Council on Medical Student Education in Pediatrics


Search This Site

Journal Club

Ely JW. Osheroff JA, Maviglia SM, and Rosenbaum ME.
Patient-Care Questions that Physicians are Unable to Answer. Joumal of the American Medicallnfonnatics Association.
Volume 14, Number 4, Pages 407-414. July/August 2007.

Ely JW. Osheroff JA, Maviglia SM, and Rosenbaum ME.
Patient-Care Questions that Physicians are Unable to Answer. Joumal of the American Medicallnfonnatics Association.
Volume 14, Number 4, Pages 407-414. July/August 2007.

Reviewed by Chris Maloney

What is the problem and what is known about it so far?
The authors acknowledge other classifications for answering questions however they could not identify any literature specifically looking at developing a taxonomy for unanswered questions. Furthermore, use of PICO (patient/problem/population, intervention, comparison/control, outcome) may not be adequate for unanswerable questions.

What did the researchers do in this particular study?
Who was studied?
The author's follow-up on a previous study where they followed 48 physicians in ambulatory settings (16 internists. 17 pediatricians and 15 FP•s). In the previous study they waited outside the rooms of clinicians and when the clinician exited they wrote down their question verbatim. The authors documented 1062 questions originally asked and 477 were not attempted to be answered, because the Physician "doubted that they could find a good answer quickly". Of the remaining 585 questions that were pursued 237 could not be answered in print and electronic resources.

This is the point were the current article picks up. Of the 237 questions not pursued. the authors felt the reason for difficu lty was apparent from reading 184 questions. A reason for difficulty answering the remaining 53 questions was not apparent from reading the question.

How was the study done?
Three of the authors independently used a qualitative analysis technique called "constant comparison method" to identify themes from a random sample of 52 of the 184 questions. They identified 12 themes and then added the 13th theme of "reason for difficulty not apparent" -the other 53 questions. Therefore they attempted to answer 105 of the previously "unanswerable" questions.

A single aulhor attempted to use 10 resources to answer lhe questions:, Thomson Healthcare, Micromedex, National Guideline Clearinghouse, Harrisons textbook of Internal Medicine, Nelson Textbook of Pediatrics, Sanford Guide. 2003 Red Book, eMedicine and Google. "These resources were selected by the investigators because they are clinically oriented. generally available, and commonty used in practice". The author searched each resource for up to 10 minutes to find the answer. Although there was potential for 1050 question/resource pairs. the authors did not use all resources if th.ey deemed them inappropriate e.g. using Harrison's for a pediatric question. Therefore they 818 question/resource pairs. A complete answer (defined as enough. information to direct patient care) was found for 224 (27% -requiring an average of 3.2 minutes/resource) pairs, a partial answer for 204 (25% " requiring an average of 5.0 minutes/resource) pairs and no answer found for 390 pairs (48% -requiring an average of 3.5 minutes/resource).

What did the researchers find?
The authors added 6 more themes to provide a taxonomy consisting of 19 theme types of questions with the 3 most frequent being "undiagnosed finding" (what is the approach to finding X), "conditional" (what is the management of X, given Y) and "compound" (can X cause V). They provide recommendations for Physicians -I. Select the most appropriate resource, 2. Rephrase the question to match the resource, 3. Use more effective search terms. They finally provide recommendations for authors (clinical information resource developer)¬ I. Authors should be explicit, 2. Anticipate questions thai are likely to arise in clinical practice. A full list of the taxonomy and recommendations can be found in the electronic appendix of the article (not included with th.e article). I/1

What were the limitations of the study?
The authors acknowledge four limitations -I. There strategy may not work well for "answemble" questions, 2. Finding a "complete" answer may not be a "correct" answer, 3. They only attempted to answer 105 of the 237 originally unanswered questions and the outcome may have been different. 4. They did not distinguish between an unanswered questions and unanswerable questions (those questions with no basis in the medical literature).

What are the implications of the study?
This study demonstrates the challenge of an evidence-based approach to clinical care. Spending 30+ minutes answering a clinical question is not possible in a busy clinical setting. Therefore, as medical student educators we need to emphasize "how" to ask questions that are answerable from the common resources used in clinical medicine. An EBM course should be offered and woven through the entire medical student curriculum. The axiom of "If you don't use it, you will loose it" applies to BBM tools and search strategies. The experts in search strategies are our medical librarians and clerkship directors and course directors should collaborate with this excellent resource.

Ed note: We all accept that EBM ought to guide our provision ofcare. Perhaps we need to create several paradigms to teach the application of EBM. Searching answers on the fly may not be the best method. Maybe questions need to be saved and investigated at a more leisurely pace. Team-based EBM may be anolher approach.

There nre some real challenges to applying EBM in an ambulatory setting. and this paper highlights most of them. The original paper found that of the 1062 questions that came up in the course of the subjects' practices, 67% went unanswered -pretty daunting numbers to overcome. The authors' recommendations to both providcrs and aulhors are succinct, but can we have an impact on editors of journals to help make authors' statements explicit? Depends on what the meaning of "is" is..... BZM

Return to Journal Club