Individual Participant Data Meta-Analysis: Why Raw Data Beats Published Summaries

An individual participant data meta-analysis gathers the original line-by-line records from every eligible trial and reanalyzes them together, rather than combining the summary results each trial published. Because the raw data are in hand, the team can standardize how outcomes are defined, check whether randomization held, and ask whether an effect differs by age or severity in ways published tables cannot. That extra work is why it is often called the gold standard of evidence synthesis, though it depends entirely on trial teams agreeing to share their data.

What individual participant data actually means

Most meta-analyses work with aggregate data. Each trial reports its summary results, a mean difference here, an odds ratio there, and the reviewer stacks those published numbers together. An individual participant data meta-analysis works one level deeper. The team asks each trial for the original records, one row per person, carrying that person's treatment assignment, baseline features, and outcomes. Then they reanalyze everyone together as if the trials formed one large coordinated study.

That difference sounds administrative, but it changes what the review can see. Published summaries are already cooked. You get whatever the original authors chose to calculate and report. With the raw data in hand, the reviewer can start again from the ingredients, on their own terms.

Why the raw data change what you can check

Holding the original records lets a review do things a summary can never support. Reviewers can standardize outcomes so that every trial defines response or remission the same way, rather than trusting that a dozen teams meant the same thing. They can check whether randomization actually held by looking at the balance of baseline features, and whether anyone who was randomized quietly disappeared from the analysis.

The raw data also recover information that publication left out. A trial may have measured an outcome at two years but only reported results at one. It may have collected an adverse event it never tabulated. IPD can bring those back. For time-to-event questions, having each person's follow-up allows a proper survival analysis instead of crude counts.

The subgroup question it answers best

The strongest reason to go to this trouble is the question of who benefits. With only summary data, testing whether a treatment works better in older patients means comparing trials that enrolled older populations against trials that enrolled younger ones. That comparison is confounded by everything else that differs between trials, a trap known as aggregation bias or the ecological fallacy.

With individual data you can test the interaction inside each trial, where randomization still protects the comparison, and then combine those within-trial interactions across studies. That is the most dependable way to ask whether an effect genuinely changes with age, severity, or a biomarker, rather than merely appearing to across dissimilar trials.

What it cannot fix

Working with the original data does not launder a weak trial. If the underlying studies were biased, small, or poorly conducted, an IPD analysis inherits those flaws; it simply sees them more clearly. There is also a subtler risk. Teams cannot always obtain data from every eligible trial, and the trials that share may differ from the trials that decline, which can bias the pool much as missing studies bias any review.

And it is slow. Requesting, cleaning, and harmonizing data from many groups can take years of effort. That cost is why most questions are still answered with aggregate data, and why a topic without an IPD review is not a topic with weak evidence.

How to read one without being a statistician

When you meet a review that used individual participant data, a few checks tell you how much to trust it. Look for what fraction of the eligible participants the team actually obtained; a review that gathered ninety percent of the data stands on firmer ground than one that reached forty. Look for a statement that the data were checked for integrity, the kind of step the PRISMA-IPD reporting extension asks authors to describe.

Then read the subgroup claims with more confidence than usual. When an IPD review reports that an effect is stronger in one group, that finding rests on within-trial evidence, which is far more credible than the same claim drawn from summary tables across studies.