Detecting and correcting for publication bias in meta-analysis – A truncated normal distribution approach

2016 ◽  
Vol 27 (9) ◽  
pp. 2722-2741 ◽  
Author(s):  
Qiaohao Zhu ◽  
KC Carriere
Keyword(s):  
Normal Distribution ◽  
Publication Bias ◽  
Random Effects ◽  
Effect Size ◽  
Meta Analysis ◽  
Truncated Normal Distribution ◽  
Random Effects Models ◽  
Funnel Plot ◽  
Fixed And Random Effects ◽  
Truncated Normal

Publication bias can significantly limit the validity of meta-analysis when trying to draw conclusion about a research question from independent studies. Most research on detection and correction for publication bias in meta-analysis focus mainly on funnel plot-based methodologies or selection models. In this paper, we formulate publication bias as a truncated distribution problem, and propose new parametric solutions. We develop methodologies of estimating the underlying overall effect size and the severity of publication bias. We distinguish the two major situations, in which publication bias may be induced by: (1) small effect size or (2) large p-value. We consider both fixed and random effects models, and derive estimators for the overall mean and the truncation proportion. These estimators will be obtained using maximum likelihood estimation and method of moments under fixed- and random-effects models, respectively. We carried out extensive simulation studies to evaluate the performance of our methodology, and to compare with the non-parametric Trim and Fill method based on funnel plot. We find that our methods based on truncated normal distribution perform consistently well, both in detecting and correcting publication bias under various situations.

Meta-analysis in Sociological Research: Power and Heterogeneity

2019 ◽  
pp. 004912411988247 ◽  
Author(s):  
Guangyu Tong ◽  
Guang Guo
Keyword(s):  
Random Effects ◽  
Science Studies ◽  
Meta Analysis ◽  
Sociological Research ◽  
Random Effects Models ◽  
Sociological Inquiry ◽  
Wide Range ◽  
Fixed And Random Effects ◽  
Study Heterogeneity ◽  
Analysis Models

Meta-analysis is a statistical method that combines quantitative findings from previous studies. It has been increasingly used to obtain more credible results in a wide range of scientific fields. Combining the results of relevant studies allows researchers to leverage study similarities while modeling potential sources of between-study heterogeneity. This article provides a review of the core methodologies of meta-analysis that we consider most relevant to sociological research. After developing the foundation of the fixed- and random-effects models of meta-analysis models, this article illustrates the utility of the method with regression coefficients reported from two sets of social science studies. We explain the various steps of the process including constructing the meta-sample from primary studies, estimating the fixed- and random-effects models, analyzing the source of heterogeneity across studies, and assessing publication bias. We conclude with a discussion of steps that could be taken to strengthen the development of meta-analysis in sociological research, which will eventually increase the credibility of sociological inquiry via a knowledge-cumulative process.

Fixed- and random-effects models in meta-analysis.

1998 ◽  
Vol 3 (4) ◽  
pp. 486-504 ◽  
Author(s):  
Larry V. Hedges ◽  
Jack L. Vevea
Keyword(s):  
Random Effects ◽  
Meta Analysis ◽  
Random Effects Models ◽  
Fixed And Random Effects

scholarly journals Interleukin-18 Polymorphisms Deficiency Association with Asthma Risk: An Update Meta-analysis

2019 ◽  
Author(s):  
Huan LIU ◽  
Yu Feng ◽  
Wei Zhang ◽  
Xiao-Dong Deng ◽  
Ying Ma ◽  
...  
Keyword(s):  
Publication Bias ◽  
Meta Analysis ◽  
Interleukin 18 ◽  
Promoter Polymorphisms ◽  
Random Effects Models ◽  
Asthma Risk ◽  
Genotype Frequencies ◽  
Fixed And Random Effects ◽  
Matching Criteria ◽  
Strength Of Association

Growing evidence indicated conflicting results that Interleukin-18 (IL-18) promoter polymorphisms rs1946518 (A-607C), rs187238 (G-137C) and rs549908 (A-105C) were associated with asthma risk. The aim of this study is to comprehensively evaluate the IL-18 polymorphisms and asthma by a systematic review and meta-analysis. A total of 12 studies testing the association between these polymorphisms and asthma were examined (8 studies for A-607C, 8 studies for G-137C, and 4 studies for A-105C) in the update meta-analysis, up to Dec 30, 2017. Summary odds ratios (ORs) and 95% confidence intervals (CI) were used to estimate the strength of association between each polymorphism and asthma using fixed- and random-effects models when appropriate. Heterogeneity and publication bias were evaluated. The meta-analysis results indicated that any allele frequencies of the IL-18 polymorphisms (A-607C, G-137C and A-105C) was not associated with asthma risk (p>0.05). And no statistically significant association was observed between genotype frequencies of these polymorphisms and asthma under different genetic models (p>0.05). Subgroup analysis results were similar to the main analysis by ethnicity, sample size, genotyping methods, matching criteria and quality score. There was no evidence of publication bias. The present meta-analysis suggests that IL-18 polymorphisms (A-607C, G-137C and A-105C) were unlikely to be associated with asthma risk.

