11 PRA & Patient Safety: A Trial Application to the Risks of Organ Mismatches in Transplantation (PSAM-0323)

In 2003, the Federal agency chartered with supporting the development of improvements in healthcare safety, the Agency for Healthcare Research & Quality (AHRQ), supported the trial application of probabilistic risk assessment (PRA) in healthcare applications. One such study was the application of PRA to examine the effects of changes to the protocols used to ensure that incompatibilities did not occur between the blood types of donated organs and the organ recipients, as had happened in a widely publicized case involving a young girl who died after receiving wrong blood-type lungs and heart in February 2003.

Performing the PRA involved a very extensive exploration of the processes involved in the selection and matching of organs to donors, including numerous visits to organ procurement organizations and hospitals, to observe and understand the processes involved in preventing incompatible implanting of organs, and to understand how these have changed over time, from before the event of February 2003 to completion of the study in December 2004.

This study identified several features of the transplant process that are not obvious. First, the greatest ‘sin’ in transplantation is to waste a ‘good’ organ, since so few are available compared with the number of potential recipients. There is an extensive computer-based process to match available organs with already identified candidates (including confirmation of compatibility of blood types), but sometimes no match is found after an extensive search. In such a situation, a process called an ‘open offer’ can be made, where transplant centers are asked if they can use such an organ. It was found that under such conditions, several of the barriers against transplanting incompatible organs were bypassed without it being recognized as such. Several changes in response to the event in February 2003 have been made that have reduced the likelihood of an incompatible implant by about two orders of magnitude, as assessed in the PRA.

PRA has yet to be embraced by the healthcare industry, which relies still on simple models of accidents—particularly single-failure models embodied in tools like FMEAs—and a belief in the ‘skill of the craft’ and personal accountability. However as this study showed, the path to failure was much more complex that could ever be captured by a tool like FMEA, and did not involve any personal failures on the part of the surgical teams (even though the head surgeon took personal responsibility for the outcome). Analyses using PRA allows the complexity of events to be identified and explored, and can show how ‘simple fixes’ like adding another blood type test immediately prior to implanting an organ can have the effect of potentially increasing the likelihood of organ failure through increasing the ischmic time of the organ—the time it is without blood flow outside of the body.

However, it is recognized that, while the tools used in performing a PRA look relatively simple (as they are), it is the acquisition of the deep technical knowledge of the subject area and its representation in the PRA models that is time consuming. This study took about 4 staff-months to perform over a period of about 10 months, of which 3.5 staff months were spent understanding the processes involved and representing them in the PRA.