Search

Individual participant data meta-analysis is a commonly used alternative to the traditional aggregate data meta-analysis. It is popular because it avoids relying on published results and enables direct adjustment for relevant covariates. However, a practical challenge is that the studies being combined often vary in terms of the potential confounders that were measured. Furthermore, it will inevitably be the case that some individuals have missing values for some of those covariates. In this paper, we demonstrate how these challenges can be resolved using a propensity score approach, combined with multiple imputation, as a strategy to adjust for covariates in the context of individual participant data meta-analysis. To illustrate, we analyze data from the Bill and Melinda Gates Foundation-funded Healthy Birth, Growth, and Development Knowledge Integration project to investigate the relationship between physical growth rate in the first year of life and cognition measured later during childhood. We found that the overall effect of average growth velocity on cognitive outcome is slightly, but significantly, positive with an estimated effect size of 0.36 (95% CI 0.18, 0.55).

High levels of prenatal alcohol exposure (PAE) result in significant cognitive deficits in children, but the exact nature of the dose-response relationship is less well understood. To investigate this relationship, data were assembled from six longitudinal birth cohort studies examining the effects of PAE on cognitive outcomes from early school age through adolescence. Structural equation models (SEMs) are a natural approach to consider, because of the way they conceptualise multiple observed outcomes as relating to an underlying latent variable of interest, which can then be modelled as a function of exposure and other predictors of interest. However, conventional SEMs could not be fitted in this context because slightly different outcome measures were used in the six studies. In this paper we propose a multi-group Bayesian SEM that maps the unobserved cognition variable to a broad range of observed outcomes. The relation between these variables and PAE is then examined while controlling for potential confounders via propensity score adjustment. By examining different possible dose-response functions, the proposed framework is used to investigate whether there is a threshold PAE level that results in minimal cognitive deficit.

High levels of prenatal alcohol exposure (PAE) result in significant cognitive deficits in children, but the exact nature of the dose-response relationship is less well understood. To investigate this relationship, data were assembled from six longitudinal birth cohort studies examining the effects of PAE on cognitive outcomes from early school age through adolescence. Structural equation models (SEMs) are a natural approach to consider, because of the way they conceptualise multiple observed outcomes as relating to an underlying latent variable of interest, which can then be modelled as a function of exposure and other predictors of interest. However, conventional SEMs could not be fitted in this context because slightly different outcome measures were used in the six studies. In this paper we propose a multi-group Bayesian SEM that maps the unobserved cognition variable to a broad range of observed outcomes. The relation between these variables and PAE is then examined while controlling for potential confounders via propensity score adjustment. By examining different possible dose-response functions, the proposed framework is used to investigate whether there is a threshold PAE level that results in minimal cognitive deficit.

In psychiatric and social epidemiology studies, it is common to measure multiple different outcomes using a comprehensive battery of tests thought to be related to an underlying construct of interest. In the research that motivates our work, researchers wanted to assess the impact of in utero alcohol exposure on child cognition and neuropsychological development, which are evaluated using a range of different psychometric tests. Statistical analysis of the resulting multiple outcomes data can be challenging, because the outcomes measured on the same individual are not independent. Moreover, it is unclear, a priori, which outcomes are impacted by the exposure under study. While researchers will typically have some hypotheses about which outcomes are important, a framework is needed to help identify outcomes that are sensitive to the exposure and to quantify the associated treatment or exposure effects of interest. We propose such a framework using a modification of stochastic search variable selection, a popular Bayesian variable selection model and use it to quantify an overall effect of the exposure on the affected outcomes. The performance of the method is investigated empirically and an illustration is given through application using data from our motivating study.

