Extreme between-study homogeneity in meta-analyses could offer useful insights

Investigating the Heterogeneity of "Study Twins"

Meta-analyses are commonly performed based on random-effects models, while in certain cases one might also argue in favor of a common-effect model. One such case may be given by the example of two "study twins" that are performed according to a common (or at least very similar) protocol. Here we investigate the particular case of meta-analysis of a pair of studies, for example, summarizing the results of two confirmatory clinical trials in phase III of a clinical development program. Thereby, we focus on the question of to what extent homogeneity or heterogeneity may be discernible and include an empirical investigation of published ("twin") pairs of studies. A pair of estimates from two studies only provide very little evidence of homogeneity or heterogeneity of effects, and ad hoc decision criteria may often be misleading.

Identification of application and interpretation errors that can occur in pairwise meta-analyses in systematic reviews of interventions: a systematic review

OBJECTIVES

To generate a bank of items describing application and interpretation errors that can arise in pairwise meta-analyses in systematic reviews of interventions.

STUDY DESIGN AND SETTING

MEDLINE, Embase, and Scopus were searched to identify studies describing types of errors in meta-analyses. Descriptions of errors and supporting quotes were extracted by multiple authors. Errors were reviewed at team meetings to determine if they should be excluded, reworded, or combined with other errors, and were categorized into broad categories of errors and subcategories within.

RESULTS

Fifty articles met our inclusion criteria, leading to the identification of 139 errors. We identified 25 errors covering data extraction/manipulation, 74 covering statistical analyses, and 40 covering interpretation. Many of the statistical analysis errors related to the meta-analysis model (eg, using a two-stage strategy to determine whether to select a fixed or random-effects model) and statistical heterogeneity (eg, not undertaking an assessment for statistical heterogeneity).

CONCLUSION

We generated a comprehensive bank of possible errors that can arise in the application and interpretation of meta-analyses in systematic reviews of interventions. This item bank of errors provides the foundation for developing a checklist to help peer reviewers detect statistical errors.

Can Kinesio taping improve discomfort after mandibular third molar surgery? A systematic review and meta-analysis

OBJECTIVE

The purpose of this study was to evaluate whether Kinesio taping (KT) can improve patient discomfort after mandibular third molar surgery.

MATERIALS AND METHODS

This systematic review and meta-analysis was conducted according to the PICO strategy. We searched 4 databases for related articles. All controlled trials or randomized controlled trials that evaluated the application of KT after mandibular third molar surgery were included. Screening and article selection were carried out by two independent reviewers. The main evaluation indicators were facial swelling, pain, and trismus. All statistical analyses were conducted using Review Manager 5.3 software.

RESULT

This analysis included 8 articles. The combined results showed that compared with the control group, the postoperative application of KT significantly reduced pain in the early (early stage mean difference (MD), - 2.00; 95% confidence interval (CI), - 2.40 to - 1.60; P < 0.00001) and late (late stage MD, - 1.18; 95% CI, - 2.26 to - 0.11; P = 0.03) postoperative periods and, thus, reduced the intake of painkillers. KT also reduced facial swelling in the early and late postoperative periods (early stage standardized mean difference (SMD), - 1.34; 95% CI, - 1.99 to - 0.68; P < 0.0001; late stage SMD, - 0.31; 95% CI, - 0.51 to - 0.11; P = 0.002). In addition, the postoperative application of KT improved restricted mouth opening in the early and late postoperative periods (early stage MD, - 5.03 mm; 95% CI, - 6.32 to - 3.74 mm; P < 0.00001; late stage MD, - 3.42 mm; 95% CI, - 5.31 to - 1.52 mm; P = 0.0004).

CONCLUSION

KT can significantly reduce postoperative pain, swelling, and trismus after impacted mandibular tooth extraction. Additional high-quality and rigorously designed randomized controlled trials should be conducted to verify these conclusions.

CLINICAL RELEVANCE

KT is a low-cost, simple, effective nondrug therapy for the postoperative management of mandibular third molar extraction and has broad prospects for clinical application.

Bayesian hypothesis testing and estimation under the marginalized random-effects meta-analysis model

Meta-analysis methods are used to synthesize results of multiple studies on the same topic. The most frequently used statistical model in meta-analysis is the random-effects model containing parameters for the overall effect, between-study variance in primary study’s true effect size, and random effects for the study-specific effects. We propose Bayesian hypothesis testing and estimation methods using the marginalized random-effects meta-analysis (MAREMA) model where the study-specific true effects are regarded as nuisance parameters which are integrated out of the model. We propose using a flat prior distribution on the overall effect size in case of estimation and a proper unit information prior for the overall effect size in case of hypothesis testing. For the between-study variance (which can attain negative values under the MAREMA model), a proper uniform prior is placed on the proportion of total variance that can be attributed to between-study variability. Bayes factors are used for hypothesis testing that allow testing point and one-sided hypotheses. The proposed methodology has several attractive properties. First, the proposed MAREMA model encompasses models with a zero, negative, and positive between-study variance, which enables testing a zero between-study variance as it is not a boundary problem. Second, the methodology is suitable for default Bayesian meta-analyses as it requires no prior information about the unknown parameters. Third, the proposed Bayes factors can even be used in the extreme case when only two studies are available because Bayes factors are not based on large sample theory. We illustrate the developed methods by applying it to two meta-analyses and introduce easy-to-use software in the R package BFpack to compute the proposed Bayes factors.

Publication bias examined in meta-analyses from psychology and medicine: A meta-meta-analysis

Publication bias is a substantial problem for the credibility of research in general and of meta-analyses in particular, as it yields overestimated effects and may suggest the existence of non-existing effects. Although there is consensus that publication bias exists, how strongly it affects different scientific literatures is currently less well-known. We examined evidence of publication bias in a large-scale data set of primary studies that were included in 83 meta-analyses published in Psychological Bulletin (representing meta-analyses from psychology) and 499 systematic reviews from the Cochrane Database of Systematic Reviews (CDSR; representing meta-analyses from medicine). Publication bias was assessed on all homogeneous subsets (3.8% of all subsets of meta-analyses published in Psychological Bulletin) of primary studies included in meta-analyses, because publication bias methods do not have good statistical properties if the true effect size is heterogeneous. Publication bias tests did not reveal evidence for bias in the homogeneous subsets. Overestimation was minimal but statistically significant, providing evidence of publication bias that appeared to be similar in both fields. However, a Monte-Carlo simulation study revealed that the creation of homogeneous subsets resulted in challenging conditions for publication bias methods since the number of effect sizes in a subset was rather small (median number of effect sizes equaled 6). Our findings are in line with, in its most extreme case, publication bias ranging from no bias until only 5% statistically nonsignificant effect sizes being published. These and other findings, in combination with the small percentages of statistically significant primary effect sizes (28.9% and 18.9% for subsets published in Psychological Bulletin and CDSR), led to the conclusion that evidence for publication bias in the studied homogeneous subsets is weak, but suggestive of mild publication bias in both psychology and medicine.

