Bivariate Mixed Model Research Articles

Genetic analysis of udder conformation traits derived from automatic milking system recording in dairy cows

Cartesian teat coordinates measured by automatic milking systems (AMS) provide new opportunities to record udder conformation traits and to study changes in udder conformation genetically and phenotypically within and between parities. The objective of this study was to estimate heritabilities and repeatabilities of AMS-based udder conformation traits within parities, to estimate genetic correlations between parities for AMS-based udder conformation traits, and to estimate genetic correlations between AMS-based udder conformation traits and classifier-based udder conformation traits, longevity, and udder health. Data from 70 herds, including 12,663 first-parity cows, 10,206 second-parity cows, and 7,627 third-parity cows, were analyzed using univariate and bivariate mixed animal models. Heritabilities of the AMS udder conformation traits were large (0.37-0.67) and genetic correlations between the AMS udder conformation traits and classifier-based traits were strong (>0.91). Repeatabilities within parities were large as well (0.89-0.97), indicating that a single record on udder conformation per lactation reflects udder conformation well. Genetic correlations of AMS udder conformation traits between parities were strong (0.88-1.00) and were stronger than the permanent environmental correlations. This shows that udder conformation changes over parities, but this change is mostly due to nongenetic factors. Based on these results, the current herd classification system, where cows are scored on udder conformation once in first parity, is sufficient. The AMS udder conformation traits as defined in this study have limited value as replacement for classifier-based udder conformation traits because they have smaller genetic correlations with functional traits than classifier-based traits. In summary, udder conformation hardly changes genetically between parities and is highly repeatable within parities. Udder conformation traits based on AMS need fine-tuning before they can replace classifier-based traits, and AMS teat coordinates probably contain additional information about udder health that is yet to be explored.

Diagnostic significance of miR-210 as a potential tumor biomarker of human cancer detection: an updated pooled analysis of 30 articles.

A large number of studies have explored the diagnostic value of miR-210 as a potential diagnostic cancer biomarker to detect various cancers in patients. However, the results of its diagnostic accuracy and reliability in individual studies are still inconsistent. Therefore, we conducted this updated pooled analysis to derive a more reliable conclusion of the overall accuracy of miR-210 in cancer detection and diagnosis. A comprehensive literature search was performed using the PubMed, Cochrane Library, Web of Science, China National Knowledge Infrastructure, and Wanfang databases. The quality of all eligible studies was scored according to Quality Assessment of Diagnostic Accuracy Studies-2 guidelines. The bivariate mixed model was applied to pooled sensitivity, specificity, likelihood ratios, and diagnostic ORs. The summary receiver operator characteristic (SROC) curve and the hierarchical SROC models were used to check overall diagnostic performance. Thirty articles with 2,304 patients and 1,673 controls were included in this study. The pooled parameters calculated from all studies are as follows: sensitivity −0.74 (95% CI: 0.68–0.79), specificity −0.79 (95% CI: 0.74–0.83), positive likelihood ratio −3.57 (95% CI: 2.85–4.47), negative likelihood ratio −0.32 (95% CI: 0.26–0.40), diagnostic OR −10.98 (95% CI: 7.55–15.98), SROC −0.84 (95% CI: 0.80–0.87). All of these results revealed that miR-210 had relatively moderate accuracy in distinguishing patients with various cancers from all other individuals. However, well-designed prospective studies with large sample sizes using different groups of the population are urgently warranted to confirm our findings.

Diagnostic Accuracy of Anti-Carbamylated Protein Antibodies in Rheumatoid Arthritis: a Systematic Review and Meta-Analysis.

