CBRNe events in closed and controlled environments: an analysis of COVID-19 outbreaks in Italian prisons

The Italian management of Covid-19 in prisons caused considerable concern at the beginning of the pandemic, when many riots resulted in inmate deaths, damage and prison escapes. The aim of this study is to shed some light on this issue by analysing the infection (prevalence of infection and number of reproductive units over time (R(t))) and disease parameters (prevalence of hospitalisation and incidence) in Italian prisons over the period from the second to the fourth wave of the outbreak. The first part of this study focuses on aggregated data collected from all Italian prisons and therefore represents averages of many and varied realities. The second part of the analysis, on the other hand, focuses on the data collected from each prison, with the aim of identifying the ancillary variables that may favour the spread of the virus in these places. R(t) and hospitalisation were calculated using an Eulerian approach applied to differential equations derived from compartmental models (SIR model). Comparisons between trends were made using paired t-test and linear regression analysis. Correlation analysis between a selected dependent variable, the max R(t), and "predictive" variables, such as number of staff, prison overcrowding, staff epidemiological parameters, type of prison and presence/absence of health isolation wards, was performed using linear single and multiple regression analysis. The infection trends (infected prevalence and R(t)) show a high correlation between the prison population and the external community. Both indices seem to lag one week in prison. The R(t) values of prisoners are not statistically different from those of the general population. The hospitalisation trend for prisoners is highly correlated with that of the outside population with a lag of two weeks. The magnitude of hospitalisation in prison was smaller than in the external community. Moreover, the maximum R(t) values are poorly correlated with the overcrowding rate of each prison and the same was found when examining other variables such as the type of prison and the presence/absence of health isolation units. On the contrary, there is a significant correlation when the independent variable is the presence of staff. In conclusion, the number of infections was higher in prisons than outside, but the R(t) values did not show a significant difference. In addition, the hospitalisation rate was lower in prison. The results suggest that the higher infection prevalence in prison could be due to the constant surveillance of inmates and the lower hospitalisation rate could be related to the earlier start of the vaccination campaign. All three indices examined had a lag of one to two weeks in prison. This delay could provide a useful window of opportunity to intensify planned countermeasures. These plans could include an intensification of all measures involving the testing of staff and the use of PPE in prisons.