Health Care Utilization and Costs in Patients With Migraine Who Have Failed Previous Preventive Treatments

Abstract

Objective

To characterize health care utilization (HCU) and associated costs among patients with migraine categorized by the number of preventive treatment failures (TFs; 1 TF, 2 TFs, and 3+ TFs) using real-world data.

Methods

This retrospective analysis identified adults with incident migraine diagnosis in the IBM MarketScan Commercial and Medicare Supplemental database between January 1, 2011, and June 30, 2015. TF was defined in the 2 years after the first migraine diagnosis period. One TF, 2 TFs, and 3+ TFs were defined as patients who had received only 2 preventive treatments (PTs), 3 PTs, and 4+ PTs in the 2-year period, respectively. A negative binomial model was used to analyze HCU data, and a 2-part model was used for cost data controlling for the preindex Deyo-Charlson Comorbidity Index.

Results

Overall, 24,282 patients with incident migraine who had failed at least 1 PT were included in the analysis. Of these, 72.7% (n = 17,653) had 1 TF, 20.2% (n = 4,900) had 2 TFs, and 7.1% (n = 1,729) had 3+ TFs. Adjusted annualized rates of all-cause and migraine-specific HCU increased with an increase in the number of TFs (1.4–4 times higher; all p < 0.0001 vs 1 TF). The mean total all-cause health care costs were higher by $3,732 (95% confidence interval [CI]: $2,708–$4,588) in patients with 2 TFs and by $8,912 (95% CI: $7,141–$10,822) in patients with 3+ TFs vs those with 1 TF. Outpatient costs were the key drivers of differences in health care costs.

Conclusions

TF in patients with migraine was associated with a substantial resource and cost burden, which increased with the number of TFs.

TAKE-HOME POINTS

• → Failure to previous preventive treatments (PTs) imposes a significant challenge in migraine management. However, the economic burden of migraine in patients with a history of treatment failure (TF) has not been characterized.

• → Using claims data from a nationally representative US database (2010–2017), this study quantified the resource and cost burden associated with failed preventive treatments in patients with migraine.

• → This retrospective analysis demonstrates that health care utilization (HCU) and associated costs are substantially higher among patients with a history of 3+ TFs followed by 2 TFs and 1 TF.

• → The study findings underscore a substantial unmet need among patients with migraine who have failed previous PTs and reinforce the need for better preventive therapies, which could reduce HCU and costs associated with TF.

Migraine is a debilitating neurologic disorder that affects approximately 12% of the general population in the United States.¹ Migraine is 3 times more prevalent in women than in men² and often affects people in their most productive years (age 25–55 years),³ causing a substantial effect on the functionality and quality of life of affected individuals.^4,5

The economic burden associated with migraine in the US population has been well documented.^1,6-9 The annual cost burden has been estimated to be over $56 billion (inflated to 2013 US dollars), with over 60% attributable to direct costs.^6,7,9,10 The incremental total direct costs were estimated at approximately $9.2 billion per year for patients with migraine compared with those without migraine.⁹ Moreover, data from these claims-based analyses suggest that the mean annual direct all-cause costs are approximately 4 times higher in individuals living with migraine compared with those living without the disease.^6,7 Although these studies emphasize the economic burden associated with migraine as a whole,^6-10 recent studies suggest that failure of prior preventive treatments (PTs) may further add to the resource and cost burden in patients with migraine.^11,12 Data quantifying the burden associated with failed PTs in patients with migraine are scarce. Therefore, the purpose of the current retrospective, observational cohort study was to characterize health care utilization (HCU) and associated costs among patients with migraine stratified by the number of preventive treatment failures (TFs; 1 TF, 2 TFs, and 3+ TFs).

Methods

Study Design and Data Source

This longitudinal, retrospective cohort study used administrative medical and pharmacy claims data from the IBM MarketScan Commercial Claims and Encounters (Commercial) and the Medicare Supplemental and Coordination of Benefits (Medicare Supplemental) databases. The Commercial and Medicare supplemental databases include enrollment data and health insurance claims across the continuum of care (e.g., inpatient, outpatient, and outpatient pharmacy) for their respective covered populations and together reflect a nationally representative sample of insured individuals living in the United States. The commercial database contains health insurance claims across the continuum of care (e.g., inpatient, outpatient, and outpatient pharmacy) as well as enrollment data from large employers and health plans across the United States who provide private health care coverage for employees, their spouses, and dependents. The Medicare supplemental database profiles the health care experience of individuals with Medicare supplemental insurance paid for by employers and is representative of the national population of retirees with both Medicare and supplemental employer-sponsored health coverage.

