Hakan Demirtas

Efficacy of folic acid supplementation in stroke prevention: a meta-analysis

BACKGROUND The efficacy of treatments that lower homocysteine concentrations in reducing the risk of cardiovascular disease remains controversial. Our aim was to do a meta-analysis of relevant randomised trials to assess the efficacy of folic acid supplementation in the prevention of stroke. METHODS We collected data from eight randomised trials of folic acid that had stroke reported as one of the endpoints. Relative risk (RR) was used as a measure of the effect of folic acid supplementation on the risk of stroke with a random effect model. The analysis was further stratified by factors that could affect the treatment effects. FINDINGS Folic acid supplementation significantly reduced the risk of stroke by 18% (RR 0.82, 95% CI 0.68-1.00; p=0.045). In the stratified analyses, a greater beneficial effect was seen in those trials with a treatment duration of more than 36 months (0.71, 0.57-0.87; p=0.001), a decrease in the concentration of homocysteine of more than 20% (0.77, 0.63-0.94; p=0.012), no fortification or partly fortified grain (0.75, 0.62-0.91; p=0.003), and no history of stroke (0.75, 0.62-0.90; p=0.002). In the corresponding comparison groups, the estimated RRs were attenuated and insignificant. INTERPRETATION Our findings indicate that folic acid supplementation can effectively reduce the risk of stroke in primary prevention.

Plausibility of multivariate normality assumption when multiply imputing non-Gaussian continuous outcomes: a simulation assessment

Multiple imputation under the assumption of multivariate normality has emerged as a frequently used model-based approach in dealing with incomplete continuous data in recent years. Despite its simplicity and popularity, however, its plausibility has not been thoroughly evaluated via simulation. In this work, the performance of multiple imputation under a multivariate Gaussian model with unstructured covariances was examined on a broad range of simulated incomplete data sets that exhibit varying distributional characteristics such as skewness and multimodality that are not accommodated by a Gaussian model. Behavior of efficiency and accuracy measures was explored to determine the extent to which the procedure works properly. The conclusion drawn is that although the real data rarely conform with multivariate normality, imputation under the assumption of normality is a fairly reasonable tool, even when the assumption of normality is clearly violated; the fraction of missing information is high, especially when the sample size is relatively large. Although we discourage its uncritical, automatic and, possibly, inappropriate use, we report that its performance is better than we expected, leading us to believe that it is probably an underrated approach.

Familial aggregation of food allergy and sensitization to food allergens: a family‐based study

Background The increasing prevalence of food allergy (FA) is a growing clinical and public health problem. The contribution of genetic factors to FA remains largely unknown.

Modeling between-subject and within-subject variances in ecological momentary assessment data using mixed-effects location scale models

Ecological momentary assessment and/or experience sampling methods are increasingly used in health studies to study subjective experiences within changing environmental contexts. In these studies, up to 30 or 40 observations are often obtained for each subject. Because there are so many measurements per subject, one can characterize a subjects mean and variance and can specify models for both. In this article, we focus on an adolescent smoking study using ecological momentary assessment where interest is on characterizing changes in mood variation. We describe how covariates can influence the mood variances and also extend the statistical model by adding a subject-level random effect to the within-subject variance specification. This permits subjects to have influence on the mean, or location, and variability, or (square of the) scale, of their mood responses. These mixed-effects location scale models have useful applications in many research areas where interest centers on the joint modeling of the mean and variance structure.

A method for multivariate ordinal data generation given marginal distributions and correlations

A method is described for simulating multivariate ordinal variates with specified marginal distributions and correlation structure. The method relies on simulating correlated binary variates as an intermediate step. After collapsing the ordinal levels to the binary ones, it is straightforward to obtain binary means. Corresponding binary correlations are computed via simulation in a way to ensure that re-conversion to the ordinal scale delivers the original distributional properties. Employing binary data generation is shown to be an effective method for simulating ordinal variates for a broad range of given marginals and pairwise associations.

A Practical Way for Computing Approximate Lower and Upper Correlation Bounds

Correlations among variables are typically not free to vary between −1 and 1, with bounds determined by the marginal distributions. Computing upper and lower limits of correlations given the marginal characteristics often raises theoretical and computational challenges. We propose a simple sorting technique that is predicated upon a little-known consequence of a well-established fact from statistical distribution theory to obtain approximate correlation bounds. This approach works regardless of the data type or distribution. We believe that it has practical value in appropriately specifying the correlation structure in simulation studies.

