surgeons standing over patient

Quality Thinking About Quality – Part Three: Scientific Revolutions, Rarely Occurring Events

In Part Two, I discussed the Joint Commission’s Universal Protocol, explained why it hasn’t worked at preventing wrong-site surgery, and proposed a better approach.  I also discussed the problem of airway disasters and suggested that it is time for a new approach to airway management.  Here I’d like to discuss how revolutions in science (including the clinical sciences) often occur, and suggest that we need a revolution in airway management.  Then I’ll discuss the difficulty in trying to prevent rarely occurring events and suggest an innovative approach to their management.  Finally, I’ll wrap things up.

In his seminal work “The Structure of Scientific Revolutions” (1962), Thomas Kuhn described how scientific fields undergo periodic, often highly controversial upheavals in consensus opinion. Proponents of the prevailing paradigm believe it, defend it and structure their investigations accordingly.  Mostly they nibble at the edges of the paradigm that don’t work so well.  A challenge to dogma is often met with fierce resistance, the upstart challenger attacked personally, professionally, politically.  Steadily the challenger wins converts until there is a sudden, dramatic acceptance of the new idea.  Kuhn called this a “paradigm shift.”  Multiple examples abound in the clinical sciences: germ theory of disease (Pasteur), thermometry (Boerhaave), antiseptic handwashing (Semmelweis), newborn incubators (Couney), balloon angioplasty (Grüntzig), viral transmission of cancer (Rous), gastric bacteria (Marshall), infectious proteins (Prusiner), cancer immunotherapy (Allison), traumatic brain encephalopathy (Omalu).  If we are ever going to dramatically reduce the incidence of airway mishaps, we need a new paradigm in routine airway management.   We need to move away from an 80 year-old paradigm—rigid laryngoscopy—to a gentler, safer, more versatile, less traumatic, less invasive paradigm—flexible fiberoptic bronchoscopy.

Surgical fires, patient burns, electrical injuries, retained surgical equipment, intraoperative patient falls, catastrophic equipment failures, blood transfusion reactions, anaphylaxis, malignant hyperthermia, air emboli and other such very rarely occurring events can cause significant patient injury and financial loss.  The standard response to such occurrences is a “Root Cause Analysis”—an inquiry into what happened and educational and corrective measures to prevent a future event.  Then years or decades pass. Personnel, equipment, techniques, protocols and facilities change.  The knowledge that had been gained is steadily extinguished. Another event occurs.  Another RCA.  Can we do better?

For any one institution any one type of these incidents probably occurs very rarely if at all.  Because the data points are so few and so temporally remote, meaningful trend analysis is impossible.  To entertain conclusions from scant data is to entertain error.    With the emergence and growth of large hospital networks and conglomerates, data can be pooled across dozens or even hundreds of hospitals and analyzed intelligently.  But even with hundreds of hospitals contributing data, for any one type of event we still might have only a dozen or two data points to plot on a time series chart.  A novel approach would be to convert a binary event—either the thing occurred or it didn’t—into a daily rate of occurrence.  We do this by counting the number of days between events and dividing one (the event) by that number.  We now have a daily rate of occurrence for one event for the time between its occurrence and the event immediately prior to it.  We repeat the math for the next events, counting the number of days between each one and the immediately prior event and dividing each number into one.  Now we can plot these data points (rates of occurrence) on a time series chart, along with a centerline, representing their average, and horizontal lines above and below the centerline representing three standard deviations from the mean, or three-sigma limits.  If most or all of our data points fall between the upper and lower limits, the event is occurring in some sort of predictable fashion and we could then identify an approximate time window in which the next event might occur.  We could then undertake educational and other proactive measures among all the hospitals to prevent such an occurrence.  Success should reveal itself in the control chart, as the rate of occurrence decreases over time. 

To summarize, quality is a characteristic of physical things.  When health care administrators talk about quality, they are actually talking about performance and outcomes.  To make excessive quality demands on a system is actually to make excessive performance demands and to increase the likelihood of system failure.  Workers in the system do not own it and are not empowered to make ownership changes to it.  If a demand exceeds the system’s performance capability, the workers have only three choices: fail to meet the demand, distort the system to appear that they are meeting the demand, or distort the data to appear that they are meeting the demand.  The Joint Commission’s Universal Protocol has proven to be ineffective in reducing the incidence of wrong-site surgery because 1) it represents another demand on a heavily burdened system, and 2) it is a safety maneuver rather than a safety mechanism.  Blocking the wrong site rather than marking the correct site could dramatically reduce wrong-site surgeries. Routinely utilizing flexible fiberoptic bronchoscopy rather than rigid laryngoscopy could greatly reduce the incidence of airway mishaps. To better gain insight into very rarely occurring events, data should be pooled among large hospital networks, converted into rates of occurrence and plotted on a control chart.  An approximate time window might be identified in which the next event could be expected to occur and appropriate measures employed to preempt it.