What key lessons can be gleaned from the implementation challenges encountered by the PSML grantees? Perhaps first among these involves the time window that is likely to be needed in any future implementation and evaluation effort examining malpractice claiming effects. As we describe above, and in several of our previous evaluation reports to AHRQ, malpractice claiming and adjudication can often take 5 or more years from initial claim to resolution. As a result, it is fundamentally unrealistic to implement new clinical or policy interventions for patient safety that will impact meaningfully on malpractice claims within a 3-year timeframe. Realistically, 6 to 10 years might be needed for undertaking this kind of activity and for rigorously documenting its impact. Acknowledging and accommodating the “real-world” timeframes in the planning and design of follow-on PSML efforts would be a major step forward.
A second lesson involves recognizing the multifaceted relationship between patient safety and medical malpractice: PSML projects may need to target a specific element of that relationship to be directly comparable with one another. For example, one theoretical connection between patient safety and malpractice involves improving patient safety outcomes by reducing preventable injuries in hospitals. In principle, doing so may also reduce malpractice claiming. This is one plausible way to intervene in the PSML environment, and it has a corresponding set of implications for the types of interventions that might be fielded and for the performance data that would need to be collected to rigorously assess outcomes. By contrast, an entirely different set of PSML interventions might be designed around reducing malpractice claiming directly, as through disclosure and resolution efforts, on the assumption that a less litigious malpractice climate would then help to stimulate more effective root cause analysis and patient safety improvement efforts. These interventions, too, are intuitively reasonable, but they are also qualitatively different, and different kinds of performance data are needed to assess them. Practically speaking, each type of intervention that might be undertaken as a PSML activity involves a specific underlying conceptual relationship between patient safety and medical malpractice, and each entails correspondingly different data collection and analytic approaches to evaluate it. For future initiatives under the PSML portfolio, it would be desirable to articulate the policy focus and research aim more clearly at the outset, so that the potential for an “apples-to-apples” comparison of findings is improved or the potential for an “apples-to-oranges” comparison of findings across interventions is better recognized and planned for from the beginning.
A third lesson is that currently available national data resources are poorly aligned to support future PSML initiatives, particularly so on the malpractice claims side. As we have discussed previously at some length, the National Practitioner Data Base has serious weaknesses when it comes to using claims data for assessing the impact of PSML projects.1 Perhaps the most important of these for national evaluations is the fact that NDPB claims data are only geocoded at the State level in the public-use data file and therefore cannot readily be linked to specific hospitals and other host organizations in connection with PSML interventions.
For hospital systems and other health care entities that undertake PSML-style interventions in the future, one of the basic planning steps involves considering the outcome metrics to track and the data to generate that will be useful and appropriate in context. The upside is that some relevant performance metrics for at least some types of PSML projects may be relatively easy to track locally and are not dependent on these national data pipelines to collect (e.g., numbers of claims and/or time to resolution of claims in the context of a DRP intervention).