Home About us Contact
|
Home About us Contact | ||
|
|
||
Misclassification
Kinds of Misclassification Terms modified by Misclassification Selected AbstractsMisclassification and the use of register-based indicators for depressionACTA PSYCHIATRICA SCANDINAVICA, Issue 4 2009K. Thielen Objective:, To study the degree to which depression indicators based on register data on hospital and antidepressant treatment suffer from differential misclassification with respect to gender, age and social group. Method:, Data on 7378 persons were obtained by linking a cross-sectional survey of Danish adults aged 40 and 50 years with population-based registers. Misclassification was analysed by comparing survey data to register data on major depression using the method proposed by Rothman and Greenland. Results:, Differential misclassification was found. Adjustment for misclassification reduced women's odds ratios from 2.18 to 1.00 for hospital treatment and from 1.70 to 1.10 for antidepressants. For the lower social group, the corresponding odds ratios increased from 1.18 to 3.52, and from 1.35 to 2.32 respectively, whereas odds ratios with respect to age remained almost unchanged. Conclusion:, Differential misclassification should be considered when register-based information about hospital and antidepressant treatment are used as depression indicators. [source] Overestimation of Left Ventricular Mass and Misclassification of Ventricular Geometry in Heart Failure Patients by Two-Dimensional Echocardiography in Comparison with Three-Dimensional EchocardiographyECHOCARDIOGRAPHY, Issue 3 2010Dmitry Abramov M.D. Background: Accurate assessment of left ventricular hypertrophy (LVH) and ventricular geometry is important, especially in patients with heart failure (HF). The aim of this study was to compare the assessment of ventricular size and geometry by 2D and 3D echocardiography in normotensive controls and among HF patients with a normal and a reduced ejection fraction. Methods: One hundred eleven patients, including 42 normotensive patients without cardiac disease, 41 hypertensive patients with HF and a normal ejection fraction (HFNEF), and 28 patients with HF and a low ejection fraction (HFLEF), underwent 2DE and freehand 3DE. The differences between 2DE and 3DE derived LVM were evaluated by use of a Bland,Altman plot. Differences in classification of geometric types among the cohort between 2DE and 3DE were determined. Results: Two-dimensional echocardiography overestimated ventricular mass compared to 3D echocardiography (3DE) among normal (166 ± 36 vs. 145 ± 20 gm, P = 0.002), HFNEF (258 ± 108 vs. 175 ± 47gm, P < 0.001), and HFLEF (444 ± 136 vs. 259 ± 77 gm, P < 0.001) patients. The overestimation of mass by 2DE increased in patients with larger ventricular size. The use of 3DE to assess ventricular geometry resulted in reclassification of ventricular geometric patterns in 76% of patients with HFNEF and in 21% of patients with HFLEF. Conclusion: 2DE overestimates ventricular mass when compared to 3DE among patients with heart failure with both normal and low ejection fractions and leads to significant misclassification of ventricular geometry in many heart failure patients. (Echocardiography 2010;27:223-229) [source] Identification and Estimation of Regression Models with MisclassificationECONOMETRICA, Issue 3 2006Aprajit Mahajan This paper studies the problem of identification and estimation in nonparametric regression models with a misclassified binary regressor where the measurement error may be correlated with the regressors. We show that the regression function is nonparametrically identified in the presence of an additional random variable that is correlated with the unobserved true underlying variable but unrelated to the measurement error. Identification for semiparametric and parametric regression functions follows straightforwardly from the basic identification result. We propose a kernel estimator based on the identification strategy, derive its large sample properties, and discuss alternative estimation procedures. We also propose a test for misclassification in the model based on an exclusion restriction that is straightforward to implement. [source] Evaluation of Three Algorithms to Identify Incident Breast Cancer in Medicare Claims DataHEALTH SERVICES RESEARCH, Issue 5 2007Heather T. Gold Objective. To test the validity of three published algorithms designed to identify incident breast cancer cases using recent inpatient, outpatient, and physician insurance claims data. Data. The Surveillance, Epidemiology, and End Results (SEER) registry data linked with Medicare physician, hospital, and outpatient claims data for breast cancer cases diagnosed from 1995 to 1998 and a 5 percent control sample of Medicare beneficiaries in SEER areas. Study Design. We evaluate the sensitivity and specificity of three algorithms applied to new data compared with original reported results. Algorithms use health insurance diagnosis and procedure claims codes to classify breast cancer cases, with SEER as the reference standard. We compare algorithms by age, stage, race, and SEER region, and explore via logistic regression whether adding demographic variables improves algorithm performance. Principal Findings. The sensitivity of two of three algorithms is significantly lower when applied to newer data, compared with sensitivity calculated during algorithm development (59 and 77.4 percent versus 90 and 80.2 percent, p<.00001). Sensitivity decreases as age increases, and false negative rates are higher for cases with in situ, metastatic, and unknown stage disease compared with localized or regional breast cancer. Substantial variation also exists by SEER registry. There was potential for improvement in algorithm performance when adding age, region, and race to an indicator variable for whether the algorithm determined a subject to be a breast cancer case (p<.00001). Conclusions. Differential sensitivity of the algorithms by SEER region and age likely reflects variation in practice patterns, because the algorithms rely on administrative procedure codes. Depending on the algorithm, 3,5 percent of subjects overall are misclassified in 1998. Misclassification disproportionately affects older women and those diagnosed with in situ, metastatic, or unknown-stage disease. Algorithms should be applied cautiously to insurance claims databases to assess health care utilization outside SEER-Medicare populations because of uneven misclassification of subgroups that may be understudied already. [source] The effect of misclassification on the estimation of association: a reviewINTERNATIONAL JOURNAL OF METHODS IN PSYCHIATRIC RESEARCH, Issue 