Other Publications

Education Columns

Coffee Talks: How Physicians Balance Evidence-Based Medicine With Clinical Experience

Todd Felix, MD, Penn State University, Department of Family Medicine, Hershey Medical Center

Clinical reasoning (CR) is the application of critical thinking to the formulation of a differential diagnosis, diagnostic evaluation, and treatment plan for clinical problems. Information mastery and evidence-based medicine (IM/EBM) are the integration of clinical expertise with the best available external clinical evidence. Obtaining and critically appraising the evidence does not always equate with better decisions in clinical care, though. The evidence should be linked to the clinician’s expertise, clinical context or setting, and the patient’s individual circumstances, values, and preferences.

While CR and IM/EBM are, along with clinical experience and knowledge of the patient, essential for the practice of medicine, few medical schools have CR training or a formal IM/EBM curriculum through all 4 undergraduate years. Recent data have shown that clinical reasoning skills plateau after the second year of medical school, with a lack of accelerated progress during the clinical years of training.1

Our faculty addressed this plateau in development of CR skills by establishing a 4-hour case-based conference titled “Coffee Talks,” where third-year students bring cases seen during their month-long family medicine clerkship back to a group of 12–14 students and several faculty at the end of their rotation. The typical forum includes three to four experienced faculty with an EBM focus, seasoned clinicians, and an EBM resource librarian. The thinking skills of medical students are enhanced by faculty-led discussions of the cases in formulating a working differential that is assessed and modified after each segment of the student’s systematic case presentation. The discussions are focused around key points of the differential, diagnostic decisions, and treatment regimens based on available data presented by the student. Students identify key clinical questions that arise during the discussion, and small teams of two to three students with an assigned faculty member are then tasked to utilize EBM resources in answering the identified questions. This is followed by a student-led discussion in the larger forum that enables the students to recognize the cognitive skills utilized daily by family physicians in the care of their patients.

Consider this scenario below as an example of how CR and EBM may be incorporated into one of our Coffee Talks.

A toddler at a birthday party becomes short of breath suddenly—already you are developing a differential diagnosis—the challenge becomes how you would proceed with a diagnostic workup and treatment plan quickly and efficiently.

Clinical reasoning has been defined as the ability to ‘‘. . . sort through a cluster of features presented by a patient and accurately assign a diagnostic label, with the development of an appropriate treatment strategy as the end goal.’’2 A seasoned clinician would likely use clinical intuition, considering allergy, asthma attack, and foreign body as their top differentials. A medical student, using a more analytical approach, may yield a more expansive differential, perhaps pneumothorax, pneumonia, cardiac shunt, etc.

Regarding our case scenario, what if the child was a known asthmatic?

Most physicians move freely between pattern recognition and analytic reasoning as a given case necessitates, although the majority of our thought process is intuition/pattern recognition. Introducing Bayes theorem (pretest probability and likelihood ratio) to students reinforces the value of determining the probability for a given disease prior to testing and gives grounds to whether physicians will test, treat, or observe. In our scenario above, our pretest probability would likely be equal for the three diagnoses listed. The personal history of asthma would increase the likelihood for allergy and asthma.

Common things being common, you start treatment for an acute asthma exacerbation—but what should be your initial management option (albuterol, levalbuterol, racemic epi, steroids)?

The skill set to interpret medical data, search valid resources, and apply this to clinical scenarios in real time is an invaluable tool for students and residents. Specific training in application of EBM in clinical care is often lacking during the clinical years of medical student education. Role modeling an evidence-based approach is an important part of teaching clinical reasoning. Without data, well-founded diagnoses and management decisions are difficult to make and may become anecdotal exercises, which can leave students puzzled by the decision-making process.

But the patient worsens, so you reexamine the child and consider the birthday party environment, realizing that there was peanut butter cake, and symptoms started soon after ingestion.

Physicians frequently misdiagnose patients, despite significant improvements in diagnostic technology, including advanced imaging techniques. Groopman estimates about 15% of all people are misdiagnosed, possibly as high as 25%.3 Think of the last radiographic diagnosis made that did not seem to fit the diagnosis. Regular discussions are had with students and residents about the “pneumonia” admission without cough, fever, or leukocytosis despite a positive CXR finding. Most doctors, within the first 18 seconds of seeing a patient, will interrupt the patient’s story and generate an idea/diagnosis, which can easily be influenced by the cognitive bias of premature closure once a “diagnosis” is found.3

Our patient is administered an Epi Pen and improves quickly, then discharged with a diagnosis of peanut allergy. Oral desensitization therapy was reviewed with parents, who elected to follow total avoidance based on evidence presented.

Our Coffee Talks attempt to prepare future physicians with the tools necessary to practice in the ever-changing medical landscape. With studies showing physicians have up to two questions for every three patient visits and an average search time of less than 1 minute, developing information mastery skills is vital to student education.4 By focusing on the necessary clinical reasoning skills and appropriate use of information in a systematic fashion, students will be better positioned to enter their medical careers with a solid foundation. This unique forum has allowed for clinical experience, student collaboration, critical thinking, information mastery, and professionalism to be highlighted.

