Imputation of missing values is superior to complete case analysis and the missing-indicator method in multivariable diagnostic research: A clinical example

Heijden, Geert J. M. G. van der; Donders, A. Rogier T.; Stijnen, Theo; Moons, Karel G.M.

doi:10.1016/j.jclinepi.2006.01.015

Cited by 507 publications

(351 citation statements)

References 33 publications

Supporting

Mentioning

333

Contrasting

Unclassified

Order By: Relevance

“…Multiple imputations are superior to case-wise deletion and missing-indicator analysis (64). Missing values in a w j were multiply imputed using the R package Amelia II (version 1.2-14), with the following specified options: logistic transformation of rh (on a 0-to-1 scale); square root transformatioin of r; a third-order polynomial over time; and a 1% ridge prior (assists with numerical stability by shrinking the covariances toward zero without affecting means or variances).…”

Section: Methodsmentioning

confidence: 99%

Predicting Fusarium Head Blight Epidemics With Weather-Driven Pre- and Post-Anthesis Logistic Regression Models

Shah¹,

Molineros²,

Paul³

et al. 2013

Phytopathology®

View full text Add to dashboard Cite

Shah, D. A., Molineros, J. E., Paul, P. A., Willyerd, K. T., Madden, L. V., and De Wolf, E. D. 2013. Predicting Fusarium head blight epidemics with weather-driven pre-and post-anthesis logistic regression models. Phytopathology 103:906-919.Our objective was to identify weather-based variables in pre-and postanthesis time windows for predicting major Fusarium head blight (FHB) epidemics (defined as FHB severity  10%) in the United States. A binary indicator of major epidemics for 527 unique observations (31% of which were major epidemics) was linked to 380 predictor variables summarizing temperature, relative humidity, and rainfall in 5-, 7-, 10-, 14-, or 15-daylong windows either pre-or post-anthesis. Logistic regression models were built with a training data set (70% of the 527 observations) using the leaps-and-bounds algorithm, coupled with bootstrap variable and model selection methods. Misclassification rates were estimated on the training and remaining (test) data. The predictive performance of models with indicator variables for cultivar resistance, wheat type (spring or winter), and corn residue presence was improved by adding up to four weatherbased predictors. Because weather variables were intercorrelated, no single model or subset of predictor variables was best based on accuracy, model fit, and complexity. Weather-based predictors in the 15 final empirical models selected were all derivatives of relative humidity or temperature, except for one rainfall-based predictor, suggesting that relative humidity was better at characterizing moisture effects on FHB than other variables. The average test misclassification rate of the final models was 19% lower than that of models currently used in a national FHB prediction system.Additional keywords: additive logistic regression, data mining, multiple imputation.In the United States, Fusarium head blight (FHB) of wheat (Triticum aestivum L. em. Thell) is caused primarily by Fusarium graminearum sensu stricto of the F. graminearum species complex (44). Major FHB epidemics have occurred somewhere in the United States in every decade since the disease was formally described by W. G. Smith in 1884 (60) although, in any given location, epidemics tend to occur sporadically. During the last two decades, U.S. wheat experienced large direct production losses because of FHB (35,36) and even larger indirect losses in other sectors of the economy (43), contributing to the characterization of FHB as a reemerging disease of importance (36,53). Increased corn (Zea mays) production in wheat-growing regions, concurrent with wider adoption of reduced tillage for soil conservation, were likely contributory factors to severe epidemics beginning in the latter part of the 19th century (36,60), as pathogen survival in corn residue is an acknowledged FHB risk factor (13,27). FHB epidemiological research includes (i) basic documentation of epidemics and observed weather conditions at the time, a mainly descriptive effort, followed by quantification of optimal (usually controlle...

show abstract

Section: Methodsmentioning

confidence: 99%

Predicting Fusarium Head Blight Epidemics With Weather-Driven Pre- and Post-Anthesis Logistic Regression Models

Shah¹,

Molineros²,

Paul³

et al. 2013

Phytopathology®

View full text Add to dashboard Cite

show abstract

“…CCA makes the strict assumption that data are MCAR whereas MI requires the more liberal assumption that missingness is at random, conditional on observed data. Emerging research suggests the superiority of MI [17,18]. Beyond its familiarity to authors and readers, there is little basis for preferring a technique that requires more restrictive assumptions over one with more relaxed assumptions.…”

Section: Secondary and Sensitivity Analysesmentioning

confidence: 99%

The use and reporting of multiple imputation in medical research – a review

Mackinnon

2010

Journal of Internal Medicine

176

144

View full text Add to dashboard Cite

show abstract

“…This approach improves the effi ciency of the fi nal regression through incorporating additional patients and reduces the bias in the regression coeffi cients resulting from exclusion of patients with missing data. [28][29][30][31][32] Further details of the imputation procedure are presented in the e-Appendix 1.…”

Section: Handling Of Missing Datamentioning

confidence: 99%

Predictors of Time to Death After Terminal Withdrawal of Mechanical Ventilation in the ICU

Cooke

Hotchkin

Engelberg

et al. 2010

Chest

View full text Add to dashboard Cite

A pproximately 20% of all American patients die during or shortly after a stay in the ICU, 1 the majority of whom do so in the context of a decision to forego life-sustaining therapy. [2][3][4] Once families and caregivers decide to withdraw life support, experts in end-of-life communication advocate for clinicians to inform families of what they should expect during their loved one's dying process. 5,6 This communication might include details about the expected myoclonus of dying patients, the potential for agonal respirations after discontinuation of mechanical ventilation, and the timing of death after withdrawal of life support. 6 Despite frequent family requests as well as expert recommendations for discussion of the postwithdrawal course, few data exist to guide clinicians in accurately conveying the anticipated course. [7][8][9] The majority of studies that examine the timing of death after withdrawal focus on patients with severe neurologic injury in the context of organ donation after cardiac death 10 or on whether the use of analgesics and sedatives during the dying process hasten death in patients who are critically ill. 7,11 These studies, among others, 12,13 dedicate little attention to other factors that may infl uence the timing of death after withdrawal of life support, such as age, severity of illness, or underlying diagnosis. Characterization of the factors that predict time to death may inform familycaregiver communication at the end of life and alleviate some of the anxiety and frustration resulting

show abstract

Imputation of missing values is superior to complete case analysis and the missing-indicator method in multivariable diagnostic research: A clinical example

Cited by 507 publications

References 33 publications

Predicting Fusarium Head Blight Epidemics With Weather-Driven Pre- and Post-Anthesis Logistic Regression Models

Predicting Fusarium Head Blight Epidemics With Weather-Driven Pre- and Post-Anthesis Logistic Regression Models

The use and reporting of multiple imputation in medical research – a review

Predictors of Time to Death After Terminal Withdrawal of Mechanical Ventilation in the ICU

Contact Info

Product

Resources

About