Estimating resource use

Observational databases are a valuable source of information on resource use. Hospital Episode Statistics (HES) provides data on casemix, treatments administered, length of stay and discharge destination. Linking records across databases or to trial data can enhance the scope and value of analysis. These resources have the advantage of size and ability to capture patient pathways in routine care. However, follow-up of recently treated patients is limited, potentially biasing estimates of mean costs for mortal diseases. Traditional solutions, including using Kaplan-Meier survival estimators to weight the costs of survivors, can be difficult to implement in a regression framework. Multiple Imputation (MI) provides an alternative approach which easily accommodates adjustment for covariates. Our research has applied MI in the analysis of five year secondary care costs for patients with Endometrial Cancer utilising data from HES (the HENS study).

Resource use data collected in trials often have limited follow-up period and lack information on intensity of hospital resource use. We have used data from the Case Mix Programme (CMP) co-ordinated by the Intensive Care National Audit & Research Centre (ICNARC), to estimate level and intensity of long-term resource use and costs beyond the trial period. Our research has estimated long-term resource for various critical care areas including severe sepsis, septic shock, traumatic brain injury, and hyperglycaemia.

Key people: Zia Sadique, Mark Pennington, Richard Grieve

Comments are closed.