We present a visual comparison of survival rates among centres that is based on a graphical display known as a funnel plot (1, 2). This display is used to show how consistent the rates of the different transplant units are with the national rate.
Funnel plots show the risk-adjusted survival rate plotted against the number of transplants for each centre, with the overall national unadjusted survival rate (solid line), and its 95% (thin dotted lines) and 99.8% (thick dotted lines) confidence limits superimposed. Each dot in the plot represents one of the centres.
Note that many patients return to local renal units for follow-up care after their transplant and although we report survival according to transplant unit, patients may in fact be followed up quite distantly from their transplant centre.
1. One-year graft survival rates for deceased donor transplants in adult patients, 2002 – 2006
2. One-year patient survival rates for deceased donor transplants in adult patients, 2002 – 2006
3. Five-year graft survival rates for deceased donor transplants in adult patients, 1998 – 2002
4. Five-year patient survival rates for deceased donor transplants in adult patients, 1998 – 2002
5. One-year graft survival rates for live donor transplants in adult patients, 2002 – 2006
6. One-year patient survival rates for live donor transplants in adult patients, 2002 – 2006
7. Five-year graft survival rates for live donor transplants in adult patients, 1998 – 2002
8. Five-year patient survival rates for live donor transplants in adult patients, 1998 – 2002
9. One-year graft survival rates for deceased donor transplants in paediatric patients, 2002 – 2006
10. One-year patient survival rates for deceased donor transplants in paediatric patients, 2002 – 2006
11. Five-year graft survival rates for deceased donor transplants in paediatric patients, 1998 – 2002
12. Five-year patient survival rates for deceased donor transplants in paediatric patients, 1998 – 2002
13. One-year graft survival rates for live donor transplants in paediatric patients, 2002 – 2006
14. One-year patient survival rates for live donor transplants in paediatric patients, 2002 – 2006
15. Five-year graft survival rates for live donor transplants in paediatric patients, 1998 – 2002
16. Five-year patient survival rates for live donor transplants in paediatric patients, 1998 – 2002
Interpreting the funnel plots
If a centre lies within all the limits, then that centre has a survival rate that is statistically consistent with the national rate. If a centre lies outside the 95% confidence limits, this serves as an alert that the centre may have a rate that is significantly different from the national rate. If a centre lies outside the 99.8% limits, then further investigations may be carried out to determine the reasons for the possible difference. When a centre lies above the upper limits, this indicates a survival rate that is higher than the national rate, while a centre that lies below the lower limits has a survival rate that is lower than the national rate. It is important to note that adjusting for patient mix through the use of risk-adjustment models may not account for all possible causes of centre differences. There may be other factors that are not taken into account in the risk-adjustment process that may affect the survival rate of a particular centre.
The funnel plot for five-year patient survival rates (Figure 4) shows that one of the centres, Newcastle, had a rate that was significantly lower than the national average for transplants undertaken during the time period shown (1998 – 2002). An investigation showed that unlike any of the other centres, the patient survival rate for Newcastle (unadjusted rate: 76%, 95% CI 71 – 80%; risk-adjusted rate: 79%, 95% CI 69 – 89%) was lower than their graft survival rate (unadjusted rate: 81%, 95% CI 76 – 85%; risk-adjusted rate: 82%, 95% CI 72 – 92%), a phenomenon associated with an excess of deaths with a functioning graft.
The Newcastle Renal Transplant Unit became aware of the excess death rate a number of years ago and in 2000, after investigation, introduced a revised cardiac screening programme for transplant eligibility to address the issues identified.
The other funnel plots show that, for the most part, the centres lie within the confidence limits. None of the centres lie outside the lower 99.8% confidence limits for any of the survival rates, indicating that none of the centres have survival rates that are significantly lower than the national rate. Some of the funnel plots show some centres to be above the upper 99.8% confidence limit. This suggests that these centres may have survival rates that are considerably higher than the national rate.
References
- 1. Tekkis PP, McCulloch P, Steger AC, Benjamin IS, Poloniecki JD. Mortality control charts for comparing performance of surgical units: validation study using hospital mortality data. British Medical Journal 2003; 326: 786 - 788.
- 2. Stark J, Gallivan S, Lovegrove J, Hamilton JRL, Monro JL, Pollock JCS, Watterson KG. Mortality rates after surgery for congenital heart defects in children and surgeons' performance. Lancet 2000; 355: 1004 - 1007.