Comparison of simple and multiple imputation methods using a risk model for surgical mortality as example

Revista Brasileira de Epidemiologia

Klück, Mariza Machado¹  Fachel, Jandyra Maria Guimarães¹  Nunes, Luciana Neves¹ 

[1] Universidade Federal do Rio Grande do Su, Porto Alegre, Brasil

关键词: Imputation methods;    Multiple imputation;    Missing data;    Missing at random;   

DOI  :  10.1590/S1415-790X2010000400005

学科分类：过敏症与临床免疫学

来源: SciELO

【 摘 要 】

INTRODUCTION:It is common for studies in health to face problems with missing data. Throughimputation, complete data sets are built artificially and can be analyzed bytraditional statistical analysis. The objective of this paper is to comparethree types of imputation based on real data.
METHODS: The data used came from a study on the development of risk modelsfor surgical mortality. The sample size was 450 patients. The imputation methodsapplied were: two single imputations and one multiple imputation and the assumptionwas MAR (Missing at Random).
RESULTS: The variable with missing data was serum albumin with 27.1%of missing rate. The logistic models adjusted by simple imputation were similar,but differed from models obtained by multiple imputation in relation to theinclusion of variables.
CONCLUSIONS: The results indicate that it is important to take into accountthe relationship of albumin to other variables observed, because different modelswere obtained in single and multiple imputations. Single imputation underestimatesthe variability generating narrower confidence intervals. It is important toconsider the use of imputation methods when there is missing data, especiallymultiple imputation that takes into account the variability between imputationsfor estimates of the model.

【 授权许可】

CC BY