Inference for correlated effect sizes using multiple univariate meta-analyses

Multivariate meta-analysis, which involves jointly analyzing multiple and correlated outcomes from separate studies, has received a great deal of attention. One reason to prefer the multivariate approach is its ability to account for the dependence between multiple estimates from the same study. However, nearly all the existing methods for analyzing multivariate meta-analytic data require the knowledge of the within-study correlations, which are usually unavailable in practice. We propose a simple non-iterative method that can be used for the analysis of multivariate meta-analysis datasets, that has no convergence problems, and does not require the use of within-study correlations. Our approach uses standard univariate methods for the marginal effects but also provides valid joint inference for multiple parameters. The proposed method can directly handle missing outcomes under missing completely at random assumption. Simulation studies show that the proposed method provides unbiased estimates, well-estimated standard errors, and confidence intervals with good coverage probability. Furthermore, the proposed method is found to maintain high relative efficiency compared with conventional multivariate meta-analyses where the within-study correlations are known. We illustrate the proposed method through two real meta-analyses where functions of the estimated effects are of interest.

The effect of shunt surgery on neuropsychological performance in normal pressure hydrocephalus: a systematic review and meta-analysis

We conducted a systematic review of the literature and used meta-analytic techniques to evaluate the impact of shunt surgery on neuropsychological performance in patients with normal pressure hydrocephalus (NPH). Twenty-three studies with 1059 patients were identified for review using PubMed, Web of Science, Google scholar and manual searching. Inclusion criteria were prospective, within-subject investigations of cognitive outcome using neuropsychological assessment before and after shunt surgery in patients with NPH. There were statistically significant effects of shunt surgery on cognition (Mini-Mental State Examination; MMSE), learning and memory (Rey Auditory Verbal Learning Test; RAVLT, total and delayed subtests), executive function (backwards digit span, phonemic verbal fluency, trail making test B) and psychomotor speed (trail making test A) all in the direction of improvement following shunt surgery, but with considerable heterogeneity across all measures. A more detailed examination of the data suggested robust evidence for improved MMSE, RAVLT total, RAVLT delayed, phonemic verbal fluency and trail making test A only. Meta-regressions revealed no statistically significant effect of age, sex or follow-up interval on improvement in the MMSE. Our results suggest that shunt surgery is most sensitive for improving global cognition, learning and memory and psychomotor speed in patients with NPH.

Extending DerSimonian and Laird's methodology to perform network meta-analyses with random inconsistency effects

Network meta-analysis is becoming more popular as a way to compare multiple treatments simultaneously. Here, we develop a new estimation method for fitting models for network meta-analysis with random inconsistency effects. This method is an extension of the procedure originally proposed by DerSimonian and Laird. Our methodology allows for inconsistency within the network. The proposed procedure is semi-parametric, non-iterative, fast and highly accessible to applied researchers. The methodology is found to perform satisfactorily in a simulation study provided that the sample size is large enough and the extent of the inconsistency is not very severe. We apply our approach to two real examples.

Methods to estimate the between-study variance and its uncertainty in meta-analysis

Meta-analyses are typically used to estimate the overall/mean of an outcome of interest. However, inference about between-study variability, which is typically modelled using a between-study variance parameter, is usually an additional aim. The DerSimonian and Laird method, currently widely used by default to estimate the between-study variance, has been long challenged. Our aim is to identify known methods for estimation of the between-study variance and its corresponding uncertainty, and to summarise the simulation and empirical evidence that compares them. We identified 16 estimators for the between-study variance, seven methods to calculate confidence intervals, and several comparative studies. Simulation studies suggest that for both dichotomous and continuous data the estimator proposed by Paule and Mandel and for continuous data the restricted maximum likelihood estimator are better alternatives to estimate the between-study variance. Based on the scenarios and results presented in the published studies, we recommend the Q-profile method and the alternative approach based on a 'generalised Cochran between-study variance statistic' to compute corresponding confidence intervals around the resulting estimates. Our recommendations are based on a qualitative evaluation of the existing literature and expert consensus. Evidence-based recommendations require an extensive simulation study where all methods would be compared under the same scenarios.

