Learning from Every Patient

A Newsletter for Frontline Healthcare Team Members and Quality Improvement Professionals

Real cases. Practical improvements. Stronger voices at the bedside.

From HB Healthcare Safety® and Influence Ignited!

Edited by Jeanne M. Huddleston, MD, MS

2 ÷ 20,000.

BLUF: A small numerator divided by a very large denominator is a very small proportion.

Two patients. Lost to a rare event last year. Two of twenty thousand admissions. Every hospital has a version of this fraction, including a mandate to "fix it". And every improvement plan promises to do just that. Mortality is a fraction too. Patients who died over patients admitted. That fraction has an algebra problem.

The events everyone wanted to chase: the catastrophic ones, the never-events, the unanticipated deaths. Every one is somebody's mother, somebody's son. My first instinct was to chase them too. Then I realized the math underlying our choices.

A hospital with 20,000 admissions has approximately 2% of its patients die (national average). That is 400 patients a year. Now do the impossible and wipe out the rare event - like a preventable death. Those are approximately 5% of all deaths. Erase them entirely. Zero preventable deaths. Your numerator shifts from 400 to 380.

The mortality rate on the board's dashboard falls from 2.0% to 1.9%. If the mandate was to decrease the mortality rate by even a little, then the team was successful. Lives were saved, but the change in the metric falls into the category of statistical noise. In other words, statistical variation from year to year can explain the change from 2.0% to 1.9%.

Do not confuse what I am saying. Those lives saved are important, but that work alone - focusing only on the preventable - does not get to the statistically meaningful change that will stand the test of normal variation. So yes, wipe out the preventable deaths. Important work. But you must also attack a problem that affects large proportions of your patients.
​
To move a metric with an enormous denominator, the numerator has to be large enough to register. With rare events, even erasing them completely is not enough to visibly change the high-level metric. We have to do both - rare and common. This requires a change in the approach to resource allocation.

𝙌𝙪𝙖𝙡𝙞𝙩𝙮 w𝙤𝙧𝙡𝙙, 𝙬𝙚 𝙝𝙖𝙫𝙚 𝙖𝙣 𝙖𝙡𝙜𝙚𝙗𝙧𝙖 𝙥𝙧𝙤𝙗𝙡𝙚𝙢.

Leadership is not wrong to want these metrics to improve. Boards are not wrong to demand it. Quality officers are not wrong to chase the mandated measures. The mandates were not written by people who wanted to fail patients. They were written to create reportable signals. A reportable signal requires a nameable event. A nameable event requires a system that asks: what went wrong and who did it. That is Safety I's question. Safety I produces exactly what it was designed to surface: the rare, classifiable, commission-based failure. The system gave us the right tool for what it was built to measure. The problem is not the tool. It is just the answer to the wrong question.

In 2003, the mandate I received was clear: fix mortality.

At that time, the healthcare quality world was focused on the concept of "preventability". I was intrigued by how M&M Conferences focused on "interesting cases" and peer review looked at who did what wrong, while the few mortality review processes I knew screened for unanticipated deaths. But what about the other 95% of the cases?

At the time when we were deciding how to approach the "fix mortality" mandate, I had a 90+ year old grandfather with colon cancer, an EF of <20%, and was transfusion dependent because of slow GI blood loss. I realized that if, for some reason, he was hospitalized and died in our hospital, under traditional patterns we would not review his case. My father could ask me what happened to his father; I would have to respond with, "I don't know. He was a sick old guy and we did not review his clinical course." This insinuated that there was nothing to learn or improve in cases like my grandfather. Needless to say, I bristled at the thought, and the 100% mortality review approach was born (Learning from Every Death publication).

𝗨𝗻𝗳𝗼𝗿𝘁𝘂𝗻𝗮𝘁𝗲𝗹𝘆, 𝘁𝗵𝗮𝘁 𝗿𝗲𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗱𝗼𝗲𝘀 𝗻𝗼𝘁 create a visible or statistical 𝗰𝗵𝗮𝗻𝗴𝗲 𝗶𝗻 𝘁𝗵𝗲 𝗺𝗼𝗿𝘁𝗮𝗹𝗶𝘁𝘆 𝗿𝗮𝘁𝗲.

100% mortality review takes serious resources - ones that were more available 20 years aago. I had to sell that approach because of the resource commitment needed and its difference to best practices at the time. Using our epidemiology approach - consecutive mortality reviews - we proved that opportunities for improvement happened to patients regardless of code status, age, and comorbidities. So what went from a goal of saving 5 lives per month turned into one of saving those 5 lives AND at least 20 more. We recognized that we had to work upstream, at the level of the frequent, quiet omissions — where the events are common enough to measure, attribute, and change inside a cycle short enough to matterTo honor the mandate, the scope had to expand far beyond what the reporting system was designed to classify. So we studied every death. Not what went wrong and who did it. What could have gone better, and why did the same process/pattern happening to different people on different days, cared for by different teams?

