A Novel Use of Latent Class Analysis to Identify Patterns of Workplace Hazards among Informally Employed Domestic Workers in 14 Cities, United States, 2011-2012

Introduction Few studies, mostly descriptive, have quantitatively analyzed the working conditions of domestic workers (DWers) informally employed by private households in the USA. These workers are explicitly or effectively excluded from numerous workplace protections, and scant data exist on their exposures or how best to categorize them. Methods We analyzed data from the sole nationwide survey of informally employed US DWers with work-related hazards data, conducted by the National Domestic Workers Alliance, the University of Illinois Chicago Center for Urban Economic Development, and the DataCenter in 14 US cities (2011-2012; N = 2086). We used exploratory latent class analysis to identify groups of DWers with distinct patterns of exposure to 21 self-reported economic, social, and occupational workplace hazards (e.g. pay violations, verbal abuse, heavy lifting). We then used multinomial logistic latent class regression to examine associations between workers' individual, household, and occupational characteristics and latent class membership. Results Among the 2086 DWers, mean age was 42.6 years, 97.3% were women, 56.0% Latina/o, 26.5% White, 33.2% undocumented immigrants, and 11.7% live-in. 53.5%, 32.0%, and 14.5% primarily worked doing housecleaning, child care, and adult care, respectively. 49.9% of workers reported >= 3 hazards. Latent class analysis identified four groups of DWers doing: 'Low hazard domestic work' (lowest exposure to all hazards), 'Demanding care work' (moderate exposure to pay violations [item response probability (IRP) = 0.42] and contagious illness care [IRP = 0.39]), 'Strenuous cleaning work' (high exposure to cleaning-related occupational hazards, such as climbing to clean [IRP = 0.87]), and 'Hazardous domestic work' (highest exposure to all but one hazard). Covariates were strongly associated, in many cases, with latent class membership. For example, compared to other DWers, DWers doing 'hazardous domestic work' had the largest predicted probability of being economically insecure (0.53) and living-in with their employers (0.17). Conclusions Results indicate that informally employed US DWers experience distinct patterns of workplace hazards, and that it is informative to characterize DWers' exposures to different sets of multiple hazards using latent class analysis.