The Doctors Company Internal Medicine Closed Claims Study analyzed 1,180 claims that closed from 2007–2014. The study found that the top allegation, representing 39% of claims against internists, was diagnosis related and resulted from a delay or failure to diagnose. This finding is consistent with data published in Improving Diagnosis in Health Care (National Academies of Sciences, Engineering, and Medicine), which found that 34% of nonsurgical specialty claims are diagnosis related.
Physicians fail to diagnose accurately for many reasons. The dilemma can be understood best in the context of the complexity of clinical medicine. Illnesses present with an infinite number of variations, illustrated by the 68,000 ICD-10 diagnostic codes and 8,000 recognized diseases and syndromes—many of which are uncommon. The average primary care physician diagnoses about 400 different diseases a year and, every now and then, encounters a rare medical condition that he or she may have never seen before.
It is in this context that failure to diagnose may be viewed as an error or lapse in reasoning rather than just a failure of clinical skill. Therefore, diagnostic accuracy can be improved with a better understanding of how to avoid pitfalls in medical decision-making.
Improving Diagnosis in Health Care
The monograph Improving Diagnosis in Health Care characterizes failure to diagnose in terms of two types of thinking processes—rapid and slow—and the effects of psychological biases on medical decision-making.
- Type I, or rapid decision-making, involves pattern recognition (heuristics) that allows the clinician to successfully diagnose and treat most patients efficiently. For example, a female patient with dysuria and frequency will most often have an uncomplicated urinary tract infection.
- Type II, or slow decision making, requires recognition by the clinician of the possibility of a complex medical problem and the need for careful thought, a differential diagnosis, lab and imaging studies, reference resources (such as UpToDate), and/or consultation with a specialist. Recognition of risk factors is essential.
Psychological biases may undermine accurate diagnosis and treatment. Some common examples include the following:
- Anchoring bias: The tendency to rely too heavily on, or “anchor” to, one trait or piece of information when making decisions—usually the first piece of information or diagnosis that is acquired.
- Premature closure: The tendency to apply premature closure to the decision-making process by accepting a diagnosis or treatment before it has been fully verified.
- Overconfidence bias: A universal tendency to believe we know more than we do.
- Optimism bias: The tendency to be overly optimistic by overestimating favorable and pleasing outcomes. This can also be considered a form of denial.
The following illustrations are taken from The Doctors Company Internal Medicine Closed Claims Study.
A 53-year-old male presented to the hospital with acute chest, epigastric, and back pain. Risk factors included hypertension, diabetes, tobacco use, and a family history of coronary artery disease. An EKG was negative for acute changes. Lab studies included a normal CPK and minimally borderline troponin. The lipase was 1,455, and a diagnosis of acute pancreatitis was made. The epigastric pain improved, but the patient continued to report lower chest pain associated with chest palpation. Two weeks after discharge, he presented with an acute myocardial infarction.
The physician correctly diagnosed pancreatitis but, in retrospect, missed subtle suggestions of myocardial ischemia, including a slightly elevated troponin and persistent, although atypical, chest pain in the setting of multiple risk factors. The physician anchored on the single diagnosis of pancreatitis, which led to premature closure of the diagnostic process.
A 60-year-old female presented to the internist with abdominal pain and rectal bleeding. She was referred to a gynecologist, who diagnosed a likely uterine fibroid on ultrasound. An endometrial biopsy was benign. Symptoms persisted, and several months later the internist ordered an abdominal CT scan that revealed a malignant rectal mass displacing the uterus.
The internist appears to have been reassured by the gynecologist’s finding of benign pelvic disease. This is an example of premature closure, demonstrating that when the referral was made, the thinking stopped. Both patient and physician want pleasing outcomes, but a differential diagnosis—in this case, focusing on the common causes of rectal bleeding—would have probably led to a more timely diagnosis of rectal cancer.
A 65-year-old female presented with nausea, fever, and a dark area in the visual field of the right eye. She was diagnosed with a viral infection. Four days later, she presented to an ophthalmologist with the loss of central vision in the right eye and was diagnosed with a retinal detachment, resulting in permanent loss of vision.
Primary care physicians see many patients with nonspecific symptoms of nausea and fever. Most of these patients have an acute and self-limited viral illness. However, complaints of acute visual loss are relatively uncommon in a general practice, and most primary care physicians do not have the training or equipment to properly evaluate those patients. This case illustrates overconfidence bias in which the physician appears to have failed to recognize the potential significance of an unusual visual complaint, concentrating instead on the more common viral illness.
A 45-year-old male, febrile, with poorly controlled diabetes, was admitted to the hospital with vomiting and weight loss. Blood cultures revealed gram-positive cocci in chains. The patient was discharged on antibiotics before the final culture and sensitivity report was available, but he was readmitted a week later with hemodynamic decompensation and fever and diagnosed with mitral and aortic valve endocarditis. The results of the prior culture demonstrated Streptococcus viridans. He underwent valve replacement, developed severe left ventricular decompensation, and died from end-stage congestive heart failure before a heart transplant could be performed.
This patient’s initial improvement appears to have led to another example of premature closure and optimism bias. It is, of course, essential to review final blood culture results and not make the assumption that the patient’s initial improvement is a predictor of a successful outcome.
A 59-year-old male admitted with abdominal pain was diagnosed with acute diverticulitis and treated with Garamycin, Avelox, and Flagyl. The patient was discharged on the same three antibiotics without an order to monitor serum gentamicin levels. Subsequent symptoms of vertigo were ultimately diagnosed as gentamicin vestibular toxicity.
Traditional physician education has emphasized memorization and “thinking on your feet.” Stopping to consult with a reliable reference in the middle of rounds has not been part of that tradition. No clinician can possibly know all of the information required to practice medicine. There should be a low threshold for reviewing references to help with diagnosis and treatment—even for relatively common conditions such as diverticulitis. This case provides another example of overconfidence bias.
The bottom line
Accurate diagnosis and treatment are often challenging—particularly in the context of time limitations and multitasking required in today’s practice environment. Having a better understanding of current theories on how to improve the diagnostic process may help clinicians reduce errors and improve outcomes.
Read the full study, including expanded case examples and risk mitigation strategies, at http://www.thedoctors.com/internalmedicinestudy.
This post was sponsored by The Doctors Company, the nation’s largest physician-owned medical malpractice insurer.