Evidence from animal models and epidemiological studies has linked prenatal alcohol exposure (PAE) to a broad range of long-term cognitive and behavioural deficits. However, there is a paucity of evidence regarding the nature and levels of PAE associated with increased risk of clinically significant cognitive deficits. To derive robust and efficient estimates of the effects of PAE on cognitive function, we have developed a hierarchical meta-analysis approach to synthesize information regarding the effects of PAE on cognition, integrating data on multiple outcomes from six U.S. longitudinal cohort studies. A key assumption of standard methods of meta-analysis, effect sizes are independent, is violated when multiple intercorrelated outcomes are synthesized across studies. Our approach involves estimating the dose–response coefficients for each outcome and then pooling these correlated dose–response coefficients to obtain an estimated “global” effect of exposure on cognition. In the first stage, we use individual participant data to derive estimates of the effects of PAE by fitting regression models that adjust for potential confounding variables using propensity scores. The correlation matrix characterizing the dependence between the outcome-specific dose–response coefficients estimated within each cohort is then run, while accommodating incomplete information on some outcome. We also compare inferences based on the proposed approach to inferences based on a full multivariate analysis.

BACKGROUND: Cognitive and behavioral sequelae of prenatal alcohol exposure (PAE) continue to be prevalent in the United States and worldwide. Because these sequelae are also common in other neurodevelopmental disorders, researchers have attempted to identify a distinct neurobehavioral profile to facilitate the differential diagnosis of fetal alcohol spectrum disorders (FASD). We used an innovative, individual participant meta-analytic technique to combine data from six large U.S. longitudinal cohorts to provide a more comprehensive and reliable characterization of the neurobehavioral deficits seen in FASD than can be obtained from smaller samples. METHODS: Meta-analyses were performed on data from 2236 participants to examine effects of PAE (measured as oz absolute alcohol/day (AA/day)) on IQ, four domains of cognition function (learning and memory, executive function, reading achievement, and math achievement), sustained attention, and behavior problems, after adjusting for potential confounders using propensity scores. RESULTS: The effect sizes for IQ and the four domains of cognitive function were strikingly similar to one another and did not differ at school age, adolescence, or young adulthood. Effect sizes were smaller in the more middle-class Seattle cohort and larger in the three cohorts that obtained more detailed and comprehensive assessments of AA/day. PAE effect sizes were somewhat weaker for parent- and teacher-reported behavior problems and not significant for sustained attention. In a meta-analysis of five aspects of executive function, the strongest effect was on set-shifting. CONCLUSIONS: The similarity in the effect sizes for the four domains of cognitive function suggests that PAE affects an underlying component or components of cognition involving learning and memory and executive function that are reflected in IQ and academic achievement scores. The weaker effects in the more middle-class cohort may reflect a more cognitively stimulating environ

This paper discusses several modern approaches to regression analysis involving time series data where some of the predictor variables are also indexed by time. We discuss classical statistical approaches as well as methods that have been proposed recently in the machine learning literature. The approaches are compared and contrasted, and it will be seen that there are advantages and disadvantages to most currently available approaches. There is ample room for methodological developments in this area. The work is motivated by an application involving the prediction of water levels as a function of rainfall and other climate variables in an aquifer in eastern Australia.

Introduction

Methods

Results

Conclusion

© 2019 Biometrika Trust. This article concerns a class of generalized linear mixed models for two-level grouped data, where the random effects are uniquely indexed by groups and are independent. We derive necessary and sufficient conditions for the marginal likelihood to be expressed in explicit form. These models are unified under the conjugate generalized linear mixed models framework, where conjugate refers to the fact that the marginal likelihood can be expressed in closed form, rather than implying inference via the Bayesian paradigm. The proposed framework allows simultaneous conjugacy for Gaussian, Poisson and gamma responses, and thus can accommodate both unit-and group-level covariates. Only group-level covariates can be incorporated for the binomial distribution. In a simulation of Poisson data, our framework outperformed its competitors in terms of computational time, and was competitive in terms of robustness against misspecification of the random effects distributions.

