In meta-analysis, what is the key difference between using individual patient data (IPD) and aggregate data?

In meta-analysis, what is the key difference between using individual patient data (IPD) and aggregate data?

• A. IPD uses published statistics only; aggregate uses raw data
• B. IPD uses raw patient data from each trial; aggregate uses summary statistics from each trial
• C. IPD is only used for diagnostic accuracy studies
• D. IPD cannot handle time-to-event data

Answer: (B)

The key difference is that individual patient data uses the raw data for every participant from each trial, while aggregate data relies on the summary statistics reported in each trial’s publication. With IPD you can reanalyze all trials with a consistent approach, harmonize definitions, and handle time-to-event data and censoring uniformly. You can also adjust for patient-level covariates to explore subgroups, improving flexibility and precision beyond what trial-level summaries alone can provide. This is why IPD is considered more powerful for meta-analysis when available. One alternative is incorrect because IPD involves raw data rather than only published statistics, and another option incorrectly limits IPD to diagnostic accuracy studies. The final option is also incorrect, since IPD can indeed accommodate time-to-event data through proper survival analyses.

The key difference is that individual patient data uses the raw data for every participant from each trial, while aggregate data relies on the summary statistics reported in each trial’s publication. With IPD you can reanalyze all trials with a consistent approach, harmonize definitions, and handle time-to-event data and censoring uniformly. You can also adjust for patient-level covariates to explore subgroups, improving flexibility and precision beyond what trial-level summaries alone can provide. This is why IPD is considered more powerful for meta-analysis when available.

One alternative is incorrect because IPD involves raw data rather than only published statistics, and another option incorrectly limits IPD to diagnostic accuracy studies. The final option is also incorrect, since IPD can indeed accommodate time-to-event data through proper survival analyses.