How to Read a Paper

Trisha Greenhalgh is a general practitioner and Professor of Primary Care Health Sciences and Fellow of Green Templeton College at the University of Oxford. Paul Dijkstra is a sport and exercise medicine physician and Director of Medical Education at Aspetar Orthopaedic and Sports Medicine Hospital in Doha, Qatar. He has an academic affiliation with the Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences at the University of Oxford.

• Foreword to the first edition by Professor Sir David Weatherall xiiPreface to the seventh edition xivPreface to the first edition xviiAcknowledgements xixChapter 1 Why read papers at all? 1Does ‘evidence- based medicine’ simply mean ‘reading papers in medical journals’? 1Why do people sometimes groan when you mention evidence- based healthcare? 4Before you start: formulate the problem 11Exercises based on this chapter 13References 14Chapter 2 Searching the literature 15The information jungle 15What are you looking for? 16Levels upon levels of evidence 17Synthesised sources: systems, summaries and syntheses 18Pre-appraised sources: synopses of systematic reviews and primary studies 21Specialised resources 22Primary studies: tackling the jungle 23One-stop shopping: federated search engines 25Using artificial intelligence to search the literature 25Asking for help and asking around 26Online tutorials for effective searching 26Exercises based on this chapter 27References 28Chapter 3 Getting your bearings: what is this paper about? 30The science of ‘trashing’ papers 30Three preliminary questions to get your bearings 32What are randomised controlled trials and why do they matter? 34What are cohort studies? 38What are case–control studies? 40What are cross-sectional surveys? 40What are case reports? 41The traditional hierarchy of evidence 42Exercises based on this chapter 43References 43Chapter 4 Assessing methodological quality 45Was the study original? 45Who is the study about? 46Was the design of the study sensible? 47Was bias avoided or minimised? 49Was assessment ‘blind’? 54Were preliminary statistical questions addressed? 55A note on ethical considerations 58Summing up 59Exercises based on this chapter 60References 60Chapter 5 Statistics for the non-statistician 63How can non-statisticians evaluate statistical tests? 63Have the authors set the scene correctly? 65Paired data, tails and outliers 71Correlation, regression and causation 72Probability and confidence 74The bottom line (quantifying the chance of benefit and harm) 77Summary 79Exercises based on this chapter 79References 80Chapter 6 Papers that report clinical trials of simple interventions 82What is a clinical trial? 82Drug trials: ‘evidence’ and marketing 83Making decisions about therapy 86Surrogate endpoints 87What information to expect in a paper describing a randomised controlled trial: the CONSORT statement 91Getting worthwhile evidence from pharmaceutical representatives 91A note on vaccine trials 94Exercises based on this chapter 95References 95Chapter 7 Papers that report trials of complex interventions 99Complex interventions 99Ten questions to ask about a paper describing a complex intervention 101Exercises based on this chapter 106References 107Chapter 8 Papers that report diagnostic or screening tests 109Ten suspects in the dock 109Validating diagnostic tests against a gold standard 110Ten questions to ask about a paper that claims to validate a diagnostic or screening test 115Likelihood ratios 119Clinical prediction models 122Exercises based on this chapter 124References 125Chapter 9 Papers that summarise other papers (systematic reviews and meta-analyses) 128When is a review systematic? 128Evaluating systematic reviews: five questions to ask 131Meta-analysis for the non-statistician 137Explaining heterogeneity 142New approaches to systematic review 145Exercises based on this chapter 146References 146Chapter 10 Papers that advise you what to do (guidelines) 151The great guidelines debate 151Ten questions to ask about a clinical guideline 155Exercises based on this chapter 162References 162Chapter 11 Papers that estimate what things cost (health economic evaluations) 164What is an economic evaluation? 164Health economics studies: two key approaches 166Costs and benefits of health interventions 167Measuring the value of health states 168Quality-adjusted life-years 169Low-value health: choosing wisely 171Twelve questions to ask about a health economic evaluation 172Conclusion 176Exercises based on this chapter 176References 177Chapter 12 Papers that go beyond numbers (qualitative research) 179What is qualitative research? 179Summarising and synthesising qualitative research 183Nine questions to ask about a qualitative research paper 184Conclusion 191Exercises based on this chapter 192References 192Chapter 13 Papers that report questionnaire research 195The rise and rise of questionnaire research 195Ten questions to ask about a paper describing a questionnaire study 196Exercises based on this chapter 205References 206Chapter 14 Papers that report quality improvement case studies 208What are quality improvement studies and how should we research them? 208Ten questions to ask about a paper describing a quality improvement initiative 210Conclusion 217Exercises based on this chapter 217References 218Chapter 15 Papers that describe genetic association studies 220The three eras of human genetic studies (so far) 220What is a genome-wide association study? 222Clinical applications of genome-wide association studies 225Direct- to- consumer genetic testing 226Mendelian randomisation studies 227Epigenetics: a space to watch 228Ten questions to ask about a genetic association study 230Exercises based on this chapter 234References 234Chapter 16 Applying evidence with patients 237The patient perspective 237Patient- reported outcome measures 239Shared decision- making 240Option grids 243n-of-1 trials and other individualised approaches 244Exercises based on this chapter 246References 247Contents xiChapter 17 Papers on artificial intelligence in healthcare 249Introduction 249Artificial intelligence 251Big data 253Machine learning 254Generative artificial intelligence: large language and multimodal models 254Ethical principles for the use of artificial intelligence for health 255Appraising artificial intelligence papers: a plethora of checklists 256Ten questions to ask about a paper that reports AI studies in healthcare 260Summary 264Exercises based on this chapter 264References 265Chapter 18 EBM+: the importance of mechanistic evidence 268What is mechanistic evidence? An example 268The many types of mechanistic evidence and a preliminary hierarchy 269EBM+ means ‘both and’, not ‘either or’ 270Mechanistic evidence in the COVID-19 pandemic 272Exercises based on this chapter 275References 276Chapter 19 Papers that report consensus exercises 278Why are consensus method papers important? 279How do experts choose and reach consensus on a specific topic? 279Consensus methods 281Ten questions to ask about a paper that reports a consensus statement 285Exercises based on this chapter 290References 291Chapter 20 Criticisms of evidence-based healthcare 293What’s wrong with evidence-based healthcare when it’s done badly? 293What’s wrong with evidence-based healthcare when it’s done well? 296Why is ‘evidence-based policymaking’ so hard to achieve? 299Exercises based on this chapter 301References 301Appendix 1 Checklists for finding, appraising and implementing evidence 304Appendix 2 Assessing the effects of an intervention 316Index 317