Fixed Effect Covariates Research Articles

Examining the association between prenatal and perinatal adversity and the psychotic experiences in childhood.

Prenatal and perinatal complications are established risk factors for psychotic disorder, but far less is known about these measures and psychotic experiences (PEs). We investigated the longitudinal effect of prenatal risk factors (maternal behavior, medication complications) and perinatal risk factors (birth weight, medical complications) on frequency of PEs. We also examined the cumulative risk of prenatal/perinatal risk factors, and differences between transient PE, persistent PE, and controls. The Adolescent Brain Cognitive Development study is a large child cohort (age 9-10 at baseline; n = 11 872 with PE data). PEs were measured longitudinally using the Prodromal Questionnaire-Brief, Child version, and included only if reported as distressing. Mixed-effects models were used for analysis, controlling for random effects, and a substantial number of fixed-effects covariates. Urinary tract infection (β = 0.11, 95% confidence interval [CI] 0.03-0.19) and severe anemia (β = 0.18, 95% CI 0.07-0.29) increased frequency of distressing PEs in childhood. Number of prenatal complications increased frequency of PEs (β = 0.03, 95% CI 0.01-0.06) and risk of persistent PEs (odds ratio [OR] = 1.08, 95% CI 1.01-1.15). Maternal smoking was associated with an increased frequency of PEs (β = 0.11, 95% CI 0.04-0.18) and persistent PEs (OR = 1.31, 95% CI 1.04-1.66). Maternal substance use was a risk factor for a 48% increased risk of persistent PEs (OR = 1.48, 95% CI 1.08-2.01). Perinatal complications showed no effect on PEs. This study provides evidence that certain prenatal medical complications (severe nausea, severe anemia), cumulative number of prenatal medical complications, and maternal behaviors (smoking during pregnancy), increased frequency of distressing PEs in childhood. Maternal smoking and substance use, as well as cumulative number of prenatal complications increased risk of persistent PEs.

Identification of Bradycardia Following Remdesivir Administration Through the US Food and Drug Administration American College of Medical Toxicology COVID-19 Toxic Pharmacovigilance Project

The rapid spread and mortality associated with COVID-19 emphasized a need for surveillance system development to identify adverse events (AEs) to emerging therapeutics. Bradycardia is a remdesivir infusion-associated AE listed in the US Food and Drug Administration-approved prescribing information. To evaluate the magnitude and duration of bradycardic events following remdesivir administration. A multicenter cohort study of patients with recorded heart rate less than 60 beats per minute within 24 hours after administration of a remdesivir dose was conducted between November 23, 2020, and October 31, 2021. Participants included patients hospitalized with COVID-19 at 15 medical centers across the US. Patients excluded had AEs unrelated to bradycardia, AEs in addition to bradycardia, or first onset of bradycardia after 5 remdesivir doses. Remdesivir administration. Linear mixed-effect models for the minimum HR before starting remdesivir and within 24 hours of each dose included doses as fixed effects. Baseline covariates were age (≥65 years vs <65 years), sex (male vs female), cardiovascular disease history (yes vs no), and concomitant use of bradycardia-associated medications. The interactions between variables and doses were considered fixed-effects covariates to adjust models. A total of 188 patients were included in the primary analysis and 181 in the secondary analysis. The cohort included 108 men (57.4%); 75 individuals (39.9%) were non-Hispanic White and mean (SD) age was 61.3 (15.4) years. Minimum HR after doses 1 to 5 was lower than before remdesivir. Mean minimum HR was lowest after dose 4, decreasing by -15.2 beats per minute (95% CI, -17.4 to -13.1; P < .001) compared with before remdesivir administration. Mean (SD) minimum HR was 55.6 (10.2) beats per minute across all 5 doses. Of 181 patients included in time-to-event analysis, 91 had their first episode of bradycardia within 23.4 hours (95% CI, 20.1-31.5 hours) and 91 had their lowest HR within 60.7 hours (95% CI, 54.0-68.3 hours). Median time to first bradycardia after starting remdesivir was shorter for patients aged 65 years or older vs those younger than 65 years (18.7 hours; 95% CI, 16.8-23.7 hours vs 31.5 hours; 95% CI, 22.7-39.3 hours; P = .04). Median time to lowest HR was shorter for men vs women (54.2 hours; 95% CI, 47.3-62.0 hours vs 71.0 hours; 95% CI, 59.5-79.6 hours; P = .02). In this cohort study, bradycardia occurred during remdesivir infusion and persisted. Given the widespread use of remdesivir, practitioners should be aware of this safety signal.

