Measuring Inconsistency in Meta-Analyses

MRC Biostatistics Unit, Institute of Public Health, Cambridge CB2 2SR.
BMJ (online) (Impact Factor: 16.38). 10/2003; 327(7414):557-60. DOI: 10.1136/bmj.327.7414.557
Source: PubMed
  • Source
    • "In the case of substantial heterogeneity (I 2 > 50%), pooled OR was calculated using a random-effects model (REM). Otherwise, the fixed-effects model (FEM) was used when the inverse variance was I 2 less than 50% (Higgins et al., 2003). Participants were grouped by races (Asians and Europeans/Americans) and types of mental disorders (mood disorder, alcohol dependence, and schizophrenia disorder) to observe the differences among races and mental disorders. "
    • "Pooled proportions were calculated by a random effects model using metaprop, a statistical procedure for meta-analysis of binomial data [26]. The statistical heterogeneity was assessed using the I 2 statistic, which measures the variation across studies that is due to inter-study heterogeneity rather than chance [27]. Relative rates (proportion self /proportion control ) and absolute differences (proportion self –proportion control ) were assessed by applying random effects models using metan [28] [29]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Population coverage for cervical cancer screening is an important determinant explaining differences in the incidence of cervical cancer between countries. Offering devices for self-sampling has the potential to increase participation of hard-to-reach women. A systematic review and meta-analysis were performed to evaluate the participation after an invitation including a self-sampling device (self-sampling arm) versus an invitation to have a sample taken by a health professional (control arm), sent to under-screened women. Sixteen randomised studies were found eligible. In an intention-to-treat analysis, the pooled participation in the self-sampling arm was 23.6% (95% confidence interval (CI)=20.2-27.3%), when self-sampling kits were sent by mail to all women, versus 10.3% (95% CI=6.2-15.2%) in the control arm (participation difference: 12.6% [95% CI=9.3-15.9]). When women had to opt-in to receive the self-sampling device, as used in three studies, the pooled participation was not higher in the self-sampling compared to the control arm (participation difference: 0.2% [95% CI=-4.5-4.9%]). An increased participation was observed in the self-sampling arm compared to the control arm, if self-sampling kits were sent directly to women at their home address. However, the size of the effect varied substantially among studies. Since participation was similar in both arms when women had to opt-in, future studies are warranted to discern opt-in scenarios that are most acceptable to women. Copyright © 2015 Elsevier Ltd. All rights reserved.
    European Journal of Cancer 08/2015; DOI:10.1016/j.ejca.2015.07.006 · 4.82 Impact Factor
  • Source
    • "Heterogeneity among the effect sizes was assessed using the Q statistic and the I 2 index (Borenstein et al., 2009), where I 2 ≤ 25% indicates little heterogeneity, 50% moderate heterogeneity, and ≥ 75% high heterogeneity due to real differences among studies (Higgins et al., 2003). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background: Deep transcranial magnetic stimulation (DTMS) is a relatively new, non-invasive method of stimulating larger and, presumably, deeper brain regions. The current study investigated if DTMS delivered with H-coils has acute antidepressant effects in major depression using a systematic literature review and a quantitative meta-analysis. Methods: Seventeen studies on ‘DTMS or H-coil’ and ‘depression’ were identified on Medline, PsycInfo, and Google Scholar (until November 2014). Data from nine open-label studies were meta-analysed using a random-effects model with inverse-variance weights. The outcome measures were the standardised paired mean difference (Cohen’s d) in depression scores on Hamilton Depression Rating Scale (HDRS), response, remission, and dropout rates after acute DTMS treatment compared to baseline. Results: There was a large antidepressant effect after 20 acute, high-frequency DTMS sessions compared to baseline according to HDRS change scores (overall mean weighted d=2.04, 95% confidence interval: 1.53-2.55; nine studies; 150 patients). Overall weighted response, remission, and dropout rates were 60%, 29%, and 18% respectively. HDRS change scores and response rates tended to be higher in four studies with 68 patients on concurrent antidepressants compared to two studies with 26 patients who received DTMS as a monotherapy. Limitations: These results are based on data from a low number of open-label studies. Conclusion: High-frequency DTMS appears to have acute antidepressant effects after 20 sessions in mostly unipolar and treatment-resistant patients. Concurrent treatment with antidepressants might enhance the efficacy of DTMS.
    Journal of Affective Disorders 08/2015; DOI:10.1016/j.jad.2015.08.033 · 3.71 Impact Factor
Show more

Questions & Answers about this publication