A rapid review of early guidance to prevent and control COVID-19 in custodial settings

Background With over 11 million people incarcerated globally, prevention and control of COVID-19 in custodial settings is a critical component of the public health response. Given the risk of rapid transmission in these settings, it is important to know what guidance existed for responding to COVID-19 in the early stages of the pandemic. We sought to identify, collate, and summarise guidance for the prevention and control of COVID-19 in custodial settings in the first six months of 2020. We conducted a systematic search of peer-reviewed and grey literature, and manually searched relevant websites to identify publications up to 30 June 2020 outlining recommendations to prevent and/or control COVID-19 in custodial settings. We inductively developed a coding framework and assessed recommendations using conventional content analysis. Results We identified 201 eligible publications containing 374 unique recommendations across 19 domains including: preparedness; physical environments; case identification, screening, and management; communication; external access and visitation; psychological and emotional support; recreation, legal, and health service adaptation; decarceration; release and community reintegration; workforce logistics; surveillance and information sharing; independent monitoring; compensatory measures; lifting control measures; evaluation; and key populations/settings. We identified few conflicting recommendations. Conclusions The breadth of recommendations identified in this review reflects the complexity of COVID-19 response in custodial settings. Despite the availability of comprehensive guidance early in the pandemic, important gaps remain in the implementation of recommended prevention and control measures globally, and in the availability of evidence assessing their effectiveness on reducing COVID-19 disease, impact on people in custody and staff, and implementation. Supplementary Information The online version contains supplementary material available at 10.1186/s40352-021-00150-w.


Background
With over 11 million people incarcerated on any given day and an estimated 30 million people released from custody each year globally (Penal Reform International, 2020b; United Nations Office on Drugs and Crime, 2013), the prevention and control of COVID-19 in custodial settings is a critical component of the public health response. However, custodial settingsincluding prisons, jails and police cells, youth detention, immigration detention, and forensic psychiatric facilitiespresent a multitude of challenges for the prevention and control of COVID-19. These settings are often characterised by overcrowding, poor ventilation, inadequate access to sanitation, and substandard access to, and quality of, healthcare relative to the community (Dolan et al., 2016;Penal Reform International, 2020b). Furthermore, infectious diseases can be easily transmitted between people in custody, staff, and visitors through facility transfers and staff cross-deployment, and to and from the community via intakes and releases. These conditions make custodial settings high-risk environments for COVID-19 transmission, and subsequent community spread (Beaudry et al., 2020;Penal Reform International, 2020b).
Custodial settings concentrate marginalised populations with disproportionately high rates of mental illness (Fazel et al., 2016), substance dependence (Fazel et al., 2006), communicable (Dolan et al., 2016) and non-communicable disease (Herbert et al., 2012), intellectual disability (Fazel et al., 2008), and multimorbidity (Kinner & Young, 2018;Penal Reform International, 2020b;World Health Organization Europe, 2014). People in custody are therefore more likely than people in the general population to be susceptible to severe COVID-19 disease (World Health Organization and the United Nations Development Programme, 2020). However, they are also highly vulnerable to the deleterious physical and mental health impacts of intensified and sustained confinement, which are common to COVID-19 prevention and control measures. Strategies that restrict freedoms and meaningful social interaction, such as facility lockdown and isolation, increase psychological distress and adverse outcomes for people in custody, with particularly profound impacts on those with pre-existing mental illness Stewart et al., 2020). These are important considerations for COVID-19 response in custodial settings, which must balance obligations to protect the health and human rights of people in custody (United Nations Human Rights Office of the High Commissioner, 1985; United Nations Office on Drugs and Crime, 2011; United Nations Office on Drugs and Crime, 2015) with the need to minimise COVID-19 morbidity and mortality.
Recognising these vulnerabilities and their implications for public health, the COVID-19 pandemic prompted a rapid influx of published guidance to prevent the introduction and transmission of COVID-19 in custodial settings. Based on this guidance, governments and correctional authorities were tasked to quickly mount a response to COVID-19 within justice and immigration detention systems that, in many settings, historically operated in isolation from community and public health sectors. To date, there has been no systematic effort to collate and summarise these initial recommendations and to assess the extent to which clear guidance was available in the crucial early stages of the pandemic. This is important to identify areas requiring clarification or additional guidance, and to inform evaluation and research. We therefore aimed to identify, collate, and summarise guidance for the prevention and control of COVID-19 in custodial settings in the first six months of 2020.

