Background and objective The publication of clinical outcomes for consultant surgeons in 10 specialties within the NHS has, along with national clinical audits, highlighted the importance of measuring and reporting outcomes with the aim of monitoring quality of care. Such information is vital to be able to identify good and poor practice and to inform patient choice. The need to adequately adjust outcomes for differences in case-mix has long been recognised as being necessary to provide ‘like-for-like’ comparisons between providers. However, directly comparing values of the standardised mortality ratio (SMR) between different healthcare providers can be misleading even when the risk-adjustment perfectly quantifies the risk of a poor outcome in the reference population. An example is shown from paediatric intensive care.

Methods Using observed case-mix differences for 33 paediatric intensive care units (PICUs) in the UK and Ireland for 2009–2011, SMRs were calculated under four different scenarios where, in each scenario, all of the PICUs were performing identically for each patient type. Each scenario represented a clinically plausible difference in outcome from the reference population.

Results Despite the fact that the outcome for any patient was the same no matter which PICU they were to be admitted to, differences between the units were seen when compared using the SMR: scenario 1, 1.07–1.21; scenario 2, 1.00–1.14; scenario 3, 1.04–1.13; scenario 4, 1.00–1.09.

Conclusions Even if two healthcare providers are performing equally for each type of patient, if their patient populations differ in case-mix their SMRs will not necessarily take the same value. Clinical teams and commissioners must always keep in mind this weakness of the SMR when making decisions.