Heaton, Pamela C.*; Tundia, Namita L.*; Schmidt, Nicole†; Wigle, Patricia R.‡; Kelton, Christina M.L.§
Comprehensive burden-of-illness studies, by quantifying the magnitude of the disease problem, guide the innovative efforts of medical and pharmaceutical companies and generate public interest in funding research efforts. They also provide a solid foundation for cost-effectiveness and comparative-effectiveness studies for new treatments versus established methods of care. Because the pediatric population represents a special subgroup of patients who often respond differently from adults to medical and pharmacotherapeutic regimens, estimating the costs borne by them and their parents or caregivers requires a dedicated effort with an appropriate research design and database. In the case of pediatric inflammatory bowel disease (IBD), which refers collectively to Crohn disease (CD) and ulcerative colitis (UC), the most burdensome aspect, at least in terms of direct cost, is hospitalization. Hospitalization may occur multiple times during the course of the illness and may involve various surgical procedures (1).
Two previous burden-of-illness studies that did not focus on children but included children with IBD, reported total costs, but they did not report inpatient costs separately (2,3). Bickston et al (4) reported mean inpatient costs for adolescents with UC but did not provide estimates for all of the adolescents with IBD nor did they provide estimates for younger children. Kappelman et al (5) estimated inpatient costs, along with outpatient and pharmaceutical costs, for both UC and CD and for both children and adults using a PharMetrics database (IMS Health, Danbury, CT). Although their study gave an overview of the pediatric cost of illness, there was little detail offered. First, the study did not stratify the pediatric population by age. Second, it did not determine the effect of other patient characteristics, hospital characteristics, or comorbidities. Finally, the database used in the study included only a small number of publicly insured children and, hence, could not speak to differences in outcomes across payer types. Moreover, the database in the Kappelman et al study could not be used to produce nationally representative estimates. The present study fills in the details left out by the earlier study, identifying important pediatric cost drivers through an in-depth analysis. Moreover, it provides a nationally representative estimate of the costs of pediatric CD and UC.
In the United States, the prevalence of CD and UC in children younger than 20 years is 43 and 28/100,000, respectively (6). In approximately 25% to 30% of patients with CD and 20% of patients with UC, symptoms are present before the age of 20 years (7), and 4% of pediatric IBD cases are diagnosed in children younger than 5 years (8). Children with IBD experience reduced quality of life because of their disease and associated depression and other psychological comorbidities (9). It is the goal of this research to estimate the total pediatric inpatient IBD burden in the United States. Using the 2006 Kids’ Inpatient Database (KID), the objectives were to quantify the burden, both overall and stratified by patient and hospital characteristics, by determining the number of hospitalizations, the number of days spent in the hospital, and hospitalization costs. All of the estimates are reported for CD and UC separately because of their unique illness patterns.
The 2006 KID (the latest available version of KID at the time of writing) was used to examine the economic burden of IBD in hospitalized children and adolescents ages 20 years and younger. Developed by the Healthcare Cost and Utilization Project (HCUP), which was sponsored by the Agency for Healthcare Research and Quality, the 2006 KID includes data from 3739 hospitals in 38 states. Of this total, 11.0% of the hospitals were in the north east region of the country, 34.1% were in the midwest, 35.8% were in the south, and 19.0% were in the west (10). The KID is the only hospital administrative dataset designed specifically to evaluate the use of hospital services by newborns, children, and adolescents. As a result, the KID is ideal for studying specific conditions, procedures, and subpopulations that often cannot be assessed with other databases because of the relatively small proportion of hospital stays that include children. The KID generates national estimates through the use of a discharge weight variable, which is a multiplier applied to each discharge. Because the KID provides only charge data, costs were estimated using HCUP cost-to-charge ratios and analytical techniques outlined in Agency for Healthcare Research and Quality documentation.
According to the KID, there were a total of 7,558,812 pediatric hospitalizations in the United States in 2006. The mean cost was $4935 (95% confidence interval [CI] $4712–$5158), and mean length of hospital stay in days (LOS) was 3.67 (95% CI 3.59–3.75) days (11). The institutional review board at the University of Cincinnati determined that the KID data were exempt from review.
