Using multiple imputation to deal with missing data and attrition in longitudinal studies with repeated measures of patient-reported outcomes

Abstract

ObjectiveMissing data is a ubiquitous problem in studies using patient-reported measures, decreasing sample sizes and causing possible bias. In longitudinal studies, special problems relate to attrition and death during follow-up. We describe a methodological approach for the use of multiple imputation (MI) to meet these challenges.MethodsIn a cohort of patients treated with percutaneous coronary intervention followed with use of repetitive questionnaires and information from national registers over 3 years, only 417 out of 1,726 patients had complete data on all measure points and covariates. We suggest strategies for use of MI and different methods for dealing with death along with sensitivity analysis of deviations from the assumption of missing at random, all with the use of standard statistical software. The Mental Component Summary from Short Form 12-item survey was used as an example.ConclusionIgnoring missing data may cause bias of unknown size and direction in longitudinal studies. We have illustrated that MI is a feasible method to try to deal with bias due to missing data in longitudinal studies, including attrition and nonresponse, and should be considered in combination with analysis of sensitivity in longitudinal studies. How to handle dropout due to death is still open for debate.

Impact of dietary fat composition on prediabetes: a 12-year follow-up study

“…However, at V2, V3 and V4, 3262 dietary data and 6566 blood glucose data were missing. This is 42 % of the total n, which is beyond the imputations that have been reported thus far (33)(34)(35)(36)(37) . Multiple imputation has been attempted and used in epidemiological studies, especially in nutritional epidemiology with repeated measures of variables in longitudinal design (35,38) .…”

Journal Club: Long-term functional outcome in patients with acquired infections after acute spinal cord injury

“…Attrition, or loss to follow-up, can occur for numerous reasons, and is of particular concern in longitudinal studies. 5 These missing data can introduce bias when the lack of data is related to the outcome measure, and is not missing at random. 5 One strategy to avoid this bias is multiple imputation, in which missing values are predicted through other observed values in the dataset, creating multiple imputed datasets, and averaging these imputed estimates.…”

“…5 These missing data can introduce bias when the lack of data is related to the outcome measure, and is not missing at random. 5 One strategy to avoid this bias is multiple imputation, in which missing values are predicted through other observed values in the dataset, creating multiple imputed datasets, and averaging these imputed estimates. 5 This is a common method of accounting for missing data, with varying outcomes.…”

“…5 One strategy to avoid this bias is multiple imputation, in which missing values are predicted through other observed values in the dataset, creating multiple imputed datasets, and averaging these imputed estimates. 5 This is a common method of accounting for missing data, with varying outcomes. 5,6 Presented in conjunction with the complete case sensitivity analysis, the imputed data allow for a transparent interpretation of the results.…”

“…5 This is a common method of accounting for missing data, with varying outcomes. 5,6 Presented in conjunction with the complete case sensitivity analysis, the imputed data allow for a transparent interpretation of the results. In addition, the authors also adjust for variables known to be related to functional recovery after SCI (e.g., level of injury, AIS).…”