Budget impact of antiretroviral therapy in a French clinic cohort.

OBJECTIVES

In this study, we first assessed costs associated with the use of antiretroviral therapy (ART) in an infectious diseases University Hospital Clinic; second, we evaluated characteristics associated with these costs and finally simulated the impact on the overall ART budget of switching first-line and second-line regimens to less-costly regimens (as effective and well tolerated).

DESIGN

Cohort analysis including persons living with HIV (PLHIV) aged at least 18 years on ART to estimate ART costs during 2014.

METHODS

The current study was conducted in the Bichat-Claude Bernard University Hospital Clinic in Paris, France, where 4501 PLHIV consulted in 2014. We used the medical database Nadis to describe patients' ART, characteristics and estimated costs. When assessing the budgetary impact of potential switches, we considered patients' history of failure, CD4 cell count, plasma viral load, resistance mutations, hepatitis B surface antigen or HLAB5701 profile.

RESULTS

A total of 4238 of 4501 patients were on ART (94%). The total annual cost of ART prescribed was estimated at &OV0556;48 280 200 in 2014; first/second (simplification)-line regimens represented 25% (1076/4238) of the treated PLHIV and 23% (&OV0556;11 209 000) of the annual cost. For these PLHIV, we considered switches from the most common ART regimens (protease inhibitor boosted by ritonavir or nonnucleoside reverse transcriptase inhibitor + two nucleoside reverse transcriptase inhibitors) to less-expensive regimens. We found savings ranging from &OV0556;36 100 to 1472 600/year. Savings were the highest when we considered switching to generic-based regimens or from protease inhibitor-based triple therapy to protease inhibitor monotherapy.

CONCLUSION

Costs associated with ART prescriptions are very high. Switches to generic-based regimens are associated with large savings. However, those targeting protease inhibitor regimens are also associated with substantial savings and should be considered.