Erschienen in:
01.03.2013 | Original Research
Symptom clusters in patients with brain metastases—a reanalysis comparing different statistical methods
verfasst von:
Emily Chen, Luluel Khan, Liying Zhang, Janet Nguyen, Liang Zeng, Gillian Bedard, May Tsao, Cyril Danjoux, Elizabeth Barnes, Arjun Sahgal, Lori Holden, Flo Jon, Kristopher Dennis, Edward Chow
Erschienen in:
Journal of Radiation Oncology
|
Ausgabe 1/2013
Einloggen, um Zugang zu erhalten
Abstract
Objective
The aim of this study was to examine whether symptom clusters identified among patients with brain metastases treated with whole brain irradiation were consistent using different statistical methods, and to determine the stability of symptom clusters from baseline to 12 weeks following whole brain radiotherapy.
Methods
Reanalysis of an existing data set compiled from 170 patients with brain metastases was performed using exploratory factor analysis (EFA) and hierarchical cluster analysis (HCA) to extract symptom clusters at baseline, 1, 2, 4, 8, and 12 weeks’ follow-up time points. Symptom clusters identified with these two methods were compared with findings employing principal component analysis (PCA) from our previously published study.
Results
Symptom clusters extracted at baseline and each subsequent follow-up generally varied depending on the analytical method employed. Twelve unique clusters were found at each follow-up with each method, and at only one period in time (8-week follow-up) did all three methods demonstrate the same cluster. Cluster findings using PCA and HCA correlated more strongly with each other than either did with the findings of EFA. Inconsistency in symptom cluster composition was also observed at different time intervals. While constituents of symptom clusters differed over time between the three analytical methods employed, symptoms within determined clusters such as anxiety and depression or fatigue and drowsiness consistently clustered together.
Conclusion
The stability of symptoms pairs observed indicates a robust interrelationship existed between the symptoms involved. Generally, symptom cluster analysis yields different results depending on which statistical method is employed. A key step in achieving consistency in symptom cluster research involves the utilization of a common statistical analysis method. Further research is warranted to determine the best analytical method that should be employed to provide the most clinical relevance.