Overview
We conducted a rapid review of peer-reviewed and grey literature to identify guidance for COVID-19 prevention and control in custodial settings. We defined custodial settings as inclusive of prisons, jails and police cells, youth detention settings, immigration detention settings, and forensic psychiatric facilities. We registered a protocol with the International Prospective Register of Systematic Reviews (PROSPERO; CRD42020191735).

Information sources
We searched the following databases: Medline (Ovid), PsycINFO, Embase, Web of Science, CINAHL, Global Health (CABI), Criminal Justice Abstracts, LILACS, Lit-Covid, and Google Scholar. We searched an additional three databases indexing grey literature: WorldWi-deScience, TRIP Database, and Google Search Engine. The first 20 pages of Google results were screened to identify relevant publications. We manually searched COVID-19 'information hubs' that indexed resources relevant to custodial settings, websites of organisations with an interest in the health or human rights of people in custody, and reference lists.

Search strategy
Our search strategy is detailed in Appendix S3. We used variants and combinations of search terms relating to two key concepts: COVID-19, and custodial settings. We used all common variants of COVID-19, based on a preliminary literature search. We included terms related to the criminal justice system, deprivation of liberty, and specific types of custodial settings. Terms relating to 'recommendations' or 'guidance' were used to focus Google results. All peer-reviewed database searches were conducted on 1 July 2020, capturing publications up to and including 30 June 2020. Grey literature data sources were searched between 1 July 2020 and 10 July 2020.

Eligibility criteria
Eligible publications had to outline recommendations for the prevention and/or control of COVID-19 in custodial settings and be published in English. To ensure a feasible number of grey literature publications, we limited eligibility to those published by federal and state/ provincial/territorial governments, international organisations (e.g., United Nations), specialised intergovernmental agencies (e.g., World Health Organization [WHO]), and organisations with an interest in the health and/or human rights of people in custody. We also excluded from grey literature news articles and blog posts; opinion pieces and commentaries not published in peerreviewed journals (to act as a quality control via the peer-review process); publications focusing on policing measures; and legal documents including proceedings, summaries, and case reports.

Selection process
Initial eligibility of grey literature identified from Google and manual searches was decided by one reviewer (LP or JF) at the time of the search, by applying a priori eligibility criteria. References were uploaded to Covidence systematic review management software (Covidence, 2021) and duplicates removed. Title and abstract screening were conducted by two independent reviewers (CK, LP). If an abstract or summary was not available, the reviewer conducted a brief full-text screening to assess eligibility. Disagreements in eligibility were resolved through discussion with a third reviewer (LCF). Full-text review was conducted by two independent reviewers (CK, LP, AV) and disagreements in eligibility were resolved through consensus with a third reviewer. The selection process is illustrated in Fig. 1.

Quality assessment
No formal quality assessment was applied due to our focus on extracting recommendations rather than assessing results.

Analysis
We developed a data extraction form to collate information on all included publications (Appendix S2). We conducted a conventional content analysis to systematically examine text and develop domains within which to categorise recommendations (Hall & Steiner, 2020;Hsieh & Shannon, 2005). To inductively develop an initial coding framework, we purposively sampled 14 publications that comprehensively provided recommendations across key custodial settings and population subgroups. Three authors (CK, LP, AV) independently reviewed and coded each publication. Independent codes were combined and grouped into domains to develop a preliminary coding framework, which was reviewed and refined by coauthors. An additional 17 publications were coded against the interim framework to identify gaps and facilitate further refinement into a final coding framework (Table 2). Each remaining publication was coded by one author (CK, LP, AV) and any uncertainties were resolved by consensus with a second author. Recommendations were summarised by domain and sub-domain. We did not code detailed guidance for the clinical management of COVID-19 cases, testing, and contact tracing, unless recommendations were specific to custodial settings. We used NVivo 12 (QSR International, 2021) for coding and document storage.

