Direct and indirect healthcare resource utilization and costs associated with ulcerative colitis in a privately-insured employed population in the US

Abstract

Objective:

To assess direct and indirect healthcare resource utilization and costs of privately insured US employees with ulcerative colitis (UC) from a societal perspective.

Research design and methods:

Employees aged 18–64 with ≥2 UC diagnoses and no more than one diagnosis of Crohn’s disease (CD) were identified from a large, de-identified, private insurance US claims database from January 1, 2005 through March 31, 2013. Patients with UC were matched 1:1 to non-IBD controls based on demographics and index date (a randomly selected UC diagnosis). All patients were required to have continuous eligibility for ≥1 year before (baseline period) and after (study period) the index date. Descriptive analyses compared baseline characteristics and study period outcomes. Multivariate cost analysis adjusted for baseline comorbidities. Sub-group analyses compared patients with moderate-to-severe UC with matched controls.

Main outcome measures:

Costs (2013 US dollars) were measured from a societal perspective, which included direct (patient and payer costs) and indirect (lost wages because of time away from work) costs.

Results:

Patients with UC (n = 4314; mean age = 45.1 years, 63.6% male) had significantly higher baseline comorbidity rates compared with controls. In the study period, significantly more patients with UC (p < 0.0001) had higher hospitalization rates (16.9% vs 6.2%), emergency department visits (31.1% vs 22.0%), prescription drug use (95.3% vs 72.0%), and work loss (100% vs 81.4%). Patients with UC also had significantly higher adjusted total direct ($15,548 vs $4812) and indirect costs ($4125 vs $1961). Patients with moderate-to-severe UC (n = 1728) had significantly (p < 0.0001) higher hospitalization rates (26.5% vs 6.2%) and adjusted total direct ($23,085 vs $4932) and indirect costs ($5666 vs $1960).

Conclusions:

Patients with UC had higher resource utilization and direct and indirect costs compared with matched controls. The excess burden was greatest in patients with moderate-to-severe UC.

Keywords::

• Colitis
• Ulcerative
• Economics
• Cost of illness
• Healthcare costs
• Health expenditures
• Absenteeism

Introduction

Ulcerative colitis (UC) is a chronic, idiopathic, inflammatory disease that primarily involves the mucosal layer and occasionally the submucosal layer of the colon. Symptoms include bloody diarrhea, abdominal cramping during bowel movements, and abdominal pain1. Partly because of the lack of standard criteria for diagnosis, inconsistent case ascertainment, disease awareness, and disease misclassification, the estimated incidence rates for UC vary greatly, from 0.5–24.5 cases per 100,000 persons worldwide2. A recent epidemiologic study3 revealed that the prevalence of UC may be increasing in the US, with an estimation of 263 adult UC patients per 100,000 patients in 2009.

The aim of medical treatment in UC is to achieve remission. Anti-inflammatory agents such as 5-aminosalicyclic acid derivatives, corticosteroids, and the immunomodulatory agent cyclosporine are often used to induce remission4–17. Azathioprine and 6-mercaptopurine are used to maintain remission4^,18–22. In addition, anti-tumor necrosis factor biologics (e.g., adalimumab, infliximab, and golimumab) are used both to induce clinical remission and for long-term therapy in adult patients with moderately-to-severely active UC who had an inadequate response to conventional therapy21–23. A recent study showed that, after initiation of immunological therapy, the percentage of patients with UC experiencing severe symptoms fell from 66% at diagnosis to 15% at 1-year follow-up and the percentage in remission rose from 11% at diagnosis to 72% at 1-year follow-up24.

Although pharmacotherapy can be used to control inflammation and reduce symptoms, many patients with UC eventually require surgery. A recent review of population-based studies found that the risk of intestinal surgery 1, 5, and 10 years after a diagnosis of UC was 4.9%, 11.6%, and 15.6%, respectively25. Surgery is associated with several short-term and long-term complications that could significantly impair patients’ quality-of-life, including sepsis, small bowel obstruction, alterations in urinary function, failure of perineal wound healing, sexual dysfunction, and infertility in women26^,27.

In addition to the clinical burden of UC, the cost and healthcare resource utilization associated with UC can be substantial. The lost time at work, hospitalization, and other direct and indirect costs to society contribute to the economic burden of UC. A 2010 systematic review reported that the annual direct medical costs of UC ranged from $6217–$11,477 per patient (in 2008 US currency) in the US28. Hospitalizations accounted for 41–55% of direct medical costs, and the total economic burden was estimated at $8.1–14.9 billion annually (in 2008 US currency)28.

