These findings were presented as a poster at Digestive Disease Week 2007.
Rising hospitalization rates for inflammatory bowel disease in the United States between 1998 and 2004†
Article first published online: 7 SEP 2007
Copyright © 2007 Crohn's & Colitis Foundation of America, Inc.
Inflammatory Bowel Diseases
Volume 13, Issue 12, pages 1529–1535, December 2007
How to Cite
Nguyen, G. C., Tuskey, A., Dassopoulos, T., Harris, M. L. and Brant, S. R. (2007), Rising hospitalization rates for inflammatory bowel disease in the United States between 1998 and 2004. Inflamm Bowel Dis, 13: 1529–1535. doi: 10.1002/ibd.20250
- Issue published online: 5 NOV 2007
- Article first published online: 7 SEP 2007
- Ruth L. Kirschstein individual postdoctoral fellowship
- National Institutes of Health. Grant Number: F32DK076257
- Harvey M. and Lyn P. Meyerhoff Inflammatory Bowel Disease Center
- Crohn's disease;
- general clinical;
- economics of IBD therapies;
- outcomes research/measurements;
- surgery for IBD;
- ulcerative colitis
Background: Recent epidemiological studies suggest that the prevalences of Crohn's disease (CD) and ulcerative colitis (UC) are increasing in the United States. We sought to determine whether nationwide rates of inflammatory bowel disease (IBD) hospitalizations have increased in response to temporal trends in prevalence.
Methods: We identified all admissions with a primary diagnosis of CD or UC, or 1 of their complications in the Nationwide Inpatient Sample between 1998 and 2004. National estimates of hospitalization rates and rates of surgery were determined using the U.S. Census population as the denominator.
Results: There were an estimated 359,124 and 214,498 admissions for CD and UC, respectively. The overall hospitalization rate for CD was 18.0 per 100,000 and that for UC was 10.8 per 100,000. There was a 4.3% annual relative increase in hospitalization rate for CD (P < 0.0001) and a 3.0% annual increase for UC (P < 0.0001). Surgery rates were 3.4 bowel resections per 100,000 for CD and 1.2 colectomies per 100,000 for UC and remained stable. There were no temporal patterns for average length of stay for CD (5.8 days) or for UC (6.8 days). The national estimate of total inpatient charges attributable to CD increased from $762 million to $1,330 million between 1998 and 2004, and that for UC increased from $592 million to $945 million.
Conclusions: Hospitalization rates for IBD, particularly CD, have increased within a 7-year period, incurring a substantial rise in inflation-adjusted economic burden. The findings reinforce the need for effective treatment strategies to reduce IBD complications.
(Inflamm Bowel Dis 2007)
The inflammatory bowel diseases (IBDs), comprised of ulcerative colitis (UC) and Crohn's disease (CD), are idiopathic disorders that affect over 1 million Americans. Although there have been reports of a rising incidence of IBD since the 1940s, a recent population study conducted in Olmsted County, Minnesota, reported that the incidence of CD and UC had stabilized at 8.8 per 100,000 and 7.9 per 100,000, respectively, by 1990.1–3 More recent epidemiological data from Olmsted County suggest that both CD and UC have increased in prevalence about 20% between 2001 and 2004.4
The clinical course of IBD is marked by relapses that result in frequent hospitalizations and bowel surgery. The advent of biological therapies such as infliximab has been shown to reduce the number of hospitalizations and surgical procedures for CD,5–8 although these trends have not been demonstrated at a national level in the U.S. A Canadian study showed that nationwide hospitalization rates had modestly decreased for CD and remained stable for UC between 1994 and 2001.9
As the impact of a rising prevalence of CD may be balanced by more effective medical therapies, it is unclear how these secular changes may affect nationwide hospitalization patterns for IBD in the U.S. In this study we sought to determine rates and temporal patterns of hospitalization for both CD and UC in the 7-year period following the FDA approval of infliximab in 1998. Our secondary goal was to characterize temporal patterns in surgical utilization, length of stay, and economic burden of IBD in the same time period.
