Bootstrap-Based Between-Study Heterogeneity Tests in Meta-Analysis.

Meta-analysis combines pertinent information from existing studies to provide an overall estimate of population parameters/effect sizes, as well as to quantify and explain the differences between studies. However, testing between-study heterogeneity is one of the most challenging tasks in meta-analysis research. Existing methods for testing heterogeneity, such as the Q test and likelihood ratio (LR) test, have been criticized for their failure to control Type I error rate and/or failure to attain enough statistical power. Although better reference distribution approximations have been proposed in the literature, their application is limited. Additionally, when the interest is to test whether the size of the heterogeneity is larger than a specific level, existing methods are far from mature. To address these issues, we propose new heterogeneity tests. Specifically, we combine bootstrap methods with existing heterogeneity tests (i.e., the maximum LR test, the restricted maximum LR test, and the Q test) to overcome the reference distribution issue and denote them as B-ML-LRT, B-REML-LRT, and B-Q, respectively. Simulation studies were conducted to examine and compare the performance of the proposed methods with the regular LR test, the regular Q test, and the Kulinskaya's improved Q test in both random- and mixed-effects meta-analyses. Based on the results of Type I error rates and statistical power, B-REML-LRT is recommended. Additionally, the improved Q test is also recommended when it is applicable. An R package boot.heterogeneity is provided to facilitate the implementation of the proposed tests.

Bayesian mendelian randomization with study heterogeneity and data partitioning for large studies.

BACKGROUND: Mendelian randomization (MR) is a useful approach to causal inference from observational studies when randomised controlled trials are not feasible. However, study heterogeneity of two association studies required in MR is often overlooked. When dealing with large studies, recently developed Bayesian MR can be computationally challenging, and sometimes even prohibitive. METHODS: We addressed study heterogeneity by proposing a random effect Bayesian MR model with multiple exposures and outcomes. For large studies, we adopted a subset posterior aggregation method to overcome the problem of computational expensiveness of Markov chain Monte Carlo. In particular, we divided data into subsets and combined estimated causal effects obtained from the subsets. The performance of our method was evaluated by a number of simulations, in which exposure data was partly missing. RESULTS: Random effect Bayesian MR outperformed conventional inverse-variance weighted estimation, whether the true causal effects were zero or non-zero. Data partitioning of large studies had little impact on variations of the estimated causal effects, whereas it notably affected unbiasedness of the estimates with weak instruments and high missing rate of data. For the cases being simulated in our study, the results have indicated that the "divide (data) and combine (estimated subset causal effects)" can help improve computational efficiency, for an acceptable cost in terms of bias in the causal effect estimates, as long as the size of the subsets is reasonably large. CONCLUSIONS: We further elaborated our Bayesian MR method to explicitly account for study heterogeneity. We also adopted a subset posterior aggregation method to ease computational burden, which is important especially when dealing with large studies. Despite the simplicity of the model we have used in the simulations, we hope the present work would effectively point to MR studies that allow modelling flexibility, especially in relation to the integration of heterogeneous studies and computational practicality.

Jackknife empirical likelihood confidence intervals for assessing heterogeneity in meta-analysis of rare binary event data.

In meta-analysis, the heterogeneity of effect sizes across component studies is typically described by a variance parameter in a random-effects (Re) model. In the literature, methods for constructing confidence intervals (CIs) for the parameter often assume that study-level effect sizes are normally distributed. However, this assumption might be violated in practice, especially in meta-analysis of rare binary events. We propose to use jackknife empirical likelihood (JEL), a nonparametric approach that uses jackknife pseudo-values, to construct CIs for the heterogeneity parameter. To compute jackknife pseudo-values, we employ a moment-based estimator and consider two commonly used weighing schemes (i.e., equal and inverse variance weights). We prove that with each scheme, the resulting log empirical likelihood ratio follows a chi-square distribution asymptotically. We further examine the performance of the proposed JEL methods and compare them with existing CIs through simulation studies and data examples that focus on data of rare binary events. Our numerical results suggest that the JEL method with equal weights compares favorably to alternatives, especially when (observed) effect sizes are non-normal and the number of component studies is large. Thus, it is worth serious consideration in statistical inference.

Bounds for the weight of external data in shrinkage estimation.