253 Physical activity patterns in adolescents and young adults with intellectual and developmental disabilities

OBJECTIVES/GOALS: Individuals with intellectual and developmental disabilities (IDD) have lower levels of moderate-to-vigorous physical activity (MVPA) and a greater risk for sedentary-related comorbidities compared to their typically developing peers. Understanding activity patterns may provide opportunities for targeted physical activity interventions. METHODS/STUDY POPULATION: Secondary analyses were performed on baseline accelerometer data pooled from 2 clinical trials and a pilot study in adolescents (11-17 years) and young adults (18-21 years) with IDD. MVPA was assessed using accelerometers worn on the non-dominant hip during waking hours over 7 consecutive days. Data were collected at 60 hertz and activity counts were aggregated over 60 second epochs. Wear time was determined with the Choi algorithm and MVPA was classified using the Troiano adult or Freedson age-specific child cut-points. Mixed effects linear regressions were used to determine the effects of day of the week, time of the day, and season on MVPA. Diagnosis, gender, and age were used as fixed effect covariates with random intercepts varying among the participants and days of observation within each participant. RESULTS/ANTICIPATED RESULTS: There were 231 individuals (15.6 ± 2.8 years, 51.5% female) who had IDD (36.8% Autism, 48.1% Down syndrome) with 22,498 minutes of MVPA. Individuals with IDD wore the accelerometers an average of 592 ± 254 min./day and completed 13.5 ± 17.9 min./day of MVPA. Average MVPA was lowest in individuals with Autism (12.6 ± 11.4 min./day) and Down syndrome (13.2 ± 9.3 min./day) when compared to those with other IDDs (16.8 ± 10.8 min./day). Participation in MVPA was similar in males (13.4 ± 10.7 min./day) and females (13.7 ± 9.9 min./day). Mixed effects linear regressions showed that individuals participated in fewer minutes of MVPA on the weekend (β = -0.75, p &lt; 0.001) and from 12-3 pm (reference) when compared to before 12 pm (β = 0.87, p &lt; 0.001) and 3-7 pm (β = 0.66, p=0.007). No significant seasonal effects were found. DISCUSSION/SIGNIFICANCE: Individuals with IDD were significantly less active on the weekend, but they did participate in more minutes of MVPA in the morning and late afternoon/early evening. Physical activity interventions aiming to increase MVPA on the weekend and during the early afternoon may increase the number of weekly minutes of MVPA in individuals with IDD.

Cytokine Hemoadsorption During Cardiac Surgery Versus Standard Surgical Care for Infective Endocarditis (REMOVE): Results From a Multicenter Randomized Controlled Trial.

Cardiac surgery often represents the only treatment option in patients with infective endocarditis (IE). However, IE surgery may lead to a sudden release of inflammatory mediators, which is associated with postoperative organ dysfunction. We investigated the effect of hemoadsorption during IE surgery on postoperative organ dysfunction. This multicenter, randomized, nonblinded, controlled trial assigned patients undergoing cardiac surgery for IE to hemoadsorption (integration of CytoSorb to cardiopulmonary bypass) or control. The primary outcome (change in sequential organ failure assessment score [ΔSOFA]) was defined as the difference between the mean total postoperative SOFA score, calculated maximally to the 9th postoperative day, and the basal SOFA score. The analysis was by modified intention to treat. A predefined intergroup comparison was performed using a linear mixed model for ΔSOFA including surgeon and baseline SOFA score as fixed effect covariates and with the surgical center as random effect. The SOFA score assesses dysfunction in 6 organ systems, each scored from 0 to 4. Higher scores indicate worsening dysfunction. Secondary outcomes were 30-day mortality, duration of mechanical ventilation, and vasopressor and renal replacement therapy. Cytokines were measured in the first 50 patients. Between January 17, 2018, and January 31, 2020, a total of 288 patients were randomly assigned to hemoadsorption (n=142) or control (n=146). Four patients in the hemoadsorption and 2 in the control group were excluded because they did not undergo surgery. The primary outcome, ΔSOFA, did not differ between the hemoadsorption and the control group (1.79±3.75 and 1.93±3.53, respectively; 95% CI, -1.30 to 0.83; P=0.6766). Mortality at 30 days (21% hemoadsorption versus 22% control; P=0.782), duration of mechanical ventilation, and vasopressor and renal replacement therapy did not differ between groups. Levels of interleukin-1β and interleukin-18 at the end of integration of hemoadsorption to cardiopulmonary bypass were significantly lower in the hemoadsorption than in the control group. This randomized trial failed to demonstrate a reduction in postoperative organ dysfunction through intraoperative hemoadsorption in patients undergoing cardiac surgery for IE. Although hemoadsorption reduced plasma cytokines at the end of cardiopulmonary bypass, there was no difference in any of the clinically relevant outcome measures. URL: https://www. gov; Unique identifier: NCT03266302.