MATERIALS AND METHODS
All data were extracted from the Nationwide Inpatient Sample (NIS) between 1998 and 2004. The NIS is maintained as part of the Healthcare Cost and Utilization Project (HCUP) by the Agency for Healthcare Research and Quality (AHRQ) and is the largest all-payer database of national hospital discharges. It is a 20% stratified sample of nonfederal, acute-care hospitals in the U.S. This sample includes community and general hospitals and academic medical centers but excludes long-term facilities. Each record in the NIS represents a single hospital discharge and includes a unique identifier, demographic data (age, gender, and race), hospital transfer status, admission type (emergent, urgent, or elective), primary and secondary diagnoses (up to 15), primary and secondary procedures (up to 15), expected primary and secondary insurance payers, total hospital charges, length of stay, and hospital characteristics (region, urban versus rural location, bed-size, teaching status). The NIS data concurs with the National Hospital Discharge Survey, supporting data reliability.10
We included all hospital discharges between the years of 1998 and 2004 with a diagnosis of IBD (CD or UC). Eligibility criteria required a primary diagnosis of: 1) CD (555.x); 2) UC (556.x); or 3) a complication of either CD or UC (intestinal perforation and fistula formation, bowel obstruction, anemia, malnutrition, dehydration) with a secondary diagnosis of CD or UC. Diagnoses were identified by the Clinical Modification of the International Classification of Diseases, 9th Revision (ICD-9-CM) diagnostic codes.
Predictors and Outcome Variables
Case-mix adjustment for resource utilization was performed using the Deyo modification of the Charlson Index.11, 12 The Charlson Index is a widely used instrument used to characterize and adjust for disease burden and case-mix in administrative data. This scale incorporates 17 comorbid conditions that are weighted to yield a summary index.13 Bowel surgery status was ascertained by ICD-9-CM procedure codes and included small and large bowel resections and anastomotic revisions (456.1–456.3, 457.1–457.9, 458, 469.3–469.4). The validity of these codes for major surgical procedures has been previously documented.14
Data were analyzed using the Stata 9.0 SE software package (College Station, TX). Analyses took into account the stratified 2-stage cluster design using Stata's SVY (survey data) commands. National estimates of the total number of hospitalizations for each calendar year were obtained by extrapolating the number of discharges in the NIS to the overall sampling frame of all acute-care hospitals using an algorithm that accounted for the complex survey design and incorporated individual discharge sampling weights. This value was subsequently divided by the U.S. census population estimate for that year to derive the annual rate of hospitalization. The total population census used to calculate incidence of hospitalization was derived from mid-year (July) intercensus population estimates for each calendar year from the U.S. Census Bureau. Because there could be increases in elective IBD hospital admissions for the administration of infliximab, we performed a sensitivity analysis in which elective 1-day admissions were excluded from the analysis. Population estimates between 1998 and 2004 were similarly derived for the 4 regions of the U.S. (Northeast, Midwest, South, and West) and the following age categories (≤20 years, 21–30 years 31–40 years, 41–50 years, 51–60 years, 60–80 years, and >80 years). All hospital charges were expressed in 2005 US dollars and adjusted using the Medical Consumer's Price Index. In addition, test of time trends for incidence rates and aggregate measures of total charges and hospitalization days were performed using Poisson and linear regression, respectively, with time as the independent variable.
Outcomes for hospitalization included national estimates of total hospitalization days, mean and median length of hospital stay (LOS), total hospital charges, and mean charges per hospitalization. These measures were calculated separately for CD and UC. Factors that impacted hospital charges were assessed for both CD and UC using multiple linear regression incorporating need for surgery in addition to the above covariates. Because hospital charge data were skewed, the models were constructed for these outcomes after logarithmic transformation. Coefficients from these models were exponentiated to yield a percent change in charges associated with each predictor.
The research protocol was approved by the Institutional Review Board of the Johns Hopkins Medical Institutions.
Rates of Hospitalization
The demographics of the 359,124 CD and 214,498 UC admissions admitted between 1998 and 2004 are shown in Table 1. The overall hospitalization rate among all individuals in the U.S. census for CD was 18.0 per 100,000 and that for UC was 10.8 per 100,000. Stratified by geographic U.S. census region, the Northeast had the highest rate of hospitalizations for CD (24.1 per 100,000) compared to the Midwest (19.6 per 100,000), the South (18.5 per 100,000), and the West (11.1 per 100,000). Relative to the Northeast region, the incidence rate ratio for hospitalization was 0.81 (95% confidence interval [CI]: 0.80–0.82) for the Midwest, 0.77 (95% CI: 0.76–0.78) for the South, and 0.46 (95% CI: 0.45–0.47) for the West. Likewise, for UC hospitalizations were most frequent in the Northeast (14.7 per 100,000) compared to the Midwest (10.8 per 100,000), the South (10.2 per 100,000), and the West (8.6 per 100,000). Compared to the Northeast, the incidence rate ratio (IRR) for being admitted in the Midwest was 0.73 (95% CI: 0.72–0.74), in the South was 0.69 (95% CI: 0.68–0.70), and in the West was 0.58 (95% CI: 0.57–0.59). Age-specific hospitalization rates were also determined for both CD and UC and are shown in Figure 1. The incidence of hospitalization for CD peaked at 26.6 per 100,000 between the ages of 21 and 30 years and declined gradually thereafter. Conversely, UC hospitalization rates reached a plateau after age 20 years and remained stable until age 60, where it rose sharply.
