Disease-modifying drug initiation patterns in commercially insured multiple sclerosis patients: a retrospective cohort study

This retrospective cohort study used administrative healthcare claims from a database of US-based employers and health insurers to identify individuals newly diagnosed with MS during 2001-2007 and to construct a detailed claims history for each individual. Patients were characterized based on their demographics and baseline clinical indicators (diagnoses, comorbidities, concomitant medications) during the year prior to the index diagnosis, and then followed for at least one year to assess healthcare resource utilization, medical expenditures, treatments, and adherence to therapy.

Data came from two Thomson Reuters MarketScan^® research databases: the Commercial Claims and Encounters (Commercial) database and the Medicare Supplemental and Coordination of Benefits (Medicare) database, from January 1, 2000, through December 31, 2007. The Commercial database contained the inpatient, outpatient, and outpatient prescription drug experience of 44.5 million employees and their dependents under a variety of insurance coverage types. The Medicare database contained the healthcare experience of 4.5 million individuals with Medicare supplemental insurance paid for by employers, including both the Medicare-covered portion of payment and the employer-paid portion. All database records were de-identified and fully compliant with US patient confidentiality requirements (HIPAA).

Selected patients were a minimum of 18 years old with a diagnosis of MS (ICD-9-CM code 340) between January 1, 2001, and January 1, 2007 (patient selection period). A minimum of one inpatient or emergency department claim, or more than one outpatient claim at least 30 days apart with ICD-9-CM 340 (excluding diagnostic laboratory and diagnostic radiology claims; also referred to herein as "non-diagnostic" claims) was required. The index diagnosis date was assigned to the earlier of the first medical claim with an MS diagnosis or the date of the first medical or pharmacy claim for one of the DMDs during the patient selection period. Patients were required to have at least 12 months of continuous health plan eligibility with both medical and pharmacy benefits both before and following their index diagnosis date. To identify patients as "incident" or newly diagnosed, they were required to have no claims with a diagnosis of MS or any DMD therapy during the 12-months prior to their index diagnosis date. The study used administrative claims data in the absence of the patient medical charts, therefore relying on the diagnoses coded by the provider and the patient history found only within the healthcare claims. Patients were followed for a minimum of 12 months from their index date and then for as long as the data were available or until the end of calendar year 2007. Follow-up of subjects was done by ICD-9-CM code identification and not by actual chart review.

Diagnoses, medical procedures and pharmacologic therapy were derived from medical and pharmacy claims received according to service dates in the patient's pre-index and post-index periods. MS diagnoses and comorbidities were determined from medical claims (excluding diagnostic laboratory and radiology claims). A diagnosis code entered in any position for the claim was used. Pharmacologic therapy was identified from pharmacy claims and from HCPCS codes in medical claims. Medical procedures or diagnostic evaluations were identified from CPT-4 codes or ICD-9-CM procedure codes in medical claims. Drugs for treating MS were analyzed individually and by drug class.

Demographic data included age, gender, insurance plan type (comprehensive, EPO, HMO, POS, POS with capitation, PPO, and other), geographic region and length of follow-up. Clinical data included monitoring for selected comorbidities in medical claims, computation of the Deyo-Charlson comorbidity index (measured pre-index), use of DMD therapy, other pharmacological treatments for MS symptoms, and medical procedures and diagnostic tests associated with MS.

Treatment pattern analyses included time to DMD therapy initiation (for patients treated), persistence on therapy and discontinuation patterns. The length of continuous treatment with the index DMD was determined by considering the days supplied listed on the index drug claim(s), as well as the fill date and days supplied for subsequent claims of the same medication, including medical claims showing DMD administration. The length of index DMD therapy was defined as "continuous" by calculating days across consecutive prescription or medical claims of the index drug with no gaps of more than 30 days. A gap between consecutive prescriptions of the index drug is typically allowed because patients may not fill prescriptions the exact day they are due for a refill. The number of patients with gaps in therapy of 31-60 days, 61-90 days, 91-120 days and >120 days were also tracked. Gaps for patients receiving natalizumab and mitoxantrone were adjusted to accommodate for their longer dosing intervals, assuming 4 weeks for each medical claim date for natalizumab and 12 weeks from each medical claim date for mitoxantrone [30, 31]. As a part of the gap analysis, the number of patients with evidence of the index DMD after a gap of >60 days ("restarts") were identified and the average number of days to restart was measured. The length of DMD therapy from the first until the last drug claim of the index DMD was also measured. This measure counted the number of days between the treatment index date (i.e., the first drug claim for the index medication) and the last drug claim for the index medication (including days of supply) during the study period, regardless of gaps in therapy. The length of continuous treatment with all DMD therapy was measured, both with and without gaps of greater than 30 days. For each patient's drug regimen, the time from the first DMD claim to the date of the last applicable DMD claim (including days of supply) was measured, including any DMD therapy that was received. Switching among DMDs and discontinuation of all DMDs (i.e., no evidence of any DMD for a 60-day period following the end of the last observed prescription) was also analyzed.

