic staff would have been eligible for testing in the community as per national guidance.
The timing of staff tests is shown in Figure 3 . The number of tests was greatest following the start of the asymptomatic mass testing but remained relatively constant from around day 40 onwards. Positive tests were returned at all stages of the evaluation, with no obvious peaks. It should be noted that staff testing did not start until day 29 of the outbreak, 3 days after the asymptomatic mass testing for residents.
The positive tests by day of testing and divided into residents or staff ( Figure 1 ) show a steady number of positive tests, albeit a low number, until day 26. At this point, there was a sharp increase. Cases then dropped quickly, with a smaller peak on day 39    and a large number on day 59. The majority of cases were seen in residents.

Test positivity by testing round
As described previously, mass testing was also conducted in line with government recommendations, using days 0, 7, and 28. In this testing schedule, day 0 was set as November 7, 2020, day 7 as November 14, 2020, and day 28 as December 5, 2020, although there were some delays: up to 4 days to conduct mass testing for day 0 tests, up to 5 days for day 7 tests, and up to 13 days for day 28 tests. The number of tests conducted by day, and results, are detailed in Table 4 , which is a subset of the larger dataset described above. All test results reported here are in residents, and no staff were included.
The test positivity was highest on the first round of testing (22.8%), but this dropped sharply in the second round to 3.8% and was 4.2% in the final round. Significant differences were found using chi-square test of independence between the proportion of positive tests on day 0 and day 7 (X 2 = 54.10, P < 0.0 0 01) and between the proportion of positive tests on day 0 and day 28 (X 2 = 61.41, P < 0.0 0 01), but there was no significant difference observed between positive tests on day 7 and day 28 (X 2 = 0.18, P < 0.670).

Discussion
The pattern of COVID-19 cases is in keeping with a propagated outbreak, as seen in other custodial establishments ( Wilburn et al . , 2021 ). The increase in cases between days 26 and 30 corresponds to the first round of the mass testing scheduled as per the national guidance. This spike in cases shows how mass testing can lead to rapid identification of additional cases, particularly asymptomatic cases. The usual reproduction number of SARS-CoV-2 is just under three ( Billah et al . , 2020 ), but the Diamond Princess cruise ship outbreak reported a reproductive number of 14.8 ( Rocklöv et al . , 2020 ), and the relatively closed environment of prison can be seen to have some similarities to those on a cruise ship. Implementation of infection prevention and control (IPC) measures should prevent these scenarios ( O'Moore and Farrar, 2020 ,) but situations may arise when these are not possible. In the case of this prison, it may have been exacerbated by layout and environment, making IPC measures and good ventilation difficult.
Only about one in four cases had reported symptoms. This demonstrates that relying on residents and staff to report symptoms will underestimate the true extent of the outbreak, leading to further transmission and a prolonged outbreak, meaning that stricter IPC measures must be enforced for a longer time. This is likely to have a negative impact on the mental well-being of prison residents ( Johnson et al . , 2021 ).
If infections were introduced by new entrants, the highest rate of infection would have been seen on C wing, specifically spurs C1 and C2, which were used as the reverse cohorting unit where many imprisoned people would spend their first nights in the prison. However, the rate of infection on this wing was lower than all but one of the wings, suggesting asymptomatic infection within the establishment was unlikely to have been introduced by the new entrants to the prison. Staff are the most likely source of infection, with one study using sequencing information showing multiple introductions of the virus into the prison from the community ( Czachorowski et al., 2022 ). Although the positivity rate was slightly higher on C1 than other spurs on the same wing, C2 had a lower positivity rate in testing than another spur that was not used for reverse cohorting.
The national recommendations when this service evaluation took place stated that mass testing should take place as soon as an outbreak is declared ('day 0'), 7 days later ('day 7') and 28 days after the putative last case (this was later changed to 14 days after the last case). It can be difficult to adhere strictly to testing schedules, particularly where there are large numbers of people to test and short intervals between cycles. It is a huge logistical challenge to mobilize testing at short notice, particularly in a large national prison system and when there is competition for mass testing procedures. Furthermore, when there are nearly 10 0 0 people to be tested, and particularly where there are restrictions on movement, this can be difficult to undertake in a single day. However, it is a valuable tool in providing a systematic method of testing to identify cases where testing of symptomatic people or contacts will miss a large proportion of cases. In a prison environment, it may be more pragmatic to treat individual wings as separate environments, with each wing having its own schedule. However, data from this evaluation suggests that if one wing is affected, it is likely that there will be infected individuals across the whole prison.
Key challenges have been highlighted in the literature for mass testing in prisons: assumed participation of consent, test administration, and technological barriers ( Lambert and Wilkinson, 2021 ). With the implementation of national guidelines and increasing testing capacity within the UK, technological barriers had minimal impact, but test administration is likely to have been a significant challenge to the testing schedule and may explain testing needing to be implemented over several days. In England, UK Health Security Agency has worked with Her Majesty's Prison and Probation Service, National Health Services (NHS) England and NHS Improvement to enable the rapid mobilization of mass testing units to help the local services in conducting the necessary testing.
Another issue is that of uptake: our evaluation showed poor response rate to testing. It should be noted that the figure reported for residents (48.3%) was likely to have been affected by the transient population of the prison. This would not have affected staff members, whose uptake was even lower (30.4%). This is a key issue for both prevention of infection incursion and outbreak control; measures such as reverse cohorting can mitigate the risk of introducing infection from incoming residents, but staff are likely to introduce infection from the community and can introduce infection to several areas of the prison estate ( Kinner et al . , 2020 ). The number of staff entering and leaving the prison into the community is also much greater than the number of residents entering and leaving the prison. Infection rates in prison staff have been found to more closely mirror that of the prison they work in than their local community ( Nowotny et al . , 2021 ), and so asymptomatic testing needs much higher uptake and coverage of staff members to be effective. It is not surprising that this outbreak occurred at a time when the prevalence of COVID was highest in North West of England ( 2020 ), and also covers the period where the Alpha variant became the dominant variant throughout England, which may have exacerbated the outbreak further, though cases in the prison do not mirror the regional trend in North West England at this time (see Figure 4 ). This may suggest that establishments such as prisons may amplify the spread of viruses even before marked peaks in community transmission are observed, as has been described in other prison outbreaks ( Duarte et al . , 2022 ;Henry, 2021 ;Vicente-Alcalde et al . , 2022 ), with prisons being described as "epicenters of transmission to the community" in some circumstances ( Henry, 2021 ).

