A recent article in Journal of the American Medical Association delved into the touchy subject of mandatory alcohol and drug screening programs for physicians. In their piece titled “Identification of Physician Impairment”, the authors Drs. Pham, Provonost, and Skipper concluded healthcare lags behind other high-risk industries in detecting high-risk factors causing injury or accident.


When a critical event occurs in most high-risk industries (such as airlines, nuclear power, or railways), a detailed investigation examines a variety of system and individual factors (such as fatigue and substance abuse) that caused or contributed to the event. Directly involved individuals are commonly tested for alcohol and other drugs. Airplane pilots and truck drivers are tested following crashes and near misses. Some law enforcement officers are tested following fatal shooting incidents.


They noted that mandatory alcohol and drug testing is not conducted for clinicians involved in an unexpected patient death or sentinel event (such as wrong-sided surgery). The authors suggested over their careers, up to 1/3 of physicians will experience a condition that impairs their ability to practice medicine safely. They argued that current systems, such as peer review, are inadequate to detect and treat the underlying problem until a number of patients have been injured.


Their recommended solutions:

  1. Mandatory physical examination, drug testing, or both prior to providing hospital privileges;
  2. Random alcohol-drug testing;
  3. Routine drug-alcohol testing on all doctors involved in sentinel event leading to patient death;
  4. And so on.


Many hospitals have already initiated drug and alcohol screening for its employees; and presumably physicians who are employed by such hospitals already agree to such testing.


The larger question is whether such proposals would have a favorable cost / benefit ratio. Arguably, the biggest risk factor for physician performance is plain-old fatigue. Not having enough time to sleep or rest. If this is accurate, it might be easier to embrace mini-cognitive tests as a solution, for example, asking clinicians to perform quick calculations on a monitor (or some equivalent), to detect real-time cognitive impairment. This could be as time-consuming as typing in a username and password into an electronic medical record system. In this model, one is merely measuring potential performance before performance is to be rendered.


This metric would be more meaningful than a urine or blood test; tests that might have zero correlation to performance. A positive drug screen for marijuana might be related to a joint that was smoked legally while on vacation in Colorado. And, a doctor might have a valid prescription for low dose narcotics for chronic pain.


The goal of a patient safety program should be to identify performance. There are a million ways performance can be affected. I argue it would be simpler – and more cost effective – to identify ability to perform, real-time, rather than measuring the million causes underlying impaired performance.