References
1. Williams RG, Klamen DL, White CB, et al. Tracking development of clinical reasoning ability across five medical schools using a progress test. Acad Med 2011;86(9):1148-54.

2. Rencic J. Twelve tips for teaching expertise in clinical reasoning. Med Teach 2011;33(11):887-92.

3. Groopman J, ed. How doctors think. Boston: Houghton Mifflin Company, 2007.

4. Ely JW, osheroff JA, Ebell MH, et al. Analysis of questions asked by family doctors regarding patient care. BMJ 1999;319(7206):358-61.

Other resources
Trowbridge R. Twelve tips for teaching avoidance of diagnostic errors. Med Teach 2008;30(5):496-500.

Ely JW, Graber ML, Croskerry P. Checklists to reduce diagnostic errors. Acad Med 2011;86(3):307-13.

Croskerry P. The theory and practice of clinical decision making. Can J Anesth 2005;52:R1-R8.

Mamade S, Schmidt HG. The structure of reflective practice in medicine. Med Educ 2004;38(12):1302-8.

Sackett D, ed. Evidenced-based medicine: how to practice and teach EBM. New York: Churchill Livingston, 1997.

Slawson D, Shaughnessy A, Bennett J. Becoming a medical information master: feeling good about not knowing everything. J Fam Pract 1999;48(2):135-9.

Slawson D, Shaughnessy A. Teaching EBM: should we be teaching information mastery instead? Acad Med 2005;80:685-9.

Ask a Question
AI Chatbot Tips

Tips for Using STFM's AI Assistant

STFM's AI Assistant is designed to help you find information and answers about Family Medicine education. While it's a powerful tool, getting the best results depends on how you phrase your questions. Here's how to make the most of your interactions:

1. Avoid Ambiguous Language

Be Clear and Specific: Use precise terms and avoid vague words like "it" or "that" without clear references.

Example:
Instead of: "Can you help me with that?"
Try: "Can you help me update our Family Medicine clerkship curriculum?"
Why this is important: Ambiguous language can confuse the AI, leading to irrelevant or unclear responses. Clear references help the chatbot understand exactly what you're asking.

2. Use Specific Terms

Identify the Subject Clearly: Clearly state the subject or area you need information about.

Example:
Instead of: "What resources does STFM provide?"
Try: "I'm a new program coordinator for a Family Medicine clerkship. What STFM resources are available to help me design or update clerkship curricula?"
Why this is better: Providing details about your role ("program coordinator") and your goal ("design or update clerkship curricula") gives the chatbot enough context to offer more targeted information.

3. Don't Assume the AI Knows Everything

Provide Necessary Details:The STFM AI Assistant has been trained on STFM's business and resources. The AI can only use the information you provide or that it has been trained on.

Example:
Instead of: "How can I improve my program?"
Try: "As a program coordinator for a Family Medicine clerkship, what resources does STFM provide to help me improve student engagement and learning outcomes?"
Why this is important: Including relevant details helps the AI understand your specific situation, leading to more accurate and useful responses.

4. Reset if You Change Topics

Clear Chat History When Switching Topics:

If you move to a completely new topic and the chatbot doesn't recognize the change, click the Clear Chat History button and restate your question.
Note: Clearing your chat history removes all previous context from the chatbot's memory.
Why this is important: Resetting ensures the AI does not carry over irrelevant information, which could lead to confusion or inaccurate answers.

5. Provide Enough Context

Include Background Information: The more context you provide, the better the chatbot can understand and respond to your question.

Example:
Instead of: "What are the best practices?"
Try: "In the context of Family Medicine education, what are the best practices for integrating clinical simulations into the curriculum?"
Why this is important: Specific goals, constraints, or preferences allow the AI to tailor its responses to your unique needs.

6. Ask One Question at a Time

Break Down Complex Queries: If you have multiple questions, ask them separately.

Example:
Instead of: "What are the requirements for faculty development, how do I register for conferences, and what grants are available?"
Try: Start with "What are the faculty development requirements for Family Medicine educators?" Then follow up with your other questions after receiving the response.
Why this is important: This approach ensures each question gets full attention and a complete answer.

Examples of Good vs. Bad Prompts

Bad Prompt

"What type of membership is best for me?"

Why it's bad: The AI Chat Assistant has no information about your background or needs.

Good Prompt

"I'm the chair of the Department of Family Medicine at a major university, and I plan to retire next year. I'd like to stay involved with Family Medicine education. What type of membership is best for me?"

Why it's good: The AI Chat Assistant knows your role, your future plans, and your interest in staying involved, enabling it to provide more relevant advice.

Double Check Important Information

While the AI Chat Assistant is a helpful tool, it can still produce inaccurate or incomplete responses. Always verify critical information with reliable sources or colleagues before taking action.

Technical Limitations

The Chat Assistant:

  • Cannot access external websites or open links
  • Cannot process or view images
  • Cannot make changes to STFM systems or process transactions
  • Cannot access real-time information (like your STFM Member Profile information)

STFM AI Assistant
Disclaimer: The STFM Assistant can make mistakes. Check important information.