Standard Protocol Approvals, Registrations, and Patient Consents

All study data are fully compliant with US patient confidentiality requirements, including the Health Insurance Portability and Accountability Act of 1996. As this study used only retrospective deidentified patient data, it was exempted from institutional review board approval, and formal informed patient consent was not obtained.

Patient Selection

Adult patients (aged ≥18 years) with incident migraine between January 1, 2011, and June 30, 2015 (identification period) were identified. Patients with incident migraine were defined as having at least 2 diagnoses of migraine on different dates ≥7 days apart in OP claims or at least 1 emergency department (ED) or hospitalization claim with a primary diagnosis of migraine in the identification period and no migraine diagnosis in the 12 months before the first migraine diagnosis (index 1). The date of the first migraine diagnosis was defined as index 1 (figure 1). All patients were required to have continuous enrollment with medical and pharmacy coverage for 12 months before and 24 months after index 1 and have continuous medical and pharmacy coverage in the 12 months after the index date (i.e., initiation of the third PT in patients with 2 TFs). Patients with a migraine diagnosis and neuropathy, anxiety, arrhythmias, hypertension, depression, or epilepsy in the 12 months before the first migraine diagnosis (i.e., index 1) were excluded from the analysis. These patients were excluded as they might have received any of the listed preventive medications (appendix e-1, links.lww.com/CPJ/A274) as treatment for other illnesses rather than for the prevention of migraine, and the databases did not outline which drugs were used for which diagnosis. The list of migraine PTs in appendix e-1, links.lww.com/CPJ/A274, was developed adopting a series of steps. The initial list of PTs was based on the AHS 2019 position statement and AAN 2012 treatment guidelines.^13,14 Following this, medical experts who specialized in migraine treatment and a team of internal experts reviewed and finalized the list of preventives included in this study. Only pharmacologic treatment that is prescribed in the prevention of migraine was included in this study.

Figure 1. Study Design.

^aDefinition of index date: (1) for patients with 1 TF in the 2 years after the index 1 period, the index date is the start of the second PT. (2) For patients with 2 TFs in the 2 years after the index 1 period, the index date is the start of the third PT. (3) For patients with 3+ TFs in the 2 years after the index 1 period, the index date is the start of the fourth PT. PT = preventive treatment; TF = treatment failure.

Based on the number of PTs received during the 2-year follow-up period after index 1, patients with incident migraine were stratified into 3 cohorts: (1) 1 TF—patients who received only 2 PTs, (2) 2 TFs—patients who received 3 PTs, and (3) 3+ TFs—patients who received at least 4 PTs. The index date for patients with 1 TF was the date of initiation of the second PT. The index date for patients with 2 TFs and 3+ TFs was the date of initiation of the third and fourth PTs, respectively. The PT is in the individual generic drug level. Because these PTs are often used for other nonmigraine conditions, to qualify as a migraine PT, any claimed medication was required to have a non–rule out migraine diagnosis within 14 days before the drug claim and at least 28 days of supply.

Study Variables

All-cause and migraine-specific HCU and costs in the 12-month follow-up period (i.e., from the index date to 12 months after the index date) were measured overall and by service category (IP, OP, ED visits, and pharmacy claims). Migraine-specific HCU and associated costs were defined as ED/IP or OP claims with a diagnosis of migraine (primary diagnosis for ED/IP visits and diagnosis 1 or 2 for OP visits). Across all-cause and migraine-specific costs, the total medical (ED + IP + OP) costs and health care (medical + prescription drug) costs were reported. Costs referred to the total reimbursed amount, including patients' copays, deductibles, and coinsurance. All costs were inflation adjusted to 2017 US dollars using the medical care component of the Consumer Price Index.

Other Variables

Demographic characteristics including age, sex, and geographic region (US Census division) were measured on the index date. Clinical characteristics were measured in the 12 months before the index date and included the Deyo-Charlson Comorbidity Index (CCI; a measure of general health status that categorizes 17 major comorbidities in patients based on the International Classification of Diseases, Ninth/Tenth Revisions, Clinical Modification [ICD-9/10-CM]),¹⁵ general comorbid conditions, CM without aura, and migraine medication use (based on the National Drug Code and Healthcare Common Procedure Coding System claims).

Statistical Analyses

All study measures were summarized descriptively. Categorical variables were presented as counts and percentages, and continuous variables were summarized as means with SDs.