The objective of this study was to assess the disease burden in employed UC patients from a societal perspective. Specifically, this study compared direct and indirect costs, resource utilization, and work loss between employed adult patients with UC and matched controls without UC. Additionally, sub-group analyses were conducted in employed UC patients with moderate-to-severe disease severity. Because the treatment options for UC have recently expanded with the introduction of anti-tumor necrosis factor agents, this study provides an updated estimate of the disease burden associated with UC.

Patients and methods

Data source

This retrospective, observational study utilized data from 41 US-based large companies providing short- and long-term disability data in the OptumHealth Reporting and Insights claims database from January 1, 2005 through March 31, 2013. The claims database included direct costs, healthcare resource use, and work loss information for over 4.2 million privately insured employees. These 41 companies had operations nationwide in a broad array of job classifications and industries (e.g., financial services, manufacturing, telecommunications, energy, and food and beverage). The database contained de-identified information on patient demographics (age and sex); monthly enrollment status; and medical, pharmacy, and disability claims. Utilization of medical services was determined from dates of services, associated diagnoses (up to two codes, using the International Classification of Diseases, Ninth Revision, ICD-9)29, medical procedures performed (using the Current Procedural Terminology), and total payments. The database also included pharmacy claims with prescribed medications that were identified by the National Drug Codes30, dates of prescription fills, number of days the medication was supplied, quantity of medication supplied, and total payment amounts. Dates of work loss and wages were obtained from short- and long-term disability claims.

Sample selection

Employees with UC

The UC patient sample was drawn from employees aged 18–64 years who were the primary policy holders and had at least two UC diagnoses (ICD-9-CM: 556.x) from January 1, 2005 through March 31, 2013. To capture an overall snapshot of UC patients with varying levels of disease severity, the index date for UC patients was the date of a randomly chosen UC diagnosis with at least 1 year of continuous eligibility before and at least 1 year after the diagnosis. The 1-year period before the diagnosis was defined as the baseline period, and the 1-year period after the index date was defined as the study period. Patients were required to be employed for the duration of the baseline period and the study period. UC patients with two or more Crohn’s disease (CD) diagnoses (ICD-9-CM: 555.x) during the 1-year baseline and the 1-year study period were excluded from the UC patient sample.

A sub-group of patients with moderate-to-severe UC was identified from the overall UC group. Moderate-to-severe UC was defined as a hospitalization with a primary diagnosis of UC or treatment with biologics, immunosuppressants, or systemic corticosteroids31 during the 1-year study period (Supplemental Table 1).

Employee controls without UC

Potential controls were identified among employees aged 18–64 years who had no UC or CD diagnoses in their claims histories from January 1, 2005 through March 31, 2013. An employee control with the same age ±1 year, sex, and region was randomly matched without replacement to each eligible employee with UC so that every identified employee with UC had his or her own unique control. Each control was assigned the index date of the UC patient to whom he or she was matched. Controls were also required to have continuous eligibility for at least 1 year before and 1 year after the assigned index date, to match the distribution of index dates among employees with UC.

Patient characteristics

Patient characteristics included those used to match patients (i.e., age, sex, and region), baseline conditions, and a disease severity indicator. Baseline conditions included diseases commonly seen in patients with UC, such as autoimmune diseases, back pain, chronic pulmonary diseases, congestive heart failure, diabetes, hypertension, intestinal disorders, malignancy, non-infectious gastroenteritis and colitis, osteoarthritis, respiratory or other chest symptoms, and symptoms of the abdomen and pelvis. Baseline comorbidity burden was determined using the Charlson Comorbidity Index (CCI), an index of 17 physical conditions predictive of 1-year mortality32^,33.

Direct and indirect resource use

Direct healthcare resource use during the 1-year study period was categorized by place or type of service. Inpatient visits were identified by claims within the inpatient setting, whereas emergency department visits were identified by claims in the emergency department setting. All other visits (e.g., outpatient surgery, physician services, laboratory, and other ancillary services, etc.) were categorized as outpatient/other visits. Resource use was also reported for prescription drugs. Indirect resource use included work loss due to disability and estimated medically-related absenteeism. Days of disability were computed by identifying the total time covered by short- or long-term disability claims.

