A population‐averaged approach to diagnostic test meta‐analysis

The meta‐analysis of diagnostic accuracy studies is often of interest in screening programs for many diseases. The typical summary statistics for studies chosen for a diagnostic accuracy meta‐analysis are often two dimensional: sensitivities and specificities. The common statistical analysis approach for the meta‐analysis of diagnostic studies is based on the bivariate generalized linear‐mixed model (BGLMM), which has study‐specific interpretations. In this article, we present a population‐averaged (PA) model using generalized estimating equations (GEE) for making inference on mean specificity and sensitivity of a diagnostic test in the population represented by the meta‐analytic studies. We also derive the marginalized counterparts of the regression parameters from the BGLMM. We illustrate the proposed PA approach through two dataset examples and compare performance of estimators of the marginal regression parameters from the PA model with those of the marginalized regression parameters from the BGLMM through Monte Carlo simulation studies. Overall, both marginalized BGLMM and GEE with sandwich standard errors maintained nominal 95% confidence interval coverage levels for mean specificity and mean sensitivity in meta‐analysis of 25 of more studies even under misspecification of the covariance structure of the bivariate positive test counts for diseased and nondiseased subjects.     

A meta‐analysis of dispersal in butterflies

Dispersal has recently gained much attention because of its crucial role in the conservation and evolution of species facing major environmental changes such as habitat loss and fragmentation, climate change, and their interactions. Butterflies have long been recognized as ideal model systems for the study of dispersal and a huge amount of data on their ability to disperse has been collected under various conditions. However, no single ‘best’ method seems to exist leading to the co‐occurrence of various approaches to study butterfly mobility, and therefore a high heterogeneity among data on dispersal across this group. Accordingly, we here reviewed the knowledge accumulated on dispersal and mobility in butterflies, to detect general patterns. This meta‐analysis specifically addressed two questions. Firstly, do the various methods provide a congruent picture of how dispersal ability is distributed across species? Secondly, is dispersal species‐specific? Five sources of data were analysed: multisite mark‐recapture experiments, genetic studies, experimental assessments, expert opinions, and transect surveys. We accounted for potential biases due to variation in genetic markers, sample sizes, spatial scales or the level of habitat fragmentation. We showed that the various dispersal estimates generally converged, and that the relative dispersal ability of species could reliably be predicted from their relative vagrancy (records of butterflies outside their normal habitat). Expert opinions gave much less reliable estimates of realized dispersal but instead reflected migration propensity of butterflies. Within‐species comparisons showed that genetic estimates were relatively invariable, while other dispersal estimates were highly variable. This latter point questions dispersal as a species‐specific, invariant trait.     

A matrix‐based method of moments for fitting the multivariate random effects model for meta‐analysis and meta‐regression

Multivariate meta‐analysis is becoming more commonly used. Methods for fitting the multivariate random effects model include maximum likelihood, restricted maximum likelihood, Bayesian estimation and multivariate generalisations of the standard univariate method of moments. Here, we provide a new multivariate method of moments for estimating the between‐study covariance matrix with the properties that (1) it allows for either complete or incomplete outcomes and (2) it allows for covariates through meta‐regression. Further, for complete data, it is invariant to linear transformations. Our method reduces to the usual univariate method of moments, proposed by DerSimonian and Laird, in a single dimension. We illustrate our method and compare it with some of the alternatives using a simulation study and a real example.     

Psychotherapies for depression in low‐ and middle‐income countries: a meta‐analysis

Most psychotherapies for depression have been developed in high‐income Western countries of North America, Europe and Australia. A growing number of randomized trials have examined the effects of these treatments in non‐Western countries. We conducted a meta‐analysis of these studies to examine whether these psychotherapies are effective and to compare their effects between studies from Western and non‐Western countries. We conducted systematic searches in bibliographical databases and included 253 randomized controlled trials, of which 32 were conducted in non‐Western countries. The effects of psychotherapies in non‐Western countries were large (g=1.10; 95% CI: 0.91‐1.30), with high heterogeneity (I²=90; 95% CI: 87‐92). After adjustment for publication bias, the effect size dropped to g=0.73 (95% CI: 0.51‐0.96). Subgroup analyses did not indicate that adaptation to the local situation was associated with the effect size. Comparisons with the studies in Western countries showed that the effects of the therapies were significantly larger in non‐Western countries, also after adjusting for characteristics of the participants, the treatments and the studies. These larger effect sizes in non‐Western countries may reflect true differences indicating that therapies are indeed more effective; or may be explained by the care‐as‐usual control conditions in non‐Western countries, often indicating that no care was available; or may be the result of the relative low quality of many trials in the field. This study suggests that psychotherapies that were developed in Western countries may or may not be more effective in non‐Western countries, but they are probably no less effective and can therefore also be used in these latter countries.     

