Biases in describing residents in long-term residential aged care

In New Zealand, no reliable information describes use of long-term residential aged care. Instead, when estimating use, records of government subsidy payments are upscaled to adjust for private payers. This paper assesses consequential bias in reporting use of long-term care and considers the implications.

Data from OPAL, a census-type survey of residents of aged-care facilities in Auckland in 2008, linked to routinely-collected hospitalisation, mortality and subsidy data from national databases. Demographic, functional and service use characteristics of unsubsidised residents were compared to subsidised.

Records of 5961 OPAL residents aged 65+ years were matched with subsidy data; 25% were unsubsidised. In low-level care (51% of all), unsubsidised residents had similar care needs to subsidised residents, but were 1.7 years older on average (p<0.001) with shorter length of stay. In high-level care (41% of all), unsubsidised residents had significantly lower care needs on six different measures and were less likely to die during the follow-up period. Upscaling yields undercounts at all care levels.

National reports derived from current upscaling methods undercount residents. Stratification by region and age group would improve estimates. Age and care needs are misrepresented. Population policies that depend upon upscaled counts should, where possible, ascertain the biases introduced.