Correlation between work productivity loss and EORTC QLQ-C30 and -BR23 domains from the MONALEESA-7 trial of premenopausal women with HR+/HER2- advanced breast cancer.

Background:The phase III MONALEESA-7 trial (NCT02278120) assessed ribociclib + endocrine therapy (ET) versus ET in premenopausal women with HR+/HER2− advanced breast cancer (ABC). The relationship between work productivity loss (WPL) and domains of European Organisation for Research and Treatment of Cancer Quality of Life (EORTC QLQ-C30) and the breast cancer (BC)-specific module (QLQ-BR23) has not been explored in ABC. In this post hoc analysis (data cutoff, November 30, 2018), we assessed the correlation between the WPL component of the Work Productivity and Activity Impairment: General Health (WPAI:GH) questionnaire and EORTC QLQ-C30/BR23 domains.Methods:We analyzed EORTC and WPAI:GH data from 329 patients in both treatment arms of MONALEESA-7 who were employed during the trial. Separate univariable mixed-model repeated measures (MMRM) regression models were fitted for each domain, with WPL as dependent variable and each EORTC domain score as a single fixed-effect covariate. Linear and quadratic relationships were considered based on the Akaike information criterion. Next, two separate multivariable MMRM regression models were fitted with WPL a dependent variable and all QLQ-C30/BR23 domain scores as fixed-effect covariates. The strength of correlation between WPL and EORTC domains was assessed in terms of minimally important differences for the QLQ-C30/BR23 modules.Results:Our univariable analysis showed that greater WPL was statistically significantly associated with lower levels of overall quality of life (QoL) and other functional domains and with higher levels of all symptomatic domains of the QLQ-C30/BR23 modules. Our multivariable analysis determined that this correlation was primarily driven by changes in QoL; physical, role, social, and future perspective domains; and BC-specific symptomatic domains.Conclusion:This analysis determined the QoL domains that correlate with WPL in premenopausal patients with HR+/HER2− ABC. These results may inform prognostic tools to identify and characterize patients with greater risk for WPL and help design interventional strategies to minimize WPL.

Seasonal patterns of malaria, genital infection, nutritional and iron status in non-pregnant and pregnant adolescents in Burkina Faso: a secondary analysis of trial data

BackgroundAdolescents are considered at high risk of developing iron deficiency. Studies in children indicate that the prevalence of iron deficiency increased with malaria transmission, suggesting malaria seasonally may drive iron deficiency. This paper examines monthly seasonal infection patterns of malaria, abnormal vaginal flora, chorioamnionitis, antibiotic and antimalarial prescriptions, in relation to changes in iron biomarkers and nutritional indices in adolescents living in a rural area of Burkina Faso, in order to assess the requirement for seasonal infection control and nutrition interventions.MethodsData collected between April 2011 and January 2014 were available for an observational seasonal analysis, comprising scheduled visits for 1949 non-pregnant adolescents (≤19 years), (315 of whom subsequently became pregnant), enrolled in a randomised trial of periconceptional iron supplementation. Data from trial arms were combined. Body Iron Stores (BIS) were calculated using an internal regression for ferritin to allow for inflammation. At recruitment 11% had low BIS (< 0 mg/kg). Continuous outcomes were fitted to a mixed-effects linear model with month, age and pregnancy status as fixed effect covariates and woman as a random effect. Dichotomous infection outcomes were fitted with analogous logistic regression models.ResultsSeasonal variation in malaria parasitaemia prevalence ranged between 18 and 70% in non-pregnant adolescents (P < 0.001), peaking at 81% in those who became pregnant. Seasonal variation occurred in antibiotic prescription rates (0.7–1.8 prescriptions/100 weekly visits, P < 0.001) and chorioamnionitis prevalence (range 15–68%, P = 0.026). Mucosal vaginal lactoferrin concentration was lower at the end of the wet season (range 2–22 μg/ml, P < 0.016), when chorioamnionitis was least frequent. BIS fluctuated annually by up to 53.2% per year around the mean BIS (5.1 mg/kg2, range 4.1–6.8 mg/kg), with low BIS (< 0 mg/kg) of 8.7% in the dry and 9.8% in the wet seasons (P = 0.36). Median serum transferrin receptor increased during the wet season (P < 0.001). Higher hepcidin concentration in the wet season corresponded with rising malaria prevalence and use of prescriptions, but with no change in BIS. Mean Body Mass Index and Mid-Upper-Arm-Circumference values peaked mid-dry season (both P < 0.001).ConclusionsOur analysis supports preventive treatment of malaria among adolescents 15–19 years to decrease their disease burden, especially asymptomatic malaria. As BIS were adequate in most adolescents despite seasonal malaria, a requirement for programmatic iron supplementation was not substantiated.