Medically-related absenteeism days (e.g., an office visit or a hospitalization event) were estimated using medical claims that occurred during weekdays. In particular, outpatient/other visits were counted as a half-day of absenteeism, but each hospitalization day and emergency department visit was counted as a full day of work loss. Medically-related absenteeism days did not include days with medical services occurring during a period of disability, but did include the qualifying days missed from work before the start of the disability period, excluding any overlap (e.g., if an employee had a short-term disability claim and the employer plan specified that short-term disability begins after 5 work days of illness, then 5 days of medically-related absenteeism were added, excluding any days already included in the calculation of medically-related absenteeism).

Direct and indirect cost calculation

Direct healthcare costs, including medical and prescription drug costs, were calculated during the 1-year study period. Direct cost estimation was calculated from a societal perspective and included what the patient paid (i.e., co-payment), as well as the amount covered by both primary and secondary insurance. Medical costs were calculated for inpatient, emergency department, and outpatient/other medical services.

Indirect cost estimation was calculated from a societal perspective using the employee’s wage information rather than the employer’s disability payments. The cost of work loss days was calculated on the basis of individual employee wage information obtained from eligibility files, and was inflated to 2013 US dollars using average hourly compensation data from the Bureau of Labor Statistics34.

All costs were indexed to 2013 US dollars, the most recent year of available medical and pharmacy claims, using the Consumer Price Index for Medical Care34.

Statistical analyses

Patient characteristics of employees with UC were compared with those of matched employees without UC. Categorical variables were compared using McNemar tests. Continuous variables, such as the CCI, resource use (e.g., number of inpatient visits, work loss days), and costs that likely had skewed distributions, were compared using non-parametric Wilcoxon signed rank tests between cohorts.

Multivariate generalized linear models were conducted with a gamma distribution of the error term and log link function to adjust the dollar amount of cost estimation when controlling for the CCI and medical conditions, excluding gastro-related conditions such as intestinal disorders, non-infectious gastroenteritis and colitis, and symptoms of the abdomen and pelvis, that differed significantly at p < 0.05.

All resource use and cost analyses were repeated for the moderate-to-severe disease severity sub-group. All analyses were conducted using SAS version 9.3 (SAS Institute Inc., Cary, NC). p-values less than or equal to 0.05 were considered statistically significant differences.

Results

Baseline characteristics and medical conditions

Upon implementation of the sample selection criteria, a total of 4314 employees with UC and 4314 matched controls were identified (Figure 1). Employees in the overall study sample were on average 45.1 years old and were mostly men (63.6%). With the exception of congestive heart failure (p > 0.99) and diabetes (p = 0.12), employees with UC had significantly higher rates (p < 0.05) of baseline medical conditions, including autoimmune diseases, back pain, chronic pulmonary diseases, hypertension, intestinal disorders, malignancy, non-infectious gastroenteritis and colitis, osteoarthritis, respiratory or other chest symptoms, and symptoms of the abdomen and pelvis compared to matched controls. Employees with UC also had a greater baseline CCI than employee controls, indicating the presence of a greater comorbidity burden (Table 1).

Figure 1. Sample selection process. CD, Crohn’s disease; UC, ulcerative colitis.

Table 1. Baseline characteristics of study population.