Multivariate meta‐analysis with an increasing number of parameters

Meta‐analysis can average estimates of multiple parameters, such as a treatment's effect on multiple outcomes, across studies. Univariate meta‐analysis (UVMA) considers each parameter individually, while multivariate meta‐analysis (MVMA) considers the parameters jointly and accounts for the correlation between their estimates. The performance of MVMA and UVMA has been extensively compared in scenarios with two parameters. Our objective is to compare the performance of MVMA and UVMA as the number of parameters, p, increases. Specifically, we show that (i) for fixed‐effect (FE) meta‐analysis, the benefit from using MVMA can substantially increase as p increases; (ii) for random effects (RE) meta‐analysis, the benefit from MVMA can increase as p increases, but the potential improvement is modest in the presence of high between‐study variability and the actual improvement is further reduced by the need to estimate an increasingly large between study covariance matrix; and (iii) when there is little to no between‐study variability, the loss of efficiency due to choosing RE MVMA over FE MVMA increases as p increases. We demonstrate these three features through theory, simulation, and a meta‐analysis of risk factors for non‐Hodgkin lymphoma.     

Inference for meta‐analysis with a suspected temporal trend

There is sometimes a clear evidence of a strong secular trend in the treatment effect of studies included in a meta‐analysis. In such cases, estimating the present‐day treatment effect by meta‐regression is both reasonable and straightforward. We however consider the more common situation where a secular trend is suspected, but is not strongly statistically significant. Typically, this lack of significance is due to the small number of studies included in the analysis, so that a meta‐regression could give wild point estimates. We introduce an empirical Bayes meta‐analysis methodology, which shrinks the secular trend toward zero. This has the effect that treatment effects are adjusted for trend, but where the evidence from data is weak, wild results are not obtained. We explore several frequentist approaches and a fully Bayesian method is also implemented. A measure of trend analogous to I² is described, and exact significance tests for trend are given. Our preferred method is one based on penalized or h‐likelihood, which is computationally simple, and allows invariance of predictions to the (arbitrary) choice of time origin. We suggest that a trendless standard random effects meta‐analysis should routinely be supplemented with an h‐likelihood analysis as a sensitivity analysis.     

Evaluation of association of maternal IL‐10 polymorphisms with risk of preeclampsia by A meta‐analysis

Emerging evidence shows that interleukin (IL)‐10 gene polymorphisms can regulate its expression level and thus influence person's susceptibility to preeclampsia. However, various published results were inconsistent. To explore the association between maternal IL‐10 gene polymorphisms and preeclampsia, we performed a meta‐analysis based upon 11 individual studies here. Our meta‐analysis results indicated that IL‐10 ‐819C/T (C versus T, OR = 1.28, 95% CI = 1.08–1.50, P = 0.003) and ‐592C/A (C versus A, OR = 1.28, 95% CI = 1.03–1.59, P = 0.03) polymorphisms were associated with preeclampsia. Although there was no overall association between ‐1082A/G polymorphism and preeclampsia (G versus A, OR = 0.93, 95% CI = 0.77–1.13, P = 0.49), such association existed among Asian (G versus A, OR = 1.29, 95% CI = 1.04–1.60, P = 0.02) and South American (G versus A, OR = 0.72, 95% CI = 0.54–0.94, P = 0.02) populations in the subgroup analysis stratified by continents.     

Bivariate random‐effects meta‐analysis models for diagnostic test accuracy studies using arcsine‐based transformations

Diagnostic or screening tests are widely used in medical fields to classify patients according to their disease status. Several statistical models for meta‐analysis of diagnostic test accuracy studies have been developed to synthesize test sensitivity and specificity of a diagnostic test of interest. Because of the correlation between test sensitivity and specificity, modeling the two measures using a bivariate model is recommended. In this paper, we extend the current standard bivariate linear mixed model (LMM) by proposing two variance‐stabilizing transformations: the arcsine square root and the Freeman–Tukey double arcsine transformation. We compared the performance of the proposed methods with the standard method through simulations using several performance measures. The simulation results showed that our proposed methods performed better than the standard LMM in terms of bias, root mean square error, and coverage probability in most of the scenarios, even when data were generated assuming the standard LMM. We also illustrated the methods using two real data sets.     

