Please activate JavaScript!
Please install Adobe Flash Player, click here for download

ISCB2014_abstract_book

128 ISCB 2014 Vienna, Austria • Abstracts - Poster PresentationsWednesday, 27th August 2014 • 15:30-16:00 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust Wednesday, 27th August 2014 - 15.30-16.00 Poster session P4 P4.1 Meta-analysis and network meta-analysis P4.1.4 Developing and validating risk prediction models in an individual participant data meta-analysis I Ahmed1 , TPA Debray2 , KGM Moons2 , RD Riley1 1 University of Birmingham, Birmingham, United Kingdom, 2 University Medical Center Utrecht, Utrecht, The Netherlands   Background: Risk prediction models estimate the risk of developing future outcomes for individuals based on one or more underlying char- acteristics (predictors). We review how researchers develop and validate risk prediction models within an individual participant data (IPD) meta- analysis, in order to assess the feasibility and conduct of the approach. Methods: A qualitative review of the aims, methodology, and reporting in 15 articles that developed a risk prediction model using IPD from multiple studies. Results: The IPD approach offers many opportunities but methodological challenges exist, including: unavailability of requested IPD, missing pa- tient data and predictors, and between-study heterogeneity in methods of measurement, outcome definitions and predictor effects. Most articles develop their model using IPD from all available studies and perform only an internal validation (on the same set of data). Ten of the 15 articles did not allow for any study differences in baseline risk (intercepts), potentially limiting their model’s applicability and performance in some populations. Only two articles used external validation (on different data), including a novel method which develops the model on all but one of the IPD stud- ies, tests performance in the excluded study, and repeats by rotating the omitted study. Conclusions: An IPD meta-analysis offers unique opportunities for risk prediction research. Researchers can make more of this by allowing sepa- rate model intercept terms for each study (population) to improve gen- eralisability, and by using‘internal-external cross-validation’to simultane- ously develop and validate their model. Methodological challenges can be reduced by prospectively planned collaborations that share IPD for risk prediction.   P4.1.38 Meta-analysis of single-arm survival studies: a distribution-free approach for estimating summary survival curves with random effects C Combescure1 , Y Foucher2 , D Jackson3 1 University of Geneva & University Hospitals of Geneva, Geneva, Switzerland, 2 University of Nantes, Nantes, France, 3 Institute of Public Health, Cambridge, United Kingdom   Regression models have been proposed to combine survival curves from various studies in a single summary survival curve. This approach is useful for meta-analysis of survival studies. We propose an alternative approach, which is distribution-free, for meta-analysis of single-arm survival studies. It is an extension of the product-limit estimator of survival for aggregate data accounting for between-study variability. The survival estimates are read off from survival curves at a set of time- points. The conditional survival estimates, derived from the collected survival estimates, go through an arc-sine transformation and then are pooled. The between-study variability is accounted by using the exten- sion of the DerSimonian and Laird’s methodology for multiple outcomes. A back transformation is applied to obtain the pooled conditional survival estimates. The pooled survival at each time point is obtained by a product of pooled conditional survival estimates. Statistics I2 and H2 are used to quantify the impact of the heterogeneity in the published survival curves and we propose a statistical test for the between-strata comparison. The performance of the proposed approach was evaluated in a simulation study with various Weibull survival models, sample sizes and censoring rates. This simulation study showed that our approach is biased when events are rare and sample sizes are small. It performs well in other situations. An application of our approach on aggregate data of 27 studies will be given, with the aim to synthesize the survival of untreated patients with hepatocellular carcinoma. P4.1.42 Publication bias tests for survival data: a simulation study TPA Debray1 , KGM Moons1 , H Koffijberg1 , RD Riley2 1 UMC Utrecht, Utrecht, The Netherlands, 2 University of Birmingham, Birmingham, United Kingdom   The presence of publication bias is often verified in meta-analyses by vi- sual inspection of the funnel plot. Statistical tests may estimate the asso- ciation between the reported effect size and their standard error (Egger´s test), total sample size (Macaskill´s test) or inverse of the total sample size (Peter´s test). Although these tests have been evaluated for pooling odds ratios, their application may be less appropriate for survival data where censoring influences statistical significance (and thus selective reporting) of the hazard ratio. Here, we propose and evaluate two new publication bias tests for survival data that are based on the total number of events (Test-1) and the total survival time (Test-2). We compare their performance to existing tests in an extensive simulation study where more than 20,000,000 meta-analyses were generated. Here, we varied the true hazard ratio (HR=0.20-1.00), the number of available studies (N=10-100), the censoring proportion (cp=0.10-0.50) and the scale of the hazard distribution. Furthermore, we used a set of predefined reflecting meta-analyses of randomised clinical trials in the medical literature. When treatment groups are balanced, simulation results demonstrate that type-I errors are problematic for Egger´s test (averaging from 0.110 for N=10 to 0.195 for N=100), but consistently good (around 0.10) for Peter´s test and Test-1. The power of all tests was low; for example Test-1 yielded power from 0.112 (for N=10) to 0.208 (for N=100). Finally, we compare and discuss the performance of Peter´s test and Test-1 in imbalanced treat- ment groups, and make recommendations for practice.   P4.1.80 Meta-analysis of mobile phone-based interventions for smoking cessation trials in different countries Y Jiang1 , Y Huang1 , M Ybarra2 , L Abroms3 , C Free4 , R Whittaker5 1 The University of Auckland, Auckland, New Zealand, 2 Center for Innovative Public Health, San Clemente, United States, 3 George Washington University, Washington, United States, 4 London School of Hygiene & Tropical Medicine, London, United Kingdom, 5 National Institute for Health Innovation, Auckland, New Zealand   Tobacco use contributes to 12% of all deaths among adults ages 30 years and older. Smoking cessation programs delivered via mobile phone text messaging are used worldwide to help people quit smoking. Although re- sults suggest increases in self-reported quitting, at least in the short-term, little is known how this relates to different cultures. Individual participants’ data (N=8,315) collected in five randomised con- trol trials in New Zealand, UK, USA and Turkey were included for meta-

Pages Overview