Correlation between work productivity loss (WPL) and European Organization for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire (QLQ-C30) domains from the MONALEESA-7 (ML-7) trial of premenopausal women with HR+/HER2- advanced breast cancer (ABC).

1051 Background: The international, randomized, double-blind, placebo-controlled, Phase III ML-7 trial (NCT02278120) assessed ribociclib + endocrine therapy (ET) vs ET alone in premenopausal women with HR+/HER2– ABC. To our knowledge, the relationship between WPL and domains of the EORTC QLQ-C30 and the tumor-specific module for breast cancer (QLQ-BR23) has not been explored in ABC. In this post hoc analysis (data cutoff, November 30, 2018) of all patients (pts) enrolled in ML-7, we assessed the correlation between the WPL component of the Work Productivity and Activity Impairment: General Health (WPAI:GH) questionnaire and domains of the EORTC QLQ-C30/BR23. Methods: We analyzed EORTC and WPAI:GH data from all pts enrolled in ML-7 who were employed at any point during the trial (N = 329 of 672 total pts). Domains of the EORTC QLQ-C30 and QLQ-BR23 that had the greatest correlation (pairwise Pearson correlation) with WPL were prioritized for analysis. Separate univariable mixed-model repeated-measures regression models were fitted for each domain, with WPL as the dependent variable and each EORTC domain as a single fixed-effect covariate. Linear and quadratic relationships were considered. Model selection was based on the Akaike information criterion (AIC). Results: Linear models were favored over quadratic models. WPL was negatively correlated with global health status (GHS) and the physical, role, social, and emotional functioning domains and was positively correlated with the fatigue and pain domains of the QLQ-C30 ( P &lt;.001; Table). The coefficients indicated the estimated mean change in WPL was associated with a 1-unit increase in each QLQ-C30 domain. For example, a 10-point increase in GHS was associated with an estimated mean decrease of 7.8% (95% CI, 7.1%-8.5%) in WPL. Conclusions: Greater WPL was associated with higher levels of fatigue and pain and with lower levels of overall quality of life and physical, role, social, and emotional functioning among pts with HR+/HER2− ABC in ML-7. Further investigation of the correlation with QLQ-BR23 and multivariable analysis could determine which EORTC domains and items independently drive these findings. Clinical trial information: NCT02278120 .[Table: see text]

Genomic prediction of agronomic traits in wheat using different models and cross-validation designs.

Genomic predictions across environments and within populations resulted in moderate to high accuracies but across-population genomic prediction should not be considered in wheat for small population size. Genomic selection (GS) is a marker-based selection suggested to improve the genetic gain of quantitative traits in plant breeding programs. We evaluated the effects of training population (TP)composition, cross-validation design, and genetic relationship between the training and breeding populations on the accuracy of GS in spring wheat (Triticum aestivum L.). Two populations of 231 and 304 spring hexaploid wheat lines that were phenotyped for six agronomic traits and genotyped with the wheat 90K array were used to assess the accuracy of seven GS models (RR-BLUP, G-BLUP, BayesB, BL, RKHS, GS + de novo GWAS, and reaction norm) using different cross-validation designs. BayesB outperformed the other models for within-population genomic predictions in the presence of few quantitative trait loci (QTL) with large effects. However, including fixed-effect marker covariates gave better performance for an across-population prediction when the same QTL underlie traits in both populations. The accuracy of prediction was highly variable based on the cross-validation design, which suggests the importance to use a design that resembles the variation within a breeding program. Moderate to high accuracies were obtained when predictions were made within populations. In contrast, across-population genomic prediction accuracies were very low, suggesting that the evaluated models are not suitable for prediction across independent populations. On the other hand, across-environment prediction and forward prediction designs using the reaction norm model resulted in moderate to high accuracies, suggesting that GS can be applied in wheat to predict the performance of newly developed lines and lines in incomplete field trials.