scholarly journals Improving the accuracy of likelihood-based inference in meta-analysis and meta-regression

Biometrika ◽  
2017 ◽  
Vol 104 (2) ◽  
pp. 489-496 ◽  
Author(s):  
I. Kosmidis ◽  
A. Guolo ◽  
C. Varin
Keyword(s):  
Random Effects ◽  
Effect Size ◽  
Variance Component ◽  
General Framework ◽  
Meta Analysis ◽  
Asymptotic Bias ◽  
Likelihood Estimator ◽  
Random Effects Models ◽  
The Mean ◽  
Meta Regression

Summary Random-effects models are frequently used to synthesize information from different studies in meta-analysis. While likelihood-based inference is attractive both in terms of limiting properties and of implementation, its application in random-effects meta-analysis may result in misleading conclusions, especially when the number of studies is small to moderate. The current paper shows how methodology that reduces the asymptotic bias of the maximum likelihood estimator of the variance component can also substantially improve inference about the mean effect size. The results are derived for the more general framework of random-effects meta-regression, which allows the mean effect size to vary with study-specific covariates.

scholarly journals Cortical Bone Trajectory Screw Fixation of Lumbar Spine in Adult Patients with Degenerative and Traumatic Spine Disorders

QJM ◽  
2020 ◽  
Vol 113 (Supplement_1) ◽  
Author(s):  
A M Elbadrawy ◽  
T M S Elkhateeb ◽  
E M M A Hassan
Keyword(s):  
Lumbar Spine ◽  
Random Effects ◽  
Screw Fixation ◽  
Meta Analysis ◽  
Pedicle Screws ◽  
Fusion Rate ◽  
Random Effects Models ◽  
Fixed And Random Effects ◽  
Spine Disorders ◽  
Lumbar Pedicle

Abstract Background the CBT for placing lumbar pedicle screws is a technique used to improve fixation during instrumented fusion of the lumbar spine. In comparison with traditional trajectory (TT) for pedicle screws, CBT screws (otherwise known as pars screws or cortical screws) have a more medial starting point and are aimed in a medial to lateral, caudal to cranial direction. First reported in 2009 as a method to increase the purchase of lumbar pedicle screws within bone.(1) Aim of the Work to perform a systematic review and meta-analysis to determine whether traditional Pedicles Screw Fixation (PS Fx) or Cortical Bone Trajectory Screw Fixation (CBT Fx); has been successful for the treatment and fixation of lumbar spine in adult patients with degenerative and traumatic spine disorders; and to compare the 2 techniques to identify risk factor for unfavorable outcome through the recent researches about that issue. Methodology this review was done using standard methodology outlined in the Cochrane Handbook and reported the findings in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement guidelines. Results meta-analysis study showed that; Successful fusion rate in fixed and random-effects models were (92.24% respectively); in SP group.Successful fusion rate in fixed and random-effects models were (92.44% respectively); in CBT group.Fixed and random-effects models showed non-significant difference in successful fusion rate; between the 2 groups of studies (p > 0.05). We calculated safety for each technique through post-operative (failed fusion rate). Conclusion Although there were insignificant p-values in the most of the comparative items but the CBT showed lower average of intraoperative blood loss, operation time and higher average of decrease in VAS & increase in ODI, slightly higher fusion rate in comparison with PS. So we recommend, the use of CBT as an acceptable alternative of PS in lumbar spine fixation.

scholarly journals Meta-Analysis of Sensorimotor Gating Deficits in Patients With Schizophrenia Evaluated by Prepulse Inhibition Test

2020 ◽  
Vol 46 (6) ◽  
pp. 1482-1497
Author(s):  
Rodrigo San-Martin ◽  
Leonardo Andrade Castro ◽  
Paulo Rossi Menezes ◽  
Francisco José Fraga ◽  
Priscyla Waleska Simões ◽  
...  
Keyword(s):  
Prepulse Inhibition ◽  
Random Effects ◽  
Effect Size ◽  
Meta Analysis ◽  
Sensorimotor Gating ◽  
Control Group ◽  
Random Effects Models ◽  
Meta Regression Analysis ◽  
High Level ◽  
Meta Analyses

