Societal burden of cluster headache in the United States: a descriptive economic analysis

Abstract

Aim: To estimate direct and indirect costs in patients with a diagnosis of cluster headache in the US.

Methods: Adult patients (18–64 years of age) enrolled in the Marketscan Commercial and Medicare Databases with ≥2 non-diagnostic outpatient (≥30 days apart between the two outpatient claims) or ≥1 inpatient diagnoses of cluster headache (ICD-9-CM code 339.00, 339.01, or 339.02) between January 1, 2009 and June 30, 2014, were included in the analyses. Patients had ≥6 months of continuous enrollment with medical and pharmacy coverage before and after the index date (first cluster headache diagnosis). Three outcomes were evaluated: (1) healthcare resource utilization, (2) direct healthcare costs, and (3) indirect costs associated with work days lost due to absenteeism and short-term disability. Direct costs included costs of all-cause and cluster headache-related outpatient, inpatient hospitalization, surgery, and pharmacy claims. Indirect costs were based on an average daily wage, which was estimated from the 2014 US Bureau of Labor Statistics and inflated to 2015 dollars.

Results: There were 9,328 patients with cluster headache claims included in the analysis. Cluster headache-related total direct costs (mean [standard deviation]) were $3,132 [$13,396] per patient per year (PPPY), accounting for 17.8% of the all-cause total direct cost. Cluster headache-related inpatient hospitalizations ($1,604) and pharmacy ($809) together ($2,413) contributed over 75% of the cluster headache-related direct healthcare cost. There were three sub-groups of patients with claims associated with indirect costs that included absenteeism, short-term disability, and absenteeism + short-term disability. Indirect costs PPPY were $4,928 [$4,860] for absenteeism, $803 [$2,621] for short-term disability, and $3,374 [$3,198] for absenteeism + disability.

Conclusion: Patients with cluster headache have high healthcare costs that are associated with inpatient admissions and pharmacy fulfillments, and high indirect costs associated with absenteeism and short-term disability.

Keywords:

• Direct and indirect cost analysis
• cluster headache
• claims
• absenteeism
• short-term disability
• healthcare resource utilization

Introduction

Cluster headache is a relatively rare neurological disease known to affect ∼1 in 1,000 people in the US¹. Cluster headache is characterized by short-lasting attacks (15–180 min) of severe to excruciating unilateral orbital, supraorbital, and/or temporal pain, that are accompanied by cranial autonomic symptoms such as tearing, conjunctival injection, rhinorrhea/nasal congestion ipsilateral to the pain, and restlessness². Multiple cluster headache attacks may occur daily or every other day with circadian periodicity. Bouts can last for weeks or months and are separated by pain-free remission periods².

Although cluster headache attacks are periodic, the personal burden can be substantial, due to lifestyle restrictions during bouts, increased healthcare utilization, and negative impact on work^3,4. A European study showed that the average direct and indirect costs of cluster headache in 2010 were €5,963 per person during a 6-month period, and direct costs and burden of disease were significantly correlated with attack frequency⁵. In the US, the Cluster Headache Survey was the largest survey ever completed by patients with cluster headache. From this survey, conducted in 2008, Rozen and Fishman³ reported that almost 20% of the patients had lost a job due to cluster headache, while another 8% were out of work or on disability. However, information on the socioeconomic impact of this disease in the US has yet to be described.

Herein, we report for the first time the direct and indirect costs associated with a diagnosis of cluster headache in the US, based on an examination of health insurance claims data. The results of this study contribute much needed information on the socioeconomic burden of this disease for patients with cluster headache.

Methods

This retrospective observational study utilized US insurance claims data to assess direct and indirect costs for patients with cluster headache. All database records were de-identified and fully compliant with US patient confidentiality requirements, including the Health Insurance Portability and Accountability Act of 1996. Because the study used only de-identified patient records and did not involve the collection, use, or transmittal of individually identifiable data, Institutional Review Board approval to conduct this study was not necessary.

The study samples were identified from two Truven Health Analytics MarketScan Research Databases that contain data from large employers and managed care organizations: the Commercial Database and the Health and Productivity Management (HPM) database (Ann Arbor, MI). The claims in the Commercial Database provide detailed cost and utilization data for healthcare services performed in both inpatient and outpatient settings. Insurance coverage is provided under a variety of fee-for-service and capitated health plans, including exclusive provider organizations, preferred provider organizations, point-of-service plans, indemnity plans, and health maintenance organizations. It includes all pharmacy fills with positive health plan payment or patient co-payment. Written prescriptions that were not filled were not included in the claims database. The HPM Database includes detailed information on workplace absences, short-term disability, and workers’ compensation for a sub-set of individuals in the larger Commercial Database.