2 2005Michael Höfler Abstract Misclassification, the erroneous measurement of one or several categorical variables, is a major concern in many scientific fields and particularly in psychiatric research. Even in rather simple scenarios, unless the misclassification probabilities are very small, a major bias can arise in estimating the degree of association assessed with common measures like the risk ratio and the odds ratio. Only in very special cases , for example, if misclassification takes place solely in one of two binary variables and is independent of the other variable (,non-differential misclassification') , is it guaranteed that the estimates are biased towards the null value (which is 1 for the risk ratio and the odds ratio). Furthermore, misclassification, if ignored, usually leads to confidence intervals that are too narrow. This paper reviews consequences of misclassification. A numerical example demonstrates the problem's magnitude for the estimation of the risk ratio in the easy case where misclassification takes place in the exposure variable, but not in the outcome. Moreover, uncertainty about misclassification can broaden the confidence intervals dramatically. The best way to overcome misclassification is to avoid it by design, but some statistical methods are useful for reducing bias if misclassification cannot be avoided. Copyright © 2005 Whurr Publishers Ltd. [source] The Impact of Performance Level Misclassification on the Accuracy and Precision of Percent at Performance Level MeasuresJOURNAL OF EDUCATIONAL MEASUREMENT, Issue 2 2008Damian W. Betebenner No Child Left Behind (NCLB) performance mandates, embedded within state accountability systems, focus school AYP (adequate yearly progress) compliance squarely on the percentage of students at or above proficient. The singular importance of this quantity for decision-making purposes has initiated extensive research into percent proficient as a measure of school quality. In particular, technical discussions have scrutinized the impact of sampling, measurement, and other sources of error on percent proficient statistics. In this article, we challenge the received orthodoxy that measurement error associated with individual students' scores is inconsequential for aggregate percent proficient statistics. Synthesizing current classification accuracy research with techniques from randomized response designs, we establish results which specify the extent to which measurement error,manifest as performance level misclassifications,produces bias and increases error variability for percent at performance level statistics. The results have direct relevance for the design of coherent and fair accountability systems based upon assessment outcomes. [source] The role of fat mass index in determining obesityAMERICAN JOURNAL OF HUMAN BIOLOGY, Issue 5 2010Gerson Peltz Objectives: The objective of this study is to compare body mass index (BMI), percent body fat (PBF), and fat mass index (FMI) and to investigate the accuracy of FMI as a convenient tool for assessing obesity. Design: Anthropometric measurements and bioelectrical impedance analyses were performed on 538 Mexican Americans (373 women and 165 men). Correlations between BMI and PBF and between FMI and PBF were investigated. The percentage of persons misclassified as obese using different classifications was calculated. Multiple linear regression analysis was performed to generate predictive models of FMI for males and females separately. Results: BMI and PBF were correlated in men (, = 0.877; P < 0.0001) and women (, = 0.966; P < 0.0001); however, 20 and 67.2% of the men and 9.2 and 84.2% of women, classified as normal weight and overweight by BMI, respectively, were diagnosed as obese by PBF. FMI and PBF were also correlated in men (, = 0.975; P < 0.0001) and women (, = 0.992; P < 0.0001). Four percent of the men classified as normal weight and 65.5% classified as overweight by BMI were obese by FMI, while 71.3% of women classified as overweight by BMI were obese by FMI. Misclassification of obesity between FMI and PBF categories was observed in 5.4% of men and 7.8% of women. Conclusions: The discrepancy observed between BMI and PBF reflects a limitation of BMI. Conversely, FMI accurately assessed obesity in our study of Mexican Americans, but further studies are necessary to confirm our findings in different ethnic groups. Am. J. Hum. Biol. 22:639,947, 2010. © 2010 Wiley-Liss, Inc. [source] Haplotype Misclassification Resulting from Statistical Reconstruction and Genotype Error, and Its Impact on Association EstimatesANNALS OF HUMAN GENETICS, Issue 5 2010Claudia Lamina Summary Haplotypes are an important concept for genetic association studies, but involve uncertainty due to statistical reconstruction from single nucleotide polymorphism (SNP) genotypes and genotype error. We developed a re-sampling approach to quantify haplotype misclassification probabilities and implemented the MC-SIMEX approach to tackle this as a 3 × 3 misclassification problem. Using a previously published approach as a benchmark for comparison, we evaluated the performance of our approach by simulations and exemplified it on real data from 15 SNPs of the APM1 gene. Misclassification due to reconstruction error was small for most, but notable for some, especially rarer haplotypes. Genotype error added misclassification to all haplotypes resulting in a non-negligible drop in sensitivity. In our real data example, the bias of association estimates due to reconstruction error alone reached ,48.2% for a 1% genotype error, indicating that haplotype misclassification should not be ignored if high genotype error can be expected. Our 3 × 3 misclassification view of haplotype error adds a novel perspective to currently used methods based on genotype intensities and expected number of haplotype copies. Our findings give a sense of the impact of haplotype error under realistic scenarios and underscore the importance of high-quality genotyping, in which case the bias in haplotype association estimates is negligible. [source] The Measurement of the QT and QTc on the Neonatal and Infant Electrocardiogram: A Comprehensive Reliability AssessmentANNALS OF NONINVASIVE ELECTROCARDIOLOGY, Issue 2 2009B.S., Robert M. Gow M.B. Background: An electrocardiogram has been proposed to screen for prolonged QT interval that may predispose infants to sudden death in the first year of life. Understanding the reliability of QT interval measurement will inform the design of a screening program. Methods: Three pediatric cardiologists measured the QT/RR intervals on 60 infant electrocardiograms (median age 46 days), from leads II, V5 and