Advertisement

Impact of Clostridium difficile-associated diarrhea on acute care length of stay, hospital costs, and readmission: A multicenter retrospective study of inpatients, 2009-2011

      Highlights

      • We used the Premier database to determine Clostridium difficile-associated diarrhea burden on U.S. acute care hospitals.
      • Patients with Clostridium difficile-associated diarrhea had higher lengths of stay, intensive care unit admission, and inpatient mortality.
      • Clostridium difficile-associated diarrhea also conferred higher adjusted costs and all-cause readmissions.
      • Clostridium difficile-associated diarrhea-attributable costs and readmissions were also higher in high-risk subgroups.
      • Prevention of initial and recurrent Clostridium difficile-associated diarrhea is essential to lessen burden to hospitals.

      Background

      The recent epidemiologic changes of Clostridium difficile-associated diarrhea (CDAD) have resulted in substantial economic burden to U.S. acute care hospitals. Past studies evaluating CDAD-attributable costs have been geographically and demographically limited. Here, we describe CDAD-attributable burden in inpatients, overall, and in vulnerable subpopulations from the Premier hospital database, a large, diverse cohort with a wide range of high-risk subgroups.

      Methods

      Discharges from the Premier database were retrospectively analyzed to assess length of stay (LOS), total inpatient costs, readmission, and inpatient mortality.

      Results

      Patients with CDAD had significantly worse outcomes than matched controls in terms of total LOS, rates of intensive care unit (ICU) admission, and inpatient mortality. After adjustment for risk factors, patients with CDAD had increased odds of inpatient mortality, total and ICU LOS, costs, and odds of 30-, 60- and 90-day all-cause readmission versus non-CDAD patients. CDAD-attributable costs were higher in all studied vulnerable subpopulations, which also had increased odds of 30-, 60- and 90-day all-cause readmission than those without CDAD.

      Conclusion

      Given the significant economic impact CDAD has on hospitals, prevention of initial episodes and targeted therapy to prevent recurrences in vulnerable patients are essential to decrease the overall burden to hospitals.

      Key Words

      In the last 15 years, the epidemiology of Clostridium difficile-associated diarrhea (CDAD) has changed, resulting in increased incidence and severity.
      • Khanna S.
      • Pardi D.S.
      The growing incidence and severity of Clostridium difficile infection in inpatient and outpatient settings.

      Lucado J, Gould C, Elixhauser A. Clostridium difficile infections (CDI) in hospital stays, 2009. Healthcare Cost and Utilization Project (HCUP) Statistical Brief #124. Available from: http://www.hcup-us.ahrq.gov/reports/statbriefs/sb124.pdf. Accessed August 14, 2014.

      • Song X.
      • Bartlett J.G.
      • Speck K.
      • Naegeli A.
      • Carroll K.
      • Perl T.M.
      Rising economic impact of Clostridium difficile-associated disease in adult hospitalized patient population.
      Between 2000 and 2009, C difficile infection hospitalizations increased by 237% in the United States,
      • Peery A.F.
      • Dellon E.S.
      • Lund J.
      • Crockett S.D.
      • McGowan C.E.
      • Bulsiewicz W.J.
      • et al.
      Burden of gastrointestinal disease in the United States: 2012 update.
      with nearly 1% of all hospitalizations involving CDAD in 2009.

      Lucado J, Gould C, Elixhauser A. Clostridium difficile infections (CDI) in hospital stays, 2009. Healthcare Cost and Utilization Project (HCUP) Statistical Brief #124. Available from: http://www.hcup-us.ahrq.gov/reports/statbriefs/sb124.pdf. Accessed August 14, 2014.

      Additionally, nearly 250,000 people require hospital care for CDAD each year.

      Centers for Disease Control and Prevention. Antibiotic resistance threats in the United States, 2013. Available from: http://www.cdc.gov/drugresistance/threat-report-2013/pdf/ar-threats-2013-508.pdf. Accessed April 9, 2014.

