Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
By the end of this chapter the reader should:
Understand basic principles of evidence-based medicine in order to implement it into clinical practice
Be able to formulate a clinical question
Be able to undertake a hierarchical search strategy using online search databases
Be able to critically appraise a randomized controlled trial (RCT)
Be able to interpret the commonly used measures of treatment efficacy
Be able to give an evidence-based presentation
Evidence-based medicine (EBM) has now established itself as a key principle at the heart of modern clinical life. But what is it? In this chapter we will look at the principles of evidence-based medicine and how to practise it.
When we make a clinical decision (e.g. should I give this wheezy child a nebulized or spaced β 2 agonist?), we need to think about the patient and the overall outcome. There could be beneficial outcomes, but these should be weighed against the possibility of negative effects. As clinicians, we instinctively assess the chances of these outcomes, weigh them, and conclude on a course of action. If we are treating a child with an acute exacerbation of asthma, we may want to know what is the best mode of delivery for a β 2 agonist? But what does best mean? Patient satisfaction? Ease of delivery? Fewer symptoms? Fewest side effects? Fewer admissions? Most cost-effective? Least expensive?
For the clinician, the process of practising EBM can be difficult, time consuming, and (dare we say it) boring. In this chapter, these barriers will be tackled with examples of EBM in practice, revealing that the five minutes spent thinking this through may have saved you hours of work whilst improving the care provided to your patients.
EBM was defined by one of the founders of the EBM movement, David Sackett, as ‘the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients’. But what does that mean in real terms? EBM is a shorthand term for five linked ideas:
Our practice should be a meeting of ( Fig. 39.1 ):
Our clinical skills (including history, examination and diagnosis building)
Our patient's own values, preferences and beliefs
The best available evidence
Seeking information/evidence is key to learning and should mostly be ‘just in time’ (see below), determined by an individual patient or population's problem.
The care we offer should integrate quality clinical research with clinical experience, rather than relying on habit, dogma or tradition.
Searching and appraising evidence is only meaningful if it is applied to decisions and actions that benefit patients.
Clinicians, including those in training, should continuously evaluate their performance.
Some misconceptions about EBM are listed in Box 39.1 .
There are a series of misconceptions regarding EBM which should to be dismissed as untrue:
EBM belittles or removes clinical expertise
EBM ignores patients' values, preferences and beliefs
EBM promotes ‘cookbook’ medicine
EBM is simply a cost-cutting tool
EBM is only for tertiary hospital clinicians/specialists
EBM is exclusively limited to undertaking research rather than using research findings
EBM leads to the renunciation of therapies when there is an absence of evidence from RCTs
EBM is performed by statisticians sitting in offices and not by clinicians working in wards, clinics or operating theatres.
Practicing EBM improves patient care. There really should be no other incentive. Considering this in terms of the four pillars of medical ethics can be helpful:
Autonomy : All patients have the right to make decisions about their own care. This is central to EBM. Their values, preferences and beliefs should always be taken into account when making decisions.
Beneficence : When starting therapies you need the best available evidence to be certain that they have a meaningful benefit to the patient.
Non-maleficence : If a therapy or test has no meaningful benefit but is harming the patient (e.g. through side effects) then it should be stopped or not used in the first place.
Justice : Scarce resources must be fairly distributed. When rationing provisions, there must be good evidence that they are effective. If they are not, this resource could have been used elsewhere.
There are also selfish reasons for wanting to practise EBM:
Personal development/examinations : EBM is included in the RCPCH Curriculum for Paediatric Training, which states that: ‘In addition to a detailed knowledge and understanding of diseases in children and young people, paediatricians must ensure they are up-to-date, conform with highest standards of practice, aim to promote evidence-based medicine where possible and audit practice (assessment standards 18–20).’
Self-preservation : Clinicians are presented with huge volumes of research evidence. Without the time to read all this information, skills are required to filter out the good from the bad.
Reduced workload: Practising EBM will hopefully help patients to be diagnosed more effectively and get better quicker, therefore reducing workload.
There are 11 new systematic reviews and 75 new trials published every day ( ). It is impossible to keep up-to-date with all medical advancements, so long-term professional strategies are required to meet our learning and information needs.
Reading this book is an example of just in case information: packing in a range of general nuggets to provide a good underlying understanding of paediatrics in order to be good clinicians. When taken to the extreme, it is the diligent, regular reading of a series of journals ‘just in case’ a particular case was to present. This is inefficient and induces guilt when you cannot manage it.