The productivity and medical claims data are linked to outpatient drug claims and person-level enrollment data. When a prescription was filled, the associated demographic, physician specialties consulted, insurance plan type, prescription charges, co-payments, and reimbursed payments were available in the database.

Physician specialty assignments associated with prescription claims were made based on the physician specialty of the prescriber, which included all-cause and cluster headache-related. If there was no physician specialty on the index date, then the physician specialty associated with office visits within 45 days prior to the index date was used. When there were multiple providers, the following hierarchy was applied: primary care provider→neurologist→other specialist. In other words, if a patient saw both a primary care provider and neurologist, then the patient was assigned to the primary care provider group.

Claims for pharmaceutical treatment of cluster headache were identified as those medications that were aligned with the European Federation of the Neurological Societies (EFNS) guidelines⁶, which were the only guidelines available at the time this study was conducted. Medication for acute treatment of cluster headache included sumatriptan subcutaneous (6 mg) or nasal (20 mg), zolmitriptan nasal (5 mg and 10 mg) or oral (5 mg and 10 mg), lidocaine intranasal, and octreotide subcutaneous. Preventive medications included verapamil, steroids (prednisone, dexamenthasone), lithium carbonate, methysergide, topiramate, ergotamine tartrate, valproic acid, melatonin, and baclofen. Cluster headache-related procedures included occipital nerve block, percutaneous radiofrequency ablation, trigeminal gangliorhyzolysis, deep brain stimulation, sphenopalatine ganglion stimulation, and hyperbaric oxygen therapy.

Comorbidities were identified by the presence of at least one inpatient or non-diagnostic outpatient medical claim with an International Classification of Diseases, Ninth revision, Clinical Modification (ICD-9-CM) diagnosis code.

Patient selection

To be included in the analysis, patients were to have at least two non-diagnostic outpatient claims at least 30 days apart with ICD-9-CM diagnosis code for cluster headache (ICD-9-CM code 339.00, 339.01, or 339.02) between January 1, 2009 and June 30, 2014, or at least one inpatient admission with a cluster headache diagnosis code in any position during the same time period; ≥6 months of continuous enrollment with medical and pharmacy coverage before the index date; age between 18 and 64 years on the index date; and ≥6 months of continuous enrollment with medical and pharmacy coverage in the post-index period. For each patient, the index date was set to the date of the first qualifying cluster headache diagnosis during the selection window between January 1, 2009 and June 30, 2014.

Outcomes

Three types of outcomes were evaluated: (1) healthcare resource utilization, (2) costs associated with healthcare resource utilization including pharmacy costs, and (3) indirect costs associated with work days lost due to absenteeism and short-term disability. For each of these outcomes, we included all-cause claims, as well as claims specifically associated with the cluster headache diagnosis (ICD-9-CM 339.00, 339.01, or 339.02). Pharmacy costs were analyzed a priori as any pharmacy cost and were not further evaluated as acute or preventive medications. All-cause indirect costs were estimated for work time loss for sub-sets of patients who met eligibility criteria for the indirect cost analysis. Eligible patients were those with ≥6 months of eligibility in the absenteeism and/or short-term disability database before and after the index date, were active full-time employees, and had no evidence of pregnancy or giving birth during the entire study period. Indirect costs per patient per year (PPPY) were estimated from the cost associated with absenteeism and short-term disability that was based on an average daily wage, which was estimated from the 2014 US Bureau of Labor Statistics and inflated to 2015 dollars using the Medical Care Component of the Consumer Price Index. To adjust for the variable length of the post-index periods across claims, the total amount of an individual’s claim was divided by the length of follow-up and multiplied by 365 (days in a year) to get the estimated cost per year.

Statistical analysis

Descriptive statistics for demographic and clinical characteristics, healthcare resource utilization, direct healthcare costs, and work loss and indirect costs are presented as the number of observations and percentage or as means ± standard deviation (SD). The statistical analysis software program used for this study was SAS version 9.4.