The purpose of this study was to estimate the diagnostic accuracy of anti-carbamylated protein (anti-CarP) antibodies in rheumatoid arthritis. We searched the PubMed, EMBASE, Cochrane Library, Web of Science, and Scopus databases for studies published before January 1, 2019. Two investigators independently evaluated studies to determine their inclusion in the analysis, assess their quality, and extract the relevant data. The articles were assessed with the Quality Assessment of Diagnostic Accuracy Studies tool, and a bivariate mixed effects model was used to estimate the diagnostic indexes across studies. We included 16 published studies in this meta-analysis. The pooled sensitivity and specificity of anti-CarP were 43.1% and 94.4%, respectively. The area under the summary receiver operator characteristic curve was 0.55. The specificity estimates were highly heterogeneous, which could be partly explained by the higher specificity in the healthy control group (43.0%, 96.8%) than in the other disease group (43.4%, 89.8%). Anti-CarP antibodies have a relatively low sensitivity and high specificity for rheumatoid arthritis. However, the specificity was lower in the other disease subgroups than in the healthy controls.

THE USE OF CONTRASTS IN MULTIVARIATE NONLINEAR MIXED MODELS TO COMPARE TREATMENTS IN LONGITUDINAL FACTORIAL EXPERIMENTS

The purpose of this study was to establish contrasts in multivariate nonlinear mixed models to verify the effects of treatments in experiments with longitudinal data and multiple responses. The evaluated nonlinear functions were the three parameters curves logistic, Gompertz and von Bertalanffy. The random variables were added to the fixed parameters, asymptote α , abscissa of the inflection point β, and parameter γ. The best fitted model was expanded with covariates, which establish orthogonal contrasts, in order to verify main effects and interactions in factorial experiments. The methodology was applied to analyse data of an experiment with citrus, in which case the logistic bivariate mixed effects model was the best fit. The chosen model allowed comparisons between treatments in a global context of more than one dependent variable and throughout the measurement period.

Diagnostic accuracy of serum inflammatory markers in late fracture-related infection: a systematic review and meta-analysis.

To assess the diagnostic value of C-reactive protein (CRP), leucocyte count (LC), and erythrocyte sedimentation rate (ESR) in late fracture-related infection (FRI). PubMed, Embase, and Cochrane databases were searched focusing on the diagnostic value of CRP, LC, and ESR in late FRI. Sensitivity and specificity combinations were extracted for each marker. Average estimates were obtained using bivariate mixed effects models. A total of 8284 articles were identified but only six were suitable for inclusion. Sensitivity of CRP ranged from 60.0% to 100.0% and specificity from 34.3% to 85.7% in all publications considered. Five articles were pooled for meta-analysis, showing a sensitivity and specificity of 77.0% and 67.9%, respectively. For LC, this was 22.9% to 72.6%, and 73.5% to 85.7%, respectively, in five articles. Four articles were pooled for meta-analysis, resulting in a 51.7% sensitivity and 67.1% specificity. For ESR, sensitivity and specificity ranged from 37.1% to 100.0% and 59.0% to 85.0%, respectively, in five articles. Three articles were pooled in meta-analysis, showing a 45.1% sensitivity and 79.3% specificity. Four articles analyzed the value of combined inflammatory markers, reporting an increased diagnostic accuracy. These results could not be pooled due to heterogeneity. The serum inflammatory markers CRP, LC, and ESR are insufficiently accurate to diagnose late FRI, but they may be used as a suggestive sign in its diagnosis.

Genetic and phenotypic correlation for breeding relevant traits in Dianthus caryophyllus L.

Indirect selection is a useful tool for the selection of complex traits like vase life (VL) of cut carnations. Our objective, therefore, was to assess the relative merit of indirect selection compared to direct selection. Phenotypic and genetic correlations were calculated for traits such as VL, flower diameter (FD) of standard carnation, stem length (SL) and count of flowers (CF) in mini carnations, phenotyped in three different experiments. The experiments are special in that some traits are measured in a second phase. Data were analyzed by bivariate mixed models considering in which phase the trait was measured. Some relevant genetic correlations were found in VL and FD and the traits SL and CF, but our results indicate that indirect selection would show no benefit compared to direct selection. However, the combination of direct and indirect selection could improve breeding of long-lasting cut carnations.

The Evaluation of Bivariate Mixed Models in Meta-analyses of Diagnostic Accuracy Studies with SAS, Stata and R.

