Concept Architecture
Concept
Theoretically, Aggregate Data Meta-Analysis is a statistical evidence-synthesis method that combines study-level summary estimates from multiple independent studies to estimate an overall treatment effect. It is founded on sampling theory and the principle that each study provides an estimate of a common or related underlying effect with a quantifiable degree of uncertainty. The method exists to increase statistical precision, assess consistency across studies and support comparative-effectiveness conclusions when individual participant data are unavailable.
Mathematically, aggregate data meta-analysis represents the pooled effect as a weighted average of study-specific effect estimates. Under a fixed-effect model, weights are determined by the inverse of each estimate?s sampling variance. Under a random-effects model, weights incorporate both within-study variance and estimated between-study variance, allowing the underlying treatment effect to vary across studies. Heterogeneity is commonly quantified using Cochran?s Q, �� and I�.
In practice, treatment effects and standard errors are extracted from published reports, trial registries or regulatory submissions and transformed onto a common scale. Studies are weighted, pooled and assessed for statistical heterogeneity, sensitivity and potential publication bias. In health economics, pooled clinical effects are frequently used as efficacy inputs in decision-analytic models, cost-effectiveness analyses and health technology assessments.
Purpose
Used to synthesise study-level evidence, estimate pooled treatment effects, quantify heterogeneity and provide comparative-effectiveness parameters for health economic evaluation and decision modelling.
Mathematical Formulae
Primary Formula
Fixed-effect pooled estimate:
?? = ????? w???? / ????? w?
where:
w? = 1 / v?
Random-effects pooled estimate:
??RE = ????? w???? / ????? w?
where:
w?* = 1 / (v? + ��)
Supporting Formulae
Cochran's heterogeneity statistic:
Q = ????? w?(??? ? ??)�
Between-study variance (DerSimonian?Laird):
�� = max{0, (Q ? (k ? 1)) / (?w? ? ?w?� / ?w?)}
Inconsistency statistic:
I� = max{0, (Q ? (k ? 1)) / Q} ? 100%
Standard error of pooled estimate:
SE(??) = �(1 / ?w?)
95% confidence interval:
?? � 1.96 ? SE(??)
Related Mathematical Methods
- Fixed-Effect Meta-Analysis
- Random-Effects Meta-Analysis
- Inverse-Variance Weighting
- DerSimonian?Laird Estimation
- Restricted Maximum Likelihood Estimation (REML)
- Meta-Regression
- Network Meta-Analysis
- Publication Bias Assessment
Example
Four randomised trials report log hazard ratios for a new oncology treatment.
| Study | Log(HR) | Variance |
|---|---|---|
| 1 | ?0.223 | 0.040 |
| 2 | ?0.162 | 0.050 |
| 3 | ?0.287 | 0.060 |
| 4 | ?0.105 | 0.045 |
Using inverse-variance weights:
w? = 25.00
w? = 20.00
w? = 16.67
w? = 22.22
Pooled estimate:
?? = (25??0.223 + 20??0.162 + 16.67??0.287 + 22.22??0.105) / (25 + 20 + 16.67 + 22.22)
?? = ?0.189
Pooled hazard ratio:
HR = exp(?0.189) = 0.83
The meta-analysis estimates a 17% reduction in the hazard of the outcome relative to the comparator.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| SUMPRODUCT | =SUMPRODUCT(B2:B5,C2:C5)/SUM(C2:C5) | Calculates the inverse-variance weighted pooled estimate. |
| SUM | =SUM(C2:C5) | Calculates the total study weight. |
| SQRT | =SQRT(1/SUM(C2:C5)) | Calculates the standard error of the pooled estimate. |
| EXP | =EXP(F2) | Converts pooled log effect to hazard ratio or odds ratio. |
| SUMPRODUCT | =SUMPRODUCT(C2:C5,(B2:B5-$F$2)^2) | Calculates Cochran's Q statistic. |
VBA (Optional)
Automate meta-analysis by importing study-level summary estimates, calculating pooled fixed-effect and random-effects estimates, heterogeneity statistics and forest plot inputs.
Sources
- Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Introduction to Meta-Analysis.