Shrinkage estimation in a meta-analysis framework may be used to facilitate dynamical borrowing of information. This framework might be used to analyze a new study in the light of previous data, which might differ in their design (e.g., a randomized controlled trial and a clinical registry). We show how the common study weights arise in effect and shrinkage estimation, and how these may be generalized to the case of Bayesian meta-analysis. Next we develop simple ways to compute bounds on the weights, so that the contribution of the external evidence may be assessed a priori. These considerations are illustrated and discussed using numerical examples, including applications in the treatment of Creutzfeldt-Jakob disease and in fetal monitoring to prevent the occurrence of metabolic acidosis. The target study's contribution to the resulting estimate is shown to be bounded below. Therefore, concerns of evidence being easily overwhelmed by external data are largely unwarranted.

Comparison of commonly used methods in random effects meta-analysis: application to preclinical data in drug discovery research.

BACKGROUND: Meta-analysis of preclinical data is used to evaluate the consistency of findings and to inform the design and conduct of future studies. Unlike clinical meta-analysis, preclinical data often involve many heterogeneous studies reporting outcomes from a small number of animals. Here, we review the methodological challenges in preclinical meta-analysis in estimating and explaining heterogeneity in treatment effects. METHODS: Assuming aggregate-level data, we focus on two topics: (1) estimation of heterogeneity using commonly used methods in preclinical meta-analysis: method of moments (DerSimonian and Laird; DL), maximum likelihood (restricted maximum likelihood; REML) and Bayesian approach; (2) comparison of univariate versus multivariable meta-regression for adjusting estimated treatment effects for heterogeneity. Using data from a systematic review on the efficacy of interleukin-1 receptor antagonist in animals with stroke, we compare these methods, and explore the impact of multiple covariates on the treatment effects. RESULTS: We observed that the three methods for estimating heterogeneity yielded similar estimates for the overall effect, but different estimates for between-study variability. The proportion of heterogeneity explained by a covariate is estimated larger using REML and the Bayesian method as compared with DL. Multivariable meta-regression explains more heterogeneity than univariate meta-regression. CONCLUSIONS: Our findings highlight the importance of careful selection of the estimation method and the use of multivariable meta-regression to explain heterogeneity. There was no difference between REML and the Bayesian method and both methods are recommended over DL. Multiple meta-regression is worthwhile to explain heterogeneity by more than one variable, reducing more variability than any univariate models and increasing the explained proportion of heterogeneity.

Dynamically borrowing strength from another study through shrinkage estimation.

Meta-analytic methods may be used to combine evidence from different sources of information. Quite commonly, the normal-normal hierarchical model (NNHM) including a random-effect to account for between-study heterogeneity is utilized for such analyses. The same modeling framework may also be used to not only derive a combined estimate, but also to borrow strength for a particular study from another by deriving a shrinkage estimate. For instance, a small-scale randomized controlled trial could be supported by a non-randomized study, e.g. a clinical registry. This would be particularly attractive in the context of rare diseases. We demonstrate that a meta-analysis still makes sense in this extreme case, effectively based on a synthesis of only two studies, as illustrated using a recent trial and a clinical registry in Creutzfeld-Jakob disease. Derivation of a shrinkage estimate within a Bayesian random-effects meta-analysis may substantially improve a given estimate even based on only a single additional estimate while accounting for potential effect heterogeneity between the studies. Alternatively, inference may equivalently be motivated via a model specification that does not require a common overall mean parameter but considers the treatment effect in one study, and the difference in effects between the studies. The proposed approach is quite generally applicable to combine different types of evidence originating, e.g. from meta-analyses or individual studies. An application of this more general setup is provided in immunosuppression following liver transplantation in children.

The Safety of Local Hormonal Treatment for Vulvovaginal Atrophy in Women With Estrogen Receptor-positive Breast Cancer Who Are on Adjuvant Aromatase Inhibitor Therapy: Meta-analysis.

Atrophic vaginitis is a relatively common adverse effect of aromatase inhibitors used as an adjunctive treatment for breast cancer. Vaginal estrogen therapy is a treatment option, but the safety of its use in estrogen receptor-positive breast cancer remains understudied. The aim of our study was to determine the safety of local hormonal treatment of vulvovaginal atrophy in women treated with aromatase inhibitors. Our meta-analysis was based on a systematic search of the literature and selection of high-quality evidence. The safety of local hormonal therapy of vaginal atrophy in women on aromatase inhibitors were summarized using calculators built by the authors; heterogeneity was assessed by the Cochrane Q test and I2 values. Several types of bias were assessed; publication bias was calculated by a funnel plot and the Egger regression. Eleven studies fulfilled the inclusion criteria for our study. After 8 weeks of local hormonal treatment, there was no change in the serum levels of luteinizing hormone and estradiol, whereas sex hormone binding globulins were low, and follicle stimulating hormone was almost doubled compared with the baseline. Adverse effect rates of vaginal discharge, facial hair growth, urinary tract or yeast infection, and vaginal or vulvar itching and/or irritation did not show significant changes in the sensitivity analysis, with exception of a single trial. Current evidence suggests that vaginal estrogen administration in postmenopausal women with a history of breast cancer is not associated with systemic absorption of sex hormones and may provide indirect evidence for the safety of their use.