V6 on three separate occasions, 7 days apart, according to a standard protocol. The QTc was corrected by Bazett's (QTcB), Fridericia's (QTCFrid), and Hodges' (QTcH) formulae. Intraobserver and interobserver reliability were assessed by intraclass correlation coefficients (ICC), limits of agreement and repeatability coefficients for single, average of two and average of three measures. Agreement for QTc prolongation (> 440 msec) was assessed by kappa coefficients. Results: QT interval intraobserver ICC was 0.86 and repeatability coefficient was 25.9 msec; interobserver ICC increased from 0.88 for single observations to 0.94 for the average of 3 measurements and repeatability coefficients decreased from 22.5 to 16.7 msec. For QTcB, intraobserver ICC was 0.67, and repeatability was 39.6 msec. Best interobserver reliability for QTcB was for the average of three measurements (ICC 0.83, reproducibility coefficient 25.8 msec), with further improvement for QTcH (ICC 0.92, reproducibility coefficient 16.69 msec). Maximum interobserver kappa for prolonged QTc was 0.77. Misclassification around specific cut points occurs because of the repeatability coefficients. Conclusions: Uncorrected QT measures are more reliable than QTcB and QTCFrid. An average of three independent measures provides the most reliable QT and QTc measurements, with QTcH better than QTcB. [source] Power frequency magnetic fields and risk of childhood leukaemia: Misclassification of exposure from the use of the ,distance from power line' exposure surrogateBIOELECTROMAGNETICS, Issue 3 2009Myron Maslanyj Abstract A recent study examining the relationship between distance to nearby power lines and childhood cancer risk re-opened the debate about which exposure metrics are appropriate for power frequency magnetic field investigations. Using data from two large population-based UK and German studies we demonstrate that distance to power lines is a comparatively poor predictor of measured residential magnetic fields. Even at proximities of 50 m or less, the positive predictive value of having a household measurement over 0.2 µT was only 19.4%. Clearly using distance from power lines, without taking account of other variables such as load, results in a poor proxy of residential magnetic field exposure. We conclude that such high levels of exposure misclassification render the findings from studies that rely on distance alone uninterpretable. Bioelectromagnetics 30:183,188, 2009. © 2008 Wiley-Liss, Inc. [source] Misclassification in Logistic Regression with Discrete CovariatesBIOMETRICAL JOURNAL, Issue 5 2003Ori Davidov Abstract We study the effect of misclassification of a binary covariate on the parameters of a logistic regression model. In particular we consider 2 × 2 × 2 tables. We assume that a binary covariate is subject to misclassification that may depend on the observed outcome. This type of misclassification is known as (outcome dependent) differential misclassification. We examine the resulting asymptotic bias on the parameters of the model and derive formulas for the biases and their approximations as a function of the odds and misclassification probabilities. Conditions for unbiased estimation are also discussed. The implications are illustrated numerically using a case control study. For completeness we briefly examine the effect of covariate dependent misclassification of exposures and of outcomes. [source] Misclassification and the use of register-based indicators for depressionACTA PSYCHIATRICA SCANDINAVICA, Issue 4 2009K. Thielen Objective:, To study the degree to which depression indicators based on register data on hospital and antidepressant treatment suffer from differential misclassification with respect to gender, age and social group. Method:, Data on 7378 persons were obtained by linking a cross-sectional survey of Danish adults aged 40 and 50 years with population-based registers. Misclassification was analysed by comparing survey data to register data on major depression using the method proposed by Rothman and Greenland. Results:, Differential misclassification was found. Adjustment for misclassification reduced women's odds ratios from 2.18 to 1.00 for hospital treatment and from 1.70 to 1.10 for antidepressants. For the lower social group, the corresponding odds ratios increased from 1.18 to 3.52, and from 1.35 to 2.32 respectively, whereas odds ratios with respect to age remained almost unchanged. Conclusion:, Differential misclassification should be considered when register-based information about hospital and antidepressant treatment are used as depression indicators. [source] Differences between European birthweight standards: impact on classification of ,small for gestational age'DEVELOPMENTAL MEDICINE & CHILD NEUROLOGY, Issue 11 2006K Hemming PhD We describe a quantitative and comparative review of a selection of European birthweight standards for gestational age for singletons, to enable appropriate choices to be made for clinical and research use. Differences between median values at term across standards in 10 regions and misclassification of 'small for gestational age'(SGA), were studied. Sex and parity differences, exclusion criteria, and methods of construction were considered. There was wide variation between countries in exclusion criteria, methods of calculating standards, and median birthweight at term. The lightest standards (e.g. France's medians are 255g lower than Norway's medians) were associated with fewer exclusion criteria. Up to 20% of the population used in the construction of the Scottish standard would be classified as SGA using the Norwegian standard. Substantial misclassification of SGA is possible. Assumptions about variation used in the construction of some standards were not justified. It is not possible to conclude that there are real differences in birthweight standards between European countries. Country-based standards control for some population features but add misclassification due to the differing ways in which standards are derived. Standards should be chosen to reflect clinical or research need. If standards stratified by sex or parity are not available, adjustments should be made. In multinational studies, comparisons should be made between results using both a common standard and country-based standards. [source] Incorrect and incomplete coding and classification of diabetes: a systematic reviewDIABETIC MEDICINE, Issue 5 2010M. A. Stone Diabet. Med. 27, 491,497 (2010) Abstract Aims, To conduct a systematic review to identify types and implications of incorrect or incomplete coding or classification within diabetes or between diabetes and other conditions; also to determine