Conducting Meta-Analyses Based on p Values: Reservations and Recommendations for Applying p-Uniform and p-Curve

Because of overwhelming evidence of publication bias in psychology, techniques to correct meta-analytic estimates for such bias are greatly needed. The methodology on which the p-uniform and p-curve methods are based has great promise for providing accurate meta-analytic estimates in the presence of publication bias. However, in this article, we show that in some situations, p-curve behaves erratically, whereas p-uniform may yield implausible estimates of negative effect size. Moreover, we show that (and explain why) p-curve and p-uniform result in overestimation of effect size under moderate-to-large heterogeneity and may yield unpredictable bias when researchers employ p-hacking. We offer hands-on recommendations on applying and interpreting results of meta-analyses in general and p-uniform and p-curve in particular. Both methods as well as traditional methods are applied to a meta-analysis on the effect of weight on judgments of importance. We offer guidance for applying p-uniform or p-curve using R and a user-friendly web application for applying p-uniform.

Confidence intervals for the between-study variance in random-effects meta-analysis using generalised heterogeneity statistics: should we use unequal tails?

BACKGROUND

Confidence intervals for the between study variance are useful in random-effects meta-analyses because they quantify the uncertainty in the corresponding point estimates. Methods for calculating these confidence intervals have been developed that are based on inverting hypothesis tests using generalised heterogeneity statistics. Whilst, under the random effects model, these new methods furnish confidence intervals with the correct coverage, the resulting intervals are usually very wide, making them uninformative.

METHODS

We discuss a simple strategy for obtaining 95 % confidence intervals for the between-study variance with a markedly reduced width, whilst retaining the nominal coverage probability. Specifically, we consider the possibility of using methods based on generalised heterogeneity statistics with unequal tail probabilities, where the tail probability used to compute the upper bound is greater than 2.5 %. This idea is assessed using four real examples and a variety of simulation studies. Supporting analytical results are also obtained.

RESULTS

Our results provide evidence that using unequal tail probabilities can result in shorter 95 % confidence intervals for the between-study variance. We also show some further results for a real example that illustrates how shorter confidence intervals for the between-study variance can be useful when performing sensitivity analyses for the average effect, which is usually the parameter of primary interest.

CONCLUSIONS

We conclude that using unequal tail probabilities when computing 95 % confidence intervals for the between-study variance, when using methods based on generalised heterogeneity statistics, can result in shorter confidence intervals. We suggest that those who find the case for using unequal tail probabilities convincing should use the '1-4 % split', where greater tail probability is allocated to the upper confidence bound. The 'width-optimal' interval that we present deserves further investigation.

Ethical publishing in intensive care medicine: A narrative review

Ethical standards in the context of scientific publications are increasingly gaining attention. A narrative review of the literature concerning publication ethics was conducted as found in PubMed, Google Scholar, relevant news articles, position papers, websites and other sources. The Committee on Publication Ethics has produced guidelines and schedules for the handling of problem situations that have been adopted by professional journals and publishers worldwide as guidelines to authors. The defined requirements go beyond the disclosure of conflicts of interest or the prior registration of clinical trials. Recommendations to authors, editors and publishers of journals and research institutions were formulated with regard to issues of authorship, double publications, plagiarism, and conflicts of interest, with special attention being paid to unethical research behavior and data falsification. This narrative review focusses on ethical publishing in intensive care medicine. As scientific misconduct with data falsification damage patients and society, especially if fraudulent studies are considered important or favor certain therapies and downplay their side effects, it is important to ensure that only studies are published that have been carried out with highest integrity according to predefined criteria. For that also the peer review process has to be conducted in accordance with the highest possible scientific standards and making use of available modern information technology. The review provides the current state of recommendations that are considered to be most relevant particularly in the field of intensive care medicine.

Citrus Fruit Intake Substantially Reduces the Risk of Esophageal Cancer: A Meta-Analysis of Epidemiologic Studies

Many epidemiologic studies indicate a potential association between fruit and vegetable intake and various cancers. The purpose of this meta-analysis is to investigate the association between citrus fruit intake and esophageal cancer risk. The authors conducted a comprehensive search on PubMed, EMBASE, and the Cochrane Library from inception until July 2014. Studies presenting information about citrus intake and esophageal cancer were analyzed. The authors extracted the categories of citrus intake, study-specific odds ratio or relative risk, and the P value and associated 95% confidence intervals for the highest versus lowest dietary intake of citrus fruit level. The association was quantified using meta-analysis of standard errors with a random-effects model. Thirteen case-control studies and 6 cohort studies were eligible for inclusion. Citrus intake may significantly reduce risk of esophageal cancer (summary odds ratio = 0.63; 95% confidence interval = 0.52-0.75; P = 0), without notable publication bias (intercept = -0.79, P = 0.288) and with significant heterogeneity across studies (I = 52%). The results from epidemiologic studies suggest an inverse association between citrus fruit intake and esophageal cancer risk. The significant effect is consistent between case-control and cohort studies. Larger prospective studies with rigorous methodology should be considered to validate the association between citrus fruits and esophageal cancer.

Beautiful small: Misleading large randomized controlled trials? The example of colloids for volume resuscitation

In anesthesia and intensive care, treatment benefits that were claimed on the basis of small or modest-sized trials have repeatedly failed to be confirmed in large randomized controlled trials. A well-designed small trial in a homogeneous patient population with high event rates could yield conclusive results; however, patient populations in anesthesia and intensive care are typically heterogeneous because of comorbidities. The size of the anticipated effects of therapeutic interventions is generally low in relation to relevant endpoints. For regulatory purposes, trials are required to demonstrate efficacy in clinically important endpoints, and therefore must be large because clinically important study endpoints such as death, sepsis, or pneumonia are dichotomous and infrequently occur. The rarer endpoint events occur in the study population; that is, the lower the signal-to-noise ratio, the larger the trials must be to prevent random events from being overemphasized. In addition to trial design, sample size determination on the basis of event rates, clinically meaningful risk ratio reductions and actual patient numbers studied are among the most important characteristics when interpreting study results. Trial size is a critical determinant of generalizability of study results to larger or general patient populations. Typical characteristics of small single-center studies responsible for their known fragility include low variability of outcome measures for surrogate parameters and selective publication and reporting. For anesthesiology and intensive care medicine, findings in volume resuscitation research on intravenous infusion of colloids exemplify this, since both the safety of albumin infusion and the adverse effects of the artificial colloid hydroxyethyl starch have been confirmed only in large-sized trials.

