The Real Problem With Misdiagnosis

Charles W. Smith


J Participat Med. 2013;5 

In a recent editorial, my colleague and co-editor, Terry Graedon, reviews the substantial harm that results from misdiagnosis and argues that information overload is a major contributing factor. She also reports Larry Weed's belief that the lack of adequate software tools is one reason that clinicians may be unable to correctly match symptoms with diagnosis. But, with or without such a system in place, it is rare that clinicians can compile a list of potential diagnoses and simply choose one that "fits." One must always take into account the clinical process an experienced, skilled provider employs during an office visit. If all we needed was better software, we wouldn't need a provider with years of education and training.

The notion of matching symptoms with diagnoses and employing software to deal with information overload risks oversimplifying the problem and the solution. While information overload is a reality of practice, it can be mitigated by allowing sufficient time to carefully consider the possibilities and chart a plan. After taking the history and performing a physical examination, the clinician builds a list of problems, followed by testing to confirm possible diagnoses. Unfortunately, we may not have enough time in a single encounter to go through these steps completely and, thus, the risk of misdiagnosis rises substantially. But fortunately, we do have an opportunity during follow up visits to review and revise the problem list. Identification of problems and diagnoses is an ongoing and often-changing process, with omitted diagnoses or problems added, current ones changed, and old ones deleted from the list. Rarely, does an office visit occur when I don't revise one or more items on my patient's problem list. I believe the best defense against the danger of misdiagnosis is to assure that a comprehensive history is taken, a thorough physical exam is performed and any relevant lab or imaging tests are obtained. Then, for those issues that remain "undiagnosed" (eg, abdominal pain, headaches, joint pain, weight loss, etc), followup visits are scheduled, initial laboratory tests are reviewed, additional tests are ordered as indicated, referrals are made and, over time, most of the problems will be correctly identified.

Terry goes on to state that "computers show greater accuracy in diagnosis than physicians." While this may sometimes be true, we must also remember that humans are very complex, and usually present with a mix of physical and emotional issues that intermingle in confusing ways that don't readily lend themselves to Dr. Weed's approach. For example, somatization is a mechanism in which a patient with emotional pain exhibits that problem as if it were a purely physical problem, but no physical problem can be found. These complicated clinical problems don't neatly fit into a specific diagnosis and certainly don't fit into a tidy computer algorithm. Rather, in this setting, the clinician makes an assessment of "what is going on" considering medical history, social history, family background, current stresses, and evaluation of coping strategies. These issues cannot be passed to the domain of the mental health professional, since they make up a substantial (some say at least half) of the primary care physician's practice. Further, there are a number of syndromes that have not been well defined or described and, certainly, as in the case of IBS and fibromyalgia, are not well understood as to etiology; and for which there is only "clinical," not definitive, diagnosis. Where, then, do these vague and "ill defined" entities fit into the "misdiagnosis" discussion?

Clearly, in the setting of an acute presenting symptom, such as the patient with gastrointestinal bleeding or severe, acute pain, the need for a timely, accurate diagnosis is critically important. But, in the majority of ambulatory encounters, primary care patients either have lifestyle issues that need to be addressed through partnership with the patient, or they have chronic disease such as diabetes, hypertension, and/or hyperlipidemia, which pose more of a management problem than a diagnostic dilemma.

So what is the message? That reducing misdiagnosis is, indeed, a worthy goal and information technology is a useful tool to assist the physician in this task. But it will only assist in a small portion of the job of assessing the "whole patient" and managing the complicated array of problems presented.

Although misdiagnosis is going to occasionally occur, a commitment to the proper doctor-patient relationship and care process will usually suffice to prevent it. Along the way, EMRs, clinical data bases on smart phones, and software programs such as the one Dr. Weed describes, will be welcome assistants to conscientious physicians.