Domain 1: planning and preparedness
Early intervention was recommended to quickly identify and respond to COVID-19 outbreaks. Guidelines and checklists for facility-level planning and preparedness (Justice and Corrections Service, 2020; World Health Organization, 2020a) were developed and recommended. Outbreak management plans that identify steps for rapid case identification, isolation, and treatment; communication plans that facilitate rapid decision making; and contingency plans for staff shortages were recommended. This included collaboration with other custodial facilities for joint surveillance and between-facility transfers (Communicable Diseases Network Australia, 2020; US Centres for Disease Control and Prevention, 2020a; US Centres for Disease Control and Prevention, 2020b; US Centres for Disease Control and Prevention, 2020c), and with local health services to support regional contact tracing and medical case management (Communicable Diseases Network Australia, 2020; Kinner et al., 2020;Penal Reform International, 2020a;Wang et al., 2020;Yang & Thompson, 2020).
Domain 2: creating safer physical environments 'Cohorting', the practice of restricting movement between groups of people in custody (e.g., housing units, close contacts and confirmed cases, or medically vulnerable people from the remainder of the custodial population), was recommended to increase biosecurity (AMEND, 2020b; Association for the Prevention of

Summary of guiding principles for the prevention and control of COVID-19 in custodial settings
Immediate action to reduce prison population density is needed to address widespread overcrowding in correctional settings, and must be applied with adequate transition planning to facilitate safe reintegration into the community (Alohan & Calvo, 2020; American Academy of Pediatrics, 2020; Amnesty International, 2020a; Annie E. Casey Foundation, 2020; Council of Europe Commissioner for Human Rights, 2020aRights, , 2020bHenry, 2020;Piel, 2020;Rubenstein, 2020;Simpson & Butler, 2020;Sivashanker et al., 2020; Terres des hommes, 2020; The Alliance for Child Protection in Humanitarian Action, 2020; UNICEF, 2020; World Organization Against Torture, 2020c) A sustainable response to the COVID-19 pandemic requires carceral system reform rooted in health equity to address ongoing crises of overcrowding, poor living conditions, and substandard health care in custodial settings (Alohan & Calvo, 2020;Crowley et al., 2020;Minkler et al., 2020; The Alliance for Child Protection in Humanitarian Action, 2020) 2020; Penal Reform International, 2020a) (see 'Remote access & visitation' in Appendix S1). Other measures that protect the emotional and social wellbeing of people in medical isolation, including prevention of violence and discrimination towards suspected or confirmed cases , were recommended to convey the non-punitive nature of treatment, encourage symptom reporting and early healthcare intervention,  2020). It was recommended that decisions to restrict inperson visitation recognise the diverse roles of visitors, including the provision of money, food, and other essential supplies; two publications recommended that custodial authorities adapt protocols so that these resources continue to safely reach people in custody (Amnesty International, 2020a; Amnesty International, 2020b; Association for the Prevention of Torture, 2020).