The effects of high perioperative inspiratory oxygen fraction for adult surgical patients

BACKGROUND

Available evidence on the effects of a high fraction of inspired oxygen (FIO2) of 60% to 90% compared with a routine fraction of inspired oxygen of 30% to 40%, during anaesthesia and surgery, on mortality and surgical site infection has been inconclusive. Previous trials and meta-analyses have led to different conclusions on whether a high fraction of supplemental inspired oxygen during anaesthesia may decrease or increase mortality and surgical site infections in surgical patients.

OBJECTIVES

To assess the benefits and harms of an FIO2 equal to or greater than 60% compared with a control FIO2 at or below 40% in the perioperative setting in terms of mortality, surgical site infection, respiratory insufficiency, serious adverse events and length of stay during the index admission for adult surgical patients.We looked at various outcomes, conducted subgroup and sensitivity analyses, examined the role of bias and applied trial sequential analysis (TSA) to examine the level of evidence supporting or refuting a high FIO2 during surgery, anaesthesia and recovery.

SEARCH METHODS

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, BIOSIS, International Web of Science, the Latin American and Caribbean Health Science Information Database (LILACS), advanced Google and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) up to February 2014. We checked the references of included trials and reviews for unidentified relevant trials and reran the searches in March 2015. We will consider two studies of interest when we update the review.

SELECTION CRITERIA

We included randomized clinical trials that compared a high fraction of inspired oxygen with a routine fraction of inspired oxygen during anaesthesia, surgery and recovery in individuals 18 years of age or older.

DATA COLLECTION AND ANALYSIS

Two review authors extracted data independently. We conducted random-effects and fixed-effect meta-analyses, and for dichotomous outcomes, we calculated risk ratios (RRs). We used published data and data obtained by contacting trial authors.To minimize the risk of systematic error, we assessed the risk of bias of the included trials. To reduce the risk of random errors caused by sparse data and repetitive updating of cumulative meta-analyses, we applied trial sequential analyses. We used Grades of Recommendation, Assessment, Development and Evaluation (GRADE) to assess the quality of the evidence.

MAIN RESULTS

We included 28 randomized clinical trials (9330 participants); in the 21 trials reporting relevant outcomes for this review, 7597 participants were randomly assigned to a high fraction of inspired oxygen versus a routine fraction of inspired oxygen.In trials with an overall low risk of bias, a high fraction of inspired oxygen compared with a routine fraction of inspired oxygen was not associated with all-cause mortality (random-effects model: RR 1.12, 95% confidence interval (CI) 0.93 to 1.36; GRADE: low quality) within the longest follow-up and within 30 days of follow-up (Peto odds ratio (OR) 0.99, 95% CI 0.61 to 1.60; GRADE: low quality). In a trial sequential analysis, the required information size was not reached and the analysis could not refute a 20% increase in mortality. Similarly, when all trials were included, a high fraction of inspired oxygen was not associated with all-cause mortality to the longest follow-up (RR 1.07, 95% CI 0.87 to 1.33) or within 30 days of follow-up (Peto OR 0.83, 95% CI 0.54 to 1.29), both of very low quality according to GRADE. Neither was a high fraction of inspired oxygen associated with the risk of surgical site infection in trials with low risk of bias (RR 0.86, 95% CI 0.63 to 1.17; GRADE: low quality) or in all trials (RR 0.87, 95% CI 0.71 to 1.07; GRADE: low quality). A high fraction of inspired oxygen was not associated with respiratory insufficiency (RR 1.25, 95% CI 0.79 to 1.99), serious adverse events (RR 0.96, 95% CI 0.65 to 1.43) or length of stay (mean difference -0.06 days, 95% CI -0.44 to 0.32 days).In subgroup analyses of nine trials using preoperative antibiotics, a high fraction of inspired oxygen was associated with a decrease in surgical site infections (RR 0.76, 95% CI 0.60 to 0.97; GRADE: very low quality); a similar effect was noted in the five trials adequately blinded for the outcome assessment (RR 0.79, 95% CI 0.66 to 0.96; GRADE: very low quality). We did not observe an effect of a high fraction of inspired oxygen on surgical site infections in any other subgroup analyses.

AUTHORS' CONCLUSIONS

As the risk of adverse events, including mortality, may be increased by a fraction of inspired oxygen of 60% or higher, and as robust evidence is lacking for a beneficial effect of a fraction of inspired oxygen of 60% or higher on surgical site infection, our overall results suggest that evidence is insufficient to support the routine use of a high fraction of inspired oxygen during anaesthesia and surgery. Given the risk of attrition and outcome reporting bias, as well as other weaknesses in the available evidence, further randomized clinical trials with low risk of bias in all bias domains, including a large sample size and long-term follow-up, are warranted.

HELOW: a program for testing extreme homogeneity in meta-analysis

Meta-analysis aims to synthesize results from different studies. Although, in a meta-analysis the presence of large between-study heterogeneity is routinely evaluated, in some instances is also important to probe whether there is extreme between-study homogeneity (i.e. extreme low between-study heterogeneity). HELOW (HEterogeneity LOW) is a program for testing extreme homogeneity in a meta-analysis of risk ratios when binary outcome and Mantel-Haenszel fixed effects summary risk ratio estimate are employed. The significance of extreme homogeneity is assessed using a Monte Carlo test. Extreme homogeneity may yield insights for the statistical and clinical interpretation of the data.

Vitamin D supplementation for prevention of cancer in adults

BACKGROUND

The evidence on whether vitamin D supplementation is effective in decreasing cancers is contradictory.

OBJECTIVES

To assess the beneficial and harmful effects of vitamin D supplementation for prevention of cancer in adults.

SEARCH METHODS

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, LILACS, Science Citation Index Expanded, and the Conference Proceedings Citation Index-Science to February 2014. We scanned bibliographies of relevant publications and asked experts and pharmaceutical companies for additional trials.

SELECTION CRITERIA

We included randomised trials that compared vitamin D at any dose, duration, and route of administration versus placebo or no intervention in adults who were healthy or were recruited among the general population, or diagnosed with a specific disease. Vitamin D could have been administered as supplemental vitamin D (vitamin D₃ (cholecalciferol) or vitamin D₂ (ergocalciferol)), or an active form of vitamin D (1α-hydroxyvitamin D (alfacalcidol), or 1,25-dihydroxyvitamin D (calcitriol)).