Copyright © 2017 John Wiley & Sons, Ltd. In many settings, an analysis goal is the identification of a factor, or set of factors associated with an event or outcome. Often, these associations are then used for inference and prediction. Unfortunately, in the big data era, the model building and exploration phases of analysis can be time-consuming, especially if constrained by computing power (ie, a typical corporate workstation). To speed up this model development, we propose a novel subsampling scheme to enable rapid model exploration of clustered binary data using flexible yet complex model set-ups (GLMMs with additive smoothing splines). By reframing the binary response prospective cohort study into a case-control–type design, and using our knowledge of sampling fractions, we show one can approximate the model estimates as would be calculated from a full cohort analysis. This idea is extended to derive cluster-specific sampling fractions and thereby incorporate cluster variation into an analysis. Importantly, we demonstrate that previously computationally prohibitive analyses can be conducted in a timely manner on a typical workstation. The approach is applied to analysing risk factors associated with adverse reactions relating to blood donation.

© 2017 by the authors; licensee MDPI, Basel, Switzerland. The field of spatio-temporal modelling has witnessed a recent surge as a result of developments in computational power and increased data collection. These developments allow analysts to model the evolution of health outcomes in both space and time simultaneously. This paper models the trends in ischaemic heart disease (IHD) in New South Wales, Australia over an eight-year period between 2006 and 2013. A number of spatio-temporal models are considered, and we propose a novel method for determining the goodness-of-fit for these models by outlining a spatio-temporal extension of the Moran’s I statistic. We identify an overall decrease in the rates of IHD, but note that the extent of this health improvement varies across the state. In particular, we identified a number of remote areas in the north and west of the state where the risk stayed constant or even increased slightly.

© 2017 American Statistical Association. The big data era demands new statistical analysis paradigms, since traditional methods often break down when datasets are too large to fit on a single desktop computer. Divide and Recombine (D&R) is becoming a popular approach for big data analysis, where results are combined over subanalyses performed in separate data subsets. In this article, we consider situations where unit record data cannot be made available by data custodians due to privacy concerns, and explore the concept of statistical sufficiency and summary statistics for model fitting. The resulting approach represents a type of D&R strategy, which we refer to as summary statistics D&R; as opposed to the standard approach, which we refer to as horizontal D&R. We demonstrate the concept via an extended Gamma–Poisson model, where summary statistics are extracted from different databases and incorporated directly into the fitting algorithm without having to combine unit record data. By exploiting the natural hierarchy of data, our approach has major benefits in terms of privacy protection. Incorporating the proposed modelling framework into data extraction tools such as TableBuilder by the Australian Bureau of Statistics allows for potential analysis at a finer geographical level, which we illustrate with a multilevel analysis of the Australian unemployment data. Supplementary materials for this article are available online.

Spatial data have become increasingly common in epidemiology and public health research thanks to advances in GIS (Geographic Information Systems) technology. In health research, for example, it is common for epidemiologists to incorporate geographically indexed data into their studies. In practice, however, the spatially defined covariates are often measured with error. Naive estimators of regression coefficients are attenuated if measurement error is ignored. Moreover, the classical measurement error theory is inapplicable in the context of spatial modeling because of the presence of spatial correlation among the observations. We propose a semiparametric regression approach to obtain bias-corrected estimates of regression parameters and derive their large sample properties. We evaluate the performance of the proposed method through simulation studies and illustrate using data on Ischemic Heart Disease (IHD). Both simulation and practical application demonstrate that the proposed method can be effective in practice.

© 2016 The Royal Statistical Society Making effective public policy decisions is challenging at the best of times, but especially in the context of environmental regulation, which typically requires managing opposing interests and strong opinions from industry and private citizens. In this case study, Louise Ryan, Matt Wand and Alan Malecki show how statistical analysis can help resolve conflict and inform effective decision-making under uncertainty.