Patient Visit Selection
IBD-related hospitalizations were identified using International Classification of Disease-9 (ICD-9) codes recorded in the KID as the primary diagnosis. Specifically, the hospitalization was selected if the primary diagnosis was characterized by one of the following ICD-9 codes: 555.0, 555.1, 555.2, or 555.9 (regional enteritis of the small intestine, large intestine, small intestine with large intestine, or unspecified site, respectively); or 556.0, 556.1, 556.2, 556.3, 556.4, 556.5, 556.6, 556.8, or 556.9 (ulcerative chronic enterocolitis, ulcerative chronic ileocolitis, ulcerative chronic proctitis, ulcerative chronic proctosigmoiditis, pseudopolyposis of colon, left-sided ulcerative chronic colitis, universal ulcerative chronic colitis, other UC, or UC unspecified, respectively).
LOS and charges for the hospitalization were found in KID for each of the IBD-related hospitalizations. Hospitalization cost was found using the hospital's cost-to-charge ratio provided by HCUP.
Patient characteristics included age, sex, income, and health insurance coverage. We defined 4 age groups: children (0–5 years), youth (6–10 years), young adolescents (11–15 years), and older adolescents (16–20). Income was classified according to HCUP documentation for income quartiles for counties of residence: very low (<$38,000), low ($38,000–$46,999), high ($47,000–$61,999), and very high (≥$62,000). Primary health insurance coverage was classified as private, Medicaid, Medicare (children with end-stage renal disease are covered by Medicare), self-pay, no charge, and “other” (including Title V, CHAMPUS, CHAMPVA, and other government programs) for purposes of descriptive stratification (10). For the regression analyses, the latter 3 groups were considered together. Race was not included because data on race were missing for 9 of the 38 states included in the KID.
Hospital characteristics included teaching status, urban versus rural location, region, hospital bed count, and whether the hospital was a stand-alone children's hospital. A teaching hospital was defined as having an American Medical Association–approved residency program, by being a member of the Council of Teaching Hospitals and Health Systems, or by having a ratio of full-time equivalent interns and residents to beds of 0.25 or higher. HCUP documentation defined hospital location as either rural or urban using core-based statistical area (CBSA) codes from the 2000 census data. Hospitals residing in counties with a CBSA type of metropolitan were considered urban, whereas hospitals with a CBSA type of micropolitan or noncore were classified as rural (10). The hospital region was identified as northeast, midwest, south, or west. A hospital was classified as small if it was rural and had 1 to 49 beds, urban and nonteaching and had 1 to 99 beds, or urban and teaching with fewer than 300 beds. A hospital was considered medium size if it was rural and had 50 to 99 beds, urban and nonteaching with 100 to 199 beds, or urban and teaching with 300 to 499 beds. A large-size hospital could be rural with >99 beds, urban and nonteaching with >199 beds, or urban and teaching with >499 beds.
Severity of illness (minor, medium, major, or extreme) was captured from the fourth digit of the hospitalization's All Patient Refined-Drug-related Groups (APR-DRG) code (12). The DRG system was developed for Medicare as part of a prospective payment system to classify hospital cases into 1 of approximately 500 groups expected to have similar hospital resource use. The APR-DRG is a refinement of the system that accounts as well for severity of illness (minor, moderate, major, or extreme) and risk of mortality (minor, moderate, major, or extreme). APR-DRGs were found by RAND Health to explain substantially more variation in hospital cost than other DRG classification systems (13).
The comorbidities experienced by patients hospitalized for IBD were identified using comorbidity software developed by HCUP (14). The number of diagnoses, DRGs, and ICD-9 codes were used to create 29 comorbidity measures (15). Visits during which surgery was performed were also identified by procedure code. The online-only appendix (http://links.lww.com/MPG/A72) lists specific bowel resection surgeries and their associated procedure codes.