Systematic review and meta-analysis of the association between frailty and outcome in surgical patients

INTRODUCTION

Frailty is becoming increasingly prevalent in the elderly population although a lack of consensus regarding a clinical definition hampers comparison of clinical studies. More elderly patients are being assessed for surgical intervention but the effect of frailty on surgical related outcomes is still not clear.

METHODS

A systematic literature search for studies prospectively reporting frailty and postoperative outcomes in patients undergoing surgical intervention was performed with data collated from a total of 12 studies. Random effects meta-analysis modelling was undertaken to estimate the association between frailty and mortality rates (in-hospital and one-year), length of hospital stay and the need for step-down care for further rehabilitation/nursing home placement.

RESULTS

Frailty was associated with a higher in-hospital mortality rate (pooled odds ratio [OR]: 2.77, 95% confidence interval [CI]: 1.62-4.73), a higher one-year mortality rate (pooled OR: 1.99, 95% CI: 1.49-2.66), a longer hospital stay (pooled mean difference: 1.05 days, 95% CI: 0.02-2.07 days) and a higher discharge rate to further rehabilitation/step-down care (pooled OR: 5.71, 95% CI: 3.41-9.55).

CONCLUSIONS

The presence of frailty in patients undergoing surgical intervention is associated with poorer outcomes with regard to mortality and return to independence. Further in-depth studies are required to identify factors that can be optimised to reduce the burden of frailty in surgical patients.

Pattern mixture models for the analysis of repeated attempt designs

It is not uncommon in follow-up studies to make multiple attempts to collect a measurement after baseline. Recording whether these attempts are successful or not provides useful information for the purposes of assessing the missing at random (MAR) assumption and facilitating missing not at random (MNAR) modeling. This is because measurements from subjects who provide this data after multiple failed attempts may differ from those who provide the measurement after fewer attempts. This type of "continuum of resistance" to providing a measurement has hitherto been modeled in a selection model framework, where the outcome data is modeled jointly with the success or failure of the attempts given these outcomes. Here, we present a pattern mixture approach to model this type of data. We re-analyze the repeated attempt data from a trial that was previously analyzed using a selection model approach. Our pattern mixture model is more flexible and is more transparent in terms of parameter identifiability than the models that have previously been used to model repeated attempt data and allows for sensitivity analysis. We conclude that our approach to modeling this type of data provides a fully viable alternative to the more established selection model.

New models for describing outliers in meta-analysis

An unobserved random effect is often used to describe the between‐study variation that is apparent in meta‐analysis datasets. A normally distributed random effect is conventionally used for this purpose. When outliers or other unusual estimates are included in the analysis, the use of alternative random effect distributions has previously been proposed. Instead of adopting the usual hierarchical approach to modelling between‐study variation, and so directly modelling the study specific true underling effects, we propose two new marginal distributions for modelling heterogeneous datasets. These two distributions are suggested because numerical integration is not needed to evaluate the likelihood. This makes the computation required when fitting our models much more robust. The properties of the new distributions are described, and the methodology is exemplified by fitting models to four datasets. © 2015 The Authors. Research Synthesis Methods published by John Wiley & Sons, Ltd.

The design-by-treatment interaction model: a unifying framework for modelling loop inconsistency in network meta-analysis

In this note, we clarify and prove the claim made Higgins et al. (2012) that the design‐by‐treatment interaction model contains all possible loop inconsistency models. This claim provides a strong argument for using the design‐by‐treatment interaction model to describe loop inconsistencies in network meta‐analysis. © 2015 The Authors. Research Synthesis Methods published by John Wiley & Sons, Ltd.

Dextrocardia in Adults with Congenital Heart Disease

BACKGROUND/OBJECTIVES

Dextrocardia is rare in the general population, and may be associated with significant additional cardiac malformations. We aimed to identify the prevalence and patterns of additional cardiac defects, as well as the associated long-term morbidity and mortality, in adult patients with dextrocardia, in a specialised Adult Congenital Heart Disease (ACHD) service.

