Never Say Never: Knowledge Discovery for Clinical Quality Improvement
Here’s a cheery thought for the holiday season: according to new estimates, every week a surgeon in the US performs surgery on the wrong body site 20 times, performs the wrong procedure another 20 times, and leaves a foreign object in a surgical patient 39 times! Events like these are termed “never events” in healthcare oversight circles because, in fact, they should NEVER occur. What is even scarier is that these statistics are based on a study of paid malpractice settlements and judgments reported to the National Practitioner Data Bank, a source widely believed to under-represent actual events.
To be fair, the National Library of Medicine states that 15 million people a year have surgery in the US, so 4,000 never events is an incredibly small fraction of the total, unless you happen to be one of the people who die or are injured. In the scheme of things, there are lots of other opportunities for quality improvement that might make a more significant difference in overall patient outcomes and the health of our population.
The Institute for Healthcare Improvement (IHI), one of the most prominent voices in the quest for better healthcare quality, has spent more than two decades helping organizations identify systemic process issues that lead to suboptimal care. They like to talk about the “No needless list,” that is, no needless deaths, pain or suffering, helplessness in those served or serving, unwanted waiting, waste, and no one left out. Information technology is only one component of that ongoing process improvement effort.
Why am I interested, besides the fear we all share that one day we might be one of the patient’s associated with one of these events? It represents a set of intersection points for me:
- the junction between healthcare and law
- the contrast between the medical record for purposes of care and the Rule 803 business record, and
- the use of knowledge discovery for clinical performance improvement vs. discovery to support a legal action.
What if the thousands of records examined, and the years of discovery invested in bringing malpractice lawsuits to the point of trial or settlement could instead be directed to clinical knowledge discovery within those same records? To understanding where variations in the process of care lead to errors? To identifying connections between clinical events that lead to positive outcomes, in contrast to those that lead to horrendous outcomes? The evaluation processes are actually pretty similar, though in the one case we are looking for relevant patient cohorts, and in the other, relevant documents. When you actually start looking at patient data within the Recommind tool set, it gets pretty exciting! I’m looking forward to what we can discover in the data so we don’t have to discover it in the courtroom.