What Stumps Primary Care Clinicians? An Analysis of Diagnostic Uncertainty Cases Discussed in Practice Inquiry Meetings
Running Head: What Stumps Primary Care Clinicians?
Authors: Sommers, Lucia Siegel1; Vaghani, Viralkumar2,3; Giardina, Traber D.2; Singh, Hardeep 4,2
Affiliations:
1. University of California System, Family and Community Medicine; San Francisco, CA, USA
2. Michael E. DeBakey Veterans Affairs Medical Center, Houston VA Center for Innovations in Quality, Effectiveness and Safety; Houston, USA
3. Baylor College of Medicine, Department of Medicine, Houston, TX, USA
4. Baylor College of Medicine, Section of Health Services Research, Department of Medicine, Houston, TX, USA 77030
Corresponding Author:
Dr. Lucia Siegel Sommers
164 Underhill Ave., Apt. 1
Brooklyn, NY 11238
Phone: 415 710 9172
Email: Lucia.Sommers@ucsf.edu
INTRODUCTION
Diagnostic difficulty is frequent in primary care yet little is known about what clinical scenarios present uncertainty to primary care practitioners (PCPs) and how they manage them.1,2Uncertainty can arise from a patient presenting with worrisome symptoms that, over the course of the visit, form the substrate for an efficient investigation and reasonable path forward. Alternatively, uncertainty can arise from a patient presenting with the same symptoms that this time, given patient context, cause worry, frustration, cognitive overload, and hasty decisions. Such primary care scenarios, how they evolve and resolve, are rarely shared among clinicians since medical culture has undervalued acknowledgment and communication of uncertainty.3,4 Discomfort in confronting ‘not knowing’ engenders self-doubt, guilt and shame.5Exacerbated by shortened patient visits and electronic health record (EHR) burden, inadequate management of diagnostic uncertainty is associated with medical error, clinician burnout and increased costs.6
We provide an initial description of patients causing diagnostic uncertainty that PCPs brought to regularly scheduled, facilitated, confidential practice meetings.7 Findings could advance understanding of how physicians talk about and manage patients with diagnostic uncertainty. Information about these patients’ characteristics and how their clinicians responded could better define diagnostic uncertainty in primary care, explain and acknowledge the value of ‘uncertainty work’ PCPs do, and promote deliberate attention to its management.
METHODS
We analyzed data from 459 patient scenarios where PCPs presented current, case-based instances of uncertainty as defined by themto peers in “Practice Inquiry (PI) Colleague Group” meetings. Meetings were conducted at 15 San Francisco-Bay Area primary care practices between 2002 - 2015 as part of a university-based continuing medical education program.2 Practices were recruited incrementally as the first author contacted practices with graduates in Family Medicine and General Internal Medicine from the University of California, San Francisco. Most groups met on a monthly basis; the oldest group was 13 years old, the youngest group, 2 years old. Average number of cases provided by a group was 17. (Median: 9, Mode: 19)
The lead investigator (LS) facilitated 99.5% of meetings based on a structured group process (See Figure 1.) and recorded data for consecutive patients presented into a case log organized into four categories: 1) uncertainty statement/question; 2) patient and clinician information provided by case presenter during first 2-3 minutes of presentation, including symptoms, physical findings, lab/imaging findings, medical/social history, and patient-clinician relationship; 3) colleague group’s responses to presentation; and 4) patient follow-up offered by presenter at subsequent meetings. (See Figure 2.) At these meetings, colleagues reviewed log entries to edit and update.
We used data from categories 1 and 2 (excluding colleague responses and follow-up information) to select diagnostic uncertainty cases from the larger case cohort. To select cases, we used previously developed criteria from published work for PCPs’ direct and indirect expressions of certainty (e.g., question marks, absence of diagnosis at visit end) in electronic medical record notes. We refer to these in the paper as the “Bhise criteria.” 8 Data for our study came directly from PCP case descriptions of what confused or puzzled them and verbalized within the first 2-3 minutes of their presentations. Different and rudimentary experiences of not knowing would occasionally be presented revealing diagnostic uncertainty indirectly as more general statements (e.g., “Am I doing everything I should do for this patient? What should I do next?” (See Figure 1, Case 1.) Such cases, lacking the key words/phrases/punctuation used in the Bhise criteria, nonetheless, dealt with the same topics as those coded by the Bhise criteria. Cases identified using the Bhise criteria and those characterized by more general statements were selected by two authors (LS and V); a third (HS) adjudicated differences. For each case selected, data were extracted from the case log in categories 1 and 2 above: the uncertainty statement and patient information provided in first 2-3 minutes. Two authors (LS and V) coded these data to describe patients’ presenting symptoms, abnormal physical exam findings, abnormal lab/ imaging; patients’ known diagnoses/conditions; and diagnoses/conditions identified by presenter as possible uncertainty explanations. We used a consensus approach to reconcile differences. Diseases/conditions were categorized using ICD-11 codes.
