Quality of life profile and psychometric properties of the EQ-5D-5L in HIV/AIDS patients
- PDF / 199,454 Bytes
- 8 Pages / 595.28 x 793.7 pts Page_size
- 37 Downloads / 216 Views
RESEARCH
Open Access
Quality of life profile and psychometric properties of the EQ-5D-5L in HIV/AIDS patients Bach Xuan Tran1,2*, Arto Ohinmaa1† and Long Thanh Nguyen3†
Abstract Objectives: We assessed health-related quality of life (HRQOL), its associated factors, and examined measurement properties of the EuroQol - 5 Dimensions - 5 Levels (EQ-5D-5L) in HIV/AIDS patients. Methods: A cross-sectional multi-site survey was conducted in 1016 patients (age: 35.4 ± 7.0 years; 63.8% male) in three epicenters of Vietnam. Internal consistency reliability, convergent validity, and discriminative validity of the EQ-5D-5L and a visual analogue scale (VAS) were evaluated. Tobit censored regression models were used to identify predictors of HRQOL in HIV/AIDS patients. Results: The mean EQ-5D-5L single index and VAS were 0.65 (95% Confidence Interval (CI) = 0.63; 0.67) and 70.3 (95% CI = 69.2; 71.5). Cronbach’s alpha of five dimensions was 0.85. EQ-5D-5L has a good convergent validity with VAS (0.73). It discriminated patients at different HIV/AIDS stages, duration of ART, and CD4 cell count. Predictors of poorer HRQOL included being female, lower education level, unemployment, alcohol and drug use, CD40.8) indicates small, moderate and large differences in HRQOL. The percentage of respondents at floor and ceiling scores were examined for each dimension of the EQ-5D. The EQ-5D-5L single index was correlated with VAS score and a global
Page 3 of 8
rating of health-related quality of life to examine the convergent validity of the measurement. Multivariable linear regression was used to determine factors associated with HRQOL domain’s scores. Since the range of EQ-5D-5L single index and VAS ranged at (−0.452; 1) and (0; 100), these scales were left- and right censored. Censoring from above and below the HRQOL scores did not allow us to measure exactly the values which were higher or lower than the range thresholds. Therefore, in regression analysis, we employed censored regression models (Tobit models) to determined factors associated with EQ-5D-5L single index and VAS [37]. Candidates for multivariate models included those variables that met one of the following three criteria: (1) socio-economic status, (2) biological association with the outcome of interest; and (3) previously shown to be associated with the health-related quality of life among Vietnamese population [18,27,33]. We identified a set of candidate variables, including: (1) patient’s characteristics: gender, age, education, marital status, employment and income; (2) HIV/AIDS-related characteristics: CD4 cell count, HIV stages, and ART; and (3) patient’s adherence to ART, alcohol use, and drug use [18,24,27,33,38]. This set of candidate variables were entered in the full regression models. After that, we applied a stepwise forward selection to determine the reduced model that selected variables based on the log-likelihood ratio test at a p-value < 0.1, and excluded variables at p-values >0.2. Ethical consideration
The use of survey data was approved by the Autho
Data Loading...