|Demographic Variables||Crohn's Disease (N = 359,124)||Ulcerative Colitis (N = 214,498)|
|Age (SD)*||41.3 (17.9)||47.5 (21)|
|Female||42,625 (58)||23,367 (54)|
|Private||44,259 (61)||25,028 (58)|
|Medicare||13,903 (19)||11,720 (27)|
|Medicaid||8,215 (11)||3,272 (8)|
|Self-pay||3,899 (5)||1,891 (4)|
|Other||2,338 (3)||1,355 (3)|
|Charlson Index||0.3 (0.7)||0.4 (0.9)|
|Geographic U.S. region|
|Northwest||17,798 (25)||10,867 (26)|
|Midwest||17,364 (25)||9,522 (23)|
|South||27,724 (36)||15,225 (34)|
|West||10,311 (14)||8,031 (18)|
|Rural||9,236 (13)||5,131 (12)|
|Urban||63,921 (87)||38,484 (88)|
|Small or medium||27,688 (37)||16,466 (37)|
|Large||45,469 (63)||27,149 (63)|
|Nonteaching||37,532 (50)||22,383 (50)|
|Teaching||35,625 (50)||21,232 (50)|
Time Trends in Hospitalizations
Time trends in hospitalization for both CD and UC are shown in Figure 2. There was a statistically significant 4.3% annual relative increase in hospitalization rate for CD (P < 0.0001) and a 3.0% annual increase for UC (P < 0.0001). In a sensitivity analysis excluding elective 1-day admissions, there were persistent increasing trends in hospitalization for CD (4.0% annual increase) and UC (2.8% annual increase). When stratified by geographic region, there was a significant average annual increase in rate of hospitalizations for CD in the Northeast (6.4%; P < 0.001) and the Midwest (6.2%; P < 0.001), but not the other regions. Similarly, for UC, the annual relative increase in rate for the Northeast was 5.0% (P < 0.0001) and for the Midwest was 2.9% (P < 0.0001). Time trends analyses were also performed stratified by age (≤20 years, 21–40 years, 41–60 years, and >60 years) for both CD and UC. In CD, those between the ages of 21 and 40 years of age experienced the steepest absolute and relative increase in hospitalization rates with an annual 6% (P < 0.001) rise from 21.5 per 100,000 to 30.5 per 100,000 over 7 years (Fig. 3). The relative annual increases in hospitalization rates were 5% in those younger than or equal to 20 years (P < 0.001) and 3.3% in those between 41 and 60 years of age (P < 0.001), while no significant trends were observed for those older than 60 years (Fig. 3). For UC, modest relative annual increases of 3.9%, 3.8%, and 3.5% were observed for those younger than or equal to 20 years, between 21 and 40 years, and between 41 and 60 years, respectively (P < 0.001 for all 3 groups). No significant time trends were observed for those older than 60 years (Fig. 4).
Rates of Surgery
The overall rate of bowel resection during the 7-year period for CD was 3.4 surgeries per 100,000, and the rate of colectomy for UC was 1.2 per 100,000. There were no significant temporal trends for surgical rates in either CD or UC (Fig. 5). Stratified by geographic region, population-based surgical rates for CD were lower in the West (2.0 per 100,000) compared to the Northeast (4.0 per 100,000; P < 0.001), to the Midwest (4.1 per 100,000; P < 0.001), and the South (3.1 per 100,000; P < 0.001). Colectomy rates for UC were 1.2 per 100,000 for both the Northeast and the West. Relative to these regions, colectomy rates were higher in the Northeast (1.4 per 100,000; P < 0.001) and lower in the West (1.0 per 100,000; P < 0.001).
The total number of hospital days attributable to CD had increased from 262,138 days to 335,962 days over the 7-year period, reflecting a 4.5% relative annual increase (P < 0.001). For UC admissions, there was an annual 3.2% increase (P < 0.001) in hospital days from 194,968 days to 237,523 days. The overall length of stay was 5.8 days for CD and 6.8 days for UC. For both CD and UC the average length of stay had slightly decreased between 1998 and 1999, and it remained steady for the remainder of the study period (Table 2). The inflation-adjusted total charges throughout the U.S. attributable to CD hospitalizations increased from $762 million in 1998 to $1,330 million in 2004, reflecting an average 10.5% annual increase (P < 0.001). There was a similar 8.6% annual increase (P < 0.001) in total hospital charges for UC from $592 million to $945 million in the 7-year period (Table 2). The average hospital charge per admission also increased significantly by 5.0% per year in CD and 4.6% per year in UC (Table 2). Table 3 shows the results of a multivariate analysis evaluating the impact of demographic variables on average hospital charges while simultaneously adjusting for other predictor variables. Notably, undergoing bowel surgery increased average hospital charges by 93% for CD and 81% for UC. Each additional day in the LOS increased the average hospital charge in CD and UC by 8% and 9%, respectively. Hospital charges were on average 29% higher for CD and 35% higher for UC in the West compared to the Northeast. Admission to an urban hospital was associated with a 44% increase in costs compared to rural centers for both UC and CD.
