Abstract
Objective
Estimate health care resource utilization and costs associated with medication overuse headache and potential acute medication overuse.
Methods
A retrospective analysis was conducted with Clinformatics Data Mart data (1 January 2019–31 December 2019) that included continuously enrolled commercially insured adults with migraine (International Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM] code G43.xxx). Medication overuse headache was defined as ≥1 inpatient or ≥2 outpatient claims with an ICD-10-CM code G44.41/40 (drug-induced headache). Potential acute medication overuse was defined as possessing sufficient medication for >10 mean treatment days/month for ergots, triptans, opioids, or combination analgesics or >15 mean cumulative days/month for simple prescription analgesics (e.g., acetaminophen, aspirin, other non-opioid analgesics) for >6 consecutive months. All-cause and migraine-related health care resource utilization and costs were compared after adjusting for demographic and clinical characteristics.
Results
Among 90,017 individuals with migraine, the frequency of medication overuse headache/potential acute medication overuse was 12.6% (diagnosed medication overuse headache: 0.6%; potential acute medication overuse: 12.1%). Adjusted all-cause total costs ($31,235 vs $21,486; difference: $9,749 [
Conclusions
Individuals with diagnosed medication overuse headache/potential acute medication overuse had higher all-cause and migraine-related health care resource utilization and costs versus individuals without medication overuse headache/potential acute medication overuse, suggesting that improved migraine management is needed to reduce associated costs.
Introduction
Migraine is a common neurological condition characterized by recurrent headaches and associated symptoms (1). Several acute or symptomatic treatment options are available for migraine attacks, including triptans, ergots, butalbital combinations, opioids/opioid combinations, simple analgesics, and non-steroidal anti-inflammatory drugs (2). The International Classification of Headache Disorders, 3rd edition (ICHD-3), describes medication overuse headache (MOH) as “headache occurring on 15 or more days/month in a patient with a pre-existing primary headache and developing as a consequence of regular overuse of acute or symptomatic headache medication (on 10 or more or 15 or more days/month, depending on the medication) for more than 3 months” (1). Therefore, MOH is considered a secondary headache disorder. The prevalence of MOH is estimated to be about 1% of the general population in Europe, North America, and Asia (3,4). Other terms closely related to MOH include analgesic rebound headache, drug-induced headache, or medication-misuse headache (5).
While MOH is the diagnosis of a secondary headache disorder, acute medication overuse (AMO) is defined based on the frequency of acute medication use and describes medication-taking behavior. AMO is a major risk factor in migraine progression (6). Barbiturates, opioids, and triptans have been associated with migraine progression (7). Barbiturates and opioids are used in the United States but are not recommended (8); their use is discouraged in Europe (9). AMO is thought to lead to MOH and interfere with successful migraine treatment (4,9). While numerous studies have highlighted the significant economic burden associated with migraine (10–12), little is known about the excess health care resource utilization (HCRU) and costs associated with AMO or MOH in migraine. This study examines the HCRU and costs associated with diagnosed MOH and potential AMO, defined based on acute treatment possession, in commercially insured individuals with migraine in the United States.
Methods
Study design
This retrospective, administrative, claims–based analysis aimed to estimate the economic burden (evaluated using HCRU and costs) associated with MOH and potential AMO using a prevalence-based, cross-sectional approach. We chose this approach for several reasons. First, identifying incident MOH/potential AMO is cumbersome. Second, for a condition present from the beginning of enrollment, it is not possible to determine when the condition began or was first diagnosed. Hence, assignment of the first medical claim for MOH during the study period as the “index” date (and then following individuals for 12 months to estimate annual costs) are artificial starting points to follow individuals over time. Furthermore, annual calendar year–based estimates of the economic burden of MOH may be more relevant to payers interested in understanding the budget impact in their annual enrolled population. Third, 2019 was the first full calendar year when the three injectable calcitonin gene–related peptide (CGRP) monoclonal antibodies (mAbs) were available on the market, given the US Food and Drug Administration approval of these new migraine treatments around mid-2018 and after. The availability of these relatively expensive agents may have changed the costs associated with migraine and MOH. Thus, the use of 2019 data provided a more relevant and generalizable perspective on the economic burden of MOH to the payer.
