Multiple imputation in veterinary epidemiological studies: a case study and simulation

Abstract

The problem of missing data occurs frequently in veterinary epidemiological studies. Most studies use a complete case (CC) analysis which excludes all observations for which any relevant variable have missing values. Alternative approaches (most notably multiple imputation (MI)) which avoid the exclusion of observations with missing values are now widely available but have been used very little in veterinary epidemiology.This paper uses a case study based on research into dairy producers' attitudes toward mastitis control procedures, combined with two simulation studies to evaluate the use of MI and compare results with a CC analysis. MI analysis of the original data produced results which had relatively minor differences from the CC analysis. However, most of the missing data in the original data set were in the dependent variable and a subsequent simulation study based on the observed missing data pattern and 1000 simulations showed that an MI analysis would not be expected to offer any advantages over a CC analysis in this situation. This was true regardless of the missing data mechanism (MCAR -missing completely at random, MARmissing at random, or NMAR -not missing at random) underlying the missing values. Surprisingly, recent textbooks dealing with MI make little reference to this limitation of MI for dealing with missing values in the dependent variable.An additional simulation study (1000 runs for each of the three missing data mechanisms) compared MI and CC analyses for data in which varying levels (n=7) of missing data were created in predictor variables. This study showed that MI analyses generally produced results that were less biased on average, were more precise (smaller SEs), were more consistent (less variability between simulation runs) and consequently were more likely to produce estimates that were close to the "truth" (results obtained from a data set with no missing values). While 3 the benefit of MI varied with the mechanism used to generate the missing data, MI always performed as well as, or better than, CC analysis.

Little association between birth weight and health of preweaned dairy calves

“…Multiple imputation 25 followed by generalised estimating equations (MI‐GEE analysis) 26 was used for analysis. Data were stored and processed in Access and Excel.…”

Variation in hospital performance for heart failure management in the National Heart Failure Audit for England and Wales

“…No covariate was missing at a rate that exceeded 15%. Simulation studies have suggested that multiple imputation provides equivalent or better coverage and bias than complete case analysis, even when missingness is not at random 12 13. We therefore used multiple imputation rather than a complete case analysis.…”

Farm characteristics and management routines related to cow longevity: a survey among Swedish dairy farmers

“…However, instead of calling the veterinarian when discovering a sick cow, this data set showed a negative association between the average longevity of the herd and farmers that most likely “Initiated treatment on one’s own” when discovering an unhealthy cow. As multiple imputation analyses generally produce less biased results than complete case analyses [ 24 ], more weight is put on the results from the multiple imputation model. It is generally recommended to use a large number of predictors in the imputation process, and in this case we used all available data from the questionnaires.…”

“…Multiple imputation is a flexible, simulation-based statistical technique for handling missing data [ 23 , 24 ]. It has been shown that multiple imputation analyses generally produce less biased results than complete case analyses [ 24 ]. Multiple imputation of missing predictor values was carried out using the outcome variable and all predictors for which data were missing.…”