RESULTS
Of 459 cases PCPs presented to colleagues, 258 (56%) involved diagnostic uncertainty. Of these, 85% (220) were identified by the Bhise criteria and the remaining 15% (38) using clinicians’ more general expressions of not knowing. The latter cases dealt with 19 patients (14%) presented as reflected-upon, suspected, adverse diagnostic events. (See Case 5, Figure 1.) These included 6 deaths. The remaining cases comprised patients where clinicians were uncertain about how to diagnose cognitive deficits (e.g.,“ I don’t know what’s going on with this patient…”), adult or child abuse ( e.g., “I fear that something bad is going to happen.”), substance abuse ( e.g., “This patient may be addicted to what I’ve been prescribing.”), and preventable risk ( “Never seen a vitamin D level this low!”).
Sixty-one percent of the 258 cases involved women, the median age decade was 50-59 years. Forty-four percent presented with two or more prior diseases/conditions. Fifty-three percent of patients had one or more non-pain, symptoms; 7% had only pain symptoms in the absence of abnormal physician exam or lab/imagining findings. A third of cases were both diagnostic and management dilemmas (See Case 3, Figure 2.)
Characteristics describing patients’ current status, past known diagnoses, and possible diagnoses explaining uncertainty are displayed in Table 1. Presenters discussed patients’ past known diagnoses in 186 (72%) cases; of these, the most prevalent, previously known diagnoses were mental health/ behavioral/ neurodevelopmental conditions (17%). Presenters speculated about possible diagnoses that could explain uncertainty in 133 cases (52%); of these, the most prevalent, possible diagnoses were mental health/behavioral/neurodevelopmental conditions (27%).
DISCUSSION
In this report, we describe an initial analysis of real-time, diagnostic uncertainty case discussions among PCPs at regularly scheduled, Practice Inquiry Colleague Group meetings. The 259-case dataset is unique: it was created over 13 years as PCPs came to meetings and, with no criteria defining ‘clinical uncertainty,’ presented patients that perplexed them for any reason. We uncovered new and different experiences of uncertainty not covered by the Bhise criteria and requiring further analysis and validation. For this initial study, we used limited criteria for identifying uncertainty in primary care. Future work with this dataset should consider how uncertainty has been conceptualized by several clinical uncertainty researchers and theorists such as Fox (uncertainty “forms” 9 and later, “themes”10), Beresford (“dimensions” 11), Cassell (“roots” 12), Djulbegovic (“knowledge deficiency” types13 ), and Han (“sources,” “issues,” “locus” 14 ). The cases we found that did not meet the Bhise criteria often identified uncertainty experiences that were presented as broader in scope than a single symptom or finding (e.g., diagnosing abuse, assessing disease risk); reflections on a untoward diagnostic outcome; treating a patient while the diagnostic process progresses; and dealing with one’s emotions while coping with not knowing. An analysis that starts with PCPs’ actual questions and statements of not knowing, coupled with appreciation of prior work to conceptualize clinical uncertainty, could result in a theoretical framework for defining and managing diagnostic uncertainty that could better support PCPs in daily practice.
Certain clinical scenarios, such as patients with mental health, behavioral/neurodevelopmental conditions who present with a new symptom appear to have posed special diagnostic challenges. What are these patients’ co-morbidities and how do they affect diagnostic thinking? Although pain alone appears to be less likely, what are their presenting complaints? Analysis of the third category of data collected - Discussion Points - could reveal colleagues’ response patterns to these patients and how they suggest proceeding. For example, how did colleagues, using abductive reasoning in addition to intuitive and analytic thinking, generate new thinking about these patients?15,16 Did they comment about the new symptoms (in the context of already-known mental illness) as in need of immediate intervention or watchful waiting? How did they opine about shared decision-making for new symptom work-up and safety-netting? (See Case 4 in Figure 1.)