Meta-analyses require a thoroughly planned procedure to obtain unbiased overall estimates. From a statistical point of view not only model selection but also model implementation in the software affects the results. The present simulation study investigates the accuracy of different implementations of general and generalized bivariate mixed models in SAS (using proc mixed, proc glimmix and proc nlmixed), Stata (using gllamm, xtmelogit and midas) and R (using reitsma from package mada and glmer from package lme4). Both models incorporate the relationship between sensitivity and specificity - the two outcomes of interest in meta-analyses of diagnostic accuracy studies - utilizing random effects. Model performance is compared in nine meta-analytic scenarios reflecting the combination of three sizes for meta-analyses (89, 30 and 10 studies) with three pairs of sensitivity/specificity values (97%/87%; 85%/75%; 90%/93%). The evaluation of accuracy in terms of bias, standard error and mean squared error reveals that all implementations of the generalized bivariate model calculate sensitivity and specificity estimates with deviations less than two percentage points. proc mixed which together with reitsma implements the general bivariate mixed model proposed by Reitsma rather shows convergence problems. The random effect parameters are in general underestimated. This study shows that flexibility and simplicity of model specification together with convergence robustness should influence implementation recommendations, as the accuracy in terms of bias was acceptable in all implementations using the generalized approach.

Allowing for the structure of a designed experiment when estimating and testing trait correlations

AbstractCrop scientists occasionally compute sample correlations between traits based on observed data from designed experiments, and this is often accompanied by significance tests of the null hypothesis that traits are uncorrelated. This simple approach does not account for effects due to the randomization layout and treatment structure of the experiments and hence statistical inference based on standard procedures is not appropriate. The present paper describes how valid inferences accounting for all relevant effects can be obtained using bivariate mixed linear model procedures. A salient feature of the approach is that the bivariate model is commensurate with the model used for univariate analysis of individual traits and allows bivariate correlations to be computed at the level of effects. Heterogeneity of correlations between effects can be assessed by likelihood ratio tests or by graphical inspection of bivariate scatter plots of effect estimates. if heterogeneity is found to be substantial, it is crucial to focus on the correlation of effects, and usually, the main interest will be in the treatment effects. If heterogeneity is judged to be negligible, the marginal correlation can be estimated from the bivariate model for an overall assessment of association. The proposed methods are illustrated using four examples. Hints are given to alternative routes of analysis accounting for all treatment and design effects such as regression with groups and analysis of covariance.

Performance of spleen stiffness measurement in prediction of clinical significant portal hypertension: A meta-analysis

Our purpose was to evaluate the correlation between spleen stiffness (SS) measured by ultrasound-based elastography and hepatic venous pressure gradient (HVPG) and assess the accuracy of SS in detecting clinical significant portal hypertension (CSPH) and severe portal hypertension. Nine studies were included from thorough literature research and selection processes. A random model was used to analyze the correlation between HVPG and SS. We adopted the bivariate mixed effects model to assess the diagnostic performance. Regarding to correlation between SS and HVPG, the summary correlation coefficient was 0.72 (95% confidence interval [CI], 0.63-0.80). In detection of CSPH, the sensitivity, specificity, AUC and DOR were: 0.88 (0.70-0.96), 0.84 (0.72-0.92), 0.92 (0.89-0.94) and 38 (17-84) for CSPH, respectively; and 0.92 (0.82-0.96), 0.79 (0.72-0.85), 0.87 (0.84-0.90) and 41 (17-100) for severe portal hypertension, respectively. Correlation between SS and HVPG was good. Although SS showed good sensitivity and specificity, the different cut-off values and techniques among studies might limit the impact of our results on clinical practice. Therefore, more high-quality prospective studies are required to evaluate the role of SS in predicting portal hypertension.