Categorical variables were presented as counts and proportions. Continuous variables were summarized by providing the number of observations, the mean, the standard deviation, the median, and the range. Statistical tests of significance for differences in these distributions included Chi-square tests to evaluate the statistical significance of differences for categorical variables, t-tests and ANOVA for normally distributed continuous variables, while nonparametric Wilcoxon and Kruskal-Wallis tests were used for continuous variables that are not normally distributed. P-values of 0.050 or less were considered statistically significant.

Multivariate regressions were conducted to control for imbalances in observed baseline demographics and clinical characteristics. Cox proportional hazards regressions were used to assess characteristics associated with treatment initiation with adjustment for time from diagnosis to first treatment as well as to assess treatment persistence, measured as the time to a 31 day gap in therapy, among those treated. A logistic regression model was also used to predict patient characteristics affecting the initiation of treatment with DMDs. For each of the three models, age and gender were included as covariates and a backward elimination method was employed to select among the many other characteristics, with the variables remaining in each model each statistically significant at a level of 5%. All statistical analyses were performed using SAS version 9.2.

Initially there were 67,869 Commercial and Medicare patients with at least one inpatient or two non-diagnostic outpatient claims at least 30 days apart with an MS diagnosis between January 1, 2001 and January 1, 2007. Of these, 34,699 (51.1%) were excluded after applying twelve months pre- and post-index continuous eligibility requirements, 306 (0.5%) were excluded who were under the age of 18, and 21,803 patients (32.1%) were excluded who had an MS diagnosis or DMD treatment during the pre-index period, resulting in a final cohort of 11,061 patients.

Population demographics are shown in Table 1. Females comprised 74.2% of the study population. The average length of follow up was 35.7 ± 17.5 months. There were 40.3% (n = 4,462) of patients treated with at least one of the DMDs during the study period. The mean population age was 50.3 ± 13.3 years, with DMD-treated patients significantly younger (mean age 44.4 ± 10.6 years) than those without DMD treatment (mean age 54.3 ± 13.4 years, p < 0.001). The age distributions show skewing of treatment to younger age groups, with 48.8% of DMD-treated patients under the age of 44 compared with 23.3% of untreated patients, and 2.2% of patients over 65 with DMD treatment versus 20.7% of untreated patients.

Chronic pain conditions were the most frequently noted comorbidities in the medical claims of this MS population (Table 2), with over half of the cohort reporting chronic skeletal pain (joint derangements, dorsopathies, or polymyalgia rheumatica). With the exception of fatigue, headache, depression, skin cancers, and sleep disorders, all other comorbidities were noted in significantly higher percentages of patients not treated with DMDs. The Deyo-Charlson Comorbidity Index was calculated for all patients, measured using pre-index comorbidities, and was significantly lower for DMD-treated patients, computed at 0.36 ± 0.86 for DMD-treated patients and 0.64 ± 1.24 for untreated patients, p < 0.001.

The most common index DMD used in treating these incident patients was glatiramer (33.7% of treated incident patients) followed by interferon-beta-1a IM (32.1%) and interferon-beta-1a SQ (20.5%) (Table 2). The start of availability for interferon beta-1a SQ in the US in February, 2002 is acknowledged as a factor possibly underestimating its relative reported usage over the five-year time span of this study. Similarly, it is noted that natalizumab was approved by the US FDA in November 2004, withdrawn in February 2005, and then re-approved for relapsing MS in June 2006.