Domain 7: psychological and emotional support
The protection of the mental and emotional health of people in custody and staff was a recurrent theme. It was recommended that additional psychological support and compensatory measures be made available during periods of sustained restriction or isolation (see Domain 16 'Compensatory measures'). Several publications (Knox, 2020;Kothari et al., 2020;Wang et al., 2020;World Health Organization, 2020b; World Organization Against Torture, 2020a) recommended that custodial settings develop capacity to monitor stress, burnout, and fatigue among staff, and/or counteract these through enhanced, no-cost psychological support programs and opportunities for debriefing with colleagues. Domain 9: adapting recreation, programming, and services Educational, vocational, social, and religious programs can help to prepare people in custody for successful integration into the community (León et al., 2020) and reduce recidivism (Pyrooz et al., 2020). In circumstances where these programs cannot be adapted to meet infection prevention and control standards, it was recommended that steps be taken to compensate for suspended vocational programs and provide electronic entertainment and social activities, online education, and virtual religious services (Council of Europe, 2020; Innovative Prison Systems, 2020; León et al., 2020). In accordance with human rights standards (United Nations Human Rights Office of the High Commissioner, 1985; United Nations Office on Drugs and Crime, 2011; United Nations Office on Drugs and Crime, 2015), it was asserted that outdoor access not fall below a minimum of 1 hour per day (United Nations Office on Drugs and Crime (UNODC), 2020; World Health Organization, 2020b).

Domain 10: adapting legal services and processes
Continued functioning of courts and access to legal services, including the establishment of emergency courts (Inter-Agency Standing Committee, 2020), was recommended to support decarceration (Association for the Prevention of Torture, 2020; Tahrir Institute for Middle East Policy (TIMEP) and Middle East and North Africa (MENA) Rights Group, 2020; The Alliance for Child Protection in Humanitarian Action, 2020; The International Legal Foundation, 2020; United Nations, 2020) by reducing numbers of unsentenced people held in pretrial detention -currently over 3 million people globally (Penal Reform International, 2020b). In contrast, one publication recommended the temporary suspension of judicial hearings to reduce transmission, with exception of remote hearings for urgent cases (Innovative Prison Systems, 2020). Virtual court hearings were recommended with careful consideration of due process, data security, and the vulnerabilities of children and people with a disability (United Nations Office on Drugs and Crime, 2020a). An important consideration was that adaptations to legal proceedings must not compromise the right to a fair trial and the safety of defendants, witnesses, and victims (United Nations Office on Drugs and Crime, 2020a).

Domain 11: Decarceration
Decarceration strategies (Henry, 2020) were widely recommended to reduce prison overcrowding (  Wurcel et al., 2020). Recommended non-custodial measures to reduce detention at pre-trial, sentencing, and post-trial are detailed in Appendix S1.

Domain 12: release and community re-integration
Post-release transitional support was recommended to ensure that people leaving custody during the pandemic are able to access health, social, and housing services that allow them to comply with local public health advice (Montoya- Barthelemy et al., 2020;Mukherjee & El-Bassel, 2020; United Nations Network on Migration, 2020). Pre-release needs assessment and transitional planning were recommended for all people leaving custody, particularly for those at highest risk of harm postrelease, including people with substance use disorders, mental illness, chronic illness, and housing instability (Gorman & Ramaswamy, 2020). It was recommended that custodial authorities carefully weigh the benefits of early release with capacity for transitional planning and support (Piel, 2020;Shepherd & Spivak, 2020).

Domain 13: workforce logistics
Publications emphasised the protection of staff safety and wellbeing (Openshaw & Travassos, 2020

Domain 15: independent monitoring and inspection
The World Health Organization (WHO) stated that COVID-19 "must not be used as a justification for objecting to external inspection of prisons and other places of detention"(p.6) (World Health Organization, 2020b). It was recommended that monitoring and inspection continue with due caution regarding infection prevention and control (Association for the Prevention of Torture, 2020; International Corrections and Prisons Association (ICPA), 2020), attention towards the justified and appropriate use of medical isolation and lockdown measures (Commonwealth Human Rights Initiative, 2020), and increased implementation of remote reporting and complaint mechanisms for people in custody and staff (International Corrections and Prisons Association (ICPA), 2020; New South Wales Government, 2020; United Nations Human Rights Office of the High Commissioner, 2020a; United Nations Human Rights Office of the High Commissioner, 2020b; World Organization Against Torture, 2020a). Guidance for remote inspection was developed (International Corrections and Prisons Association (ICPA), 2020).