Therefore, a prevalence-based incremental cost-of-illness analysis, which estimates the excess costs attributable to MOH/potential AMO by comparing individuals with MOH/potential AMO to individuals without these conditions, was used to estimate the economic burden from a payer perspective.
Data source
This study used medical claims data, pharmacy claims data, and enrollment information from Optum’s Clinformatics Data Mart (CDM). The CDM is an administrative health claims database comprised of members from a large national insurance company affiliated with Optum. Providers and pharmacies submit administrative health claims for payment. The administrative health claims are then verified, adjudicated, adjusted, and de-identified prior to inclusion in the CDM database. The database is comprised of claims data from commercial and Medicare Advantage health plan members. Only the commercial health plan data were used for this study and contained a geographically diverse population across all 50 states.
Study sample
The criteria for the study sampling frame included adults (age 18–64 years) who were commercially insured, were continuously eligible for medical and prescription benefits, and had ≥1 inpatient claim or ≥2 outpatient claims on different dates with a diagnosis of migraine (International Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM] code G43.xxx) in any position between 1 January 2019, and 31 December 2019. Individuals with negative values for any of the cost outcomes were excluded.
From this overall migraine sample, we identified individuals with MOH/potential AMO and a comparison group of those without MOH/potential AMO. Individuals with MOH were identified using ≥1 inpatient OR ≥2 outpatient claims on different dates with an ICD-10-CM diagnosis for drug-induced headache (G44.41 drug-induced headache, not elsewhere classified, intractable; G44.40 drug-induced headache, not elsewhere classified, not intractable) in any position between 1 January 2019, and 31 December 2019.
To identify individuals with potential AMO, acute migraine prescription fills were examined over six-month sliding windows between 1 January 2019, and 31 December 2019. The analysis included the following five acute migraine medication classes: triptans, ergots, simple (nonopioid) analgesics, noncombination opioids, and combination analgesics (defined as formulations combining ≥2 medication classes). For each prescription, the number of days of available medication was calculated based on the quantity of medication divided by a daily dosing factor for the drug class. Individuals meeting any of the following ICHD-3 criteria in ≥1 of the six-month sliding windows during the year were coded as potential AMO patients: (a) Mean cumulative days of available medication >10 days/month across all prescriptions for triptans, opioids, ergotamines, or combination analgesics, alone or in combination, over the six-month window; (b) Mean cumulative days of available medication >15 days/month across all prescriptions for other simple analgesics, including acetaminophen, aspirin, nonsteroidal anti-inflammatory drugs (NSAIDs), or other non-opioid analgesics, alone or in combination, over the six-month window. Individuals with a diagnosis code for MOH were excluded from this group.
Individuals with diagnosed MOH or potential AMO were pooled into the MOH/potential AMO group, and those not meeting the inclusion criteria for either diagnosed MOH or potential AMO were used as the non-MOH/potential AMO comparison group.
Outcomes
The primary study outcomes included all-cause and migraine-related HCRU and costs. HCRU measures of interest included inpatient visits, emergency department (ED) visits, and physician visits for evaluation and management (E&M). E&M refers to the range of primarily cognitive tasks that providers perform based on the knowledge required for diagnosis and treatment. All-cause HCRU included all services of interest, regardless of diagnosis. Migraine-related HCRU included any medical claim for the services of interest with a diagnosis of migraine or headache in any position. Migraine-related prescription use included use of acute and preventive medications for migraine. The assessed acute medication classes included acetaminophen, triptans, butalbital combinations, ergots, NSAIDs, opioids, and opioid combinations. Preventive medication classes included oral migraine preventive medications (OMPMs, further categorized as antidepressants, antihypertensives, antiepileptics, and anticonvulsants), CGRP mAbs, and onabotulinumtoxinA.
Cost measures included all-cause and migraine-related total, medical, and prescription costs. Costs were calculated based on paid amounts of adjudicated claims, including plan payments and patient out-of-pocket costs in the form of copayments, deductibles, and coinsurance. In Optum’s CDM database, the paid amounts are converted to standardized costs with standardization across all facilities, providers, and regions based on detailed methodology available in the CDM data manual. All costs were reported in 2019 US dollars.
