Meta-regression with partial information on summary trial or patient characteristics

Combining randomized and non-randomized data to predict heterogeneous effects of competing treatments

Some patients benefit from a treatment while others may do so less or do not benefit at all. We have previously developed a two-stage network meta-regression prediction model that synthesized randomized trials and evaluates how treatment effects vary across patient characteristics. In this article, we extended this model to combine different sources of types in different formats: aggregate data (AD) and individual participant data (IPD) from randomized and non-randomized evidence. In the first stage, a prognostic model is developed to predict the baseline risk of the outcome using a large cohort study. In the second stage, we recalibrated this prognostic model to improve our predictions for patients enrolled in randomized trials. In the third stage, we used the baseline risk as effect modifier in a network meta-regression model combining AD, IPD randomized clinical trial to estimate heterogeneous treatment effects. We illustrated the approach in the re-analysis of a network of studies comparing three drugs for relapsing-remitting multiple sclerosis. Several patient characteristics influence the baseline risk of relapse, which in turn modifies the effect of the drugs. The proposed model makes personalized predictions for health outcomes under several treatment options and encompasses all relevant randomized and non-randomized evidence.

Meta-Analysis with Few Studies and Binary Data: A Bayesian Model Averaging Approach

In meta-analysis, the existence of between-sample heterogeneity introduces model uncertainty, which must be incorporated into the inference. We argue that an alternative way to measure this heterogeneity is by clustering the samples and then determining the posterior probability of the cluster models. The meta-inference is obtained as a mixture of all the meta-inferences for the cluster models, where the mixing distribution is the posterior model probabilities. When there are few studies, the number of cluster configurations is manageable, and the meta-inferences can be drawn with BMA techniques. Although this topic has been relatively neglected in the meta-analysis literature, the inference thus obtained accurately reflects the cluster structure of the samples used. In this paper, illustrative examples are given and analysed, using real binary data.

Bivariate evaluation of thromboembolism and bleeding in clinical trials of anticoagulants in patients with atrial fibrillation

Clinical trials of antithrombotic therapy require a cohesive assessment of benefit and risk. A new graphical method to represent the bivariate relation of benefit and risk in trials of antithrombotic drugs is described and illustrated using published data from the four major registration clinical trials of non-vitamin K oral anticoagulants (NOACs) totalling 71,683 patients for prevention of thromboembolic events (TE) in patients with atrial fibrillation (RE-LY, ROCKET AF, ARISTOTLE, and ENGAGE-AF TIMI48). A curve representing a null hypothesis defines a region of benefit on a two-dimensional plane. Trial results are summarised by a rectangle defined by standard 95 % confidence intervals (CI) for thrombosis and bleeding risks. Benefit is judged by whether the confidence rectangle contains the null curve. The treatment effect is measured by the distance from the null curve to the opposing corners of the confidence rectangle (termed “corner distance (CD)”). Across trials NOACs reduced the absolute risk of TE compared to warfarin by 0.30 % (95 % CI: –0.56 % to –0.05 %) and reduced major bleeding by 0.88 % (95 % CI: –1.26 % to –0.51 %). Bivariate evaluation showed NOAC superiority to warfarin overall and elucidated dose differences; low dose edoxaban increased bivariate TE-bleeding risk 0.08 % (CD = –0.85 % to 0.78 %), whereas high dose edoxaban reduced risk 1.41 % (CD = –2.07 % to –0.70 %). In conclusion, bivariate evaluation facilitates visual assessment of the safety-efficacy profile of antithrombotic drugs. Its application to trials in atrial fibrillation found NOACs superior to warfarin without substantial differences between agents.

Supplementary Material to this article is available online at www.thrombosis-online.com.

Objective Bayesian meta-analysis for sparse discrete data

This paper presents a Bayesian model for meta-analysis of sparse discrete binomial data, which are out of the scope of the usual hierarchical normal random-effect models. Treatment effectiveness data are often of this type. The crucial linking distribution between the effectiveness conditional on the healthcare center and the unconditional effectiveness is constructed from specific bivariate classes of distributions with given marginals. This assures coherency between the marginal and conditional prior distributions utilized in the analysis. Further, we impose a bivariate class of priors that is able to accommodate a wide range of heterogeneity degrees between the multicenter clinical trials involved. Applications to real multicenter data are given and compared with previous meta-analysis.

The impact of trial characteristics on premature discontinuation of antipsychotics in schizophrenia

Patient dropout is common in mental health trials. It is important to understand why patients drop out from trials, so that measures can be taken to minimize its occurrence. This research sought to identify trial characteristics that have an impact on premature discontinuation in antipsychotic trials for schizophrenia.

METHODS

Poisson regression analysis was applied with dropout rate per patient-week as the dependent variable and trial characteristics as independent variables. Multinomial logistic regression analysis was performed to examine whether the same characteristics predict whether patients drop out without providing any outcome data and whether they drop out with sufficient early data for a 'last observation carried forward' analysis to be performed.

RESULTS

trials with adequate allocation concealment, double blinding, placebo as control, higher precision, larger trial size, at least three treatment arms, recent publication, conduct in the United States and enrollment of inpatients were all associated with higher dropout rates. Similar factors were associated with whether a patient was more likely to be evaluated at least once, or be excluded entirely from the analysis. However, blinding status did not predict the former type of dropout, and allocation concealment, higher precision and larger sample size, number of arms, recent publication and recruiting inpatient did not predict the latter type of dropout.

