Hidden selection bias deriving from donor organ characteristics does not affect performance evaluations of kidney transplant centers.

BACKGROUND

Transplant center performance evaluations have garnered substantial attention in recent years. Among sources of bias that may affect measured performance are underlying characteristics of donor organs. An unresolved question is whether centers accepting higher-risk donations are placed in jeopardy for lower evaluations independent of actual quality of care.

OBJECTIVE

The primary aim was to assess whether unmeasured characteristics of donor organs impact risk-adjusted outcomes used for center performance evaluations.

SUBJECTS

The study included adult kidney transplant recipients (n = 53,791) from 1994 to 2008 from a national registry.

RESEARCH DESIGN

We compared adjusted graft survival with use of paired-donor kidneys (allocated to high- and low-performing centers) and unpaired donor kidneys to investigate whether measured center performance was consistent with organs derived from the same donor (minimizing the influence of noncodified risk factors).

RESULTS

The primary finding was that differences between centers were unaffected by use of paired or unpaired donations (hazard ratio for patients transplanted at high performing centers with paired kidneys = 0.63 [95% CI, 0.53-0.74] and with unpaired kidneys = 0.66 [95% CI, 0.62-0.70], P value for interaction = 0.52). This finding was consistent over 5 consecutive cohorts, based on either concurrent or prospective outcomes and by altering the threshold criteria for identification of performance outliers.

CONCLUSIONS

Results indicate that underlying selection bias from donor characteristics does not impact transplant center evaluations. This is important evidence that donor selection is not a primary driver for evaluated quality of care among transplant centers and acceptance of higher-risk kidneys should not be perceived as a primary threat to measured performance.