Assessing the degree of stratification between closely related Holstein-Friesian populations

Genomic information is an important part of the routine evaluation of dairy cattle and provides the wide availability of animals genotyped using single nucleotide polymorphism (SNP) microarrays. We analyzed 2243 Polish and 2294 German Holstein-Friesian bulls genotyped using the Illumina BovineSNP50 BeadChip. For each bull, estimated breeding values (EBVs) calculated from national routine genetic evaluation were available for production traits and for somatic cell score (SCS). Separately for each population, we estimated SNP haplotypes, pairwise linkage disequilibrium (LD), and SNP effects. The SNP genetic covariance between both populations was estimated using a bivariate mixed model. The average LD was lower in the Polish than in the German population and, with increasing genomic distance, LD decays 1.7 times more rapidly in German than in Polish cattle. The comparison of SNP allele frequencies for base populations estimated separately using Polish and German data revealed a very good agreement. The comparison of genetic effects corresponding to various window lengths defined in bp emerged a systematic pattern: regardless of the length of the compared region, few significant differences were found for production traits, while many were observed for SCS. For each trait, the German population had much higher SNP variances than the Polish population and the genetic covariance estimates were all positive. Depending on traits’ inheritance mode, the additive genetic variation can be stored in many genes following the infinitesimal model (like for SCS) or distributed between genes with high effects and the polygenic “background” (like for production traits). Accounting for those differences has implications on the prospective international genomic evaluation.

Diagnostic ability of intraoperative ultrasound for identifying tumor residual in glioma surgery operation.

Achieving total glioma resection represents a major challenge to neurosurgeons with no distinct margin between tumor and surrounding brain tissue. Many imaging methods are employed in surgery visualization and resection control. We performed this meta-analysis to assess the diagnosis value of intraoperative ultrasound and judged whether ultrasound is a suitable tool in detecting glioma residual. The databases including PubMed, Embase, Web of Science, China National Knowledge Infrastructure (CNKI), Wanfang and Weipu were systematically searched to find out relevant studies and published up to May 5, 2017. A total of 14 studies involving 542 participants met the selection criteria and bivariate mixed effects models were used for analysis. The parameters and their corresponding 95% confidence interval (CI) were computed on Stata 12.0 software. The pooled sensitivity was 0.75 (95%CI: 0.62–0.84), specificity was 0.88 (95%CI: 0.79–0.94), positive likelihood ratios was 6.27 (95%CI: 3.76–10.47), negative likelihood ratios was 0.29 (95%CI: 0.20–0.42), diagnostic odds ratios was 21.83 (95%CI: 14.20–33.55) and area under the curve of summary receiver operator characteristic was 0.89. Stratified meta-analysis showed sensitivity and area under the curve in low-grade glioma were both higher than high-grade glioma. The Deek's plot showed no significant publication bias (t = −1.03, P = 0.33). Intraoperative ultrasound has high overall diagnostic value to identify glioma remnants, especially in low-grade glioma, which shows a benefit for prognosis and life quality of patients. In general, Intraoperative ultrasound is an effective tool for maximizing the extent of glioma resection.

A statistical methodology for estimating assortative mating for phenotypic traits that are labile or measured with error

Abstract Assortative mating in wild populations is commonly reported as the correlation between males’ and females’ phenotypes across mated pairs. Theories of partner selection and quantitative genetics assume that phenotypic resemblance of partners captures associations in “intrinsically determined” trait values. However, when considering traits with a repeatability below one (labile traits or traits measured with error), the correlation between phenotypes of paired individuals can arise from shared environmental effects on the phenotypes of paired individuals or correlated measurement error. We introduce statistical approaches to estimate assortative mating in labile traits or traits measured with error in the presence of shared environmental effects. These approaches include (1) the correlation between the mean phenotypes of males and females, (2) the correlation between randomized values of individuals and (3) the between‐pair correlation derived from a bivariate mixed model. We use simulations to show that the performance of these different approaches depends on the number of repeated measures within individuals or pairs, which is determined by study design, and rates of survival and divorce. We conclude that short‐term environmental effects on phenotypes of paired individuals likely inflate estimates of assortative mating when not statistically accounted for. Our approach allows investigation of this important issue in assortative mating studies for labile traits (e.g. behaviour, physiology, or metabolism) in both socially monogamous and other mating systems, and groupings of individuals outside a mating context.

