Article Text

PDF
How to adjust for case-mix when comparing outcomes across healthcare providers
  1. Shalini Santhakumaran
  1. Correspondence to Shalini Santhakumaran, Neonatal Data Analysis Unit, Imperial College London, 4th Floor Lift Bank D, Chelsea and Westminster Hospital, 369 Fulham Road, London SW10 9NH, UK; s.santhakumaran{at}imperial.ac.uk

Statistics from Altmetric.com

The reporting of clinical outcomes is essential for quality improvement, patient choice and accountability, all of which are becoming increasingly important within healthcare systems across the world. Comparisons of healthcare measures across regions of England are already freely available through the National Health Service (NHS) Atlas of Variation in Healthcare,1 and individual cardiac surgeons working in the NHS have had survival rates following surgery published for several years.2 Variation in outcomes can be due to many factors, so the ‘signal’ (variation due to the factors we are interested in) needs to be identified separately from the ‘noise’ (other variation). One major source of noise is a difference in patient case-mix. Patients are not randomised to providers, so some providers will have sicker patients or more challenging cases than others. It is vital to adjust for case-mix in order to make fair comparisons of outcomes across hospitals, regions or practitioners.

Table 1 shows some fictional data on neonatal mortality of very preterm infants from two neonatal units over 1 year, along with data from the whole region over the same period. Unit A is a neonatal intensive care unit treating the sickest infants whereas Unit B has a case-mix more similar to the region as a whole. A manager is investigating variation in mortality rates across …

View Full Text

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.