Abstracts Prepulse inhibition (PPI) of startle is an operational measure of sensorimotor gating that is often impaired in patients with schizophrenia. Despite the large number of studies, there is considerable variation in PPI outcomes reported. We conducted a systematic review and meta-analysis investigating PPI impairment in patients with schizophrenia compared with healthy control subjects, and examined possible explanations for the variation in results between studies. Major databases were screened for observational studies comparing healthy subjects and patients with schizophrenia for the prepulse and pulse intervals of 60 and 120 ms as primary outcomes, ie, PPI-60 and PPI-120. Standardized mean difference (SMD) and 95% confidence intervals (CI) were extracted and pooled using random effects models. We then estimated the mean effect size of these measures with random effects meta-analyses and evaluated potential PPI heterogeneity moderators, using sensitivity analysis and meta-regressions. Sixty-seven primary studies were identified, with 3685 healthy and 4290 patients with schizophrenia. The schizophrenia group showed reduction in sensorimotor gating for both PPI-60 (SMD = −0.50, 95% CI = [−0.61, −0.39]) and PPI-120 (SMD = −0.44, 95% CI = [−0.54, −0.33]). The sensitivity and meta-regression analysis showed that sample size, gender proportion, imbalance for gender, source of control group, and study continent were sources of heterogeneity (P < .05) for both PPI-60 and PPI-120 outcomes. Our findings confirm a global sensorimotor gating deficit in schizophrenia patients, with overall moderate effect size for PPI-60 and PPI-120. Methodological consistency should decrease the high level of heterogeneity of PPI results between studies.

scholarly journals Combined Effects of Aerobic Exercise and Diet on Lipids and Lipoproteins in Overweight and Obese Adults: A Meta-Analysis

2012 ◽  
Vol 2012 ◽  
pp. 1-16 ◽  
Author(s):  
George A. Kelley ◽  
Kristi S. Kelley ◽  
Susan Roberts ◽  
William Haskell
Keyword(s):  
Aerobic Exercise ◽  
Random Effects ◽  
Effect Size ◽  
Meta Analysis ◽  
Combined Effects ◽  
Obese Adults ◽  
Men And Women ◽  
Random Effects Models ◽  
Multiple Groups ◽  
Lipids And Lipoproteins

This study used the aggregate data meta-analytic approach to determine the combined effects of aerobic exercise and diet on lipids and lipoproteins in overweight and obese adults. Twelve studies representing 859 men and women (443 intervention, 416 control) were included. Using random-effects models, statistically significant, intervention minus control reductions were found for TC (−12.8 mg/dL, 95% CI, −19.9 to −5.7), TC : HDL-C (−0.5 mg/dL, 95% CI, −0.8 to −0.1), LDL-C (−6.8 mg/dL, 95% CI, −11.8 to −1.8), and TG (−13.1 mg/dL, 95% CI, −21.2 to −5.0) but not HDL-C (−0.4 mg/dL, 95% CI, −2.3 to 1.6). Results remained robust when adjusted for publication bias, deleting each study from the model once, and collapsing results for multiple groups from the same study into one effect size. These findings suggest that concurrent aerobic exercise and diet improve TC, LDL-C, TC : HDL-C, and TG, but not HDL-C, in overweight and obese adults.

scholarly journals P-uniform*

2018 ◽  
Author(s):  
Robbie Cornelis Maria van Aert ◽  
Marcel A. L. M. van Assen
Keyword(s):  
Publication Bias ◽  
Random Effects ◽  
Effect Size ◽  
Meta Analysis ◽  
Statistical Properties ◽  
Selection Model ◽  
Effect Sizes ◽  
Random Effects Model ◽  
Average Effect ◽  
Model Approach

Publication bias is a major threat to the validity of a meta-analysis resulting in overestimated effect sizes. P-uniform is a meta-analysis method that corrects estimates for publication bias but overestimates average effect size if heterogeneity in true effect sizes (i.e., between-study variance) is present. We propose an extension and improvement of p-uniform called p-uniform*. P-uniform* improves upon p-uniform in three important ways, as it (i) entails a more efficient estimator, (ii) eliminates the overestimation of effect size in case of between-study variance in true effect sizes, and (iii) enables estimating and testing for the presence of the between-study variance. We compared the statistical properties of p-uniform* with p-uniform, the selection model approach of Hedges (1992), and the random-effects model. Statistical properties of p-uniform* and the selection model approach were comparable and generally outperformed p-uniform and the random-effects model if publication bias was present. We demonstrate that p-uniform* and the selection model approach estimate average effect size and between-study variance rather well with ten or more studies in the meta-analysis when publication bias is not extreme. P-uniform* generally provides more accurate estimates of the between-study variance in meta-analyses containing many studies (e.g., 60 or more) and if publication bias is present. However, both methods do not perform well if the meta-analysis only includes statistically significant studies. P-uniform performed best in this case but only when between-study variance was zero or small. We offer recommendations for applied researchers, and provide an R package and an easy-to-use web application for applying p-uniform*.

Sign in / Sign up

Export Citation Format

Share Document