• Neal, David E.
• Verne, Julia E. C. W.
• Group, The Cap Trial
• Down, Liz
• Thorn, Joanna
• Turner, Emma
• Hounsome, Luke
• Martin, Richard
• Hamdy, Freddie C.
• Noble, Sian
• Donovan, Jenny L.
• Walsh, Eleanor

Abstract

<p><strong>Objectives</strong> To evaluate the accuracy of routine data for costing inpatient resource use in a large clinical trial and to investigatecosting methodologies.</p><p><strong>Design</strong> Final-year inpatient cost profiles were derived using (1) data extracted from medical records mapped to the National HealthService (NHS) reference costs via service codes and (2) Hospital Episode Statistics (HES) data using NHS reference costs.Trust finance departments were consulted to obtain costs for comparison purposes.</p><p><strong>Setting</strong> 7 UK secondary care centres.</p><p><strong>Population</strong> A subsample of 292 men identified as having died at least a year after being diagnosed with prostate cancer in Cluster randomisedtriAl of PSA testing for Prostate cancer (CAP), a long-running trial to evaluate the effectiveness and cost-effectivenessof prostate-specific antigen (PSA) testing.</p><p><strong>Results</strong> Both inpatient cost profiles showed a rise in costs in the months leading up to death, and were broadly similar. The differencein mean inpatient costs was £899, with HES data yielding ∼8% lower costs than medical record data (differences compatiblewith chance, p=0.3). Events were missing from both data sets. 11 men (3.8%) had events identified in HES that were all missingfrom medical record review, while 7 men (2.4%) had events identified in medical record review that were all missing from HES.The response from finance departments to requests for cost data was poor: only 3 of 7 departments returned adequate data setswithin 6 months.</p><p><strong>Conclusions</strong> Using HES routine data coupled with NHS reference costs resulted in mean annual inpatient costs that were very similar tothose derived via medical record review; therefore, routinely available data can be used as the primary method of costingresource use in large clinical trials. Neither HES nor medical record review represent gold standards of data collection.Requesting cost data from finance departments is impractical for large clinical trials.</p><p><strong>Trial registration number</strong> ISRCTN92187251; Pre-results.</p>

Topics

• impedance spectroscopy
• cluster
• gold