Treatment Persistency With Rivastigmine and Donepezil in a Large State Medicaid Program

To the Editor: Cholinesterase (ChE) inhibitors can provide improvement in symptoms, temporary stabilization of cognition, or reduction in the rate of cognitive decline in patients with mild to moderate Alzheimer's disease (AD).1,2 Although persistency with treatment is an important aspect of drug therapy and is likely to reflect the balance between efficacy and tolerability, there is limited information regarding persistency with AD therapy outside the clinical trial setting. We studied the drug persistency patterns for patients treated with rivastigmine or donepezil in usual-care settings in a large state Medicaid program. Longitudinal data from California-Medicaid program were used to study patients newly initiated on rivastigmine or donepezil from April 2000. All treated patients were required to be aged 40 and older, have had continuous eligibility in MediCal for at least 3 months before the first prescription, and could not have taken any ChE inhibitor treatment during this time. Drug discontinuation was defined as a drug switch or no prescription refill within 60 days of estimated completion of prior prescription. Time to treatment discontinuation was analyzed using Kaplan-Meier survival plots for patients on the two therapies. The observation period ended on December 31, 2002. A total of 17,742 patients met the eligibility criteria of the study. Of these, 15,128 were treated with donepezil, and 2,614 received rivastigmine. The two groups were similar in age (mean±standard error 76.9±0.2) for rivastigmine and 76.9±0.08 for donepezil) and sex distribution (69.6% women for rivastigmine and 68.7% women for donepezil). The mean duration of follow up was 431 days. A total of 10,186 patients (67.3%) taking donepezil discontinued their treatment, compared with 1,586 (60.7%) patients taking rivastigmine. The Kaplan Meier analysis (Figure 1) showed no statistically significant difference between the two drugs. Cox proportional hazards analysis also showed no significant difference between the two drugs, after adjusting for age and sex. The median time to treatment discontinuation was 120 days for donepezil and 135 days for rivastigmine (not significant). Kaplan-Meier plot of time to drug discontinuation among subjects on rivastigmine and donepezil. The persistency results from real-world data in this study are in contrast to those from a recent randomized open-label study that compared treatment outcomes and persistency after 12 weeks for patients receiving rivastigmine with those of patients taking donepezil.3 The results indicated that more donepezil-treated patients (89.3%) completed the study than rivastigmine-treated patients (69.1%), even though both groups showed comparable cognitive improvements. Persistency results from randomized clinical trials are often difficult to generalize to real-world practice. Randomized clinical trials often have stringent inclusion and exclusion criteria and emphasize homogeneity at the expense of generalizability. Protocol-necessitated monitoring may also affect patient compliance and persistence. Therefore, it is difficult to compare the two studies. There are several limitations to this study. First, prescription refill rates indicate that a prescription was filled but do not prove that the patient actually took the drug. However, studies in other disease areas have shown that pharmacy-dispensing records correlate well with patient drug exposure.4,5 Second, studies based on administrative databases do not have data on measures of cognitive and functional status, such as the Mini-Mental State Examination and activity of daily living index. Therefore, it is not possible to assess the severity of AD activity in the two treatment groups and control for any differences. However, the fact that only newly treated patients with AD who had not been exposed to any ChE treatment were studied may offset to some extent this concern. There are several strengths to the observational design of this study. Claims databases offer access to large populations of patients that might be excluded from randomized clinical trials. The longitudinal nature of these data allows for longer-term analyses than can generally be conducted with clinical data. Patients enrolled in Medicaid are less likely to drop out or change their insurance plans than patients in commercial insurance plans. Medicaid databases are also more likely to contain information on patients aged 65 and older than databases based on commercial insurance plans because Medicaid rules allow dual coverage with Medicare in patients aged 65 and older. In conclusion, in this study of a Medicaid population, adherence to treatment with rivastigmine was found to be similar to adherence to treatment with donepezil. Financial Disclosure: Novartis Pharmaceuticals Corporation funded this project through funds provided to the Institute of Clinical Outcomes Research and Education. Drs. Thomas and Arcona are employees of Novartis Pharmaceuticals. No other authors have any conflict of interest. Authors' Contribution: Gurkirpal Singh, principal investigator, was responsible for study concept, design, analysis, interpretation, and preparation of manuscript. Simu Thomas was responsible for study concept, design, and interpretation. Vijayabharathi Lingala, statistician and programmer, was responsible for analysis and interpretation. Alka Mithal was responsible for study concept, design, interpretation, and preparation of manuscript. Steve Arcona was responsible for study concept, design, and interpretation. Sponsor's Role: Two coauthors who are employees of Novartis Pharmaceuticals had a role in suggesting study concept, design, and interpretation, but the principal investigator (GS) had the final authority on all issues of study concept, design, interpretation, decision to publish, and preparation of manuscript.