Statistical analysis
All analyses compared the study group of migraine patients with diagnosed MOH or potential AMO vs the control group of migraine patients without diagnosed MOH or potential AMO. Sample characteristics and unadjusted HCRU and cost outcomes were generated and compared between the study group and control group. Categorical variables were reported as counts and percentages of individuals in each category; continuous variables were reported as means and standard deviations.
Multivariate analyses were conducted to control for differences in demographic and clinical characteristics (age, sex, race, region, plan type, and Elixhauser comorbidity count (13)) between the two groups. Plan type included a variety of health plan types commonly used in the United States, such as point-of-service, preferred provider organization, and exclusive provider organization, that determine the providers that are in a given network. The Elixhauser comorbidity count is an established means of quantifying comorbidities that are distinct from the primary reason for hospitalization and determining their impact on the individual. Logistic regressions were estimated for binary outcomes of any inpatient visit and any ED visit for both all-cause and migraine-related measures. Given that the number of health care visits and health care cost data are often right skewed, adjusted analyses of these outcomes involved using multivariate generalized linear models with links selected by using Pearson correlation test, the Pregibon link test, the modified Hosmer and Lemeshow test, and families selected by using the modified Park test (14).
Adjusted estimates using the method of recycled predictions were produced for all outcomes, and comparisons were drawn between the two groups (14). Differences with a
Results
A total of 90,017 individuals met selection criteria for inclusion in the overall migraine group (Online Supplemental Table 1). Individually, MOH and potential AMO diagnosed using the primary ICHD-3 criteria were observed in 0.6% (N = 531) and 12.1% (N = 10,875), respectively, of individuals in the overall migraine group. The combined MOH/potential AMO migraine group comprised 12.6% (N = 11,300) of the overall migraine sample.
Most individuals in the MOH/potential AMO migraine group and non-MOH/potential AMO migraine control group were 45–64 years of age (64% and 42%, respectively) (Table 1). Both groups were predominantly female (83%) and White (70% and 63% in the combined and control groups, respectively). About three-quarters of individuals in both groups were in point-of-service health plans. The combined MOH/potential AMO migraine group had a higher prevalence of ≥4 Elixhauser comorbidities compared with the non-MOH/potential AMO migraine control group (Table 2). The most common migraine-related comorbidities in the combined MOH/potential AMO migraine group were other pain disorders (44.1%), mood disorders (42.7%), and sleep disorders (36.4%), all of which were prevalent at higher rates than in the non-MOH/potential AMO migraine control group (other pain disorders, 10.9%; mood disorders, 28.8%; sleep disorders, 22.3%). The combined MOH/potential AMO migraine group had a higher rate of chronic migraine diagnosis (36.2%) than the non-MOH/potential AMO migraine control group (23.5%).
Baseline characteristics among individuals with migraine with and those without MOH/potential AMO.
AMO, acute medication overuse; EPO, exclusive provider organization; HMO, health maintenance organization; MOH, medication overuse headache; POS, point of service; PPO, preferred provider organization.
Clinical characteristics among individuals with migraine with and those without MOH/potential AMO.
AMO, acute medication overuse; IBS, irritable bowel syndrome; MOH, medication overuse headache.
Greater than 5%.
Health care resource utilization
Unadjusted all-cause HCRU was higher in the combined MOH/potential AMO migraine group compared with the non-MOH/potential AMO migraine control group, with a larger proportion of individuals having any inpatient (20.2% vs 11.5%) or ED (43.8% vs 35.1%) visit, and higher mean number of E&M visits (14.3 vs 8.8) (Online Supplemental Figure 1). Similarly, unadjusted migraine-related HCRU was higher in the combined MOH/potential AMO migraine group compared with the non-MOH/potential AMO migraine control group, with a larger proportion of individuals having any migraine-related inpatient (13.6% vs 7.9%) or ED (22.7% vs 18.5%) visit, and higher mean number of migraine-related E&M visits (3.1 vs 2.1) (Online Supplemental Figure 2).