METHODS

A retrospective study of patients with dextrocardia managed by our tertiary ACHD service, since January 2000, was performed. Medical records were reviewed and the National Death Index was consulted to confirm survival status.

RESULTS

Of 3698 adults in our ACHD Service, 19 (0.5%) had dextrocardia. Mean follow-up duration was 7±7.5 years. The mean age at last review was 36.8±10.5 years (range 20-63 years). Situs was solitus in 14 (74%) and inversus in five (26%). Eleven patients (58%) had functional single ventricles, of whom five had atrioventricular (AV)-ventriculoarterial (VA) discordance and two had VA discordance only. Four patients with two ventricles had AV-VA discordance. All patients had at least one additional cardiac malformation. Fourteen patients (74%) required surgical intervention. Eleven patients (58%) underwent a Fontan-type operation. Five patients (26%) required ablation procedures for arrhythmia. One patient had infective endocarditis and two deaths occurred, both in patients who also had AV-VA discordance.

CONCLUSION

Dextrocardia remains a rare finding in adults, even in a highly select group of patients with known congenital heart disease. Those with associated congenital heart abnormalities are likely to have complex lesions, which may require multiple surgical and medical interventions. Despite this, our series demonstrated that patients surviving to adulthood and then managed in an ACHD centre may have good medium-term survival.

Strain features and distributions in pneumococci from children with invasive disease before and after 13-valent conjugate vaccine implementation in the USA

The effect of second-generation pneumococcal conjugate vaccines on invasive pneumococcal disease (IPD) strain distributions have not yet been well described. We analysed IPD isolates recovered from children aged <5 years through Active Bacterial Core surveillance before (2008–2009; n = 828) and after (2011–2013; n = 600) 13-valent pneumococcal conjugate vaccine (PCV13) implementation. We employed conventional testing, PCR/electrospray ionization mass spectrometry and whole genome sequence (WGS) analysis to identify serotypes, resistance features, genotypes, and pilus types. PCV13, licensed in February 2010, effectively targeted all major 19A and 7F genotypes, and decreased antimicrobial resistance, primarily owing to removal of the 19A/ST320 complex. The strain complex contributing most to the remaining β-lactam resistance during 2011–2013 was 35B/ST558. Significant emergence of non-vaccine clonal complexes was not evident. Because of the removal of vaccine serotype strains, positivity for one or both pilus types (PI-1 and PI-2) decreased in the post-PCV13 years 2011–2013 relative to 2008–2009 (decreases of 32–55% for PI-1, and >95% for PI-2 and combined PI-1 + PI-2). β-Lactam susceptibility phenotypes correlated consistently with transpeptidase region sequence combinations of the three major penicillin-binding proteins (PBPs) determined through WGS analysis. Other major resistance features were predictable by DNA signatures from WGS analysis. Multilocus sequence data combined with PBP combinations identified progeny, serotype donors and recipient strains in serotype switch events. PCV13 decreased the frequency of all PCV13 serotype clones and concurrently decreased the frequency of strain subsets with resistance and/or adherence features conducive to successful carriage. Our results serve as a reference describing key features of current paediatric IPD strains in the USA after PCV13 implementation.

Approximate confidence intervals for moment-based estimators of the between-study variance in random effects meta-analysis

Moment‐based estimators of the between‐study variance are very popular when performing random effects meta‐analyses. This type of estimation has many advantages including computational and conceptual simplicity. Furthermore, by using these estimators in large samples, valid meta‐analyses can be performed without the assumption that the treatment effects follow a normal distribution. Recently proposed moment‐based confidence intervals for the between‐study variance are exact under the random effects model but are quite elaborate. Here, we present a much simpler method for calculating approximate confidence intervals of this type. This method uses variance‐stabilising transformations as its basis and can be used for a very wide variety of moment‐based estimators in both the random effects meta‐analysis and meta‐regression models. © 2015 The Authors. Research Synthesis Methods published by John Wiley & Sons, Ltd.