Further qualitative work should study how clinicians presented their experiences of diagnostic uncertainty. What were their specific concerns? (e.g., symptom cause, completeness of work-up, managing the diagnostic process while providing care, disclosing uncertainty, handling their own emotions) Additionally, understanding clinicians’ responses to their colleagues’ cases (e.g., advice given, literature suggested, support offered) could offer a useful window into potential value of PI Colleague groups and how they can be enhanced. For example, how should facilitators be trained to lead discussions of critical thinking, knowledge gaps, system deficits, and clinicians’ emotional reactions to uncertainty? Lastly, approximately one third of the cases had patient follow-up presented at subsequent meetings. Analysis of these could contribute to understanding how clinicians experience uncertainty resolution.
Longevity of PI Colleague Groups, new group formation (20 since 2015) and high scores on annual satisfaction surveys 1suggest they are useful forums for supporting PCPs and their organizations in improving care while revealing the challenging work that PCPs do. They merit continued evaluation and broader implementation to become safe, ‘slowing down’ places for colleagues to engage uncertainty through experiencing the paradox of expert practice: “to act with confidence while simultaneously remaining uncertain.”16
CONFLICT OF INTEREST: Neither the first author nor the co-authors have any conflict of interest to report in regards to this manuscript.
BIBLIOGRAPHY
Alam R, Cheraghi-Sohi S, Panagioti M, Esmail A, Campbell S, Panagopoulou E.
Managing diagnostic uncertainty in primary care: a systematic critical review. BMC Fam Pract. 2017 Aug 7;18(1):79.
  1. Sommers LS, Launer J. Clinical uncertainty in primary care: the challenge of collaborative engagement. erlin: Springer Science & Business Media; 2013.
  2. McIntyre N, Popper K. The critical attitude in medicine: the need for a new
ethics. Br Med J (Clin Res Ed). 1983 Dec 24-31;287(6409):1919-23.
Gheihman G, Johnson M, Simpkin AL. Twelve tips for thriving in the face of
clinical uncertainty. Med Teach. 2019 Mar 26:1-7.
Hoffman JR, Kanzaria HK. Intolerance of error and culture of blame drive
medical excess. BMJ. 2014 Oct 14;349.
Singh H, Sittig DF. Were my diagnosis and treatment correct? No news is not
necessarily good news. J Gen Intern Med. 2014 Aug;29(8):1087-9.
  1. Sommers, L. S., Morgan, L., Johnson, L., et al. (2007). Practice inquiry: Clinical uncertainty as a focus for small-group learning and practice improvement. Journal of General Internal Medicine, 22 (2), 246–252.
  2. Bhise V, Rajan SS, Sittig DF, Vaghani V, Morgan RO, Khanna A, Singh H.
Electronic health record reviews to measure diagnostic uncertainty in primary care. J Eval Clin Pract. 2018 Jun;24(3):545-551.
  1. Fox, R. C. (1957). Training for uncertainty. In R. K. Merton, G. G. Reader, & P. Kendall (Eds.), The student-physician: Introductory studies in the sociology of medical education . Cambridge: Harvard University Press.
  2. Fox, R. C. (2000). Medical uncertainty revisited. In G. Albrecht, R. Fitzpatrick, & S. Scrimshaw (Eds.), Handbook of social studies in health and medicine (pp. 409–25). London: Sage.
  3. Beresford EB. Uncertainty and the shaping of medical decisions. Hastings Cent
Rep. 1991 Jul-Aug;21(4):6-11.
  1. Cassell EJ. The sorcerer’s broom. Medicine’s rampant technology. Hastings Cent Rep. 1993 Nov-Dec;23(6):32-9. PubMed PMID: 8307745.
  2. Djulbegovic, B, Iztok H, Greenland, S. ‘Uncertainty in Clinical Medicine” in D.M. Gabbay. J Woods. Editors. Handbook of The Philosophy of Science: Philosophy of Medicine. San Diego: New Holland, 2011, pp.299-356.
  3. Han, P. K., Klein, W. M., & Arora, N. K. (2011). Varieties of uncertainty in health care: A concep- tual taxonomy. Medical Decision Making, 31 (6), 828–838.
  4. Stanley DE, Nyrup R. Strategies in Abduction: Generating and Selecting
Diagnostic Hypotheses. J Med Philos. 2020 Jan 7. pii: jhz041. doi: 10.1093/jmp/jhz041. [Epub ahead of print]
  1. Vanstone M at al. Experienced physician descriptions of intuition in clinical reasoning: a typology. Diagnosis 2019 Aug 27;6(3):259-268.
  2. Ilgen JS, Eva KW, de Bruin A, Cook DA, Regehr G. Comfort with uncertainty:
reframing our conceptions of how clinicians navigate complex clinical situations. Adv Health Sci Educ Theory Pract. 2018 Nov 2