Resistance against infectious pancreatic necrosis exhibits significant genetic variation and is not genetically correlated with harvest weight in rainbow trout (Oncorhynchus mykiss)

Infectious pancreatic necrosis (IPN) is one of the most prevalent and economically important diseases in rainbow trout aquaculture (Oncorhynchus mykiss). Vaccines as a conventional control measure have shown variable results under production conditions. Genetic improvement for resistance to IPN represents an alternative for the prevention of disease outbreaks. The objective of the present work was to estimate the heritability and genetic correlation for IPN resistance and harvest weight (HW) in rainbow trout. To determine the genetic resistance to the IPN virus, a total of 2278 fingerlings from 58 full-sib families were used, which were challenged with IPN virus to induce the disease. Resistance refers to survival to the disease and was defined as the day of death of each fish. HW was also recorded in 13,241 genetically related individuals from the same population. For the genetic analysis we fitted a bivariate mixed linear model including HW and resistance to IPN as dependent variables; tank:year:sex as a fixed contemporary group and age at harvest as a covariate for HW; and final weight as a covariate for resistance to IPN. The animal effect was included as a random effect for both traits. A random effect associated with common environment was also included for HW. The estimated heritability for IPN resistance was 0.39±0.08 and 0.35±0.06 for HW. Genetic correlation between IPN and HW resistance was not significant (0.05±0.25). The results indicate that the heritability for both traits is moderately high in this population, and that there is no significant genetic correlation between them. The presence of significant genetic variation for both IPN and HW resistance and the absence of genetic correlation between both traits indicate the feasibility of improving them simultaneously by means of artificial selection.

A new R-based statistical software package for fitting multivariate generalized linear mixed models to large and complex datasets

ABSTRACT
 ObjectiveTo demonstrate the efficiency and efficacy of the new freeware MGLMM in the statistical modelling of large and complex datasets.
 ApproachThe new R-based software MGLMM (shortly to become available on the CRAN website https://cran.r-project.org/) can be used to fit a range of different multivariate generalized linear mixed models, thereby permitting a better representation of the multivariate nature of many complex socio-economic processes which may involve a range of different response types. The software MGLMM will also set records for the time taken to fit these complex models to large sets of data. The efficiency and efficacy of this new software will be demonstrated by way of a number of applications.
 One such example will come from the STAR Project in which pupils were followed from kindergarten in 1985 (aged 5) to 1989 (third grade, aged 8). The pupils were assessed on their ability in both mathematics and English on an annual basis between 1985 and 1989. At the end of the kindergarten year, some of the students were reallocated to different class sizes in order ‘to achieve sexual and racial balance and to separate incompatible children’. In the fourth grade, the pupils returned to regular classes and the experiment ended.
 We can regard the two score variables ‘Maths’ and ‘English’ as constituting repeated, correlated, bivariate continuous data. We will use the new package MGLMM to apply a series of bivariate normal mixed models with correlated random effects. These models will allow us to distinguish the role of prior attainment in mathematics and English on current attainment in mathematics and English in an experimental setting, for pupils in small and regular class sizes. The models will permit an examination of the effect of class size on this bivariate response, whilst controlling for a set of individual-level confounding factors such as gender and age.
 ResultsWe will present the results of these analyses, thereby demonstrating the efficacy of MGLMM. We will also compare the performance of the new software with existing packages such as Stata.
 ConclusionsWe will have demonstrated the gain in efficiency through the use of MGLMM, compared to other standard software packages, in the statistical modelling of large and complex datasets.

