A REML method for the evidence-splitting model in network meta-analysis.

Checking for possible inconsistency between direct and indirect evidence is an important task in network meta-analysis. Recently, an evidence-splitting (ES) model has been proposed, that allows separating direct and indirect evidence in a network and hence assessing inconsistency. A salient feature of this model is that the variance for heterogeneity appears in both the mean and the variance structure. Thus, full maximum likelihood (ML) has been proposed for estimating the parameters of this model. Maximum likelihood is known to yield biased variance component estimates in linear mixed models, and this problem is expected to also affect the ES model. The purpose of the present paper, therefore, is to propose a method based on residual (or restricted) maximum likelihood (REML). Our simulation shows that this new method is quite competitive to methods based on full ML in terms of bias and mean squared error. In addition, some limitations of the ES model are discussed. While this model splits direct and indirect evidence, it is not a plausible model for the cause of inconsistency.

Notes on correctness of p-values when analyzing experiments using SAS and R.

It is commonly believed that if a two-way analysis of variance (ANOVA) is carried out in R, then reported p-values are correct. This article shows that this is not always the case. Results can vary from non-significant to highly significant, depending on the choice of options. The user must know exactly which options result in correct p-values, and which options do not. Furthermore, it is commonly supposed that analyses in SAS and R of simple balanced experiments using mixed-effects models result in correct p-values. However, the simulation study of the current article indicates that frequency of Type I error deviates from the nominal value. The objective of this article is to compare SAS and R with respect to correctness of results when analyzing small experiments. It is concluded that modern functions and procedures for analysis of mixed-effects models are sometimes not as reliable as traditional ANOVA based on simple computations of sums of squares.

Biological variation of biochemical urine and serum analytes in healthy dogs.

BACKGROUND: Biological variation (BV) of urinary (U) biochemical analytes has not been described in absolute terms, let alone as a ratio of the U-creatinine or fractional excretion in healthy dogs. These analytes are potential diagnostic tools for different types of kidney damage and electrolyte disorders in dogs. OBJECTIVES: We aimed to investigate the BV of specific gravity, osmolality, creatinine, urea, protein, glucose, chloride, sodium, potassium, calcium, and phosphate in urine from healthy pet dogs. METHODS: Blood and urine samples from 13 dogs were collected once weekly for 8 weeks. Samples were analyzed in duplicate and in randomized order. For each sample, U-analyte and serum concentrations were measured, and U-analyte/U-creatinine and fractional excretion (FE) were calculated. Components of variance, estimated by restricted maximum likelihood, were used to determine within-subject variation (CVI ), between-subject variation (CVG ), and analytical variation (CVA ). Index of individuality (II) and reference change values were calculated. RESULTS: CVI for all urine analytes varied between 12.6% and 35.9%, except for U-sodium, U-sodium/U-Cr, and FE-sodium, which had higher CVI s (59.5%-60.7%). For U-protein, U-sodium, U-potassium, U-sodium/U-creatinine, FE-urea, FE-glucose, FE-sodium, FE-potassium, and FE-phosphate II were low, indicating that population-based RIs were appropriate. The remaining analytes had an intermediate II, suggesting that population-based RIs should be used with caution. CONCLUSION: This study presents information on the biological variation of urinary and serum biochemical analytes from healthy dogs. These data are important for an appropriate interpretation of laboratory results.

Projecting results of zoned multi-environment trials to new locations using environmental covariates with random coefficient models: accuracy and precision.

KEY MESSAGE: We propose the utilisation of environmental covariates in random coefficient models to predict the genotype performances in new locations. Multi-environment trials (MET) are conducted to assess the performance of a set of genotypes in a target population of environments. From a grower's perspective, MET results must provide high accuracy and precision for predictions of genotype performance in new locations, i.e. the grower's locations, which hardly ever coincide with the locations at which the trials were conducted. Linear mixed modelling can provide predictions for new locations. Moreover, the precision of the predictions is of primary concern and should be assessed. Besides, the precision can be improved when auxiliary information is available to characterize the targeted locations. Thus, in this study, we demonstrate the benefit of using environmental information (covariates) for predicting genotype performance in some new locations for Swedish winter wheat official trials. Swedish MET locations can be stratified into zones, allowing borrowing information between zones when best linear unbiased prediction (BLUP) is used. To account for correlations between zones, as well as for intercepts and slopes for the regression on covariates, we fitted random coefficient (RC) models. The results showed that the RC model with appropriate covariate scaling and model for covariate terms improved the precision of predictions of genotypic performance for new locations. The prediction accuracy of the RC model was competitive compared to the model without covariates. The RC model reduced the standard errors of predictions for individual genotypes and standard errors of predictions of genotype differences in new locations by 30-38% and 12-40%, respectively.

Homeostasis model assessment, serum insulin and their relation to body fat in cats.

