John W. Ely, MD, MSPH

As physicians we do not have our act together on some things. Our medical records are often illegible, disorganized, or lost . Our medical errors are treated like sporadic embarrassments rather than events to be studied and prevented.¹ And we do a terrible job of answering our clinic al questions. Most of our questions are never answered, sometimes because the answers do not exist, but more often because they are not pursued.^2,3 As busy physicians we quickly learn that being aware of our knowledge gaps and trying to fill those gaps does not lead to efficient patient flow. Better to do the best we can with our current knowledge. But this is not really better, because it stifles lifelong learning and professional growth and may adversely affect patient care. The trick is to get fast and accurate answers from immediately available resources.

Two studies in this issue of JFP aim to help physicians answer their questions, but use different approaches. Alper and colleagues⁴ started with real questions and looked for the computer databases most likely to answer them. Rosser and coworkers⁵ started with answers in the form of guidelines and designed a rigorous method for picking the best ones. Both groups recognized the time pressures faced by practicing physicians. Alper and colleagues concluded that although most questions could be answered by using computer databases, the process often took too long to be practical in a busy office. Computers have tremendous potential to provide needed information, and this potential will be realized when enthusiastic programmers understand the importance of speed and usability. Rosser and coworkers recognized the value of writing short summaries of lengthy guidelines. These summaries distill multiple bits of evidence into statements that can be used by clinicians. Unfortunately, guidelines often answer questions that do not occur in practice and fail to address those that do. For example, physicians who ask how to screen for prostate cancer might consult the recommended prostate cancer guideline on the Guideline Advisory Committee Web site (www.gacguidelines.ca). This guideline states that the existing evidence is inadequate to recommend for or against prostate specific antigen (PSA) testing—which makes it a guideline that provides no guidance. Counseling patients about the pros and cons of PSA screening sounds good on paper; it is not so easy when you are behind schedule, and the nurse is pointing to your full waiting room. The clinicia n still must decide what to do. It is much faster to just say, “I think we should get a PSA today.” Everyone is happier at least in the shor t term, because the schedule does not fall apart, and the patient is not burdened with your uncertainties. The practical advantages of opinion-based medicine can be complling.

Physicians should ask good questions, and authors should write good answers. For example: “How should you work up a newborn with a 2-vessel umbilical cord whose physical examination is otherwise normal?” Is this a good question? It does not fit the PICO (Patient-Intervention-Comparison-Outcome) model.⁶ But I think physicians should ask what they want to know, rather than transforming their questions to fit the available evidence. The physician could have asked “What is the prevalence of serious malformations in babies with 2-vessel cords?” but that is not the information that is needed. The question is what to do with this baby. I could not find a clear answer to this question using the guidelines from Rosser and coworkers (or any other guidelines) or by using the databases of Alper and colleagues (after an hour of searching). I have my own answer but would not dare write it here, because it is based on opinion rather than evidence.

Physicians are overwhelmed with medical information, yet they cannot answer their patient care questions. We need not be discouraged by the impossible job of keeping up with new information, because most of it is irrelevant to our practices. Instead, we should focus on our questions because they will guide us toward more efficient lifelong learning. Finding good answers to our questions can be time consuming and difficult work that is often done at the end of the day while our suppers get cold. Someday we will get our act together by systematically recording real questions and organizing them in a large database; the Family Practice Inquiries Network (www.fpin.org) and others are moving toward that goal. Authors could search this database for questions related to their topics and include the answers in review articles, textbooks, and Web sites. Instead of blindly adding to an overwhelming body of available information, we need to document our information needs so that those needs can be met.

As physicians we do not have our act together on some things. Our medical records are often illegible, disorganized, or lost . Our medical errors are treated like sporadic embarrassments rather than events to be studied and prevented.¹ And we do a terrible job of answering our clinic al questions. Most of our questions are never answered, sometimes because the answers do not exist, but more often because they are not pursued.^2,3 As busy physicians we quickly learn that being aware of our knowledge gaps and trying to fill those gaps does not lead to efficient patient flow. Better to do the best we can with our current knowledge. But this is not really better, because it stifles lifelong learning and professional growth and may adversely affect patient care. The trick is to get fast and accurate answers from immediately available resources.

Two studies in this issue of JFP aim to help physicians answer their questions, but use different approaches. Alper and colleagues⁴ started with real questions and looked for the computer databases most likely to answer them. Rosser and coworkers⁵ started with answers in the form of guidelines and designed a rigorous method for picking the best ones. Both groups recognized the time pressures faced by practicing physicians. Alper and colleagues concluded that although most questions could be answered by using computer databases, the process often took too long to be practical in a busy office. Computers have tremendous potential to provide needed information, and this potential will be realized when enthusiastic programmers understand the importance of speed and usability. Rosser and coworkers recognized the value of writing short summaries of lengthy guidelines. These summaries distill multiple bits of evidence into statements that can be used by clinicians. Unfortunately, guidelines often answer questions that do not occur in practice and fail to address those that do. For example, physicians who ask how to screen for prostate cancer might consult the recommended prostate cancer guideline on the Guideline Advisory Committee Web site (www.gacguidelines.ca). This guideline states that the existing evidence is inadequate to recommend for or against prostate specific antigen (PSA) testing—which makes it a guideline that provides no guidance. Counseling patients about the pros and cons of PSA screening sounds good on paper; it is not so easy when you are behind schedule, and the nurse is pointing to your full waiting room. The clinicia n still must decide what to do. It is much faster to just say, “I think we should get a PSA today.” Everyone is happier at least in the shor t term, because the schedule does not fall apart, and the patient is not burdened with your uncertainties. The practical advantages of opinion-based medicine can be complling.

