A longitudinal analysis of symptom clusters in cancer patients and their sociodemographic predictors

Bejoy C Thomas; Amy Waller; Rebecca L Malhi; Tak Fung; Linda E Carlson; Shannon L Groff; Barry D Bultz

doi:10.1016/j.jpainsymman.2013.04.007

A longitudinal analysis of symptom clusters in cancer patients and their sociodemographic predictors

J Pain Symptom Manage. 2014 Mar;47(3):566-78. doi: 10.1016/j.jpainsymman.2013.04.007. Epub 2013 Sep 12.

Authors

Bejoy C Thomas¹, Amy Waller², Rebecca L Malhi³, Tak Fung⁴, Linda E Carlson⁵, Shannon L Groff⁶, Barry D Bultz⁵

Affiliations

¹ Department of Psychosocial Resources, Tom Baker Cancer Centre, University of Calgary, Calgary, Alberta, Canada; Department of Oncology, University of Calgary, Calgary, Alberta, Canada. Electronic address: bejoy.thomas@albertahealthservices.ca.
² Faculty of Health, University of Newcastle, Callaghan, New South Wales, Australia.
³ Department of Oncology, University of Calgary, Calgary, Alberta, Canada.
⁴ Department of Information Technologies, University of Calgary, Calgary, Alberta, Canada.
⁵ Department of Psychosocial Resources, Tom Baker Cancer Centre, University of Calgary, Calgary, Alberta, Canada; Department of Oncology, University of Calgary, Calgary, Alberta, Canada.
⁶ Department of Psychosocial Resources, Tom Baker Cancer Centre, University of Calgary, Calgary, Alberta, Canada.

PMID: 24035068
DOI: 10.1016/j.jpainsymman.2013.04.007

Abstract

Context: Exploring the relationships between concurrent symptoms or "symptom clusters" (SCs) longitudinally may complement the knowledge gained from the traditional approach of examining individual symptoms or SCs crosssectionally.

Objectives: To identify consistent SCs over the course of one year and determine the possible associations between SCs and demographic and medical characteristics, and between SCs and emotional distress.

Methods: This study was an exploratory longitudinal analysis of SCs in a large sample of newly diagnosed cancer patients. Patients provided symptom assessment data at baseline, three, six, and 12 months. A factor analysis was conducted (controlling for the patient over time) on pain, fatigue, anxiety, depression, sleep, weight change, and food intake items to identify clusters. A panel regression on each cluster explored associations with demographic and medical characteristics and distress.

Results: In total, 877 patients provided baseline data, with 505 retained at 12 months. Three SCs explained 71% of the variance. The somatic cluster included pain, fatigue, and sleep; the psychological cluster included anxiety and depression; and the nutrition cluster consisted of weight and food intake. Low income and treatment with radiation or chemotherapy predicted higher somatic symptom burden. Younger age, being female, low income, and treatment with surgery predicted more psychological symptomatology. Older age and treatment with surgery predicted higher nutritional burden. Patients with higher somatic, psychological, and nutritional symptom burden reported higher distress.

Conclusion: The presence of SCs across the first year of diagnosis supports the need for routine and ongoing screening for the range of symptoms that may be experienced by patients. Further work is needed to develop interventions that better target individual symptoms that cluster, as well as the entire cluster itself.

Keywords: Symptom clusters; cancer; distress; factor analysis; longitudinal study; panel regression.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Disease Progression
Female
Humans
Longitudinal Studies
Male
Middle Aged
Neoplasms / epidemiology*
Neoplasms / physiopathology*
Neoplasms / psychology
Risk Factors
Socioeconomic Factors
Time Factors