GPs encounter three knowledge gaps in an average surgery, but reading every relevant study requires more than 600 hours a month.
So how are GPs to fill these gaps with reliable, up-to-date information? Evidence-based practice (EBP) has some solutions.
This article deals with the first two steps of EBP (see figure 1) and an important source of information, systematic reviews.
Experienced GPs tend to ask questions that push at the boundaries of certainty, such as, 'What is the risk of prescribing bisphosphonates beyond five years?' rather than basic questions such as, 'Which drugs are used for the prevention of osteoporotic fractures?'
To answer the first type of question, we start by defining the question. The PECOT structure is a good way to do this (see table 1).
|Table 1: The PECOT approach|
|Population (P)||In people who have received prophylaxis with bisphosphonates ...|
|Exposure (E)||... who continue to take bisphosphonates beyond five
|Comparison (C)||... compared to those who stop after five years ...|
|Outcome (O)||... what is the risk of an atypical fracture ...|
|Time (T)||... over a five- to 10-year period?|
Our next move is to search for evidence. Three sources spring to mind, but they have drawbacks:
- Ask a colleague. But at the boundaries of certainty, a colleague may not know better than you.
- Search the web. But Google returns 230,000 hits.
- Go to a textbook. But this particular information has not entered most textbooks and is vague in those it has.
So a different approach is called for. We need to:
- Know where to find the evidence.
- Have a strategy to obtain the evidence as quickly and effortlessly as possible.
- Understand if the evidence found is the type of study that is appropriate to our question.
Certain types of primary studies are better suited to some questions. Table 2 is a rough guide.
|Table 2: Study types and questions|
In our bisphosphonate example, we would expect to find cohort studies and there are indeed two studies of patients who, at the end of trials, either continued with the therapy (E) or stopped it (C).
Not too many to read, but what if we found 6, 12, 24 or more papers? We would turn to types of evidence that synthesise the findings of primary studies - systematic reviews.
The methods employed for systematic reviews differ from those of narrative reviews. This makes systematic reviews more reliable.
Analysis of the quality of studies weighs up how good a piece of evidence is, so strength of conclusions can be judged. Another feature is the ability to make an estimate of an effect more precise.
Research aims to measure the size of an effect; for example, by how much does a treatment decrease time to recovery or by how much does an exposure increase risk of a disease?
Research measures the effect size in a sample that provides an estimate of the effect size in the population. Around this estimate, a confidence interval can be calculated, which contains, by convention, 95% of plausible estimates.
By combining the results of several studies, one type of systematic review, a meta-analysis, narrows the confidence interval and so gives a more precise estimate of the effect in the population.
However, there are problems. A big problem is publication bias, the tendency for studies that find effects to be published in preference to those that do not.
Another problem is that not all systematic reviews are conducted satisfactorily. However, systematic reviews are invaluable as long as we differentiate the good from the bad, which we do by scrutinising how well they adhere to the methods listed in table 3.
|Table 3: Study methodology|
|Question||Often broad||Often focused|
|Search for studies||Often not specified||Explicit and comprehensive|
|Selection of studies||Often not specified||Criteria specified|
|Data extraction Uncontrolled Pre-specified||Uncontrolled||Pre-specified|
|Appraisal of quality of studies||Variable||Rigorous|
How to search for evidence
Our strategy (see figure 2) for finding suitable evidence should:
- Be easy and quick to use.
- Offer systematic reviews when available; otherwise, offer primary studies.
- Display results of searches in useful categories.
If your question type is therapy, it might be quickest to search the Cochrane library. If that fails or your question is a different type, there are several meta-search engines that search numerous databases.
Some, such as Pubmed Clinical Queries, provide helpful filters for aetiology, diagnosis, prognosis and therapy questions.
Most engines, including the TRIP database, return hits in categories that include guidelines, systematic reviews, primary studies, textbooks and synopses.
Synopses are another helpful category for busy practitioners. They summarise reviews and add value in the form of a commentary. This usually includes a judgment of their reliability and assessment of the importance of the findings.
For the bisphosphonate example, this strategy found several narrative reviews that provided opinions worth considering. However, the two cohort studies allow readers to draw their own conclusions.
Which search engine you use is a matter of personal taste, but it is worth becoming acquainted with several, because each has its own advantages.
It is possible for a GP to do such searches, although the usual worry is that we cannot conduct in-depth searches for every question.
For many questions, a quick look at an online textbook may be sufficient. However, there are times when a search is essential - questions in areas of uncertainty, topics that are important and often arise, and when practice is guided more by custom than knowledge.
- Dr Hopayian is a GP in Leiston, Suffolk, and author of 'Making Your Practice Evidence-Based: A Self-Study Guide for Primary Care', RCGP Publications, 2010
|CPD IMPACT: EARN MORE CREDITS|
These further action points may allow you to earn more credits by increasing the time spent and the impact achieved.