Physicians should ask good questions, and authors should write good answers. For example: “How should you work up a newborn with a 2-vessel umbilical cord whose physical examination is otherwise normal?” Is this a good question? It does not fit the PICO (Patient-Intervention-Comparison-Outcome) model.⁶ But I think physicians should ask what they want to know, rather than transforming their questions to fit the available evidence. The physician could have asked “What is the prevalence of serious malformations in babies with 2-vessel cords?” but that is not the information that is needed. The question is what to do with this baby. I could not find a clear answer to this question using the guidelines from Rosser and coworkers (or any other guidelines) or by using the databases of Alper and colleagues (after an hour of searching). I have my own answer but would not dare write it here, because it is based on opinion rather than evidence.

Physicians are overwhelmed with medical information, yet they cannot answer their patient care questions. We need not be discouraged by the impossible job of keeping up with new information, because most of it is irrelevant to our practices. Instead, we should focus on our questions because they will guide us toward more efficient lifelong learning. Finding good answers to our questions can be time consuming and difficult work that is often done at the end of the day while our suppers get cold. Someday we will get our act together by systematically recording real questions and organizing them in a large database; the Family Practice Inquiries Network (www.fpin.org) and others are moving toward that goal. Authors could search this database for questions related to their topics and include the answers in review articles, textbooks, and Web sites. Instead of blindly adding to an overwhelming body of available information, we need to document our information needs so that those needs can be met.

As physicians we do not have our act together on some things. Our medical records are often illegible, disorganized, or lost . Our medical errors are treated like sporadic embarrassments rather than events to be studied and prevented.¹ And we do a terrible job of answering our clinic al questions. Most of our questions are never answered, sometimes because the answers do not exist, but more often because they are not pursued.^2,3 As busy physicians we quickly learn that being aware of our knowledge gaps and trying to fill those gaps does not lead to efficient patient flow. Better to do the best we can with our current knowledge. But this is not really better, because it stifles lifelong learning and professional growth and may adversely affect patient care. The trick is to get fast and accurate answers from immediately available resources.

Two studies in this issue of JFP aim to help physicians answer their questions, but use different approaches. Alper and colleagues⁴ started with real questions and looked for the computer databases most likely to answer them. Rosser and coworkers⁵ started with answers in the form of guidelines and designed a rigorous method for picking the best ones. Both groups recognized the time pressures faced by practicing physicians. Alper and colleagues concluded that although most questions could be answered by using computer databases, the process often took too long to be practical in a busy office. Computers have tremendous potential to provide needed information, and this potential will be realized when enthusiastic programmers understand the importance of speed and usability. Rosser and coworkers recognized the value of writing short summaries of lengthy guidelines. These summaries distill multiple bits of evidence into statements that can be used by clinicians. Unfortunately, guidelines often answer questions that do not occur in practice and fail to address those that do. For example, physicians who ask how to screen for prostate cancer might consult the recommended prostate cancer guideline on the Guideline Advisory Committee Web site (www.gacguidelines.ca). This guideline states that the existing evidence is inadequate to recommend for or against prostate specific antigen (PSA) testing—which makes it a guideline that provides no guidance. Counseling patients about the pros and cons of PSA screening sounds good on paper; it is not so easy when you are behind schedule, and the nurse is pointing to your full waiting room. The clinicia n still must decide what to do. It is much faster to just say, “I think we should get a PSA today.” Everyone is happier at least in the shor t term, because the schedule does not fall apart, and the patient is not burdened with your uncertainties. The practical advantages of opinion-based medicine can be complling.

Physicians should ask good questions, and authors should write good answers. For example: “How should you work up a newborn with a 2-vessel umbilical cord whose physical examination is otherwise normal?” Is this a good question? It does not fit the PICO (Patient-Intervention-Comparison-Outcome) model.⁶ But I think physicians should ask what they want to know, rather than transforming their questions to fit the available evidence. The physician could have asked “What is the prevalence of serious malformations in babies with 2-vessel cords?” but that is not the information that is needed. The question is what to do with this baby. I could not find a clear answer to this question using the guidelines from Rosser and coworkers (or any other guidelines) or by using the databases of Alper and colleagues (after an hour of searching). I have my own answer but would not dare write it here, because it is based on opinion rather than evidence.

Physicians are overwhelmed with medical information, yet they cannot answer their patient care questions. We need not be discouraged by the impossible job of keeping up with new information, because most of it is irrelevant to our practices. Instead, we should focus on our questions because they will guide us toward more efficient lifelong learning. Finding good answers to our questions can be time consuming and difficult work that is often done at the end of the day while our suppers get cold. Someday we will get our act together by systematically recording real questions and organizing them in a large database; the Family Practice Inquiries Network (www.fpin.org) and others are moving toward that goal. Authors could search this database for questions related to their topics and include the answers in review articles, textbooks, and Web sites. Instead of blindly adding to an overwhelming body of available information, we need to document our information needs so that those needs can be met.