This study, the first to report real-world comparative effectiveness data across all available oral and injectable DMTs, indicates that DMF and fingolimod have similar effectiveness for relapse prevention in patients with MS and demonstrate superior effectiveness to IFNβ, GA, and teriflunomide in routine clinical practice. This finding was consistent across all subgroup and sensitivity analyses indicating that, irrespective of the patient subgroup assessed, DMF and fingolimod provide substantial and significant reductions in ARR that are superior to those provided by IFNβ, GA, and teriflunomide.
The findings in the present study suggest an effect of the pre-treatment relapse rate on the on-treatment relapse rate. ARR in the pre-index period was highest in the DMF and fingolimod cohorts, with the largest ARR reductions observed in the same cohorts. The findings also suggest an effect of prior therapy on post-DMT initiation relapse rates; approximately 85% of patients initiating IFNβ or GA had not received treatment in the previous year compared with 31–36% of patients receiving DMF, fingolimod, or teriflunomide. Interpretation of these results should therefore be conducted in the context of the differences between cohorts as determinants of pre-index ARR, such as the level of disability, MS phenotype, and time since first MS diagnosis. As claims databases do not include this information, the current analysis is unable to control for these factors. However, these data derived from routine clinical practice are consistent with those from a mixed treatment comparison of the comparative efficacy of DMF, IFNβ, GA, teriflunomide, and fingolimod based on clinical trial data . Our results now extend these findings to the real-world setting and confirm that, in routine clinical practice, DMF is likely to provide superior outcomes for patients compared with teriflunomide, GA, and IFNβ, as well as the benefits of oral therapy, in terms of patient convenience, compared with the latter two parenterally administered DMTs. The use of long-term, self-administered injections is burdensome to some patients, especially those with a needle phobia, and is associated with injection-related adverse events . Inconvenience, needle phobia, and injection-site reactions, among others, are factors recognized to negatively affect adherence to therapy . By contrast, the use of oral therapies overcomes patient concerns relating to needle phobia and injection-site reactions and offers a convenient method of administration .
The similar effectiveness data for DMF and fingolimod in our analysis are consistent with prior analyses based on the clinical trial data, indicating that both DMTs provide an appropriate choice of therapy for patients with RRMS. Moreover, the findings in the present study are consistent with an analysis of 775 propensity-matched patients with MS receiving either DMF (n = 458) or fingolimod (n = 317) in routine clinical practice in the USA . In this analysis, there was no significant difference in ARR, overall brain magnetic resonance imaging activity, and discontinuations at 1 year of therapy for patients receiving DMF or fingolimod .
Results reported in the present study were robust, irrespective of the subgroup or sensitivity analysis performed, and consistent with the large number of patients included in the analysis. Neither age (<40 years) nor adherence to therapy (MPR of ≥0.8, ≥0.7, or ≥0.6) substantially impacted on the conclusions from the primary analysis. The minimal influence of DMT adherence on the relapse rates is important given the potential role of adherence in DMT effectiveness. Our findings indicate that differences in DMT effectiveness remain robust even with moderate decreases in DMT adherence. Sensitivity analyses also had no relevant impact on the findings from the primary analysis, irrespective of changes to the method of analysis used (negative binomial regression or logistics regression) and to the time frame considered as a single relapse event. Together, the subgroup and sensitivity analyses confirm our findings that DMF provides substantial and significant reductions in ARR that are comparable with those provided by fingolimod and superior to those provided by IFNβ, GA, and teriflunomide.
As clinical evidence of MS relapse is not included in claims data, relapses were identified based on a validated algorithm derived from the ICD-9-CM diagnosis code and details of medications dispensed. This algorithm has been used in previous studies to determine relapse rates in patients with RRMS based on claims data and allows the identification of moderate-to-severe relapses that result in a clinical encounter. However, patients with a mild relapse may not seek an appointment with a physician; thus, mild relapses may be difficult to identify.
Validation of the algorithm by Chastek and colleagues was achieved by a comparison with medical charts and clinician review, resulting in the classification of 67.3% of patients with relapses (positive predictive value) and 70.0% of patients without relapses (negative predictive value) in a sample of 300 patients . Although this approach may underestimate the overall relapse incidence, this effect is anticipated to be uniform across all comparisons. We also assessed algorithm robustness and sensitivity by varying input parameters (including using the originally published algorithm), considering outpatient visits involving plasma exchange and high-dose oral steroid use as identifying a relapse event and considering clinical encounters occurring within 7 days (as opposed to 30 days) of each other as a single event. ARRs derived by all variations of the algorithm were in good agreement, differing by less than 0.1 (data not shown). Furthermore, the ARRs reported here are in the same range as those reported from clinical trials and are consistent with the ARRs reported in studies of real-world data [17, 25]. Our study, therefore, provides further confirmation of the validity of this approach for determining ARR from claims data.
The present study is subject to a number of the limitations associated with retrospective studies utilizing administrative claims data. First, data are collected for the purposes of billing and monitoring the quality of care, not for assessing treatment effectiveness, and are subject to coding errors, although these are anticipated to occur at random. Second, the data do not contain information on the diagnostic criteria used, severity and duration of MS, rate of progression, or results of laboratory or imaging tests; differences in neurologic or MS-related disability were not measured or accounted for as this information is not captured in claims data. These data would allow the analysis to control for differences in cohort clinical characteristics. Future studies are needed to develop and validate an algorithm to control for these confounders. Finally, MS relapses were identified from administrative claims and do not necessarily correspond to events as identified in clinical trials (e.g., a relapse can be defined as new or recurrent neurologic symptoms lasting for ≥24 h and accompanied by new objective neurologic findings) . However, the influence of potential misclassification is not expected to bias specific DMTs.
In this study, patients were not randomized to treatment, and Poisson and negative binomial regression was applied to account for differences in some of the clinical and demographic variables observed in the data set. Matching patients, through propensity score matching, provides an alternative approach to controlling for these variables, but would not have eliminated any potential confounding for the matched variables . The use of regression analysis in this study maintains statistical precision by allowing the use of all patient information rather than discarding information for unmatched patients . Importantly, as discussed, the results of this study were comparable with those reported for DMF and fingolimod in propensity-matched patients with MS . However, further studies are needed to confirm whether the clinical and demographic variables not captured in claims data influence relapse rates in patients with MS receiving an oral or injectable DMT.