TITLE:
Symptom Cluster Research in Women with Breast Cancer: A Comparison of Three Subgrouping Techniques
AUTHORS:
Angela R. Starkweather, Debra E. Lyon, R. K. Elswick Jr., Alison Montpetit, Yvette Conley, Nancy L. McCain
KEYWORDS:
Breast Neoplasms; Cluster Analysis; Symptom Clusters; Psychoneurological Symptoms
JOURNAL NAME:
Advances in Breast Cancer Research,
Vol.2 No.4,
September
9,
2013
ABSTRACT:
Aims: To examine how symptom cluster subgroups defined by extreme
discordant composite scores, cut-off scores, or a median split influence
statistical associations with peripheral cytokine levels in women with breast
cancer. Background: Systemic cytokine dysregulation has been posited as a potential biological
mechanism underlying symptom clusters in women with breast cancer. Symptom
characteristics may play an important role in identifying cytokines of
significant etiological importance, however, there is no consensus regarding to the
ideal subgrouping technique to use. Design: A secondary analysis of data collected from a cross-sectional
descriptive study of women with stage I-II breast cancer was used to examine
and compare the relationships between peripheral cytokine levels and symptom
subgroups defined by extreme discordant composite scores, cut-off scores, or a
median split. Methods: Participant
symptom scores were transformed into a composite score to account for
variability in symptom intensity, frequency and interference. Cytokine levels
in subgroups defined by composite scores within the highest and lowest 20% were
contrasted with those composed from cut-off scores and a median split. Results: Subgroups defined by the
composite score or cut-off scores resulted in similar statistical relationships
with cytokine levels in contrast to the median split technique. The use of
a median split for evaluating relationships between symptoms clusters and
cytokine levels may increase the risk of a type I error. Conclusion: Composite and
cut-off scores represent best techniques for defining symptom cluster subgroups
in women with breast cancer. Using a consistent approach to define symptom
clusters across studies may assist in identifying relevant biological
mechanisms.