Classical and Bayesian random-effects meta-analysis models with sample quality weights in gene expression studies.

BACKGROUND: Random-effects (RE) models are commonly applied to account for heterogeneity in effect sizes in gene expression meta-analysis. The degree of heterogeneity may differ due to inconsistencies in sample quality. High heterogeneity can arise in meta-analyses containing poor quality samples. We applied sample-quality weights to adjust the study heterogeneity in the DerSimonian and Laird (DSL) and two-step DSL (DSLR2) RE models and the Bayesian random-effects (BRE) models with unweighted and weighted data, Gibbs and Metropolis-Hasting (MH) sampling algorithms, weighted common effect, and weighted between-study variance. We evaluated the performance of the models through simulations and illustrated application of the methods using Alzheimer's gene expression datasets. RESULTS: Sample quality adjusting within study variance (wP6) models provided an appropriate reduction of differentially expressed (DE) genes compared to other weighted functions in classical RE models. The BRE model with a uniform(0,1) prior was appropriate for detecting DE genes as compared to the models with other prior distributions. The precision of DE gene detection in the heterogeneous data was increased with the DSLR2wP6 weighted model compared to the DSLwP6 weighted model. Among the BRE weighted models, the wP6weighted- and unweighted-data models and both Gibbs- and MH-based models performed similarly. The wP6 weighted common-effect model performed similarly to the unweighted model in the homogeneous data, but performed worse in the heterogeneous data. The wP6weighted data were appropriate for detecting DE genes with high precision, while the wP6weighted between-study variance models were appropriate for detecting DE genes with high overall accuracy. Without the weight, when the number of genes in microarray increased, the DSLR2 performed stably, while the overall accuracy of the BRE model was reduced. When applying the weighted models in the Alzheimer's gene expression data, the number of DE genes decreased in all metadata sets with the DSLR2wP6weighted and the wP6weighted between study variance models. Four hundred and forty-six DE genes identified by the wP6weighted between study variance model could be potentially down-regulated genes that may contribute to good classification of Alzheimer's samples. CONCLUSIONS: The application of sample quality weights can increase precision and accuracy of the classical RE and BRE models; however, the performance of the models varied depending on data features, levels of sample quality, and adjustment of parameter estimates.

The alfa and beta of tumours: a review of parameters of the linear-quadratic model, derived from clinical radiotherapy studies.

BACKGROUND: Prediction of radiobiological response is a major challenge in radiotherapy. Of several radiobiological models, the linear-quadratic (LQ) model has been best validated by experimental and clinical data. Clinically, the LQ model is mainly used to estimate equivalent radiotherapy schedules (e.g. calculate the equivalent dose in 2 Gy fractions, EQD2), but increasingly also to predict tumour control probability (TCP) and normal tissue complication probability (NTCP) using logistic models. The selection of accurate LQ parameters α, ß and α/ß is pivotal for a reliable estimate of radiation response. The aim of this review is to provide an overview of published values for the LQ parameters of human tumours as a guideline for radiation oncologists and radiation researchers to select appropriate radiobiological parameter values for LQ modelling in clinical radiotherapy. METHODS AND MATERIALS: We performed a systematic literature search and found sixty-four clinical studies reporting α, ß and α/ß for tumours. Tumour site, histology, stage, number of patients, type of LQ model, radiation type, TCP model, clinical endpoint and radiobiological parameter estimates were extracted. Next, we stratified by tumour site and by tumour histology. Study heterogeneity was expressed by the I2 statistic, i.e. the percentage of variance in reported values not explained by chance. RESULTS: A large heterogeneity in LQ parameters was found within and between studies (I2 > 75%). For the same tumour site, differences in histology partially explain differences in the LQ parameters: epithelial tumours have higher α/ß values than adenocarcinomas. For tumour sites with different histologies, such as in oesophageal cancer, the α/ß estimates correlate well with histology. However, many other factors contribute to the study heterogeneity of LQ parameters, e.g. tumour stage, type of LQ model, TCP model and clinical endpoint (i.e. survival, tumour control and biochemical control). CONCLUSIONS: The value of LQ parameters for tumours as published in clinical radiotherapy studies depends on many clinical and methodological factors. Therefore, for clinical use of the LQ model, LQ parameters for tumour should be selected carefully, based on tumour site, histology and the applied LQ model. To account for uncertainties in LQ parameter estimates, exploring a range of values is recommended.