What did we learn then and later repeat at a national level? 89% of opportunities for improvement were omissions. Not acts of commission (Safety I). But omissions. Things that we missed. Things that we did not do. Things that do not get a voluntary report submission exactly because "nothing happened". So our leaders were missing 89% of the harm or opportunities for improvement occuring in their facilities - not because they weren't paying attention. But because they never recieved the information needed to act.

𝟴𝟵% 𝗼𝗳 𝘄𝗵𝗮𝘁 𝗻𝗲𝗲𝗱𝗲𝗱 𝘁𝗼 𝗯𝗲 𝗳𝗶𝘅𝗲𝗱 𝘄𝗮𝘀 𝗶𝗻𝘃𝗶𝘀𝗶𝗯𝗹𝗲 𝘁𝗼 𝘁𝗵𝗲 𝘀𝘆𝘀𝘁𝗲𝗺 𝗱𝗲𝘀𝗶𝗴𝗻𝗲𝗱 𝘁𝗼 𝗳𝗶𝗻𝗱 𝗶𝘁.

𝗦𝗮𝗳𝗲𝘁𝘆 𝗜 𝗮𝘀𝗸𝘀: What went wrong and who did it? It finds the rare events.

𝗦𝗮𝗳𝗲𝘁𝘆 𝗜𝗜 𝗮𝘀𝗸𝘀: What could have gone better, and why does our process keep breaking down? It reaches the frequent.

What could actually move a mortality rate was NEVER the rare event. It was what was happening to many patients, admission after admission: the goals-of-care conversation that never happened, the diagnosis that arrived late, the treatment that waited, the recognition and rescue of the deteriorating patient that came too slowly.

These occur far more frequently than any hospital-acquired condition. These are the events that, improved at scale, that will statistically decrease the board-level mandate of decreasing mortality rate.

Frequent events make large numerators.

Shifts in large numerators = real change.

QI resources have been pointed at the rare things. Because that is where the mandates and public reporting aim them. Federal and state mandatory reporting directed everything toward CAUTI, CLABSI, falls, pressure injuries - acts of commission, all of them. Rare, all of them. All of them, except falls, improved over the decades with intense, prolonged focus.

This has created a structural problem with how we express our metric goals.

Board-level directive to decrease mortality rate in the next fiscal year. The quality leadership believes the driver should be zero preventable deaths. Admirable. But doing only that, putting all focus and energy on the identification and eradication of those few, and thankfully relatively rare, deaths, at the end of the year, the mortality rate is unchanged.

We must learn from the "preventable" death... and every other patient experience.

Not a commitment problem. Not a capability problem. There is a lot to learn from cases like my grandfather and father. But a meaningful review to surface commisions and omissions of care takes me a solid 60 minutes in our existing EHR. The capacity to learn from every patient who has something to teach us is gone. Times are different now.

There is a misalignment between the frequency of the events we are required to measure and the frequency required to change the number. In order to provide the data needed, we must do things differently. I realize that very few healthcare systems, if any, will have the resources to dedicate to 100% mortality review with 100% multidisciplinary, multispecialty commtment to the findings. That kind of time just does not exist. So we built something new.

Service as Software: we are ending the administrative burden of case reviews

We developed a model that replicates my standardized 7-step case review process. It tee's up OFIs for the human in the loop to accept or decline. You get to read a summary and provide your input. Every OFI has an auditable connection to the location in the medical record that justifies the OFI presented.
​
This is now my life's work. I am still going to "learn from every death", but we are also going to learn from every life - patient and frontline care team member. Aggregating lessons learned and OFIs identified at that level will give us a real Pareto diagram to guide our interventions. The OFI with the largest numerator (happening most frequently) wins. I

f we stay in Safety I... if we stay with the small numerators, that will never create meaningful shifts in proportions... twenty years from now, someone will publish the exact same paper. No improvement. Same metrics. Same conclusion. The math will not have changed.

Quality Leaders, we have an algebra problem!

Newsletter Trivia Question

Just repy to this email with your answer and get entered into a $100 at the end of the quarter. Any answer work - right or wrong!​
Question: The essay leans on a distinction between two ways of asking about safety. Who coined the terms "Safety-I" and "Safety-II"?

About this Newsletter

Learning From Every Patient is a case-based patient safety and quality newsletter for frontline clinicians and healthcare quality improvement professionals who want to stop preventable harm. Each issue features rotating content, including real cases, expert commentary, a short “Loved One Lens” reflection, a brief look at “what worked” in successful improvement efforts, tips and tricks for quality improvement and patient safety projects, and quick scans of relevant new evidence. Published by HB Healthcare Safety, SBC, and powered by Influence Ignited!, the newsletter is edited by Jeanne M. Huddleston, MD, MS, and is designed to turn everyday stories and process defects into practical, actionable lessons you can use on your next shift.

Editor: Jeanne M Huddleston, MD, MS

Founder, Influence Ignited, LLC

Co-Founder, HB Healthcare Safety, SBC

Professor of Medicine, Mayo Clinic College of Medicine

huddleston@hbhealthcaresafety.org OR huddleston@influenceignited.org