This study involved a retrospective analysis of approximately 174 million unique de-identified patients (from as early as 1995) in the MarketScan Commercial Claims and Encounters (CCE) Database, a medical and drug insurance claims database, including active employees, early retirees, COBRA continuers, and dependents insured by employer-sponsored plans. The database was established in 1988 and contains inpatient admission records, outpatient services, prescription drugs, populations, eligibility status, and costs of services .
Patient Selection and Study Design
The study population was identified from the MarketScan CCE Database using administrative claims data, from January 1, 2010, to March 31, 2015, for patients with MS (defined as having at least one MS diagnosis [ICD-9: 340] within the year prior to initiating an IFN beta therapy), aged 18–64 years, who were relapse free for at least 1 year while continuously treated with IFN beta therapy (SC IFN beta-1a, SC IFN beta-1b, or IM IFN beta-1a). Patients also had to be continuously enrolled for 1 year prior to the index date (first claim = index date; baseline) and 1 year after the index date (follow-up) (Fig. 1).
Patients were propensity score matched 3:1, using the greedy method, for those who stayed on their baseline therapy (no-switch) versus those who switched to another IFN therapy (switch). Exact matching was based on the following: age category, sex, month/year of index date (first claim = index date), previous IFN used, and the prior year’s adherence category (medication possession ratio [MPR], < 0.6, 0.6–0.7, ≥ 0.8).
Among switch patients who were matched in terms of the above categories, the three closest no-switch patients were selected on the basis of a logistic propensity model using the above characteristics, as well as insurance plan type, the number of doctor visits in the year prior to the index date, total medical costs (log transformed), and MS symptoms as defined by experts in the field.
Among patients in each group, the Charslon comorbidity index  was using to assess the extent of comorbidities.
Adherence was assessed using patient refill records to ascertain the MPR (defined as the sum of all days’ supplies for all fills of the drug in a particular time period [pre or post index date], divided by the number of days in the time period) and the proportion of days covered (PDC; defined as the number of covered days in a particular time period divided by the total number of days in the time period).
The operational definition of relapses, adapted from Chastek et al. , included any inpatient hospital stay with a primary diagnosis of MS or any outpatient visit (either an emergency room or office visit) with the use of corticosteroids (high dose of oral steroid [daily dose of ≥ 500 mg prednisone] or intravenous steroid), adrenocorticotropic hormone, or total plasma exchange within 30 days after the stay and/or visit.
Outcomes were compared for switch versus no-switch groups for patients stable on any IFN beta therapy at baseline and then for the subset of patients stable on IM IFN beta-1a at baseline. Descriptive statistics were provided for cohorts before and after matching. Mean and standard deviation were presented for continuous measures, and count and proportion were presented for categorical measures.
Statistical testing was conducted to detect statistically significant difference (i.e., P < 0.05) between switch and no-switch groups before and after matching. Comparisons between treatment groups were analyzed using the chi-squared test or Fisher exact test for categorical measures and nonparametric Wilcoxon rank-sum test for continuous measures (e.g., costs).
The data contained in the MarketScan CCE Database are statistically de-identified and have been certified to satisfy the conditions set forth in section 164.514 (a)–(b)1ii of the Health Insurance Portability and Accountability Act. As such, the patient information used in this study is exempt from the US Department of Health and Human Services regulations that require institutional review board approval, and patient consent was not deemed necessary.