The report by Makary and Daniel from the Surgery Department at Johns Hopkins University (BMJ, May 3, 2016) states that Medical Errors are the third leading cause of death in the U.S. One of their ideas for improving the situation is “better reporting.” “Better reporting” is always a good idea. But who does the reporting? And to whom do they report? Most of the proposals for better reporting give short shrift to these questions.
For example, messengers Makary and Daniel offer the specific recommendation of adding a field to the death certificate that would allow doctors to list “medical error” as the cause of death. While the idea represents a step in the right direction, the truth is it is dead on arrival. Why would a provider be any more responsive to this new field when they have had a sub-optimal response to the other fields? The fact is many providers are already overwhelmed by the vast array of metrics that have become a key part of their jobs. Adding one more is not likely to work. We need an approach that not only focuses on the right things, but also makes providers’ lives easier. That’s the only kind of solution that will actually be adopted.
To be clear, hospitals and physicians want to—and often go to great lengths to—measure their results. But they face two obstacles. First, they face the “fox & henhouse” problem. When data is self-reported there is a tendency to minimize negative outcomes and maximize positive ones. Even the best organizations struggle with this. So getting an independent assessment of outcomes is key.
The second obstacle is that many deaths happen post-discharge and are difficult to track. In an era where inpatient lengths-of-stay have been remarkably compressed from 20-30 years ago, many complications of care including deaths are not identified until the post-discharge period. Surveillance by hospitals and physicians of those post-discharge events are very difficult. As has been reported in another study from Johns Hopkins, 20-40% of 90-day readmissions following discharge occur at hospitals other than the site of the index operation (Kim et al, Annals of Surgery, 2015). Our own research has identified that more deaths occur in the 90-days following discharge in Medicare patients than occurred during the inpatient length-of-stay for many surgical procedures.
What is really needed is a state-based, all-payer database with encrypted patient identifiers that permits the actual tracking of care outcomes over time. Political and legal wrangling makes this an unlikely opportunity any time soon. So here is the next best thing.
Establish National Benchmarks for Performance
National benchmarks for hospital performance offer a fair and effective way to improve outcomes on both a local and national scale. The way they would work is simple. Establish national public benchmarks for performance, then compare individual hospital performance to those benchmarks. This would not only allow patients to make educated choices about where they receive care. It would also allow hospitals to know how they compare with their peers. The risk of self-reporting would still exist. But transparency would bring more scrutiny. Public transparency will also bring more competition, which would ensue would make the entire system more cost-effective. Such a system isn’t intended to focus on solely reducing medical errors, but it would surely have that effect.
Making it a Reality
How would you build such a system? We have found that the Medicare database serves as an important tool to assess risk-adjusted outcomes across the continuum of care. It obviously includes only Medicare patients, but for most hospitals this can prove to be a meaningful sampling of care outcomes among the highest risk population of patients likely to be seen.
We identify four elements in particular to focus on for the national benchmarks: inpatient deaths, major inpatient complications of care, 90-day or longer post-discharge deaths, and post-discharge readmissions for relevant causes associated with the index hospitalization. By developing national risk-adjustment models for each of these four elements, each hospital that meets minimum volume thresholds can be evaluated by observed-to-expected outcomes. Each hospital’s performance can then be gauged against national benchmarks for each category of case that is studied. Preventable adverse outcomes can be measured by comparing best performing-to-suboptimal performing hospitals based upon risk-adjusted metrics.
The end result is a simpler system that reduces the administrative burden on providers while increasing the focus on the outcomes that matter most. But why wait for a national system to get up and running? MPA can do this analysis now. Can we build this system for you?