This is a lifelong skill – the acquisition of information ‘just’ as you need it, e.g. reading a patient's clinic notes before they arrive. In the EBM context, it is identifying information that will help us best manage our patients by seeking rapid answers to specific queries. Research has shown that a doctor in training will have two unanswered questions for every three patients they consult (Green et al 2000). An inquisitive clinician could have many more questions. As our clinical experience improves, this number is unlikely to change, but the content may become more focused.
The practice of EBM is a multi-step process. Each of these steps ( Table 39.1 ) requires individual skills and practice, though some resources will allow us to shortcut some of these steps. Throughout this chapter, there are examples of the ‘five steps’ in practice (see also Tables 39.3, 39.4, 39.6 and 39.9 ).
Step | Skills required | Barriers | Consequences of inadequate implementation |
---|---|---|---|
Step 1: Asking a question |
|
|
|
Step 2: Acquiring information/evidence |
|
|
|
Step 3: Appraising the information/evidence |
|
|
|
Step 4: Applying the evidence to your patient |
|
|
|
Step 5: Assessing your performance |
|
|
|
It is all too easy to practise medicine without asking questions, as asking questions exposes potentially embarrassing gaps in our knowledge. The first step of EBM is to address this challenge and admit ignorance or uncertainty, then convert our information needs into answerable questions. This means having an inquisitive mind. Looking at the anatomy of enquiry, ask initially ‘What sort of question am I asking?’ If it is a clinical question then it can be grossly categorized as ‘foreground’ or ‘background’. Background questions are broad, and are often ‘what is’ or ‘what causes’ type questions, e.g. ‘What causes asthma in childhood?’ Foreground questions are specific and pointed, and can be fitted into a ‘PICO’ framework (patient-problem, intervention, comparison, outcome). This art is known as ‘framing a clear question’ and is an essential skill. A well-framed question must be directly related to the patient and structured in order to search for a relevant and precise answer.
A popular method for framing a clinical question is the PICO method ( Box 39.2 ):
P – Patient, Pathology, Problem or Population : What are the key features that describe the patient or population? Be specific.
I – Intervention or Interest : Be specific about the intervention, test or risk factor you are considering.
C – Control or Comparison : What would be appropriate alternatives? This may be placebo or more often currently used treatments.
O – Outcomes : Consider patient-oriented short-term and long-term outcomes; remember negative effects too. Avoid surrogate markers (e.g. improved CRP).
Search at the point of care.
If you are too busy to search immediately, write your questions down (e.g. a notebook, smartphone or portfolio).
Present your questions to your colleagues for feedback.
Write educational prescriptions on ward rounds ( http://www.cebm.net/wp-content/uploads/2014/04/educational_prescription_1.pdf )
Choosing the right outcomes to search is incredibly important. If you have a strong opinion on a topic, it is likely to be drawn from experience, e.g. always giving inhaled salbutamol to wheezy children in A&E. But what do we actually want to know about our intervention? Will it stop the patient dying? Will it keep them out of hospital for longer? Will they feel better? These are all important questions.
The aim of this step is the acquisition of good quality information/evidence to answer your skilfully constructed question. This can be difficult, but with practice and a few tips on where to look, it gets easier. The process therefore follows three steps:
Converting your PICO question into searchable terms
Searching for secondary sources (e.g. guidelines, Cochrane/DARE, etc.)
Finding primary sources if secondary sources are not available (e.g. PubMed). (See http://www.youtube.com – search for ‘PubMed Advanced Search Builder’ in YouTube for a 3-minute tutorial on how to search in PubMed.).
Getting your search right is both an art and a science. A good search is both sensitive and specific. A sensitive search will not miss any relevant papers, a specific search will not have too many irrelevant articles.
Once you have written a sound PICO question you must convert the question into searchable terms.
To form a search strategy:
Convert each arm of your PICO into search terms (including alternate spellings, synonyms, and truncations).
Search for the correct MeSH term (if you are using a database that does not automatically map). ‘MeSH’ is essentially the National Institute of Health (NIH) thesaurus of medical terms that guides towards the ‘correct medical term’. This means a searching clinician is able to find the relevant research, including when the papers' authors have not used the ‘preferred’ medical term). (See http://www.nlm.nih.gov/mesh/ ; Tutorial: http://www.nlm.nih.gov/bsd/viewlet/mesh/searching/mesh1.html ).