Objective: To describe the use of surveillance and forecasting models to predict and track epidemics (and, potentially, pandemics) of influenza. Methods: We collected 5 years of historical data (2005-2009) on emergency department presentations and hospital admissions for influenza-like illnesses (International Classification of Diseases [ICD-10-AM] coding) from the Emergency Department Information System (EDIS) database of 27 Queensland public hospitals. The historical data were used to generate prediction and surveillance models, which were assessed across the 2009 southern hemisphere influenza season (June-September) for their potential usefulness in informing response policy. Three models are described: (i) surveillance monitoring of influenza presentations using adaptive cumulative sum (CUSUM) plan analysis to signal unusual activity; (ii) generating forecasts of expected numbers of presentations for influenza, based on historical data; and (iii) using Google search data as outbreak notification among a population. Results: All hospitals, apart from one, had more than the expected number of presentations for influenza starting in late 2008 and continuing into 2009. (i) The CUSUM plan signalled an unusual outbreak in December 2008, which continued in early 2009 before the winter influenza season commenced. (ii) Predictions based on historical data alone underestimated the actual influenza presentations, with 2009 differing significantly from previous years, but represent a baseline for normal ED influenza presentations. (iii) The correlation coefficients between internet search data for Queensland and statewide ED influenza presentations indicated an increase in correlation since 2006 when weekly influenza search data became available. Conclusion: This analysis highlights the value of health departments performing surveillance monitoring to forewarn of disease outbreaks. The best system among the three assessed was a combination of routine forecasting methods cou

OBJECTIVE: The objective of this study was to estimate the association between caffeine consumption and in vitro fertilization (IVF) outcomes. METHODS: A total of 2474 couples were prospectively enrolled prior to undergoing their first cycle of IVF, contributing a total of 4716 IVF cycles. Discrete survival analysis adjusting for observed confounders was applied to quantify the relation between caffeine consumption and livebirth. Secondary outcomes of interest were oocyte retrieval, peak estradiol level, implantation rate, and fertilization rate. RESULTS: Overall, caffeine consumption by women was not significantly associated with livebirth (ptrend=0.74). Compared with women who do not drink caffeine, the likelihood of livebirth was not significantly different for women who drank low (>0-800 mg/week; odds ratio [OR]=1.00, 95% confidence interval [CI])=0.83-1.21), moderate (>800-1400 mg/week; OR=0.89, 95% CI=0.71-1.12), or high levels of caffeine (>1400 mg/week; OR=1.07, 95% CI=0.85-1.34). Greater caffeine intake by women was associated with a significantly lower peak estradiol level (ptrend=0.03), but was not associated with the number of oocytes retrieved (ptrend=0.75), fertilization rate (ptrend=0.10), or implantation rate (ptrend=0.23). There was no significant association between caffeine intake by men and livebirth (ptrend=0.27), fertilization (ptrend=0.72), or implantation (ptrend=0.24). The individual effects of consumption of coffee, tea, or soda by women or men were not related to livebirth. CONCLUSION: Caffeine consumption by women or men was not associated with IVF outcomes.

The number of in vitro fertilization (IVF) cycles in the United States increased from fewer than 46,000 in 1995 to more than 120,000 in 2005. IVF and other assisted reproductive technology (ART) data are routinely collected and used to identify outcome predictors. However, researchers do not always make full use of the data due to their complexity. Design approaches have included restriction to first-cycle attempts only, which reduces power and identifies effects only of those factors associated with initial success. Many statistical techniques have been used or proposed for analysis of IVF data, ranging from simple t tests to sophisticated models designed specifically for IVF. We applied several of these methods to data from a prospective cohort of 2687 couples undergoing ART from 1994 through 2003. Results across methods are compared and the appropriateness of the various methods is discussed with the intent to illustrate methodologic validity. We observed a remarkable similarity of coefficient estimates across models. However, each method for dealing with multiple cycle data relies on assumptions that may or may not be expected to hold in a given IVF study. The robustness and reported magnitude of effect for individual predictors of IVF success may be inflated or attenuated due to violation of statistical assumptions, and should always be critically interpreted. Given that risk factors associated with IVF success may also advance our understanding of the physiologic processes underlying conception, implantation, and gestation, the application of valid methods to these complex data is critical. © 2011 by Lippincott Williams & Wilkins.

