BSc and Beyond: Critical Appraisal and Literature Review Presented by: usamaali Email: firstname.lastname@example.org
Learning Objectives By the end of this tutorial, you will be able to: • Gain an appreciation for what sorts of things you should be looking for when critically appraising a paper. • Understand what a literature review is. • Understand how to write a literature review.
Warning It’s quite a dry topic
Plan • Literature reviews • Critical Appraisal
What is a literature review? • Searching and analysing the current literature on a topic. • Summarising all the available information in a certain number of words.
Main difficulties of a literature review Lots of information Not enough words
Writing a literature review (my suggestion) Gather information on each key topic Write the review without worrying about the word count List the key topics you need to research Make notes on the key findings in your search Cut words down Timeline
Tips on cutting down and being concise • Re-read every sentence. If you can remove words and retain the meaning of the sentence, remove the words. • Avoid sentences with >20 words. Examples: • “Is associated with a reduction in” “Reduced” • “It is widely known that…” “Evidence shows…” • “Resulted in a significant increase” “Significantly increased”
Plan • Literature reviews • Critical Appraisal
What is a critical appraisal? • Evaluate the STRENGTHS and the WEAKNESSES of a paper. • Must justify the proposed strengths and weaknesses.
The three C’s… • For ALL parts of a paper (introduction, methods, results, discussion), ask yourself 3 key questions: • Is it CLEAR? • Is it COMPREHENSIBLE? • Is it COMPREHENSIVE?
The Introduction Is there sufficient background information? Is the background information correct? Are there clear aims and a hypothesis? Is the study original? Is there a clear rationale?
Methods Reviews/ Meta-Analyses Animal studies Human studies Sample size? Sample size? All relevant literature addressed? Groups randomised/ blinded? Groups randomised/ blinded? Sound inclusion/exclusion criteria? Choice of animal – translational potential? Sound inclusion/exclusion criteria? Confounders? Confounders? Are methods showing a physiological or pharmacological effect? Are the outcome measures appropriate?
Physiological vs Pharmacological Effect Hypothesis: Molecule X, found intrinsically in mice, increases body temperature in the day-to-day temperature control of mice Temperature Mol. X Control Temperature Conclusion: Our results show that molecule X causes an increase in temperature compared to the control group. Therefore, molecule X is key in the day-to-day increase of body temperature in mice.
Physiological vs Pharmacological Effect Hypothesis: Molecule X, found intrinsically in mice, increases body temperature in the day-to-day temperature control of mice Mol. X antagonist Control Temperature Temperature Conclusion: Our results show that antagonizing molecule X causes a reduction in temperature compare to the control group. Therefore, our results suggest that molecule X is key in the day-to-day increase of body temperature in mice.
Results – things to consider Baseline characteristics Statistics Do they state which tests? Are P-values stated in all figures/legends? Are the tests appropriate?
Results – things to consider Figures Is it clear what error bars show? Is the type of figure appropriate? Are all figures necessary?
Discussion Pharmacological vs Physiological? Is the importance of data exaggerated (e.g. no statistical significance in any data?) Is there an attempt to make an unreasonable generalization? Does the conclusion follow from the results?
Discussion – other things to consider • Are the limitations clearly stated? • Does the paper actually tell us something new about a particular field?
Funding Other things to consider Ethics Authors conflicts of interest
It’s not just about what you see… It’s also about what you DON’T see Example: Subjects were split into two groups. In one group (mean age 47), subjects received the drug. In the other group, subjects received the placebo. No mention of randomisation Were the mean ages controlled?
Types of biases to be aware of Investigated Total Population Where the sample investigated do not represent the population Sampling Bias
Types of biases to be aware of Treatment group Placebo group You feel better after taking the drug, right? The drug isn’t making you feel better, is it? When outcome measurements are influenced by the investigator Observer Bias
Types of biases to be aware of Are you racist? No Response Bias Where subjects give answers that they think the interviewer want to hear
Types of biases to be aware of Thank you Please fill out my feedback form! tiny.cc/tutorialfeedback Procedural Bias Where subjects are put under pressure to complete responses quickly