DATA COLLECTION AND ANALYSIS

Two review authors extracted data independently. We conducted random-effects and fixed-effect model meta-analyses. For dichotomous outcomes, we calculated the risk ratios (RRs). We considered risk of bias in order to assess the risk of systematic errors. We conducted trial sequential analyses to assess the risk of random errors.

MAIN RESULTS

Eighteen randomised trials with 50,623 participants provided data for the analyses. All trials came from high-income countries. Most of the trials had a high risk of bias, mainly for-profit bias. Most trials included elderly community-dwelling women (aged 47 to 97 years). Vitamin D was administered for a weighted mean of six years. Fourteen trials tested vitamin D₃, one trial tested vitamin D₂, and three trials tested calcitriol supplementation. Cancer occurrence was observed in 1927/25,275 (7.6%) recipients of vitamin D versus 1943/25,348 (7.7%) recipients of control interventions (RR 1.00 (95% confidence interval (CI) 0.94 to 1.06); P = 0.88; I² = 0%; 18 trials; 50,623 participants; moderate quality evidence according to the GRADE instrument). Trial sequential analysis (TSA) of the 18 vitamin D trials shows that the futility area is reached after the 10th trial, allowing us to conclude that a possible intervention effect, if any, is lower than a 5% relative risk reduction. We did not observe substantial differences in the effect of vitamin D on cancer in subgroup analyses of trials at low risk of bias compared to trials at high risk of bias; of trials with no risk of for-profit bias compared to trials with risk of for-profit bias; of trials assessing primary prevention compared to trials assessing secondary prevention; of trials including participants with vitamin D levels below 20 ng/mL at entry compared to trials including participants with vitamin D levels of 20 ng/mL or more at entry; or of trials using concomitant calcium supplementation compared to trials without calcium. Vitamin D decreased all-cause mortality (1854/24,846 (7.5%) versus 2007/25,020 (8.0%); RR 0.93 (95% CI 0.88 to 0.98); P = 0.009; I² = 0%; 15 trials; 49,866 participants; moderate quality evidence), but TSA indicates that this finding could be due to random errors. Cancer occurrence was observed in 1918/24,908 (7.7%) recipients of vitamin D₃ versus 1933/24,983 (7.7%) in recipients of control interventions (RR 1.00 (95% CI 0.94 to 1.06); P = 0.88; I² = 0%; 14 trials; 49,891 participants; moderate quality evidence). TSA of the vitamin D₃ trials shows that the futility area is reached after the 10th trial, allowing us to conclude that a possible intervention effect, if any, is lower than a 5% relative risk reduction. Vitamin D₃ decreased cancer mortality (558/22,286 (2.5%) versus 634/22,206 (2.8%); RR 0.88 (95% CI 0.78 to 0.98); P = 0.02; I² = 0%; 4 trials; 44,492 participants; low quality evidence), but TSA indicates that this finding could be due to random errors. Vitamin D₃ combined with calcium increased nephrolithiasis (RR 1.17 (95% CI 1.03 to 1.34); P = 0.02; I² = 0%; 3 trials; 42,753 participants; moderate quality evidence). TSA, however, indicates that this finding could be due to random errors. We did not find any data on health-related quality of life or health economics in the randomised trials included in this review.

AUTHORS' CONCLUSIONS

There is currently no firm evidence that vitamin D supplementation decreases or increases cancer occurrence in predominantly elderly community-dwelling women. Vitamin D₃ supplementation decreased cancer mortality and vitamin D supplementation decreased all-cause mortality, but these estimates are at risk of type I errors due to the fact that too few participants were examined, and to risks of attrition bias originating from substantial dropout of participants. Combined vitamin D₃ and calcium supplements increased nephrolithiasis, whereas it remains unclear from the included trials whether vitamin D₃, calcium, or both were responsible for this effect. We need more trials on vitamin D supplementation, assessing the benefits and harms among younger participants, men, and people with low vitamin D status, and assessing longer duration of treatments as well as higher dosages of vitamin D. Follow-up of all participants is necessary to reduce attrition bias.

Vitamin D supplementation for prevention of mortality in adults

BACKGROUND

Available evidence on the effects of vitamin D on mortality has been inconclusive. In a recent systematic review, we found evidence that vitamin D3 may decrease mortality in mostly elderly women. The present systematic review updates and reassesses the benefits and harms of vitamin D supplementation used in primary and secondary prophylaxis of mortality.

OBJECTIVES

To assess the beneficial and harmful effects of vitamin D supplementation for prevention of mortality in healthy adults and adults in a stable phase of disease.

SEARCH METHODS

We searched The Cochrane Library, MEDLINE, EMBASE, LILACS, the Science Citation Index-Expanded and Conference Proceedings Citation Index-Science (all up to February 2012). We checked references of included trials and pharmaceutical companies for unidentified relevant trials.

SELECTION CRITERIA

Randomised trials that compared any type of vitamin D in any dose with any duration and route of administration versus placebo or no intervention in adult participants. Participants could have been recruited from the general population or from patients diagnosed with a disease in a stable phase. Vitamin D could have been administered as supplemental vitamin D (vitamin D3 (cholecalciferol) or vitamin D2 (ergocalciferol)) or as an active form of vitamin D (1α-hydroxyvitamin D (alfacalcidol) or 1,25-dihydroxyvitamin D (calcitriol)).

DATA COLLECTION AND ANALYSIS

Six review authors extracted data independently. Random-effects and fixed-effect meta-analyses were conducted. For dichotomous outcomes, we calculated the risk ratios (RRs). To account for trials with zero events, we performed meta-analyses of dichotomous data using risk differences (RDs) and empirical continuity corrections. We used published data and data obtained by contacting trial authors.To minimise the risk of systematic error, we assessed the risk of bias of the included trials. Trial sequential analyses controlled the risk of random errors possibly caused by cumulative meta-analyses.