Domain 16: compensatory measures
Several publications recommended that COVID-19 prevention and control measures that further restrict the freedoms of people in custody be offset by compensatory measures that maintain the rehabilitative qualities of custody and an acceptable quality of life (e.g., increase frequency and/or time allowances for phone calls; access to virtual education, vocational programs, and free psychological support services). A comprehensive list of recommended compensatory measures is provided in Appendix S1.  Control and Prevention, 2020c). Few concrete recommendations were made for the termination of prevention and control measures such as cohorting and changes to recreation, programming, and services. Recommendations for lifting restrictions on inperson visitation were context-dependent; they ranged from once screening and containment policies are in place (AMEND, 2020b), to after an outbreak is declared over (Communicable Diseases Network Australia, 2020), to as long as COVID-19 remains prevalent in the community (European Centre for Disease Prevention and Control, 2020a; United Nations Human Rights Office of the High Commissioner, 2020d). Recommendations for terminating medical isolation are provided in Appendix S1.

Domain 18: learning systems and evaluative frameworks
Data-driven evaluation and policy analysis were recommended to assess the effectiveness of response measures in reducing COVID-19 infection, to understand their impacts on the health and human rights of people in custody and staff, and to identify beneficial policy changes to adopt in standard operations Buchanan et al., 2020;Communicable Diseases Network Australia, 2020;Dalton et al., 2009;Martyn et al., 2020;Nature, 2020; New Zealand Office of the Ombudsman, 2020). Regional and international knowledge sharing was recommended to inform future planning and response for similar health crises (AMEND, 2020b; Communicable Diseases Network Australia, 2020; United Nations Office on Drugs and Crime, 2020a) and to advocate for broader carceral system reform (Inter-Agency Standing Committee, 2020; Mukherjee & El-Bassel, 2020; United Nations Office on Drugs and Crime, 2020a; World Organization Against Torture, 2020a). It was recommended that both people in custody and staff be involved in evaluation Buchanan et al., 2020;Gagnon, 2020).

Domain 19: key populations and settings
We identified several recommendations that were specific to key populations (women, elderly, Indigenous peoples, people with a disability, people who use drugs, and people with mental illness) and custodial settings (youth detention, immigration detention, forensic psychiatric facilities, and in low-middle income countries [LMIC]) (Appendix S1). Notably, this review found fewer recommendations specific to LMIC (Amnesty International, 2020a; Amnesty International, 2020b; European Centre for Disease Prevention and Control, 2020a; FIA-CAT, 2020), which may require more pragmatic interventions that account for resource constraints, and forensic psychiatry (Innovative Prison Systems, 2020; , where release may not be possible.

Discussion
From over 200 eligible publications, we documented a considerable volume of recommendations to prevent and/or control COVID-19 in custodial settings during the first six months of 2020. In total, we identified 374 unique recommendations spanning 19 domains; each domain represented a distinct and important area to consider for a comprehensive COVID-19 response. We determined that, overall, comprehensive guidance was available. However, no individual publication addressed all identified domains. Numerous publications called for immediate reductions in the number of people incarcerated, and a moratorium on immigration detention to address pre-existing crises of widespread overcrowding in custodial settings (Alohan & Calvo, 2020; American Academy of Pediatrics, 2020; Amnesty International, 2020a; Amnesty International, 2020b; Annie E. Casey Foundation, 2020; Council of Europe Commissioner for Human Rights, 2020a; Council of Europe Commissioner for Human Rights, 2020b;Henry, 2020;Piel, 2020;Rubenstein, 2020;Simpson & Butler, 2020;Sivashanker et al., 2020; Terres des hommes, 2020; The Alliance for Child Protection in Humanitarian Action, 2020; UNICEF, 2020; World Organization Against Torture, 2020b). These recommendations build on a strong history of advocacy to reduce prison overcrowding globally (Penal Reform International, 2019;World Health Organization, 2007), with COVID-19 starkly revealing the health and human rights implications of overcrowding. National prison systems in 124 countries exceed their maximum occupancy rate, with 22 countries reporting that their prisons contain over twice as many people as they were designed to house (Penal Reform International, 2020b). Overcrowding to this degree renders physical distancing and other infection prevention and control measures near impossible, severely inhibits access to healthcare, and undermines all aspects of COVID-19 response (Penal Reform International, 2020b). It was therefore unsurprising that several publications reinforced that the COVID-19 pandemic provides an unprecedented opportunity for carceral system reform to reduce overreliance on mass incarceration and create safer physical environments that enable the prevention and control of infectious disease (McKenzie & Mishori, 2020;Minkler et al., 2020). With a strong focus on prison depopulation during the COVID-19 pandemic, efforts to reduce populations in custody need to be matched with adequate supports that facilitate safe community reintegration after release (Franco-Paredes et al., 2021;Gorman & Ramaswamy, 2020;Montoya-Barthelemy et al., 2020;Mukherjee & El-Bassel, 2020;Piel, 2020;Shepherd & Spivak, 2020;United Nations Network on Migration, 2020).

