Critical appraisal is the assessment of the quality, role, importance and impact of research articles. This is predicated on an understanding of scientific thought, methodology and statistics.

When critiquing a manuscript or abstract, begin by assessing the research question. Has it been clearly defined? Is it a worthwhile question? A worthwhile question is one to which the answer will change our knowledge base or practice. It must be specific enough such that acquiring empirical data through an experiment will answer it.

The next step is to evaluate whether the study design is appropriate to answer the research question. Do the inclusion and exclusion criteria successfully delineate study groups of interest? Are the groups well matched (see section ‘Study Design’)? Is the study powered adequately? Is the design subject to bias?

In evaluating the results of a study, one should identify the significant and non-significant results and assess whether they directly answer the research question posed. Have the outcomes of the study been achieved? Statistically significant results should be evaluated in the context of the study’s methodology and should not be elevated to a pedestal; statistically non-significant results should not be ignored. For example, is a statistically significant result clinically or scientifically meaningless? Or is a statistically non-significant result important? For example, the recent finding that permissive anaemia does not affect outcomes in sepsis (NEJM DOI: 10.​1056/​NEJMoa1406617) is a negative, statistically non-significant result of great clinical significance.

The final step is to evaluate the conclusions of the study. Do the conclusions match the data? Do they err beyond the confines of the data? Do the limitations of the study negate any proposed conclusions? Overall, consider what the study adds to the existing literature.

A literature search is performed for a number of different reasons including to help formulate and focus a research question or when writing a literature review. With the wealth of information contained in books and journals, performing an effective literature search necessitates a systematic approach.

Relevant and specific search terms should be selected and noted. The use of MEDLINE (pubmed.com) is standard. The best way to search MEDLINE is to use the advanced function, rather than the generic homepage. In this way, one can select articles of interest more specifically (for example, by searching for only original research articles, or articles within a specific publication period). Further articles can be found using the “related articles” panel, which appears alongside abstracts. Once you are satisfied that you have found all the relevant articles, you should examine the reference lists of these articles to search for any missed papers. Recent review articles often refer to useful sources of information, which may have been missed in an initial literature search.

Not all pieces of evidence are of equal scientific worth. Evidence is sorted into scientific “strength” using a system of levels (defined, in the UK, by the Oxford Centre for Evidence-based Medicine). It is generally accepted that there are four levels of evidence (I-IV). Level IV evidence is of limited use: it may consist of the opinions of a panel of experts on a particular topic. Level III evidence is one rung higher in strength and includes non-interventional published material such as case reports, case series and retrospective comparisons. More robust evidence is from experimental studies, which should be well designed (IIb). The presence of a control group, even if non-randomized, enhances the strength of the research (IIa). Level I evidence consists of randomized control trials (RCT), which are often mistakenly described as providing the highest level of evidence. However, a single good RCT is rated as level Ib, which is below a meta-analysis of multiple RCTs (Ia). Inherent in this rating scale is an assessment of quality. A poorly designed or conducted RCT is worth less than an excellent non-randomised controlled trial. It is now routine to be asked about levels of evidence at academic interviews. These ratings are also commonly seen in clinical guidelines to show the level of evidence supporting a certain recommendation.

The principles of medical ethics may be summarised as respect for autonomy (allowing people to do as they please with their own bodies and make their own choices), beneficence (doing good), non-maleficence (not harming patients) and justice (fair treatment including in resource allocation).

Medical research has a tainted history with respect to medical ethics, from the illegal harvesting of dead bodies for the study of anatomy to the intentional withholding of available treatment for patients with syphilis. While a full discussion of medical ethics is beyond the scope of this chapter, it is pertinent to summarise the major ethical and legal components of medical research.

First, we consider the use of animals in medical research. Research on animals is fundamental to the science that informs human and veterinary medicine. Much of our understanding of anatomy, physiology, pharmacology and genetics comes from work in animals. As animals are vulnerable and experimentation may cause pain or suffering, we have a duty to carefully consider their use in research. In the UK, the Home Office governs the use of animals for experimentation. All projects using animals must be approved by the Home Office and granted a licence to go ahead. The Home Office also performs random inspections to ensure that animals are being treated humanely. The UK research community adheres to the principles of “the three Rs” – replacement of animals with non-animal research models, refinement of techniques to reduce suffering, and reduction of numbers needed. In the UK, the vast majority of animal research uses species such as mice, rats and flies. Research on primates is very strictly controlled and research on great apes is essentially banned.

The ethical foundations of human research are based on the Declaration of Helsinki, originally adopted in 1964. This document emerged from the Nuremberg Code, and dealt with the medical research atrocities committed by the Nazis.