      Despite current therapies, CDAD-related morbidity and mortality rates remain high,
      • Hensgens M.P.
      • Goorhuis A.
      • Dekkers O.M.
      • van Benthem B.H.
      • Kuijper E.J.
      All-cause and disease-specific mortality in hospitalized patients with Clostridium difficile infection: a multicenter cohort study.
      with worse outcomes and higher acute care costs than in patients without CDAD.
      • Dubberke E.R.
      • Butler A.M.
      • Reske K.A.
      • Agniel D.
      • Olsen M.A.
      • D'Angelo G.
      • et al.
      Attributable outcomes of endemic Clostridium difficile-associated disease in nonsurgical patients.
      • Kyne L.
      • Hamel M.B.
      • Polavaram R.
      • Kelly C.P.
      Health care costs and mortality associated with nosocomial diarrhea due to Clostridium difficile.
      This is particularly true in patients with risk factors, including renal disease, malignant neoplasms, immunocompromising conditions, inflammatory bowel disease (IBD), or concomitant antibiotic use, where CDAD is associated with substantially increased economic burden.
      • Campbell R.
      • Dean B.
      • Nathanson B.
      • Haidar T.
      • Strauss M.
      • Thomas S.
      Length of stay and hospital costs among high-risk patients with hospital-origin Clostridium difficile-associated diarrhea.
      • Quimbo R.A.
      • Palli S.R.
      • Singer J.
      • Strauss M.E.
      • Thomas S.M.
      Burden of Clostridium difficile-associated diarrhea among hospitalized patients at high risk of recurrent infection.
      Studies evaluating the U.S. health care costs attributable to CDAD have been limited to individual hospitals, specific populations, and small geographic areas.
      • Dubberke E.R.
      • Olsen M.A.
      Burden of Clostridium difficile on the healthcare system.
      • Gabriel L.
      • Beriot-Mathiot A.
      Hospitalization stay and costs attributable to Clostridium difficile infection: a critical review.
      One recent review noted that most previous studies included small sample sizes or inadequate control of confounders, such as comorbidities and increased age and illness acuity, factors that are more likely in patients with CDAD than in those without, and that cost differences may vary by region.
      • Dubberke E.R.
      • Olsen M.A.
      Burden of Clostridium difficile on the healthcare system.
      Another found that attributable outcomes (costs and length of stay [LOS]) were erratic among studies and not consistently reported, making it difficult to draw meaningful conclusions.
      • Gabriel L.
      • Beriot-Mathiot A.
      Hospitalization stay and costs attributable to Clostridium difficile infection: a critical review.
      The current study was designed to address the shortcomings of earlier assessments of CDAD-related burden on U.S. acute care hospitals. The primary strengths of the Premier hospital database include geographic diversity and its representative sampling of teaching and nonteaching hospitals. Additionally, its large size is likely adequate to provide meaningful conclusions regarding vulnerable subgroups. Our primary aim was to describe the burden attributable to CDAD in hospitalized patients, overall, and in specific vulnerable subpopulations.

      Materials and methods

      Study design

      This retrospective, observational study used data from the Premier hospital database, a deidentified patient database containing a complete census of inpatients from geographically diverse hospitals, with patient demographic information, hospital characteristics, and all discharge ICD-9-CM diagnoses and procedure codes. Date-stamped information was available for all billed services, including medications and diagnostic and therapeutic services in patient daily service records. The database is compliant with the Health Insurance Portability and Accountability Act of 1996.

      Patient selection

      The inpatient population with CDAD was identified using the first inpatient discharge (index discharge) between January 1, 2009, and December 31, 2011, in which the patient met the following criteria: aged ≥18 years at discharge; principal or secondary discharge diagnosis of ICD-9-CM 008.45 (intestinal infection caused by C difficile); received fidaxomicin, metronidazole, or vancomycin during index hospitalization; and no previous hospital admission 90 days before index admission.
      The non-CDAD control population was selected using the first inpatient discharge for patients who met the following criteria: index discharge between January 1, 2009, and December 31, 2011; aged ≥18 years at discharge; no previous hospital admission 90 days before index admission; and no record of ICD-9-CM code 008.45.
      Patients without CDAD were matched 1:1 to patients with CDAD in a 2-step process. First, patients from both cohorts were categorized by Medicare severity diagnosis-related group. Within each group, individuals were matched using the Mahalanobis caliper method for propensity score matching.
      • D'Agostino Jr., R.B.
      Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group.
      All propensity score logistic models used the same covariates: patient demographics (age, sex, race, admission source, admit type, and discharge year) and hospital characteristics (geographic region, teaching status, urban-rural status, and number of beds). All models were assessed for goodness-of-fit using the concordance c statistic. All matched patients were then aggregated, and the data were reviewed for outliers in LOS and total costs, which were removed from the analysis file, from which all analyses were conducted.

      Subgroup analysis

      Several subgroups were identified from the matched analysis file, most of which were identified by ICD-9-CM codes. These included patients with renal impairment (ICD-9-CM codes: 403.01, 403.11, 403.91, 404.02, 404.03, 404.12, 404.13, 404.92, 404.93, 582.x, 583.0-583.7, 585.x, 586.x, 588.0, V42.0, V45.1, V56.x), malignant neoplasms (140.x-172.x, 174.x-195.8, 200.x-208.x, 238.6), and IBD (556.x, 555.x). The exceptions were patients with immunocompromised status (identified as those exposed to selected alkylating agents, platinum compounds, antimetabolites, antimitotics, epipodophyllotoxins, pegaspargase, asparaginase, DNA topoisomerase inhibitors, biologic response modifiers, monoclonal antibodies, bortezomib, and tyrosine kinase inhibitors) and patients with concomitant antibiotic usage (identified as those exposed to carbapenems, cephalosporins, penicillins, aminoglycosides, tetracyclines, macrolides, fluoroquinolones, and β-lactams). The subgroups were not mutually exclusive, and patients could be included in >1 subgroup.

