Accounting for nonignorable verification bias in assessment of diagnostic tests

A "gold" standard test, providing definitive verification of disease status, may be quite invasive or expensive. Current technological advances provide less invasive, or less expensive, diagnostic tests. Ideally, a diagnostic test is evaluated by comparing it with a definitive gold standard test. However, the decision to perform the gold standard test to establish the presence or absence of disease is often influenced by the results of the diagnostic test, along with other measured, or not measured, risk factors. If only data from patients who received the gold standard test were used to assess the test performance, the commonly used measures of diagnostic test performance--sensitivity and specificity--are likely to be biased. Sensitivity would often be higher, and specificity would be lower, than the true values. This bias is called verification bias. Without adjustment for verification bias, one may possibly introduce into the medical practice a diagnostic test with apparent, but not truly, high sensitivity. In this article, verification bias is treated as a missing covariate problem. We propose a flexible modeling and computational framework for evaluating the performance of a diagnostic test, with adjustment for nonignorable verification bias. The presented computational method can be utilized with any software that can repetitively use a logistic regression module. The approach is likelihood-based, and allows use of categorical or continuous covariates. An explicit formula for the observed information matrix is presented, so that one can easily compute standard errors of estimated parameters. The methodology is illustrated with a cardiology data example. We perform a sensitivity analysis of the dependency of verification selection process on disease.     

Comparative Molecular Evolution of Primary (<Emphasis Type="BoldItalic">Buchnera</Emphasis>) and Secondary Symbionts of Aphids Based on Two Protein-Coding Genes

A+T content, phylogenetic relationships, codon usage, evolutionary rates, and ratio of synonymous versus non-synonymous substitutions have been studied in partial sequences of the atpD and aroQ/pheA genes of primary (Buchnera) and secondary symbionts of aphids and a set of selected non-symbiotic bacteria, belonging to the five subdivisions of the Proteobacteria. Compared to the homologous genes of the last group, both genes belonging to Buchnera behave in a similar way, showing a higher A+T content, forming a monophyletic group, a loss in codon bias, especially in third base position, an evolutionary acceleration and an increase in the number of non-synonymous substitutions, confirming previous results reported elsewhere for other genes. When available, these properties have been partly observed with the secondary symbionts, but with values that are intermediate between Buchnera and free living Proteobacteria. They show high A+T content, but not as high as Buchnera, a non-solved phylogenetic position between Buchnera, and the other γ-Proteobacteria, a loss in codon bias, again not as high as in Buchnera and a significant evolutionary acceleration in the case of the three atpD genes, but not when considering aroQ/pheA genes. These results give support to the hypothesis that they are symbionts at different stages of the symbiotic accommodation to the host.     

Phylogeny Derived from Coding Retroviral Genome Organization

When divergence between viral species is large, the analysis and comparison of nucleotide or protein sequences are dependent on mutation biases and multiple substitutions per site leading, among other things, to the underestimation of branch lengths in phylogenetic trees. To avoid the problem of multiply substituted sites, a method not directly based on the nucleic or protein sequences has been applied to retroviruses. It consisted of asking questions about genome structure or organization, and gene function, the series of answers creating coded sequences analyzed by phylogenic software. This method recovered the principal retroviral groups such as the lentiviruses and spumaviruses and highlighted questions and answers characteristic of each group of retroviruses. In general, there was reasonable concordance between the coded genome methodology and that based on conventional phylogeny of the integrase protein sequence, indicating that integrase was fixing mutations slowly enough to marginalize the problem of multiple substitutions at sites. To a first approximation, this suggests that the acquisition of novel genetic features generally parallels the fixation of amino acid substitutions. Received: 18 May 2001 / Accepted: 7 September 2001     

The effect of single nucleotide polymorphism identification strategies on estimates of linkage disequilibrium

At present there is tremendous interest in characterizing the magnitude and distribution of linkage disequilibrium (LD) throughout the human genome, which will provide the necessary foundation for genome-wide LD analyses and facilitate detailed evolutionary studies. To this end, a human high-density single-nucleotide polymorphism (SNP) marker map has been constructed. Many of the SNPs on this map, however, were identified by sampling a small number of chromosomes from a single population, and inferences drawn from studies using such SNPs may be influenced by ascertainment bias (AB). Through extensive simulations, we have found that AB is a potentially significant problem in estimating and comparing LD within and between populations. Specifically, the magnitude of AB is a function of the SNP discovery strategy, number of chromosomes used for SNP discovery, population genetic characteristics of the particular genomic region considered, amount of gene flow between populations, and demographic history of the populations. We demonstrate that a balanced SNP discovery strategy (where equal numbers of chromosomes are sampled from multiple subpopulations) is the optimal study design for generating broadly applicable SNP resources. Finally, we validate our theoretical predictions by comparing our results to publicly available data from ten genes sequenced in 24 African American and 23 European American individuals.     