|Total hospital days|
|Crohn's disease (days)a||262,138||270,400||273,088||296,711||304,514||328,241||335,962|
|Ulcerative colitis (days)b||194,968||193,181||196,256||210,516||206,205||218,795||237,523|
|Average length of stay|
|Crohn's disease (days)||6.1||5.8||5.7||5.7||5.6||5.8||5.7|
|Ulcerative colitis (days)||7.2||6.6||6.9||6.9||6.8||6.6||6.7|
|Total Charges ($million)|
|Mean charge per patient|
|Predictors Variables||Crohn's Disease % Increase (95% CI)||Ulcerative Colitis % Increase (95% CI)|
|Age (per 10 years)||0% (0% to 1%)||2% (1% to 3%)|
|Female gender||0% (-1% to 1%)||-1% (-2% to 0%)|
|Medicare||6% (3% to 9%)||0% (-3% to 5%)|
|Medicaid||2% (-1% to 5%)||4% (0% to 9%)|
|Self-pay||-1% (-4% to 2%)||-5% (-10% to -1%)|
|Resective bowel surgery||93% (84% to 103%)||81% (74% to 89%)|
|Length of stay (per additional day)||8% (7% to 9%)||9% (8% to 10%)|
|Midwest||-5% (-11% to 2%)||-4% (-15% to 9%)|
|South||-2% (-9% to 5%)||1% (-11% to 15%)|
|West||29% (18% to 41%)||36% (18% to 56%)|
|Urban||44% (38% to 51%)||44% (35% to 54%)|
|Large vs. small/medium hospital size||13% (8% to 18%)||16% (8% to 24%)|
|Teaching hospital||2% (-3% to 7%)||8% (0% to 18%)|
We have shown using nationwide hospital discharge data that the rate of hospitalization for IBD, particularly CD, has significantly increased within a 7-year period following the FDA approval of infliximab. The rise in CD hospitalizations is unlikely due to elective admissions for infliximab infusions since a sensitivity analysis excluding 1-day admissions yielded similar trends in hospitalizations. The temporal trends in hospitalizations can be partly explained by nationwide increases in prevalence of both CD and UC. Recent data from a population-based study from Olmsted County has shown increases in the prevalence of both CD and UC by 20% between 2001 and 2004.4 These data, however, may not be readily extrapolated to the remainder of the U.S. A study using more nationally representative health claims data suggested that IBD prevalence varies by geographic location.15 However, this cross-sectional study was unable to evaluate for temporal trends in prevalence. Epidemiological data from Manitoba, Canada, have also shown increases in: prevalence of CD from 198 per 100,000 to 271 per 100,000 between 1994 and 2000. A similar rise in prevalence of UC from 170 per 100,000 to 249 per 100,000 was also observed in Manitoba during the same time period.16, 17 Alternatively, a temporal change in disease severity may lead to rising IBD hospitalizations, although there are no data to support this latter hypothesis.
Despite rising IBD hospitalizations, rates of surgery remained steady between 1998 and 2004, implying an increase in response to medical therapy, increase in admissions with milder disease severity (i.e., lower threshold for hospital admissions), or both. There are several lines of evidence suggesting that infliximab, approved by the FDA in 1998, reduces surgical utilization in CD.5, 6 More recently, infliximab has been shown to be potentially surgery-sparing among patients admitted with severe, steroid-refractory UC.18 However, there are sparse data on how widespread the use of biological agents is throughout the nation, and whether it is sufficient to explain the steady surgical rates. Stable surgical rates do suggest that the rise in hospitalizations is not due to admissions specifically for surgery.
