Premium
Assessing intra‐individual changes in health related quality of life data in psychiatric clinical trials
Author(s) -
Mazumdar Sati,
Amanda Dew Mary,
Houck Patricia R.,
Reynolds Charles F.
Publication year - 2004
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/env.682
Subject(s) - medicine , quality of life (healthcare) , depression (economics) , clinical trial , population , psychology , clinical psychology , psychiatry , environmental health , nursing , pathology , economics , macroeconomics
Assessment of intra‐individual changes in health related quality of life (HRQOL) measures is important to develop individual change standards and to provide help in clinical practice. Methodological issues related to the analysis of HRQOL data depend on the nature of medical illnesses and the target population. However, one issue—the need to assess HRQOL changes across individuals over time—spans most conditions. Modest improvement in HRQOL may be harder to assess in elderly depressed persons who also have high rates of co‐existing medical and neurobiological illnesses. Powerful statistical techniques are necessary to assess these improvements. Focusing on methods related to the issue of measuring change, we describe the potential of a mixed‐modeling approach for the assessment of intra‐individual changes using HRQOL data measured in a large group of elderly patients with recurrent major depression. These patients were treated with combined pharmacotherapy and interpersonal psychotherapy. We used a mixed‐modeling approach and two pre‐post difference score methods. The percentage of intra‐individual statistically significant improvement rates was found to be largest for the mixed‐modeling approach. We conducted simulation studies to compare these three approaches for assessing these changes. The simulated data reflected the observed data. Results from the two simulation studies are in agreement with the observed results and provide positive evidence for an earlier hypothesis (Speer and Greenbaum, 1995) that the mixed‐modeling approach is more sensitive for assessing changes across individuals over time compared to the pre‐post difference score methods. Thus, we recommend the use of the mixed‐modeling approach to measure intra‐individual changes, based on a larger array of supporting evidence than has been available previously. Copyright © 2004 John Wiley & Sons, Ltd.