the availability of evidence regarding frequency of occurrence. Methods, Medical Subject Headings (MeSH) and free-text terms were used to search relevant electronic databases for papers published to the end of August 2008. Two researchers independently reviewed titles and abstracts and, subsequently, the full text of potential papers. Reference lists of selected papers were also reviewed and authors consulted. Three reviewers independently extracted data. Results, Seventeen eligible studies were identified, including five concerned with distinguishing between Type 1 and Type 2 diabetes. Evidence was also identified regarding: the distinction between diabetes and no-diabetes, failure to specify type of diabetes, and diagnostic errors or difficulties involving maturity-onset diabetes of the young, latent autoimmune diabetes in adults, pancreatic diabetes, persistence of foetal haemoglobin and acquired immune deficiency syndrome (AIDS). The sample was too heterogeneous to derive accurate information about frequency, but our findings suggested that misclassification occurs most commonly in young people. Implications relating to treatment options and risk management were highlighted, in addition to psychological and financial implications and the potential impact on the validity of quality of care evaluations and research. Conclusions, This review draws attention to the occurrence and implications of incorrect or incomplete coding or classification of diabetes, particularly in young people. A pragmatic and clinically relevant approach to classification is needed to assist those involved in making decisions about types of diabetes. [source] Overestimation of Left Ventricular Mass and Misclassification of Ventricular Geometry in Heart Failure Patients by Two-Dimensional Echocardiography in Comparison with Three-Dimensional EchocardiographyECHOCARDIOGRAPHY, Issue 3 2010Dmitry Abramov M.D. Background: Accurate assessment of left ventricular hypertrophy (LVH) and ventricular geometry is important, especially in patients with heart failure (HF). The aim of this study was to compare the assessment of ventricular size and geometry by 2D and 3D echocardiography in normotensive controls and among HF patients with a normal and a reduced ejection fraction. Methods: One hundred eleven patients, including 42 normotensive patients without cardiac disease, 41 hypertensive patients with HF and a normal ejection fraction (HFNEF), and 28 patients with HF and a low ejection fraction (HFLEF), underwent 2DE and freehand 3DE. The differences between 2DE and 3DE derived LVM were evaluated by use of a Bland,Altman plot. Differences in classification of geometric types among the cohort between 2DE and 3DE were determined. Results: Two-dimensional echocardiography overestimated ventricular mass compared to 3D echocardiography (3DE) among normal (166 ± 36 vs. 145 ± 20 gm, P = 0.002), HFNEF (258 ± 108 vs. 175 ± 47gm, P < 0.001), and HFLEF (444 ± 136 vs. 259 ± 77 gm, P < 0.001) patients. The overestimation of mass by 2DE increased in patients with larger ventricular size. The use of 3DE to assess ventricular geometry resulted in reclassification of ventricular geometric patterns in 76% of patients with HFNEF and in 21% of patients with HFLEF. Conclusion: 2DE overestimates ventricular mass when compared to 3DE among patients with heart failure with both normal and low ejection fractions and leads to significant misclassification of ventricular geometry in many heart failure patients. (Echocardiography 2010;27:223-229) [source] Identification and Estimation of Regression Models with MisclassificationECONOMETRICA, Issue 3 2006Aprajit Mahajan This paper studies the problem of identification and estimation in nonparametric regression models with a misclassified binary regressor where the measurement error may be correlated with the regressors. We show that the regression function is nonparametrically identified in the presence of an additional random variable that is correlated with the unobserved true underlying variable but unrelated to the measurement error. Identification for semiparametric and parametric regression functions follows straightforwardly from the basic identification result. We propose a kernel estimator based on the identification strategy, derive its large sample properties, and discuss alternative estimation procedures. We also propose a test for misclassification in the model based on an exclusion restriction that is straightforward to implement. [source] Eukaryotic diversity and phylogeny using small- and large-subunit ribosomal RNA genes from environmental samplesENVIRONMENTAL MICROBIOLOGY, Issue 12 2009William Marande Summary The recent introduction of molecular techniques in eukaryotic microbial diversity studies, in particular those based in the amplification and sequencing of small-subunit ribosomal DNA (SSU rDNA), has revealed the existence of an unexpected variety of new phylotypes. The taxonomic ascription of the organisms bearing those sequences is generally deduced from phylogenetic analysis. Unfortunately, the SSU rDNA sequence alone has often not enough phylogenetic information to resolve the phylogeny of fast-evolving or very divergent sequences, leading to their misclassification. To address this problem, we tried to increase the phylogenetic signal by amplifying the complete eukaryotic rDNA cluster [i.e. the SSU rDNA, the internal transcribed spacers, the 5.8S rDNA and the large-subunit (LSU) rDNA] from environmental samples, and sequencing the SSU and LSU rDNA part of the clones. Using marine planktonic samples, we showed that surveys based on either SSU or SSU + LSU rDNA retrieved comparable diversity patterns. In addition, phylogenetic trees based on the concatenated SSU + LSU rDNA sequences showed better resolution, yielding good support for major eukaryotic groups such as the Opisthokonta, Rhizaria and Excavata. Finally, highly divergent SSU rDNA sequences, whose phylogenetic position was impossible to determine with the SSU rDNA data alone, could be placed correctly with the SSU + LSU rDNA approach. These results suggest that this method can be useful, in particular for the analysis of eukaryotic microbial communities rich in phylotypes of difficult phylogenetic ascription. [source] Inducing safer oblique trees without costsEXPERT SYSTEMS, Issue 4 2005Sunil Vadera Abstract: Decision tree induction has been widely studied and applied. In safety applications, such as determining whether a chemical process is safe or whether a person has a medical condition, the cost of misclassification in one of the classes is significantly higher than in the other class. Several authors have tackled