CONCLUSIONS

high dropout rates in antipsychotic trials can be associated with various characteristics, and appears to be particularly associated with use of placebo and study size.

Investigating clinical heterogeneity in systematic reviews: a methodologic review of guidance in the literature

BACKGROUND

While there is some consensus on methods for investigating statistical and methodological heterogeneity, little attention has been paid to clinical aspects of heterogeneity. The objective of this study is to summarize and collate suggested methods for investigating clinical heterogeneity in systematic reviews.

METHODS

We searched databases (Medline, EMBASE, CINAHL, Cochrane Library, and CONSORT, to December 2010) and reference lists and contacted experts to identify resources providing suggestions for investigating clinical heterogeneity between controlled clinical trials included in systematic reviews. We extracted recommendations, assessed resources for risk of bias, and collated the recommendations.

RESULTS

One hundred and one resources were collected, including narrative reviews, methodological reviews, statistical methods papers, and textbooks. These resources generally had a low risk of bias, but there was minimal consensus among them. Resources suggested that planned investigations of clinical heterogeneity should be made explicit in the protocol of the review; clinical experts should be included on the review team; a set of clinical covariates should be chosen considering variables from the participant level, intervention level, outcome level, research setting, or others unique to the research question; covariates should have a clear scientific rationale; there should be a sufficient number of trials per covariate; and results of any such investigations should be interpreted with caution.

CONCLUSIONS

Though the consensus was minimal, there were many recommendations in the literature for investigating clinical heterogeneity in systematic reviews. Formal recommendations for investigating clinical heterogeneity in systematic reviews of controlled trials are required.

Meta-analysis methods and models with applications in evaluation of cholesterol-lowering drugs

In this paper, we propose a class of multivariate random effects models allowing for the inclusion of study-level covariates to carry out meta-analyses. As existing algorithms for computing maximum likelihood estimates often converge poorly or may not converge at all when the random effects are multi-dimensional, we develop an efficient expectation-maximization algorithm for fitting multi-dimensional random effects regression models. In addition, we also develop a new methodology for carrying out variable selection with study-level covariates. We examine the performance of the proposed methodology via a simulation study. We apply the proposed methodology to analyze metadata from 26 studies involving statins as a monotherapy and in combination with ezetimibe. In particular, we compare the low-density lipoprotein cholesterol-lowering efficacy of monotherapy and combination therapy on two patient populations (naïve and non-naïve patients to statin monotherapy at baseline), controlling for aggregate covariates. The proposed methodology is quite general and can be applied in any meta-analysis setting for a wide range of scientific applications and therefore offers new analytic methods of clinical importance.

Impact of Intraoperative Stimulation Brain Mapping on Glioma Surgery Outcome: A Meta-Analysis

Purpose

Surgery for infiltrative gliomas aims to balance tumor removal with preservation of functional integrity. The usefulness of intraoperative stimulation mapping (ISM) has not been addressed in randomized trials. This study addresses glioma surgery outcome on the basis of a meta-analysis of observational studies.

Methods

A systematic search retrieved 90 reports published between 1990 and 2010 with 8,091 adult patients who had resective surgery for supratentorial infiltrative glioma, with or without ISM. Quality criteria consisted of postoperative neurologic examination details and follow-up timing. New postoperative neurologic deficits were categorized on the basis of timing and severity. Meta-analysis with a Bayesian random effects model determined summary event rates of deficits as well as gross total resection rate and eloquent locations. Meta-regression analysis explored heterogeneity among studies.

Results

Late severe neurologic deficits were observed in 3.4% (95% CI, 2.3% to 4.8%) of patients after resections with ISM, and in 8.2% (95% CI, 5.7% to 11.4%) of patients after resections without ISM (adjusted odds ratio, 0.39; 95% CI, 0.23 to 0.64). The percentages of radiologically confirmed gross total resections were 75% (95% CI, 66% to 82%) with ISM and 58% (95% CI, 48% to 69%) without ISM. Eloquent locations were involved in 99.9% (95% CI, 99.9% to 100%) of resections with ISM and in 95.8% (95% CI, 73.1% to 99.8%) of resections without ISM. Relevant sources of heterogeneity among studies were ISM, continent, and academic setting.

Conclusion

Glioma resections using ISM are associated with fewer late severe neurologic deficits and more extensive resection, and they involve eloquent locations more frequently. This indicates that ISM should be universally implemented as standard of care for glioma surgery.

Bayesian sensitivity models for missing covariates in the analysis of survival data

RATIONALE

This paper presents a Bayesian approach using WinBUGS for analysing survival data in which observations have missing information on some covariates. Modelling the joint density of the survival time and the covariates allows for the missing covariate data to be missing at random (MAR), as opposed to the more restrictive assumptions imposed by a complete case analysis.

METHODS

Here, the survival times are modelled by the accelerated failure time model and the joint covariate density is factorized as a series of conditional densities, each modelled by logistic regression. We propose a class of models to examine MAR assumption by using Bayesian informative priors or auxiliary data.

RESULTS

In the example considered, the complete case analysis underestimates the proportion of severely impaired cases. Furthermore, models evaluating sensitivity to the MAR assumption suggest that median life expectancies could be increased. Gains in precision are small in the application considered, possibly a reflection of extra uncertainty because of the inclusion of all cases, including those of unknown severity.

CONCLUSIONS

Through simple Bayesian models, more realistic assumptions concerning the nature of the missing data can be made. Implementation in WinBUGS means that this model is accessible to practicing statisticians.