Review 1: "Dynamic Contact Networks of Residents of an Urban Jail in the Era of SARS-CoV-2"

Ashley Hazel

doi:doi:10.1162/2e3983f5.827f71d1

Dynamic Contact Networks of Residents of an Urban Jail in the Era of SARS-CoV-2

by Samuel M. Jenness, Karina Wallrafen-Sam, Isaac Schneider, Shanika Kennedy, Matthew J. Akiyama, and Anne C. Spaulding

Show Description
Published on Oct 02, 2023
www.medrxiv.org

Description

ABSTRACT Background In custodial settings such as jails and prisons, infectious disease transmission is heightened by factors such as overcrowding and limited healthcare access. Specific features of social contact networks within these settings have not been sufficiently characterized, especially in the context of a large-scale respiratory infectious disease outbreak. The study aims to quantify contact network dynamics within the Fulton County Jail in Atlanta, Georgia, to improve our understanding respiratory disease spread to informs public health interventions.Methods As part of the Surveillance by Wastewater and Nasal Self-collection of Specimens (SWANSS) study, jail roster data were utilized to construct social contact networks. Rosters included resident details, cell locations, and demographic information. This analysis involved 6,702 residents over 140,901 person days. Network statistics, including degree, mixing, and turnover rates, were assessed across age groups, race/ethnicities, and jail floors. We compared outcomes for two distinct periods (January 2022 and April 2022) to understand potential responses in network structures during and after the SARS-CoV-2 Omicron variant peak.Results We found high cross-sectional network degree at both cell and block levels, indicative of substantial daily contacts. While mean degree increased with age, older residents exhibited lower degree during the Omicron peak, suggesting potential quarantine measures. Block-level networks demonstrated higher mean degrees than cell-level networks. Cumulative degree distributions for both levels increased from January to April, indicating heightened contacts after the outbreak. Assortative age mixing was strong, especially for residents aged 20–29. Dynamic network statistics illustrated increased degrees over time, emphasizing the potential for disease spread, albeit with a lower growth rate during the Omicron peak.Conclusions The contact networks within the Fulton County Jail presented ideal conditions for infectious disease transmission. Despite some reduction in network characteristics during the Omicron peak, the potential for disease spread remained high. Age-specific mixing patterns suggested unintentional age segregation, potentially limiting disease spread to older residents. The study underscores the need for ongoing monitoring of contact networks in carceral settings and provides valuable insights for epidemic modeling and intervention strategies, including quarantine, depopulation, and vaccination. This network analysis offers a foundation for understanding disease dynamics in carceral environments.

RR:C19 Evidence Scale rating by reviewer:

Strong. The main study claims are very well-justified by the data and analytic methods used. There is little room for doubt that the study produced has very similar results and conclusions as compared with the hypothetical ideal study. The study’s main claims should be considered conclusive and actionable without reservation.

***************************************

Review: This is an important paper—clear in its objectives, straightforward in its analysis, and highly valuable in its potential impact. The authors found that the residential setting of an urban jail setting was highly conducive to SARS-CoV-2 transmission in terms of key network characteristics. The authors found that key network features, such as degree (number of contacts), heterophily (assortative mixing), and contact dissolution (change in residential location), were supportive of SARS-CoV-2 transmission. This is particularly important because jails are limited in what they can do to minimize transmissible contacts in overcrowded populations, especially if reducing inflow (i.e., new inmates) is not sufficient for stopping transmission within the facility. By comparing network statistics during and after an outbreak, this paper also evaluates how much interventions that directly affect the network—e.g., quarantine—were undertaken and what further steps could be taken, given the constraints of an overcrowded jail system.

By evaluating contact patterns at the cell and block levels, this paper reveals both the frequency and intensity of contact in an incarcerated setting. Frequency and intensity are distinct features of social contact that are often conflated in transmission models. Providing a data-informed approach to parameterize both is an important contribution to the public health of incarcerated populations.

Moreover, the authors utilized data sources that are commonly generated by prison systems, such as multi-level resident location (cell, block), resident demographics, and, in the event of outbreaks, case and vaccination data. Therefore, this analysis is based on commonly available, high resolution spatial information, and their approach can be adopted by other incarceration systems to improve surveillance and prevention for future outbreaks. Furthermore, this paper can serve as a reference for future models of respiratory virus transmission that require more realistic parameterization of contact patterns in congregate-living settings.

The authors acknowledge certain limitations, like absence of prison staff data. These data are less readily available, and while they are important for studies of virus importation and outbreak initiation, their absence from this evaluation does little harm to the primary objects, which are to understand the social structures of residents that support potential SARS-CoV-2 transmission.

While I really appreciate the overall simplicity and clarity of the paper, I wonder if it would not be worthwhile to offer evaluations of whether network characteristics shown in Tables 2–4 are statistically different according to stratifications laid out in the tables? Some basic univariate statistics for degree and perhaps a log-linear model for mixing? This would help give a stronger sense of whether any types of residents are at greater structural risk for exposure than others.

Overall, this is an important paper and I look forward to seeing it published and referencing it in the future.