Rory Leisegang
Uppsala University
Objectives: Differentiated service delivery (DSD) models, which focus more on individual patient preferences and the needs of vulnerable subpopulations. Courier delivery of chronic medication to a patient’s home (home-refill) is an attractive and scalable intervention to potentially improve adherence to medication and prolong treatment durability, but in some settings, a personalized medicine rather than a population-based approach may be more appropriate given the potential for harm from unwanted disclosure.
Methods: Building on a previous retrospective analysis on ART naïve starting a first line regimen looking at the self-refill versus home-refill of ART, a marginal structural analysis was used to establish evidence for the intervention at a population level. An Emax model was then applied to establish a referent CD4 response prior to viral failure (VL>1000 copies/ml) or regimen switch. We then propose a hidden Markov model (HMM) (Plan et. al, PAGE, 2015). which was informed the residuals derived from the referent CD4 model, which could be used to identify individuals who had failed treatment, and validated the results against the viral load.
Results: 40,939 patients, contributing over follow-up 66,000 years were evaluated. In a baseline analysis only, courier was associated with improved survival (adjusted hazard ratio = 0.90 [95% CI: 0.84-0.96], p-value for log-rank test < 0.001) after adjusting for baseline differences. Within an MSM framework, which addresses time-varying aspects, courier was associated with higher benefit (adjusted hazard ratio = 0.66 [95% CI: 0.55-0.78]) than with a typical regression analysis. In terms of CD4 response, EC50 was lower (time to 50% of CD4 gain) in the courier group suggesting a faster response.
Conclusions: Our findings support the adoption of home-refill (courier) within the DSD models and the potential to impact patient care at at an individual level using an HMM approach. Further research is needed on the potential impact of home-refill in vulnerable groups with known transportation barriers – e.g., unemployed, postpartum woman, and adolescents – and in other illnesses.
References:
[1] Plan, E., et al., Handling Underlying Discrete variables with Mixed Hidden Markov Models in NONMEM. , in PAGE. 2015.
Reference: PAGE () Abstr 9553 [www.page-meeting.org/?abstract=9553]
Poster: Clinical Applications