Article Text

SP0093 Using Literature to Solve Clinical Problems
  1. M. Boers1
  1. 1Epidemiology & Biostatistics; Rheumatology, Vu University Medical Center, Amsterdam, Netherlands


Rapid access to online databases available in most Western countries can really change the way we search for answers to our clinical problems. However, most of us have no time to search these resources during or after our clinics. And if we do search we usually stick to… Google. This presentation is an overview of tips and tricks to do better without losing too much time. However, to optimally apply these, one must be willing to spend some time developing basic skills.

Framing the question The first step is categorizing and framing the question. Most clinical questions refer to Diagnosis, Prognosis or Treatment. Each is best framed in PICO style (Patient/Population, Intervention/Exposure, Comparison/Control, Outcome). For example when a patient with erosive hand osteoarthritis (OA) asks you what type of glucosamine she should start taking, the PICO question would be: ‘Is oral glucosamine added to standard therapy better than standard therapy alone in improving … of patients with hand OA?’ the format reveals we have to be specific and ask the patient which features of her OA are most important to treat, eg. pain, stiffness or function.

Searching We should search for information with the highest level of evidence, but also aim for extreme efficiency. For example, a systematic review of randomized trials comprises the highest level of evidence, but an up-to-date clinical guideline summarizes the review in a few sentences. In other words, we must balance sensitivity (too high means finding everything in a pool of rubbish) and specificity (too high means finding only one article with too narrow a focus). Finally, we choose the source to search in.

The most rapid source of information is Google. You can type almost anything and find something relevant in the first few pages. improves on this by looking only at scientific sources. Also, knowing about operators, filters and word order in Google can make your results much more specific.

A good alternative is if your library has access. This database contains a wealth of brief but very complete texts on many conditions, and is very frequently updated.

A step down, but still very efficient is to search the site of your national guidelines committee. A quick (Google) search revealed several of them right away, but there is also a registry of guidelines ( You do need to check the date of the guideline, because your patient will always claim to have very recent evidence. Other possibilities include:

- a UK search machine of 70 EBM websites;

- : UK PubMed &EU guidelines

- : MedLine, DARE (Database of Abstracts of Reviews of Effects) included in Cochrane), US guidelines

The next step is to search, the most extensive database of systematic reviews for therapy. It is also increasingly the source for systematic reviews on studies of diangostic test accuracy and studies on prognostic factors.

If all of this doesn’t help (the waiting room is filling up) you will have to stop and reconsider. The next step is searching the medical databases yourself, i.e. PubMed, EMBASE (more EU and pharmacology), (more paramedic) or (more psychologic). Searching these databases is best done with the help of a librarian, but a quick course can greatly help if you want to do some searching yourself. The general tactic with these databases is to split your research question into smaller parts, then use a sensitive strategy to find all potentially relevant literature in each. By intersecting the subsearches (finding literature that is present in all of them) and then applying methodologic (e.g. ‘systematic review’) and other filters (eg. English language) you can whittle down the results to a more manageable level. Building the search requires some knowledge of free text and Medical Subject Heading (MeSH) keywords search methodology.

Critical appraisal And then, finally: you have to critically appraise what you found, for example by using the well-known questions from the Users’ Guides to the Medical Literature (

- Are the results valid?

- level of evidence

- methodological appraisal

- What are the results?

- level of uncertainty

- Are the results relevant to my patient?

- research population

- intervention/test

- outcome

Disclosure of Interest None Declared

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.