Appropriate estimation of staff costs for economic evaluations: a case study in haemodialysis

Abstract

The thesis examines methods to measure or attribute resource use and costing for economic evaluations in health care. The literature review found minimal evidence comparing top-down and bottom-up (micro) costing. To cost nursing inputs for patients, studies rarely measured staff time and poorly reported methods. In chronic haemodialysis (HD), 'case mix' variations in nursing between patients were ignored.

The empirical work evaluated nurses' self-recording using barcode scanners and observer work sampling to measure the nursing time per patient. Initial piloting eliminated patient-level work sampling due to problems linking data to patients. Barcode scanning captured 80% of nurses' hours; data quality was acceptable. It covered 4 weeks for 169 patients. Costs, in 2006, included employers' National Insurance and superannuation.

Relative to the 'top-down' nursing expenditure per HD session (£44.56 to £50.79), the bottom-up cost was underestimated by up to 10%: 4% due to the unit cost using expected rather than actual working hours, and 6% due to missing patient-level resource use data. Multiple linear regression clustered by patient found those ineligible for care at satellite units needed extra nursing input (mean 8 minutes, 95% Cl 4-11, or £2.30 to £7.22 per session) compared with those eligible.

Conclusions were that top-down (expenditure based) and bottom-up estimates of staff costs cannot reconcile due to averaging at different points, their attribution of resource use or costs to patients, and valuation of unit costs. More guidance is required on which unit cost of staff time (per hour paid, worked or patient-related) best reflects the opportunity cost of staff time. Barcode scanning successfully captured data, but required considerable research effort, making it impractical for most multicentre studies. Cost differences between patients were 5-14% of the nursing cost per session or 1-5% of the overall cost per session. Hence, they had minimal effect on results of economic evaluations.

More information

Identifiers

ORCID for Paul Roderick: orcid.org/0000-0001-9475-6850

Catalogue record

Export record