An individual patient data meta-analysis obtains the raw participant-level records from each eligible trial and re-analyses them together, rather than pooling the summary results the trials published. Working from the original data lets us harmonise outcome definitions across studies, apply one consistent model, check randomisation integrity, and study effect modifiers at the patient level. Our service spans the whole project: negotiating data-sharing agreements, cleaning and harmonising datasets, fitting one-stage or two-stage models, and reporting to PRISMA-IPD. This is often called the gold-standard approach to synthesis.
Individual patient data meta-analysis services
When trial-level averages cannot answer who benefits, we obtain and harmonise the raw participant records and re-analyse them together to a defensible standard.
4.9 / 5 across 1,194+ delivered projects
- Cochrane and PRISMA 2020 methods
- PhD methodologists
- 100% human-written, no generative AI
- Reproducible R and Stata code
- Free quote within 24 hours
- Mutual NDA on request
When the raw data is worth the effort
An individual patient data meta-analysis is the most demanding form of evidence synthesis, and it is not the right tool for every question. Its value appears when trial-level averages are simply the wrong unit of analysis: when you need to know which patients benefit, when outcomes were defined inconsistently across trials, or when a time-to-event analysis has to be done properly rather than reconstructed from a published curve. In those cases the raw data answers questions an aggregate meta-analysis cannot.
It also carries a real cost in time and coordination, because it depends on trial teams agreeing to share data that took years to collect. We are candid about that trade-off from the first conversation. If your question can be answered well from published results, we will say so rather than steer you into a long collection effort you do not need. Where the participant-level data genuinely changes the answer, the effort is justified.
How an individual patient data project runs
Each stage below has its own risks and safeguards, and the collection and harmonisation work is the part that most often decides whether the project succeeds. We report every step so the synthesis is reproducible and defensible.
- 1
Identify trials and secure data-sharing agreements
We run the search, confirm eligibility, then approach each trial team or data custodian and negotiate a formal data-sharing agreement before any participant records are transferred.
- 2
Clean and harmonise the datasets
We align variables, units, and outcome definitions across every dataset so participants from different trials enter one coherent analysis, resolving mismatches with the original teams rather than assuming.
- 3
Check randomisation and data integrity
We verify baseline balance, screen for implausible or duplicated values, reconcile each dataset against its own publication, and account for participants lost to follow-up before pooling.
- 4
Fit the model and analyse effect modifiers
We choose a one-stage or two-stage approach on a stated rationale, estimate the overall effect, and examine patient-level subgroups and interactions to see how the effect varies.
One-stage and two-stage models
The two analytical routes give consistent answers when applied carefully, but they differ in flexibility and in what your reviewers will find easiest to follow. We match the approach to your data and your question rather than defaulting to one.
One-stage hierarchical models
Every participant from every trial is modelled together in a single mixed model with trial as a clustering factor, which is the stronger route for studying interactions and effect modifiers.
Two-stage synthesis
The model is fitted within each trial first, then the per-trial estimates are pooled like a conventional meta-analysis, a transparent choice that works when datasets cannot be fully merged.
Effect-modifier and subgroup analysis
We separate genuine within-trial interactions from misleading across-trial patterns, so subgroup findings reflect real patient-level effect modification rather than ecological artefact.
Time-to-event and mixed outcomes
We handle survival, continuous, and binary outcomes in the participant data directly, including proper handling of censoring that a published summary cannot support.
What you receive
You get a complete, reportable synthesis built on the raw data: the harmonised analysis datasets and a data dictionary, the integrity-check record, the pooled effect with its interval, the patient-level subgroup and interaction analyses, the forest plots, and a methods-and-results write-up aligned with the PRISMA reporting standard and its individual patient data extension. Every figure ships with the analysis script, so the work reruns and your reviewers can trace each number back to the participant-level source.
- A record of the data-sharing agreements and the provenance of every dataset
- Harmonised analysis datasets with a documented data dictionary
- An integrity report covering randomisation balance, plausibility, and reconciliation with publications
- The pooled effect from a one-stage or two-stage model with a stated rationale for the choice
- Patient-level subgroup and effect-modifier analyses that separate within-trial from across-trial signals
- A methods-and-results section drafted to PRISMA-IPD, plus the annotated analysis code
Who commissions an individual patient data meta-analysis
The brief varies by who is asking and what the synthesis has to support, and we shape the collection and the analysis to match. What every commissioner shares is a question that trial-level summaries cannot answer and a willingness to invest in obtaining the raw data.
Collaborative research consortia
Multi-team projects pooling data across trials, where consistent harmonisation and a single defensible analysis matter more than speed.
Guideline and policy bodies
A gold-standard synthesis to underpin a recommendation, with patient-level effect modification examined so guidance can be targeted rather than averaged.
Doctoral and academic researchers
A participant-level synthesis chapter you can defend, with every harmonisation and modelling decision explained so the work is genuinely yours.
Clinical trial groups
Teams holding several related trials who want a rigorous pooled re-analysis, including the integrity checks the raw data make possible.
