Snapshot of Statistical Methods Used in Geriatric Cohort Studies: How Do We Treat Missing Data in Publications?

Authors

  • Diklah Geva Department of Epidemiology and Health Science Evaluation, Faculty of Health Science, Ben Gurion University of the Negev, P.O. Box 653 Beer-Sheva 84105, Israel
  • Danit Shahar Department of Epidemiology and Health Science Evaluation, Faculty of Health Science, Ben Gurion University of the Negev, P.O. Box 653 Beer-Sheva 84105, Israel
  • Tamara Harris Laboratory of Epidemiology, Demography, and Biometry, Gateway Building, 3C309, 7201 Wisconsin Avenue, Bethesda, MD 20892, USA
  • Sigal Tepper Department of Epidemiology and Health Science Evaluation, Faculty of Health Science, Ben Gurion University of the Negev, P.O. Box 653 Beer-Sheva 84105, Israel
  • Michael Friger Department of Epidemiology and Health Science Evaluation, Faculty of Health Science, Ben Gurion University of the Negev, P.O. Box 653 Beer-Sheva 84105, Israel

DOI:

https://doi.org/10.6000/1929-6029.2013.02.04.5

Keywords:

Missing data, geriatric cohort studies, methodologies review, longitudinal analysis

Abstract

Background: Geriatric studies often miss data of frail participants. The aim of this paper is to explore which missing data methodologies have entered current practice and to discuss the potential impact of ignoring the issue.

Methods: A Sample of 103 articles was drawn from key cohort studies: Health ABC, InCHIANTI, LASA, BLSA, EPESE, and KLoSHA. The studies were classified according to missing data methodologies used.

Results: Seventy-seven percent described the selected analysis data set and only 28% used a method of handling all available observations per case. Missing data dedicated methods were rare (< 10%), applying single or multiple imputations for baseline variables. Studies with longer follow-up periods more often employed longitudinal analysis methodologies.

Conclusions: Despite the recognition that missing data is a major problem in studies of older persons, few published studies account for missing data using limited methodologies; this could affect the validity of study conclusions. We propose researchers apply Joint Modeling of longitudinal and time-to-event data, using shared-parameter model.

Author Biographies

Diklah Geva, Department of Epidemiology and Health Science Evaluation, Faculty of Health Science, Ben Gurion University of the Negev, P.O. Box 653 Beer-Sheva 84105, Israel

Department of Epidemiology and Health Science Evaluation, Faculty of Health Science,

Danit Shahar, Department of Epidemiology and Health Science Evaluation, Faculty of Health Science, Ben Gurion University of the Negev, P.O. Box 653 Beer-Sheva 84105, Israel

Department of Epidemiology and Health Science Evaluation, Faculty of Health Science,

Sigal Tepper, Department of Epidemiology and Health Science Evaluation, Faculty of Health Science, Ben Gurion University of the Negev, P.O. Box 653 Beer-Sheva 84105, Israel

Department of Epidemiology and Health Science Evaluation, Faculty of Health Science,

Michael Friger, Department of Epidemiology and Health Science Evaluation, Faculty of Health Science, Ben Gurion University of the Negev, P.O. Box 653 Beer-Sheva 84105, Israel

Center for Statistics (CenStat)

References

Panel on Handling Missing Data in Clinical Trials. The prevention and treatment of missing data in clinical trials. National Academy Press 2010.

Ahrens W, Pigeot I. Handbook of epidemiology. Springer 2005. DOI: https://doi.org/10.1007/978-3-540-26577-1

Ferraro KF, Kelley-Moore JA. A half century of longitudinal methods in social gerontology: Evidence of change in the journal. J Gerontol Series B: Psychol Sci Soc Sci 2003; 58(5): S264. DOI: https://doi.org/10.1093/geronb/58.5.S264

Yaffe K, Weston A, Graff-Radford NR, Satterfield S, Simonsick EM, Younkin SG, et al. Association of plasma β-amyloid level and cognitive reserve with subsequent cognitive decline. JAMA 2011; 305(3): 261. http://dx.doi.org/10.1001/jama.2010.1995 DOI: https://doi.org/10.1001/jama.2010.1995

Allison PD. Missing data. Thousand Oaks, CA: Sage Publications 2001.

Little R, Rubin D. Statistical analysis with missing data. 2nd edn. Wiley & Sons 2002. DOI: https://doi.org/10.1002/9781119013563

Rubin DB. Multiple imputation for nonresponse in surveys. Wiley & Sons 1987. http://dx.doi.org/10.1002/9780470316696 DOI: https://doi.org/10.1002/9780470316696

Zeger SL, Liang KY, Albert PS. Models for longitudinal data: A generalized estimating equation approach. Biometrics 1988; 44(4): 1049-60. http://dx.doi.org/10.2307/2531734 DOI: https://doi.org/10.2307/2531734

Hogan JW, Roy J, Korkontzelou C. Handling drop‐out in longitudinal studies. Stat Med 2004; 23(9): 1455-97. http://dx.doi.org/10.1002/sim.1728 DOI: https://doi.org/10.1002/sim.1728

Rotnitzky A, Robins JM, Scharfstein DO. Semiparametric regression for repeated outcomes with nonignorable nonresponse. J Am Statist Assoc 1998; 93(444): 1321-39. http://dx.doi.org/10.1080/01621459.1998.10473795 DOI: https://doi.org/10.1080/01621459.1998.10473795

Tsiatis A. Semiparametric theory and missing data. Springer 2006.

Tsiatis AA, Davidian M. Joint modeling of longitudinal and time-to-event data: An overview. Statistica Sinica 2004; 14(3): 809-34.

Rizopoulos D. Joint models for longitudinal and time-to-event data: With applications in R. CRC Press 2012. http://dx.doi.org/10.1201/b12208 DOI: https://doi.org/10.1201/b12208

Rizopoulos D. JM: An R package for the joint modelling of longitudinal and time-to-event data. J Statist Soft 2010; 35(9): 1-33. DOI: https://doi.org/10.18637/jss.v035.i09

Verbeke G, Molenberghs G. Linear mixed models for longitudinal data. Springer 2009.

Diggle P, Heagerty P, Liang K, Zeger S. Analysis of longitudinal data. Oxford University Press 2013.

Rizopoulos D. Dynamic predictions and prospective accuracy in joint models for longitudinal and Time‐to‐Event data. Biometrics 2011; 67(3): 819-29. http://dx.doi.org/10.1111/j.1541-0420.2010.01546.x DOI: https://doi.org/10.1111/j.1541-0420.2010.01546.x

Proust-Lima C, Taylor JM. Development and validation of a dynamic prognostic tool for prostate cancer recurrence using repeated measures of posttreatment PSA: A joint modeling approach. Biostatistics 2009; 10(3): 535-49. http://dx.doi.org/10.1093/biostatistics/kxp009 DOI: https://doi.org/10.1093/biostatistics/kxp009

Downloads

Published

2013-10-31

How to Cite

Geva, D., Shahar, D., Harris, T., Tepper, S., & Friger, M. (2013). Snapshot of Statistical Methods Used in Geriatric Cohort Studies: How Do We Treat Missing Data in Publications?. International Journal of Statistics in Medical Research, 2(4), 289–296. https://doi.org/10.6000/1929-6029.2013.02.04.5

Issue

Section

General Articles