Intended for healthcare professionals

Practice Practice Pointer

Five strategies for clinicians to advance diagnostic excellence

BMJ 2022; 376 doi: (Published 16 February 2022) Cite this as: BMJ 2022;376:e068044
  1. Hardeep Singh, professor of medicine1,
  2. Denise M Connor, associate professor of medicine2 3,
  3. Gurpreet Dhaliwal, professor of medicine 2 3
  1. 1Center for Innovations in Quality, Effectiveness and Safety, Michael E DeBakey Veterans Affairs Medical Center and Baylor College of Medicine Houston, TX, USA
  2. 2Department of Medicine, University of California San Francisco, San Francisco, CA, USA
  3. 3Medical Service, San Francisco Veterans Affairs Medical Center, San Francisco, CA, USA
  1. Correspondence to: H Singh hardeeps{at} Twitter @HardeepSinghMD, @Denise_M_Connor, @Gurpreet2015

What you need to know

  • The World Health Organization and the National Academy of Medicine (US) have identified measuring and reducing diagnostic error as a patient safety priority

  • Diagnosis is a process that is influenced by systems, cognitive, teamwork, and social factors that may either enhance or reduce diagnostic accuracy

  • Clinicians can integrate diagnostic performance feedback into their day-to-day work

  • Clinicians can take steps to mitigate biases (regarding race, ethnicity, gender, and other identities) that run counter to their values and impair diagnostic performance

  • Clinicians can integrate the expertise of other health professionals, patients, and families to reimagine the routines and culture around diagnosis

Diagnostic accuracy is an important component of clinical excellence. However, diagnostic errors—failures to establish an accurate and timely explanation of a patient’s health problem or to communicate that explanation to the patient—harm patients worldwide.12 In a recent UK study, diagnostic errors occurred in 4.3% of primary care consultations.3 A meta-analysis estimated that nearly 250 000 harmful diagnostic errors occur annually among hospitalised adults in the United States.4 The National Academies of Science, Engineering, and Medicine’s report Improving Diagnosis in Health Care highlighted that most people will experience at least one diagnostic error in their lifetime.1 The World Health Organization now identifies measuring and reducing diagnostic error as a patient safety priority.5

Diagnostic excellence involves making a correct and timely diagnosis using the fewest resources while maximising patient experience and managing uncertainty.6 Compared with other patient safety problems such as medication errors, procedural complications, and hospital acquired infections, diagnostic errors have received less investigation.78910 Diagnoses are rendered by clinicians but health system vulnerabilities frequently influence clinical reasoning and contribute to diagnostic error.1112

While system interventions are essential to achieving diagnostic excellence,13 this article focuses on the individual clinician. Based on current scientific literature, expert opinion, and theoretical frameworks pertaining to diagnostic decision making and learning, we propose five strategies for clinicians to consider to achieve diagnostic excellence and reduce diagnostic error in their practice.

Why diagnostic errors occur

Diagnosis is a complex categorisation task driven by mental models (illness scripts and diagnostic schemas) that reside in long term memory. Through education and experience, clinicians form illness scripts that encapsulate their knowledge of specific conditions (such as osteoarthritis) and develop diagnostic schemas that structure their approach to a specific health problem (such as knee pain). This cognitive process intersects with systems, teamwork, and social factors that can enhance or reduce diagnostic accuracy.14 Diagnosis is often a dynamic longitudinal process in which patients’ clinical features evolve, multiple interactions with the healthcare system occur, and additional information surfaces.15

Multiple factors can contribute to a delayed or inaccurate diagnosis. Clinicians may be unaware of a disease, or a disease may manifest in an atypical way. However, most diagnostic errors involve common conditions that clinicians are aware of.4161718 In these situations, diagnostic errors may result from inadequate data gathering, inadequate information processing, or inadequate application of knowledge.19202122 Clinicians may misestimate the likelihood of diseases they are familiar with (for example, overestimate the probability of cellulitis and underestimate the probability of venous stasis when evaluating a patient’s oedematous erythematous leg).23 Time and productivity pressures can adversely influence thinking. The electronic health record may hold diagnostic clues, but data may be disorganised, inaccurate, or fragmented across multiple care settings.2425 Clinicians often do not have time to search medical records and reference materials because of taxing data-entry demands. Patients and other health professionals often hold critical diagnostic information and hypotheses, but this team expertise is regularly overlooked.26 The way clinicians solicit, interpret, and pursue diagnostic information with patients can be influenced by stereotypes about various populations and communities.