A negative binomial model with baseline Deyo-CCI as a covariate was used to assess the HCU for patients with migraine with 3+ TFs vs 1 TF and for patients with migraine with 2 TFs vs 1 TF. Rate ratios, 95% confidence intervals (CIs), and p values were presented. A linear gamma regression with 2-part models was used for costs with ≥10% of value observations as 0; when <10% of observations were 0, only gamma regression was used. The baseline Deyo-CCI was used as an adjusting covariate. Mean costs for each cohort and the incremental cost difference between the cohorts (i.e., 3+ TFs vs 1 TF and 2 TFs vs 1 TF) were summarized with means and 95% CIs.

Data Availability

The original deidentified data used in this retrospective analysis were obtained from and are the property of IBM. The raw data were provided by IBM, which were used to create the analytic files for the study. If any researcher requires access to the analytic files that were derived from the MarketScan database, a third-party use agreement must be executed by the third party, Novartis, and IBM, and access fees may apply. Researchers interested in accessing the raw data that were used to generate the analytical files should contact lifesciences@us.ibm.com.

Results

Patient Disposition

A total of 24,282 patients with incident migraine with at least 1 PT between January 1, 2011, and June 30, 2015, were identified. Of these, 17,653 (73%) were classified as having 1 TF, 4,900 (20%) as having 2 TFs, and 1,729 (7%) as having 3+ TFs (figure 2).

Figure 2. Patient Attrition.

^aAt least 2 outpatient diagnoses of migraine on dates ≥7 days apart or at least 1 ED or inpatient claim with a primary diagnosis (outpatient ED all diagnosis) of the disease. ^bThe date of the first migraine diagnosis between January 1, 2011, and June 30, 2015. ^cDefinition of index date: (1) For patients with 1 TF in the 2 years after the index 1 period, the index date is the start of the second PT. (2) For patients with 2 TFs in the 2 years after the index 1 period, the index date is the start of the third PT. (3) For patients with 3+ TFs in the 2 years after the index 1 period, the index date is the start of the fourth PT. ^dPT must have a non–rule out migraine diagnosis within 14 days before the drug claim and at least 28 days of supply. ED = emergency department; PT = preventive treatment; TF = treatment failure.

Patient Characteristics

Baseline demographic and clinical characteristics are presented in table 1. The mean (SD) age of the patients across the 3 TF cohorts ranged between 43.4 (12.9) and 44.0 (12.6) years. Most patients were female (83.6%–85.6%) and resided in the Southern region of the United States (43.3%–45.3%). The 3+ TF cohort had a relatively higher proportion of patients with CM (identified based on the diagnosis code for CM without aura; 47.1%) than the 2 TF (28.3%) and 1 TF (13.1%) cohorts.

Table 1.

Characteristics of Patients With Migraine Stratified by Number of TFs

The mean Deyo-CCI score in the 12 months before the index period was higher in patients in the 3+ TF cohort (0.51) compared with patients in the 2 TF (0.44) and 1 TF (0.37) cohorts. Comorbidities were relatively higher in the 3+ TF cohort vs the 1 TF and 2 TF cohorts. The most prevalent selected comorbidities across the 3 cohorts included cardiovascular disease (22%–28.6%), depression (18.3%–28.9%), anxiety (13.2%–24.4%), fibromyalgia (10.8%–18.2%), and complications of constipation (9.4%–17.7%) (table 1).

HCU

During the 12-month period following the index date, the unadjusted all-cause and migraine-specific HCU increased with the number of TFs. The proportion of patients with all-cause and migraine-specific ED + IP visits, office visits, and other OP visits was relatively higher among patients in the 3+ TF cohort compared with patients in the 1 TF and 2 TF cohorts. Patients with 3+ TFs also had greater mean all-cause and migraine-specific HCU than those with fewer TFs (table 2).

Table 2.

Unadjusted HCU and Costs During the 12-Month Postindex Period

After controlling for Deyo-CCI, the estimated annualized rates of all-cause and migraine-specific HCU increased with an increase in the number of TFs (all p < 0.0001 vs 1 TF). The annualized rates of all-cause ED + IP visits, office visits, and other visits in patients with 3+ TFs and 2 TFs were approximately 1.2–1.5 times higher compared with patients with 1 TF. Similarly, the annualized rates of migraine-specific ED + IP visits, office visits, and other visits were approximately 3 times, 2 times, and over 4 times higher in patients with 3+ TFs compared with patients with 1 TF, respectively. Compared with all-cause HCU, the rate ratios for migraine-specific HCU were much higher for patients with 3+ TFs and 2 TFs than for patients with 1 TF (figure 3, A and B).

