In Package-Metaan-Rep/Metaan: Meta-Analysis of Published Epidemiological Studies

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

Metaan

The goal of Metaan package is to help user to compute a Meta-Analysis of Published Epidemiological Studies.

Description:

This package helps to combine estimates from published individual studies and provide a summarized or a pooled estimate of evidence. It makes use of statistical methods of meta-analysis computation such as the fixed effect model and the random effect model. These methods are ready for risk estimates (e.g relative risk (RR), odds ratio (OR) and hazard ratio (HR)) meta-analyses but are improved using the alternative method of Richardson et al. 2020 for the computation of excess relative risk (ERR) estimates metanalyses.

Installation

You can install the released version of Metaan from CRAN with:

install.packages("Metaan")
library(Metaan)

Install directly from github :

install.packages("devtools") # if you do not have it already
devtools::install_github("Package-Metaan-Rep/Metaan")
library(Metaan)

library(Metaan)

Data format

Formate the dataset in order to have your individual studies in lines with the following columns :

A column with your ERR/EOR for each study
A column with the upper bound of the confidence interval of the ERR/EOR for each study
A column with the lower bound of the confidence interval of the ERR/EOR for each study
A column reporting the maximum dose from each study
A column with the study characteristics (Authors, Year of publication etc.). This column is optional and can be replaced by numbers.

An example from the ischemic heart disease (IHD) dataset:

data(IHD)
print(IHD)

In this example, the "Study" column contains information on the Authors and on the date of the study. "err" is the value expressing the excess relative risk (err) reported from the study. "lower_ci" is the value of the lower bound of the confidence interval of the risk reported from the study. "upper_ci" is the upper bound of the confidence interval of the risk reported from the study and the "max_dose" column gives the maximum dose (in millisievert mSv) reported from the study.

Estimate the summary effect

To estimate the pooled effect from this set of studies, we will make use of functions "alpexfix" or "alpexrand" for respectively estimate using the alternative fixed effect model or the alternative random effect model. For more details about the alternative model, I recommend you this interesting article Richardson et al. 2020, and to know more the fixed and the random effect models, there is some references at the end of the tutorial.

Estimate the fixed effect summary :

alpexfix(err=IHD$err,
         u=IHD$upper_ci,
         l=IHD$lower_ci,
         d=IHD$max_dose)

Estimate the random effect summary :

alpexrand(err=IHD$err,
         u=IHD$upper_ci,
         l=IHD$lower_ci,
         d=IHD$max_dose)

These functions estimate the summary effect using the correction method proposed by Richardson et al. 2020. However user can choose to use the standard model without correction. To do so, user can choose the following functions "pexfix" and "pexrand".

Estimate the fixed effect summary using the standard model :

pexfix(err=IHD$err,
         u=IHD$upper_ci,
         l=IHD$lower_ci)

Estimate the random effect summary using the standard model :

pexrand(err=IHD$err,
         u=IHD$upper_ci,
         l=IHD$lower_ci)

Estimate the summary effect in a sensitivity analysis

Random effect using the alternative model

exsens(study = IHD$Study,
       err=IHD$err,
         u=IHD$upper_ci,
         l=IHD$lower_ci,
         d=IHD$max_dose,
         test = "RANDOM", # you can choose test="FIXED" for the fixed effect model
       model = "alternative") # you can choose model = "standard" for the fixed effect model

Random effect using the standard model

exsens(study = IHD$Study,
       err=IHD$err,
         u=IHD$upper_ci,
         l=IHD$lower_ci, # d=NULL when using the standard model with model = "standard"
         test = "RANDOM", # you can choose test="FIXED" for the fixed effect model
       model = "standard")

References

DerSimonian R, Laird N (1986) Meta-analysis in clinical trials. Controlled clinical trials 7:177–188.

Larry V Hedges and Ingram Olkin (1985), Statistical methods for meta-analysis. ISBN 978-0-08-057065-5, https://doi.org/10.1016/C2009-0-03396-0

Richardson, D. B., Abalo, K., Bernier, M. O., Rage, E., Leuraud, K., Laurier, D., ... & Little, M. P. (2020). Meta-analysis of published excess relative risk estimates. Radiation and Environmental Biophysics, 1-11.

Little MP, Azizova TV, Bazyka D, et al. Systematic review and meta-analysis of circulatory disease from exposure to low-level ionizing radiation and estimates of potential population mortality risks. Environ Health Perspect 2012;120(11):1503-11.