Meta-analysis approaches to combine multiple gene set enrichment studies.

In the field of gene set enrichment analysis (GSEA), meta-analysis has been used to integrate information from multiple studies to present a reliable summarization of the expanding volume of individual biomedical research, as well as improve the power of detecting essential gene sets involved in complex human diseases. However, existing methods, Meta-Analysis for Pathway Enrichment (MAPE), may be subject to power loss because of (1) using gross summary statistics for combining end results from component studies and (2) using enrichment scores whose distributions depend on the set sizes. In this paper, we adapt meta-analysis approaches recently developed for genome-wide association studies, which are based on fixed effect and random effects (RE) models, to integrate multiple GSEA studies. We further develop a mixed strategy via adaptive testing for choosing RE versus FE models to achieve greater statistical efficiency as well as flexibility. In addition, a size-adjusted enrichment score based on a one-sided Kolmogorov-Smirnov statistic is proposed to formally account for varying set sizes when testing multiple gene sets. Our methods tend to have much better performance than the MAPE methods and can be applied to both discrete and continuous phenotypes. Specifically, the performance of the adaptive testing method seems to be the most stable in general situations.

Discussion on Quantifying publication bias in meta-analysis.

In this discussion, I will describe some issues that are related to the article presented by Lin and Chu. In particular, I discuss three concerns that should be addressed before their methodology may be accepted for general use.

The prevalence of suicidal thoughts and behaviours among college students: a meta-analysis.

BACKGROUND: Adolescence and young adulthood carry risk for suicidal thoughts and behaviours (STB). An increasing subpopulation of young people consists of college students. STB prevalence estimates among college students vary widely, precluding a validated point of reference. In addition, little is known on predictors for between-study heterogeneity in STB prevalence. METHODS: A systematic literature search identified 36 college student samples that were assessed for STB outcomes, representing a total of 634 662 students [median sample size = 2082 (IQR 353-5200); median response rate = 74% (IQR 37-89%)]. We used random-effects meta-analyses to obtain pooled STB prevalence estimates, and multivariate meta-regression models to identify predictors of between-study heterogeneity. RESULTS: Pooled prevalence estimates of lifetime suicidal ideation, plans, and attempts were 22.3% [95% confidence interval (CI) 19.5-25.3%], 6.1% (95% CI 4.8-7.7%), and 3.2% (95% CI 2.2-4.5%), respectively. For 12-month prevalence, this was 10.6% (95% CI 9.1-12.3%), 3.0% (95% CI 2.1-4.0%), and 1.2% (95% CI 0.8-1.6%), respectively. Measures of heterogeneity were high for all outcomes (I 2 = 93.2-99.9%), indicating substantial between-study heterogeneity not due to sampling error. Pooled estimates were generally higher for females, as compared with males (risk ratios in the range 1.12-1.67). Higher STB estimates were also found in samples with lower response rates, when using broad definitions of suicidality, and in samples from Asia. CONCLUSIONS: Based on the currently available evidence, STB seem to be common among college students. Future studies should: (1) incorporate refusal conversion strategies to obtain adequate response rates, and (2) use more fine-grained measures to assess suicidal ideation.

Meta-analysis of few small studies in orphan diseases.

Meta-analyses in orphan diseases and small populations generally face particular problems, including small numbers of studies, small study sizes and heterogeneity of results. However, the heterogeneity is difficult to estimate if only very few studies are included. Motivated by a systematic review in immunosuppression following liver transplantation in children, we investigate the properties of a range of commonly used frequentist and Bayesian procedures in simulation studies. Furthermore, the consequences for interval estimation of the common treatment effect in random-effects meta-analysis are assessed. The Bayesian credibility intervals using weakly informative priors for the between-trial heterogeneity exhibited coverage probabilities in excess of the nominal level for a range of scenarios considered. However, they tended to be shorter than those obtained by the Knapp-Hartung method, which were also conservative. In contrast, methods based on normal quantiles exhibited coverages well below the nominal levels in many scenarios. With very few studies, the performance of the Bayesian credibility intervals is of course sensitive to the specification of the prior for the between-trial heterogeneity. In conclusion, the use of weakly informative priors as exemplified by half-normal priors (with a scale of 0.5 or 1.0) for log odds ratios is recommended for applications in rare diseases. © 2016 The Authors. Research Synthesis Methods published by John Wiley & Sons Ltd.

