|Home | About | Journals | Submit | Contact Us | Français|
To categorise questions that emergency department physicians have during patient encounters.
An observational study of 26 physicians at two institutions. All physicians were followed for at least two shifts. All questions that arose during patient care were recorded verbatim. These questions were then categorised using a taxonomy of clinical questions.
Physicians had 271 questions in the course of the study. The most common questions were about drug dosing (35), what drug to use in a particular case (28), “what are the manifestations of disease X” (23), and what laboratory test to do in a situation (21). Notably lacking were questions about medication costs, administrative questions, questions about services in the community, and pathophysiology questions.
Emergency department physicians tend to have questions that cluster around practical issues such as diagnosis and treatment. In routine practice they have fewer epidemiologic, pathophysiologic, administrative, and community services questions.
Point of care access to information is particularly critical in the emergency department (ED) where one does not have the luxury of multiple patient visits to diagnose and treat a patient's disorders. In a prior study we reported on information sources that ED physicians use to answer questions at the point of care and barriers to answering these questions.1 In this study, we categorise the types of questions that ED physicians have based on a taxonomy of clinical questions previously developed by Ely et al.2 Understanding the gaps in physicians' knowledge has several practical applications. First, if it is found that physicians are uniformly lacking knowledge in one area (such as interpretation of diagnostic tests, for example), residency and continuing education can be oriented towards addressing these knowledge deficiencies. Second, understanding holes in physician knowledge will allow planners to develop point of care libraries that closely reflect the type of cognitive information that physicians are lacking.
This was an observational study undertaken in two EDs in the midwest of the USA with patient volumes of 35000 and 50000 per year. Each institution sees a broad range of patients including major trauma. The University of Iowa is a tertiary care and level 1 trauma centre with a major referral population from throughout the state and beyond. St Luke's Hospital is an inner city community hospital. Sixteen academic (University of Iowa, Iowa City, Iowa) and 10 community (St Luke's, Cedar Rapids, Iowa) based emergency physicians participated in the study. Physicians and institutions were initially recruited by mail with a follow up telephone call. After agreeing to participate, each physician was observed for at least two shifts by one of the investigators (BR). All study physicians were board certified in emergency medicine except for two who are faculty in emergency medicine at the academic institution. These two physicians each have over 15 years of emergency medicine experience. Physicians were asked after each patient visit if they had any questions related to the patient encounter. Questions could be about diagnosis, medication use, available social resources, need for admission, etc. All questions were recorded verbatim on a standardised form.
Questions were categorised based on a previously described taxonomy developed by Ely et al for classifying clinical questions.2 Questions were categorised by three physicians (CJ, DA, BP) acting independently without prior knowledge of each other's classification. If all three physicians agreed on the question's taxonomy assignment, the assignment was reviewed by MG to make sure that it had face validity. Any questions for which all three scorers did not agree were adjudicated by MG and JWE and then discussed with other scorers until a consensus was obtained. This project was approved by each institution's human subjects review board and Health Insurance Portability and Accountability Act (HIPPA) compliance committees.
All three scorers (DA, CJ, BP) agreed on question classification for 172 of the 271 questions, two of the three agreed 82 times, and none of the three initial scorers agreed 19 times. All classifications for which all reviewers agreed met “face validity”. After adjudication by MG and JE, all five scorers agreed with the final classification. We did not attempt to divide questions into those asked in the academic and community institution. Our goal is to report the types of questions asked in overall practice rather than in a particular type of practice.
The results are summarised in table 11.. Briefly, treatment questions predominated, especially questions about drug dosing (35) and what drug might be appropriate to use in a situation (28). Questions about drug side effects, “what is drug X?” and drug safety questions were also common (18, 15 and 12 questions respectively). Diagnostic questions were dominated by questions about the manifestations of a disease (“What are the manifestations of disease X”) and by what laboratory/studies are appropriate in a particular situation (23 and 21, respectively). The significance of symptoms and signs (“What causes symptom X?”, “Is physical finding Y from disease X”) accounted for 7 and 8 questions, respectively. Relatively lacking were questions about drug cost (1), social or community services (1), ethics (2), pathophysiology (0), and questions of an administrative nature (such as “does plan X pay for drug Y”) (1). Physicians tried to answer 81% of the point‐of‐care questions and were successful 87% of the time.1
To our knowledge, this is the first study undertaken of questions in the ED. It confirms what many of us in emergency medicine believe. The questions that we have in the course of our practice are mostly about clinical matters that apply at the point of care. Notably absent are questions about community services, drug costs, and pathophysiology and legal questions (among others). The absence of questions about community services does not meet face validity because patient disposition is a major issue at both of these institutions, especially when discharging patients from the ED. It is likely that physicians did not consider disposition as a clinical question or consider this in the purview of others such as nursing or social work. A less likely, but more concerning, alternative explanation is that ED physicians do not consider what happens once patients leave the ED. The second notable lack is questions about drug costs, something that physicians are poorly informed about.3,4 Again, this has a couple of possible explanations. First, it is possible that ED physicians do not consider cost of care in their calculus. One recent study suggests that a majority of US physicians do not feel that it is their responsibility to help limit the amount of money that patients must pay for drugs (even though they believe costs should be controlled).5 The second possibility is that ED physicians are particularly cost conscious and already prescribe a predominant number of generic drugs. While we would like to delude ourselves into believing the second explanation, the correct interpretation is probably the first. Cost effective drug prescribing will require moving questions of cost onto the radar screen of practitioners. It would be interesting to see if physicians would consult a list of drug costs prominently posted at the point of care.
The lack of pathophysiology questions is surprising since one of the centres involved in this study was an academic institution. This likely reflects the immediacy of need in a busy ED. Time to ponder non‐clinical questions is a luxury that is often lacking.
This study also helps to point the way towards information resources that would be particularly useful in the ED. Not surprisingly, physicians' questions correlate well with the resources used to find answers.1 In our prior report, we found that drug information resources were the most commonly used resources to answer clinical questions.1 Rapid access to authoritative drug information would go a long way towards meeting the information needs of ED physicians.
There are several potential sources of error in our study. First, our study is vulnerable to the “Hawthorne effect”. Subjects who are being observed in a study tend to behave differently when they know they are being watched. It is possible that this study underestimated the number of questions that physicians have because the participants were unwilling to admit their lack of knowledge to a medical student. It is also likely that this study underestimated the total number of questions since it was impossible to capture every possible question during busy times in the ED. Finally, as noted above, it may be that physicians did not believe topics such as drug costs, patient services in the community and legal questions were in their sphere of practice.
This study was undertaken in only two moderate volume EDs in the US midwest and the results may not be generalisable to EDs with higher or lower volumes. We attempted to include a range of practices by including an academic and a non‐academic ED. The study followed 26 physicians in these two settings so that the overall number of questions reflects the practice of these 26 physicians. Even so, the external validity of this study is difficult to measure.
ED physicians have clinical questions that revolve primarily around patient diagnosis and treatment. Clinical questions, and the ability to answer them, are an important area of research with implications for patient safety. This study and others can help maximise the development of information resources and educational interventions in the ED.
Competing interests: None declared.