Associations between gut microbiota and genetic risk for rheumatoid arthritis in the absence of disease: a cross-sectional study

SummaryBackgroundRheumatoid arthritis is a chronic inflammatory autoimmune disease that is associated with reduced life expectancy. The disease is heritable and an extensive repertoire of genetic variants have been identified. The gut microbiota might represent an environmental risk factor for rheumatoid arthritis. We aimed to assess whether known rheumatoid arthritis risk alleles were associated with the gut microbiota in a large population who do not have rheumatoid arthritis.MethodsIn this cross-sectional study done in the UK and Switzerland, we used genotyping and microbiota data from previous studies of the TwinsUK cohort, excluding participants who had ever had a diagnosis of rheumatoid arthritis, as well as their unaffected co-twins. We used blood samples for genotyping and stool samples for the assessment of the gut microbiota. We generated a polygenic risk score (PRS) for rheumatoid arthritis in 1650 TwinsUK participants without the disease, based on 233 GWAS-identified single nucleotide polymorphisms associated with rheumatoid arthritis. We validated the PRS using logistic regression against rheumatoid arthritis diagnosis in 2686 UK Biobank individuals with a confirmed diagnosis of rheumatoid arthritis. Amplicon sequence variants (ASVs) were generated from 16S rRNA gene sequencing of stool samples and assessed for association with the PRS for rheumatoid arthritis. We validated the findings in an independent sample comprised of first-degree relatives of patients with rheumatoid arthritis from the SCREEN-RA cohort. Differential abundance of ASVs present in more than 5% of samples, grouped by ASV taxon annotation, against the rheumatoid arthritis PRS as a continuous variable was assessed using fixed-effects covariates. To account for multiple testing, the false discovery rate calculation was applied to all p values to generate q values, with a significance threshold of 0·05 determined a priori.FindingsWe found that presence of Prevotella spp were positively associated with the rheumatoid arthritis PRS in TwinsUK participants (q<1 × 10−7). This finding was validated in SCREEN-RA participants (n=133) carrying established shared epitope risk alleles (q=0·0011). We also found an association between Prevotella spp and presence of preclinical rheumatoid arthritis phases (q=0·021).InterpretationPrevotella spp in the gut microbiota are associated with the rheumatoid arthritis genotype in the absence of rheumatoid arthritis, including in individuals at high risk of developing rheumatoid arthritis. Our findings suggest that host genotype is associated with microbiota profile before disease onset.FundingVersus Arthritis.

Estimation and Inference in Mixed Effect Regression Models Using Shape Constraints, with Application to Tree Height Estimation

SummaryEstimation of tree height given diameter is an important part of the forest inventory analysis of the US Forest Service. Existing methods use parametric models to estimate the curve. We propose a semiparametric model in which log (height) is a smooth, increasing and concave function of diameter, with a random-plot component and fixed effect covariates. Large sample properties and inference methods that work well in practice are derived. Proposed inference methods use approximate normal distributions for the fixed effects and a likelihood ratio test for the significance of the random effect. A closed form approximate prediction method is provided and overall it outperformed competitors for both a simulation and a real data application. The methods are implemented by the cgamm routine in the R package cgam and can be used for a wide range of mixed model applications.

The impact of disregarding family structure on genome-wide association analysis of complex diseases in cohorts with simple pedigrees.

The generalized linear mixed models (GLMMs) methodology is the standard framework for genome-wide association studies (GWAS) of complex diseases in family-based cohorts. Fitting GLMMs in very large cohorts, however, can be computationally demanding. Also, the modified versions of GLMM using faster algorithms may underperform, for instance when a single nucleotide polymorphism (SNP) is correlated with fixed-effects covariates. We investigated the extent to which disregarding family structure may compromise GWAS in cohorts with simple pedigrees by contrasting logistic regression models (i.e., with no family structure) to three LMMs-based ones. Our analyses showed that the logistic regression models in general resulted in smaller P values compared with the LMMs-based models; however, the differences in P values were mostly minor. Disregarding family structure had little impact on determining disease-associated SNPs at genome-wide level of significance (i.e., P < 5E-08) as the four P values resulted from the tested methods for any SNP were all below or all above 5E-08. Nevertheless, larger discrepancies were detected between logistic regression and LMMs-based models at suggestive level of significance (i.e., of 5E-08 ≤ P < 5E-06). The SNP effects estimated by the logistic regression models were not statistically different from those estimated by GLMMs that implemented Wald's test. However, several SNP effects were significantly different from their counterparts in LMMs analyses. We suggest that fitting GLMMs with Wald's test on a pre-selected subset of SNPs obtained from logistic regression models can ensure the balance between the speed of analyses and the accuracy of parameters.