Simultaneous Generation of Binary and Normal Data with Specified Marginal and Association Structures

Situations in which multiple outcomes and predictors of different distributional types are collected are becoming increasingly common in biopharmaceutical practice, and joint modeling of mixed types has been gaining popularity in recent years. Evaluation of various statistical techniques that have been developed for mixed data in simulated environments necessarily requires joint generation of multiple variables. This article is concerned with building a unified framework for simulating multiple binary and normal variables simultaneously given marginal characteristics and association structure via combining well-established results from the random number generation literature. We illustrate the proposed approach in two simulation settings where we use artificial data as well as real depression score data from psychiatric research, demonstrating a very close resemblance between the specified and empirically computed statistical quantities of interest through descriptive and model-based tools.

Multiple Imputation Under Power Polynomials

Although the normality assumption has been regarded as a mathematical convenience for inferential purposes due to its nice distributional properties, there has been a growing interest regarding generalized classes of distributions that span a much broader spectrum in terms of symmetry and peakedness behavior. In this respect, Fleishmans power polynomial method seems to have been gaining popularity in statistical theory and practice because of its flexibility and ease of execution. In this article, we conduct multiple imputation for univariate continuous data under Fleishman polynomials to explore the extent to which this procedure works properly. We also make comparisons with normal imputation models via widely accepted accuracy and precision measures using simulated data that exhibit different distributional features as characterized by competing specifications of the third and fourth moments. Finally, we discuss generalizations to the multivariate case. Multiple imputation under power polynomials that cover most of the feasible area in the skewness-elongation plane appears to have substantial potential of capturing real missing-data trends.

Demolition of High-Rise Public Housing Increases Particulate Matter Air Pollution in Communities of High-Risk Asthmatics

Abstract Public housing developments across the United States are being demolished, potentially increasing local concentrations of particulate matter (PM) in communities with high burdens of severe asthma. Little is known about the impact of demolition on local air quality. At three public housing developments in Chicago, IL, PM with an aerodynamic diameter <10 μm (PM10) and <2.5 μm were measured before and during high-rise demolition. Additionally, size-selective sampling and real-time monitoring were concurrently performed upwind and downwind of one demolition site. The concentration of particulates attributable to demolition was estimated after accounting for background urban air pollution. Particle microscopy was performed on a small number of samples. Substantial increases of PM10 occurred during demolition, with the magnitude of that increase varying based on sampler distance, wind direction, and averaging time. During structural demolition, local concentrations of PM10 42 m downwind of a demolition site increased 4- to 9-fold above upwind concentrations (6-hr averaging time). After adjusting for background PM10, the presence of dusty conditions was associated with a 74% increase in PM10 100 m downwind of demolition sites (24-hr averaging times). During structural demolition, short-term peaks in real-time PM10 (30-sec averaging time) occasionally exceeded 500 μg/m3. The median particle size downwind of a demolition site (17.3 μm) was significantly larger than background (3 μm). Specific activities are associated with real-time particulate measures. Microscopy did not identify asbestos or high concentrations of mold spores. In conclusion, individuals living near sites of public housing demolition are at risk for exposure to high particulate concentrations. This increase is characterized by relatively large particles and high short-term peaks in PM concentration.

Association of serum prolactin and oxytocin with milk production in mothers of preterm and term infants.

The present study was designed to compare milk production and hormone responses (prolactin [PRL], oxytocin [OT]) and to determine associations of hormone levels with milk production in mothers of preterm (PT) and term (TM) infants during the first 6 weeks postpartum. Mothers of PT infants (n = 95) were all pump dependent; mothers of TM infants (n = 98) were all feeding their infant at breast. Mothers of nonnursing PT infants produced less milk over time compared to mothers of TM infants. A higher proportion of PT mothers had lower basal PRL levels compared with TM mothers. PRL and frequency of breast stimulation combined positively influenced milk production in PT mothers. OT levels were higher in PT versus TM mothers, but OT was not related to milk production. Further study is warranted regarding interventions to enhance milk production, particularly in pump-dependent mothers of PT infants.