MAIN RESULTS

We identified 159 randomised clinical trials. Ninety-four trials reported no mortality, and nine trials reported mortality but did not report in which intervention group the mortality occurred. Accordingly, 56 randomised trials with 95,286 participants provided usable data on mortality. The age of participants ranged from 18 to 107 years. Most trials included women older than 70 years. The mean proportion of women was 77%. Forty-eight of the trials randomly assigned 94,491 healthy participants. Of these, four trials included healthy volunteers, nine trials included postmenopausal women and 35 trials included older people living on their own or in institutional care. The remaining eight trials randomly assigned 795 participants with neurological, cardiovascular, respiratory or rheumatoid diseases. Vitamin D was administered for a weighted mean of 4.4 years. More than half of the trials had a low risk of bias. All trials were conducted in high-income countries. Forty-five trials (80%) reported the baseline vitamin D status of participants based on serum 25-hydroxyvitamin D levels. Participants in 19 trials had vitamin D adequacy (at or above 20 ng/mL). Participants in the remaining 26 trials had vitamin D insufficiency (less than 20 ng/mL).Vitamin D decreased mortality in all 56 trials analysed together (5,920/47,472 (12.5%) vs 6,077/47,814 (12.7%); RR 0.97 (95% confidence interval (CI) 0.94 to 0.99); P = 0.02; I(2) = 0%). More than 8% of participants dropped out. 'Worst-best case' and 'best-worst case' scenario analyses demonstrated that vitamin D could be associated with a dramatic increase or decrease in mortality. When different forms of vitamin D were assessed in separate analyses, only vitamin D3 decreased mortality (4,153/37,817 (11.0%) vs 4,340/38,110 (11.4%); RR 0.94 (95% CI 0.91 to 0.98); P = 0.002; I(2) = 0%; 75,927 participants; 38 trials). Vitamin D2, alfacalcidol and calcitriol did not significantly affect mortality. A subgroup analysis of trials at high risk of bias suggested that vitamin D2 may even increase mortality, but this finding could be due to random errors. Trial sequential analysis supported our finding regarding vitamin D3, with the cumulative Z-score breaking the trial sequential monitoring boundary for benefit, corresponding to 150 people treated over five years to prevent one additional death. We did not observe any statistically significant differences in the effect of vitamin D on mortality in subgroup analyses of trials at low risk of bias compared with trials at high risk of bias; of trials using placebo compared with trials using no intervention in the control group; of trials with no risk of industry bias compared with trials with risk of industry bias; of trials assessing primary prevention compared with trials assessing secondary prevention; of trials including participants with vitamin D level below 20 ng/mL at entry compared with trials including participants with vitamin D levels equal to or greater than 20 ng/mL at entry; of trials including ambulatory participants compared with trials including institutionalised participants; of trials using concomitant calcium supplementation compared with trials without calcium; of trials using a dose below 800 IU per day compared with trials using doses above 800 IU per day; and of trials including only women compared with trials including both sexes or only men. Vitamin D3 statistically significantly decreased cancer mortality (RR 0.88 (95% CI 0.78 to 0.98); P = 0.02; I(2) = 0%; 44,492 participants; 4 trials). Vitamin D3 combined with calcium increased the risk of nephrolithiasis (RR 1.17 (95% CI 1.02 to 1.34); P = 0.02; I(2) = 0%; 42,876 participants; 4 trials). Alfacalcidol and calcitriol increased the risk of hypercalcaemia (RR 3.18 (95% CI 1.17 to 8.68); P = 0.02; I(2) = 17%; 710 participants; 3 trials).

AUTHORS' CONCLUSIONS

Vitamin D3 seemed to decrease mortality in elderly people living independently or in institutional care. Vitamin D2, alfacalcidol and calcitriol had no statistically significant beneficial effects on mortality. Vitamin D3 combined with calcium increased nephrolithiasis. Both alfacalcidol and calcitriol increased hypercalcaemia. Because of risks of attrition bias originating from substantial dropout of participants and of outcome reporting bias due to a number of trials not reporting on mortality, as well as a number of other weaknesses in our evidence, further placebo-controlled randomised trials seem warranted.

Genetic Associations of Angiotensin-Converting Enzyme with Primary Intracerebral Hemorrhage: A Meta-analysis

Background

A number of studies have reported an association of angiotensin-converting enzyme (ACE) gene polymorphism with primary intracerebral hemorrhage (PICH), however the reports have demonstrated inconclusive results. To clarify this conflict, we updated the previously performed meta-analysis by Peck et al., which revealed negative results, by investigating the ACE polymorphism and its correlation to PICH.

Methods

PubMed and Embase databases (through Dec 2012) were searched for English articles on the relationship of the I/D polymorphism in ACE with PICH in humans. Summary odds ratios (ORs) were estimated and potential sources of heterogeneity and bias were explored.

Results

A total of 805 PICH cases and 1641 control cases obtained from 8 case-control studies were included. The results suggest that in dominant genetic models, the ACE I/D polymorphic variant was associated with a 58% increase in susceptibility risk of PICH (OR = 1.58; 95% CI = 1.07–2.35 for DD vs. DI+II). However, in the subgroup analysis based on race, a significant increased risk was found in Asian DD homozygote carriers (OR = 1.76 and 95% CI = 1.16–2.66 for DD vs. DI+II), but not in Caucasian DD homozygote carriers (OR = 1.18, 95% CI = 0.36–3.88, P = 0.784 for DD vs. DI+II). The heterogeneity between studies was remarkable, and its major sources of heterogeneity were due to the year in which the study was published. No potential publication bias was observed in dominant genetic models.

Conclusions

These data demonstrated evidence of a positive association between ACE I/D polymorphism with PICH, and suggested that the ACE gene is a PICH susceptible gene in Asian populations.

Scientific inbreeding and same-team replication: type D personality as an example

Replication is essential for validating correct results, sorting out false-positive early discoveries, and improving the accuracy and precision of estimated effects. However, some types of seemingly successful replication may foster a spurious notion of increased credibility, if they are performed by the same team and propagate or extend the same errors made by the original discoveries. Besides same-team replication, replication by other teams may also succumb to inbreeding, if it cannot fiercely maintain its independence. These patterns include obedient replication and obliged replication. I discuss these replication patterns in the context of associations and effects in the psychological sciences, drawing from the criticism of Coyne and de Voogd of the proposed association between type D personality and cardiovascular mortality and other empirical examples.

Association of angiotensin-converting enzyme I/D polymorphism with heart failure: a meta-analysis

Heart failure (HF) is a complex clinical syndrome and is thought to have a genetic basis. Numerous case-control studies have investigated the association between heart failure and polymorphisms in candidate genes. Most studies focused on the angiotensin-converting enzyme insertion/deletion (ACE I/D) polymorphism, however, the results were inconsistent because of small studies and heterogeneous samples. The objective was to assess the association between the ACE I/D polymorphism and HF. We performed a meta-analysis of all case-control studies that evaluated the association between ACE I/D polymorphism and HF in humans. Studies were identified in the PUBMED and EMBASE databases, reviews, and reference lists of relevant articles. Two reviewers independently assessed the studies. Seventeen case-control studies with a total of 5576 participants were included in the meta-analysis, including 2453 cases with HF and 3123 controls. The heterogeneity between studies was significant. No association was found under all the four genetic models (D vs. I, DD vs. ID and II, DD and ID vs. II, DD vs. ID). Subgroup analyses for ischemic HF (IHF) and HF because of dilated cardiomyopathy (DHF) also showed no significant association between ACE I/D polymorphism and HF. No significant association between the ACE I/D polymorphism and risk of HF was found in this meta-analysis. The future studies should focus on large-scale prospective and case-control studies which designed to investigate gene-gene and gene-environment interactions to shed light on the genetics of HF.