Early Versus Late Surgical Decompression for Acute Spinal Cord Injury: An Individual Patient Data Meta-Analysis of 1548 Patients

Abstract INTRODUCTION We sought to leverage the statistical power derived from pooling 4 high-quality prospective datasets to compare sensorimotor recovery with early (&lt; 24 hr) vs late (≥ 24 hr) surgical decompression for acute traumatic spinal cord injury (SCI). METHODS Patients with acute SCI who underwent surgical decompression were identified from 4 prospective, multi-center SCI datasets (NACTN, STASCIS, Sygen, and NASCIS III). Patients were dichotomized into early (&lt; 24 hr) and late (≥ 24 hr) surgery groups. The primary end point was change in ASIA motor score (AMS) at 1-yr. Secondary outcomes included AIS grade and change in ASIA light touch and pin prick scores at 1-yr. One-stage meta-analyses for each outcome were performed by hierarchical mixed-effects regression using a stratified intercept to account for clustering of patients within studies. Fixed-effect covariates were specified for baseline score, age, injury mechanism, AIS grade, neurological level, and steroids. The treatment (early vs late surgery) was specified as a random-effect. RESULTS A total of 1548 patients were eligible. The early surgery group experienced greater improvement than the late surgery group at 1-yr for AMS (MD 4.0, 95% CI 1.7-6.2, P = .001), light touch score (MD 4.6, 95% CI 1.9-7.2, P = .001), and pin prick score (MD 4.2, 95% CI 1.5-6.9, P = .003). Further, on ‘shift analysis’, the early surgery group achieved a more favorable distribution of AIS grades at 1-yr compared to the late surgery group (cOR 1.46, 95% CI 1.14-1.87, P = .003). The effect of early surgery was strongest for cervical SCI (P = .003); however, we observed a trend toward improved recovery with early versus late surgery for thoracic SCI as well (MD 5.2, 95% CI -0.8-11.2, P = .088). CONCLUSION In an individual patient data meta-analysis adjusting for potential confounders, we found early surgery, within 24 hr of injury, to be associated with superior sensorimotor recovery at 1-yr following acute SCI, as compared to late surgery.

Independent Association of Glucose Variability With Hospital Mortality in Adult Intensive Care Patients: Results From the Australia and New Zealand Intensive Care Society Centre for Outcome and Resource Evaluation Binational Registry.

Retrospective, registry data analyses of outcomes. Large, binational registry (Australia and New Zealand Intensive Care Society Centre for Outcome and Resource Evaluation Adult Patient Database repository) of 176 ICUs across Australia and New Zealand. We used 10-year data on nonhyperglycemic patients registered in the Australia and New Zealand Intensive Care Society Centre for Outcome and Resource Evaluation Adult Patient Database repository (n = 290,966). None. Glucose variability was captured using glucose width defined as the difference between highest and lowest blood glucose concentration within first 24 hours of ICU admission. We used hierarchical, mixed effects logistic regression models that accounted for ICU variation and several fixed-effects covariates. Glucose width was specifically and independently associated with hospital mortality. The association of blood glucose variability with mortality remained significant (odds ratio for highest vs lowest quartile of glucose, 1.43; 95% CI, 1.32-1.55; p < 0.001) even after adjusting for the baseline risk of mortality, midpoint blood glucose level, occurrence of hypoglycemia and inter-ICU variation. Mixed effects modeling showed that there was a statistically significant variation in this association across ICUs. Our study demonstrates that glucose variability is independently associated with hospital mortality in critically ill adult patients. Inclusion of correction for glucose variability in glycemic control protocols needs to be investigated in future studies.

Alleviating Spatial Confounding for Areal Data Problems by Displacing the Geographical Centroids

Spatial confounding between the spatial random effects and fixed effects covariates has been recently discovered and showed that it may bring misleading interpretation to the model results. Techniques to alleviate this problem are based on decomposing the spatial random effect and fitting a restricted spatial regression. In this paper, we propose a different approach: a transformation of the geographic space to ensure that the unobserved spatial random effect added to the regression is orthogonal to the fixed effects covariates. Our approach, named SPOCK, has the additional benefit of providing a fast and simple computational method to estimate the parameters. Also, it does not constrain the distribution class assumed for the spatial error term. A simulation study and real data analyses are presented to better understand the advantages of the new method in comparison with the existing ones.

Decreased Injection Site Pain Associated with Phosphate-Free Etanercept Formulation in Rheumatoid Arthritis or Psoriatic Arthritis Patients: A Randomized Controlled Trial.

