The scientific method has given us modern medicine, with astounding abilities to diagnose, treat and even cure. Until the last century, medical treatment was futile at best and fatal at worst. Cure was not possible and the touchstone of our profession was comforting the patient.
Things have changed. Cure is possible. In some areas, cure is common. Many previously devastating and fatal illnesses have been reduced to nuisances or even eliminated. Just as we do not care about our airline pilot’s conversational skills or personality as long as she expertly delivers us unharmed to our destination, we care little that the surgeon is imperious as long as the operation is successful. The tantalizing prospect of good outcomes trumps process.
We have discovered that modern medicine is too complex and too rapidly changing for any individual - or even a team - to acquire, remember, and apply all the best and newest information. We know that when we move beyond the simplest medical problems, the use of algorithms, standards of care, protocols and checklists that are based on evidence will result in more consistent and better care for populations than expert individuals. Doctor knows best has been replaced by data knows best. Unfortunately for our patients, the data is increasingly applied using management principles based on the industrial revolution. One of these principles is the use of standardized processes to eliminate variability.
Starting with John Wennberg, researchers at Dartmouth have shown how muchunjustified variability there is in health care. Atul Gawande has been a great proponent ofchecklists and protocols to eliminate unwarranted variability (here and here). In pursuit of quality, medicine is feverishly adopting industrial management processes aimed at standardization, in the belief that variability is the enemy of quality. (The term feverish is particularly apt, as there is sometimes more heat than light, like the thought and behavior born of delirium.) The elimination of variability through standardization is a good example of the old saying that every complex problem has a simple solution that won’t work
For starters, we should not forget that variability is not avoidable. No matter how hard we try we will never succeed in imposing order on inherently complex and emergent processes and problems. One danger in standardization is that it can prevent innovation, creativity, growth and learning. It can lock us into old solutions and blind us to new possibilities that arise from new knowledge, or the need for new approaches as the problems themselves evolve.
Also overlooked is the concept that there are two kinds of variability in medicine: one arising from the patient and one arising outside the patient.
The Atul Gawandes of the world are rightly crusading to identify and eliminate the variability that (1) arises from the system (institutions and providers); and (2) does not confer a benefit. This is much harder than it sounds, but is worth the effort.
The variability that arises from the patient is a different story. The growing emphasis (bordering at times on religion) on standardization can obscure the fact that the data on which evidence based medicine rests is population data and applies well to populations, but is very tricky to use individually. Every high school student knows that 1000 coin flips will be split very close to half heads and half tails, but that the next coin toss will either be 100% heads or 100% tails, not a coin with half a head and half a tail showing. Five percent of a population may be pregnant, but each individuals is either pregnant or not pregnant. There is no such thing as 5% pregnant. Variability based on patients should not be ignored or denied. It should be sought out and used to individualize the response. Patient variability demands treatment variability.
A recent experience in my practice is a good illustration. I saw a Brent, young firefighter and EMT in his late 20s. He came in for his annual health review and we noted that his previously borderline blood pressures were now clearly abnormal in the 155-160/95 range. He was trim and fit, a non-smoker and non-drinker. He had a family history of hypertension but not of coronary artery disease or stroke. His EKG, lipids, calcium, sugar and renal function were normal and he had no evidence of ‘target organ’ damage on history, exam or lab. As there was little in his life style to change, we agreed to start medication and I prescribed an ACE (lisinopril, a generic angiotensin converting enzyme inhibitor). I usually start with a diuretic, either chlorthalidone or hydrochlorthiazide, as these have been around a long time and have been shown (population data) to be the most reliably effective and safest of the medications available. However, he was a firefighter and often worked long hours in heavy protective gear in a superheated environment doing extremely physical (and dangerous) work without access to water for rehydration. Diuretics can compromise volume. For most people this is not an issue, but I was concerned that it put him and his partners at risk in a firefighting environment.
His insurance refused to pay for the ACE unless he tried and failed what they said the literature proved was the standard first line agent. I had a polite but unproductive phone conversation with a representative of his insurer. The patient then was prescribed a diuretic and after two weeks he was involved in a full equipment demonstration drill on a test fire. He became dehydrated and orthostatic (light headed and nearly passed out) while climbing a ladder to enter a burning building. The insurance company accepted this as a failure of diuretic therapy and his blood pressure is now well controlled on lisinopril, which he tolerates. No one was harmed, but people were clearly put at risk.
In this instance, the algorithm and data made sure I thought about a diuretic as a first line approach, but should have then allowed me to individualize, recognizing that Brent was not well represented by the ‘average’ patient in the population data. This does not justify variation. It demands variation, consistent with the variation of the individual patient from the population mean. Rigid adherence to the standardized protocol cost hime and his insurer money (several additional office visits as well as the emergency room visit and testing for his syncopal episode), disrupted my patient’s life (he missed a week of work and a pay check), and could have cost lives.
I am all for examining variability and looking for ways to standardize best practices, but until our patients are mass produced on an assembly line within narrow tolerances, and their lives are regulated to avoid deviations from the mean, rigid adherence to evidence based protocols is irrational and unscientific, and represents maladaptive behavior more suited to a cult than a caring profession using science as a tool.