Medication persistence of targeted immunomodulators for plaque psoriasis: A retrospective analysis using a U.S. claims database

Kaplan-Meier curves of persistence. A, 150/90 model, targeted drug-naïve, B, 150/90 model, targeted drug-experienced, C, 120 model, targeted drug-naïve, D, 120 model, targeted drug-experienced, E, days' supply model, targeted drug-naïve, F, days' supply model, targeted drug-experienced



Studies of medication persistence in plaque psoriasis have shown inconsistent results, likely due to differing definitions of nonpersistence and of the permissible gap between refills. Also, medication persistence information for two recently approved drugs, apremilast and ixekizumab, is limited.


We use the Truven Health MarketScan claims database to assess persistence for six drugs: adalimumab, apremilast, etanercept, ixekizumab, secukinumab, and ustekinumab. We define the permissible gap in three ways: 150 days for ustekinumab and 90 days for all other drugs (150/90 model); 120 days for all drugs (120 model); and twice the days' supply for all drugs (days' supply model). To estimate unadjusted persistence, we use Kaplan‐Meier curves, and a proportional hazards model to estimate the adjusted risk of non‐persistence.


Ustekinumab is most sensitive to changes in the definition of permissible gap, likely because of its longer maintenance dosing interval. Among targeted drug‐experienced patients using ustekinumab, median persistence is 358 days (95% confidence interval: 343‐371) in the 150/90 model and 189 days (179‐199) in the days' supply model. Among targeted drug‐experienced patients, median persistence in the days' supply model is longest for ixekizumab and secukinumab at 252 (217‐301) and 222 (210‐244) days, respectively. We also find that adjusted risk of nonpersistence increases by approximately 1% per year at treatment start.


The definition of permissible gap meaningfully changes both absolute and ordinal estimates of medication persistence. Each definition has unique limitations, which should be considered when interpreting persistence data.

Pharmacoepidemiology & Drug Safety
Nathaniel Hendrix
Nathaniel Hendrix
Researcher and data scientist