The total pediatric inpatient burden of IBD (and CD and UC, separately), in terms of the number of IBD-related hospitalizations, the total number of days hospitalized, and cost of hospitalizations, was calculated taking into account the complex weighting and sample design of the KID. Mean LOS and cost were calculated for all of the IBD discharges, categorized by hospital characteristics, comorbidities, and other patient characteristics. We computed 95% CIs to capture the degree of variability around the national estimates of mean LOS and cost. No statistical comparisons were made across subpopulations. An ordinary least-squares regression model was developed and estimated to explain hospitalization cost. The dependent variable was log transformed cost. The decision to include LOS as an explanatory variable was made after preliminary analysis of its statistical relation to the other independent variables. The log transformation was necessary to ensure the validity of the error-term normality assumption. Statistical significance was evaluated at the 5% level. After inspection of all of the between-variable estimated correlation coefficients, we determined that multicollinearity was not a problem. Because of the large number of regressors, we included only their main effects in the model. We did not investigate the effects of potential interactions. We used SAS software (version 9.1, SAS Institute Inc, Cary, NC) for all of the analyses, specifically using the survey procedures available in SAS that take account of the sampling design of KID.
In 2006, there were 10,777 IBD-related pediatric hospital visits, of which 6599 were CD related and 4178 were UC related (Table 1). For CD and UC, respectively, there were 37,175 and 27,810 days spent in the hospital; mean LOS was 5.63 (95% CI 5.41–5.85) days and 6.66 (95% CI 6.27–7.05) days, respectively, greater than the mean LOS for all types of discharges (3.67 days). The total and mean costs for CD- and UC-related visits were $66.3 million and $10,176 (95% CI $9647–$10,705), and $48.6 million and $11,836 (95% CI $10,760–$12,912), respectively. Using the Hospital Inpatient Services Consumer Price Index, the total IBD-related inpatient cost was inflated to 2010 dollars and equaled US$152.4 million (16).
Table 2 breaks down the IBD-related hospitalizations by patient and hospital characteristics. For each subgroup of visits, mean LOS and cost were calculated for CD and UC separately and together. Older adolescents, ages 16 to 20 years, had the highest number of both CD- and UC-related hospitalizations. For CD-related hospitalizations, 0- to 5-year-old patients had the highest mean LOS (8.10, 95% CI 5.53–10.67, days) and mean cost ($13,894, 95% CI $9053–$18,735), whereas, for UC-related visits, 11- to 15-year-old patients had the highest mean LOS (7.49, 95% CI 6.88–8.10, days) and mean cost ($13,407, 95% CI $11,704–$15,110) across the age groups. The percentages of hospitalizations accounted for by girls and boys were almost identical. A slightly higher percentage (50.6% vs 48.6%) of CD-related hospitalizations was experienced by boys, whereas the opposite was true for UC-related visits. Of all of the IBD-related visits, 67% were visits for patients with private insurance; however, Medicaid patients had a slightly higher mean LOS (6.35, 95% CI 5.98–6.72, days) and mean cost ($11,103, 95% CI $10,072–$12,134) than patients with private insurance. Some regional variation in cost was evident. For both CD- and UC-related hospitalizations, mean cost was higher in the west and northeast regions relative to the south and midwest. Our analysis showed that in 13.7% of CD-related visits and 11.3% of UC-related visits, surgery was performed. Surgery during the visit increased mean LOS by about 4 days and essentially doubled the cost. The number of IBD-related visits and visit-related costs were seen to increase with level of income. If the IBD-related visit was in a teaching hospital, the LOS was at least 2 days longer and cost at least $3000 more compared with a nonteaching hospital. Only 18.2% of IBD-related discharges were from stand-alone pediatric hospitals. The mean LOS and costs for patients in pediatric hospitals were 6.88 (95% CI 6.08–7.68) days and $13,858 (95% CI $11,861–$15,855), respectively, higher than those in general hospitals (5.82, 95% CI 5.62–6.02, days, and $10,183, 95% CI $9552–$10,814, respectively).
TABLE 2-a Frequency,...Image Tools
Regression results are presented in Table 3. An additional day in the hospital is estimated to increase costs $773 (P < 0.01) for a CD-related hospitalization and $633 (P < 0.01) for a UC-related visit. Age was not a significant predictor of cost for CD-related visits, but it was for UC-related visits in which patients in the 11- to 15-year-old age group cost $669 (P < 0.01) more than older patients, the reference group. Surgery was found to be greatly predictive of both CD- and UC-related hospitalization costs. If the patient had surgery, then the hospitalization costs for CD increased by $4655 (P < 0.01) and for a UC-related visit increased by $7693 (P < 0.01). Severity of illness affected the cost of the hospitalization. Compared with children with extreme severity, children with minor or moderate severity had statistically significant (P < 0.01) lower costs. This result held for both CD and UC.