Why are Children Still Being Infected with HIV? Impact of an Integrated Public Health and Clinical Practice Intervention on Mother-to-Child HIV Transmission in Las Vegas, Nevada, 2007-2012

Background:

During a 9 months period, September 2005 through June 2006, Nevada documented six cases of pediatric HIV acquired through mother-to-child transmission. Subsequently, a community-based approach to the care of women and children living with or exposed to HIV was implemented.

Subjects and Methods:

A detailed review of mother-infant pairs where HIV transmission occurred was performed to identify missed opportunities for prevention of mother-to-child HIV transmission. An intervention program was developed and implemented using the six-step process. Data were collected prospectively over a 6 years period (2007–2012) and were evaluated for six core outcomes measures: (1) adequacy of prenatal care (2) HIV diagnoses of expectant mothers prior to delivery (3) appropriate use of antiretroviral (ARV) therapy before delivery (4) appropriate use of cesarean section for delivery (5) adequacy of zidovudine prophylaxis to newborn (6) HIV transmission rate.

Results:

Twenty-six infants were born to HIV-infected mothers from July 2005 to June 2006 with 6 documented infections. One hundred and five infants were born to HIV infected mothers from January 2007 to December 2012. Postimplementation, adequacy of prenatal care increased from 58% (15/26) to 85% (89/105); appropriate use of ARV therapy before delivery increased from 73% (19/26) to 86% (90/105); cesarean section as the method for delivery increased from 62% (16/26) to 74% (78/105); adequacy of zidovudine prophylaxis to newborn increased from 54% (14/26) to 87% (91/105). HIV transmission rate dropped from 23% (6/26) to 0%.

Conclusion:

Integrating public health and clinical services in the care of HIV-infected pregnant women and exposed infants leads to better coordination of care and improved quality of care.

Systematic Review and Meta-analysis of Long-term survival After Elective Infrarenal Abdominal Aortic Aneurysm Repair 1969-2011: 5 Year Survival Remains Poor Despite Advances in Medical Care and Treatment Strategies

Background

Improved critical care, pre-operative optimization, and the advent of endovascular surgery (EVAR) have improved 30 day mortality for elective abdominal aortic aneurysm (AAA) repair. It remains unknown whether this has translated into improvements in long-term survival, particularly because these factors have also encouraged the treatment of older patients with greater comorbidity. The aim of this study was to quantify how 5 year survival after elective AAA repair has changed over time.

Methods

A systematic review was performed identifying studies reporting 5 year survival after elective infrarenal AAA repair. An electronic search of the Embase and Medline databases was conducted to January 2014. Thirty-six studies, 60 study arms, and 107,814 patients were identified. Meta-analyses were conducted to determine 5 year survival and to report whether 5 year survival changed over time.

Results

Five-year survival was 69% (95% CI 67 to 71%, I² = 87%). Meta-regression on study midpoint showed no improvement in 5 year survival over the period 1969–2011 (log OR −0.001, 95% CI −0.014–0.012). Larger average aneurysm diameter was associated with poorer 5 year survival (adjusted log OR −0.058, 95% CI −0.095 to −0.021, I² = 85%). Older average patient age at surgery was associated with poorer 5 year survival (adjusted log OR −0.118, 95% CI −0.142 to −0.094, I² = 70%). After adjusting for average patient age, an improvement in 5 year survival over the period that these data spanned was obtained (adjusted log OR 0.027, 95% CI 0.012 to 0.042).

Conclusion

Five-year survival remains poor after elective AAA repair despite advances in short-term outcomes and is associated with AAA diameter and patient age at the time of surgery. Age-adjusted survival appears to have improved; however, this cohort as a whole continues to have poor long-term survival. Research in this field should attempt to improve the life expectancy of patients with repaired AAA and to optimise patient selection.

