The Process of Diagnosis

Understanding why so many people face persistent nondiagnosis or misdiagnosis and working to address this problem requires examination of the process of diagnosis itself. By improving our insight into the end goal of diagnosis and how we often get there, we can better understand the strengths and weaknesses of this process and work to improve its effectiveness for doctors and patients alike.

In my opinion, DIAGNOSIS IS THE MOST IMPORTANT ELEMENT OF HEALTH CARE. Why? Every subsequent care decision for a patient depends on a timely and accurate diagnosis. The goal of diagnosis is to identify a condition so that it can be appropriately treated, making it the limiting factor affecting many health outcomes. Delayed diagnosis results in more advanced disease before treatment begins and therefore worse health outcomes. Inaccurate diagnoses result in patients being treated with interventions that are ineffective and potentially harmful. There are estimated to be 100,000 deaths per year in the US due to medical error, with many of these due to inappropriate treatments. Only when an accurate diagnosis is secured can appropriate and targeted treatment be effectively used.

Despite its importance, diagnosis may be one of the least understood elements of health care. Even defining diagnosis becomes difficult. At its simplest, it can be thought of as the identification of disease within an individual. It is a description and explanation of a condition as well as a guide for treatment. But, knowledge about conditions varies, as do diagnostic tools and methods. Some diagnoses are categorical and definitive, such as whether a patient has HIV or not, while others (such as Lupus) rely on a physician’s clinical judgment and cannot be proved with certainty. But, how sure is sufficient? It is a diagnosis if it is 75% likely? 51%? This highlights the fact that diagnosis is often as much art as science.

There are three main ways that doctors approach the process of diagnosis: Bayesian diagnosis and branching logic. Bayesian diagnosis is a probabilistic model that correlates symptoms with conditions to identify a diagnosis. This relative probability evolves as additional information is collected, such as history, physical examination, and test results. Branching logic is a categorical model that involves constructing a series of logical decisions intended to rule out all other possibilities. This can be envisioned as a decision tree of hypothesis generation and testing that moves from general to specific. Most doctors do not overtly follow one or both of these processes in their heads, but subconsciously use elements of each in narrowing down the potential cause of a condition to secure a diagnosis.

While both of these models offer valuable frameworks to conceive of how doctors make diagnostic decisions, they fail to capture an essential component of diagnosis: intuition and expertise. Diagnostic insight is the cumulative effect of years of experience and often, good diagnosticians will just have a hunch about a patient. Combining a logical and scientific approach with intuition and insight can result in important diagnostic breakthroughs.

Given the volume of information that doctors must process, they are also forced to develop mental algorithms and heuristics (rules of thumb) to simplify many situations. The best known of these, Occam’s Razor, essentially states that the simplest answer is the most likely answer. This is also expressed through the common medical sayings, “When you hear hoofbeats, think horses, not zebras” and “An abnormal manifestation of a common condition is more likely than a normal manifestation of an uncommon condition.” While in the majority of cases these frameworks serve diagnosticians well, they can represent intellectual traps that inhibit the recognition of unique cases and outliers.

There is much that patients and doctors can do to improve the process of diagnosis and therefore the outcomes of this process. The first and most important is to recognize that diagnosis is extremely difficult and often imprecise. This highlights the need for patients to seek out appropriate caregivers and be proactive partners in the health care process rather than passive recipients of care. To do so, patients must do a better job collecting, managing, and presenting information relevant to their case. They must also follow through with diagnostics and treatments prescribed by their doctor. Doctors should remind themselves to question diagnostic assumptions, be proactive about seeking out alternative viewpoints or exhaustive second opinions, and make use of clinical decision support tools when appropriate. Earlier and more accurate diagnosis will lead to improved health outcomes, significant cost avoidance, and a reduction in pain and suffering. To achieve this goal, we must improve our understanding of the diagnostic process and develop tools to enhance its effectiveness.