*Conclusion of the UKCTA report*

86 Perioperative Considerations in Cardiac Surgery

*3.5 The rate of PAEs 7 within 30 days was significantly higher at Harefield than other UK centres: 67.2% (95% CI 53.7%-79%) for the whole period compared to 44.9% (38.3%-51.7%) elsewhere. In 2008, the difference was more marked: 93.3% (95 CI 68-99.8%) at Harefield and 42% (28.1%- 56.8%) elsewhere. The UKCTA report commented that the high PAE rate "may be due, at least in* 

*3.6 The median total ischaemia time for the whole period was similar but slightly lower at Harefield than elsewhere: 196 minutes (interquartile range 175-218) compared to 220 minutes (190-251) for the other centres combined. The corresponding figures for 2008 were 213 (160-229) and 227 (192- 263). Transport and implant times were also slightly lower at Harefield. These times were stable at* 

*3.7 The risk profile of heart transplant patients at Harefield was higher in 2008 than in 2007 and 2006 (Figure 1). The risk model has been developed by the UKCTA from a dataset that extends back 13 years and is used to adjust for factors found to be associated with an increased risk of early death after heart transplant. The risk factors are diabetes, reduced creatinine clearance (renal function),* 

*3.8 Figure 2 shows that the proportion of transplant patients with individual risk factors (except* 

*3.9 The proportion of patients, who at the time of transplant had been receiving inotrope drugs to improve heart contraction or mechanical circulatory support by ventricular assist devices (VADs), extracorporeal membrane oxygenation (ECMO) and/or intra-aortic balloon pump (IABP), was significantly higher at Harefield than at other centres: 93% compared to 45% (p=0.002). These* 

*3.10 The UKCTA report commented that "the data reported to the audit suggests that the majority of patients who underwent transplantation were high risk; only one of the patients transplanted* 

*3.11 Risk adjustment using the UKCTA risk model reduced the difference between the observed and expected number of deaths within 30 days at Harefield and at other UK centres in 2008. The centre effect estimate for Harefield reduced by more than 50% with risk adjustment and the UKCTA analysis found that "although mortality at Harefield remained 52% higher than expected using the risk model, the number of cases was low and Harefield was not identified as significantly* 

*3.12 The sequence of deaths in July and August 2008 was enough to cause the more sensitive technique of sequential monitoring using cumulative sum (CUSUM) method to signal an alert for* 

*this calendar year was not on inotropes, a VAD, ECMO or IABP at transplant".* 

*part, to the risk profile of patients transplanted" (see paragraphs 3.7-3.11 below).* 

*previous open heart operation, older donor age and longer ischaemic times.* 

*for ischaemic time) was higher in 2008 than the previous two years.* 

*Post-operative adverse events (PAE)* 

*Harefield during 1 January 2006 - 31 August 2008.* 

*factors are not included in the UKCTA risk model.* 

*Risk-adjusted mortality* 

*Sequential monitoring of mortality* 

*divergent."* 

*Risk profile of heart transplant patients* 

*Ischaemic time* 

*3.13 "The number of operations reviewed is small as [heart] transplantation in the UK is a low volume procedure, with only around 288 transplants carried out nationally over the 32 months of the review. As a consequence differences and changes in mortality, which may appear large, are not necessarily identified as significant from a statistical perspective. Nonetheless, the rise in 30-day mortality at the end of the series was sufficient to trigger an alarm using the tabular CUSUM methodology. The rise in early mortality seen at Harefield can be explained, at least in part, by the risk profile of the patients, the majority of whom were high risk."*

The additional statistical analyses were able to inform the review process and highlight the issues when basing analyses on very small numbers. They also highlight the importance of robust and appropriate case-mix adjustment which is yet another challenge we face in highly complex services that involve small numbers of cases.