Antithrombotic treatment for acute extracranial carotid artery dissections: a meta-analysis

INTRODUCTION

Carotid artery dissection is a leading cause of stroke in younger patients, with an associated prevalence of 2.6-3.0 per 100,000 population. This meta-analysis aims to determine whether in patients managed medically, treatment with anticoagulants or antiplatelet agents was associated with a better outcome with respect to mortality, ischaemic stroke, and major bleeding episodes.

PATIENTS AND METHODS

A comprehensive search strategy was employed of the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE (January 1966 to March 2015), and EMBASE (January 1980 to March 2015) databases. Primary outcomes were death (all causes) or disability. Secondary outcomes were ischaemic stroke, symptomatic intracranial haemorrhage, and major extracranial haemorrhage during the reported follow-up period.

RESULTS

No completed randomized trials were found. Comparing antiplatelets with anticoagulants across 38 studies (1,398 patients), there were no significant differences in the odds of death (effects size, ES, -0.007, p = .871), nor in the death and disability comparison or across any secondary outcomes.

CONCLUSION

There were no randomised trials comparing either anticoagulants or antiplatelets with control, thus there is no level 1 evidence to support their routine use for the treatment of carotid artery dissection. Also, there were no randomised trials that directly compared anticoagulants with antiplatelet drugs, and the reported non-randomised studies did not show any evidence of a significant difference between the two.

Consistency and inconsistency in network meta-analysis: model estimation using multivariate meta-regression

Network meta-analysis (multiple treatments meta-analysis, mixed treatment comparisons) attempts to make the best use of a set of studies comparing more than two treatments. However, it is important to assess whether a body of evidence is consistent or inconsistent. Previous work on models for network meta-analysis that allow for heterogeneity between studies has either been restricted to two-arm trials or followed a Bayesian framework. We propose two new frequentist ways to estimate consistency and inconsistency models by expressing them as multivariate random-effects meta-regressions, which can be implemented in some standard software packages. We illustrate the approach using the mvmeta package in Stata. Copyright © 2012 John Wiley & Sons, Ltd.

Consistency and inconsistency in network meta-analysis: concepts and models for multi-arm studies

Meta-analyses that simultaneously compare multiple treatments (usually referred to as network meta-analyses or mixed treatment comparisons) are becoming increasingly common. An important component of a network meta-analysis is an assessment of the extent to which different sources of evidence are compatible, both substantively and statistically. A simple indirect comparison may be confounded if the studies involving one of the treatments of interest are fundamentally different from the studies involving the other treatment of interest. Here, we discuss methods for addressing inconsistency of evidence from comparative studies of different treatments. We define and review basic concepts of heterogeneity and inconsistency, and attempt to introduce a distinction between ‘loop inconsistency’ and ‘design inconsistency’. We then propose that the notion of design-by-treatment interaction provides a useful general framework for investigating inconsistency. In particular, using design-by-treatment interactions successfully addresses complications that arise from the presence of multi-arm trials in an evidence network. We show how the inconsistency model proposed by Lu and Ades is a restricted version of our full design-by-treatment interaction model and that there may be several distinct Lu–Ades models for any particular data set. We introduce novel graphical methods for depicting networks of evidence, clearly depicting multi-arm trials and illustrating where there is potential for inconsistency to arise. We apply various inconsistency models to data from trials of different comparisons among four smoking cessation interventions and show that models seeking to address loop inconsistency alone can run into problems. Copyright © 2012 John Wiley & Sons, Ltd.

Paediatric CT dose: a multicentre audit of subspecialty practice in Australia and New Zealand

OBJECTIVES

To evaluate paediatric CT dosimetry in Australia and New Zealand and calculate size-specific dose estimates (SSDEs) for chest and abdominal examinations.

METHODS

Eight hospitals provided data from 12 CT systems for 1462 CTs in children aged 0-15. Imaging data were recorded for eight examinations: head (trauma, shunt), temporal bone, paranasal sinuses, chest (mass) and chest HRCT (high-resolution CT), and abdomen/pelvis (mass/inflammation). Dose data for cranial examinations were categorised by age and SSDEs by lateral dimension. Diagnostic reference ranges (DRRs) were defined by the 25th and 75th percentiles. Centralised image quality assessment was not undertaken.