After performing a covariate-adjusted regression, differences in HCRU between the two groups were attenuated but, in most cases, significantly higher in the combined MOH/potential AMO migraine group compared with the non-MOH/potential AMO migraine control group. After adjusting for covariates, all-cause HCRU remained higher in the combined MOH/potential AMO migraine group compared with the non-MOH/potential AMO migraine control group for any inpatient visit (+1.8%,
Adjusted all-cause HCRU among individuals with migraine with and those without MOH/potential AMO. AMO, acute medication overuse; E&M, evaluation and management; ED, emergency department; HCRU, health care resource utilization; MOH, medication overuse headache. *
Adjusted migraine-related HCRU among individuals with migraine with and those without MOH/potential AMO. AMO, acute medication overuse; E&M, evaluation and management; ED, emergency department; HCRU, health care resource utilization; MOH, medication overuse headache. *
Online Supplemental Table 2 shows the descriptive results comparing migraine-related prescription use between groups. Acute medication use was higher in the combined MOH/potential AMO migraine group (98.6%) than the non-MOH/potential AMO migraine control group (70.6%). Individuals in the combined MOH/potential AMO migraine group had higher use of all acute medication classes than individuals in the control group. In particular, opioid use was substantially higher in the combined MOH/potential AMO migraine group (75.8%) than in the non-MOH/potential AMO migraine control group (27.1%). Preventive medication use also was higher in the combined MOH/potential AMO migraine group than in the non-MOH/potential AMO migraine control group. The MOH/potential AMO migraine group was more likely to use OMPMs, CGRP mAbs, onabotulinumtoxinA, antiepileptics, antihypertensives, and antidepressants than the control group.
Health care costs
Mean unadjusted all-cause annual total health care costs were $45,124 for individuals in the combined MOH/potential AMO migraine group and $19,819 for individuals in the non-MOH/potential AMO migraine control group. All-cause medical costs represented >80% of total costs, while all-cause prescription costs ranged from 16%-18% in both groups (Online Supplemental Figure 3). Unadjusted migraine-related mean total costs were $12,936 in the combined MOH/potential AMO migraine group and $5833 in the non-MOH/potential AMO migraine control group (Online Supplemental Figure 4). Outpatient service costs comprised more than half of the all-cause medical costs in both groups, followed by inpatient and all-cause prescription costs (Online Supplemental Table 3A). Among total migraine-related costs, inpatient and prescription costs were noticeably higher in the combined MOH/potential AMO migraine group (Online Supplemental Table 3B).
After adjusting for covariates in the regression model, the differences in annual health care costs between the two groups were attenuated but still substantially higher in the combined MOH/potential AMO migraine group compared with the non-MOH/potential AMO migraine control group. For instance, the differences in mean all-cause annual total, medical, and pharmacy costs were substantially higher in the combined MOH/potential AMO migraine group compared with the non-MOH/potential AMO migraine control group (+$9749, +$6746, and +$3107, respectively; all comparisons
Adjusted all-cause costs among individuals with migraine with and those without MOH/potential AMO. AMO, acute medication overuse; MOH, medication overuse headache. *
Adjusted migraine-related costs among individuals with migraine with and those without MOH/potential AMO. AMO, acute medication overuse; MOH, medication overuse headache. *
Discussion
To our knowledge, this claims-based study is the first attempt to identify MOH and potential AMO and quantify the associated excess direct health care costs among commercially insured patients with migraine in the United States. In addition to diagnosed MOH, the inclusion of potential AMO patients based on prescription acute migraine treatment fill patterns further strengthened our study by expanding the focus on patients at increased risk of chronic migraine as well as MOH. Our study has several important findings with potential implications for private payers and clinicians in the United States.
We found that approximately one in eight commercially insured patients with migraine had diagnosed MOH or had potential AMO. Although many acute migraine medications can cause MOH, barbiturates, opioids, and triptans carry the most risk and may lead to MOH or AMO in a shorter period of time (6–8,15). The ability of CGRP-targeted acute and preventive migraine treatments to reduce migraine days without increasing the risk for AMO suggests that the risk for experiencing MOH may be lower with these treatment options (16–18). Prior studies have shown high discontinuation rates of triptans and higher opioid use among individuals insufficiently responding to triptans (19,20). In our study in particular, patients with MOH/potential AMO were nearly three-fold more likely to use opioids than were those without MOH/potential AMO. These findings suggest that improvements in clinical management of patients with migraine are needed to reduce their risk of MOH and migraine progression. For these reasons, the use of opioids and barbiturates should be discouraged in migraine.
