Symptom-based clusters in patients with advanced chronic organ failure identify different trajectories of symptom variations

• Published in: Aging clinical and experimental research 2021-02, Vol.33 (2), p.419-428
• Main Authors: Finamore, Panaiotis, Janssen, Daisy J. A., Schols, Jos M. G. A., Verstraeten, Els R. N., Antonelli Incalzi, Raffaele, Wouters, Emiel F. M., Spruit, Martijn A.
• Format: Article
• Language: English
• Subjects: Chronic illnesses; Chronic obstructive pulmonary disease; Cluster analysis; Disease; Geriatrics/Gerontology; Health services; Heart failure; Heart Failure - diagnosis; Hemodialysis; Hospitals; Humans; Longitudinal Studies; Medicine; Medicine & Public Health; Original Article; Palliative care; Patient admissions; Pulmonary Disease, Chronic Obstructive; Quality of Life; R&D; Research & development; Syndrome
• ISSN: 1720-8319; 1594-0667; 1720-8319
• DOI: 10.1007/s40520-020-01711-z

Background Healthcare needs are complex and heterogeneous in advanced chronic organ failure. However, based on symptom clusters, groups of patients with similar quality of life, care dependency and life-sustaining treatment preferences can be identified. Aims To evaluate the stability of symptom-based clusters over time, and whether and to what extent the clusters are able to predict patients’ 2-year survival and hospitalization rates. Methods This is a secondary analysis of a longitudinal observational study including 95 outpatients with chronic obstructive pulmonary disease (COPD) GOLD stage III–IV, 80 outpatients with chronic heart failure (CHF) NYHA stage III–IV and 80 outpatients with chronic renal failure (CRF) requiring dialysis. Patients were clustered into three groups applying K-means algorithm on baseline symptoms’ severity and were then longitudinally evaluated. 2-year survival and hospital admissions during 1 year were estimated using Kaplan–Meier curves and Cox models. 1-year tendencies in symptom variation, using mixed linear models, and clusters comparison over time were performed. Results The three clusters were unable to predict patients’ survival and hospital admissions. Noteworthy, they show different trajectories of symptom variation, with Cluster 1 patients experiencing a worsening of symptoms, associated with an increased care dependency, and Cluster 2 and Cluster 3 patients being stable or having a relief in some symptoms. Although Cluster 1 is becoming more similar to Cluster 2, the three clusters preserve the overall characteristics and differences. Discussion Symptom-based clusters might help to identify patients with different trajectories of symptom variations. Conclusion Symptom clusters do not predict survival and hospital admissions and are stable over time.