Bottom‐up vs. top‐down effects on terrestrial insect herbivores: a meta‐analysis

Primary consumers are under strong selection from resource (‘bottom‐up’) and consumer (‘top‐down’) controls, but the relative importance of these selective forces is unknown. We performed a meta‐analysis to compare the strength of top‐down and bottom‐up forces on consumer fitness, considering multiple predictors that can modulate these effects: diet breadth, feeding guild, habitat/environment, type of bottom‐up effects, type of top‐down effects and how consumer fitness effects are measured. We focused our analyses on the most diverse group of primary consumers, herbivorous insects, and found that in general top‐down forces were stronger than bottom‐up forces. Notably, chewing, sucking and gall‐making herbivores were more affected by top‐down than bottom‐up forces, top‐down forces were stronger than bottom‐up in both natural and controlled (cultivated) environments, and parasitoids and predators had equally strong top‐down effects on insect herbivores. Future studies should broaden the scope of focal consumers, particularly in understudied terrestrial systems, guilds, taxonomic groups and top‐down controls (e.g. pathogens), and test for more complex indirect community interactions. Our results demonstrate the surprising strength of forces exerted by natural enemies on herbivorous insects, and thus the necessity of using a tri‐trophic approach when studying insect‐plant interactions.     

Interleukin‐8 ‐251A/T gene polymorphism and lung cancer susceptibility: a meta‐analysis

Many studies have examined the association between the interleukin‐8 ‐251T/A ( rs4073 ) gene polymorphism and lung cancer risk in various populations, but the results have been inconsistent. In this meta‐analysis, PubMed was searched for case–control studies published through 01 December 2013. The data were extracted, and pooled odds ratios (OR) with 95% confidence intervals (CI) were calculated. We assessed six published studies on the association between the interleukin‐8 ‐251T/A polymorphism and lung cancer risk. The included studies yielded a total of 3265 lung cancer cases and 3607 controls. For the homozygous A/A and A allele carriers (T/A + A/A), the pooled ORs for all studies combining 3265 cases and 3607 controls were 1.03 (95% CI = 0.92–1.14; P = 0.235 for heterogeneity) and 1.07 (95% CI = 0.96–1.19; P = 0.245 for heterogeneity) when compared with the homozygous wild‐type genotype (T/T). When the analysis was stratified by ethnicity, significant risks were found among Asians for both the A allele carriers and the homozygous A/A individuals. However, no significant associations were found in non‐Asian populations using any of the genetic models. This meta‐analysis suggests that the interleukin‐8 ‐251A allele confer an increased risk for the development of lung cancer among Asians.     

Virulence‐driven trade‐offs in disease transmission: A meta‐analysis*

The virulence–transmission trade‐off hypothesis proposed more than 30 years ago is the cornerstone in the study of host–parasite co‐evolution. This hypothesis rests on the premise that virulence is an unavoidable and increasing cost because the parasite uses host resources to replicate. This cost associated with replication ultimately results in a deceleration in transmission rate because increasing within‐host replication increases host mortality. Empirical tests of predictions of the hypothesis have found mixed support, which cast doubt about its overall generalizability. To quantitatively address this issue, we conducted a meta‐analysis of 29 empirical studies, after reviewing over 6000 published papers, addressing the four core relationships between (1) virulence and recovery rate, (2) within‐host replication rate and virulence, (3) within‐host replication and transmission rate, and (4) virulence and transmission rate. We found strong support for an increasing relationship between replication and virulence, and replication and transmission. Yet, it is still uncertain if these relationships generally decelerate due to high within‐study variability. There was insufficient data to quantitatively test the other two core relationships predicted by the theory. Overall, the results suggest that the current empirical evidence provides partial support for the trade‐off hypothesis, but more work remains to be done.     

Sylvatic plague management and prairie dogs – a meta‐analysis

Yersinia pestis, a bacterial pathogen that causes sylvatic plague, is present in the prairie dogs (Cynomys spp.) of North America. Epizootics of sylvatic plague through transmission in vectors (fleas) commonly completely extirpate colonies of prairie dogs. Wildlife managers employ a wide variety of insecticidal treatments to suppress plague and conserve prairie dog colonies. I compiled and statistically compared the available literature describing methods of plague control and their relative effectiveness in managing plague outbreaks by using meta‐analyses. Natural log response ratios were used to calculate insecticide‐induced vector mortality and vaccine‐conferred survival increases in prairie dogs in 37 publications. Further, subgroupings were used to explore the most effective of the available vector suppression insecticides and plague suppression vaccines. After accounting for the type of treatment used and the method by which it was applied, I observed plague reduction through use of both insecticides and vaccines. Insecticides resulted in a significant reduction of the abundance of vectors by 91.34% compared to non‐treated hosts (p<0.0001). Vaccines improved survival of prairie dog hosts by 4.00% (p<0.0001) compared to control populations. The use of insecticides such as deltamethrin and carbaryl is recommended to stop actively spreading epizootics, and dual antigen oral vaccines to initially suppress outbreaks.     

