The impact of covariate measurement error on risk prediction

In the development of risk prediction models, predictors are often measured with error. In this paper, we investigate the impact of covariate measurement error on risk prediction. We compare the prediction performance using a costly variable measured without error, along with error‐free covariates, to that of a model based on an inexpensive surrogate along with the error‐free covariates. We consider continuous error‐prone covariates with homoscedastic and heteroscedastic errors, and also a discrete misclassified covariate. Prediction performance is evaluated by the area under the receiver operating characteristic curve (AUC), the Brier score (BS), and the ratio of the observed to the expected number of events (calibration). In an extensive numerical study, we show that (i) the prediction model with the error‐prone covariate is very well calibrated, even when it is mis‐specified; (ii) using the error‐prone covariate instead of the true covariate can reduce the AUC and increase the BS dramatically; (iii) adding an auxiliary variable, which is correlated with the error‐prone covariate but conditionally independent of the outcome given all covariates in the true model, can improve the AUC and BS substantially. We conclude that reducing measurement error in covariates will improve the ensuing risk prediction, unless the association between the error‐free and error‐prone covariates is very high. Finally, we demonstrate how a validation study can be used to assess the effect of mismeasured covariates on risk prediction. These concepts are illustrated in a breast cancer risk prediction model developed in the Nurses' Health Study. Copyright © 2015 John Wiley & Sons, Ltd.     

Meta-analysis for linear and nonlinear dose-response relations: examples, an evaluation of approximations, and software

Two methods for point and interval estimation of relative risk for log-linear exposure-response relations in meta-analyses of published ordinal categorical exposure-response data have been proposed. The authors compared the results of a meta-analysis of published data using each of the 2 methods with the results that would be obtained if the primary data were available and investigated the circumstances under which the approximations required for valid use of each meta-analytic method break down. They then extended the methods to handle nonlinear exposure-response relations. In the present article, methods are illustrated using studies of the relation between alcohol consumption and colorectal and lung cancer risks from the ongoing Pooling Project of Prospective Studies of Diet and Cancer. In these examples, the differences between the results of a meta-analysis of summarized published data and the pooled analysis of the individual original data were small. However, incorrectly assuming no correlation between relative risk estimates for exposure categories from the same study gave biased confidence intervals for the trend and biased P values for the tests for nonlinearity and between-study heterogeneity when there was strong confounding by other model covariates. The authors illustrate the use of 2 publicly available user-friendly programs (Stata and SAS) to implement meta-analysis for dose-response data.     

Antigenic Domains of the Open Reading Frame 2-Encoded Protein of Hepatitis E Virus

The antigenic composition of the hepatitis E virus (HEV) protein encoded by open reading frame 2 (ORF2) was determined by using synthetic peptides. Three sets of overlapping 18-, 25-, and 30-mer peptides, with each set spanning the entire ORF2 protein of the HEV Burma strain, were synthesized. All synthetic peptides were tested by enzyme immunoassay against a panel of 32 anti-HEV-positive serum specimens obtained from acutely HEV-infected persons. Six antigenic domains within the ORF2 protein were identified. Domains 1 and 6 located at the N and C termini of the ORF2 protein, respectively, contain strong immunoglobulin G (IgG) and IgM antigenic epitopes that can be efficiently modeled with peptides of different sizes. In contrast, antigenic epitopes identified within the two central domains (3 and 4) were modeled more efficiently with 30-mer peptides than with either 18- or 25-mers. Domain 2 located at amino acids (aa) 143 to 222 was modeled best with 25-mer peptides. A few 30-mer synthetic peptides derived from domain 5 identified at aa 490 to 579 demonstrated strong IgM antigenic reactivity. Several 30-mer synthetic peptides derived from domains 1, 4, and 6 immunoreacted with IgG or IgM with more than 70% of anti-HEV-positive serum specimens. Thus, the results of this study demonstrate the existence of six diagnostically relevant antigenic domains within the HEV ORF2 protein.     