Synopsis and meta-analysis of genetic association studies in osteoporosis for the focal adhesion family genes: the CUMAGAS-OSTEOporosis information system

BACKGROUND

Focal adhesion (FA) family genes have been studied as candidate genes for osteoporosis, but the results of genetic association studies (GASs) are controversial. To clarify these data, a systematic assessment of GASs for FA genes in osteoporosis was conducted.

METHODS

We developed Cumulative Meta-Analysis of GAS-OSTEOporosis (CUMAGAS-OSTEOporosis), a web-based information system that allows the retrieval, analysis and meta-analysis (for allele contrast, recessive, dominant, additive and codominant models) of data from GASs on osteoporosis with the capability of update. GASs were identified by searching the PubMed and HuGE PubLit databases.

RESULTS

Data from 72 studies involving 13 variants of 6 genes were analyzed and catalogued in CUMAGAS-OSTEOporosis. Twenty-two studies produced significant associations with osteoporosis risk under any genetic model. All studies were underpowered (<50%). In four studies, the controls deviated from the Hardy-Weinberg equilibrium. Eight variants were chosen for meta-analysis, and significance was shown for the variants collagen, type I, α1 (COL1A1) G2046T (all genetic models), COL1A1 G-1997T (allele contrast and dominant model) and integrin β-chain β3 (ITGB3) T176C (recessive and additive models). In COL1A1 G2046T, subgroup analysis has shown significant associations for Caucasians, adults, females, males and postmenopausal women. A differential magnitude of effect in large versus small studies (that is, indication of publication bias) was detected for the variant COL1A1 G2046T.

CONCLUSION

There is evidence of an implication of FA family genes in osteoporosis. CUMAGAS-OSTEOporosis could be a useful tool for current genomic epidemiology research in the field of osteoporosis.

Randomized trials of dopamine agonists in restless legs syndrome: a systematic review, quality assessment, and meta-analysis

BACKGROUND

The use of dopamine agonists (DAs) for the treatment of restless legs syndrome (RLS) has been assessed in numerous randomized clinical trials (RCTs).

OBJECTIVES

The aims of this study were to assess the reporting quality of published RCTs according to the Consolidated Standards of Reporting Trials (CONSORT) statement and to synthesize the study results in terms of efficacy and tolerability to inform the clinical management of RLS.

METHODS

PubMed and Cochrane Controlled Trials Register were searched for English-language RCTs that assessed the effects of DAs in RLS. Quality of reporting was measured using the proportion of 17 CONSORT checklist items included in each study. The 2 primary outcomes were pooled mean change from baseline in International RLS (IRLS) Study Group rating scale score (Deltamu) (95% CI) and relative risk (RR) (95% CI) of response based on the Clinical Global Impression-Improvement (CGI-I) scale score. The pooled proportions of adverse events (PAEs) (95% CI) were also estimated.

RESULTS

Eighteen RCTs (N = 2848 patients) were included. Two of the 17 CONSORT checklist items were reported in 7 studies (39%) and 9 of the 17 items were reported in all 18 studies (100%). The differences in the IRLS scores and RR for CGI-I were significantly greater with pramipexole, ropinirole, rotigotine, and cabergoline compared with placebo. Results for heterogeneity were nonsignificant. The difference in Deltamu (95% CI) was significant with pramipexole (-6.63 [-9.15 to -4.10]) versus ropinirole (-3.64 [-4.76 to 2.51]) (P = 0.04). The difference between pramipexole and rotigotine was nonsignificant. The pooled PAEs (95% CI) for pramipexole, ropinirole, and rotigotine were 4.8% (2.0% to 8.7%), 10.2% (2.6% to 22.1%), and 7.6% (1.3% to 18.5%), respectively. In the trial of sumanirole, the PAE value was 2% (0% to 5.4%).

CONCLUSION

Based on the findings from the meta-analysis, DAs were significantly more efficacious in the treatment of RLS compared with placebo.

Birth weight and risk of neuroblastoma: a meta-analysis

BACKGROUND

Neuroblastoma is the most common solid tumour in infancy but its aetiology is largely unknown. Prenatal factors might play a key role in its pathogenesis. Previous studies investigated whether birth weight is associated with risk of neuroblastoma, with conflictive results. We conducted a meta-analysis to quantitatively summarize the published evidence.

METHODS

Results from 10 case-control studies and one cohort study (1966 to December 2008) were included, involving a total of 3004 children with neuroblastoma. We constructed random-effects and fixed-effects models, performed 'pool-first' analyses, assessed heterogeneity and publication bias and performed sensitivity and influence analyses.

RESULTS

High birth weight (>4000 g) was associated with increased risk of neuroblastoma [odds ratio (OR) 1.19; 95% confidence interval (CI) 1.04-1.36]. Results for high birth weight were highly homogenous (I(2) = 0%). Low birth weight (<2500 g) was also related to increased risk of neuroblastoma (OR 1.24; 95% CI 1.0-1.55), but results were more heterogeneous (I(2 )= 30%). No evidence for particularly influential studies or for publication bias was found. However, sensitivity analysis indicated the presence of bias in studies on the association with low birth weight. Above 2500 g each 1000-g increase in birth weight was associated with a 13% (95% CI 3-25) increase in risk of neuroblastoma.

CONCLUSIONS

This meta-analysis shows that high birth weight is highly reproducibly associated with increased risk of neuroblastoma. The association with low birth weight was found to be less robust and deserves further studies.

Estimating required information size by quantifying diversity in random-effects model meta-analyses

BACKGROUND

There is increasing awareness that meta-analyses require a sufficiently large information size to detect or reject an anticipated intervention effect. The required information size in a meta-analysis may be calculated from an anticipated a priori intervention effect or from an intervention effect suggested by trials with low-risk of bias.

METHODS

Information size calculations need to consider the total model variance in a meta-analysis to control type I and type II errors. Here, we derive an adjusting factor for the required information size under any random-effects model meta-analysis.

RESULTS

We devise a measure of diversity (D2) in a meta-analysis, which is the relative variance reduction when the meta-analysis model is changed from a random-effects into a fixed-effect model. D2 is the percentage that the between-trial variability constitutes of the sum of the between-trial variability and a sampling error estimate considering the required information size. D2 is different from the intuitively obvious adjusting factor based on the common quantification of heterogeneity, the inconsistency (I2), which may underestimate the required information size. Thus, D2 and I2 are compared and interpreted using several simulations and clinical examples. In addition we show mathematically that diversity is equal to or greater than inconsistency, that is D2 >or= I2, for all meta-analyses.