Meta-analysis,funnel plots and sensitivity analysis

Publication bias is a major problem, perhaps the major problem, in meta-analysis (or systematic reviews). Small studies are more likely to be published if their results are 'significant' than if their results are negative or inconclusive, and so the studies available for review are biased in favour of those with positive outcomes. Correcting for this bias is not possible without making untestable assumptions. In this paper, a sensitivity analysis is suggested which is based on fitting a model to the funnel plot. Some examples are discussed.     

Meta-analysis in applied ecology

This overview examines research synthesis in applied ecology and conservation. Vote counting and pooling unweighted averages are widespread despite the superiority of syntheses based on weighted combination of effects. Such analyses allow exploration of methodological uncertainty in addition to consistency of effects across species, space and time, but exploring heterogeneity remains controversial. Meta-analyses are required to generalize in ecology, and to inform evidence-based decision-making, but the more sophisticated statistical techniques and registers of research used in other disciplines must be employed in ecology to fully realize their benefits.     

What are we missing with only ground‐level mist nets? Using elevated nets at a migration stopover site

ABSTRACT Mist nets deployed in a standard ground‐level fashion capture birds approximately 0.5–2.6 m above the ground. In habitats where the vegetation extends above this height, standard mist net deployment may inadequately sample the targeted avian community and age‐ and sex‐classes within species. Such sampling biases may raise questions regarding studies based on data from mist‐net captures. To determine if birds were equally likely to be captured by mist nets at different heights, we constructed a series of paired ground‐level and elevated mist nets (hereafter “net rigs”) at a research station in western New York State. Net rigs were operated during 14 migration seasons from 2000 to 2006 (spring and fall each year), and 19,735 birds of 118 species were captured. Capture rates were significantly higher in ground‐level nets, but 12 species were only captured in elevated nets. Of 44 species with at least 50 captures, 25 species were more likely to be captured in the ground‐level nets and two species in the elevated nets. For four of 18 species, more birds were captured in the elevated nets during fall migration than during spring migration. We conclude that standard ground‐level net placement was more efficient in capturing birds in the secondary growth habitats that we sampled. However, ground‐level nets may not adequately sample the entire targeted community or all age‐ or sex‐classes within species.     

SERT and uncertainty: serotonin transporter expression influences information processing biases for ambiguous aversive cues in mice

The long allele variant of the serotonin transporter (SERT, 5‐HTT) gene‐linked polymorphic region (5‐HTTLPR) is associated with higher levels of 5‐HTT expression and reduced risk of developing affective disorders. However, little is known about the mechanisms underlying this protective effect. One hypothesis is that 5‐HTT expression influences aversive information processing, with reduced negative cognitive bias present in those with higher 5‐HTT expression. Here we investigated this hypothesis using genetically‐modified mice and a novel aversive learning paradigm. Mice with high levels of 5‐HTT expression (5‐HTT over‐expressing, 5‐HTTOE mice) and wild‐type mice were trained to discriminate between three distinct auditory cues: one cue predicted footshock on all trials (CS+); a second cue predicted the absence of footshock (CS?); and a third cue predicted footshock on 20% of trials (CS20%), and was therefore ambiguous. Wild‐type mice exhibited equivalently high levels of fear to the CS+ and CS20% and minimal fear to the CS?. In contrast, 5‐HTTOE mice exhibited high levels of fear to the CS+ but minimal fear to the CS? and the CS20%. This selective reduction in fear to ambiguous aversive cues suggests that increased 5‐HTT expression reduces negative cognitive bias for stimuli with uncertain outcomes.     

Kinship and similarity in residency status structure female social networks in black‐and‐white colobus monkeys (colobus vellerosus)

Kinship shapes female social networks in many primate populations in which females remain in their natal group to breed. In contrast, it is unclear to which extent kinship affects the social networks in populations with female dispersal. Female Colobus vellerosus show routine facultative dispersal (i.e., some females remain philopatric and others disperse). This dispersal pattern allowed us to evaluate if facultative dispersed females form social networks shaped by an attraction to kin, to social partners with a high resource holding potential, or to similar social partners in terms of maturational stage, dominance rank, and residency status. During 2008 and 2009, we collected behavioral data via focal and ad libitum sampling of 61 females residing in eight groups at Boabeng‐Fiema, Ghana. We determined kinship based on partial pedigrees and genotypes at 17 short tandem repeat loci. Kinship influenced coalition and affiliation networks in three groups consisting of long‐term resident females with access to a relatively high number of female kin. In contrast, similar residency status was more important than kinship in structuring the affiliation network in one of two groups that contained recent female immigrants. In populations with female dispersal, the occurrence of kin structured social networks may not only depend on the kin composition of groups but also on how long the female kin have resided together. We found no consistent support for females biasing affiliation toward partners with high resource holding potential, possibly due to low levels of contest competition and small inter‐individual differences in resource holding potential. Am J Phys Anthropol 153:365–376, 2014. © 2013 Wiley Periodicals, Inc.