The paper describes the use of a longitudinal tobit model to characterize cognitive decline over a 13-year period in a cohort of 2087 elderly Australians. Use of a tobit formulation allows for the so-called 'ceiling effect' wherein many subjects achieve perfect test scores. A Bayesian hierarchical joint model is presented that allows for random subject-specific intercepts and slopes, as well as for informative dropout. Results suggest several potential areas of intervention. For example, there is a clear dose-response effect of exercise whereby increasing levels of exercise are associated with higher cognitive scores. © 2010 Royal Statistical Society.

In a study of 133 volunteer subjects, demographic, physiologic and pharmacokinetic data through exposure to 1,3-Butadiene (BD) were collected in order to estimate the percentage of BD concentration metabolized at steady state, and to determine whether this percentage varies across gender, racial, and age groups. During the 20 min of continuous exposure to 2 parts per million (ppm) of BD, five measurements of exhaled concentration were made on each subject. In the following 40 min washout period, another five measurements were collected. A Bayesian hierarchical compartmental physiologically-based pharmacokinetic model (PKPB) was used. Using prior information on the model parameters, Markov Chain Monte Carlo (MCMC) simulation was conducted to obtain posterior distributions. The overall estimate of the mean percent of BD metabolized at steady state was 12.7% (95% credible interval: 7.7-17.8%). There was no significant difference in gender with males having a mean of 13. 5%, and females 12.3%. Among the racial groups, Hispanic (13.9%), White (13.0%), Asian (12.1%), and Black (10.9%), the significant difference came from the difference between Black and Hispanic with a 95% credible interval from -5.63 to -0.30%. Those older than 30 years had a mean of 12.2% versus 12.9% for the younger group; although this was not a statistically significant difference. Given a constant inhalation input of 2 ppm, at steady state, the overall mean exhaled concentration was estimated to be 1.75ppm (95% credible interval: 1.64-1.84). An equivalent parameter, first-order metabolic rate constant, was also estimated and found to be consistent with the percent of BD metabolized at steady state across gender, race, and age strata. © 2009 Springer Science+Business Media, LLC.

Background It has been widely accepted that sun exposure is a risk factor of squamous cell carcinoma (SCC) among fair-skinned populations. However, sun exposure and sun reaction have not been explored in Asians and no gender-specific data were available. Method In a case-control study, 176 incident skin cancer cases were recruited from National Cheng-Kung University Medical Center from 1996 to 1999. Controls included 216 age-, gender-, and residency-matched subjects from the southwestern Taiwan. A questionnaire was administered to collect information on life style and other risk factors. Logistic regression analysis was performed to evaluate the association between sun exposure or sun reaction and the risk of SCC by gender. Results Early-age (age 15 to 24) and lifetime sun exposure were significantly associated with increased risk of SCC in a dose-response pattern [odds ratio (OR) = 1.49-3.08, trend p = 0.009 and 0.0007, respectively]. After stratified by gender, the third tertile of early-age sun exposure was significantly associated with the SCC risk among men (OR = 3.08). The second and third tertiles of lifetime sun exposure was significantly associated with SCC risk among women (OR = 3.78 and 4.53, respectively). Skin reaction after 2-h sun exposure during childhood and adolescence was not significantly associated with the risk of SCC. Conclusions Lifetime sun exposure was more related to SCC risk in women, while early-age sun exposure was more relevant to men's SCC risk. This may be attributable to different lifestyle between men and women. © Springer Science+Business Media B.V. 2010.

