Patient Experience as a Source for Understanding the Origins, Impact, and Remediation of Diagnostic Errors

Questions for Future Research

Methods exist to elicit more reliable and complete accounts about patients’ diagnostic experiences. Using these methods is a necessary but not sufficient step for more effective learning about patients’ perspectives on diagnostic safety.

Understanding patient and family perspectives on diagnostic experiences also requires recognition that the diagnostic experiences being recounted can often be complex and will sometimes never be fully resolved in their own minds, even when viewed in retrospect. But additional research may help further enhance our capacity to learn from patient experiences with diagnostic safety.

Enhancing the Fidelity and Reliability of Narrative Accounts: Several aspects of collecting narratives about diagnosis would also benefit from additional study. The first involves potentially fine-tuning elicitation techniques to learn from experiences of patients in the throes of extended diagnostic odysseys.⁶ For example, these frequently occur for newly emerging conditions (e.g., “long COVID”) or rare conditions, both cases where clinical knowledge is limited and clear protocols for timely diagnosis often remain ill defined. Comparably extended searches for a diagnosis can, of course, emerge for a variety of conditions when patients present with an unusual combination of symptoms or these symptoms manifest in atypical ways.

These circumstances also challenge patients and families to interpret diagnostic experiences and make it more difficult to attribute delays or other shortfalls in diagnosis to a particular clinician’s expertise or thoroughness. Some patients or families will explore the internet to clarify these boundaries, but many patients still rely exclusively on clinicians to inform them about what is known about a given condition.^v Whether elicitation techniques need to be adapted to these circumstances requires further study.

Contexts in which clinicians and patients might have agreed to not aggressively pursue certain diagnoses present a second issue. These situations are not uncommon in the care of older patients and might grow even more prevalent were clinicians and patients to discuss forms of watchful waiting or palliative care more openly.

However common, these circumstances raise the specter of age-related biases inhibiting appropriate diagnoses for some patients,³⁵ and watchful waiting might be viewed very differently by family respondents than by the patients themselves. Precisely because these circumstances are complex and potentially fraught, research is needed to clarify how best to ask respondents to describe and explore them.

A third set of considerations involves the role of organizational or systemic factors that limit the capacity of clinicians to pursue diagnoses in ways they would prefer. Past research on how patients make attributions regarding clinical encounters shows that it is often hard for patients to discern the impact of system-level factors. These factors include organizational constraints on referrals or limitations imposed by health insurers on coverage of out-of-network specialists.¹⁷

Narratives of patient encounters suggest that patients often suspect that these constraints have affected their diagnosis but typically do not discuss these matters with their clinicians to verify this speculation.³⁶ Research is needed to understand how best to learn about these effects and to promote clinician-patient dialogue that clarifies the impact of system-level factors.

Addressing the Limitations of Narrative Feedback: When the diagnostic process goes awry, patients will often experience the consequences in the form of less effective treatments than anticipated. But the efficacy of initial treatment can also be a perfectly reasonable way of “testing” a diagnosis, thereby learning more about the condition in question. That makes it difficult for patients or families to reliably assess the linkage between diagnosis and their subsequent experiences with treatment. It also makes it especially challenging to determine whether initial treatments were sustained too long, leading to inappropriate delays in considering alternative diagnoses.

For all these reasons, the nexus between diagnosis and treatment may remain opaque. Whether something potentially useful can still be learned from patients and families about this nexus remains unclear. That leaves it as a suitable topic for additional study but one that is likely to have relatively limited yield from even the most effective narrative elicitation.

A second and related set of limitations involves assessments by patients and families of diagnostic delays more generally. In this case, a potentially meaningful distinction can be made between narrative feedback that can identify sources of delay and narrative accounts that can determine the magnitude of that delay. (Delays may involve clinicians unwilling to make a referral or test results that are misplaced or for which no one “closed the loop” in terms of followup communication.)

Patient narratives have ample capacity to identify sources of delay, although how prevalent and how reliable these markers of delay are remain in need of additional research. But the capacity to use narrative accounts to calculate metrics of the magnitude of delay appears far less promising, in large part because patients and family have limited capacity to parse out appropriate from inappropriate lags in altering diagnoses.

Third, patients and families sometimes report an absence of effective care coordination (see Exhibit 1 in Volume 1 of this issue brief). But many aspects of care coordination, including sharing information through electronic health records and other forms of interprovider communication, are hard for patients to reliably discern. Therefore, narrative accounts are likely to be a relatively unreliable source of information about care coordination.

Patient or family perceptions could readily miss forms of communication or overlook breakdowns in communication. Additional research may help determine the direction of this bias, if any. But here, too, we anticipate that the yield will be limited.

Thus, many opportunities are open for future research that could either enhance existing approaches to narrative elicitation or better define the limits of reliable elicitation. We summarize these options in Exhibit 3.

Exhibit 3. Research priorities related to patient experiences with diagnosis

^v. Our own (unpublished) data suggest that about one-third of all patients prefer to not turn to the internet to augment what their clinician informs them about diagnoses.