Marseille 2007
| Abstract #177 - DEVELOPMENT OF A HAART NON-ADHERENCE INTENSITY RISK SCORE | ||||
|
Session: 26.77: Posters B (Poster) on Tuesday in Chaired by Authors: Presenting Author: Dr Adriana Ammassari - INMI "L.Spallanzani", Italy |
||||
| Additional Authors: MD Adriana Ammassari, MD Maria Paola Trotta, MD Patrizia Marconi, MD Rita Murri, MD Antonella d'Arminio Monforte, MD Andrea Antinori, | ||||
| Aim: Although the detrimental effect of HAART non-adherence on clinical and viro-immunological outcome of antiretroviral therapy is clear, predictors of HAART non-adherence are yet not fully defined. A proper non-adherence risk identification and stratification is necessary to identify those persons, who are most likely to experience non-adherence and for whom targeted intervention is absolutely needed. This study is aimed to develop an HAART non-adherence intensity risk score (NAIRS) aimed to predict non-adherence in a cohort study. | ||||
| Method / Issue: Data on HAART non-adherence used for this analysis were derived from the AdICoNA cohort (a nested study within the I.Co.N.A. cohort). Adherence was assessed using a previously validated self-reported questionnaire, which investigated also the most common reasons for non-adherence and a list of self-reported symptoms. Non-adherence has been defined as having skipped at least one dose in the last three days or reporting at least occasional interruptions in drug supply. To build the NAIRS, different risk points were assigned to the predictors based on their relative odds ratio (OR) of a patient being non-adherent at univariable analysis. Accuracy measures of NAIRS (sensitivity, specificity, positive and negative predictive value) and the area under the ROC curve were also calculated. | ||||
| Results / Comments: A total of 542 patients were included in the present analysis: 37.6% reported a non-adherence behaviour. At univariable analysis, predictors of non-adherence were: 1) younger age (1.54; 95CI 1.36-2.76); 2) reporting non-adherence due to treatment complexity (1.58; 95CI 1.10-2.27); 3) reporting non-adherence due to unsuited regimen with daily routine (1.45; 95%CI 1.01-2.09); 4) reporting an higher symptoms/medication side-effects score (1.94; 95CI 1.36-2.76); 5) reporting active drug abuse (3.44; 95%CI 1.79-6.62); 6) unemployment (1.71; 95%CI 1.12-2.61). A strong association between increasing number of risk factors and risk of non-adherence was found, with an OR of 1.47 (95%CI 1.27-1.70) for each adjunctive risk factor at logistic regression model. Weighted NAIRS was constructed by assigning 1 point for each 0.5 increase in OR and obtaining an intensity score from 0 (absence of risk factors) to 11.5 (co-presence of all risk factors). The obtained score was robustly associated with non-adherence (1.26; 95% 1.16-1.38) and showed an area under the ROC curve of 0.64. Accuracy measures of NAIRS identified a cut-off of >3 as a reasonable balance between sensitivity (66.7%) and specificity (56.1%). Based on NAIRS quartiles, three risk groups were identified: low risk (<25%=NAIRS 0-1), medium risk (26-74%=NAIRS 1.5-4), high risk (>75%=NAIRS >4.5). Considering the low risk group as reference, medium and high risk groups had a significantly increased risk of non-adherence: 1.79 (95%CI 1.12-2.87) and 3.24 (95%CI 1.92-5.48), respectively. | ||||
| Discussion: Increasing number of selected patient- and treatment-related predictors is associated with higher risk of HAART non-adherence. Patients with high weighted risk score are 3-times more likely to be non-adherent. Overall, the non-adherence intensity risk score showed a satisfactory accuracy in predicting HAART non-adherence and, because it is a simplification of the information on which it is based, it could be easily remembered and implemented in both a clinical and research setting. | ||||
| Go Back | ||||