	All UC patients (n = 4314)	Controls (n = 4314)	p-value^a	Moderate-to-severe UC patients (n = 1728)	Controls (n = 1728)	p-value^a
Demographic characteristics
Age, years, mean ± SD	45.1 ± 10.1	45.1 ± 10.1	–	44.8 ± 10.1	44.8 ± 10.1	–
Age distribution, n (%)			–			–
18–30 years	484 (11.2)	481 (11.1)		194 (11.2)	194 (11.2)
31–44 years	1486 (34.4)	1476 (34.2)		627 (36.3)	626 (36.2)
45–54 years	1571 (36.4)	1592 (36.9)		609 (35.2)	607 (35.1)
55–64 years	773 (17.9)	765 (17.7)		298 (17.2)	301 (17.4)
Men, n (%)	2745 (63.6)	2745 (63.6)	–	1072 (62.0)	1072 (62.0)	–
Baseline conditions, n (%)
Autoimmune diseases^b	127 (2.9)	63 (1.5)	<0.0001	72 (4.2)	29 (1.7)	<0.0001
Back pain^c	995 (23.1)	767 (17.8)	<0.0001	433 (25.1)	311 (18.0)	<0.0001
Chronic pulmonary diseases^d	294 (6.8)	191 (4.4)	<0.0001	163 (9.4)	90 (5.2)	<0.0001
Congestive heart failure	17 (0.4)	17 (0.4)	>0.99	8 (0.5)	8 (0.5)	>0.99
Diabetes, type I and II	250 (5.8)	217 (5.0)	0.12	107 (6.2)	83 (4.8)	0.07
Hypertension	800 (18.5)	711 (16.5)	0.01	362 (20.9)	261 (15.1)	<0.0001
Intestinal disorders^e	3084 (71.5)	206 (4.8)	<0.0001	1369 (79.2)	89 (5.2)	<0.0001
Malignancy^f	162 (3.8)	82 (1.9)	<0.0001	67 (3.9)	34 (2.0)	0.0009
Non-infectious gastroenteritis and colitis	1076 (24.9)	50 (1.2)	<0.0001	547 (31.7)	19 (1.1)	<0.0001
Osteoarthritis	201 (4.7)	154 (3.6)	0.01	97 (5.6)	56 (3.2)	0.0007
Respiratory or other chest symptoms	768 (17.8)	519 (12.0)	<0.0001	388 (22.5)	213 (12.3)	<0.0001
Symptoms of the abdomen and pelvis	1093 (25.3)	318 (7.4)	<0.0001	553 (32.0)	138 (8.0)	<0.0001
Charlson Comorbidity Index, mean (SD)	0.37 ± 1.0	0.20 ± 0.7	<0.0001	0.45 ± 1.1	0.21 ± 0.7	<0.0001

SD, standard deviation; UC, ulcerative colitis.

^aMcNemar tests and Wilcoxon signed rank tests were used for comparisons of categorical and continuous variables, respectively.

^bAutoimmune diseases included ankylosing spondylitis, multiple sclerosis, myasthenia gravis, psoriasis, and rheumatoid arthritis.

^cBack pain included dorsopathies, unspecified neuralgia neuritis and radiculitis, curvature of spine, acquired spondylolisthesis, acquired deformity of back or spine, non-allopathic lesions, spina bifida, other specified congenital anomalies of the spinal cord, sprains and strains of the sacroiliac region, and sprains and strains of other and unspecified parts of the back.

^dChronic pulmonary diseases included bronchitis, emphysema, asthma, bronchiectasis, extrinsic allergic alveolitis, chronic airway obstruction, asbestosis, pneumoconiosis, and chronic respiratory conditions due to fumes and vapors.

^eIntestinal disorders included anal and rectal polyp, rectal prolapse, stenosis of the rectum and anus, abscess of the intestine, colostomy and enterostomy complications, and other disorders of the anus, rectum, and intestine.

^fMalignancy included malignant neoplasms of the lip, oral cavity, pharynx, digestive organs, peritoneum, respiratory and intrathoracic organs, bone, connective tissue, skin, breast, genitourinary organs, lymphatic, or hematopoietic tissue.

The moderate-to-severe UC disease severity sub-sample included 1728 employees with UC meeting the study criteria and 1728 matched controls. The sub-sample of patients with moderate-to-severe UC disease was, on average, 44.8 years old, and more than half were men (62.0%). Baseline comparisons were similar for this sub-sample and their matched controls (Table 1). As seen in the overall population, employees with moderate-to-severe UC had significantly higher rates (p < 0.05) compared with their matched controls for all baseline conditions, with the exception of congestive heart failure (p > 0.99) and diabetes (p = 0.07).

Study period resource use

All UC employees

Compared with employee controls, significantly more employees with UC (p < 0.0001) had hospitalization events (16.9% vs 6.2%), emergency department visits (31.1% vs 22.0%), outpatient/other visits (99.9% vs 81.5%), and prescription medication use (95.3% vs 72.0%; Table 2). Employees with UC also had 4-times the mean number of hospital days (1.2 vs 0.3, p < 0.0001), and approximately twice the mean number of emergency department visits (0.5 vs 0.3, p < 0.0001) and outpatient/other visits per year compared with employee controls without UC (15.0 vs 7.5, p < 0.0001).

Table 2. Direct and indirect resource utilization during the study period for all patients, regardless of UC severity.

