Bridging the gaps in evidence based diagnosisBMJ 2006; 333 doi: https://doi.org/10.1136/bmj.38945.464722.80 (Published 24 August 2006) Cite this as: BMJ 2006;333:405
- Sharon E Straus (), associate professor
The process of diagnosis is complex and poses many challenges to doctors and other clinicians attempting to practise in an evidence based manner. When making a diagnosis in patients who are already ill we should be able to draw on evidence about the accuracy of diagnostic tests. When trying to make an early diagnosis of presymptomatic disease in well people through population screening we rely on evidence from randomised trials on whether patients benefit from such screening. Looking for presymptomatic disease among patients with an unrelated disorder (case finding) or trying to generate a differential diagnosis for patients' signs and symptoms require different types of evidence. We need to match each question to the type of evidence by using a diagnostic strategy.
Furthermore, diagnosis seldom relies on a single test. Ideally a clinician would like to find valid evidence about a cluster of tests, including the clinical examination, along with measures of the accuracy of combinations of these tests. It may also be useful to know how a new test compares with older tests, although this will depend partly on the role of the new test—whether it is a replacement, triage, or add-on test.1
Reliable evidence about the precision and accuracy of a diagnostic test is hard to find, however, given the recent increase in new laboratory tests and the lack of evidence to support many of them.2–4 Despite improving standards in research on diagnostic tests in the past decade, the methodological quality of many diagnostic studies remains low.w1-w5 Even apparently evidence based sources are flawed—in this week's BMJ Mallett and colleagues report that the reliability and relevance of current systematic reviews of diagnostic tests for cancer are compromised by poor reporting and review methods.5 This matters because biased estimates of diagnostic accuracy and poorly designed and reported studies can mislead clinical decision making.3 w1
At the top of the pyramid are systems using explicit review processes to present high quality, pre-appraised evidence. For questions of diagnosis this can be difficult though, because some systems such as Clinical Evidence (http://www.clinicalevidence.com/) do not include material on diagnosis, although this is set to change soon. The second tier of the pyramid or search cascade comprises high quality journals of secondary publications—such as ACP Journal Club (http://www.acpjc.org/) and Evidence Based Medicine (http://ebm.bmjjournals.com/)—and bmjupdates (http://bmjupdates.mcmaster.ca/), a service that pre-rates primary articles for quality, clinical relevance, and interest. The next tier includes databases of systematic reviews such as the Cochrane Library, but this will not be of much help until the Cochrane Collaboration launches its new process for systematically reviewing evidence on diagnosis. The final tier consists of original studies, which can be identified by using the PubMed clinical queries (www.ncbi.nlm.nih.gov/entrez/query/static/clinical.shtml) diagnosis search filter to search Medline. This search filter has been validated for accuracy,7 but not everything relevant to the topic will be found using this approach.8 If all else fails a general Medline search might prove useful. However, finding the evidence will provide answers only if the evidence is valid, important, and applicable to the patient in question.9 10
An additional search of the Rational Clinical Examination website (http://www.sgim.org/clinexam.cfm#RCE) may answer questions about the accuracy of the clinical examination. This website presents a series of systematic reviews on the precision and accuracy of the clinical examination and provides a search filter for finding relevant articles, although this filter has not yet been validated for accuracy.
All this work to answer diagnostic questions is beyond the scope of most doctors and other practising clinicians. They need evidence that is available quickly in a concise and intelligible form because they cannot spare more than a few seconds per patient to find and assimilate relevant evidence. Indeed, if answers to clinical questions are not found within 90 seconds, searches for evidence are often abandoned.11 Even when evidence is synthesised and packaged, barriers to making evidence based decisions still exist.12 The creation of meta-search engines that integrate the main evidence resources (through collaboration among the various publishers) could allow effective and efficient searches and facilitate rapid completion of the search cascade.
Evidence based diagnosis needs more primary evidence on diagnosis, more systematic reviews, and appropriate tools to translate the evidence into action. The challenge to clinicians, educators, researchers, funders, journal editors, and publishers is to work together to make this happen. Doctors and other clinicians should demand action now.