Conflicting recommendations
Four main conflicts in recommendations were identified. First, one guideline recommended the temporary suspension of judicial hearings with exception of remote hearings for urgent cases, a recommendation implemented in Brazil, Latvia, the Netherlands, and Pakistan (Innovative Prison Systems, 2020). However, no information was provided to outline when judicial hearings should be resumed or how to distinguish 'urgent' cases from 'non-urgent' cases. This contrasted with the majority of publications on this topic calling for the continuation of all hearings during the pandemic (Association for the Prevention of Torture, 2020; Tahrir Institute for Middle East Policy (TIMEP) and Middle East and North Africa (MENA) Rights Group, 2020; The Alliance for Child Protection in Humanitarian Action, 2020; The International Legal Foundation, 2020; United Nations, 2020), including the immediate establishment of emergency courts to reduce the number of unsentenced people held in pre-trial detention (Inter-Agency Standing Committee, 2020). Second, we identified mixed recommendations regarding in-person visitation. The restriction of all non-essential visitors from entering custodial settings was recommended in several publications on this topic Innovative Prison Systems, 2020;Penal Reform International, 2020a; Prison department of the Republic of Lithuania, 2020; Seal, 2020  ; World Health Organization, 2020b) discouraged its use due to potentially harmful mental health impacts and preference towards comprehensive testing and screening. Lastly, three publications recommended the use of fines as an alternative to incarceration (Danish Institute Against Torture, 2020a; Danish Institute Against Torture, 2020b; Seal, 2020; The International Legal Foundation, 2020) whereas others expressed concerns about the disproportionate impacts of these measures on Indigenous peoples (National Aboriginal & Torres Strait Islander Legal Services, 2020b) and people living in poverty (United Nations Office on Drugs and Crime (UNODC), 2020).

Key gaps identified
Definitive guidance on when to remove or reduce the intensity of COVID-19 prevention and control measures was notably absent, except for terminating medical isolation and declaring an outbreak to be over. While there was clear consensus that these decisions should be based upon close monitoring of the local epidemiological context and public health advice, no specific benchmarks (e.g., local infection rates) were provided in any of the included publications. Furthermore, there was limited guidance over who should be responsible for these decisions; designated officials (US Centres for Disease Control and Prevention, 2020b), correctional authorities (Danish Institute Against Torture, 2020a; Danish Institute Against Torture, 2020b; United Nations Office on Drugs and Crime, 2020d), and outbreak management teams (Communicable Diseases Network Australia, 2020) were identified. Clear and externally verifiable criteria to guide the de-escalation of COVID-19 prevention and control measures in custody are needed to ensure that they are not in place for longer than necessary. This is especially pertinent for measures that restrict personal freedoms (e.g., out-of-cell time) and social interaction (e.g., visitation), due to their profound impacts on the psychological wellbeing of people in custody Stewart et al., 2020). Another gap related to guidance for specific populations. We developed a subdomain within the 'Key populations and settings' domain, when coding recommendations that were targeted for specific population subgroups. Recommendations for some groups were notably absent from this domain (e.g., LGBTQ people), were categorised broadly (e.g., Indigenous peoples), or provided minimal guidance (e.g. people with a disability). Guidance that addresses the specific needs of all people in custody and acknowledges local diversity and cultural differences within the broad subgroups identified in this review are needed. Finally, to our knowledge, few publications actively included the perspectives of people in custody, their families, or staff in the development of recommendations.