A meta‐analysis of bird and mammal response to short‐rotation woody crops

Short‐rotation woody cropping (SRWC) refers to silvicultural systems designed to produce woody biomass using short harvest cycles (1–15 years), intensive silvicultural techniques, high‐yielding varieties, and often coppice regeneration. Recent emphasis on alternatives to fossil fuels has spurred interest in producing SRWC on privately owned and intensively managed forests of North America. We examined potential bird and small mammal response at the stand level to conversion of existing, intensively managed forests to SRWCs using meta‐analysis of existing studies. We found 257 effect sizes for birds (243 effect sizes) and mammals (14 effect sizes) from 8 studies involving Populus spp. plantations. Diversity and abundance of bird guilds were lower on short‐rotation plantations compared with reference woodlands, while abundance of individual bird species was more variable and not consistently higher or lower on SRWC plantations. Shrub‐associated birds were more abundant on SRWC plantations, but forest‐associated and cavity‐nesting birds were less abundant. Effects on birds appeared to decrease with age of the SRWC plantation, but plantation age was also confounded with variation in the type of reference forest used for comparison. Both guilds and species of mammals were less abundant on SRWC plantations. These conclusions are tentative because none of these studies directly compared SRWC plantations to intensively managed forests. Plantations of SRWCs could contribute to overall landscape diversity in forest‐dominated landscapes by providing shrubby habitat structure for nonforest species. However, extensive conversion of mature or intensively managed forests to SRWC would likely decrease overall diversity, especially if they replace habitat types of high conservation value.     

Application of meta‐analysis methods for identifying proteomic expression level differences

We present new statistical approaches for identification of proteins with expression levels that are significantly changed when applying meta‐analysis to two or more independent experiments. We showed that the Euclidean distance measure has reduced risk of false positives compared to the rank product method. Our Ψ‐ranking method has advantages over the traditional fold‐change approach by incorporating both the fold‐change direction as well as the p‐value. In addition, the second novel method, Π‐ranking, considers the ratio of the fold‐change and thus integrates all three parameters. We further improved the latter by introducing our third technique, Σ‐ranking, which combines all three parameters in a balanced nonparametric approach.     

Drought effect on plant biomass allocation: A meta‐analysis

Drought is one of the abiotic stresses controlling plant function and ecological stability. In the context of climate change, drought is predicted to occur more frequently in the future. Despite numerous attempts to clarify the overall effects of drought stress on the growth and physiological processes of plants, a comprehensive evaluation on the impacts of drought stress on biomass allocation, especially on reproductive tissues, remains elusive. We conducted a meta‐analysis by synthesizing 164 published studies to elucidate patterns of plant biomass allocation in relation to drought stress. Results showed that drought significantly increased the fraction of root mass but decreased that of stem, leaf, and reproductive mass. Roots of herbaceous plants were more sensitive to drought than woody plants that reduced reproductive allocation more sharply than the former. Relative to herbaceous plants, drought had a more negative impact on leaf mass fraction of woody plants. Among the herbaceous plants, roots of annuals responded to drought stress more strongly than perennial herbs, but their reproductive allocation was less sensitive to drought than the perennial herbs. In addition, cultivated and wild plants seemed to respond to drought stress in a similar way. Drought stress did not change the scaling exponents of the allometric relationship between different plant tissues. These findings suggest that the allometric partitioning theory, rather than the optimal partitioning theory, better explains the drought‐induced changes in biomass allocation strategies.     

Methane and nitrous oxide emissions under no‐till farming in China: a meta‐analysis

No‐till (NT) practices are among promising options toward adaptation and mitigation of climate change. However, the mitigation effectiveness of NT depends not only on its carbon sequestration potential but also on soil‐derived CH₄ and N₂O emissions. A meta‐analysis was conducted, using a dataset involving 136 comparisons from 39 studies in China, to identify site‐specific factors which influence CH₄ emission, CH₄ uptake, and N₂O emission under NT. Comparative treatments involved NT without residue retention (NT0), NT with residue retention (NTR), compared to plow tillage (PT) with residue removed (PT0). Overall, NT0 significantly decreased CH₄ emission by ~30% (P < 0.05) compared to PT0 with an average emission 218.8 kg ha⁻¹ for rice paddies. However, the increase in N₂O emission could partly offset the benefits of the decrease in CH₄ emission under NT compared to PT0. NTR significantly enhanced N₂O emission by 82.1%, 25.5%, and 20.8% (P < 0.05) compared to PT0 for rice paddies, acid soils, and the first 5 years of the experiments, respectively. The results from categorical meta‐analysis indicated that the higher N₂O emission could be mitigated by adopting NT within alkaline soils, for long‐term duration, and with less N fertilization input when compared to PT0. In addition, the natural log (lnR) of response ratio of CH₄ and N₂O emissions under NT correlated positively (enhancing emission) with climate factors (temperature and precipitation) and negatively (reducing emission) with experimental duration, suggesting that avoiding excess soil wetness and using NT for a long term could enhance the benefits of NT. Therefore, a thorough understanding of the conditions favoring greenhouse gas(es) reductions is essential to achieving climate change mitigation and advancing food security in China.