Meta-analysis of two studies in the presence of heterogeneity with applications in rare diseases.

Random-effects meta-analyses are used to combine evidence of treatment effects from multiple studies. Since treatment effects may vary across trials due to differences in study characteristics, heterogeneity in treatment effects between studies must be accounted for to achieve valid inference. The standard model for random-effects meta-analysis assumes approximately normal effect estimates and a normal random-effects model. However, standard methods based on this model ignore the uncertainty in estimating the between-trial heterogeneity. In the special setting of only two studies and in the presence of heterogeneity, we investigate here alternatives such as the Hartung-Knapp-Sidik-Jonkman method (HKSJ), the modified Knapp-Hartung method (mKH, a variation of the HKSJ method) and Bayesian random-effects meta-analyses with priors covering plausible heterogeneity values; R code to reproduce the examples is presented in an appendix. The properties of these methods are assessed by applying them to five examples from various rare diseases and by a simulation study. Whereas the standard method based on normal quantiles has poor coverage, the HKSJ and mKH generally lead to very long, and therefore inconclusive, confidence intervals. The Bayesian intervals on the whole show satisfying properties and offer a reasonable compromise between these two extremes.

Detection of human papillomavirus in laryngeal squamous cell carcinoma: Systematic review and meta-analysis.

BACKGROUND: Recent studies have reported a human papillomavirus (HPV) prevalence of 20% to 30% in laryngeal squamous cell carcinoma (LSCC), although clinical data on HPV involvement remain largely inconsistent, ascribed by some to differences in HPV detection methods or in geographic origin of the studies. OBJECTIVE: To perform a systematic review and formal meta-analysis of the literature reporting on HPV detection in LSCC. METHODS: Literature was searched from January 1964 until March 2015. The effect size was calculated as event rates (95% confidence interval [CI]), with homogeneity testing using Cochran's Q and I(2) statistics. Meta-regression was used to test the impact of study-level covariates (HPV detection method, geographic origin) on effect size. Potential publication bias was estimated using funnel plot symmetry. RESULTS: One hundred seventy nine studies were eligible, comprising a sample size of 7,347 LSCCs from different geographic regions. Altogether, 1,830 (25%) cases tested HPV-positive considering all methods, with effect size of 0.269 (95% CI: 0.242 to 0.297; random-effects model). In meta-analysis stratified by the 1) HPV detection technique and 2) geographic study origin, the between-study heterogeneity was significant only for geographic origin (P = .0001). In meta-regression, the HPV detection method (P = .876) or geographic origin (P = .234) were not significant study-level covariates. Some evidence for publication bias was found only for studies from North America and those using non-polymerase chain reaction methods, with a marginal effect on adjusted point estimates for both. CONCLUSIONS: Variability in HPV detection rates in LSCC is explained by geographic origin of study but not by HPV detection method. However, they were not significant study-level covariates in formal meta-regression. LEVEL OF EVIDENCE: NA.

Association of IL-6 polymorphisms with chronic obstructive pulmonary disease risk: meta-analysis of genetic association, gene expression and expression quantitative trait locus analysis.

AIM: IL-6 might play an important role in the mechanism of chronic obstructive pulmonary disease (COPD). This study assessed the relationship of rs1800796 and rs1800797 of IL-6 with COPD. MATERIALS & METHODS: We conducted meta-analysis and gene expression analysis using published datasets to examine the associations between IL-6 SNPs and COPD. RESULTS: rs1800796 was significantly associated with COPD, yielding a pooled odds ratio of 0.52 (95% CI: 0.33-0.84; p = 0.007), and showed cis-expression quantitative trait locus associations (p = 0.02148). Differential gene expression analysis found that IL-6 was upregulated in COPD cases compared with controls. The associations of rs1800797 with COPD were not significant. CONCLUSION: The findings showed that rs1800796 was associated with COPD in Europeans and might affect COPD risk through disturbing IL-6 gene expression.

Small studies are more heterogeneous than large ones: a meta-meta-analysis.