	All UC patients [A] (n = 4314)	Controls [B] (n = 4314)	Difference [A] − [B]	p-value^a [A] vs [B]
Direct resource use, n (%)
Hospitalization event	731 (16.9)	268 (6.2)	10.7	<0.0001
Emergency department visit	1343 (31.1)	951 (22.0)	9.1	<0.0001
Outpatient/other visit	4311 (99.9)	3517 (81.5)	18.4	<0.0001
Any prescription medication use	4113 (95.3)	3106 (72.0)	23.3	<0.0001
Indirect resource use, n (%)
Disability day	470 (10.9)	289 (6.7)	4.2	<0.0001
Medically-related absenteeism day	4256 (98.7)	3457 (80.1)	18.5	<0.0001
Either	4312 (100.0)	3511 (81.4)	18.6	<0.0001
Direct resource use (mean [median] ± SD)
Hospitalization event	0.3 [0.0] ± 1.0	0.1 [0.0] ± 0.7	0.2	<0.0001
Number of days of hospitalization	1.2 [0.0] ± 5.5	0.3 [0.0] ± 1.6	0.9	<0.0001
Emergency department visits	0.5 [0.0] ± 1.3	0.3 [0.0] ± 1.0	0.2	<0.0001
Outpatient/other visits	15.0 [10.0] ± 15.0	7.5 [4.0] ± 10.6	7.4	<0.0001
Indirect resource use (mean [median] ± SD)
Disability days	7.9 [0.0] ± 35.3	4.5 [0.0] ± 26.3	3.4	<0.0001
Medically-related absenteeism days	8.5 [5.0] ± 12.5	4.3 [2.0] ± 10.4	4.2	<0.0001
Total days	16.4 [5.5] ± 37.1	8.8 [2.0] ± 28.4	7.6	<0.0001

SD, standard deviation; UC, ulcerative colitis.

^aMcNemar tests and Wilcoxon signed rank tests were used for comparisons of categorical and continuous variables, respectively.

Significantly (p < 0.0001) more employees with UC had disability days (10.9% vs 6.7%) and medically-related absenteeism days (98.7% vs 80.1%) compared to matched controls. Specifically, these patients experienced approximately twice the total mean number of disability days and medically-related absenteeism days (16.4 vs 8.8; p < 0.0001) per year compared with their matched controls (Table 2).

Moderate-to-severe UC employees

Compared with employee controls, significantly (p < 0.0001) more employees with moderate-to-severe UC had hospitalization events (26.5% vs 6.2%), emergency department visits (37.5% vs 22.0%), outpatient/other visits (99.9% vs 81.9%), and prescription medication use (98.8% vs 72.6%; Table 3). Employees with moderate-to-severe UC had ∼7-times the mean number of hospital days (2.2 vs 0.3, p < 0.0001) and ∼2-times the mean number of emergency department visits (0.7 vs 0.3, p < 0.0001) and outpatient/other visits (19.4 vs 7.7, p < 0.0001) per year compared with employee controls without UC.

Table 3. Direct and indirect resource utilization during the study period for patients with moderate-to-severe UC.

	Moderate-to-severe UC patients [A] (n = 1728)	Controls [B] (n = 1728)	Difference [A] − [B]	p-value^a [A] vs [B]
Direct resource use (n, %)
Hospitalization event	458 (26.5)	107 (6.2)	20.3	<0.0001
Emergency department visit	648 (37.5)	380 (22.0)	15.5	<0.0001
Outpatient/other visit	1726 (99.9)	1416 (81.9)	17.9	<0.0001
Any prescription medication use	1708 (98.8)	1254 (72.6)	26.3	<0.0001
Indirect resource use (n, %)
Disability day	289 (16.7)	128 (7.4)	9.3	<0.0001
Medically-related absenteeism day	1697 (98.2)	1399 (81.0)	17.2	<0.0001
Either	1728 (100.0)	1420 (82.2)	17.8	<0.0001
Direct resource use (mean [median] ± SD)
Hospitalization event	0.5 [0.0] ± 1.4	0.1 [0.0] ± 0.8	0.4	<0.0001
Number of days of hospitalization	2.2 [0.0] ± 7.2	0.3 [0.0] ± 1.4	1.9	<0.0001
Emergency department visits	0.7 [0.0] ± 1.6	0.3 [0.0] ± 0.9	0.4	<0.0001
Outpatient/other visits	19.4 [14.0] ± 16.9	7.7 [4.0] ± 11.1	11.7	<0.0001
Indirect resource use (mean [median] ± SD)
Disability days	12.3 [0.0] ± 43.3	4.5 [0.0] ± 25.0	7.8	<0.0001
Medically-related absenteeism days	11.5 [7.5] ± 16.1	4.6 [2.0] ± 11.1	6.9	<0.0001
Total days^b	23.8 [8.5] ± 45.4	9.0 [2.5] ± 27.6	14.8	<0.0001

SD, standard deviation.

