Physicians as a profession strongly emphasize the importance of making the correct diagnosis, as that act has meaningful implications over the course of a patient’s life. The cognitive work of making a diagnosis implies both an art that is refined by a residency of focused observation of diagnostic masters as well as a science using Bayesian reasoning with various testing approaches. Alternatively, making the wrong diagnosis adversely affects patient safety. Conversations about overusing diagnostic tools rely on some measure of test appropriateness. If diagnostic tests are applied without regard to pre-test disease probability, individuals with symptoms that may not cause distress or shorten their life may be labeled with a diagnosis that does not lead to interventions to improve their health state (to be fair, overuse could also apply to screening [thyroid cancer], treatment [some cardiovascular procedures and medications], site of care delivery, and end-of-life care). Kale and Korenstein indicate overdiagnosis disproportionately affects primary care practitioners due to the volume of screening in those specialties for conditions like cancer screening, attention-deficit/hyperactivity-disorder and kidney disease (disclaimer: I am a general internist and consider myself a primary care provider). Kale and Korenstein admit that overdiagnosis is the consequence of trying to find disease before it is clinically evident. They list six drivers of overdiagnosis:
- Broadening disease definitions (lowering diagnostic thresholds, recognition of risk factors as pre-diseases),
- Technology (use of advanced technology for diagnosis, use of more sensitive screening tests),
- Public health interventions (widespread screening),
- Culture of medical care (value of diagnosis for its own sake),
- Clinician cognitive errors (overestimation of benefit of therapy in mild or low-risk disease),
- System factors (financial incentives for more testing), and
- Lack of clarity regarding disease spectrum in studies of diagnostic accuracy.
This article will focus on symptoms and diagnoses not detected by screening tests, a subset of the larger population of overdiagnosis.
Bhise et al. propose this definition of diagnostic uncertainty: “subjective perception of an inability to provide an accurate explanation of the patient’s health problem.” The group found three attributes of diagnostic uncertainty in their literature review. Diagnostic uncertainty is:
- a perception or emotional response, highlighting the subjective component of experiencing it in medical practice,
- impedes the clinician’s ability to act or think appropriately to initiate definitive treatment for the stated problem, and
- dynamic and changes with time.
Alam et al. suggest clinicians react to diagnostic uncertainty cognitively (e.g., using heuristics), emotionally (stress and anxiety), and ethically (disclosing uncertainty to the patient). Andre et al. interviewed 25 general practitioners in Sweden and found they used four strategies to deal with uncertainty:
- avoided through guideline adherence (about one-third of the sample),
- avoided through clinical picture and C-reactive protein (just over half the sample),
- avoided through expanded control (e.g., prolonging the examination as well as increasing their workload), and
Andre et al. posit that guidelines may not help manage uncertainty because of the perceived distance between guidelines and clinical practice. Bhise et al. argue that some uncertainty may be tolerated as long as there is enough certainty to exceed a threshold of definitive action. The authors go on to suggest that acknowledging diagnostic uncertainty might reduce subsquent testing if framed within a specialty’s guidelines (e.g., the American College of Physicians recommends against imaging for nonspecific back pain without red flags).
Managing a conversation about diagnostic uncertainty
Ideally, the dialogue between a patient and their clinician around symptoms that are not explained by available test results or therapeutic trials would acknowledge the lack of a unifying diagnosis or set of diagnoses to explain the symptoms. Clinicians might use their judgement and understanding of the medical literature along with any unique charactertistics of the patient’s particular presentation to address any concerns that a diagnosis might be missed. I would suspect that patient-clinician dyads with similar personality traits (specifically conscientiousness and neuroticism) would be less likely to disagree on the diagnostic process and current outcome. Clinicians with emotional intelligence might be able to adjust their approach based on their patients’ reaction to this information and increase the likelihood that a patient might agree maintain the relationship by using one or more tactics to facilitate a diagnosis (allow the symptoms to progress, see another clinician [within the same specialty or a new specialty], try another diagnostic test or therapeutic trial) or look for another clinician to transfer their care. Like most interactions between individuals with limited encounters, initial impressions and how each party explores disagreement are likely to influence each party’s perceptions of the other (e.g., trust, willingness to press their preferred choice, perceived openness to change one’s point of view).
From an employer or payer perspective
From the employer and payer’s perspective, reducing diagnostic uncertainty without increasing incorrect diagnoses or overdiagnosis would the best way to maximize employees’/members’ health while spending the fewest resources (dollars or employee availability). This optimization could appear as:
- Among patient-clinician encounters that do not include diagnostic uncertainty, focus attention on understanding an employee’s/member’s preferences for treatment based on efficacy, disease prognosis, convenience, cost and overall quality of life and deliver the treatment most consistent with the employee’s/member’s preferences at the lowest possible cost.
- Among patient-clinician encounters that include diagnostic uncertainty, for those encounters that allow both parties to express their diagnostic preferences and explore concerns, minimize the healthcare spend (e.g., telemedicine, site-of-service redirection) to reach the appropriate diagnosis, and
- Among patient-clinician encounters that include diagnostic uncertainty, for those (hopefully, rare) encounters where either party is unable to express their diagnostic preferences and explore concerns, facilitate transitioning the employee’s/member’s care to a new clinician.
The last group of encounters may represent a small group of individuals with a disproportionately larger number of encounters and healthcare spend. Employers and payers may see more value in matching their employees/members to clinicians with similar temperments around diagnostic certainty rather than trying to identify characteristics that uniquely identify the “best” clinicians based on publicly-available performance metrics.
Diagnostic uncertainty may frustrate patients and clinicians alike. Both parties may need to more clearly articulate their positions around next best steps to reduce the risk of severing the relationship. In those instances where the relationship fails, rather than trying to repair the relationship, employers and payers may help their employees/members more by identifying other clinicians who approach diagnostic uncertainty in a way that is more acceptable to the employee/member.