Individual patient data versus aggregate data synthesis
The honest comparison is about fit, not prestige. Individual patient data is powerful but costly, and an aggregate synthesis is often the responsible choice. The table below shows where the two part so you can see which suits your question.
| Consideration | Aggregate data meta-analysis | Individual patient data meta-analysis |
|---|---|---|
| Data used | Published summary results from each trial | Raw participant-level records obtained from trial teams |
| Subgroup analysis | Trial-level meta-regression, prone to ecological bias | Genuine within-trial effect-modifier analysis at the patient level |
| Typical timeline | Months, limited by the search and extraction | Considerably longer, driven by data-sharing and harmonisation |
| Best used when | The overall effect is the question and outcomes are reported consistently | You need to know who benefits, or outcomes need re-defining and verifying |
How the project is scoped and quoted
We quote a fixed fee per project rather than by the hour, so the cost is known before any work begins. Behind that number sits the real load: the number of trials whose data you need, how much harmonisation the datasets demand, whether a one-stage or two-stage analysis fits, whether a data preparation stage and search already exist or need building, and how much of the reporting you want us to draft. Because collection is the longest and least predictable stage, we scope it honestly and flag early if the data will not arrive in the time you have. Send us your question and the trials you have in mind, and we will return a plan, a timeline, and a fixed quote with no obligation to proceed.
Frequently asked questions
- What is the difference between an individual patient data meta-analysis and a standard meta-analysis?
- A standard meta-analysis combines the summary results, such as a mean difference or an odds ratio, that each trial has already published. An individual patient data meta-analysis instead obtains the raw participant-level records from each trial and re-analyses them together. Working from the original data lets us standardise outcome definitions and eligibility across trials, apply one consistent analysis to every dataset, handle missing data at the participant level, and study how the treatment effect varies with patient characteristics. It is more powerful but far more demanding, because it depends on trialists agreeing to share their data.
- What is the difference between one-stage and two-stage IPD meta-analysis?
- These are the two ways to analyse the pooled participant data. A two-stage analysis fits the chosen model within each trial first, producing one effect estimate per trial, then combines those estimates exactly as a conventional meta-analysis would. A one-stage analysis models all participants from all trials together in a single hierarchical model, keeping trial as a clustering factor. One-stage models are generally preferred for studying interactions and effect modifiers, while two-stage models are simpler, more transparent, and practical when datasets cannot be fully harmonised into one structure.
- Why do an individual patient data meta-analysis instead of an aggregate data one?
- The main reasons are power and honesty about how treatment effects vary between patients. Aggregate meta-regression across trial-level averages is prone to ecological bias and is usually underpowered to detect who benefits most. With participant-level data we can examine genuine effect modifiers within trials, standardise definitions across studies, verify reported results, and adjust consistently for baseline characteristics. Individual patient data meta-analysis is often called a gold-standard synthesis for these reasons, but it is only worth the effort when the added flexibility answers a question aggregate data cannot.
- How do you obtain the raw data from the original trials?
- We identify the eligible trials, approach the trial teams or data custodians, and negotiate a formal data-sharing agreement for each dataset before any records change hands. Some data arrive through managed-access repositories rather than directly. Every dataset is then cleaned, checked, and harmonised so that variables, units, and outcome definitions line up across trials. This collection and harmonisation stage is the longest part of the project and the reason an individual patient data meta-analysis takes considerably longer than an aggregate-data synthesis of the same question.
- How do you check the integrity of the trial data you receive?
- Before any pooled analysis we run integrity checks on each dataset. We confirm that randomisation produced balanced baseline characteristics, look for implausible values, duplicate records, and digit-preference patterns, reconcile the numbers against the trial's own publication, and account for participants lost to follow-up. Where a dataset fails these checks we go back to the trial team rather than quietly analyse around the problem. This scrutiny is one of the real advantages of holding the raw data, because it surfaces issues an aggregate synthesis would never see.
- When is an aggregate data meta-analysis enough instead?
- Often. If your question is a straightforward overall effect, the published trials report it consistently, and you do not need patient-level subgroup or effect-modifier analysis, an aggregate synthesis answers it faster and at far less cost. Individual patient data earns its keep when you need to study how effects differ between patients, standardise inconsistent outcomes, analyse time-to-event data properly, or verify results. We will tell you honestly when the raw data would not change your conclusion, so you do not fund a two-year collection effort for an answer aggregate data already gives.
Tell us your question and the trials you hope to include, and we will scope an individual patient data meta-analysis and return a fixed quote.
Free quote within 24 hours. 100% human-written by PhD methodologists.
Methodology reviewed by
Senior Review Statistician
Runs the quantitative synthesis: pooling models, heterogeneity, network meta-analysis, and the figures that go in the paper.
Why researchers bring in a PhD methodologist
80+
systematic reviews are published every day (Hoffmann et al., 2021)
67.3 weeks
average time to complete a review in-house (Borah et al., BMJ Open 2017)
~70%
of published reviews rate critically low on AMSTAR 2 quality appraisal
4.9 / 5
our client rating across 1,194+ delivered projects
Explore our services
Systematic review writing servicesThe complete systematic review, run by methodologists from protocol to a submission-ready manuscript.Meta-analysis servicesPooled effect estimates, heterogeneity, forest and funnel plots, and the full quantitative synthesis, done correctly.Protocol and PROSPERO preparationA pre-specified protocol, prepared registration-ready for PROSPERO, that locks your methods before screening begins.Literature search and search strategyReproducible, peer-reviewable search strategies across every database your review needs.