^aMcNemar tests and Wilcoxon signed rank tests were used for comparisons of categorical and continuous variables, respectively.

^bValues may not sum to the total due to rounding.

Significantly (p < 0.0001) more employees with moderate-to-severe UC had disability days (16.7% vs 7.4%) and medically-related absenteeism days (98.2% vs 81.0%) compared with matched controls. Specifically, these patients had 2.6-times the total mean number of disability and medically-related absenteeism days per year (23.8 vs 9.0; p < 0.0001; Table 3) compared with their matched controls.

Study period direct and indirect cost burden

All UC employees

Employees with UC had, on average, significantly higher adjusted total direct costs compared to employee controls ($15,548 vs $4812; p < 0.0001; Table 4). Outpatient/other and hospitalization costs were the main contributors to direct costs for both UC employees and controls. Additionally, the total adjusted indirect costs were, on average, 2-times higher for employees with UC compared with employee controls ($4125 vs $1961; p < 0.0001). After controlling for the CCI and medical conditions that differed significantly at p < 0.05 (excluding those gastro-related conditions such as intestinal disorders, non-infectious gastroenteritis and colitis, and symptoms of the abdomen and pelvis), the adjusted total cost burden for employees with UC was almost 3-times the value of their matched controls ($19,619 vs $6787; p < 0.0001).

Table 4. Annual total, direct, and indirect costs for all patients, regardless of UC severity.

	All UC patients [A] (n = 4314)	Controls [B] (n = 4314)	Difference [A] − [B]	p-value [A] vs [B]
Unadjusted costs^a
Total direct costs (mean [median] ± SD)	$15,378 [$7596] ± $31,342	$4625 [$1,371] ± $13,109	$10,753	<0.0001
Hospitalization costs	$4078 [$0] ± $23,251	$886 [$0] ± $7438	$3212	<0.0001
Emergency department costs	$375 [$0] ± $1313	$204 [$0] ± $907	$171	<0.0001
Outpatient/other costs	$6861 [$3603] ± $13,310	$2534 [$674] ± $7556	$4327	<0.0001
All drug costs	$4063 [$2251] ± $7035	$1020 [$143] ± $2801	$3043	<0.0001
Total indirect costs (mean [median] ± SD)	$4103 [$1422] ± $10,021	$1911 [$442] ± $6949	$2192	<0.0001
Disability costs	$1727 [$0] ± $8832	$829 [$0] ± $6072	$897	<0.0001
Medically-related absenteeism costs	$2376 [$1202] ± $4722	$1082 [$395] ± $2966	$1295	<0.0001
Adjusted costs^b
Total costs (mean [median] ± SD)	$19,619 [$12,964] ± $18,214	$6787 [$4485] ± $6301	$12,832	<0.0001
Total direct costs	$15,548 [$10,196] ± $15,730	$4812 [$3155] ± $4868	$10,737	<0.0001
Total indirect costs	$4125 [$2775] ± $3006	$1961 [$1319] ± $1429	$2164	<0.0001

SD, standard deviation; UC, ulcerative colitis.

^aWilcoxon signed rank tests were used for comparisons of unadjusted costs.

^bA generalized linear model regression was used to estimate the impact of UC on the magnitude of the total, direct or indirect costs. The model controlled for factors that were significantly different (p < 0.05) between UC patients and controls at baseline.