Other drugs used for acute MS exacerbations but without established evidence of altering the course of disease progression are also shown in Table 2. Glucocorticoids usage was highest in patients receiving DMDs post-index (61.4%) compared with patients who were not treated with DMDs post-index (40.0%). Usage of the other agents was generally in 2% or fewer of patients. No patients were found in this timeframe who had received cladribine, daclizumab, or anti-lymphocyte globulin post-index.

Muscle relaxants, antidepressants, and anti-fatigue agents were frequently prescribed post-index in this population, also shown in Table 2, particularly among DMD-treated patients whose percentage of patients with drug use was significantly higher for all three of these drug classes compared to untreated patients (p < 0.001). Pre-index use of these medications was significantly more prevalent in patients who did not receive DMD treatment (muscle relaxants 16.1% DMD vs. 20.8% non-DMD, p < 0.001; antidepressants 20.3% DMD vs. 22.2 non-DMD, p = 0.018; anti-fatigue agents 3.5% DMD vs. 5.8% non-DMD, p < 0.001).

The mean time to therapy initiation was 5.3 ± 9.1 months following the index diagnosis (Table 3). Once started on their index medication, more than three-quarters (78.7%) of these patients stayed on their index medication without switching to another DMD for the duration of the follow-up period. Nearly three-quarters of patients (71.0%) stayed on their index therapy an average of 17.0 ± 15.9 months without gaps exceeding 30 days. Including the 21.3% of patients who switched to another DMD after an average of 14.8 ± 12.9 months on their index DMD, patients appeared to remain on all DMD therapy without gaps exceeding 30 days for an average of 18.8 ± 16.8 months. (Table 3)

There were 29.0% of patients with gaps in therapy exceeding 30 days, and over 16% with gaps in excess of 60 days (restarting therapy), as shown in Table 3. Overall, 27.7% of patients discontinued DMD therapy altogether after an average of 17.6 ± 14.6 months without restarting any DMD therapy.

Factors with statistically significant associations with the initiation of DMD therapy are shown in Table 4. The hazard ratio denotes the odds that one group of patients was treated with a DMD compared with a reference group of patients, controlling for the influence of other variables in the model and the length of follow-up. Those variables associated with an increased likelihood of initiating DMD therapy include patient age (the lower the patient age, the higher the odds of being treated), evidence of loss of coordination, and the occurrence of an NMRI or spinal tap prior to the index diagnosis. Factors associated with a decreased likelihood of DMD initiation include higher degrees of comorbidity (per the Deyo-Charlson Comorbidity Index showing a 13 percent decrease for each point rise in the index), headache, sleep disorders, or the occurrence of pressure ulcers prior to the index diagnosis, as well as the pre-index use of azathioprine, anti-fatigue medications, or muscle relaxants.

Factors with statistically significant associations with the persistence on DMD therapy are shown in Table 5. Those associated with an increased likelihood of persisting on DMD therapy include female gender, age of 18 to 34, having an HMO insurance plan type, higher degrees of comorbidity (Deyo-Charlson Comorbidity Index), pre-index comorbidities of depression, headache, or urinary tract infection, and pre-index use of antidepressants or muscle relaxants. Factors associated with a decreased likelihood of DMD persistence include age greater than 34 and pre-index neuropathic pain.

This study was subject to a number of limitations. Using administrative claims data in the absence of the patient medical charts, and limiting the look-back period to 12 months of pre-index claims history precluded establishing each patient's long term prior disease history. These factors also obscured a definitive MS diagnosis, the type of MS (e.g., relapsing-remitting, primary progressive), identification of the initial date of an MS diagnosis, therapy occurring prior to the start of the study, disease severity, and the occurrence of relapse. Claims analysis is limited in its ability to account for all possible differences in patients and providers residing in different states. In addition, misclassification errors are possible when relying on diagnosis coding from administrative claims data and where the extent of undercoding of diagnoses is unknown. In the absence of diagnosis coding on pharmacy claims, it was assumed that the drugs shown to be used for treating MS were actually used as such. Further, the data sources did not include information on patient assistance programs or investigational drug treatment programs administered outside of a health plan's coverage or benefit plan, and so only treatments submitted to a health plan for coverage were available for analysis. The databases represent primarily employed or retired individuals, and thus may not be generalizable to the entirety of the US MS population.