An Assessment of Factors Associated with High Costs Among Patients with Multiple Sclerosis (MS) Receiving Disease-Modifying Drug (DMD) Therapy

Background: There are limited data on factors associated with high medical costs of care in an MS population. Claims-based markers for health status may be useful to understand costs.

Objectives: To compare patient (pt) characteristics and frequency of diagnoses between high- and non-high-cost MS pts receiving DMDs.

Methods: MS pts (age 18–63; ≥1 MS diagnosis claim: ICD-9-CM: 340.xx) with ≥1 DMD claim (first claim=index date) and continuous eligibility 12 months pre- and post-index were identified from a random sample of 5 million lives in the IMS LifeLink Plus database from 1/1/2007 to 6/30/2012. Pts with all-cause total costs (excluding DMD costs) ≥75^th percentile were considered high-cost. Diagnoses codes (ie, “condition indicators”) were grouped into 3 domains: MS-related conditions (eg, disability), Clinical Classification System (CCS) code categories (eg, gastrointestinal) and Charlson-Deyo comorbidities. Fisher and Wilcoxon tests were used in unadjusted statistical comparisons. Logistic regression was used to evaluate likelihood of being a high-cost pt. Covariates included demographics, condition indicators, dalfampridine use, newly initiating DMDs and adherence.

Results: Analysis included 24,815 pts. 75^th percentile for high-cost status was $11,740, yielding 6207 high-cost and 18,608 non-high-cost pts (mean age: 46.3 vs 44.2, respectively; p<0.001). In unadjusted analyses, percentage of pts with each condition indicator was statistically significantly higher in the high-cost group (p<0.05). In logistic regression analyses, age and sex were not consistently predictive of being high-cost. High counts of conditions within each domain (eg, ≥5 MS-related conditions: odds ratio [OR]: 5.801; p<0.0001) and selected individual conditions (eg, disability: OR: 1.809; p<0.0001) were associated with significantly higher likelihood of being high-cost. Dalfampridine use, a symptomatic agent, was also significantly associated with being high-cost (OR: 5.744–6.062 across specifications; p<0.0001). Conversely, better adherence (medication possession ratio ≥80%) and newly initiating DMDs were associated with a lower likelihood of being high-cost (OR: 0.570–0.594 and 0.850–0.898 across specifications, respectively; all p<0.01).

Conclusions: MS-related conditions, CCS categories and Charlson-Deyo comorbidities were independently associated with high costs. Interventions targeting individuals affected by key conditions may be important to reduce costs and disease burden.