Comorbidities that significantly increased the hospitalization cost of CD-related visits were alcohol abuse ($18,165, P < 0.01), coagulopathy ($10,837, P < 0.01), peptic ulcer disease and bleeding ($7470, P < 0.01), and valvular disease ($7890, P < 0.01). UC-related visit costs were significantly increased by liver disease ($3700, P < 0.01) and paralytic ileus ($16,562, P < 0.01).
TABLE 3-b Results fr...Image Tools
Whereas our study estimated the national pediatric inpatient burden of IBD, previous studies have estimated total burden for both adults and children using the National Inpatient Sample (17,18). Nguyen et al (17) estimated that in 2004, CD accounted for 335,962 days in the hospital, with charges amounting to $1.33 billion. Comparing these figures to our results (for 2006) in Table 1 (37,175 days and charges of $168.7 million) suggests that the pediatric inpatient burden is approximately 10% of the total. A similar result holds for UC. The Nguyen et al totals of 237,523 days and charges of $945 million (17) are about 10 (less for the charges) times higher than the 27,810 days and $124.0 million estimated in the present study. In a separate study, Nguyen and Steinhart (18) looked specifically at differences between charges and admissions between high-IBD-volume admission centers and those hospitals not so classified.
The present study estimated mean pediatric inpatient costs of $10,176 (2006 US dollars) and $11,836 (2006 US dollars) for CD and UC, respectively. Kappelman et al (5) reported that in 2004, the average annual direct health care costs for children with CD and with UC were $9555 and $10,063, respectively. For comparison purposes, we inflated these costs to 2006 dollars, resulting in average costs of $10,807 (2006 US dollars) for CD and $11,381 (2006 US dollars) for UC. These figures are consistent with the results from the present study. Whereas the Kappelman et al study included approximately $3000 of annual outpatient and pharmaceutical costs (that our study did not include), it had subtracted approximately $3000 of control-patient costs, called for by their study design (but not by ours). Bickston et al (4), however, estimated the direct health care costs related to UC and reported that the average inpatient visit for an adolescent costs $15,025 (2005 US dollars). Our estimate of $13,407 (2006 US dollars) for an UC-related visit for an 11- to 15-year-old patient was somewhat lower.
The frequency and mean cost of hospitalizations for both CD and UC were highest for children in the highest-income quartile of households in 2006, a finding that is consistent with previous studies that have found increased incidence of IBD among individuals with higher socioeconomic status (19–21). It is possible that better access to care and more reporting of symptoms to physicians by higher-income parents led to earlier IBD diagnosis and course of treatment. Similar to results from previous studies, we found that more than half of the IBD-related visits were made by children covered by private insurance (4,22); however, slightly longer stays and higher costs were found for patients with Medicaid (19,23). Although one explanation may be that Medicaid patients are more severely ill, it is also possible that physicians choose to keep Medicaid patients in the hospital longer either because of the lack of access to outpatient health care or the inability of patients and families to coordinate such outpatient care. The longer hospital stays and higher costs associated with teaching (vs nonteaching) and children's (vs not children's) hospitals are not unexpected given the different patient populations. The sicker patients are more likely to be treated in tertiary pediatric centers as opposed to small, private community hospitals.
Presently available medications are crucial for managing pediatric IBD, and the expectation is that emerging pharmacotherapies may decrease the number of hospitalizations resulting from relapses and subsequent necessary surgeries (24,25). Aminosalicylates are used to extend remission periods. Immunosuppressants, such as azathioprine or methotrexate, have been used successfully to maintain remission in patients with CD. By controlling symptoms and maintaining remission, these immunosuppressant medications can delay or decrease the need for surgical intervention (26). Infliximab, an antitumor necrosis factor-α therapy, became available in the late 1990s. This medication has been shown to reduce hospitalizations and surgeries (25), although it may add to the overall cost of the disease (27). Its use, however, was not approved for children until May 2006, and the results from the major pediatric randomized controlled trial (REACH) were published in March 2007 (28). Although there may have been some off-label usage by children in our study, it is unlikely that many children in the 2006 KID database were exposed.