CONCLUSION

We conclude that D2 seems a better alternative than I2 to consider model variation in any random-effects meta-analysis despite the choice of the between trial variance estimator that constitutes the model. Furthermore, D2 can readily adjust the required information size in any random-effects model meta-analysis.

Pharmacological and combined interventions for the acute depressive episode: focus on efficacy and tolerability

BACKGROUND

Use of antidepressants is the gold standard therapy for major depression. However, despite the large number of commercially available antidepressant drugs there are several differences among them in efficacy, tolerability, and cost-effectiveness. In addition the optimal augmentation strategy is still not clear when dealing with treatment-resistant depression, a condition that affects 15% to 40% of depressed patients.

METHODS

We therefore reviewed the main characteristics of these drugs regarding their efficacy, tolerability, side effects and cost-effectiveness, by accessing all meta-analyses and systematic reviews published from 2004 to 2009. In addition, we reviewed the augmentation strategy of associated antidepressants with neurostimulation therapies (such as transcranial magnetic stimulation [TMS] and transcranial direct current stimulation [tDCS]). A search was undertaken in MEDLINE, Web of Science, Cochrane, and Scielo databases. We included: 21 meta-analyses of antidepressant trials, 15 neurostimulation clinical trials and 8 studies of pharmacoeconomics. We then performed a comprehensive review on these articles.

RESULTS AND CONCLUSION

Although recent meta-analyses suggest sertraline and escitalopram might have increased efficacy/tolerability, other studies and large pragmatic trials have not found these to be superior to other antidepressant drugs. Also, we did not identify any superior drug in terms of cost-effectiveness due to the different designs observed among pharmacoecomics studies. Side effects such as sexual dysfunction, gastrointestinal problems and weight gain were common causes of discontinuation. Tolerability was an important issue for novel neurostimulation interventions, such as TMS and tDCS. These therapies might be interesting augmentation strategies, considering their benign profile of side effects, if proper safety parameters are adopted.

Birth weight, early weight gain, and subsequent risk of type 1 diabetes: systematic review and meta-analysis

Previous studies suggest that birth weight and weight gain during the first year of life are related to later risk of type 1 diabetes. The authors performed a systematic review and meta-analysis on these associations. Twelve studies involving 2,398,150 persons of whom 7,491 had type 1 diabetes provided odds ratios and 95% confidence intervals of type 1 diabetes associated with birth weight. Four studies provided data on weight and/or weight gain during the first year of life. High birth weight (>4,000 g) was associated with increased risk of type 1 diabetes (odds ratio = 1.17, 95% confidence interval (CI): 1.09, 1.26). According to sensitivity analysis, this result was not influenced by particular study characteristics. The pooled confounder-adjusted estimate was 1.43 (95% CI: 1.11, 1.85). No heterogeneity was found (I(2) = 0%) and no publication bias. Low birth weight (<2,500 g) was associated with a nonsignificantly decreased risk of type 1 diabetes (odds ratio = 0.82, 95% CI: 0.54, 1.23). Each 1,000-g increase in birth weight was associated with a 7% increase in type 1 diabetes risk. In all studies, patients with type 1 diabetes showed increased weight gain during the first year of life, compared with controls. This meta-analysis indicates that high birth weight and increased early weight gain are risk factors for type 1 diabetes.

Interpretation of tests of heterogeneity and bias in meta-analysis

Statistical tests of heterogeneity and bias, in particular publication bias, are very popular in meta-analyses. These tests use statistical approaches whose limitations are often not recognized. Moreover, it is often implied with inappropriate confidence that these tests can provide reliable answers to questions that in essence are not of statistical nature. Statistical heterogeneity is only a correlate of clinical and pragmatic heterogeneity and the correlation may sometimes be weak. Similarly, statistical signals may hint to bias, but seen in isolation they cannot fully prove or disprove bias in general, let alone specific causes of bias, such as publication bias in particular. Both false-positive and false-negative signals of heterogeneity and bias can be common and their prevalence may be anticipated based on some rational considerations. Here I discuss the major common challenges and flaws that emerge in using and interpreting statistical tests of heterogeneity and bias in meta-analyses. I discuss misinterpretations that can occur at the level of statistical inference, clinical/pragmatic inference and specific cause attribution. Suggestions are made on how to avoid these flaws, use these tests properly and learn from them.

Genetic risk factors for placental abruption: a HuGE review and meta-analysis

BACKGROUND

Although the precise pathophysiology that leads to placental abruption is unknown, there is evidence supporting a genetic etiology.

METHODS

We searched PubMed and systematically reviewed all case-control studies that investigated the association between genetic variants and placental abruption. Pooled genetic risks were estimated using fixed and random effects odds ratios.

RESULTS

Twenty-two articles, examining a total of 14 gene polymorphisms were identified. Seven polymorphisms (F5 Arg506Gln, F5 Met385Thr, F2 G20210A, MTHFR A1298C, MTHFD1 Arg653Gln, NOS3 Glu298Asp, AGT Met235Thr) show significant association in individual studies. Six of the 7 (all except F5Met385Thr) were studied more than once and we therefore included them in our meta-analyses. A positive association under the dominant model was found for the F5 Arg506Gln and F2 G20210A polymorphisms. The random-effects odds ratio for the F5 Arg506Gln polymorphism was 3.4 (95% confidence interval = 1.4-8.3) and the fixed-effects odds ratio for the F2 G20210A polymorphism was 6.7 (3.2-13).

CONCLUSION

Considering the multifactorial etiology of abruption and the relatively small numbers of studies and participants, this review provides only the first clues of possible genetic causes. Larger case-control studies that include gene-gene and gene-environment interactions may help to elucidate the genetics of placental abruption further.

Citrus fruit intake and stomach cancer risk: a quantitative systematic review

BACKGROUND

The purpose of this systematic review was to investigate the association between dietary intake of citrus fruits and gastric cancer risk.

METHODS

We searched electronic databases and the reference lists of publications on diet and stomach cancer studies until April 2007. All of the epidemiological studies that obtained individual data on dietary intake of citrus fruits and presented risk estimates of the association between intake of citrus fruits and risk of stomach cancer were identified and included in this review. Using general variance-based methods, study-specific odds ratio (OR)/ relative risk (RR) and associated confidence interval (CI)/ standard error (SE) for highest versus lowest intake of citrus fruits level were extracted from each article.