this problem by developing cost-sensitive decision tree learning algorithms or have suggested ways of changing the distribution of training examples to bias the decision tree learning process so as to take account of costs. A prerequisite for applying such algorithms is the availability of costs of misclassification. Although this may be possible for some applications, obtaining reasonable estimates of costs of misclassification is not easy in the area of safety. This paper presents a new algorithm for applications where the cost of misclassifications cannot be quantified, although the cost of misclassification in one class is known to be significantly higher than in another class. The algorithm utilizes linear discriminant analysis to identify oblique relationships between continuous attributes and then carries out an appropriate modification to ensure that the resulting tree errs on the side of safety. The algorithm is evaluated with respect to one of the best known cost-sensitive algorithms (ICET), a well-known oblique decision tree algorithm (OC1) and an algorithm that utilizes robust linear programming. [source] Effect of including environmental data in investigations of gene-disease associations in the presence of qualitative interactionsGENETIC EPIDEMIOLOGY, Issue 6 2010Elizabeth Williamson Abstract Complex diseases are likely to be caused by the interplay of genetic and environmental factors. Despite this, gene-disease associations are frequently investigated using models that focus solely on a marginal gene effect, ignoring environmental factors entirely. Failing to take into account a gene-environment interaction can weaken the apparent gene-disease association, leading to loss in statistical power and, potentially, inability to identify genuine risk factors. If a gene-environment interaction exists, therefore, a joint analysis allowing the effect of the gene to differ between groups defined by the environmental exposure can have greater statistical power than a marginal gene-disease model. However, environmental data are subject to measurement error. Substantial losses in statistical power for detecting gene-environment interactions can arise from measurement error in the environmental exposure. It is unclear, however, what effect measurement error may have on the power of the joint analysis. We consider the potential benefits, in terms of statistical power, of collecting concurrent environmental data within large cohorts in order to enhance gene detection. We further consider whether these benefits remain in the presence of misclassification in both the gene and the environmental exposure. We find that when an effect of the gene is apparent only in the presence of the environmental exposure, the joint analysis has greater power than a marginal gene-disease analysis. This comparative increase in power remains in the presence of likely levels of misclassification of either the gene or environmental exposure. Genet. Epidemiol. 34:552,560, 2010. © 2010 Wiley-Liss, Inc. [source] On dichotomizing phenotypes in family-based association tests: quantitative phenotypes are not always the optimal choiceGENETIC EPIDEMIOLOGY, Issue 5 2007David Fardo Abstract In family-based association studies, quantitative traits are thought to provide higher statistical power than dichotomous traits. Consequently, it is standard practice to collect quantitative traits and to analyze them as such. However, in many situations, continuous measurements are more difficult to obtain and/or need to be adjusted for other factors/confounding variables which also have to be measured. In such scenarios, it can be advantageous to record and analyze a "simplified/dichotomized" version of the original trait. Under fairly general circumstances, we derive here rules for the dichotomization of quantitative traits that maintain power levels that are comparable to the analysis of the original quantitative trait. Using simulation studies, we show that the proposed rules are robust against phenotypic misclassification, making them an ideal tool for inexpensive phenotyping in large-scale studies. The guidelines are illustrated by an application to an asthma study. Genet. Epidemiol. 2007. © 2007 Wiley-Liss, Inc. [source] Quantifying bias due to allele misclassification in case-control studies of haplotypesGENETIC EPIDEMIOLOGY, Issue 7 2006Usha S. Govindarajulu Abstract Objectives Genotyping errors can induce biases in frequency estimates for haplotypes of single nucleotide polymorphisms (SNPs). Here, we considered the impact of SNP allele misclassification on haplotype odds ratio estimates from case-control studies of unrelated individuals. Methods We calculated bias analytically, using the haplotype counts expected in cases and controls under genotype misclassification. We evaluated the bias due to allele misclassification across a range of haplotype distributions using empirical haplotype frequencies within blocks of limited haplotype diversity. We also considered simple two- and three-locus haplotype distributions to understand the impact of haplotype frequency and number of SNPs on misclassification bias. Results We found that for common haplotypes (>5% frequency), realistic genotyping error rates (0.1,1% chance of miscalling an allele), and moderate relative risks (2,4), the bias was always towards the null and increases in magnitude with increasing error rate, increasing odds ratio. For common haplotypes, bias generally increased with increasing haplotype frequency, while for rare haplotypes, bias generally increased with decreasing frequency. When the chance of miscalling an allele is 0.5%, the median bias in haplotype-specific odds ratios for common haplotypes was generally small (<4% on the log odds ratio scale), but the bias for some individual haplotypes was larger (10,20%). Bias towards the null leads to a loss in power; the relative efficiency using a test statistic based upon misclassified haplotype data compared to a test based on the unobserved true haplotypes ranged from roughly 60% to 80%, and worsened with increasing haplotype frequency. Conclusions The cumulative effect of small allele-calling errors across multiple loci can induce noticeable bias and reduce power in realistic scenarios. This has implications for the design of candidate gene association studies that utilize multi-marker haplotypes. Genet. Epidemiol. 