Moderate-to-severe UC employees

On average, employees with moderate-to-severe UC had significantly higher adjusted total direct costs compared to employee controls ($23,085 vs $4932; p < 0.0001; Table 5). As in the overall population, outpatient/other and hospitalization costs were the largest contributors to direct costs for both moderate-to-severe UC employees and controls. Additionally, unadjusted total indirect costs were almost 3-times higher for employees with moderate-to-severe UC compared with employee controls ($5666 vs $1960; p < 0.0001). After controlling for the CCI and medical conditions that differed significantly at p < 0.05 (excluding those gastro-related conditions such as intestinal disorders, non-infectious gastroenteritis and colitis, and symptoms of the abdomen and pelvis), the adjusted total cost burden was 4-times that of their matched controls ($28,586 vs $6899; p < 0.0001).

Table 5. Annual total, direct, and indirect costs for patients with moderate-to-severe UC.

	Moderate-to-severe UC patients [A] (n = 1728)	Controls [B] (n = 1728)	Difference [A] − [B]	p-value [A] vs [B]
Unadjusted costs^a
Total direct costs (mean [median] ± SD)	$22,874 [$11,559] ± $39,714	$4690 [$1386] ± $12,126	$18,184	<0.0001
Hospitalization costs	$7357 [$0] ± $30,601	$881 [$0] ± $6599	$6475	<0.0001
Emergency department costs	$531 [$0] ± $1574	$192 [$0] ± $779	$339	<0.0001
Outpatient/other costs	$9245 [$4721] ± $15,345	$2561 [$694] ± $7296	$6684	<0.0001
All drug costs	$5741 [$3459] ± $8967	$1056 [$142] ± $3000	$4685	<0.0001
Total indirect costs (mean [median] ± SD)	$5784 [$1978] ± $12,914	$1900 [$451] ± $6409	$3884	<0.0001
Disability costs	$2713 [$0] ± $11,511	$772 [$0] ± $5093	$1941	<0.0001
Medically-related absenteeism costs	$3071 [$1618] ± $6084	$1128 [$399] ± $3077	$1943	<0.0001
Adjusted costs^b
Total costs (mean [median] ± SD)	$28,586 [$21,297] ± $22,883	$6899 [$5140] ± $5522	$21,687	<0.0001
Total direct costs	$23,085 [$16,957] ± $21,145	$4932 [$3623] ± $4518	$18,152	<0.0001
Total indirect costs	$5666 [$4220] ± $3150	$1960 [$1460] ± $1090	$3706	<0.0001

SD, standard deviation; UC, ulcerative colitis.

^aWilcoxon signed rank tests were used for comparisons of unadjusted costs.

^bA generalized linear model regression was used to estimate the impact of UC on the magnitude of the total, direct or indirect costs. The model controlled for factors that were significantly different (p < 0.05) between UC patients and controls at baseline.

Discussion

Drawing on data from a geographically diverse claims database of many large employers, this study estimated the direct and indirect costs, and resource use of employees with UC in the recent real-world practice. The results of this study provide further support of the large societal burden of UC and highlight the increased burden among patients with moderate-to-severe UC. Employees with UC incurred significantly higher total, direct, and indirect costs than did employees without UC during the 1-year study period. Our multivariate analysis adjusted for the CCI and medical conditions (excluding those gastro-related such as intestinal disorders, non-infectious gastroenteritis and colitis, and symptoms of the abdomen and pelvis), and revealed that employees with UC had, on average, $12,832 higher total annual costs than employees without UC (i.e., $19,619 vs $6787). The cost burden was particularly high among the sub-population of employees with moderate-to-severe UC, which had $21,687 higher total annual costs than employees without UC (i.e., $28,586 vs $6899).

A few previous studies have investigated the burden of UC in the US, using insurance claims through 200535–39. This study provides an update on the cost burden, using insurance claims from January 1, 2005 through March 31, 2013. Additionally, this study calculated the burden from the claims of a large number of privately insured employers, which provides a more comprehensive estimation of the burden of UC. By including patients’ co-payments, the UC burden estimates from the societal perspective in our study should be more comprehensive than the previous studies.

Our study improves upon the estimates of the indirect cost burden associated with UC that was previously available. First, this analysis included a larger set of employees than previous studies. Although Gibson et al.36 had a total sample of nearly 9000 UC patients; their indirect cost calculation relied on the claims data of 386 UC patients with absenteeism data and 651 patients with short-term disability data. Second, our study used the actual wages of individuals in the study sample to calculate indirect costs. On the contrary, the Gibson et al. study used $240 as a proxy for the daily wage earned for employees in ‘large firms’. This $240 daily wage estimate was made on the basis of a figure used in two previous studies40^,41. Absenteeism costs were calculated by multiplying the number of days absent from work by $240; and short-term disability costs were calculated at 70% of the $240 daily rate. Moreover, our study considered long-term disability days, while Gibson et al.’s did not, which may lead to a more comprehensive estimate of indirect costs burden.

