For many years I have counseled medical students and residents that half of what I was taught in medical school has since been proven obsolete or frankly wrong. I counsel them that I have no reason to believe that I am any better than my professors were. So I wish them luck sorting out what is true. Earlier in my career, that warning was mild hyperbole, but not anymore.

Upper respiratory infections (URIs) are the most common reason for an office visit during the winter. Bronchiolitis is the most frequent diagnosis for a winter admission of an infant to a community hospital. Pediatricians have nuanced assessments and many options when treating these diseases. Best practices have changed frequently over the past 3 decades, mostly by eliminating previously espoused treatments as ineffective. In infants and young children, those obsolete treatments include decongestants and cough suppressants for young children with common colds, inhaled beta-agonists and steroids for infants with bronchiolitis, and antibiotics for simple otitis media in older children. In other words, most of what I was originally taught.

My training, backed up by frequent experience when working as both a primary care provider and as a hospitalist, is that a significant fraction of infants after respiratory syncytial virus bronchiolitis will maintain twitchy airways for 3-6 months, during which time a simple URI seems to flare their albuterol-responsive lower respiratory tract wheezing, just as URIs do for asthmatics. This does not mean that their initial bronchiolitis was responsive to albuterol. I no longer use albuterol for initial episodes. Once fully healed, this cohort of young children has no further problem with wheezing. I therefore do not think they ever had asthma. This transient “reactive airways disease” is a manifestation of the healing time from bronchiolitis. This phenomenon is different from the statistic that one-third of children diagnosed with asthma as toddlers will outgrow it by age 5 years. This phenomenon has been somewhat supported by articles I have read, but never conclusively proven in a double blind, randomized controlled trial. My own experience vetting the phenomenon could be tainted by confirmation bias.

There is a discontinuity between guidelines that forbid routine steroids and beta-agonists for bronchiolitis in infants, and guidelines that strongly prescribe steroids, metered dose inhalers, and asthma action plans for all discharged wheezers over age 2 years. When I worked as a hospitalist in the pulmonology department, I frequently diagnosed asthma under age 1 year. As a general pediatric hospitalist, one winter I twice ran afoul of a hospital quality metric that benchmarked 100% compliance with providing steroids, inhaled corticosteroids, and asthma action plans on discharge for all wheezers over age 2. Fortunately for both me and the quality team working on that quality dashboard, my thorough documentation of why I didn’t think a particular wheezer had asthma was detailed enough to satisfy peer review.

Historically, medical knowledge has been dependent upon these types of observation which then are taught to the next generation of physicians and, if confirmed repeatedly, become memes with some degree of reliability. An all-too-typical Cochrane library entry may challenge these memes by looking at 200 articles, finding 20 relevant studies, selecting only 2 underpowered studies as meeting their randomized controlled trial criteria, and then concluding that there is “insufficient evidence” to prove the treatment works. But absence of proof is not proof of absence. Twenty five years after coining the phrase “evidence-based medicine,” our medical knowledge base has not been purified.

In the 17th century, French philosopher Rene Descartes concluded that too much of what he had been taught was wrong. He tried to purify his knowledge by starting over and only trusting what he could deduce with absolute certainty. His first deduction was “I think, therefore I am.”

In medicine, absolute certainty isn’t possible. Using 95% confidence intervals for a research paper does not even mean it is 95% likely to be right. So part of the art of medicine is finding a balance between cookbook guidelines (which might not fit a particular patient) and personal experience (which is tainted with confirmation bias.) It is a very imperfect art.

Dr. Powell is a pediatric hospitalist and clinical ethics consultant living in St. Louis. Email him at .