Meta-analytic Survival Analysis — Pooling Time-to-Event Evidence
Meta-analytic Survival Analysis · Also known as: meta-analysis of time-to-event data, pooled survival analysis, IPD survival meta-analysis, aggregate survival meta-analysis
Meta-analytic survival analysis is a quantitative synthesis method that pools hazard ratios and related time-to-event statistics from multiple independent studies to produce a single, more precise estimate of a treatment or exposure effect on survival outcomes such as overall survival, disease-free survival, or time to relapse. It can operate on aggregate published data or on individual patient data (IPD) contributed directly by study investigators.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use meta-analytic survival analysis when multiple studies have addressed the same clinical or epidemiological question with a time-to-event outcome and the individual studies lack power to be conclusive. It is particularly valuable for evidence synthesis in oncology, cardiology, and infectious disease where survival is the primary endpoint. IPD meta-analysis should be considered when aggregate data are insufficient to answer subgroup or interaction questions. Do NOT apply this method when fewer than three to five eligible studies exist (precision gains are minimal and heterogeneity cannot be reliably estimated), when survival endpoints are not consistently defined across studies, or when publication bias is severe and unaddressed — in these situations, a qualitative systematic review may be more honest.
Strengths & limitations
- Substantially increases statistical power over any single study by pooling data across multiple trials or cohorts.
- Provides the highest level of aggregate evidence for survival effects, especially when individual trials are underpowered.
- IPD meta-analysis allows patient-level subgroup and interaction analyses that are impossible with aggregate data alone.
- Systematic evidence synthesis reduces the influence of any single study's idiosyncrasies on clinical conclusions.
- Produces a summary hazard ratio that is directly interpretable as a relative risk of the event over time.
- Can incorporate unpublished data via IPD collaboration, partially addressing publication bias.
- Quality of the pooled estimate is constrained by the quality of the underlying primary studies — poor trials produce a precise but biased summary.
- Heterogeneity in how survival endpoints are defined, measured, or adjudicated across studies can introduce clinical noise even when statistical heterogeneity appears low.
- IPD meta-analysis requires substantial resources: obtaining patient-level data from many study teams is logistically demanding and time-consuming.
- Reconstruction of HRs from Kaplan-Meier curves or aggregate statistics introduces measurement error and may underestimate variance.
- Publication bias — studies with null or negative results are less likely to be published, potentially inflating the apparent treatment benefit.
Frequently asked
What is the difference between aggregate data and IPD meta-analysis of survival outcomes?
Aggregate data meta-analysis uses summary statistics (published HRs and CIs) extracted from each study's report. IPD meta-analysis obtains the raw patient-level dataset from each study and re-analyses it centrally. IPD is more powerful — it allows time-varying covariate adjustment, patient-level subgroup analyses, standardized endpoint definitions, and verification of original analyses — but requires substantially more time and investigator collaboration.
Can I reconstruct a hazard ratio from a Kaplan-Meier curve if the paper does not report one?
Yes. Methods described by Parmar et al. (1998) and implemented in tools such as DigitizeIt or the R package 'IPDfromKM' allow reconstruction of approximate log-HR and variance from published survival curves and numbers at risk. The reconstruction introduces some error but is far better than excluding the study or using a cruder effect measure.
How do I handle non-proportional hazards across studies?
When the proportional hazards assumption is violated — meaning the HR changes over time in one or more studies — the pooled HR is an average over time that may misrepresent the effect at clinically important time points. Consider reporting restricted mean survival time (RMST) differences as a complementary summary, or stratifying the meta-analysis by follow-up period.
What level of I-squared indicates that pooling is inappropriate?
There is no strict cut-off; I2 must be interpreted in context. An I2 of 25% is low, 50% moderate, and 75% high by conventional thresholds, but even high heterogeneity may be pooled if the source is understood and subgroup analyses are informative. Very high heterogeneity with no identifiable explanation is a warning sign that the studies are too clinically diverse to combine meaningfully.
Does meta-analytic survival analysis require a registered protocol?
Pre-registration is strongly recommended and increasingly required by journals. PROSPERO accepts protocols for systematic reviews and meta-analyses of interventions and observational studies with survival outcomes. Pre-registration constrains post-hoc subgroup analyses and protects the validity of conclusions.
Sources
- Parmar, M. K. B., Torri, V., & Stewart, L. (1998). Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Statistics in Medicine, 17(24), 2815–2834. DOI: 10.1002/(SICI)1097-0258(19981230)17:24<2815::AID-SIM110>3.0.CO;2-8 ↗
- Tierney, J. F., Stewart, L. A., Ghersi, D., Burdett, S., & Sydes, M. R. (2007). Practical methods for incorporating summary time-to-event data into meta-analysis. Trials, 8, 16. DOI: 10.1186/1745-6215-8-16 ↗
How to cite this page
ScholarGate. (2026, June 3). Meta-analytic Survival Analysis. ScholarGate. https://scholargate.app/en/epidemiology/meta-analytic-survival-analysis
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Cox proportional hazardsEpidemiology↔ compare
- Individual Patient Data Meta-AnalysisEvidence Synthesis↔ compare
- Kaplan-Meier AnalysisEpidemiology↔ compare
- Survival AnalysisResearch Statistics↔ compare