Figure 3. Adjusted HCU During the 12-Month Postindex Period.

(A) All-cause HCU. (B) Migraine-specific HCU; *p < 0.0001. A negative binomial model was used. The Deyo-CCI was used as an adjusting covariate. CI = confidence interval; CCI = Charlson Comorbidity Index; ED = emergency department; IP = inpatient; OP = outpatient; RR = rate ratio; TF = treatment failure.

Health Care Costs

Over the 12-month follow-up period after the index date, patients in the 3+ TF cohort had the highest unadjusted mean total all-cause costs (medical: $17,962; health care: $24,665), followed by the 2 TF cohort (medical: $14,088; health care: $19,098) and 1 TF cohort (medical: $11,294; health care: $15,384). The mean total migraine-specific medical costs were $3,679 for patients with 3+ TFs, $1,903 for patients with 2 TFs, and $896 for patients with 1 TF. Similarly, the mean prescription drug costs were higher for patients with 3+ TFs ($6,702) than for patients with 2 TFs ($5,010) and 1 TF ($4,090). Differences in unadjusted health care costs were primarily driven by OP costs (table 2).

The mean adjusted total all-cause health care costs were $3,732 higher (95% CI: $2,708–$4,588) for patients with 2 TFs and $8,912 higher (95% CI: $7,141–$10,822) for patients with 3+ TFs vs those with 1 TF. The mean adjusted total all-cause medical costs per patient were higher by $2,836 (95% CI: $2,006–$3,605) for patients with 2 TFs and by $6,493 (95% CI: $4,899–$8,503) for patients with 3+ TFs vs those with 1 TF. The mean adjusted incremental all-cause prescription drug costs were $905 (95% CI: $556–$1,273) for patients with 2 TFs and $2,410 (95% CI: $1,953–$2,860) for patients with 3+ TFs compared with those with 1 TF. Similarly, the 2 TF and 3+ TF cohorts had incremental migraine-specific costs of $1,007 (95% CI: $866–$1,165) and $2,779 (95% CI: $2,419–$3,200) compared with the 1 TF cohort, respectively. As observed with the unadjusted costs, the differences in adjusted health care costs were primarily driven by greater increases in OP costs (figure 4, A and B).

Figure 4. Adjusted Health Care Costs During the 12-Month Postindex Period.

(A) All-cause costs. (B) Migraine-specific costs. A 2-part model was used for costs when ≥10% of value observations were 0; when <10% of observations were 0, only gamma regression was used. The Deyo-CCI was used as an adjusting covariate. CI = confidence interval; CCI = Charlson Comorbidity Index; ED = emergency department; IP = inpatient; OP = outpatient; TF = treatment failure.

Discussion

This retrospective analysis evaluated the economic burden among patients with migraine in the United States who had failed previous PTs. Our analysis showed that HCU and costs (medical, prescription drug, and health care) measured over a 12-month follow-up period were higher for patients with migraine with 3+ TFs compared with those with fewer TFs (2 TFs and 1 TF). These results suggest that TF is associated with a substantial economic burden and further contributes to the total cost of care in patients with migraine.

Migraine imposes an enormous cost burden, and failure of previous PTs may further add to the economic burden.¹² Understanding the magnitude of the economic burden associated with TFs may provide useful insights to payers and health care providers on unmet needs in difficult-to-treat patients with migraine. In this analysis that assessed HCU and costs in patients with migraine who have failed one or more PTs, an increase in TFs was associated with a substantial increase in the utilization of health care services (ED + IP, and OP visits). As a consequence of a higher HCU, patients incurred higher health care costs. Specifically, the total all-cause medical costs were 24.5% higher for patients with 2 TFs and 56.1% higher for patients with 3+ TFs vs patients with 1 TF. Similarly, the prescription drug costs were 21.7% higher for patients with 2 TFs and 57.8% higher for patients with 3+ TFs vs patients with 1 TF. Furthermore, patients with migraine with 2 TFs and 3+ TFs incurred 23.7% and 55.6% higher all-cause health care costs (medical + prescription drug) compared with patients with 1 TF, respectively. In line with previous studies that estimated the cost of migraine care,^6,7,10,16 OP costs were the largest component of the total health care costs in this study population. Taken together, the data from the current study reveal that the costs of treating migraine sharply increase with the number of TFs. For example, patients with migraine with 3+ TFs are likely to incur an additional cost of $8,912 per patient compared with patients with 1 TF. Increased HCU and costs in patients with 3+ TFs could be owing to more patients with CM and a higher number of comorbidities that accrue more health care services. Patients with CM have been reported to use more health care resources and incur over 3 times the health care costs compared with patients with episodic migraine.^10,17 Moreover, preexisting comorbidities have been reported to complicate the therapeutic management of migraine, further contributing to the total health care costs related to migraine.^17-19

