Pairwise Analysis

DataParty uses Python 3.8.10 for data analysis with the following packages: Matplotlib (3.4.1), NumPy (1.20.2), Pandas (1.2.3), and SciPy (1.5.2).

Forest Plots

For pairwise analysis of dichotomous data, the Mantel-Haenszel method is used.(1) The DerSimonian-Laird method is used for the random effects model.(2) If there are zero respondents (or non-respondents) in either arm of the study, DataParty applies adjustment values of 0.5 and 1 to the numerator and denominator of both arms, respectively. DataParty advises against pooling results from studies that contain zero respondents in both arms of the pairwise comparison.

For pairwise analysis of continuous data, the inverse variance method is used. The DerSimonian-Laird method is used for the random effects model.(2) DataParty implements Hedges's g for standardized mean difference.(3)

Meta-Regression

For dichotomous data, meta-regression is performed with the logarithm of estimates. Coefficients express the impact of the independent variable(s) on the logarithm of the estimate. DataParty back-transforms the logarithm of estimates exclusively for visualization. DataParty does not transform or back-transform continuous data.

For fixed effects, the inverse variance method is used. For the random effects model, DataParty implements a DerSimonian-Laird method of moments estimator.(4)

References

1. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. JNCI: Journal of the National Cancer Institute. 1959. https://doi.org/10.1093/jnci/22.4.719

2. DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled Clinical Trials. 1986. https://doi.org/10.1016/0197-2456(86)90046-2

3. Hedges LV. Distribution theory for Glass's estimator of effect size and related estimators. Journal of Educational Statistics. 1981. https://doi.org/10.2307/1164588

4. Veronicki AA, Jackson D, Viechtbauer W, Bender R, Bowden J, Knapp G, Kuss O, Higgins JPT, Langan D, Salanti G. Methods to estimate the between-study variance and its uncertainty in meta-analysis. Research Synthesis Methods. 2014. https://doi.org/10.1002/jrsm.1164