Abstract
Bias due to selective non-response is often neglected in large-scale epidemiological studies. And, although some recent techniques enable adjustment for selective non-response, these are rarely applied. The Maastricht Cohort Study, a study on fatigue at work among 12140 respondents at baseline, enabled us to estimate the degree of bias in a real life data set. After seven subsequent measurements, spanning a 2 year period, 8070 respondents remained in the cohort. Two traditional ways of presenting longitudinal mean levels (means using all data, and means using only complete cases) are compared with adjusted mean levels, using mixed models. The difference between the complete case and overall mean levels and the adjusted means were about 2% for the continuous fatigue score and 6% for the proportion of fatigued cases. For the company mean scores the observed bias due to selective non-response might be as much as 30% for some of the company means for the continuous fatigue score and up to 160% for the estimated number of fatigued cases. We therefore conclude that bias due to selective non-response needs serious attention. Next to making vigorous attempts to minimize longitudinal non-response, the use of statistical adjustment is also recommended.
References
Britton A, Murray D, Bulstrode C, McPherson K, Denham R. Loss to follow-up: Does it matter? Lancet 1995; 345(8963): 1511-1512.
Ioannidis JP, Taha TE, Kumwenda N, et al. Predictors and impact of losses to follow-up in an HIV-1 perinatal transmission cohort in Malawi. Int J Epidemiol 1999; 28: 769-775.
Diggle P, Liang K, Zeger S. Analysis of longitudinal data. New York: Oxford University Press, 1994.
Rubin D. Inference and missing data. Biometrika 1976; 63: 581-592.
Little RJ, Wang Y. Pattern-mixture models for multivariate incomplete data with covariates. Biometrics 1996; 52: 98-111.
Hunt JR, White E. Retaining and tracking cohort study members. Epidemiol Rev 1998; 20: 57-70.
Hu FB, Goldberg J, Hedeker D, Flay BR, Pentz MA. Comparison of population-averaged and subject-specific approaches for analyzing repeated binary outcomes. Am J Epidemiol 1998; 147: 694-703.
Little R. Modeling the drop-out mechanism in repeated-measures studies. J Am Stat Assoc 1995; 90(431): 1112-1123.
Fairclough DL, Peterson HF, Chang V. Why are missing quality of life data a problem in clinical trials of cancer therapy? Stat Med 1998; 17: 667-677.
Kant IJ, Bültmann U, Schröer KAP, Beurskens AJHM, Amelsvoort LGPM van, Swaen GMH. An epidemiological approach to study fatigue in the working population: The Maastricht Cohort Study. Occup Environ Med 2003; 60(Suppl 1): i32-i39.
Bültmann U, Vries Md, Beurskens JHM, Bleijenberg G, Vercoulen JHMM, IJ K. Measurement of prolonged fatigue in the working population: Determination of a cut off point for the checklist individual strength. J Occup Health Psychol 2000; 5: 411-416.
Vercoulen JH, Swanink CM, Fennis JF, Galama JM, van der Meer JW, Bleijenberg G. Dimensional assessment of chronic fatigue syndrome. J Psychosom Res 1994; 38: 383-392.
Beurskens AJ, Bültmann U, Kant I, Vercoulen JH, Bleijenberg G, Swaen GM. Fatigue among working people: Validity of a questionnaire measure. Occup Environ Med 2000; 57: 353-357.
Bültmann U, Kant I, Beurskens A, Kasl S, Van den Brandt P. Fatigue and psychological distress in the working population: Psychometrics, prevalence, and correlates. J Psychosom Res 2002; 52: 443-450.
Wolfinger R, O'Connell M. Generalized linear mixed models: A pseudo-likelihood approach. J Stat Comput Simulat 1993; 48: 223-243.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
van Amelsvoort, L.G., Beurskens, A.J., Kant, I. et al. The effect of non-random loss to follow-up on group mean estimates in a longitudinal study. Eur J Epidemiol 19, 15–23 (2004). https://doi.org/10.1023/B:EJEP.0000013401.81078.84
Issue Date:
DOI: https://doi.org/10.1023/B:EJEP.0000013401.81078.84