IntroductionEtanercept, a tumor necrosis factor inhibitor, is used to treat rheumatoid arthritis (RA) and psoriatic arthritis (PsA), and is administered via subcutaneous injection. Injection site pain (ISP) associated with subcutaneous administration may affect compliance or hinder initiation of prescribed medications. To improve the patient experience, a new phosphate-free formulation of etanercept was evaluated for reduced ISP associated with administration.MethodsThis phase 3b, multicenter, randomized, double-blind, cross-over study compared the prior formulation of etanercept to a phosphate-free formulation. Etanercept-naïve adults with RA or PsA indicated for treatment with etanercept were eligible. Patients were randomized (1:1) to receive both etanercept formulations (50 mg) in one of two crossover sequences: prior formulation followed by phosphate-free formulation (sequence AB) or phosphate-free formulation followed by prior formulation (sequence BA) at visits 1 week apart. Patients self-reported ISP using a fit-for-purpose 100-mm visual analog scale within 30 s after injection. Safety outcomes included incidence of treatment-emergent adverse events. Mixed-effects analysis of variance model was used to assess ISP, with treatment, study period, sequence, and disease indication as fixed-effect covariates and patient-within-sequence as random effect.ResultsA total of 111 patients enrolled (56 sequence AB; 55 sequence BA). Mean ISP score for prior formulation was 23.1 mm and for phosphate-free formulation was 19.1 mm (mean difference − 4 mm; 95% confidence interval: − 8.0, 0.0; P = 0.048). Patients with the highest ISP scores from the prior formulation (by quartile cut points) had the largest reduction in pain with phosphate-free formulation. Injection site reactions were few in number and similar between formulations; no new safety signals were observed.ConclusionsThe new phosphate-free formulation of etanercept had statistically significantly lower mean pain scores than the prior formulation, with largest pain reductions observed among patients who reported highest pain with the prior formulation.Trial RegistrationClinicalTrials.gov: NCT02986139.FundingAmgen Inc, Thousand Oaks, CA USA.

Assessing the discrimination of the Kidney Donor Risk Index/Kidney Donor Profile Index scores for allograft failure and estimated glomerular filtration rate in Ireland's National Kidney Transplant Programme.

BackgroundThe Kidney Donor Risk Index (KDRI)/Kidney Donor Profile Index (KDPI) is relied upon for donor organ allocation in the USA, based on its association with graft failure in time-to-event models. However, the KDRI/KDPI has not been extensively evaluated in terms of predictive metrics for graft failure and allograft estimated glomerular filtration rate (eGFR) outside of the USA.MethodsWe performed a retrospective analysis of outcomes in the Irish National Kidney Transplant Service Registry for the years 2006–13. Associations of the KDRI/KDPI score with eGFR at various time points over the follow-up and ultimate graft failure were modelled.ResultsA total of 772 patients had complete data regarding KDRI/KDPI calculation and 148 of these allografts failed over the follow-up. The median and 25–75th centile for KDRI/KDPI was 51 (26–75). On repeated-measures analysis with linear mixed effects models, the KDRI/KDPI (fixed effect covariate) associated with eGFR over 5 years: eGFR = −0.25 (standard error 0.02; P < 0.001). The variability in eGFR mathematically accounted for by the KDRI/KDPI score was only 21%. The KDRI/KDPI score did not add significantly to graft failure prediction above donor age alone (categorized as > and <50 years of age) when assessed by the categorical net reclassification index.ConclusionsIn this cohort, while the KDRI/KDPI was predictive of eGFR over the follow-up, it did not provide additive discrimination above donor age alone in terms of graft failure prediction. Therefore it is unlikely to help inform decisions regarding kidney organ allocation in Ireland.

Relations between preschoolers’ mathematical language understanding and specific numeracy skills

Understanding of mathematical language is critical for numeracy skill development. However, prior research has focused exclusively on relations between mathematical language and numeracy skills, broadly measured. Thus, the aim of the current study was to explore more targeted relations between preschoolers’ mathematical language and specific numeracy skills (e.g., cardinality, numeral comparison). The participants were 124 preschoolers aged 3.52 to 6.03 years (M = 4.78 years, SD = 0.53). Children were assessed on a battery of early numeracy skills and mathematical language as well as expressive vocabulary. Mixed-effects regression models were conducted, with school as a random effect and age, gender, and parental education as fixed effects covariates. Results indicated that mathematical language was significantly related to most numeracy skills, including verbal counting, one-to-one correspondence, numeral identification, cardinality, comparisons of sets and/or numerals, ordering numerals, and story problems. As hypothesized, mathematical language was not significantly related to either subitizing or formal addition because these skills are independent of general language ability. Importantly, mathematical language was generally more proximal to each of these numeracy skills than was general language. These results provide direction for future research, in particular for the development of more precise measures to identify children at risk for mathematics difficulties as well as the incorporation of focused mathematical language instruction within early mathematics interventions.