This study encompassed the time period from January 1, 2005 through March 31, 2013 and included patients with varying levels of disease severity. This analysis is important because the average direct cost of employees with UC was $15,378, which is close to the costs reported from first quarter of 2000 through the third quarter of 2005 for UC employees in Gibson et al.36 and Hillson et al.39 ($16,220 and $19,860, respectively, when inflated to 2013 US dollars for comparison), despite the fact that our analysis included a time period of ∼8 years when biologics became available to treat UC (treatment of UC with biologics started on September 16, 2005). The similarity in direct costs suggests that biologics, which have higher drug costs than earlier UC treatments, may not substantially affect the overall cost of UC care, perhaps by decreasing the costs incurred in other cost components. Also of interest is the similarity in the direct cost difference between UC patients and their controls observed in our study ($10,753) and the other two studies (Gibson et al.: $10,314; Hillson et al.: $11,417, when inflated to 2013 US dollars). However, this topic was not the focus of the current analysis and may be explored in future research.

A limitation of the present study is that the results were estimated from claims data for privately-insured employee UC patients aged 18–64 years in the US, and, therefore, may not be generalizable to the population of all UC patients such as those covered by Medicare or Medicaid, or those not employed. The results are also subject to any error in the administrative claims, such as inaccurate or incomplete reporting. However, by restricting the sample to employees only, we may have minimized potential missing claims and, thus, provided a more accurate cost estimation for adult employees with UC. Despite these limitations, valid risk assessment and cost analyses have been performed using this database of privately insured employees42–45. Also to be noted is that the findings from the study may not be generalizable to non-US UC patient populations.

Although the results from this study demonstrate the substantial economic burden of UC, the estimated costs of UC presented are likely an under-estimate of the true burden of UC for a variety of reasons. This study did not account for the cost of workforce disruptions related to disability, which may include loss of productivity at work (i.e., presenteeism). Also excluded were administrative and training expenses for replacement workers, and days missed for sick time not requiring medical care. Even though our study applied a definition of moderate-to-severe UC similar to that used in previous studies in the literature31^,39, the number of patients with moderate-to-severe UC may have been under-estimated because the definition of moderate-to-severe UC required a hospitalization or specific treatment regimens.

Conclusion

This study shows that UC poses a substantial societal burden in terms of healthcare costs, medically-related absenteeism, and disability costs. UC is an important driver of overall healthcare utilization, direct costs, and work-related indirect costs. The impact is ∼4-times higher among employed patients with moderate-to-severe UC compared with matched employees without UC. This information is important for many stakeholders, including payers, employers, physicians, and patients, who consider interventions for UC which may reduce costs and healthcare resource utilization.

Transparency

Declaration of funding

Design, study conduct, and financial support for the study were provided by AbbVie; AbbVie participated in the interpretation of data, review, and approval of the publication.

Declaration of financial/other relationships

RC has served as a consultant and on scientific advisory boards for AbbVie, Celgene, Entera Health, Hospira, Janssen, Prometheus, Salix, Sandoz, Shire, Takeda, and UCB Pharma. He has served on speaker’s bureau for AbbVie, Entera Health, Salix, and Shire. MY is an employee of Analysis Group, Inc., which received consulting fees from AbbVie for this research. At the time of this study MD was an employee of Analysis Group, Inc., which received consulting fees from AbbVie for this research. JC, ABO, and MS are employed at AbbVie and may hold stock in AbbVie. At the time of this study JR was an employee of AbbVie and may hold stock in AbbVie.

Acknowledgments

Editorial services were provided by Joann Hettasch, PhD, of Arbor Communications, Inc., Ann Arbor, MI, and funded by AbbVie. The authors would like to acknowledge Jin Wei, an employee of Analysis Group, Inc., for her contribution to this research. Results of an earlier analysis of the same database were presented in part at the American College of Gastroenterology Annual Scientific Meeting, October 19–24, 2012, Las Vegas, Nevada; and Advances in Inflammatory Bowel Diseases: Crohn’s & Colitis Foundation’s Clinical & Research Conference, December 13–15, 2012, Hollywood, Florida.