A network for students interested in evidence-based health care

What are the key steps in Evidence-Based Medicine?

There are four key steps in EBM and the resources on the website are linked to these.

  1. Asking the right question
  2. Searching for the evidence
  3. Appraising the evidence
  4. Acting on the evidence

These steps are followed by a fifth: Evaluating your practice.

1. Asking the right question

To find the answer to a problem you need to start with a question. Getting the question right can be as important as getting the answer right. If you ask an inappropriate question you could end up with an answer that is not relevant to your patient (one that is not applicable to them). Or you could waste time reviewing too much information because the question is too broad and unfocused. The question should be as clear and focused as possible, in terms of the four elements.


Talking to a patient, have you asked the right question?

There are 4 elements of a good EBM question:

the patient or problem being addressed, Patient
the intervention or exposure being considered, Intervention
the comparison intervention or exposure when relevant, Comparison
the clinical outcomes of interest, Outcome

This is called the PICO from the initial letters of the four words: Patient, Intervention, Comparison and Outcome.

Here is an example of a clearly focused question:

Does aspirin reduce your risk of death after a heart attack?

Adults who have suffered a heart attack in the past month, Patient
Aspirin, Intervention
No treatment/placebo, Comparison
Death, Outcome

Here is an example of a question that is too broad and not focused enough:

Do antibiotics help children with colds?

Children with colds How old? – All children from 0 to 18 or 16 or 14. What do we mean by a ‘cold’? With a temperature or a runny nose? With a proven diagnosis of a bacterial infection?, Patient
Antibiotics – Which antibiotics? All of them?, Intervention
What comparator? A different type of antibiotic? A placebo? Nothing?, Comparison
What does “help” mean? Are we interested in symptoms/signs/quality of life/days off school? Or something else?, Outcome

Here is an example of a question that is too focused and where it is extremely unlikely there is going to be any information out there to find that addresses the question:

Does amoxycillin reduce facial pain in teenagers (13-18) with microbiologically-proven maxillary sinusitis?

Teenagers (13-18) with microbiologically-proven bacterial maxillary sinusitis… It is very unusual to find teenagers with maxillary sinusitis who have had a culture taken to prove that they have a bacterial cause. Even if there were such patients in some trials you’re not likely to come across this situation in day-to-day clinical practice. So finding studies of this sort of patient (if they exist) isn’t likely to be helpful. …and facial pain. Also if you are looking to see if treatment reduces facial pain you need to find studies that look at patients who had facial pain in the first place, Patient
Amoxycillin – It isn’t unreasonable to consider this antibiotic but it might be more helpful to look for other antibiotics as well, Intervention
What comparator? A different type of antibiotic? A placebo? Nothing?, Comparison
Reduction of facial pain – A study will only find a reduction in facial pain if the patients had this in the first place (the point made in the Patient box above). Is this the most important outcome? Should you be looking for a set of outcomes (usually one ‘primary’ outcome and a number of ‘secondary’ ones including the adverse effects of treatment)?, Outcome

2. Searching for evidence

Finding the evidence to answer your important, focused question is not straightforward. Usually you will be looking for a particular type of paper; one reporting a study conducted in a particular way. This will usually, but not always, be a randomized controlled trial (RCT) [2].


Searching for evidence

There are several large databases that include citations of published studies. One of the larger ones is PubMed [3]; it contains over 23 million articles. Some articles are simple case reports [4] and some are observational studies [5] where researchers have monitored a particular group of patients to find out more about a disease. Some are trials testing a drug or intervention against an alternative treatment or placebo [6]. Only some of these are controlled trials [7] and only some of those are randomized controlled trials [2]. Some articles bring together several studies to increase the precision and power of results and reduce bias [8] (systematic reviews [9] are one example of these).

Not only are there several databases to search, there are also several different ways to search. It can be extremely helpful to get advice from a librarian or information specialist (many librarians study information science as a discipline and see themselves as information specialists). As an example of the difficulties, similar conditions can sometimes be referred to in different ways: for example, premature babies, prematurity, preterm birth or infants or any combination of these. By searching for one you may miss papers using other terms.

3. Appraising the evidence

Do you believe everything you read in the newspaper? Do you believe everything you read in scientific journals? Almost everyone will say no to the first question. Some people will hesitate in answering the second. Unfortunately, you cannot rely 100% on papers written in even the most prestigious journals. Even if the contents of a paper are reliable, it is sometimes difficult to find the information you are looking for and to interpret it.

Not only that, different types of studies reported in papers in the literature, have different strengths and weaknesses. You need to understand these, and to know how to separate what is significant from what is not. In other words it is important to be able to critically appraise.

Further study of this topic will show that there are some general rules to help you. You will learn that there are three key questions to ask when looking at a paper:

  • Is the study valid (in other words, are the results of the study reliable because it was done in the best possible way),
  • what are the results of the study, and
  • will the results help me in looking after my patients.

In general larger studies are preferred to smaller ones as their results are less likely to be the result of chance. But this is just one aspect of a paper. It will soon become clear that the Methods section is key to understanding how the study was done.

At the end of the appraisal process you will have a better appreciation of how strong the evidence is concerning whatever the study was evaluating. You will know whether or not you should apply it to your patient.


Do you believe everything you read in the media?

4. Acting on the evidence

If you have examined all the relevant up-to-date evidence that is relevant to your focused clinical question, there should be no one in the world who is more up to date or better informed on this question than you! (Don’t get too comfortable, new evidence comes along all the time and the answer may change quickly). You should be able to explain to your patient the overall balance of evidence considering both the benefits and harms of treatment and assist the patient make a choice in what is called the shared decision-making [10] process.

Further reading (pdf)

Want to blog for us?

Register to become an S4BE Contributor