Abstract
Evidence-based medicine is the conscientious, explicit and judicious use of current best evidence to make clinical decisions. Given the pace of progress in reproductive medicine, this has become an essential part of effective care. Evidence-based practice typically involves a number of steps, including asking a question, searching for evidence, critical appraisal and application of the evidence.
13.1 Introduction
Evidence-based medicine (EBM) is the conscientious, explicit and judicious use of current best evidence to make clinical decisions [1], while integrating clinical expertise, experience and awareness of an individual patient’s preferences. Although this process of decision-making might sound laborious and contrived, much of it occurs intuitively in day-to-day clinical practice. EBM in everyday clinical practice can be challenging but, with the appropriate mindset, it is well within the capability of the average clinician entirely feasible and, given the pace of progress in reproductive medicine, increasingly, an essential part of effective care. It is, however, contingent on the ability to use digital technology to search the literature and use basic appraisal skills to judge the quality and relevance of available evidence. EBM demands clinical knowledge, both in terms of informing the parameters for a successful literature search as well as deciding whether and how the results match the patient’s clinical state, predicament and preferences. Evidence‐based guidelines can overcome the necessity for repetitive individual searches but are unable to overcome lack of basic subject-specific knowledge and poor decision-making processes or treatment skills.
Any clinical encounter generates questions about diagnostic and treatment strategies as well as concerns about side effects. The aim of a fertility consultation should be to discuss a set of options as a basis for joint decision-making, incorporating the preferences and values of both partners. For example, a couple presenting with unexplained infertility may wish to know whether they should undergo active treatment with superovulation and intrauterine insemination (SO-IUI) or in vitro fertilisation (IVF). This will require knowledge about their chances of success without treatment, the effectiveness of both treatments and expectations around the time required to conceive.
13.2 Using EBM to Inform Clinical Decision-Making
In practice, it is important to approach clinical questions through an EBM lens using a series of sequential steps (see Figure 13.1).
Figure 13.1 Sequential steps in approaching clinical questions through an EBM lens.
13.2.1 Framing a Structured Question
The first step in EBM is to convert a clinical question – for example, ‘What is the best way to manage unexplained infertility?’ – into a series of specific questions by defining key components such as the population of interest, the proposed intervention, the comparison – that is, no treatment or another intervention, the outcome of interest and the type of research design which is best suited to answering the question. For example, for the earlier question the ideal type of evidence is provided by randomised trials or, even better, by systematic reviews based on them. Table 13.1 shows examples of how this can be achieved in different areas of reproductive medicine.
Table 13.1 Asking a question
13.2.2 Searching the Literature
A comprehensive search of the literature can be a detailed and time-consuming process which can be made much simpler by following a few basic rules. A hierarchical approach aimed at identifying clinical practice guidelines and evidence-based reviews rather than primary research papers can be helpful, provided they are relevant and of high quality. National bodies such as the National Institute for Health and Care Excellence (NICE) in the United Kingdom as well as professional organisations such as the European Society for Human Reproduction and Embryology (ESHRE), the American Society for Reproductive Medicine (ASRM), the British Fertility Society (BFS) and the Royal College of Obstetricians and Gynaecologists (RCOG) generate guidelines which are periodically updated. Where high-quality evidence-based guidelines are unavailable, clinicians need to search for good quality systematic reviews, or in their absence, primary studies. Databases such as Medline and PubMed can be interrogated for systematic reviews or primary studies. A systematic search needs keywords matched to the medical subject heading terms and combined using Boolean operators such as ‘and’, ‘or’ and so forth.
13.2.2.1 Systematic Reviews
A systematic review uses prespecified methods to locate and appraise data and, where appropriate, undertake formal aggregation of quantitative data (meta-analysis). Well-conducted systematic reviews are preferable to individual studies as sources of high-quality evidence because single studies may be unrepresentative and lack the power to provide definitive answers to some questions. Inconsistent results across different publications on the same research question can be explored, and the ability to pool data allows greater precision around estimation of effects of interventions. Systematic reviews of randomised trials are considered to represent the highest quality of evidence on the effectiveness of medical and surgical treatments as well as more complex interventions such as IVF. Systematic reviews of observational studies are becoming increasingly more common, and may be the only source of evidence in situations where randomised trials do not exist or where randomisation is not feasible but results of meta-analyses need careful interpretation, as they are prone to bias.
13.2.2.2 Randomised Trials
A randomised trial can overcome many of the risks of bias associated with observational studies and provide convincing evidence to inform the choice of effective treatments. Random treatment allocation ensures that the two populations receiving different treatments are similar in terms of characteristics which could affect the results of treatment. Thus, any differences in outcomes can be assumed to be genuinely due to the intervention. Non-randomised studies with control arms may offer some information about effective treatments but are prone to confounding by patient characteristics which can influence non-random allocation of treatments. Not all randomised trials are of similar quality and some can be compromised by poor study design, execution or data analysis. A clear consort flow (Figure 13.2) is an essential component of a well reported trial which provides useful information about methodological rigour.
A traditional problem associated with infertility trials has been the multiplicity of outcomes and the way in which they are defined – something which will hopefully be addressed by the adoption of core outcome sets [2].
13.2.2.3 Observational Studies
Data from observational studies are useful in answering questions regarding the association between cause and effect, that is, exposure to smoking and its reproductive consequences. Thus, they are helpful in aiding clinicians to make decisions regarding diagnosis, causality and prognosis. In addition, as most randomised trials have relatively short periods of follow-up unlimited sample sizes larger observational studies population-based with long-term follow-up data can be useful in identifying side-effects of drugs over time. Where trials do not exist or are not feasible because of ethical considerations, observational studies may be the only source of data on the effectiveness of interventions.
13.2.2.4 Diagnostic Test Accuracy Studies
The quality and usefulness of investigations are usually assessed by means of diagnostic accuracy studies which measure sensitivity, specificity and predictive values (Figure 13.3). Patients can test positive or negative for any condition. A common way of evaluating the quality of tests is to determine the proportions of patients with normal and abnormal test results which were identified correctly. Sensitivity reflects the proportion of true positives while specificity represents the proportion of true negatives that are correctly identified by a test.
Figure 13.3 Template for calculation of test validity.
An estimate of the probability that the test will give us the correct diagnosis is better explored by means of positive and negative predictive values for a test. The former represents the proportion of patients with positive test results who are correctly diagnosed, while the latter reflects the proportion of patients with a negative test who are accurately identified. Predictive value is strongly influenced by the prevalence of the condition being tested for. If the prevalence of the disease is very low, the positive predictive value will be poor even if the sensitivity and specificity for a test are high. In population-based screening using serum anti-müllerian hormone, it is inevitable that many women with a positive test will be false positive. The ratio of the probability of getting a positive test result if the patient genuinely had a condition with the corresponding probability if she were healthy is the likelihood ratio – calculated as sensitivity/(1− specificity). The likelihood ratio indicates the value of the test for increasing certainty about a positive diagnosis. In general, the higher the likelihood ratio of an abnormal test the greater is its usefulness; for example, values of 10 or more suggest that the test could be extremely useful while a value of 1 suggests that it is useless. For a negative test a likelihood ratio of 0.1 or less suggests that it is useful while a value of 1 indicates that it is not.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