Non-concave penalization in linear mixed-effect models and regularized selection of fixed effects

Mixed-effect models are very popular for analyzing data with a hierarchical structure, e.g. repeated observations within subjects in a longitudinal design, patients nested within centers in a multicenter design. However, recently, due to the medical advances, the number of fixed effect covariates collected from each patient can be quite large, e.g. data on gene expressions of each patient, and all of these variables are not necessarily important for the outcome. So, it is very important to choose the relevant covariates correctly for obtaining the optimal inference for the overall study. On the other hand, the relevant random effects will often be low-dimensional and pre-specified. In this paper, we consider regularized selection of important fixed effect variables in linear mixed-effects models along with maximum penalized likelihood estimation of both fixed and random effect parameters based on general non-concave penalties. Asymptotic and variable selection consistency with oracle properties are proved for low-dimensional cases as well as for high-dimensionality of non-polynomial order of sample size (number of parameters is much larger than sample size). We also provide a suitable computationally efficient algorithm for implementation. Additionally, all the theoretical results are proved for a general non-convex optimization problem that applies to several important situations well beyond the mixed model set-up (like finite mixture of regressions etc.) illustrating the huge range of applicability of our proposal.

Provider and Patient Determinants of Generic Levothyroxine Prescribing: An Electronic Health Records-Based Study.

Despite the availability of generic levothyroxine products for more than a decade, uptake of these products is poor. We sought to evaluate determinants of generic prescribing of levothyroxine. In a cross-sectional analysis of electronic health records data between 2010 and 2013, we identified adult patients with a levothyroxine prescription from a primary-care physician (PCP) or endocrinologist. We used mixed-effect logistic regression models with random intercepts for prescribing provider to examine predictors of generic levothyroxine prescribing. Models include patient, prescription, and provider fixed-effect covariates. Odds ratios (ORs) and 95% CIs were generated. Between-provider random variation was quantified by the intraclass correlation coefficient (ICC). Study patients (n = 63 838) were clustered among 941 prescribing providers within 25 ambulatory care clinics. The overall prevalence of generic prescribing of levothyroxine was 73%. In the multivariable mixed-effect model, patients were significantly less likely to receive generic levothyroxine from an endocrinologist than a PCP (OR = 0.43; 95% CI = 0.33-0.55; P < 0.001). Women were less likely to receive generic levothyroxine than men from endocrinologists (OR = 0.68; 95% CI = 0.59-0.78; P < 0.001) but not from PCPs. Between-provider variation in generic prescribing was 18.3% in the absence of fixed-effect covariates and could be explained marginally by patient, prescription, and provider factors (ICC = 15.9%). Generic levothyroxine prescribing differed by PCPs and endocrinologists. Residual variation in generic prescribing, after accounting for measurable factors, indicates the need for provider interventions or patient education aimed at improving levothyroxine generic uptake.

Factors associated with milk processing characteristics predicted by mid-infrared spectroscopy in a large database of dairy cows

Despite milk processing characteristics being important quality traits, little is known about the factors underlying their variability, due primarily to the resources required to measure these characteristics in a sufficiently large population. Cow milk coagulation properties (rennet coagulation time, curd-firming time, curd firmness 30 and 60 min after rennet addition), heat coagulation time, casein micelle size, and pH were generated from available mid-infrared spectroscopy prediction models. The prediction models were applied to 136,807 spectra collected from 9,824 Irish dairy cows from research and commercial herds. Sources of variation were investigated using linear mixed models that included the fixed effects of calendar month of test; milking time in the day; linear regressions on the proportion of Friesian, Jersey, Montbéliarde, Norwegian Red, and "other" breeds in the cow; coefficients of heterosis and of recombination loss; parity; stage of lactation; and the 2-way interaction parity × stage of lactation. Within- and across-parity cow effects, contemporary group, and a residual term were also included as random effects in the model. Supplementary analyses considered the inclusion of either test-day milk yield or milk protein concentration as fixed-effects covariates in the multiple regression models. Milk coagulation properties were most favorable (i.e., short rennet coagulation time and strong curd firmness) for cheese manufacturing in early lactation, concurrent with the lowest values of both pH and casein micelle size. Milk coagulation properties and pH deteriorated in mid lactation but improved toward the end of lactation. In direct contrast, heat coagulation time was more favorable in mid lactation and less suitable (i.e., shorter) for high temperature treatments in both early and late lactation. Relative to multiparous cows, primiparous cows, on average, yielded milk with shorter rennet coagulation time and longer heat coagulation time. Milk from the evening milking session had shorter rennet coagulation time and greater curd firmness, as well as lower heat coagulation time and lower pH compared with milk from the morning session. Jersey cows, on average, yielded milk more suitable for cheese production rather than for milk powder production. When protein concentration was included in the model, the improvement of milk coagulation properties toward the end of lactation was no longer apparent. Results from the present study may aid in decision-making for milk manufacturing, especially in countries characterized by a seasonal supply of fresh milk.