Robust Hepatitis B Virus Genotyping by Mass Spectrometry

Genotyping of hepatitis B virus (HBV) is important for tracking HBV infections, prognosticating the development of severe liver disease, and predicting outcomes of therapy. Current genotyping methods can be laborious and costly and rely on subjective data interpretation. To identify less expensive but equally reliable alternatives, we compared “gold standard” sequencing to a novel mass spectrometry approach. Sera from individuals with acute or chronic HBV infection (n = 756), representing all genotypes, were used to PCR amplify the HBV S gene. All amplicons were subjected to base-specific cleavage and matrix-assisted laser desorption ionization-time-of-flight mass spectrometry (MALDI-TOF MS). The resulting mass peak patterns were used to identify HBV genotype by automated comparison to peak patterns simulated from reference sets of HBV sequences of known genotypes. The MALDI-TOF MS data and phylogenetic analysis of HBV sequences produced completely concordant results. Several parameters such as genetic relatedness of tested HBV variants to the reference set, chronic infections, and the quality of PCR products can lower the MS score but never affected the accuracy of the genotype call. This new streamlined MS-based method provides for rapid and accurate HBV genotyping, produces automated data reports, and is therefore suitable for routine use in diagnostic settings.Hepatitis B virus (HBV) infection is a major cause of morbidity and mortality across the world, 350 million people are currently estimated to be chronically infected with the virus, and 660,000 deaths have been reported from HBV-related liver disease and cancer. HBV is genetically diverse and has been classified into 8 genotypes and 24 subtypes (4, 25). HBV genotyping is an important molecular epidemiological tool for tracking HBV infections. In clinical practice, the HBV genotype has been associated with severity and progression of chronic hepatitis B, the response to anti-HBV therapy, and the risk of liver cancer (13, 19).Currently, the most accurate approach to HBV genotyping is based on direct dideoxy sequencing of DNA amplified from discrete regions of the HBV genome, followed by phylogenetic analysis of these sequences (10, 16, 29). This approach is considered the “gold standard,” but it is laborious and thus not suitable for use in routine public health or clinical practice. Restriction fragment length polymorphism (RFLP) analysis (11, 27) significantly simplified genotype detection; however, this approach has limited sensitivity. The Trugene method, a standardized and largely automated polymerase gene-sequencing assay, produces only up to 86% of interpretable data (3, 24, 33). Multiplex PCR requires the use of many sets of primers (10, 23), so its efficacy is reduced should the target templates have highly variable sequences. Real-time PCR, although applicable for genotyping, is better suited for viral load quantification (9). The DNA chip (30-32), based on a hybridization technology, efficiently detects HBV genotypes but requires the test sample to carry a viral load greater than 1,000 IU/ml, and its sensitivity is diminished to ∼90% for certain genotypes. It has low throughput and is costly, which limits its wider application (32). The InnoLipa assay, based on reverse hybridization, is one of the more widely used technologies in clinical laboratories (17, 21, 26). It is accurate and capable of discriminating genotypes in mixed-HBV infections; however, it is relatively costly and not open-ended to detect emerging variants.The MassARRAY system based on nucleic acid analysis by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) provides an alternative approach to HBV genotyping. MALDI-TOF MS has the capacity to discriminate single-point mutations. Specific primer extension into the sites of interest has successfully been applied to detect up to 60 single-nucleotide drug resistance mutations in the HBV polymerase gene (16) and to genotype HBV (12, 14). The MassARRAY system of MALDI-TOF MS is capable of detecting wild-type and mutant alleles and can identify mixes if the minority type is present at >10% (2). This technology is less costly and easy to use as it is amenable to automation and has already found application in other disease evaluations, bacterial typing, and cancer marker detection. In this study, we report the development and assessment of a novel approach to HBV genotyping based on the MassARRAY system.     

Hepatitis B virus transmissions associated with a portable dental clinic,West Virginia, 2009

BackgroundAlthough hepatitis B virus (HBV) transmission in dental settings is rare, in 2009 a cluster of acute HBV infections was reported among attendees of a two-day portable dental clinic in West Virginia.MethodsThe authors conducted a retrospective investigation by using treatment records and volunteer logs, interviews of patients and volunteers with acute HBV infection as well as of other clinic volunteers, and molecular sequencing of the virus from those acutely infected.ResultsThe clinic was held under the auspices of a charitable organization in a gymnasium staffed by 750 volunteers, including dental care providers who treated 1,137 adults. Five acute HBV infections—involving three patients and two volunteers—were identified by the local and state health departments. Of four viral isolates available for testing, all were genotype D. Three case patients underwent extractions; one received restorations and one a dental prophylaxis. None shared a treatment provider with any of the others. One case volunteer worked in maintenance; the other directed patients from triage to the treatment waiting area. Case patients reported no behavioral risk factors for HBV infection. The investigation revealed numerous infection control breaches.ConclusionsTransmission of HBV to three patients and two volunteers is likely to have occurred at a portable dental clinic. Specific breaches in infection control could not be linked to these HBV transmissions.Practical ImplicationsAll dental settings should adhere to recommended infection control practices, including oversight; training in prevention of bloodborne pathogens transmission; receipt of HBV vaccination for staff who may come into contact with blood or body fluids; use of appropriate personal protective equipment, sterilization and disinfection procedures; and use of measures, such as high-volume suction, to minimize the spread of blood.     

Variants in ABCB1, TGFB1, and XRCC1 genes and susceptibility to viral hepatitis A infection in Mexican Americans

Hepatitis A vaccination has dramatically reduced the incidence of hepatitis A virus (HAV) infection, but new infections continue to occur. To identify human genetic variants conferring a risk for HAV infection among the three major racial/ethnic populations in the United States, we assessed associations between 67 genetic variants (single nucleotide polymorphisms [SNPs]) among 31 candidate genes and serologic evidence of prior HAV infection using a population-based, cross-sectional study of 6,779 participants, including 2,619 non-Hispanic whites, 2,095 non-Hispanic blacks, and 2,065 Mexican Americans enrolled in phase 2 (1991-1994) of the Third National Health and Nutrition Examination Survey. Among the three racial/ethnic groups, the number (weighted frequency) of seropositivity for antibody to HAV was 958 (24.9%), 802 (39.2%), and 1540 (71.5%), respectively. No significant associations with any of the 67 SNPs were observed among non-Hispanic whites or non-Hispanic blacks. In contrast, among Mexican Americans, variants in two genes were found to be associated with an increased risk of HAV infection: TGFB1 rs1800469 (adjusted odds ratio [OR], 1.38; 95% confidence interval [CI], 1.14-1.68; P value adjusted for false discovery rate [FDR-P] = 0.017) and XRCC1 rs1799782 (OR, 1.57; 95% CI, 1.27-1.94; FDR-P = 0.0007). A decreased risk was found with ABCB1 rs1045642 (OR, 0.79; 95% CI, 0.71-0.89; FDR-P = 0.0007). Conclusion: Genetic variants in ABCB1, TGFB1, and XRCC1 appear to be associated with susceptibility to HAV infection among Mexican Americans. Replication studies involving larger population samples are warranted.