Understanding the pattern of length of stay in institutional long-term care has important practical implications in the management of long-term care. Furthermore, residents' attributes are believed to have significant effects on these patterns. In this paper, we present a model-based approach to extract, from a routinely gathered administrative social care dataset, high-level length-of-stay patterns of residents in long-term care. This approach extends previous work by the authors to incorporate residents' features. Two applications using data provided by a local authority in England are presented to demonstrate the potential use of this approach.