2006. © 2006 Wiley-Liss, Inc. [source] Evaluation of Three Algorithms to Identify Incident Breast Cancer in Medicare Claims DataHEALTH SERVICES RESEARCH, Issue 5 2007Heather T. Gold Objective. To test the validity of three published algorithms designed to identify incident breast cancer cases using recent inpatient, outpatient, and physician insurance claims data. Data. The Surveillance, Epidemiology, and End Results (SEER) registry data linked with Medicare physician, hospital, and outpatient claims data for breast cancer cases diagnosed from 1995 to 1998 and a 5 percent control sample of Medicare beneficiaries in SEER areas. Study Design. We evaluate the sensitivity and specificity of three algorithms applied to new data compared with original reported results. Algorithms use health insurance diagnosis and procedure claims codes to classify breast cancer cases, with SEER as the reference standard. We compare algorithms by age, stage, race, and SEER region, and explore via logistic regression whether adding demographic variables improves algorithm performance. Principal Findings. The sensitivity of two of three algorithms is significantly lower when applied to newer data, compared with sensitivity calculated during algorithm development (59 and 77.4 percent versus 90 and 80.2 percent, p<.00001). Sensitivity decreases as age increases, and false negative rates are higher for cases with in situ, metastatic, and unknown stage disease compared with localized or regional breast cancer. Substantial variation also exists by SEER registry. There was potential for improvement in algorithm performance when adding age, region, and race to an indicator variable for whether the algorithm determined a subject to be a breast cancer case (p<.00001). Conclusions. Differential sensitivity of the algorithms by SEER region and age likely reflects variation in practice patterns, because the algorithms rely on administrative procedure codes. Depending on the algorithm, 3,5 percent of subjects overall are misclassified in 1998. Misclassification disproportionately affects older women and those diagnosed with in situ, metastatic, or unknown-stage disease. Algorithms should be applied cautiously to insurance claims databases to assess health care utilization outside SEER-Medicare populations because of uneven misclassification of subgroups that may be understudied already. [source] Comparison of surrogate and direct measurement of insulin resistance in chronic hepatitis C virus infection: Impact of obesity and ethnicity,HEPATOLOGY, Issue 1 2010Khoa D. Lam Studies using surrogate estimates show high prevalence of insulin resistance in hepatitis C infection. This study prospectively evaluated the correlation between surrogate and directly measured estimates of insulin resistance and the impact of obesity and ethnicity on this relationship. Eighty-six nondiabetic, noncirrhotic patients with hepatitis C virus (age = 48 ± 7 years, 74% male, 44% white, 22% African American, 26% Latino, 70% genotype 1) were categorized into normal-weight (body mass index [BMI] < 25, n = 30), overweight (BMI = 25-29.9, n = 38), and obese (BMI , 30, n = 18). Insulin-mediated glucose uptake was measured by steady-state plasma glucose (SSPG) concentration during a 240-minute insulin suppression test. Surrogate estimates included: fasting glucose and insulin, glucose/insulin, homeostasis model assessment (HOMA-IR), quantitative insulin sensitivity check index (QUICKI), insulin (I-AUC) and glucose (G-AUC) area under the curve during oral glucose tolerance test, and the Belfiore and Stumvoll indexes. All surrogate estimates correlated with SSPG, but the magnitude of correlation varied (r = 0.30-0.64). The correlation coefficients were highest in the obese. I-AUC had the highest correlation among all ethnic and weight groups (r = 0.57-0.77). HOMA-IR accounted for only 15% of variability in SSPG in the normal weight group. The common HOMA-IR cutoff of ,3 to define insulin resistance had high misclassification rates especially in the overweight group independent of ethnicity. HOMA-IR > 4 had the lowest misclassification rate (75% sensitivity, 88% specificity). Repeat HOMA-IR measurements had higher within-person variation in the obese (standard deviation = 0.77 higher than normal-weight, 95% confidence interval = 0.25-1.30, P = 0.005). Conclusion: Because of limitations of surrogate estimates, caution should be used in interpreting data evaluating insulin resistance especially in nonobese, nondiabetic patients with HCV. HEPATOLOGY 2010 [source] Modelling small-business credit scoring by using logistic regression, neural networks and decision treesINTELLIGENT SYSTEMS IN ACCOUNTING, FINANCE & MANAGEMENT, Issue 3 2005Mirta Bensic Previous research on credit scoring that used statistical and intelligent methods was mostly focused on commercial and consumer lending. The main purpose of this paper is to extract important features for credit scoring in small-business lending on a dataset with specific transitional economic conditions using a relatively small dataset. To do this, we compare the accuracy of the best models extracted by different methodologies, such as logistic regression, neural networks (NNs), and CART decision trees. Four different NN algorithms are tested, including backpropagation, radial basis function network, probabilistic and learning vector quantization, by using the forward nonlinear variable selection strategy. Although the test of differences in proportion and McNemar's test do not show a statistically significant difference in the models tested, the probabilistic NN model produces the highest hit rate and the lowest type I error. According to the measures of association, the best NN model also shows the highest degree of association with the data, and it yields the lowest total relative cost of misclassification for all scenarios examined. The best model extracts a set of important features for small-business credit scoring for the observed sample, emphasizing credit programme characteristics, as well as entrepreneur's personal and business characteristics as the most important ones. Copyright © 2005 John Wiley & Sons, Ltd. [source] The effect of misclassification on the estimation of association: a reviewINTERNATIONAL JOURNAL OF METHODS IN PSYCHIATRIC RESEARCH, Issue 2 2005Michael Höfler Abstract Misclassification, the erroneous measurement of one or several categorical variables, is a major concern in many scientific fields and particularly in psychiatric research. Even in rather simple scenarios, unless the misclassification probabilities are very small, a major bias can arise in estimating the degree of association assessed with common measures like the risk ratio and the odds ratio. Only in very special cases , for example, if misclassification takes place solely in one of two binary variables and is independent of the other variable (,non-differential misclassification') , is it guaranteed that the estimates are biased