Results

Patient attrition is summarized in Table 1. A total of 18,303 patients with a diagnosis of cluster headache were identified in the database, of which 9,328 patients met criteria for direct cost analysis. The indirect cost analyses included three sub-groups of patients with HPM data. There were 166 patients in the sub-group for absenteeism, 844 in the short-term disability sub-group, and 139 in the sub-group of patients with both absenteeism and short-term disability records.

Table 1. Patient selection for direct and indirect cost analyses.

	n
Selection criteria for direct cost analysis
At least two pre-index^a outpatient claims with ICD-9-CM code^b for cluster headache at least 30 days apart between January 1, 2009 and June 30, 2014, OR	16,615
At least one pre-index inpatient admission claim with ICD-9-CM code for cluster headache between January 1, 2009 and June 30, 2014	2,194
Total number of patients with evidence of cluster headache between January 1, 2009 and June 30, 2014	18,303
AND ≥6 months of continuous enrollment pre-index with medical and pharmacy coverage,	10,663
AND Age between 18 and 64 years on the index date,	10,337
AND ≥6 months of continuous enrollment post-index^c with medical and pharmacy coverage	9,328
Selection criteria for indirect cost analysis
Eligible for cluster headache group	9,328
AND Eligible for ABS or STD benefit in the HPM Database for at least 6 months before and after the index date	943
AND Active full-time employees from 6 months for at least 6 months after the index date	886
AND not pregnant or giving birth during follow-up	871
Absenteeism sub-group	166
Short-term disability sub-group	844
Both absenteeism and short-term disability sub-group	139

^a Pre-index was prior to the index date, which was when the first diagnosis for cluster headache was made.

^b ICD-9-CM code for cluster headache: 339.00, 339.01, or 339.02.

^c Post-index is the period following the index date.

The mean (standard deviation) age of the patients was 44.6 (11.1) years. Most of the patients were male (61.4%, n = 5,727; male-to-female ratio = 1.6), 84.0% (n = 7,831) were from urban areas, and 36.6% were from the southern region of the US. Health plan types were known for 95% of the patients and included comprehensive/indemnity (1.6%, n = 152), exclusive provider organization/preferred provider organization (63.2%, n = 5,891), point-of-service with or without capitation (9.3, n = 863), health maintenance organization (15.9%, n = 1,486), and consumer-driven health plan/highly deductible health plan (5.1%, n = 4,750). Comorbid diseases of interest with >5% frequency included migraine (28.9%, n = 2,693), hypertension (18.0%, n = 1,683), hyperlipidemia (14.2%, n = 1,327), depression/suicide/self-harm (9.3%, n = 871), sleep disorders (9.2%, n = 857), anxiety (7.9%, n = 739), and chronic pulmonary disease (6.9%, n = 647).

Most of the patients received prescriptions (all-cause and cluster headache-related) from primary care physicians (60.6%, n = 5,657), 18.6% (n = 1,738) were from neurologists, 6.3% were from an “other specialist” (n = 590), and 14.4% (n = 1,343) were from providers whose specialty was not identified. Pharmaceutical treatments for cluster headache were prescribed for 80.0% (n = 7,465) of the patients, and included steroids (64%, n = 5,996), triptans (27.5%, n = 2,562), verapamil (22.7%, n = 2,113), anticonvulsants (22.5%, n = 2,096), oxygen therapy (7.3%, n = 679), and lithium carbonate (4.1%, n = 379).

All-cause and cluster-headache-related healthcare resource utilization is summarized in Table 2. Almost a quarter (23%) of patients had a hospitalization, and about a third of those admissions had cluster headache as the primary diagnosis at discharge. The majority (>90%) of patients had healthcare provider office visits, pharmacy fulfillments, laboratory tests, and other outpatient services. Over 80% had cluster headache-related office visits and pharmacy fulfillments. In addition, 21.8% of the patients had a cluster headache-related procedure.

Table 2. Healthcare utilization by patients (n = 9,328) that was all-cause and cluster headache-related.

Resource use	All cause^a	Cluster headache-related^b
Inpatient admission, n (%)	2,170 (23.3)	1,043 (11.2)
Emergency room visit, n (%)	4,346 (46.6)	984 (10.5)
Healthcare provider office visit, n (%)	9,269 (99.4)	7,602 (81.5)
Laboratory test, n (%)	8,527 (91.4)	1,855 (19.9)
Hospital-based outpatient visits, n (%)	6,851 (73.4)	1,562 (16.7)
Other outpatient services, n (%)	8,975 (96.2)	3,218 (34.5)
Pharmacy fulfillment, n (%)	9,055 (97.1)	8,143 (87.3)
Cluster headache-related procedures, n (%)	–	2,034 (21.8)

^a Included all claims, regardless of ICD-9-CM coding.

