In his book, Better: A Surgeon’s Notes on Performance, Atul Gawande asks the question, “What does it take to be good at something in which failure is so easy, so effortless?”1 Consider this statement for just a moment. Every day, over 355,000 patents seek care in this nation’s EDs.2 These visits have a wide-range of significance, from the low acuity and low impact self-limited problems to the cases in which every decision and every second counts. Reflect on the 1999 Institute of Medicine (IOM) report, “To Err is Human.” At that time, 16 years ago, this seminal work estimated that up to 98,000 people die each year (268 each day) as a result of errors made in US hospitals.3 Variability in documentation, for many reasons, is a plausible factor in underestimation of accurate numbers. Since 2001, the worrisome number of deaths reported by the IOM has been re-evaluated a number of times, with each successive “deep dive” looking more ominous than the last.
In 2013, John James published a more recent estimate of preventable adverse events in the Journal of Patient Safety.4 He applied a literature review method to target the Global Trigger Tool from the Institute for Healthcare Improvement as the litmus test to estimate preventable error. In this limited review, James found that between 210,000 and over 400,000 premature deaths per year (575-1,095 deaths per day) are associated with harm that is preventable in hospitals. This number accounts for approximately 17% of the annual US population mortality and exceeds the national death toll from chronic lower respiratory tract infections, strokes, and accidents.5 Estimates of serious harm events (ie, morbidity) appear to be significantly greater than mortality. The adoption of the electronic medical record has not eliminated inaccuracies due to variation in documentation, reluctance of providers to report known errors, and lack of patient perspective in the recounting of their medical stories. The enormous magnitude of public-health consequences due to medical errors thus seems clear.
We become doctors and nurses primarily to help people, and not to cause harm to anyone. When harm occurs as the result of medical errors, the gut-wrenching guilt and self-deprecation that follows for most of us, and the doubt cast on our abilities as physicians, raise the question of why errors happen, and why more is not done to prevent them or to mitigate the consequences.
An awareness of some of the circumstances that lead to error can be a tremendous help in its prevention. High reliability organizations recognize that humans are fallible and that variation in human factors contributes to error, while also focusing on building safer environments designed to create layers of defense against error and mitigate their impact. Blame, shame, and accusatory approaches fail to improve any type of error. Environmental and situational hazards such as ED overcrowding, understaffing, high-patient volumes, rigid throughput demands, lack of equipment/subspecialty services and support in the system are highly contributory and must be addressed. Systemic issues aside, there are compelling individual factors that can lead anyone to make a mistake. Although lessons learned from mistakes are paramount to improvement, an understanding and awareness of the science of error and the necessity of “mindful medicine” can help protect individuals from the personal tolls of making a mistake.
Cognitive Biases
There are five significant cognitive biases that can result in preventable errors: availability bias, anchoring bias, framing bias, confirmation bias, and premature closure. Availability bias favors the common diagnosis without proving its validity. Anchoring bias occurs when a prior diagnosis or opinion is favored and misleads one from the correct current diagnosis. Framing bias can occur when it is not recognized that the data fail to fit the diagnostic presumptions. Confirmation bias can result when information is selectively interpreted to confirm a belief. Premature closure can lead hastily to an incorrect, rushed diagnostic conclusion. The following case scenarios illustrate examples of each of these biases.
Case Scenario 1: The Most Common Diagnosis Versus Looking for the Needle in the Haystack
On a busy night in the ED, the fifth patient of an emergency physician’s (EP) fourth consecutive shift was a quiet young lady home from college during the flu season. With a temperature of 102.6˚F, a heart rate of 110 beats/minute, blood pressure of 105/68 mm Hg, respiratory rate of 16 breaths/minute, and an oxygen saturation of 100% on room air, the EP was confident she would see another case of influenza. With the patient’s body aches, fever, and cough, she clearly appeared to be suffering from the flu, just like so many others this particular week. After treating the patient with fluids, antipyretics, and reassurance, she was sent home to rest in her own bed, with care instructions for influenza.