Five strategies for diagnostic excellence

Seek diagnostic feedback

Feedback strategies, if effectively implemented, have been shown to lead to clinical practice improvement.27 Clinicians learn most from their day-to-day work, but it can be challenging to target knowledge and skill gaps that will reliably improve diagnostic performance.28 Learning from feedback can help clinicians evaluate their diagnostic accuracy and calibrate their future decisions.6293031

Two pathways can be useful for integrating feedback into day-to-day diagnostic work. First, we suggest tracking patients’ clinical outcomes by creating an electronic list of patients where the diagnosis is still evolving or questions remain, setting reminders to check diagnostic outcomes, and creating a log of lessons learnt.31 Second, we recommend clinicians learn through feedback on their diagnostic outcomes from other health professionals, patients, and families.32 This feedback can be a brief communication about a patient’s final diagnosis or outcome (“I have an update on that patient you saw”) or a conversation that highlights opportunities to improve diagnostic performance and reinforces practices that lead to diagnostic excellence.33 Routinely scheduled discussions of selected cases among peers interested in diagnostic excellence can enhance learning.34 To be effective, performance feedback from colleagues should be established as a non-punitive, cultural norm that is sensitive to clinicians’ potential discomfort around such discussions.333536

“Byte sized” learning

Digital learning activities can improve mental models and maximise diagnostic performance. Clinicians read the medical literature in order to acquire new information and maintain previous knowledge. Cognitive neuroscience has demonstrated that self testing (quizzing) by retrieving information from memory is a more effective way to enhance understanding and knowledge than re-reading the same content.37 To actualise this strategy, we recommend developing a recurring, structured routine that is grounded in tackling brief diagnostic challenges on digital platforms, including smartphone applications (apps), social media, and medical journal websites (see box 1). These resources provide concise cases that clinicians can use to practise diagnostic thinking and effectively increase the number of patients they “see” each day.38 The final diagnosis and brief teaching points in digital cases provide lessons that can be integrated to sharpen diagnostic schemas and illness scripts.

Box 1

Resources for short digital cases (“byte sized” learning)


Finding efficient training opportunities is not the main challenge; instead it is how clinicians can modify their routines such that 5-10 minutes of diagnostic training and practice is embedded in each day. This approach is typically more effective and practical than seeking large blocks of time for extended study.39

Consider bias

Recognition of the pervasiveness of cognitive biases (such as anchoring and availability bias) has raised awareness of fallibility in clinical decision making and encouraged humility during the diagnostic process. However, it remains difficult to identify our cognitive biases in real time or in hindsight,40 and debiasing interventions remain challenging to implement and require additional study.4142 While current diagnostic decision making literature focuses on understanding and addressing cognitive biases, disparities in diagnosis indicate that implicit bias and stereotyping can also lead to diagnostic errors.4344 There are numerous examples in the literature of diagnostic disparities linked with race and ethnicity, including disparities in hospitalisation for acute myocardial infarction, diagnosis of schizophrenia, diagnosis of appendicitis in children, and timely diagnosis of dementia.45464748

Harmful societal forces such as racism, sexism, transphobia, ageism, and ableism that run counter to clinicians’ values can affect diagnostic thinking despite the best intentions. Medical education lacks adequate representation of diverse individuals both as teachers and in training resources (such as textbooks). For example, teaching materials in the US typically illustrate and describe rashes as they appear on light skin, leading to failures to teach how to recognise key diagnostic clues in people with a full range of skin tones. This institutionalised racism results in knowledge gaps that can contribute directly to diagnostic error and harm.49 Prior experiences with discrimination can damage trust between patients and clinicians before the visit begins and may constrain the diagnostic information that patients feel comfortable or empowered to share.505152

While evidence on strategies to reduce the impact of implicit bias are still emerging, we suggest several pragmatic starting points. Taking time to see patients as individuals and finding common ground by learning about their lives and preferences can promote empathy and facilitate diagnostic evaluation based on individual, not stereotyped, attributes.5354 We recommend taking a diagnostic pause to consider whether alternative diagnoses would be entertained if a patient had a different background or identity. Aggregated data from clinical practices can be used to detect inequitable diagnostic approaches.55 For example, a practice may review its data to determine if urine toxicology screens are being ordered disproportionately for Black patients presenting with confusion. When data demonstrate biased diagnostic practices, individual clinicians and groups can develop systems-level solutions (such as requiring standardised criteria be met before ordering toxicology screens). Considering whether a practice reflects the background and identities of the patients it serves and making efforts to diversify clinicians and staff can foster a welcoming and equitable clinical environment. Strategies aimed at enhancing equity are most effectively implemented when clinicians commit to examining their own unconscious biases.56