- Higgins JPT, Thomas J, Chandler J, et al. Cochrane Handbook for Systematic Reviews of Interventions.
- DerSimonian R, Laird N. Meta-analysis in Clinical Trials. Controlled Clinical Trials. 1986.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation.
- NICE. Health Technology Evaluation Manual.
- ISPOR Good Practice Reports.
Related Concepts (2)
Library
Publications
1
Network Meta-Analysis for Decision Making — Dias, Ades, Welton, Jansen & Sutton, 1st Edition ed., 2018 (John Wiley & Sons)
The definitive text on network meta-analysis (mixed treatment comparisons) for decision making, presenting a coherent Bayesian framework (implemented in WinBUGS) for synthesising evidence across multiple treatments, including inconsistency, bias adjustment, and use in cost-effectiveness models.
BookView source →
Frequently Asked Questions (6)
What is aggregate data meta-analysis?
A technique combining summary results, such as reported means or event rates, from multiple studies into a single pooled estimate.
Source: DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled Clinical Trials. 1986;7(3):177-188. doi:10.1016/0197-2456(86)90046-2.
What does aggregate data meta-analysis combine from each study?
Aggregate data meta-analysis pools the summary results reported by each study, such as a mean difference or an event rate, rather than the records of individual participants. It combines these published figures into a single overall estimate, weighting each study by its size and precision so that larger, more reliable studies count for more. This makes it practical, since summary results are what journals report, but it is limited to the analyses each study happened to publish and cannot explore effects within subgroups. Pooling reported summaries is its method. Borenstein and colleagues (2009) describe this.
Source: Borenstein et al. 2009
How does aggregate data meta-analysis work?
Aggregate data meta-analysis works by extracting the summary effect estimate and its precision from each study, then combining them into a pooled estimate that weights each study according to its precision, typically giving more weight to larger, more precise studies. Fixed-effect or random-effects models are used depending on assumptions about heterogeneity. Measures of heterogeneity assess variation across studies. So aggregate data meta-analysis combines the reported summary results across studies through a weighted average, producing an overall estimate and its confidence interval, along with an assessment of how much the studies vary.
Source: DerSimonian & Laird 1986
What are the advantages of aggregate data meta-analysis?
The advantages of aggregate data meta-analysis include that it uses summary data usually available from publications, making it feasible and relatively quick without needing the original patient data; it can combine many studies to produce a more precise overall estimate than any single study; and it allows assessment of consistency across studies. It is widely applicable and practical. So aggregate data meta-analysis is advantageous for its feasibility and its ability to synthesise the available published evidence efficiently, providing a combined estimate and an assessment of heterogeneity from data that are commonly accessible.
Source: Higgins et al. 2011
What are the limitations of aggregate data meta-analysis?
The limitations of aggregate data meta-analysis include its reliance on the summary data reported, which may be incomplete or inconsistent across studies, limiting what can be combined; its inability to examine effects within subgroups or to adjust for patient-level factors without individual data; its susceptibility to bias if the reported results are affected by selective reporting or publication bias; and the challenge of heterogeneity between studies. These limitations mean aggregate data meta-analysis is interpreted with attention to reporting completeness, heterogeneity, and potential bias, and, where patient-level analysis is needed, complemented by individual patient data meta-analysis.
Source: DerSimonian & Laird 1986
How does aggregate data meta-analysis differ from individual patient data meta-analysis?
Aggregate data meta-analysis combines the summary results reported by each study, while individual patient data meta-analysis obtains and analyses the original patient-level data from each study. Aggregate data meta-analysis is more feasible, using available published data, but is limited to what is reported and cannot examine patient-level effects or subgroups directly. Individual patient data meta-analysis allows more detailed, consistent analysis, including of subgroups, but requires obtaining the raw data, which is more demanding. So the two differ in the level of data used, trading the feasibility of summary data against the flexibility of patient-level analysis.
Source: Stewart & Parmar 1993
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 1 Dec 2025
Content version: 1.0.0
Canonical Identity
- Term code
- HE-ES-ESM-001
Stable URI · Machine-readable · Resolvable · CC BY 4.0