A direct comparison of our findings with other migraine studies is difficult, as previous studies comparing HCU and costs among patients with migraine by TF status is rare. However, we can benchmark the results with a few existing studies that have reported that the migraine burden increased with the addition of a PT.^11,12,16 In 2019, using US-based claims data (2011–2013), it was reported that patients with migraine who cycled through multiple PTs accrued more health care visits (IP, OP, and ED visits). Furthermore, patients who switched to third and fourth PT classes incurred higher mean all-cause total direct costs per patient during the 12-month postindex period than those who persisted with the initial PT ($13,429 and $18,394 vs $11,941 [inflated to 2014 US dollars]; all p < 0.0001).¹⁶ In a Finnish retrospective study, patients with migraine with ≥3 PTs compared with no PT accrued higher overall visits (26.2 vs 13.8 visits per patient-year), disease-specific visits (6.0 vs 1.0 visits per patient-year), and sick-leave days (overall: 30.4 vs 16.8 per patient-year; disease-specific: 6.7 vs 0.6 per patient-year).¹¹ A global survey conducted in 31 countries to assess the real-world burden among adults with migraine experiencing ≥4 monthly migraine days showed that a higher number of monthly migraine days and history of failed PTs were associated with an increase in the utilization of health care resources.¹²

This study has several limitations inherent to administrative claims data. There is a potential for misclassification of migraine, covariates, or study outcomes as data for this analysis were retrieved from administrative claims databases as opposed to medical records. Furthermore, administrative claims databases are prone to coding and data entry errors. Owing to the observational nature of this study, the analysis may have been affected by unobserved confounding factors. A lack of adequate information to accurately evaluate the severity of migraine exists, as clinical markers are not captured in claims data. Given that TFs are not precisely captured in insurance claims data, the number of PTs that patients received was used as a proxy for the number of TFs. Moreover, the database lacks information on the reason for TF. The use of over-the-counter medications and other self-management techniques are not captured in the claims data, so the use of over-the-counter drugs, nutraceuticals, or other medications for the management of migraine was not included in our analysis. The study analyzed individuals with commercial health coverage or private Medicare supplemental coverage in the United States; thus, findings may not apply to the general migraine population. Finally, this study was conducted before the approval and availability of CGRP monoclonal antibodies for migraine prevention.

Although we acknowledge the above limitations of this study, it does have strengths that merit consideration. In contrast to previous studies that focused on HCU, costs, and treatment patterns in patients with migraine,^{3,6-12,16,17,19,20} this study more specifically quantified real-world resource and cost burden associated with failed PTs in patients with migraine. This analysis evaluated a large and diverse patient population retrieved from a sizable insurance claims database, making these estimates representative of the US-managed health care population. Finally, by investigating the economic implications of failed PTs in patients with migraine, this study adds new insights that fill an important gap in the literature.

The results of this real-world analysis show that TFs in patients with migraine were associated with a substantial resource and cost burden. Patients with migraine with 3+ TFs had a substantially higher HCU that translated to substantial costs (medical, prescription, and health care) compared with patients with fewer TFs (2 TFs and 1 TF). The findings underscore a substantial unmet need among patients with migraine who have failed previous PTs and reinforce the need for better preventive therapies, which could alleviate HCU and costs associated with TF.

Appendix. Authors

Study Funding

This study was funded by Novartis Pharma AG, Basel, Switzerland.

Disclosure

L. Newman is a consultant/advisory board member for Amgen, Allergan, Biohaven, Eli Lilly, Lundbeck, Medscape, Novartis, Teva, Theranica, WebMD, Zosano, and Xoc and receives royalties from Oxford University Press. P. Vo is a full-time employee of Novartis and holds shares of Novartis. L. Zhou is an employee of KMK Consulting Inc. C. Lopez Lopez and A. Cheadle are full-time employees of Novartis and hold shares of Novartis. M. Olson is a full-time employee of Novartis. J. Fang is a full-time employee of Novartis and holds shares of Novartis. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.