Paper
Saturday, 22 July 2006
This presentation is part of : Nursing Research Methodological Strategies
Analyzing Symptom Clusters: A Methodological Journey
Teri G. Lindgren, RN, MPH, PhD1, Yoshimi Fukuoka, RN, MS, PhD2, and Sally H. Rankin, RN, PhD, FAAN2. (1) Community Health Systems, University of California, San Francisco, San Francisco, CA, USA, (2) Family Health Care Nursing, University of California, San Francisco, San Francisco, CA, USA
Learning Objective #1: Understand the process, expectations and pitfalls of cluster analysis
Learning Objective #2: Be prepared to use cluster analysis as a methodology to interpret symptom clusters
Symptom management is a major aspect of care for many health care providers and clinicians recognize that patients commonly experience multiple symptoms that fall into patterns sometimes described as symptom constellations or clusters.  To date, research on symptoms has focused primarily on one symptom but symptom clusters have recently received more attention. However, the conceptualization of symptom clusters and the methodologies used to study this phenomenon are still evolving and there is limited information on how to begin the process of analyzing symptom clusters.  The purpose of this paper is to assist researchers interested in this new field by describing the maiden voyage of a group of UCSF researchers seeking to analyze cardiac symptom data using a particular methodology, Cluster Analysis. Data comes from a multi-site randomized intervention trial of 247 patients experiencing acute cardiac symptoms; a primary aim of the study was to test the efficacy of a nursed trained peer advisor on a variety of patient outcomes, including symptom experience. Data on symptoms were collected 5 times over 1 year. We plan to describe the clusters at all time points, how the clusters change over time and investigate if the clusters predict outcomes.  However, in starting to analyze the symptom data we encountered difficulties and ambiguities that had to be resolved in determining the number of clusters evident in the data before we could even begin to describe and interpret the clusters. We learned that even though this form of data analysis is quantitative, it is not an exact science; rather it has qualitative aspects that require interpretive approaches. Through describing our analytical process, the lessons learned and the decisions taken, we hope that other researchers will be better prepared to use cluster analysis in future studies. 

See more of Nursing Research Methodological Strategies
See more of The 17th International Nursing Research Congress Focusing on Evidence-Based Practice (19-22 July 2006)