Clinical reasoning is fundamental for the safe, effective practice of medicine. In its seminal 2015 publication “Improving Diagnosis in Healthcare,” the National Academy of Medicine urgently called for enhanced training in the diagnostic process for all healthcare professionals to help reduce the burden of diagnostic errors occurring each year.1 In the years since this publication, many resources have emerged to address the lack of clinical reasoning education available.2 Published cases, such as in the Journal of General Internal Medicine’s “Exercises in Clinical Reasoning,” have become popular vehicles for teaching clinical reasoning principles, but lack interactivity.3 Although interactive digital resources for simulating practice with unknown cases do exist, many are behind paywalls or are too long to allow meaningful participation by busy trainees.

To address the need for efficient, interactive resources for teaching clinical reasoning, the University of Pittsburgh’s internal medicine residents developed the “Pitt Puzzles” series. Written by residents and edited by faculty with training in clinical reasoning, this series presents participants with brief, interactive, case-based tutorials to enhance clinical reasoning skills on an easily accessible online platform. The concept for this resource was developed in conjunction with the University of Pittsburgh Medical Center (UPMC) Clinical Center for Medical Decision Making, as part of a comprehensive curricular intervention to create a “culture of clinical reasoning” within the residency.4 The curriculum emphasizes a common language for discussing clinical reasoning and diagnostic error while modeling a systematic approach to clinical problem solving through case conferences and didactic teaching. Adding resident-driven, interactive exercises (Pitt Puzzles) to the curriculum has increased resident engagement while also stimulating retention of reasoning concepts through spaced repetition.

Pitt Puzzles are a collection of case-based educational exercises that require 10-15 minutes each to complete and are accessible on mobile electronic devices. Cases focus specifically on atypical presentations of typical diseases or classic presentations of rarer diseases to promote the development of robust illness scripts. In contrast to traditional case simulations focusing on building medical knowledge, Pitt Puzzles focus on clinical reasoning principles as the primary learning objective. For example, a case of toxic shock syndrome may be used to demonstrate how changes to the problem representation can change the differential diagnosis through a framing effect. Throughout each Puzzle, case information is revealed sequentially, with probing questions asking participants to write summary statements, describe a differential diagnosis or a problem list, identify cognitive biases at play, or choose the next step in the diagnostic work up. As such, the instruction focuses on the process of clinical reasoning rather than the resulting diagnosis. This focus on metacognition and a common language of clinical reasoning provides a foundational framework that physicians-in-training can use to continue to develop their clinical reasoning throughout their careers.1, 5

Pitt Puzzles are constructed by resident volunteers, many of whom have a special interest in clinical reasoning and/or medical education, in cooperation with faculty mentors. Mentors assist residents in identifying an appropriate case and the associated learning objectives, after which residents outline and write the Puzzle cases. Cases are de-identified and the details changed to protect patient privacy. A final version is then negotiated over several drafts with the faculty mentor. Depending on interest from year to year, case mentors have also included senior residents or chief residents. For example, in 2022 we have two senior residents who will serve as editors with a single faculty editor-in-chief. Once finalized, cases are entered into Qualtrics, an online survey tool which allows branching question logic to further simulate the results of different choices participants make during the case. Since Pitt Puzzles’ inception in 2018, 27 cases have been written with an average of 125 participants per case. Engagement both locally and broadly has exceeded expectations. Feedback from our residency program has been overwhelmingly positive and there has been continued interest from our residents to continue to develop additional cases.

There were several challenges during the initial development of the Pitt Puzzle program: how to increase participation among busy trainees, and how to balance trainees’ desire for clinical knowledge content with our clinical reasoning-specific learning objectives. We suspect that low initial trainee participation in the program had two likely causes: first, Puzzles have never been required learning for trainees, so cases were less likely to be completed despite their short length; second, Puzzles were initially distributed to trainees by e-mail only, making them susceptible to being ignored due to e-mail fatigue. Puzzle participation subsequently increased during the COVID-19 pandemic as trainees’ appetite for digital educational increased and as familiarity with the program grew over time. We were also able to increase participation by releasing cases via the Twitter handle @MedEdPGH, thus expanding our potential audience to other institutions.

The most frequent negative feedback from post-case surveys has been the request for more disease-specific teaching. While the easiest solution would have been lengthening cases to add clinical content, we wanted to maintain each case’s brevity while balancing our educational agenda with our learners’ educational desires. Thankfully, there are multiple avenues for disease-specific clinical reasoning teaching. For example, teaching about testing characteristics (e.g., the sensitivity of pleural fluid cytology for malignancy, or D-dimer likelihood ratios for pulmonary embolism) allows for dual discussion of Bayesian reasoning and clinical content. In other instances, narrowing the scope of cases to focus on a single diagnostic decision allowed more space for clinical content discussion while still emphasizing reasoning-specific teaching points. We observe that, whereas required education can afford to have teachers solely decide content coverage, optional educational must adjust to its audience’s interests lest it risk losing the audience’s attention.

In summary, we have developed a resident-driven, case-based clinical reasoning educational tool that allows learners of all levels an opportunity for skill practice in the context of a busy clinical schedule. This educational tool is innovative in that it is created by resident physicians, for resident physicians and focuses not only on medical knowledge but also on clinical reasoning concepts. Pitt Puzzles can serve as a model for other grass-roots educational projects to engage learners in their own educational process. Our experience also reinforces the need to listen to one’s educational audience when developing an optional educational resource in order to balance audience engagement with the pre-determined learning objectives. Next steps include evaluating what educational benefits case creation has for the resident authors and editors of Pitt Puzzles, as well as exploring comparative effectiveness of this modality for teaching clinical reasoning compared to other teaching settings.


  1. Committee on Diagnostic Error in Health Care, Institute of Medicine, The National Academies of Sciences, Engineering, and Medicine. Chapter 9: The path to improve diagnosis and reduce diagnostic error. In: Balogh EP, Miller BT, Ball JR, eds. Improving Diagnosis in Health Care. Washington, DC: National Academies Press (US); 2015. Accessed November 15, 2022.
  2. Manesh R, Dhaliwal G. Digital tools to enhance clinical reasoning. Med Clin North Am. 2018;102(3):559-565. Doi:10.1016/ a.2017.12.015.
  3. Steinhilber S, Estrada CA. Exercises in clinical reasoning: a retrospective. J Gen Intern Med. 2017;32(1):1-2. Doi:10.1007/s11606-016-3906-9.
  4. DiNardo D, Bonifacino E, Tilstra SA, et al. From virtual autopsies to expedited stroke detection: How facilities are improving the diagnostic process. PA Patient Saf Advis. 2018 Oct 31;15[Suppl 1]:51-61.
  5. Dhaliwal G, Ilene J. Clinical reasoning: Talk the talk or just walk the walk. J Grad Med Educ. 2016;8(2):274-276. Doi:10.4300/JGME-D-16-00073.1



Clinical Practice, COVID-19, Medical Education, Medical Ethics, Research, SGIM

Author Descriptions

Dr. McQuade ( is a clinical assistant professor of medicine in the Division of General Internal Medicine at the University of Pittsburgh Medical Center. Dr. Wojcik ( is a fellow in the Division of Pulmonary and Critical Care at the University of Pittsburgh Medical Center. Dr. Fleshner ( is an assistant professor of medicine at the University of Colorado Anschutz Medical Campus. Dr. Smith ( is an assistant professor of critical care medicine in the Division of Critical Care Medicine at the University of Maryland School of Medicine. Dr. Bonifacino ( is an assistant professor in the Department of General Internal Medicine at the University of Pittsburgh Medical Center