RESULTS

DRRs for 201 abdominopelvic SSDEs were: 2.8-4.7, 3.6-11.5, 8.5-15.0, 7.6-15, and 10.6-16.2 for the <15 cm, 15-19 cm, 20-24 cm, 25-29 cm and >30 cm groups, respectively. For 147 chest examinations using these body width categories, SSDE DRRs were 2.0-4.4, 3.3-7.9, 4.0-9.4, 4.5-12, and 6.5-12. Kilovoltage peak (kVp), but not AEC or IR, was associated with SSDE (parameter estimate [standard error]: 0.12 (0.03); p < 0.0001).

CONCLUSIONS

Australian and New Zealand paediatric CT DRRs and abdominal SSDEs are comparable to international data. SSDEs for chest examinations are proposed. Dose variations could be reduced by adjusting kVp.

KEY POINTS

• SSDEs can be calculated for all patients, CT systems, and practices • Kilovoltage peak (kVp) has the greatest association with dose in similar-sized patients • Paediatric DRRs for CT are now available for use internationally.

Ebstein's anomaly in those surviving to adult life - a single centre experience

BACKGROUND

Ebstein's anomaly (EA) occurs in about one to five per 200 000 live births. Long-term follow-up data of adults with EA is scarce due to the relatively low frequency of the disease and the variation of its anatomic and haemodynamic severity.

METHODS

Since 1995, in our adult congenital heart disease (ACHD) centre, we have practised a uniform approach to management of adults with EA, with surgery reserved for those with refractory arrhythmia (failed medical and/or catheter-based treatment) or worsening symptoms of breathlessness. A retrospective review of medical records of all such patients with EA and normal cardiac connections was performed.

RESULTS

Fifty-one EA patients (17 males) were identified. Mean age at diagnosis was 21+/-21 years and mean follow-up time at our centre was 21±14 years. During this time, 18 patients (35%) had documented supraventricular arrhythmia. Sixteen patients (30%) underwent ablation therapy with long-term relief from arrhythmia in nine (56%). Nine patients (18%) underwent tricuspid valve (TV) surgery (four repair and five replacement), with seven patients having undergone a tricuspid valve surgery prior to referral to our unit. Three patients died, one of cardiogenic shock after redo surgery (58 years), one of progressive heart failure (45 years) and one with malignancy. Overall survival was 100% to age 40 years, 95% to age 50 years and 81% to age 60 years.

CONCLUSIONS

Ebstein's Anomaly in adulthood often has severe morphological abnormalities but is compatible with good medium-term survival, with a generally symptom driven approach to the indications for interventions.

A systematic review and meta-analysis indicates underreporting of renal dysfunction following endovascular aneurysm repair

Deterioration in renal function has been described after endovascular repair of abdominal aortic aneurysms (EVRs). The etiology is multifactorial and represents an important therapeutic target. A need exists to quantitatively summarize incidence and severity of renal dysfunction after EVR to allow better-informed attempts to preserve renal function and improve life expectancy. Here a systematic search was performed using Medline and Embase for renal function after EVR applying PRISMA statements. Univariate and multivariate random-effects meta-analyses were performed to estimate pooled postoperative changes in serum creatinine and creatinine clearance at four time points after EVR. Clinically relevant deterioration in renal function was also estimated at 1 year or more after EVR. Pooled probability of clinically relevant deterioration in renal function at 1 year or more was 18% (95% confidence interval of 14-23%, I2 of 82.5%). Serum creatinine increased after EVR by 0.05 mg/dl at 30 days/1 month, 0.09 mg/dl at 1 month to 1 year, and 0.11 mg/dl at 1 year or more (all significant). Creatinine clearance decreased after EVR by 5.65 ml/min at 1 month-1 year and by 6.58 ml/min at 1 year or more (both significant). Thus, renal dysfunction after EVR is common and merits attention.