Systematic review with individual participant data meta-analysis
What can be learned by harmonizing participant-level data across studies?
When to use this project
Use this starter when obtaining original participant data is feasible and serves a defined question, such as consistently defined outcomes or treatment–covariate interactions. Plan data access, governance and harmonization before analysis.
How it differs
IPD describes the data level, not an alternative to every other synthesis type. It can support pairwise or network methods. More granular data do not remove original study bias or bias from unavailable datasets.
Decisions before you start
- Track all eligible studies, including those that do not supply IPD; define any aggregate-data contribution.
- Preserve study membership and randomization in the model. Do not analyse a concatenated dataset as one large trial.
- Separate within-study interaction information from between-study differences; record missing-data and harmonization assumptions.
How the flows work together
Each flow below links to its diagram, steps and tables. Flow order organizes the project; it does not imply that all work is strictly sequential. Data assessment, analysis planning and protocol refinement may inform one another.
1. Participant-data review overview
Overview · Question through participant-level synthesis
Use this overview to state the purpose and link to detailed decisions. Keep a versioned protocol and decision register, and add evidence references for the rationale.
2. Study identification and access
Selection · Track studies and data availability
Study eligibility and data availability are separate. Keep requests and reasons in the register; use secure research infrastructure for participant datasets. This canvas documents the process and is not the participant-data repository.
3. Data integrity and harmonization
Data · Reconcile participant variables across studies
Retain the original variable definitions and a versioned transformation dictionary. Resolve integrity queries with study teams; unmeasured variables cannot be recovered by relabeling columns.
4. Participant-data synthesis plan
Statistics · Model participants within study structures
Specify study structure in both one-stage and two-stage approaches. The two-stage equation is illustrative; model choice, interactions and missingness need their own specifications.
5. Bias, robustness and certainty
Statistics · Assess availability and analytical assumptions
Assess unavailable studies and harmonization choices alongside study-level bias. IPD access does not make the evidence base complete or unbiased.
Equations and statistical interpretation
These explanations describe notation and assumptions. Flow does not calculate these values.
Illustrative two-stage synthesis
β̂ⱼ is the treatment effect estimated within study j using a compatible model; wⱼ reflects its uncertainty and the chosen synthesis model.
A two-stage approach first estimates within-study effects, then combines them. A one-stage model instead jointly models participants while retaining study structure. This notation does not specify an interaction or missing-data model.
Published example
A published IPD review demonstrating data acquisition and participant-level synthesis. Read its access, outcome and subgroup methods; this example is not a current treatment recommendation.
This publication illustrates a study design, not an endorsement of every method or a reproduction of its results. Follow its protocol, supplement and update history when studying the example.
Hypothetical scenario
A collaboration wants to examine whether baseline vitamin D status modifies the effect of supplementation on respiratory infection using data from several randomized trials.
- Participant-data review overview: Prespecify the interaction question, outcome window and analysis plan.
- Study identification and access: Identify eligible studies; record requests, permissions, refusals and supplied variables in the access register.
- Data integrity and harmonization: Reconcile study totals and define common units, baseline categories and outcome windows without erasing incompatibilities.
- Participant-data synthesis plan: Choose one-stage or two-stage analysis with study structure and explicit missing-data methods.
- Bias, robustness and certainty: Examine availability and harmonization sensitivity, then assess certainty of the outcome-specific conclusions.
Deliver a populated protocol, decision log, data specification, and planned outputs. Enter numerical findings only after analysis and verification. Keep decisions made before seeing results distinguishable from later amendments.
Sources
- Cochrane Handbook, Chapter 26: Individual participant data
- PRISMA-IPD: Individual participant data
- PRISMA 2020 flow diagrams: identification and selection
- Cochrane Handbook, Chapter 14: Certainty and Summary of findings
Source check: 2026-09-17. Record the version of each reporting or appraisal tool used in your actual protocol.
Compare all Evidence synthesis project types · Open editor in a new tab