towards the null value (which is 1 for the risk ratio and the odds ratio). Furthermore, misclassification, if ignored, usually leads to confidence intervals that are too narrow. This paper reviews consequences of misclassification. A numerical example demonstrates the problem's magnitude for the estimation of the risk ratio in the easy case where misclassification takes place in the exposure variable, but not in the outcome. Moreover, uncertainty about misclassification can broaden the confidence intervals dramatically. The best way to overcome misclassification is to avoid it by design, but some statistical methods are useful for reducing bias if misclassification cannot be avoided. Copyright © 2005 Whurr Publishers Ltd. [source] Agreement Between Nosologist and Cardiovascular Health Study Review of Deaths: Implications of Coding DifferencesJOURNAL OF AMERICAN GERIATRICS SOCIETY, Issue 1 2009Diane G. Ives MPH OBJECTIVES: To compare nosologist coding of underlying cause of death according to the death certificate with adjudicated cause of death for subjects aged 65 and older in the Cardiovascular Health Study (CHS). DESIGN: Observational. SETTING: Four communities: Forsyth County, North Carolina (Wake Forest University); Sacramento County, California (University of California at Davis); Washington County, Maryland (Johns Hopkins University); and Pittsburgh, Pennsylvania (University of Pittsburgh). PARTICIPANTS: Men and women aged 65 and older participating in CHS, a longitudinal study of coronary heart disease and stroke, who died through June 2004. MEASUREMENTS: The CHS centrally adjudicated underlying cause of death for 3,194 fatal events from June 1989 to June 2004 using medical records, death certificates, proxy interviews, and autopsies, and results were compared with underlying cause of death assigned by a trained nosologist based on death certificate only. RESULTS: Comparison of 3,194 CHS versus nosologist underlying cause of death revealed moderate agreement except for cancer (kappa=0.91, 95% confidence interval (CI)=0.89,0.93). kappas varied according to category (coronary heart disease, kappa=0.61, 95% CI=0.58,0.64; stroke, kappa=0.59, 95% CI=0.54,0.64; chronic obstructive pulmonary disease, kappa=0.58, 95% CI=0.51,0.65; dementia, kappa=0.40, 95% CI=0.34,0.45; and pneumonia, kappa=0.35, 95% CI=0.29,0.42). Differences between CHS and nosologist coding of dementia were found especially in older ages in the sex and race categories. CHS attributed 340 (10.6%) deaths due to dementia, whereas nosologist coding attributed only 113 (3.5%) to dementia as the underlying cause. CONCLUSION: Studies that use only death certificates to determine cause of death may result in misclassification and potential bias. Changing trends in cause-specific mortality in older individuals may be a function of classification process rather than incidence and case fatality. [source] A Chart-Based Method for Identification of Delirium: Validation Compared with Interviewer Ratings Using the Confusion Assessment MethodJOURNAL OF AMERICAN GERIATRICS SOCIETY, Issue 2 2005Sharon K. Inouye MD Objectives: To validate a chart-based method for identification of delirium and compare it with direct interviewer assessment using the Confusion Assessment Method (CAM). Design: Prospective validation study. Setting: Teaching hospital. Participants: Nine hundred nineteen older hospitalized patients. Measurements: A chart-based instrument for identification of delirium was created and compared with the reference standard interviewer ratings, which used direct cognitive assessment to complete the CAM for delirium. Trained nurse chart abstractors were blinded to all interview data, including cognitive and CAM ratings. Factors influencing the correct identification of delirium in the chart were examined. Results: Delirium was present in 115 (12.5%) patients according to the CAM. Sensitivity of the chart-based instrument was 74%, specificity was 83%, and likelihood ratio for a positive result was 4.4. Overall agreement between chart and interviewer ratings was 82%, kappa=0.41. By contrast, using International Classification of Diseases, Ninth Revision, Clinical Modification, administrative codes, the sensitivity for delirium was 3%, and specificity was 99%. Independent factors associated with incorrect chart identification of delirium were dementia, severe illness, and high baseline delirium risk. With all three factors present, the chart instrument was three times more likely to identify patients incorrectly than with none of the factors present. Conclusion: A chart-based instrument for delirium, which should be useful for patient safety and quality-improvement programs in older persons, was validated. Because of potential misclassification, the chart-based instrument is not recommended for individual patient care or diagnostic purposes. [source] Using interpubic distance for sexing manakins in the fieldJOURNAL OF FIELD ORNITHOLOGY, Issue 1 2010Chase D. Mendenhall ABSTRACT Field methods for determining the sex of birds are often limited due to morphometric overlap between sexes, intermediate plumages, seasonality, and reliance on subjective age classification. Interpubic distance, characterized in birds as the distance between the distal ends of the pubic bones, has not been formally tested as a method for determining the sex of birds, despite references among parrot breeders and the frequent use of analogous measurements in mammals. We developed a harmless and easily performed field method for measuring interpubic distance in studies involving bird capture, and compared the interpubic distances of known sex White-ruffed Manakins (Corapipo altera), Orange-collared Manakins (Manacus aurantiacus), and Blue-crowned Manakins (Lepidothrix coronata) to evaluate the possible use of this measurement to determine sex. Using interpubic distance ranges based on 85% confidence intervals where overlap existed between sexes, the sex of 92.8,100% of all manakins in our study was accurately determined with no misclassification. Interpubic distance performed better than plumage-based methods that sexed 74.0% of all individuals and misclassified 1.5%. Using linear discriminant analysis, we developed classification equations that allowed us to accurately determine the sex of all individuals with 100% accuracy using mass and interpubic distance. Additionally, we compared the interpubic distances of female White-ruffed Manakins to evaluate the potential to determine age and reproductive status. Despite an apparent relationship between interpubic distance, age and reproductive status, we concluded that interpubic distance has limited use for determining age