BACKGROUND: Obesity is associated with insulin resistance (IR) and considered a risk factor for diabetes mellitus (DM) in cats. It has been proposed that homeostasis model assessment (HOMA-IR), which is the product of fasting serum insulin (mU/L) and glucose (mmol/L) divided by 22.5, can be used to indicate IR. The objectives of this study were threefold: (i) to evaluate associations between body fat, fasting insulin, and HOMA-IR, (ii) to determine population-based reference interval of HOMA-IR in healthy lean cats, and (iii) to evaluate biological variation of HOMA-IR and fasting insulin in cats. RESULTS: 150 cats were grouped as lean or overweight based on body condition score and in 68 of the cats body fat percentage (BF%) was estimated by computed tomography. Fasting serum insulin and glucose concentrations were analysed. Statistical differences in HOMA-IR and insulin between overweight or lean cats were evaluated using Wilcoxon rank-sum test. Robust method with Box-Cox transformation was used for calculating HOMA-IR reference interval in healthy lean cats. Relations between BF% and HOMA-IR and insulin were evaluated by regression analysis. Restricted maximum likelihood ratio was used to calculate indices of biological variation of HOMA-IR and insulin in seven cats. There were significant differences between groups with overweight cats (n = 77) having higher HOMA-IR (p < 0.0001) and insulin (p = 0.0002) than lean cats (n = 73). Reference interval for HOMA-IR in lean cats was 0.1-3.0. HOMA-IR and fasting insulin concentrations showed similar significant positive association with BF% (p = 0.0010 and p = 0.0017, respectively). Within-animal coefficient of variation of HOMA-IR and insulin was 51% and 49%, respectively. CONCLUSIONS: HOMA-IR and fasting insulin higher in overweight than lean cats and correlate to BF%. The established population-based reference interval for HOMA-IR as well as the indices of biological variation for HOMA-IR and fasting insulin may be used when interpreting HOMA-IR and fasting insulin in cats. Further studies are needed to evaluate if HOMA-IR or fasting insulin is useful for identifying cats at risk of developing DM.

Testing multiplicative terms in AMMI and GGE models for multienvironment trials with replicates.

KEY MESSAGE: For analysing multienvironment trials with replicates, a resampling-based method is proposed for testing significance of multiplicative interaction terms in AMMI and GGE models, which is superior compared to contending methods in robustness to heterogeneity of variance. The additive main effects and multiplicative interaction model and genotype main effects and genotype-by-environment interaction model are commonly used for the analysis of multienvironment trial data. Agronomists and plant breeders are frequently using these models for cultivar trials repeated across different environments and/or years. In these models, it is crucial to decide how many significant multiplicative interaction terms to retain. Several tests have been proposed for this purpose when replicate data are available; however, all of them assume that errors are normally distributed with a homogeneous variance. Here, we propose resampling-based methods for multienvironment trial data with replicates, which are free from these distributional assumptions. The methods are compared with competing parametric tests. In an extensive simulation study based on two multienvironment trials, it was found that the proposed methods performed well in terms of Type-I error rates regardless of the distribution of errors. The proposed method even outperforms the robust [Formula: see text] test when the assumptions of normality and homogeneity of variance are violated.

Generalized prediction intervals for treatment effects in random-effects models.

This article derives generalized prediction intervals for random effects in linear random-effects models. For balanced and unbalanced data in two-way layouts, models are considered with and without interaction. Coverage of the proposed generalized prediction intervals was estimated in a simulation study based on an agricultural field experiment. Generalized prediction intervals were compared with prediction intervals based on the restricted maximum likelihood (REML) procedure and the approximate methods of Satterthwaite and Kenward and Roger. The simulation study showed that coverage of generalized prediction intervals was closer to the nominal level 0.95 than coverage of prediction intervals based on the REML procedure.

Generalized Confidence Intervals for Intra- and Inter-subject Coefficients of Variation in Linear Mixed-effects Models.

Linear mixed-effects models are linear models with several variance components. Models with a single random-effects factor have two variance components: the random-effects variance, i. e., the inter-subject variance, and the residual error variance, i. e., the intra-subject variance. In many applications, it is practice to report variance components as coefficients of variation. The intra- and inter-subject coefficients of variation are the square roots of the corresponding variances divided by the mean. This article proposes methods for computing confidence intervals for intra- and inter-subject coefficients of variation using generalized pivotal quantities. The methods are illustrated through two examples. In the first example, precision is assessed within and between runs in a bioanalytical method validation. In the second example, variation is estimated within and between main plots in an agricultural split-plot experiment. Coverage of generalized confidence intervals is investigated through simulation and shown to be close to the nominal value.

Parametric bootstrap methods for testing multiplicative terms in GGE and AMMI models.

The genotype main effects and genotype-by-environment interaction effects (GGE) model and the additive main effects and multiplicative interaction (AMMI) model are two common models for analysis of genotype-by-environment data. These models are frequently used by agronomists, plant breeders, geneticists and statisticians for analysis of multi-environment trials. In such trials, a set of genotypes, for example, crop cultivars, are compared across a range of environments, for example, locations. The GGE and AMMI models use singular value decomposition to partition genotype-by-environment interaction into an ordered sum of multiplicative terms. This article deals with the problem of testing the significance of these multiplicative terms in order to decide how many terms to retain in the final model. We propose parametric bootstrap methods for this problem. Models with fixed main effects, fixed multiplicative terms and random normally distributed errors are considered. Two methods are derived: a full and a simple parametric bootstrap method. These are compared with the alternatives of using approximate F-tests and cross-validation. In a simulation study based on four multi-environment trials, both bootstrap methods performed well with regard to Type I error rate and power. The simple parametric bootstrap method is particularly easy to use, since it only involves repeated sampling of standard normally distributed values. This method is recommended for selecting the number of multiplicative terms in GGE and AMMI models. The proposed methods can also be used for testing components in principal component analysis.