A comparison of the FACT-G and the Supportive Care Needs Survey (SCNS) in women with ovarian cancer: unidimensionality of constructs

Abstract

Purpose: Health-related quality of life (HRQoL) and unmet needs (needs) questionnaires offer alternative perspectives for assessing cancer patients' concerns. We examined whether the conceptual differences underlying these alternative approaches yield corresponding empirical differences.

Methods: Eight-hundred and seventy-four women with ovarian cancer completed the Functional Assessment of Cancer Therapy scale (FACT-G; HRQoL) and the Supportive Care Needs Survey (SCNS-SF34; needs) every 3 months for 2 years. Correlational analysis, exploratory and confirmatory factor analysis (EFA/CFA), and Rasch analysis tested the relationship between patients' responses to similar domains and similar items across the two questionnaires.

Results: Strong correlations were found between items with virtually identical wording (0.67-0.75), while moderate to strong correlations (0.55-0.65) were found for those with very similar wording. EFA identified two common domains across the two questionnaires: physical and psychological. For each common domain, CFA indicated models involving a single construct with systematic variation within each questionnaire fit best. Rasch analysis including very similar items within the physical and psychological domains (separately) demonstrated strong evidence of unidimensionality.

Conclusions: The high degree of similarity between patient responses to items addressing the same or very similar concerns suggests either that HRQoL and needs approaches do not reflect different constructs or that patients may not be able to differentiate between the severity of a concern and the level of need associated with that concern, especially when these are assessed in quick succession.