^b Included claims with ICD-9-CM code for cluster headache: 339.00, 339.01, or 339.02.

All-cause and cluster headache-related direct costs of healthcare resource utilization are summarized in Table 3. The all-cause cost was $17,574 PPPY, of which $3,132 (17.8%) was cluster headache-related. The largest contributors to all-cause cost included inpatient admissions ($5,201), hospital outpatient visits ($3,631), pharmacy prescriptions ($3,265), other outpatient services ($2,947), and outpatient office visits ($1,071). For cluster headache-related costs, inpatient hospitalizations ($1,604) and pharmacy fulfillments ($809) together ($2,413) contributed to over 75% of the direct cost of healthcare.

Table 3. Direct cost of healthcare per patient per year (n = 9,328).

Healthcare resource	All cause^a $, Mean (SD)	Cluster headache-related^b $, Mean (SD)
Inpatient admission	5,201 (25,970)	1,604 (13,004)
Emergency room	808 (3,115)	88 (520)
Healthcare provider office visit	1,071 (1,055)	201 (255)
Laboratory test	652 (1,886)	25 (160)
Hospital outpatient visit	3,631 (10,926)	193 (1,165)
Other outpatient services	2,947 (14,902)	95 (1,009)
Pharmacy fulfillment	3,265 (7,302)	809 (2,226)
Cluster headache procedure	–	114 (377)
Total cost	17,574 (40,970)	3,132 (13,396)

^a Included only claims with any ICD-9-CM code including cluster headache diagnosis (339.00, 339.01, or 339.02).

^b Included only claims with ICD-9-CM code for cluster headache: 339.00, 339.01, or 339.02.

Indirect cost assessment is summarized in Table 4. The percentages of patients in the sub-groups with work hours lost were 78% due to absenteeism, 23% for short-term disability, and 81% for absenteeism + short-term disability. Hours lost PPPY were 224 due to absenteeism, 61 for short-term disability, and 256 for both absenteeism + short-term disability. The estimated indirect cost PPPY was $4,928 ($4,860) for absenteeism, $803 ($2,621) for short-term disability, and $3,374 ($3,198) for absenteeism + disability.

Table 4. All-cause indirect costs for patients with absenteeism and short-term disability claims.

	Absenteeism (n = 166)	Short-term disability (n = 844)	Absenteeism + short-term disability (n = 139)
Number of patients with work hours lost, n (%)	129 (77.7)	194 (23.0)	113 (81.3)
Number of hours lost, PPPY, mean (SD)	224.0 (220.9)	60.8 (198.6)	255.6 (242.2)
Indirect costs PPPY, mean (SD)	$4,928 ($4,860)	$803 ($2,621)	$3,374 ($3,198)

Abbreviation. PPPY, per patient per year.

Discussion

This is the first claims-based study in the US to report the economic burden (both direct and indirect costs) for patients with a diagnosis of cluster headache. Consistent with prevalence studies, most of the patients were male, but the male-to-female ratio of 1.6 was lower than that reported in population-based epidemiological studies¹. The lower ratio in the current study most likely reflects the demographic makeup of a claims database as compared to the broader population examined by epidemiology that is not restricted to patients seeking medical care. However, longitudinal clinical observations of patients with cluster headache have revealed that the ratio of males to females have decreased from the 1970s to 1990s⁷, and it has been suggested that changes in lifestyle factors, such as employment and smoking habits, may have played a role in these changes⁸, as well as gender differences at the onset of cluster headache across the lifespan⁹.

Medical comorbidities contribute to the burden of illness in patients with cluster headache, their healthcare resource use, and work absenteeism. Comorbidities previously reported as common for patients with cluster headache include hypertension^10,11, anxiety and/or depression^3,11,12, suicidality³, and sleep disorders^3,13. In addition to these comorbidities, patients in the current study also frequently had migraine and chronic pulmonary disease. Comorbid migraine has not been reported very often other than on a case-by-case basis¹¹. Most recently, Gaul et al.⁵ reported that 2% of the patients from a sample of cluster headache sufferers in a headache specialty center in Germany had comorbid migraine. However, there are others who have reported the presence of migrainous symptoms that occurred during cluster attacks^3,11,14–16.

