Identifying Distinct High Unmet-Need Phenotypes and Their Associated Bladder Cancer Patient Demographic, Clinical, Psychosocial, and Functional Characteristics: Results of Two Clustering Methods

Abstract

Objectives: We explored phenotypes of high unmet need of patients with bladder cancer and their associated patient demographic, clinical, psychosocial, and functional characteristics.

Data sources: Patients (N=159) were recruited from the Bladder Cancer Advocacy Network and completed an online survey measuring unmet needs (BCNAS-32), quality of life (FACT-Bl), anxiety and depression (HADS), coping (BRIEF Cope), social support (SPS), and self-efficacy beliefs (GSE). Hierarchical agglomerative (HA) and partitioning clustering (PC) analyses were used to identify and confirm high unmet-need phenotypes and their associated patient characteristics. Results showed a two-cluster solution; a cluster of patients with high unmet needs (17% and 34%, respectively) and a cluster of patients with low-moderate unmet needs (83% and 66%, respectively). These two methods showed moderate agreement (κ=0.57) and no significant differences in patient demographic and clinical characteristics between the two groups. However, the high-need group identified by the HA clustering method had significantly higher psychological (81 vs 66, p < .05), health system (93 vs 74, p < .001), daily living (93 vs 74, P < .001), sexuality (97 vs 69, P < .001), logistics (84 vs 69, P < .001), and communication (90 vs 76, P < .001) needs. This group also had worse quality of life and emotional adjustment and lower personal and social resources (P < .001) compared with the group identified by the PC method.

Conclusion: A significant proportion of patients with bladder cancer continues to have high unique but inter-related phenotypes of needs based on the HA clustering method.

Implications for nursing practice: Identifying characteristics of the most vulnerable patients will help tailor support programs to assist these patients with their unmet needs.

Keywords: Bladder cancer; Cluster analysis; Health-related quality of life; Hierarchical agglomerative clustering; Partitioning clustering; Unmet needs.