Energy Intake, Basal Metabolic Rate, and Within-Individual Trade-Offs in Men and Women Training for a Half Marathon: A Reanalysis

Understanding the mechanisms governing energy budgets during periods of high energy expenditure is important from both an evolutionary physiology and behavioral ecology perspective. In particular, we still know too little about the linkages between key components such as daily energy expenditure (DEE), daily energy intake (DEI), and basal metabolic rate (BMR). Westerterp et al. repeatedly measured DEI (self-reported), BMR (respirometry), and body composition (fat mass) in 23 adult subjects as they transitioned from an inactive lifestyle and trained during 44 wk in preparation of running a half marathon. Here, I reanalyzed this data set using bivariate mixed models to partition the phenotypic correlation between DEI and BMR at the among- and within-individual levels. Reported DEI and BMR were positively correlated at the among-individual level (i.e., individuals with high average reported DEI also have a high average BMR). However, reported DEI and BMR were not correlated within individuals. There was a negative within-individual relationship between BMR and surplus energy (i.e., the energy intake above BMR = DEI - BMR), suggesting the presence of compensation mechanisms between BMR and other energy-demanding activities occurring within individuals. Thus, the principles governing energy budget were different at the among- and within-individual levels. To the extent that this situation is applicable to wild animals experiencing different levels of DEE throughout their annual cycle, the results presented here could explain why the relationships between BMR and other components of the energy budget (e.g., activity, growth, reproduction) are often context dependent.

Mean cerebral blood volume is an effective diagnostic index of recurrent and radiation injury in glioma patients: A meta-analysis of diagnostic test.

We conducted a meta-analysis to evaluate the diagnostic values of mean cerebral blood volume for recurrent and radiation injury in glioma patients. We performed systematic electronic searches for eligible study up to August 8, 2016. Bivariate mixed effects models were used to estimate the combined sensitivity, specificity, positive likelihood ratios, negative likelihood ratios, diagnostic odds ratios and their 95% confidence intervals (CIs). Fifteen studies with a total number of 576 participants were enrolled. The pooled sensitivity and specificity of diagnostic were 0.88 (95%CI: 0.82-0.92) and 0.85 (95%CI: 0.68-0.93). The pooled positive likelihood ratio is 5.73 (95%CI: 2.56-12.81), negative likelihood ratio is 0.15 (95%CI: 0.10-0.22), and the diagnostic odds ratio is 39.34 (95%CI:13.96-110.84). The summary receiver operator characteristic is 0.91 (95%CI: 0.88-0.93). However, the Deek's plot suggested publication bias may exist (t=2.30, P=0.039). Mean cerebral blood volume measurement methods seems to be very sensitive and highly specific to differentiate recurrent and radiation injury in glioma patients. The results should be interpreted with caution because of the potential bias.

Diagnostic Accuracy of Methylated SEPT9 for Blood-based Colorectal Cancer Detection: A Systematic Review and Meta-Analysis.

Objectives:More convenient and effective blood-based methods are believed to increase colorectal cancer (CRC) detection adoption. The effectiveness of methylated SPET9 for CRC detection has been reviewed in the newly published recommendation statement by US Preventive Services Task Force (USPSTF), while detailed instructions were not provided, which may be a result of insufficient evidence. Therefore, more evidence is needed to assist practitioners to thoroughly understand the utilization of this special maker.Methods:Based on the standard method, a systematic review and meta-analysis was performed. Quadas-2 was used to assess the methodological quality of studies. Relevant studies were searched and screened from PubMed, Embase and other literature databases up to June 1, 2016. Pooled sensitivity, specificity and diagnostic odds ratio were summarized by bivariate mixed effect model and area under the curve (AUC) was estimated by hierarchical summary receiver operator characteristic curve.Results:25 studies were included for analysis. The pooled sensitivity, specificity and AUC were 0.71, 0.92 and 0.88, respectively. Among the various methods and assays, Epipro Colon 2.0 with 2/3 algorithm was the most effective in colorectal cancer detection. Positive ratio of mSEPT9 was higher in advanced CRC (45% in I, 70% in II, 76% in III, 79% in IV) and lower differentiation (31% in high, 73% in moderate, 90% in low) tissue. However, this marker has poor ability of identifying precancerous lesions according to current evidence.Conclusions:mSEPT9 is a reliable blood-based marker in CRC detection, particularly advanced CRC. Epipro Colon 2.0 with 2/3 algorithm is currently the optimal method and assay to detect CRC.

