Background While there is evidence to suggest that household overcrowding is associated with higher levels of psychological distress, associations with serious mental illness (SMI) or depression remain unclear. We used a measure of household overcrowding derived from linked primary care and housing records across North East London and hypothesised that adults living in overcrowded households were more likely to be diagnosed with SMI or depression after accounting for individual, household and area-level factors.

Methods We linked primary care records of adults currently registered with all general practices in the North East London region on 21/03/2021 to housing data using pseudonymised unique property reference numbers (pUPRNs). Adults sharing the same address on this date, estimated from shared pUPRNs, were assigned to the same household, giving a study sample of 762,718 adults in 329,948 households. Primary outcomes were a primary care coded diagnosis of SMI or depression. Overcrowding was defined as <32.5m² per person. We used logistic regression models with robust standard errors to estimate the adjusted odds (aOR) and 95% confidence intervals (CI) of SMI or depression when living in an overcrowded household, adjusting for individual (age, sex, ethnic group), household (size, composition), and area (Index of Multiple Deprivation [IMD]) characteristics. We conducted sensitivity analyses by excluding diagnoses made before 21/3/2016. Analyses were conducted in RStudio.

Results SMI and depression were diagnosed in 6,492 (0.9%) and 45,053 (5.9%) adults respectively; 584,515 (76.6%) adults lived in overcrowded households. In univariable models, household overcrowding was inversely related to risk of SMI or depression (aOR[CI]: 0.42 [0.39,0.44]; 0.62 [0.61,0.64], with no change in direction of effect after adjustment for age, sex and ethnic group. The direction of effect reversed on adjustment for household size and composition: 1.34 [1.23,1.46] and 1.12 [1.09, 1.15] for SMI and depression respectively, with some attenuation after adjustment for area-level deprivation (1.22 [1.12,1.33]; 1.07 [1.04,1.10] respectively). Similar results were obtained in the sensitivity analyses.

Conclusion In this ethnically diverse and disadvantaged population, we found small but significant associations between SMI and depression and household overcrowding, after adjusting for individual, household- and area-level factors. We used a novel method of record linkage to estimate household overcrowding using space per person but were not able to establish causality of observed associations in this cross-sectional study. Overcrowding is only one aspect of housing quality. More research is needed to understand the importance of housing conditions for mental and physical health.