and reproductive status due to extensive overlap (31.6,100%), but shows potential in other applications. Based on these results, we endorse the use of interpubic distance to determine the sex of manakins. We encourage further study to develop additional classification equations using different morphometric measurements and to test the efficacy of interpubic distance to determine sex in other bird species. RESUMEN Los métodos del campo para distinguir el sexo de aves son limitados a causa de traslapes de medidas mórfometricas extremas entre sexos, plumajes intermedios, diferencias temporales y/o dependencia en clasificación sujeto de la edad. Distancia interpúbica, caracterizada en aves como la distancia entre los puntos distales de los huesos púbicos, no ha sido formalmente probada como un método para distinguir el sexo de las aves, a pesar de referencias por criadores de loros y uso de métodos similares en mamíferos. Diseñamos un método del campo rápido y sencillo que no tiene riesgo del daño para sacar la distancia interpúbica en estudios que capturan aves. Comparamos la distancia interpúbica de individuos de sexo conocido de Corapipo altera, Manacus aurantiacus, y Lepidothrix coronata para probar el método. Clasificamos correctamente el sexo de 92.8,100% de todos los individuos en este estudio por rangos determinados a través de intérvalos de confianza de 85%. La distancia interpúbica funcionó mejor que un método basado en plumaje, el cual que distinguió el sexo correcto de 74.0% pero falló en clasificar el sexo de 1.5% de los individuos estudiados. Usamos el análisis de discriminación linear para determinar el poder predictivo de la distancia interpúbica, longitud del ala y masa e hicimos ecuaciones de clasificación que distinguieron sexo con un 100% de éxito usando solo masa y distancia interpúbica. Además, comparamos distancia interpúbica de las hembras de C. altera para evaluar el potencial de distinguir edad y estadio reproductivo. A pesar de existir una conexión entre distancia interpúbica, edad y estadio reproductivo, concluimos que la distancia interpúbica es limitada en el contexto de distinguir edad y estadio reproductivo por traslape extensivo (31.6,100%), pero muestra potencial en otras aplicaciones. Basados en estos resultados recomendamos el uso de distancia interpúbica como un método para distinguir sexo de pipridos. Recomendamos más investigación para crear otras ecuaciones de clasificación usando medidas mórfometricas diferentes y probar la eficacia de la distancia interpúbica para distinguir el sexo de otras especies de aves. [source] Sexual Dimorphism in America: Geometric Morphometric Analysis of the Craniofacial Region,JOURNAL OF FORENSIC SCIENCES, Issue 1 2008Erin H. Kimmerle Ph.D. Abstract:, One of the four pillars of the anthropological protocol is the estimation of sex. The protocol generally consists of linear metric analysis or visually assessing individual skeletal traits on the skull and pelvis based on an ordinal scale of 1,5, ranging from very masculine to very feminine. The morphologic traits are then some how averaged by the investigator to estimate sex. Some skulls may be misclassified because of apparent morphologic features that appear more or less robust due to size differences among individuals. The question of misclassification may be further exemplified in light of comparisons across populations that may differ not only in cranial robusticity but also in stature and general physique. The purpose of this study is to further examine the effect of size and sex on craniofacial shape among American populations to better understand the allometric foundation of skeletal traits currently used for sex estimation. Three-dimensional coordinates of 16 standard craniofacial landmarks were collected using a Microscribe-3DX digitizer. Data were collected for 118 American White and Black males and females from the W.M. Bass Donated Collection and the Forensic Data Bank. The MANCOVA procedure tested shape differences as a function of sex and size. Sex had a significant influence on shape for both American Whites (F = 2.90; d.f. = 19, 39; p > F = 0.0024) and Blacks (F = 2.81; d.f. = 19, 37; p > F = 0.0035), whereas size did not have a significant influence on shape in either Whites (F = 1.69; d.f. = 19, 39; p > F = 0.08) or Blacks (F = 1.09; d.f. = 19, 37; p > F = 0.40). Therefore, for each sex, individuals of various sizes were statistically the same shape. In other words, while significant differences were present between the size of males and females (males on average were larger), there was no size effect beyond that accounted for by sex differences in size. Moreover, the consistency between American groups is interesting as it suggests that population differences in sexual dimorphism may result more from human variation in size than allometric variation in craniofacial morphology. [source] Epidemiology of irritable bowel syndrome in Asia: Something old, something new, something borrowedJOURNAL OF GASTROENTEROLOGY AND HEPATOLOGY, Issue 10 2009Kok-Ann Gwee Abstract In this review we have unearthed epidemiological data that; support the ,old' concept of irritable bowel syndrome (IBS) as a disorder of civilization, build a ,new' symptom profile of IBS for Asia, and persuade us against the use of ,borrowed' Western diagnostic criteria and illness models by Asian societies. In the 1960s, IBS was described as a disorder of civilization. Early studies from Asia suggested a prevalence of IBS below 5%. Recent studies from Asia suggest a trend for the more affluent city states like Singapore and Tokyo, to have higher prevalence of 8.6% and 9.8%, respectively, while India had the lowest prevalence of 4.2%. Furthermore, there was a trend among the better educated and more affluent strata of society in several urban Chinese populations for a higher prevalence of IBS, as well as a trend for a higher consultation rate. Across Chinese and Indian predominant populations, a majority of patients with IBS criteria report upper abdominal symptoms such as epigastric pain relieved by defecation, bloating and dyspepsia. Bloating and incomplete evacuation appear to be more important determinants of consultation behavior, than psychological factors. The failure of the Rome criteria to recognize the relationship to meals, may have led to a substantial misclassification of IBS as dyspepsia. The relevance of the Western model of psychological disturbance as a determinant of consultation behavior is questionable because of the accessibility and acceptability of medical consultation for gastrointestinal complaints in many Asian communities. [source] | |||||||||||||||||||||||||||||||||||||||||||