BACKGROUND: Epidemiology frequently relies on surrogates of long-term exposures, often either individual-level short-term measurements or group-level based on long-term characteristics of subjects and their environment. Whereas individual-level measures are often imprecise due to within-subject variability, group-level measures tend to be inaccurate due to residual between-subject variability within groups. Rather than choose between these error-prone estimates, we borrow strength from each by use of mixed-model prediction and we compare the predictive validity. METHODS: We compared alternative measures of long-term exposure to carbon monoxide (CO) among children in the RESPIRE woodstove randomized control trial during years 2003 and 2004. The main study included 1932 repeated 48-hour-average personal CO measures among 509 children from 0-18 months of age. We used a validation study with additional CO measures among a random subsample of 70 of the children to compare the predictive validity of individual-level estimates (based on observed short-term exposures), group-level estimates (based on stove type and other residential characteristics), and mixed-model predictions that combine these 2 sources of information. RESULTS: The estimated error variance for mixed-model prediction was 63% lower than the individual-level measure based on the exposure data and 58% lower than the corresponding group-level measure. CONCLUSIONS: When both individual- and group-level estimates are available but imperfect, mixed-model prediction may provide substantially better measures of long-term exposure, potentially increasing the sensitivity of epidemiologic studies to underlying causal relations. © 2009 Lippincott Williams & Wilkins, Inc.

BACKGROUND: Dominant hemimelia (Dh) is an autosomal dominant mutation that arose spontaneously in mice. Dh animals are asplenic and they exhibit asymmetric hindlimb defects in association with reduced numbers of lumbar vertebrae. These defects suggest that Dh acts early in embryonic development to affect patterning of the anterior-posterior (A-P) and left-right axes. This study was undertaken to determine whether retinoic acid (RA), which is involved in A-P patterning and coordination of bilaterally synchronized somitogenesis, affects phenotypic expression of the Dh gene. METHODS: Thirty-four pregnant females were given, by oral intubation, a single dose of 50 or 75 mg all-trans RA per kilogram body weight at GD 9, 10, or 11. The pregnant females were then euthanized at GD 18 and fetuses removed by cesarean section. A total of 326 fetuses were identified by phenotype and linked DNA and their skeletons were analyzed. RESULTS: There was a differential effect of RA on the axial skeleton and hindlimb of Dh/+ mice as compared to their wild-type littermates. Dose- and stage-specific effects on sternebrae and vertebrae were observed. CONCLUSIONS: The effects of RA dosing on numbers of sternebrae and vertebrae suggest that Dh embryos have a primary defect in retinoid-mediated A-P patterning. Dosing with RA may produce the observed effects on phenotypic expression of Dh/+ by indirectly or directly modifying an already existing altered Hox expression pattern. As the relationship between axial patterning and the asymmetric limb is unknown, Dh is an important model for studying this relationship. © 2008 Wiley-Liss, Inc.

The analysis of data arising from environmental health studies which collect a large number of measures of exposure can benefit from using latent variable models to summarize exposure information. However, difficulties with estimation of model parameters may arise since existing fitting procedures for linear latent variable models require correctly specified residual variance structures for unbiased estimation of regression parameters quantifying the association between (latent) exposure and health outcomes. We propose an estimating equations approach for latent exposure models with longitudinal health outcomes which is robust to misspecification of the outcome variance. We show that compared to maximum likelihood, the loss of efficiency of the proposed method is relatively small when the model is correctly specified. The proposed equations formalize the ad-hoc regression on factor scores procedure, and generalize regression calibration. We propose two weighting schemes for the equations, and compare their efficiency. We apply this method to a study of the effects of in-utero lead exposure on child development. © Institute of Mathematical Statistics, 2009.

Objectives. We assessed whether maternal depressive symptoms and parenting self-efficacy were associated with child growth delay. Methods. We collected data from a random sample of 595 low-income mothers and their children aged 6 to 24 months in Teresina, Piauí, Brazil, including information on sociodemographic characteristics, mothers' depressive symptoms and parenting self-efficacy, and children's anthropometric characteristics. We used adjusted logistic regression models in our analyses. Results. Depressive symptoms among mothers were associated with 1.8 times higher odds (95% confidence interval [CI]=1.1, 2.9) of short stature among children. Parenting self-efficacy was not associated with short stature, nor did it mediate or modify the relationship between depressive symptoms and short stature. Maternal depressive symptoms and self-efficacy were not related to child underweight. Conclusions. Our results showed that among low-income Brazilian families maternal depressive symptoms, but not self-efficacy, were associated with short stature in children aged 6 to 24 months after adjustment for known predictors of growth.