Improving migraine management is also highly relevant from the payer perspective given our finding that patients with diagnosed MOH/potential AMO had significantly higher all-cause and migraine-related HCRU and costs than those without MOH/potential AMO, even after adjusting for sociodemographic and clinical characteristics. Few studies have previously evaluated the cost and HCRU associated with MOH or potential AMO. Togha et al found that the mean annual cost of MOH among individuals in Iran was $2610, with the majority (55%) composed of indirect costs (21). A similar study by Raggi et al found high costs among individuals with MOH in Italy, with an annual cost of €10,553 composed mostly of indirect (51.5%) costs (22). The results were consistent in a similar study also conducted in Italy, which found that prior to treatment, the monthly cost for individuals with MOH was €950 and was attributable primarily to indirect costs (56%) (23). A cross-sectional US survey (Medication Overuse Treatment Strategy) found that individuals with more frequent headache and MOH had greater disability (Headache Impact Test-6), more symptoms of anxiety and depression (Generalized Anxiety Disorder 7 [GAD-7] and Patient Health Questionnaire 9 [PHQ-9]), and reduced quality of life (EuroQol EQ-5D) (24). Results from the Chronic Migraine Epidemiology and Outcomes (CaMEO) Study also demonstrated higher disability (Migraine Disability Assessment [MIDAS]), HCRU (emergency room and urgent care visits), anxiety (GAD-7) and depression (PHQ-9) among participants with medication overuse compared with those without (25). The COntinuous MOnitoring of Medication Overuse Headache in Europe and Latin America: development and STAndardization of an Alert and decision support System (COMOESTAS) project found that inexpensive and convenient access to acute medication was the leading cause of MOH (26). The COMOESTAS project also found that treatment of MOH was associated with a significant reduction in direct and indirect costs, as well as in disability (MIDAS), anxiety, and depression (Hospital Anxiety and Depression Scale) (27,28). However, several of these studies were conducted outside of the United States. Our study is among the first to examine the direct health care cost burden of MOH and/or potential AMO in a geographically diverse population covered by a large national commercial insurer in the United States.
Limitations of this study should also be noted. First, because the sample is limited to individuals with commercial insurance in the United States, these results may not be generalizable to individuals with other types of insurance (eg, Medicare, Medicaid), to those without health insurance, or to those living outside the United States. Second, as the study was limited to those with migraine diagnostic codes, individuals with other headache diagnostic codes, including headache not otherwise specified (NOS), were not included. Headache NOS is a widely used code for patients with migraine and its use is sometimes associated with choices of acute treatments (29,30). Third, as with all retrospective claims analyses, administrative claims data are subject to coding errors, which may result in potential misclassification of migraine/MOH status, covariates, and/or study outcomes. Since there are no specific laboratory tests that support the diagnosis of MOH, it is often underdiagnosed and under-coded in claims. Additionally, the ICD-10-CM code used to identify MOH is actually for identifying any drug-induced headache. Since the sample was limited to individuals with migraine, it was assumed that these individuals had migraine-related MOH. While we attempted to capture potential AMO in individuals who have been undiagnosed/uncoded by using prescription claims data, our algorithm may lead to misclassification bias. The use of prescription claims data is limited by the lack of information on whether individuals are actually taking their medications, especially for a medication taken as needed for migraine attacks. Hence, this analysis identifies individuals with AMO based on whether they possess a sufficient acute medication supply to meet an AMO case definition. Some individuals may exceed a medication possession threshold without actually taking the medication to excess. We assume that required co-payments would discourage the accumulation of medications that patients do not need. Classifying individuals based on medication possession rather than use may lead to inclusion of some individuals who do not overuse medication in the potential AMO group. This could attenuate between-group differences if overuse causes worsening of headache. In addition, over-the-counter (OTC) use of some medications (e.g., simple analgesics and caffeine combination products) cannot be assessed using claims data. As this study was conducted using 2019 data, it does not consider the impact of recently approved medications on the economic burden of MOH. Finally, there may be systematic differences other than MOH between the study and control groups that account for differences in health care costs across these groups. While multivariate analyses controlled for observed differences in demographic and clinical characteristics between study groups, we were unable to control for characteristics not measured in administrative claims. The reason for the limited number of individuals who are not White is unclear, but racial disparities have been observed in database analyses evaluating health care costs, in part to higher insurance coverage rates in the White population (31). It is therefore important to note that these results may not be applicable to other racial and ethnic groups, despite controlling for them in the statistical model.
