A wide range of missing imputation approaches in longitudinal data: a simulation study and real data analysis.
Longitudinal regression tree
Missing longitudinal data
Multiple imputations
Single imputation
Journal
BMC medical research methodology
ISSN: 1471-2288
Titre abrégé: BMC Med Res Methodol
Pays: England
ID NLM: 100968545
Informations de publication
Date de publication:
06 07 2023
06 07 2023
Historique:
received:
11
01
2023
accepted:
08
06
2023
medline:
10
7
2023
pubmed:
7
7
2023
entrez:
6
7
2023
Statut:
epublish
Résumé
Missing data is a pervasive problem in longitudinal data analysis. Several single-imputation (SI) and multiple-imputation (MI) approaches have been proposed to address this issue. In this study, for the first time, the function of the longitudinal regression tree algorithm as a non-parametric method after imputing missing data using SI and MI was investigated using simulated and real data. Using different simulation scenarios derived from a real data set, we compared the performance of cross, trajectory mean, interpolation, copy-mean, and MI methods (27 approaches) to impute missing longitudinal data using parametric and non-parametric longitudinal models and the performance of the methods was assessed in real data. The real data included 3,645 participants older than 18 years within six waves obtained from the longitudinal Tehran cardiometabolic genetic study (TCGS). The data modeling was conducted using systolic and diastolic blood pressure (SBP/DBP) as the outcome variables and included predictor variables such as age, gender, and BMI. The efficiency of imputation approaches was compared using mean squared error (MSE), root-mean-squared error (RMSE), median absolute deviation (MAD), deviance, and Akaike information criteria (AIC). The longitudinal regression tree algorithm outperformed based on the criteria such as MSE, RMSE, and MAD than the linear mixed-effects model (LMM) for analyzing the TCGS and simulated data using the missing at random (MAR) mechanism. Overall, based on fitting the non-parametric model, the performance of the 27 imputation approaches was nearly similar. However, the SI traj-mean method improved performance compared with other imputation approaches. Both SI and MI approaches performed better using the longitudinal regression tree algorithm compared with the parametric longitudinal models. Based on the results from both the real and simulated data, we recommend that researchers use the traj-mean method for imputing missing values of longitudinal data. Choosing the imputation method with the best performance is widely dependent on the models of interest and the data structure.
Sections du résumé
BACKGROUND
Missing data is a pervasive problem in longitudinal data analysis. Several single-imputation (SI) and multiple-imputation (MI) approaches have been proposed to address this issue. In this study, for the first time, the function of the longitudinal regression tree algorithm as a non-parametric method after imputing missing data using SI and MI was investigated using simulated and real data.
METHOD
Using different simulation scenarios derived from a real data set, we compared the performance of cross, trajectory mean, interpolation, copy-mean, and MI methods (27 approaches) to impute missing longitudinal data using parametric and non-parametric longitudinal models and the performance of the methods was assessed in real data. The real data included 3,645 participants older than 18 years within six waves obtained from the longitudinal Tehran cardiometabolic genetic study (TCGS). The data modeling was conducted using systolic and diastolic blood pressure (SBP/DBP) as the outcome variables and included predictor variables such as age, gender, and BMI. The efficiency of imputation approaches was compared using mean squared error (MSE), root-mean-squared error (RMSE), median absolute deviation (MAD), deviance, and Akaike information criteria (AIC).
RESULTS
The longitudinal regression tree algorithm outperformed based on the criteria such as MSE, RMSE, and MAD than the linear mixed-effects model (LMM) for analyzing the TCGS and simulated data using the missing at random (MAR) mechanism. Overall, based on fitting the non-parametric model, the performance of the 27 imputation approaches was nearly similar. However, the SI traj-mean method improved performance compared with other imputation approaches.
CONCLUSION
Both SI and MI approaches performed better using the longitudinal regression tree algorithm compared with the parametric longitudinal models. Based on the results from both the real and simulated data, we recommend that researchers use the traj-mean method for imputing missing values of longitudinal data. Choosing the imputation method with the best performance is widely dependent on the models of interest and the data structure.
Identifiants
pubmed: 37415114
doi: 10.1186/s12874-023-01968-8
pii: 10.1186/s12874-023-01968-8
pmc: PMC10327316
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
161Informations de copyright
© 2023. The Author(s).
Références
Stat Methods Med Res. 2019 Feb;28(2):555-568
pubmed: 29069967
Biom J. 2020 Mar;62(2):444-466
pubmed: 31919921
Stat Med. 2015 May 20;34(11):1818-33
pubmed: 25656439
Int J Family Med. 2016;2016:2639624
pubmed: 27042349
BMJ. 2009 Jun 29;338:b2393
pubmed: 19564179
BMC Med Inform Decis Mak. 2021 Nov 10;21(1):313
pubmed: 34758828
Comput Math Methods Med. 2021 Nov 09;2021:6401105
pubmed: 34795791
BMC Med Res Methodol. 2017 Jul 25;17(1):114
pubmed: 28743256
J Hum Hypertens. 2019 Nov;33(11):775-785
pubmed: 31551569
Diabetol Metab Syndr. 2021 Feb 18;13(1):20
pubmed: 33602293
BMC Med Res Methodol. 2020 Aug 12;20(1):207
pubmed: 32787781
BMC Med Res Methodol. 2015 Apr 07;15:30
pubmed: 25880850
Psychol Methods. 2016 Jun;21(2):222-40
pubmed: 26690775
Psychol Methods. 2018 Jun;23(2):298-317
pubmed: 28557466
Int J Endocrinol Metab. 2018 Oct 27;16(4 Suppl):e84744
pubmed: 30584432
Stata J. 2014 Apr 1;14(2):418-431
pubmed: 25420071
Ann Behav Med. 2003 Dec;26(3):172-81
pubmed: 14644693
Am J Phys Anthropol. 2002 Nov;119(3):257-75
pubmed: 12365038
Psychol Methods. 2017 Mar;22(1):141-165
pubmed: 27607544
Test (Madr). 2009 May 1;18(1):1-43
pubmed: 21218187
Stat Methods Med Res. 2016 Aug;25(4):1471-89
pubmed: 23698867
Cancer Causes Control. 2002 Nov;13(9):813-23
pubmed: 12462546
Prev Sci. 2007 Sep;8(3):206-13
pubmed: 17549635
Clin Epidemiol. 2017 Mar 15;9:157-166
pubmed: 28352203
Stat Med. 2017 Nov 20;36(26):4094-4105
pubmed: 28783884
Ann Transl Med. 2016 Jan;4(1):9
pubmed: 26855945
Biostat Epidemiol. 2019;3(1):1-22
pubmed: 30693349
BMC Med Res Methodol. 2018 Dec 12;18(1):168
pubmed: 30541455
Comput Methods Programs Biomed. 2016 Aug;132:29-44
pubmed: 27282225
Eat Weight Disord. 2020 Feb;25(1):25-35
pubmed: 29525920
Stat Med. 2016 Jul 30;35(17):2938-54
pubmed: 26681666
Stat Modelling. 2011 Aug;11(4):351-370
pubmed: 22271079
Clin Genet. 2019 Jul;96(1):17-27
pubmed: 30820929