Combine the search terms using the correct Boolean operators ( Table 39.2 ).
Term | Search description |
---|---|
OR | (Infant OR child) will find all articles/documents containing at least one of these keywords |
NOT | (Infant NOT child) will find articles/documents containing the keyword infant but exclude those also including the keyword child |
AND | (Infant AND child) will find articles/documents containing both of these keywords |
* (Truncation) | Infant* will find articles/documents containing the keywords infant OR infants OR infantile OR infancy, etc. |
There is probably no correct answer to the question, ‘where is best to search?’ You are likely to find databases which you are more comfortable using. Trip®, the Cochrane library® or PubMed® are good databases to be familiar with ( Box 39.3 ).
Practising EBM need not be time-consuming. Add the following websites to your bookmarks in order to speed up your hierarchical search strategy:
Trip database ( www.tripdatabase.com )
The Cochrane library MeSH search page ( http://onlinelibrary.wiley.com/cochranelibrary/search/mesh/quick )
The Cochrane library advanced search page ( http://onlinelibrary.wiley.com/cochranelibrary/search/ )
The DARE database ( http://www.crd.york.ac.uk/crdweb/ )
PubMed Clinical Queries ( http://www.ncbi.nlm.nih.gov/pubmed/clinical )
PubMed for primary searches ( http://www.ncbi.nlm.nih.gov/pubmed )
A hierarchical search aims to look for the best quality evidence first, and then work downwards if insufficient research is discovered ( Fig. 39.2 ). If you are searching in a clinical environment, a database such as Trip® will often find results quickly ( Table 39.3 ) and present them in evidence type. If this fails, the Cochrane library and DARE website should be searched ( Table 39.4 ). If this does not yield any results, then PubMed's ‘Clinical Queries’ is the next best port of call followed by a search for primary resources in PubMed.
Treatment – acute exacerbation of asthma |
---|
Scenario: A 6-year-old child is admitted with a moderate exacerbation of asthma. He has been given a nebulizer of salbutamol in the A&E department and needs admission for a trial of inhaler and holding chamber (spacer). You wonder if he could have been given the inhaler in the A&E and the child sent home sooner? Step 1: Asking a question (PICO): In a child with a moderate exacerbation of asthma [patient], is inhaled, spaced, salbutamol [intervention] as effective as nebulized [comparison] β 2 agonist in terms of time to resolution of symptoms, likelihood of admission and deterioration [outcomes]? Step 2: Acquiring evidence: Search terms: Asthma AND spacer AND nebulizer. Databases: Trip database – 110 results. Best result – Cochrane review. Step 3: Appraising the evidence: |
Study group | Intervention | Study type | Outcomes | Key results | Comments |
---|---|---|---|---|---|
2295 children in 27 trials from emergency and community setting. | Any β 2 agonist given by any nebulizer versus the same β 2 agonist given by metered-dose inhaler with any spacer. | Cochrane systematic review. Only RCTs considered for review. |
Primary outcomes : Admission to hospital or duration of stay. Secondary outcomes : Duration in emergency department, change in respiratory rate, blood gases, pulse rate, tremor, symptom score, lung function, use of steroids, relapse rates. |
Meta-analysis of probably heterogeneous results. Spacer versus nebulizer relative risk of admission was 0.72 (95% CI: 0.47 to 1.09). In children, length of stay in the emergency department was shorter with spacer, mean difference of −0.53 hours (95% CI: −0.62 to −0.44 hours). | Clear primary and secondary outcome measures. Particular emphasis on the allocation concealment, which in general appears poor in most papers. |
Commentary : Acute exacerbation of asthma is common in both hospital and primary care. The airways are narrowed due to mucosal oedema, hypersecretion and bronchospasm. β 2 agonists have been used successfully to relieve the bronchospasm. This paper included RCTs including adults and children. It can be argued that adults and children differ in their ability to use the devices being tested. Therefore, the results for adults and children were separated for each outcome. In this systematic review it was found that the method of delivery of β 2 agonist did not appear to affect hospital admission rates but did significantly reduce the duration of stay in the emergency department. Step 4: Applying the evidence (the clinical bottom line):
|
Diagnosis – cyanotic heart disease |
---|