Clinical utility of microRNA-378 as early diagnostic biomarker of human cancers: a meta-analysis of diagnostic test.

A meta-analysis was performed to evaluate the diagnostic value of miR-378 for detecting human cancers. Systematic electronic searches were conducted in PubMed, Web of Science, Embase, China National Knowledge Infrastructure, and Wanfang from the inception to January 15, 2016. We used the bivariate mixed effects models to estimate the combined sensitivity, specificity, PLRs (positive likelihood ratios), NLR (negative likelihood ratios), DORs (diagnostic odds ratios) and their 95% CI (confidence intervals) for assessing the diagnostic performance of miR-378 for cancers. Twelve studies were included in the meta-analysis, with a total number of 1172 cancer patients and 809 health controls. The overall estimated sensitivity and specificity were 0.75 and 0.74. The pooled PLR was 2.91, NLR was 0.34, DOR was 8.50, and AUC (Area Under the Curve) was 0.81. The subgroup analyses suggested that AUC for plasma-based is higher than serum-based. The overall diagnostic values of miR-378 in the present meta-analyses are moderate accurate for human cancers; The source of specimen has an effect on the diagnostic accuracy. The diagnostic value of serum-based was higher than that of plasma-based.

Screening for Depression in the General Population with the Center for Epidemiologic Studies Depression (CES-D): A Systematic Review with Meta-Analysis.

ObjectiveWe aimed to collect and meta-analyse the existing evidence regarding the performance of the Center for Epidemiologic Studies Depression (CES-D) for detecting depression in general population and primary care settings.MethodSystematic literature search in PubMed and PsychINFO. Eligible studies were: a) validation studies of screening questionnaires with information on the accuracy of the CES-D; b) samples from general populations or primary care settings; c) standardized diagnostic interviews following standard classification systems used as gold standard; and d) English or Spanish language of publication. Pooled sensitivity, specificity, likelihood ratios and diagnostic odds ratio were estimated for several cut-off points using bivariate mixed effects models for each threshold. The summary receiver operating characteristic curve was estimated with Rutter and Gatsonis mixed effects models; area under the curve was calculated. Quality of the studies was assessed with the QUADAS tool. Causes of heterogeneity were evaluated with the Rutter and Gatsonis mixed effects model including each covariate at a time.Results28 studies (10,617 participants) met eligibility criteria. The median prevalence of Major Depression was 8.8% (IQ range from 3.8% to 12.6%). The overall area under the curve was 0.87. At the cut-off 16, sensitivity was 0.87 (95% CI: 0.82–0.92), specificity 0.70 (95% CI: 0.65–0.75), and DOR 16.2 (95% CI: 10.49–25.10). Better trade-offs between sensitivity and specificity were observed (Sensitivity = 0.83, Specificity = 0.78, diagnostic odds ratio = 16.64) for cut-off 20. None of the variables assessed as possible sources of heterogeneity was found to be statistically significant.ConclusionThe CES-D has acceptable screening accuracy in the general population or primary care settings, but it should not be used as an isolated diagnostic measure of depression. Depending on the test objectives, the cut-off 20 may be more adequate than the value of 16, which is typically recommended.

Comment on 'A vine copula mixed effect model for trivariate meta-analysis of diagnostic test accuracy studies accounting for disease prevalence'.

A bivariate copula mixed model has been recently proposed to synthesize diagnostic test accuracy studies and it has been shown that it is superior to the standard generalized linear mixed model in this context. Here, we call trivariate vine copulas to extend the bivariate meta-analysis of diagnostic test accuracy studies by accounting for disease prevalence. Our vine copula mixed model includes the trivariate generalized linear mixed model as a special case and can also operate on the original scale of sensitivity, specificity, and disease prevalence. Our general methodology is illustrated by re-analyzing the data of two published meta-analyses. Our study suggests that there can be an improvement on trivariate generalized linear mixed model in fit to data and makes the argument for moving to vine copula random effects models especially because of their richness, including reflection asymmetric tail dependence, and computational feasibility despite their three dimensionality.