Differences between CD and UC were apparent in the results from the regression analyses developed and estimated to predict cost of hospitalization across IBD-related hospitalizations. Alcohol abuse, which is known to cause symptom flare-ups in patients with CD (29), increased the cost of a CD-related hospitalization by $18,165 (P < 0.01); the same was not true for a UC-related hospital visit, consistent with previous research that reported decreased risk of UC with alcohol (30,31). Other costly comorbidities associated with CD (P < 0.01) were peptic ulcer disease, valvular disease, and coagulopathy, an extraintestinal complication. For UC, liver disease and paralytic ileus were observed to be significant contributors to hospitalization costs. An uncommon but costly extraintestinal manifestation was pulmonary circulation disease, most likely the result of venous thromboembolism.
The present study has 3 major limitations, along with a number of minor ones related to the nature of the database. Most important, the KID is not a patient-level, longitudinal database. It is not possible to determine whether an individual child underwent a single or multiple hospitalizations in 2006. Moreover, patients cannot be followed through time the way they can be using a claims database to track the course of either their disease or their treatment. Second, there is no coding for IBD severity in the database. The fourth digit of the APR-DRG code attempts to control for severity of illness, but is related more to the number of comorbidities than IBD itself. Third, changes in the treatment of IBD since 2006 mean that the results may not be representative of the present hospitalization frequencies and costs. With the increasing use of biologics during the last 5 years, it is possible that hospitalizations and hospitalization costs have decreased. Future research will be able to capture changes since 2006.
In >25% of the visits, race was not reported and, therefore, could not be included as a variable in the analysis. The KID accounts for only direct inpatient costs. The direct health care costs of other health care services, such as ambulatory and emergency department services and outpatient medications, were not measured, and indirect costs were not included. The KID database is subject to misclassification errors, both inclusion of false-positives and missing IBD cases, although analyses from HCUP suggest substantial validity of the ICD-9 coding in KID (32). Moreover, the comorbidity software may allow for misclassification of in-hospital complications in the comorbidity measures. Despite these limitations, however, the KID is a multipayer data source that can be used to generate national estimates; hence, it provides results that are of widespread general interest.
For a pediatric disease with a rather low prevalence rate of 0.071%, the estimated annual inpatient pediatric burden of IBD is a sizeable $152.4 million (2010 US dollars) and 64,985 days spent in the hospital. Compared with the average length of stay across all of the pediatric hospitalizations (3.67 days) and the average cost ($4732 in 2006 US dollars), IBD resulted in substantially longer average stays of 6.03 days and more costly stays (on average, $10,817 in 2006 US dollars). Hospital costs are significantly affected by income, severity of illness, and various comorbidities that differed in effect depending on whether the hospitalization was for CD or for UC.
We are grateful to Alison Sampson for writing and editorial assistance and to Ned Berry, who graciously answered questions about database analysis throughout the project. The comments from 2 anonymous referees were invaluable in improving the manuscript.
1. Kappelman MD, Porter CQ, Galanko JA, et al. Utilization of healthcare resources by U.S. children and adults with inflammatory bowel disease. Inflamm Bowel Dis 2011;17:62–8.
2. Gibson TB, Eliza N, Ozminkowski RJ, et al. The direct and the indirect cost burden of Crohn's disease and ulcerative colitis. J Occup Environ Med 2008; 50:1261–1272.
3. Hay JW, Hay AR. Inflammatory bowel disease: costs-of-illness. J Clin Gastroenterol 1992; 14:309–317.
4. Bickston SJ, Waters HC, Dabbous O, et al. Administrative claims analysis of all-cause annual costs of care and resource utilization by age category for ulcerative colitis patients. J Manag Care Pharm 2008; 14:352–362.
5. Kappelman MD, Rifas-Shiman SL, Porter CQ, et al. Direct health care costs of Crohn's disease and ulcerative colitis in United States children and adults. Gastroenterology 2008; 135:1907–1913.
6. Kappelman MD, Rifas-Shiman SL, Kleinman K, et al. The prevalence and geographic distribution of Crohn's disease and ulcerative colitis in the United States. Clin Gastroenterol Hepatol 2007; 5:1424–1429.