Make diagnosis a team sport

Leveraging the expertise of a diagnostic team can help reduce errors.5758 For instance, a nurse may collect a nuanced time-course of illness; a physical therapist may detect an overlooked neurologic deficit; a dietician may discern a key nutrient deficiency. Patients and their families have important insights about their disease course and diagnostic hypotheses that warrant exploration.59

A true diagnostic team requires reimagining the culture around making a diagnosis, which is often viewed as the purview of doctors alone.60 Interprofessional huddles, where the hierarchy between physicians and other health professionals is flattened, can be effective forums to synthesise all diagnostic information and hypotheses. We suggest asking patients “What do you think you have?” or “What are you concerned this could be?” Such inquiry invites the patient into a conversation and can introduce new diagnostic possibilities.

Clinicians can leverage the power of group decision making to discuss diagnostic challenges with colleagues in other specialties, including laboratory medicine, pathology, and radiology, who can also help with test selection and interpretation.61 Increased experience with videoconferencing during the covid-19 pandemic has enhanced our ability to have real-time verbal discussions among colleagues,62 which is more likely to lead to satisfactory diagnostic decisions than asynchronous electronic communication through the electronic health record, email, or texting. Increasingly the diagnostic “team” will also include decision support systems and other technologies that augment diagnostic reasoning.63

Foster critical thinking

Critical thinking skills can help clinicians optimise their acquisition and interpretation of data throughout the diagnostic process.64 Critical thinking involves reflective scepticism, a propensity to consider diagnoses as provisional and subject to revision, and a willingness to seek and examine contradictory evidence. In one study of diagnostic errors in primary care, no differential diagnosis was documented in 81% of cases,16 suggesting that clinicians may not always consider alternative possibilities in their diagnostic thinking.

We suggest using intentional strategies that foster critical thinking in daily practice. When evaluating a 45 year old patient with new onset headaches, deliberately seeking both supportive and conflicting evidence on history and examination for a provisional diagnosis of migraine exemplifies critical thinking.65 Reflective scepticism is a cognitively demanding path that involves contemplating competing and mimicking conditions (such as intracranial hypotension or cluster headache), evaluating the case from different points of view (for example, if the patient’s partner expressed concern about the diagnosis because they never had headaches like this before), commitment to monitoring and collecting more data (such as reassessing for changes in cognitive function),66 and setting prompts for further investigation (such as head imaging if no improvement after two months of treatment). This metacognitive stance takes practice but may lead to enhanced diagnostic judgment.

How this article was made

The article was synthesised from multiple sources including (1) a review of peer reviewed literature, including narrative reviews and systematic reviews on diagnostic error, and interventions to reduce them; (2) research studies on feedback, calibration, clinical reasoning, teamwork, and disparities in diagnosis; (3) opinion based on commentaries and perspectives from the literature on potential strategies to reduce error with special focus on individual clinical reasoning, developing expertise, learning from feedback, and critical thinking; (4) reports from several international bodies including National Academies of Science, Engineering and Medicine, National Quality Forum, and the World Health Organization; (5) conceptual frameworks that outline approaches to components of the diagnostic process, including generating and providing feedback; and (6) input from patient reviewers.

Education into practice

  • How might you develop a system to track your diagnostic decisions and outcomes?

  • Do you examine and reflect on your biases and their potential impact on the diagnostic process?

  • Which one of the strategies in this article could you encourage as a career-long habit for a trainee you work with?

How patients were involved in the creation of this article

We invited two patient advocacy and diagnostic error experts (Helen Haskell, president of Mothers Against Medical Error, and Desiree Collins Bradley, network lead of Patient Partner Innovation Community) to provide comments on an earlier version of this paper. They advised us to clarify the intent of the “byte sized” learning strategy, be more explicit about the problem of bias, and strengthen recommendations related to patients and teamwork. They also advised us to highlight the role of the electronic health record. This input was used to make changes to the paper prior to submission.


  • Funding: HS is funded in part by the Houston Veterans Administration (VA) Health Services Research and Development (HSR&D) Center for Innovations in Quality, Effectiveness, and Safety (CIN13-413), the VA HSR&D Service (IIR17-127 and the Presidential Early Career Award for Scientists and Engineers USA 14-274), the Agency for Healthcare Research and Quality (R01HS27363), the CanTest Research Collaborative funded by a Cancer Research UK Population Research Catalyst award (C8640/A23385), and the Gordon and Betty Moore Foundation.

  • Competing interests: We have read and understood BMJ policy on declaration of interests and have no relevant interests to declare.

  • Provenance and peer review: Commissioned; externally peer reviewed.