Strengths and limitations
This paper is the first in the published literature to evaluate the recommendation of testing on days 0, 7, and 28 to provide a structured testing regimen during a COVID-19 outbreak in prison. It has shown that despite logistical challenges to implementation, particularly in larger establishments with transient populations, it is possible to implement a testing protocol to detect asymptomatic cases in a methodical manner and implement control measures. The dataset had a good level of completeness, allowing robust evaluation of testing. The only exception was relating to symptomatology. There is a risk that residents would under-report the presence of symptoms as this may lead to greater restrictions on their movement and more time in isolation. For staff, there may be a perceived lack of incentive to report symptoms (or participate in testing) as they would be unable to work extra hours and earn overtime, or there may be repercussions for the staff member's household contacts who may have to isolate in line with the guidance at that time.
Additional questions would have been useful, particularly to identify reasons for those who opted out of testing, as this would allow concerns to be addressed and could increase the uptake of testing. This information would also be valuable in investigating any possible selection bias, as it was not possible to say whether uptake was higher among particular groups. Likewise, in those being tested, additional questions around hand washing and respiratory etiquette, and the number of daily contacts, would have been useful to examine whether certain behaviors are linked to asymptomatic cases, and if additional measures could be put in place to reduce the risk of onward transmission. This would also need to take into account the prison regime.
The results presented in this service evaluation are purely those from the symptomatic testing program delivered from October 12, 2020, and the asymptomatic mass testing from November 7, 2020. During this time, different results systems were in place to record test results depending on the testing method: any staff member who was symptomatic would have had the opportunity to undertake PCR testing in the community, and any residents who were transferred to a hospital would have received tests in secondary care. Due to this, not all cases are necessarily captured in this dataset.

Implications of the paper
This evaluation has shown that it is possible to conduct a mass testing program within a wider program of testing for infectious diseases in custodial environments as part of outbreak management. While there can be operational constraints to administering mass testing over a large population, mass testing at set intervals can provide valuable information about the presence of infection within a closed setting and attempt to bring it under control quicker. Given the high proportion of asymptomatic infection in this population, mass testing of the whole prison is recommended to ensure that all cases are identified. Without the identification of such cases and the institution of appropriate IPC measures, the outbreak is likely to be prolonged, with asymptomatic cases acting as a reservoir for spread across the prison and potentially into the community through transmission to prison staff.

Declaration of competing interest
The authors have no competing interests to declare.

Funding
This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

Ethical approval
This paper reports a service evaluation and as such, it is exempt from ethical approval.

Author contributions
EP, MC and CB were responsible for design and ongoing development. CB handled the data and drafted the manuscript. MC provided guidance on data management and feedback on initial drafts of the manuscript. EP and EF provided feedback on the manuscript, with EP and EM responsible for final review.

Supplementary materials
Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.ijid.2022.10.018 .