RESULTS

Fourteen articles, including six hospital-based case-control studies, six community-based case-control studies, and two cohort studies, proved eligible. Overall summary OR using a fixed-effect model showed a 28% reduction in risk of stomach cancer associated with high intake of citrus fruits (summary OR = 0.72; 95% CI = 0.64-0.81; P value <0.0001); results were consistent across studies (I(2) = 0, where I(2) describes the percentage of total variation across studies because of heterogeneity rather than chance). Also, visual inspection of the results did not suggest a publication bias.

CONCLUSION

Pooled results from observational studies support a protective effect of high citrus fruit intake in the risk of stomach cancer.

Uncertainty in heterogeneity estimates in meta-analyses

John Ioannidis, Nikolaos Patsopoulos, and Evangelos Evangelou argue that, although meta-analyses often measure heterogeneity between studies, these estimates can have large uncertainty, which must be taken into account when interpreting evidence

Assessment of cumulative evidence on genetic associations: interim guidelines

Established guidelines for causal inference in epidemiological studies may be inappropriate for genetic associations. A consensus process was used to develop guidance criteria for assessing cumulative epidemiologic evidence in genetic associations. A proposed semi-quantitative index assigns three levels for the amount of evidence, extent of replication, and protection from bias, and also generates a composite assessment of 'strong', 'moderate' or 'weak' epidemiological credibility. In addition, we discuss how additional input and guidance can be derived from biological data. Future empirical research and consensus development are needed to develop an integrated model for combining epidemiological and biological evidence in the rapidly evolving field of investigation of genetic factors.

Methylenetetrahydrofolate reductase gene polymorphism as a risk factor for diabetic nephropathy: a meta-analysis

Investigations into the association between diabetic nephropathy (DN) and MTHFR C677T gene polymorphism in several case-control studies has yielded contradictory results. To shed light on these inconclusive findings, a meta-analysis of all available studies relating the C677T polymorphism to the risk of developing DN was conducted. The PubMed database was searched, and case-control studies investigating the association between MTHFR C677T gene polymorphism and DN were included in the meta-analysis. The meta-analysis included 15 studies, of which 8 involved Caucasians and 5 East Asians; 11 studies involved subjects with type 2 diabetes and 4 with type 1 diabetes. The main analysis (all studies) revealed significant heterogeneity between the studies (P(Q)<0.01) and a marginal association between the 677T allele and the risk of developing DN; the random effects (RE) pooled odds ratio (OR) was 1.30 (1.03-1.64). However, the sensitivity analysis (exclusion of studies not in Hardy-Weinberg equilibrium) produced non-significant results. The recessive model derived significant results in main analysis [fixed effects (FE) OR=1.32 (1.10-1.58), P(Q)=0.27], and in type 2 diabetes [FE OR=1.30 (1.06-1.60), P(Q)=0.38]. The additive model produced significant association in main analysis [RE OR=1.65 (1.13-2.42), P(Q)<0.01] in Caucasians [FE OR=1.48 (1.11-1.98), P(Q)=0.17] and in type 2 diabetes [RE OR=1.65 (1.03-2.67), P(Q)<0.01]. However, sensitivity analysis diminished the significant results in type 2 diabetes. There is no differential magnitude of effect in large versus small studies. In conclusion, although there is some evidence of association between MTHFR C677T gene polymorphism and DN, the above findings reinforce the need for further and more rigorous association studies.

The appropriateness of asymmetry tests for publication bias in meta-analyses: a large survey

BACKGROUND

Statistical tests for funnel-plot asymmetry are common in meta-analyses. Inappropriate application can generate misleading inferences about publication bias. We aimed to measure, in a survey of meta-analyses, how frequently the application of these tests would be not meaningful or inappropriate.

METHODS

We evaluated all meta-analyses of binary outcomes with é 3 studies in the Cochrane Database of Systematic Reviews (2003, issue 2). A separate, restricted analysis was confined to the largest meta-analysis in each of the review articles. In each meta-analysis, we assessed whether criteria to apply asymmetry tests were met: no significant heterogeneity, I2 < 50%, é 10 studies (with statistically significant results in at least 1) and ratio of the maximal to minimal variance across studies > 4. We performed a correlation and 2 regression asymmetry tests and evaluated their concordance. Finally, we sampled 60 meta-analyses from print journals in 2005 that cited use of the standard regression test.

RESULTS

A total of 366 of 6873 (5%) and 98 of 846 meta-analyses (12%) in the wider and restricted Cochrane data set, respectively, would have qualified for use of asymmetry tests. Asymmetry test results were significant in 7%-18% of the meta-analyses. Concordance between the 3 tests was modest (estimated k 0.33-0.66). Of the 60 journal meta-analyses, 7 (12%) would qualify for asymmetry tests; all 11 claims for identification of publication bias were made in the face of large and significant heterogeneity.

INTERPRETATION

Statistical conditions for employing asymmetry tests for publication bias are absent from most meta-analyses; yet, in medical journals these tests are performed often and interpreted erroneously.

Brain-derived neurotrophic factor gene polymorphisms and schizophrenia: a meta-analysis

OBJECTIVES

To evaluate whether the G196A and C270T polymorphisms of the brain-derived neurotrophic factor gene are associated with increased risk of schizophrenia.

METHODS

A meta-analysis of nine genetic association studies was carried out. The meta-analysis included genotype data on 1404/1597 schizophrenics/controls for G196A and 877/989 schizophrenics/controls for C270T.

RESULTS

The overall analysis for investigating the association of the G196A allele G and the risk of developing schizophrenia relative to the allele A, showed significant evidence of heterogeneity (P=0.05, I(2)=58%) between the studies and nonsignificant association [random effects odds ratio 1.08 and 95% confidence interval (0.88-1.32)]. In Caucasians, there was a trend towards heterogeneity (P=0.19, I(2)=40%), then, the random and fixed effects odds ratios were 1.24 (0.96-1.60) and 1.27 (1.06-1.53), respectively. For the C270T polymorphism, overall, there was significant evidence of heterogeneity between studies (P=0.07, I(2)=55%) and the allele T was associated with a 63% increased risk of schizophrenia compared with C allele [random effects odds ratio 1.63 (1.01-2.65)]. The dominant model for allele T produced significant association [random effects odds ratio 1.68 (1.02-2.79)]. No source of bias was seen in the selected studies and the differential magnitude of effect in large versus small studies was not significant.

CONCLUSIONS

The meta-analysis results provided a weak evidence of association between C270T polymorphism and schizophrenia, and large heterogeneity between studies, whereas there was no evidence of association for G196A polymorphism. The above findings reinforce the need for large and more rigorous association studies.