← All decks
Orthopaedics
Foundations of Research Methodology, Evidence Hierarchies, and Trial Quality
Built from Orthopedic Trauma Essentials
What’s inside
13 sections · 145 slides
Why research methods matter in trauma
High-stakes decisions, made fast, on mixed evidence
- The decisions that come first
- Why the evidence is often mixed
- What evidence-based medicine really means
- Two extra skills a trauma reader needs
- Three questions to ask of any study
- The three questions, spelled out
Framing the question
From PICO to estimand, and where the clock starts
- Start with a question you can answer
- PICO with a trauma example
- What an estimand adds
- Intercurrent events are normal in trauma
- Time zero: where the clock starts
- How a fuzzy clock fakes a benefit
- How a mis-set clock invents a benefit
Evidence hierarchies
A useful shortcut with two important cautions
- What an evidence hierarchy is for
- The classic pyramid, top to bottom
- The evidence pyramid, from case series to systematic reviews
- Two cautions before you trust the ranking
- The level depends on the question
- Four question types, four ladders
Systematic reviews and meta-analysis
Comprehensive by design, and fragile in specific ways
- What a systematic review is
- What a meta-analysis adds
- Why syntheses sit at the top
- A review inherits the level of its studies
- What makes a review credible
- Extra checks when the numbers are pooled
- Reading the study-selection flow chart
- Tracking studies from database search to final inclusion
- Pooling study results into one summary estimate
- Checking whether studies are missing
- How a systematic review goes wrong
- Biases that keep coming back in trauma research
- Common sources of bias in orthopaedic trauma research: what each one means
- Common sources of bias in orthopaedic trauma research: setting, effect and mitigation
- The two biases to spot first
Randomized controlled trials
What randomisation buys, and what it cannot protect
- What a randomized trial is
- Why randomisation is so powerful
- The label is not the proof
- Sequence generation and allocation concealment
- Blinding when the operation is visible
- What to do when the surgeon cannot be blinded
- How participants flow through a parallel randomized trial
- Five ways a randomized trial still goes wrong
- Trial variants: cluster and stepped-wedge
- Trial variants: factorial and adaptive
Quasi-experimental designs
Causal questions when randomisation is off the table
- When you cannot randomize
- What quasi-experiments buy you
- What a credible quasi-experiment needs
- Where quasi-experiments go wrong
- Three ways to build the comparison
Observational analytical designs
The workhorse of trauma literature, and its blind spots
- Why observational studies dominate trauma
- Four observational designs
- Cohort studies start with the exposure
- Why retrospective cohorts are common - and risky
- What a credible cohort needs
- How cohort studies go wrong
- Missing data in trauma registries
- Controlling confounding after the fact
- What propensity scores cannot do
- Case-control studies start with the outcome
- What a good case-control study looks like
- How case-control studies go wrong
- Cross-sectional studies: a single snapshot
- Getting a snapshot right, and what spoils it
- Case series: description without a comparator
- What a good case series still requires
- How case series mislead
- Telling the three designs apart
Diagnostic, prognostic and registry research
Three question types the therapy pyramid does not fit
- Diagnostic accuracy studies
- What a good diagnostic study needs
- Five biases that flatter a test
- Prognostic studies and prediction models
- What a prediction model must report
- How prognostic studies go wrong
- Registries and routine data
- Reporting standards for routine data
- The registry-based randomized trial
- What registry research must specify
- How registry studies go wrong
What makes any study strong
Power, endpoints, missing data and consent
- Sample size and power
- Multiplicity: the more you test, the more you find
- Surrogate versus patient-centred endpoints
- Measuring what matters to patients
- Missing data and loss to follow-up
- Intention to treat, and testing robustness
- Consent when the patient cannot give it
- Recognised routes around standard consent
- The current ethics and practice standards
Certainty of evidence
GRADE, and what to do when the evidence is thin
- GRADE separates two judgements
- The five things that lower certainty
- Four levels of certainty
- Why GRADE suits trauma
- Acting on high versus low certainty
- Strength of recommendation is more than evidence
Appraising quality and risk of bias
Which tool for which design, and what each one outputs
- What trial quality actually means
- The recurring weak points in trauma work
- What to look for without any tool
- Tools follow the design
- Risk-of-bias and methodological appraisal tools by study design
- The appraisal toolkit at a glance
- RoB 2: the standard for randomized trials
- How a RoB 2 judgement is reached
- A risk-of-bias grid across trials and domains
- Other checklists for randomized trials
- ROBINS-I: bias in non-randomized studies
- What ROBINS-I produces
- MINORS: a score for surgical studies
- ROBINS-I or MINORS?
- The Newcastle-Ottawa Scale
- Checklists for case reports and case series
- AMSTAR 2: judging the review itself
- What AMSTAR 2 tells you, and its neighbours
- Tools for diagnostic accuracy studies
- TRIPOD and its extensions
- PROBAST for prediction model bias
- Reporting guidelines, one per design
- Reporting guidelines relevant to orthopaedic trauma research
Using the evidence at the bedside
A five-step workflow, and three worked examples
- A reproducible way to read a paper
- Step 1: clarify the decision
- Steps 2 and 3: is it true, and how big?
- Steps 4 and 5: does it fit, and what now?
- Aspirin or injectable heparin after fracture surgery?
- Reading the thromboprophylaxis trial
- Which irrigation for an open tibia fracture?
- Reading the wound irrigation trial
- Reamed or unreamed nail?
- Reading a composite endpoint carefully
Take-home points
What to carry into the next paper you read
- Design and question first
- Judge credibility, then relevance
- Then decide, and stay ready to change
- References
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- References (continued)
- Orthopedic Trauma Essentials 2026