Validity of Katz Index to assess dependency in patients on oncology treatment

Abstract

Objective: To analyze construct validity of the Katz Index for patients with cancer diagnoses. Methods: Psychometric study, performed to evaluate evidences of validity of the Brazilian version of the Katz Index in patients with cancer diagnoses. The study was conducted in 691 hospitalized patients between April and June 2016. Data was analyzed by Exploratory and Confirmatory Factor Analysis. Reliability was assessed by Cronbach’s Alpha and McDonald’s Ômega. Results: Participants were mostly women (72.6%), with a mean age of 53.7±13.28 years; 55.3% reported that they did not need caregivers, and 56.6% did not have caregivers. Of the 42.5% who need help, predominant caregiver was a relative (39.6%). Exploratory and Confirmatory Factor Analysis evidenced that Katz Index is a unidimensional instrument, valid and reliable to assess dependency in this population, with a good variance and no evidence of multicollinearity or need of revision (KMO=0.89437; x2 Bartlett=3337.2 p<0.001; UNICO=0.999, EC=0.974; MIREAL=0.122). Katz Index explains 97.72% of the phenomenon and results indicate good adjustment of the model, acceptable level of residuals and good reliability (G-H=0.986; α-Cronbach=0.970560; Ω-McDonald=0.971787) to assess independence to basic activity daily living in patients with cancer diagnoses. Conclusion: The Brazilian version of the Katz Index presents excellent evidences of construct validity and reliability in its Brazilian version, and may be used as a clinical instrument to assess dependency of patients on oncology treatment.