Scenario: You start at a new hospital where you undertake ‘postnatal checks’ on newborn infants. You notice that in this hospital you do not need to perform post-ductal pulse oximetry testing. You discuss this with your consultant, who asks you to find out more exact details on the benefits of this. Step 1: Asking a question (PICO): In an asymptomatic newborn infant [patient], does post-ductal pulse oximetry [intervention] increase the number of infants correctly identified with congenital heart disease or reduce mortality rates [outcomes]? Step 2: Acquiring evidence: Search terms: (Infant, newborn OR infant* OR newborn OR ‘newborn infant’ OR neonat*) AND (heart defects, congenital OR congenital heart defect* OR Defect*, congenital heart OR heart, malformation of OR heart abnormalit* OR congenital heart disease OR cyanotic heart disease OR cyanotic heart defect OR congenital heart malformation) AND (oximetry OR oximetry, pulse OR blood gas monitoring, transcutaneous OR oximetry, transcutaneous OR oximetry, transcutaneous OR saturation*, oxygen OR oxygen saturation*) Databases: Cochrane: 164 results, 4 non-Cochrane reviews including below meta-analysis. Step 3: Appraising the evidence: |
Study group | Intervention | Study type | Outcomes | Key results | Comments |
---|---|---|---|---|---|
13 eligible studies with data for 229,421 asymptomatic newborn babies. 118 infants with critical congenital heart defects. 748 false positives. 33 false negatives. |
Pulse oximetry. 60% of studies used foot alone (postductal). |
Systematic review and meta-analysis of 12 cohort and 1 case-control study. | Detection of critical congenital heart defects. | Sensitivity 76.5% (95% CI 67.7–83.5), specificity was 99.9% (99.7–99.9), false positive rate 0.14 (0.06–0.33). Likelihood ratio positive 549 (238–1195), likelihood ratio negative 0.24 (0.17–0.33). Lower false positive rate if oximetry >24 hours (p=0.0017), but no effect on sensitivity. |
Clear description of search strategy. No statistical description of heterogeneity. Significant publication bias was reported. |
Commentary: Screening for critical congenital heart defects in newborn babies can aid early recognition, with the prospect of improved outcome. In this case, the new doctor was not interested in an individual patient but a population. As the search had the potential to lead to widespread change in the clinical assessment of all newborns, the search for evidence needed to identify the most relevant and highest quality available. The search was therefore very comprehensive. Though this systematic review found that pulse oximetry is highly specific for critical congenital heart disease, it does not look at broader outcomes. For example, do infants who have their diagnosis made earlier using screening have better long-term outcomes (e.g. mortality)? This would be an important consideration when balancing the cost (equipment, time, etc.) of implementing such a screening tool. Step 4: Applying the evidence (the clinical bottom line):
|
If performing a more detailed search (e.g. for guideline development) then an extensive search should be undertaken, preferably with the help of an information specialist at your local library.
Almost all database search engines use Boolean operators (see Table 39.2 ) to combine search concept.
When searching PubMed, a search filter can be used to find evidence relating to different areas of practice (e.g. therapy or diagnosis); these are handily available via the PubMed ‘Clinical Queries’ page. This can quickly make the search more sensitive and specific.
A common misconception regarding EBM is that only RCTs or systematic reviews constitute the ‘evidence’ in EBM. Though double-blinded RCTs are often considered the ‘gold standard’ for establishing treatment effects, they will not be the best at answering questions about diagnosis, prognosis and/or harm. Secondly, there may not be RCTs available to answer certain treatment questions. There are barriers to research in child health and this often means that good quality information is unavailable (see Box 37.1 ). If no literature is available to answer your question, then an email/discussion with an expert may be the ‘current best evidence’ for your clinical question. Do not negate the importance of qualitative research, which significantly adds to the wealth of evidence available on particular subjects.
An understanding of ‘levels of evidence’ will enable you to search for the ‘current best evidence’. Evidence hierarchies are often used to describe the relative authority of various types of medical research. There is no individual, universally accepted hierarchy, and a tiny, very poorly conducted RCT will be less use than a well-constructed cohort study.
A simple hierarchy has been described by Professor Greenhalgh in How to read a paper :
Systematic reviews and meta-analyses
Randomized controlled trials with definitive results
Randomized controlled trials with non-definitive results
Cohort studies
Case-control studies
Cross-sectional surveys
Case reports
Become a Clinical Tree membership for Full access and enjoy Unlimited articles
If you are a member. Log in here