A social network analysis approach to assess COVID19-related disruption to substance use treatment and informal social interactions among people who use drugs in Scotland.

Abstract

To assess the extent of Coronavirus-related disruption to health and social care treatment and social interactions among people with lived or living experience of substance use in Scotland, and explore potential reasons for variations in disruption. Cross sectional mixed methods interview, incorporating a social network 'egonet interview' approach asking about whether participants had interactions with a range of substance use, health, social care or third sector organisations, or informal social interactions. Five Alcohol and Drug Partnership Areas in Scotland. 57 (42% women) participants were involved in the study, on average 42 years old. Five-point Likert scale reporting whether interactions with a range of services and people had gotten much better, better, no different (or no change), worse, or much worse since COVID19 and lockdown. Ratings were nested within participants (Individuals provided multiple ratings) and some ratings were also nested within treatment service (services received multiple ratings). The nested structure was accounted for using cross classified ordinal logistic multilevel models. While the overall average suggested only a slight negative change in interactions (mean rating 2.93), there were substantial variations according to type of interaction, and between individuals. Reported change was more often negative for mental health services (Adjusted OR = 0.93 95% CI 0.17,0.90), and positive for pharmacies (3.03 95% CI 1.36, 5.93). The models found between-participant variation of around 10%, and negligible between-service variation of around 1% in ratings. Ratings didn't vary by individual age or gender but there was variation between areas. Substance use treatment service adaptations due to COVID19 lockdown led to both positive and negative service user experiences. Social network methods provide an effective way to describe complex system-wide interaction patterns, and to measure variations at the individual, service, and area level.