Conclusion
Approximately one in eight individuals with migraine in a commercially insured population had diagnosed MOH or potential AMO. Among individuals with migraine, diagnosed MOH/potential AMO was associated with higher migraine-related HCRU and costs compared with individuals with migraine without evidence of MOH or potential AMO. These findings suggest that improved migraine management is needed to reduce the risk of MOH/AMO and associated costs in this population.
Key findings
This claims-based study of commercially insured migraine patients in the United States found that diagnosed medication overuse headache (MOH) and potential acute medication overuse (AMO) was associated with higher migraine-related health care resource utilization and costs compared with no MOH/AMO. These findings suggest that improved migraine management is needed to reduce the risk of MOH/AMO and their associated costs.
Supplemental Material
sj-pdf-1-cep-10.1177_03331024241235139 - Supplemental material for Health care resource utilization and costs associated with diagnosed medication overuse headache and potential acute medication overuse in individuals with migraine
Supplemental material, sj-pdf-1-cep-10.1177_03331024241235139 for Health care resource utilization and costs associated with diagnosed medication overuse headache and potential acute medication overuse in individuals with migraine by Anand R. Shewale, Jennifer A. Brandenburg, Kate Burslem, Richard B. Lipton and Jalpa A. Doshi in Cephalalgia
Footnotes
Acknowledgments
AbbVie and the authors thank all the trial participants for their participation in and dedication to this trial.
Author contributions
Study concept and design: A.R.S., J.A.B., J.A.D., R.B.L.
Collection and assembly of data: A.R.S., J.A.B.
Data analysis: J.A.B., R.B.L.
Data interpretation: All authors
Manuscript preparation: A.R.S., J.A.B., J.A.D.
Manuscript review and revisions: All authors
Final approval of manuscript: All authors
Data sharing statement
AbbVie is committed to responsible data sharing regarding the clinical trials we sponsor. This includes access to anonymized, individual, and trial-level data (analysis data sets), as well as other information (eg, protocols, clinical study reports, or analysis plans), as long as the trials are not part of an ongoing or planned regulatory submission. This includes requests for clinical trial data for unlicensed products and indications.
These clinical trial data can be requested by any qualified researchers who engage in rigorous, independent, scientific research, and will be provided following review and approval of a research proposal, Statistical Analysis Plan (SAP), and execution of a Data Sharing Agreement (DSA). Data requests can be submitted at any time after approval in the US and Europe and after acceptance of this manuscript for publication. The data will be accessible for 12 months, with possible extensions considered. For more information on the process or to submit a request, visit the following link: 
then select “Home.”
Declaration of conflicting interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: A.R.S. is an employee of AbbVie and may hold AbbVie stock. K.B. was an employee of AbbVie at the time of study conduct and may hold AbbVie stock. J.A.B. is a consultant for AbbVie. R.B.L. receives research support from the Marx Foundation, the National Headache Foundation, the National Institutes of Health, and the US Food and Drug Administration. He serves as a consultant or advisory board member for or has received honoraria or research support from AbbVie/Allergan, Amgen, Biohaven, Dr. Reddy’s Laboratories (Promius), electroCore, Eli Lilly, GlaxoSmithKline, Lundbeck, Merck, Novartis, Teva, Vector, and Vedanta Research. He receives royalties from Informa and
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and publication of this article: Allergan (prior to its acquisition by AbbVie) funded this study and contributed to the study design, the collection, analysis, and interpretation of data, and the review and approval of the final version for publication. No honoraria or payments were made for authorship. Medical writing support was provided to the authors by Dennis Stancavish, MA, of Peloton Advantage, LLC, an OPEN Health company, and was funded by AbbVie.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