Our findings are largely consistent with a recent epidemiological study that showed a similar rise in CD-related hospitalizations and stable surgery rates using data from the National Hospital Discharge Survey.19 In addition to the previous findings, we were able to demonstrate geographic and age-specific variations in hospitalization rates. We found hospitalization rates to be the highest in the Northeast, which corresponds with the highest prevalence of IBD in the Northeast.15 The differences in hospitalization rates between the Northeast and South may reflect “a north–south gradient.” Comparative studies suggest that IBD incidence rates are higher in regions and countries of northern latitude relative to southern ones.20 Older studies of hospitalization rates among U.S. military veterans and Medicare recipients also showed a similar gradient.21, 22 Differences in environmental exposures between regions may partly explain this “north–south” gradient.
There was a peak rise in hospitalizations for CD in the 20–30-year-old age group that would be consistent with a peak incidence of IBD in the second and third decades of life.20 For UC, the first peak of hospitalization rates occurred in the 20–30-year-old age group, but there was a second more substantial peak following 60 years of age. Although there is a second peak of IBD incidence after age 60, the high number of hospitalizations observed in our study may be due to increased comorbidity and misclassification of other conditions in the elderly such as ischemic colitis as UC. Because rates of IBD-related admissions among those over the age of 60 did not change over the 7-year period in this analysis, it is unlikely that the observed rise in hospitalizations is due to an aging IBD population. In contrast, our data showed that the greatest increase in admissions for CD was among young adults between the ages of 21 and 40. The high rate of hospitalizations in this age group, which comprises a substantial fraction of the labor force, may incur significant indirect costs of IBD due to loss of work productivity.
In their analysis of trends in IBD hospitalization, Bewtra et al19 found that the rise in CD hospitalizations was counterbalanced by decreasing average length of hospitalization. In contrast, we found that average length of stay remained stable after 1998 for both CD and UC. Their study, however, extended back to 1990, and it is possible that trends for decline in average length of stay for IBD admissions took place predominantly prior to 2000. Consequently, we found total hospitalization days attributable to IBD to be rising after 1998. We subsequently demonstrated in an economic analysis that increasing hospitalization rates have had a profound impact on nationwide aggregate hospital charges attributable to IBD that accounted for over 2 billion dollars in inpatient expenditures. Of more concern is the rapid 75% rise in these inflation-adjusted hospital charges over a relatively short 7-year period.
The main strength of this study is the representativeness of the NIS hospital discharge dataset with respect to geographic region, health insurance payer, and hospital characteristics. However, use of the NIS and administrative claims data in general has several limitations. For privacy safeguards, the NIS does not contain any personal identifiers that would allow validation of diagnostic coding with a subset of medical records. Results from 2 Canadian studies have demonstrated the accuracy of administrative IBD coding for research purposes, although this has yet to be validated in the U.S.23, 24 Moreover, due to limitations of the NIS dataset, the trends in hospitalizations and economic impact were observed over a relatively short 7-year interval compared to other IBD epidemiological studies that have spanned decades. Another limitation of the study is that the unit of observation is a hospitalization and not a unique patient. Therefore, we cannot distinguish whether rising trends in hospitalizations are due to increases in the number of patients who are hospitalized, readmissions, or both. Regardless of the reason, these increasing admissions have substantial economic consequences.
Despite its limitations, the NIS has provided evidence of a growing burden of IBD at a national level. Because hospital admissions incur substantial direct and indirect costs, these recent trends in IBD-related health utilization and expenditures have significant economic impact. If these rising hospitalization rates are due predominantly to growing IBD prevalence, we must focus on identifying potentially modifiable environmental triggers and develop novel strategies to stratify individuals who are at highest risk of aggressive disease behavior and hospitalization.
- 4Increasing incidence and prevalence of inflammatory bowel disease in Olmsted County, Minnesota, during 2001-2004. Gastroenterology. 2007; 132(Suppl): A37–A38., , , et al.
- 7Remission in patients with Crohn's disease is associated with improvement in employment and quality of life and a decrease in hospitalizations and surgeries. Am J Gastroenterol. 2004; 99: 91–96., , , et al.Direct Link:
- 9Hospitalization, surgery, and readmission rates of IBD in Canada: a population-based study. Am J Gastroenterol. 2006; 101: 110–118., .Direct Link:
- 102002 HCUP Nationwide Inpatient Sample (NIS) Comparison Report. U.S. Agency for Healthcare Research and Quality; 2005: 1–89., , .
- 15Prevalence and geographical distribution of Crohn's disease and ulcerative colitis in the United States. Gastroenterology. 2007; 132(Suppl): A137., , , et al.
- 16The prevalence of extraintestinal diseases in inflammatory bowel disease: a population-based study. Am J Gastroenterol. 2001; 96: 1116–1122., , , et al.Direct Link:
- 17The epidemiology of inflammatory bowel disease in Canada: a population-based study. Am J Gastroenterol. 2006; 101: 1559–1568., , , et al.Direct Link: