Premium
Missing Data in Long‐term Follow‐up of Patients with Eating Disorders Using the Body Attitude Test
Author(s) -
Geerdens Candida,
Vanderlinden Johan,
Pieters Guido,
De Herdt Amber,
Probst Michel
Publication year - 2013
Publication title -
european eating disorders review
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.511
H-Index - 67
eISSN - 1099-0968
pISSN - 1072-4133
DOI - 10.1002/erv.2205
Subject(s) - missing data , eating disorders , normative , multivariate statistics , psychology , random effects model , multivariate analysis , test (biology) , clinical psychology , statistics , medicine , mathematics , meta analysis , paleontology , philosophy , epistemology , biology
Objective Incompleteness of data is a major problem within clinical follow‐up studies. The aim of present study was to compare different statistical models in the management of follow‐up data in patients with eating disorders using the Body Attitude Test. Method A prospective longitudinal investigation with repeated evaluations over fixed time intervals was conducted among 807 female patients with eating disorders using the Body Attitude Test as a dependent‐variable. Three types of missing data mechanisms were explored: missing completely at random, missing at random and missing not at random. Results Multivariate analysis showed that the missing completely at random type of missing data mechanism is less reliable than the missing at random or missing not at random mechanisms. Five years after admission, the body experience of eating disorder patients is less deviant. Compared with normative data, 37% of the eating disorder patients still had a more negative body experience 5 years after admission. Discussion There is no single correct method for dealing with missing data. Therefore, it is recommended that multiple methods be used under different assumptions of absenteeism. Copyright © 2012 John Wiley & Sons, Ltd and Eating Disorders Association.