7. Mendeloff AI, Calkins BM. Kirsner JB, Shorter RG. The epidemiology of idiopathic inflammatory bowel disease. Inflammatory Bowel Disease: From Bench to Bedside. Philadelphia:Lea & Febiger; 1988. 3–34.
8. Calkins BM, Lilienfield AM, Garland CG, et al. Trends in incidence rates of ulcerative colitis and Crohn's disease. Dig Dis Sci 1984; 29:913–920.
9. De Boer M, Grootenhuis M, Derkx B, et al. Health-related quality of life and psychosocial functioning of adolescents with inflammatory bowel disease. Inflamm Bowel Dis 2005; 11:400–406.
11. Tundia NL, Heaton PC, Kelton CML. The national burden of E-code-identified adverse drug events among hospitalized children using a national discharge database. Pharmacoepidemiol Drug Saf 2011;20:666–78.
15. Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care 1998; 36:8–27.
16. Bureau of Labor and Statistics. Consumer Price Index–all urban consumers. CPI for hospital inpatient services. http://www.bls.gov
. Accessed June 6, 2010.
17. Nguyen GC, Tuskey A, Dassopoulos T, et al. Rising hospitalization rates for inflammatory bowel disease in the United States between 1998 and 2004. Inflamm Bowel Dis 2007; 13:1529–1535.
18. Nguyen GC, Steinhart AH. Nationwide patterns of hospitalizations to centers with high volume of admissions for inflammatory bowel disease and their impact on mortality. Inflamm Bowel Dis 2008; 14:1688–1694.
19. Armitage EL, Aldhous MC, Anderson N, et al. Incidence of juvenile-onset Crohn's disease in Scotland: association with northern latitude and affluence. Gastroenterology 2004; 127:1051–1057.
20. Blanchard JF, Bernstein CN, Wajda A, et al. Small-area variations and sociodemographic correlates for the incidence of Crohn's disease and ulcerative colitis. Am J Epidemiol 2001; 154:328–335.
21. Gent AE, Hellier MD, Grace RH, et al. Inflammatory bowel disease and domestic hygiene in infancy. Lancet 1994; 343:766–767.
22. Cosgrove RF, Al-Atia RF, Jenkins HR. The epidemiology of pediatric inflammatory bowel disease. Arch Dis Child 1996; 74:460–461.
23. Sclar DA, Robison LM, Gavrun C, et al. Hospital length of stay for children and adolescents diagnosed with depression: is primary payer an influencing factor? Gen Hosp Psychiatry 2008; 30:73–76.
24. Lichtenstein GR, Yan S, Bala M, et al. Infliximab maintenance treatment reduces hospitalizations, surgeries, and procedures in fistulizing Crohn's disease. Gastroenterology 2005; 128:862–869.
25. Rubenstein JH, Cong RY, Cohen RD. Infliximab decreases resource use among patients with Crohn's disease. J Clin Gastroenterol 2002; 35:151–156.
26. Ramadas AV, Gunesh S, Thomas GA, et al. Natural history of Crohn's disease in a population-based cohort from Cardiff (1986–2003): a study of changes in medical treatment and surgical resection rates. Gut 2010; 59:1200–1206.
27. Saro C, Coba C, Casado MA, et al. Resource use in patients with Crohn's disease treated with infliximab. Aliment Pharmacol Ther 2007; 26:1313–1323.
28. Hyams J, Crandall W, Kugathasan S, et al. Induction and maintenance infliximab therapy for the treatment of moderate-to-severe Crohn's disease in children. Gastroenterology 2007; 132:863–873.
29. Swanson GR, Sedghi S, Farhadi A, et al. Pattern of alcohol consumption and its effect on gastrointestinal symptoms in inflammatory bowel disease. Alcohol 2010; 44:223–228.
30. Nakamura Y, Labarthe DR. A case-control study of ulcerative colitis with relation to smoking habits and alcohol consumption in Japan. Am J Epidemiol 1994; 140:902–911.
31. Boyko EJ, Perera DR, Koepsell TD, et al. Coffee and alcohol use and the risk of ulcerative colitis. Am J Gastroenterol 1989; 84:530–534.