The cost of cluster headache is just beginning to be realized. In this study, all-cause direct cost of healthcare for patients with cluster headache was $17,574 PPPY, of which 17.8% ($3,132) was attributed to cluster headache-related expenditures. Inpatient admissions, together with pharmacy prescriptions, contributed to most (>75%) of the cluster headache-related healthcare cost. One other economic analysis of cluster headache reported that direct healthcare costs in Germany were predominantly the result of expenditures for medications to treat attacks⁵. There were few inpatient admissions in the German study, so this expenditure did not contribute as much to the direct healthcare costs as it did in the current study, which reported that 11% of the patients had cluster headache-related inpatient admissions. The higher rate of inpatient admission in the current study may be due to fewer patients who were seeking care from a neurologist (18.6%); whereas all of the patients in the German study were seeking care at a neurology specialty center.

In comparison to cluster headache, a recent economic analysis of migraine headache, which is another severe neurological disease, reported direct healthcare costs that were $4,943 PPPY for chronic migraine and were $1,705 PPPY for episodic migraine¹⁷. This study also reported that a large proportion of the direct healthcare cost was attributed to pharmaceutical use among episodic and chronic migraineurs.

Indirect costs specifically related to cluster headache could not be determined because reasons for workplace absence or short-term disability were not available, so only all-cause indirect costs could be estimated for sub-groups of patients with data available for absenteeism and short-term disability. Among patients with absenteeism data, indirect costs for days claimed as missed from work were $4,928 PPPY. For patients with short-term disability data, indirect costs were $803 PPPY, and for those with both absenteeism and short-term disability data, indirect costs were $3,374 PPPY. In comparison, migraine annual indirect costs (absenteeism and presenteeism) were $3,300 PPPY for chronic migraine and were $943 PPPY for episodic migraine¹⁷.

There are several important limitations that must be considered when interpreting these results. The data included in this study were from a claims database, which limits generalization to cluster headache patients without employer-based insurance or those without health insurance coverage. Because patients were identified by administrative claims as opposed to medical records, there was the potential for misclassification of migraine as cluster headache, especially considering that nearly one-third of the patients in this study were reported to have comorbid chronic migraine. In addition, cluster headache could have been misclassified as chronic migraine, which may explain the high percentage of patients with comorbid chronic migraine. Also, even though ICD-9 codes are available for differentiation between episodic and chronic forms of the disease, ICD-9-CM 339.00 (cluster headache unspecified) is commonly selected as the billing code. Administrative claims data can be subject to data coding limitations and data entry error. Indirect cost estimations were limited by the diversity of occupations and wide range of salaries across levels of responsibility (entry level to upper management), and by sample size restrictions for absenteeism and short-term disability. In addition, estimating the cost of pharmaceutical treatment of cluster headache based on medications listed the 2006 EFNS guidelines potentially would not have captured any claims during January 2009–June 2014 for new cluster headache treatments or more recent recommendations based on recently published guidelines from the American Headache Society¹⁸. Furthermore, it would be of interest to know the impact of the socio-demographic characteristics, insurance plan coverage, co-payment arrangement, and comorbidities on the burden of cluster headache, which were beyond the scope of the present study.

Conclusions

In this real-world analysis, patients with cluster headache have high healthcare costs that are associated with inpatient admissions and pharmacy fulfillments, and high indirect costs associated with absenteeism and short-term disability.

Transparency

Declaration of funding

The study sponsor (Eli Lilly and Company, Indianapolis, IN) funded the study and was involved in the study design, data interpretation, and writing of the article. Truven Health Analytics, an IBM Company, was involved in the study design, data collection, data analysis, data interpretation, and writing of the article. All authors had full access to all data and had final responsibility for the decision to submit for publication. All authors met the ICMJE authorship criteria. Neither honoraria nor payments were made for authorship.

Declaration of other/financial relationships

JHF, JA, and JMM are employees of Eli Lilly and Company, Indianapolis, IN. GK and RF are employees of Truven Health Analytics, an IBM Company, Ann Arbor, MI. DN and BCC were employed at Truven Health Analytics at the time that the study was conducted. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.