Strengths and limitations
This review included a wide range of recommendations spanning multiple custodial settings and population subgroups, identified from a systematic search of both peerreviewed and grey literature (Appendix S3). Our broad approach allowed for the inclusion of guidance from various stakeholders represented in the COVID-19 response, and consideration of broader social, physical, and mental health needs of people in custody and staff. However, there are some limitations. First, although English is an official language of key international bodies that contributed 34% of included publications, our decision to restrict the search strategy to English-language publications may have led to exclusion of some publications from LMIC. Second, our search strategy targeted publications explicitly focused on COVID-19 and custodial settings. Recommendations relevant to custodial settings that were made in broader pandemic response publications were therefore not included in this review. Third, this review focused on initial guidance from the first 6 months of the pandemic. It therefore did not include recommendations from emergent areas such as COVID-19 vaccination or variants of concern. Recommendations identified in this review reflect information at the time of their publicationwhen governments around the world were out of necessity mounting a rapid response to the pandemicand should be considered within the context of evolving COVID-19 knowledge.

Implementation and next steps
Comprehensive guidance is necessary, but not sufficient, for effective COVID-19 response. Custodial settings and systems vary widely in their structure, stability, resource availability, and cultural context, and are impacted by external factors including political will and the broader public health response to COVID-19, which greatly influence the actions of governments and correctional authorities. Despite the availability of comprehensive guidance early in the pandemic, early reports have suggested that the global response to COVID-19 in custodial settings has largely been inadequate. A recent review of 69 countries' response to COVID-19 in custodial settings identified shortages of testing capacity, lack of preventive and protective measures, insufficient rates of release to address overcrowding, and inappropriate use of solitary confinement (Amnesty International, 2021).
We have identified several priorities to address this implementation gap that align with recommendations identified in this review. Firstly, there is an urgent need for reliable data collection, analysis, and public disclosure on basic epidemiology (e.g., numbers of infections and deaths) and responses (e.g., measures implemented) in custodial settings globally (Amnesty International, 2021). This information is rarely reported, but is critical to maximise transparency and hold governments and custodial authorities accountable for the health of people in custody and staff. Secondly, a comprehensive programme of research documenting and examining the implementation of recommended prevention and control measures in custodial settings, with due consideration to priority groups and LMIC, is needed to identify factors facilitating and inhibiting an effective COVID-19 response. To date, we are not aware of any published implementation studies. A fit-for-purpose Optional Protocol to the Convention Against Torture (OPCAT) (United Nations Human Rights Office of the High Commissioner, 2002) could act as a framework to assess the adequacy of the COVID-19 response from a human rights perspective. Finally, there remains a dearth of research evidence examining the effectiveness of recommended prevention and control measures, and combinations of measures, for reducing COVID-19 morbidity and mortality and associated impacts on people in custody and staff. This represents a critical knowledge gap that will be essential to optimising and adapting responses in custodial settings globally. The framework presented in this review (Table 2) is a useful starting point for organising this important body of work.

Conclusions
A comprehensive response to COVID-19 in custodial settings is highly complex and must carefully balance the need for restrictive infection prevention and control strategies with their potentially harmful impacts on health and human rights. Despite the availability of comprehensive guidance early in the pandemic, important gaps remain in the implementation of recommended prevention and control measures globally, and in the availability of evidence assessing their effectiveness on reducing COVID-19 disease, impact on people in custody and staff, and implementation. Evaluation of the implementation of these measures in custodial settings, and their diverse impacts on health and wellbeing, are crucial next steps.