OBJECTIVES: Between-study heterogeneity plays an important role in random-effects models for meta-analysis. Most clinical trials are small, and small trials are often associated with larger effect sizes. We empirically evaluated whether there is also a relationship between trial size and heterogeneity (τ). STUDY DESIGN AND SETTING: We selected the first meta-analysis per intervention review of the Cochrane Database of Systematic Reviews Issues 2009-2013 with a dichotomous (n = 2,009) or continuous (n = 1,254) outcome. The association between estimated τ and trial size was evaluated across meta-analyses using regression and within meta-analyses using a Bayesian approach. Small trials were predefined as those having standard errors (SEs) over 0.2 standardized effects. RESULTS: Most meta-analyses were based on few (median 4) trials. Within the same meta-analysis, the small study τS(2) was larger than the large-study τL(2) [average ratio 2.11; 95% credible interval (1.05, 3.87) for dichotomous and 3.11 (2.00, 4.78) for continuous meta-analyses]. The imprecision of τS was larger than of τL: median SE 0.39 vs. 0.20 for dichotomous and 0.22 vs. 0.13 for continuous small-study and large-study meta-analyses. CONCLUSION: Heterogeneity between small studies is larger than between larger studies. The large imprecision with which τ is estimated in a typical small-studies' meta-analysis is another reason for concern, and sensitivity analyses are recommended.

The association between polymorphism of P53 codon 72 Arg/Pro and hepatocellular carcinoma susceptibility: evidence from a meta-analysis of 15 studies with 3704 cases.

BACKGROUND: Emerging evidence has shown that p53gene participates in human carcinogenesis as tumor suppressors. Polymorphism of p53 gene codon 72 Arg/Pro (rs1042522) may influence the function of p53 protein and then affect the processing of carcinogenesis. It has been suggested that p53 codon 72 Arg/Pro polymorphism is associated with susceptibility to hepatocellular carcinoma (HCC). However, published results are inconsistent and inconclusive. To examine the validity of the association between the polymorphism and HCC risk, we performed this meta-analysis. METHODOLOGY/PRINCIPAL FINDINGS: We have conducted a search of case-control studies on the associations of p53 codon 72 polymorphism with susceptibility to HCC in PubMed, ScienceDirect, Bio-Med central, Springer-link, EBSCO, Wanfang databases and Chinese National Knowledge Infrastructure (CNKI) databases. A total of 15 studies were identified with 3704 cases and 4559 controls for codon 72 Arg/Pro polymorphism. The result did support a significant genetic association between Pro allele and susceptibility to HCC in all the genetic models. Similarly, subgroup analysis showed significant associations between the Arg/Pro polymorphism and susceptibility to HCC when stratifying by race, gender, source of controls and hepatitis virus infection status. CONCLUSIONS/SIGNIFICANCE: This meta-analysis suggests that p53 codon 72 Arg/Pro polymorphism may be associated with the risk of HCC, especially in subgroup analysis of Asian and Caucasian population, hospital-based population, the female, and the individuals infected with hepatitis virus. However, well-designed studies based on different ethnic groups with larger sample size and more detailed data are needed to confirm these conclusions.

Solitary bronchial squamous cell papilloma - another human papillomavirus (HPV)-associated benign tumor: systematic review and meta-analysis.

AIM OF THE STUDY: To perform a systematic review and formal meta-analysis of the literature reporting on HPV detection in bronchial squamous cell papillomas (SCP). MATERIAL AND METHODS: The literature was searched up to June 2012. The effect size was calculated as event rate (95% CI), with homogeneity testing using Cochran's Q and I(2) statistics. Meta-regression was used to test the impact of study-level covariates (HPV detection method, geographic origin) on effect size, and potential publication bias was estimated using funnel plot symmetry. RESULTS: Fifteen studies were eligible, covering 89 bronchial SCPs from different geographic regions. Altogether, 38 (42.7%) cases tested HPV-positive; effect size 0.422 (95% CI: 0.311-0.542; fixed effects model), and 0.495 (95% CI: 0.316-0.675; random effects model). In meta-analysis stratified by i) HPV detection technique and ii) geographic study origin, the between-study heterogeneity was not significant for either; p = 0.348, and p = 0.792, respectively. In maximum likelihood meta-regression, HPV detection method (p = 0.150) and geographic origin of the study (p = 0.164) were not significant study-level covariates. Some evidence for publication bias was found only among in situ hybridization (ISH)-based studies and among studies from Europe, but with a negligible effect on summary effect size estimates. In sensitivity analysis, the meta-analytic results were robust to all one-by-one study removals. CONCLUSIONS: In formal meta-regression, the variability in HPV detection rates reported in bronchial SCPs is not explained by the HPV detection method or geographic origin of the study.