      Outcomes

      The impact of CDAD was evaluated by assessing the following outcomes: index hospitalization LOS; total inpatient costs; readmission within 30, 60, and 90 days of index discharge; and inpatient mortality. Costs were reported by hospitals as patient care costs and were not based on charges or cost-to-charge ratios. Readmission rates reflected all-cause readmission to the same hospital within the specified time periods for patients discharged alive from the index hospitalization.

      Statistical analysis

      Unadjusted baseline characteristics for the matched CDAD and non-CDAD groups were evaluated. Categorical variables were compared using χ2 test, and continuous variables were evaluated using Student t test. Risk-adjusted models were developed for the outcomes of interest. Categorical outcomes were modeled using logistic regression, and continuous variables were modeled using generalized linear models. Because of the skewed distribution of the continuous outcomes (LOS and costs), linear models used a log link with a gamma distribution. Outputs were exponentiated to present results in the original unit of measurement. Covariates used in all models included age, sex, race, admission source, admission type, intensive care unit (ICU) admission, Charlson comorbidity index score, geographic region, teaching status, urban-rural status, and number of beds. All models were assessed for goodness-of-fit. The P values <.05 were considered statistically significant. All analyses were conducted using SAS version 9.2 (SAS Institute, Cary, NC).

      Results

      After matching, there were 171,586 eligible discharges (85,793 per cohort). There were 2,443 (1.4%) extreme outliers with total costs <$1,000 or >$200,000 or LOS >100 days (CDAD cohort, 1,568 [1.8%]; non-CDAD, 875 [1.0%]). The final analysis dataset included 169,143 discharges (84,225 CDAD; 84,918 non-CDAD).
      Patient and hospital characteristics are presented in Table 1. The matching scheme was effective in balancing several characteristics; however, significant differences still existed in part because of the very large patient population. The CDAD cohort had a greater number of patients who were white and who were admitted from skilled nursing facilities (SNFs) (both P < .01). Patients with CDAD also had a significantly elevated mean Charlson comorbidity index score (P < .01), indicating greater underlying comorbidity. Hospital characteristics were well balanced.
      Table 1Patient and hospital characteristics
      CharacteristicCDADNon-CDADP value
      P values indicate differences observed across all groups of each category.
      Patient characteristics
       Total discharges84,225 (100.0)84,918 (100.0)
       Age group, y<.01
      18-449,114 (10.8)8,893 (10.5)
      45-6422,848 (27.1)24,043 (28.3)
      65-7417,430 (20.7)17,551 (20.7)
      75-8420,808 (24.7)20,880 (24.6)
      ≥8514,025 (16.7)13,551 (16.0)
       Age, y67.7 ± 17.267.5 ± 17.0.31
       Sex, female46,429 (55.1)46,590 (54.9).10
       Race<.01
      Black10,395 (12.3)10,438 (12.3)
      Other16,280 (19.3)18,070 (21.3)
      White57,550 (68.3)56,410 (66.4)
       Admission source<.01
      Emergency department28,683 (34.1)33,063 (38.9)
      Home38,034 (45.2)38,295 (45.1)
      Other3,487 (4.1)3,238 (3.8)
      Transfer7,894 (9.4)7,912 (9.3)
      SNF6,127 (7.3)2,410 (2.8)
       Admission type.82
      Elective9,761 (11.6)9,754 (11.5)
      Emergency60,421 (71.7)61,093 (71.9)
      Other-unknown469 (0.6)469 (0.6)
      Urgent13,574 (16.1)13,602 (16.0)
       Discharge status<.01
      Hospice3,983 (4.7)2,836 (3.3)
      Transferred7,746 (9.2)6,930 (8.2)
      Expired8,556 (10.2)6,743 (7.9)
      Home37,135 (44.1)51,762 (61.0)
      SNF25,971 (30.8)15,623 (18.4)
      Other-unknown834 (1.0)1,024 (1.2)
       Charlson comorbidity index score, mean (median) ± SD2.57 (2.00) ± 2.462.19 (2.00) ± 2.33<.01
      Hospital characteristics
       Teaching.41
      Nonteaching48,870 (58.0)49,105 (57.8)
      Teaching35,355 (42.0)35,813 (42.2)
       No. of beds.36
      <1002,607 (3.1)2,514 (3.0)
      100-1998,264 (9.8)8,271 (9.7)
      200-29913,804 (16.4)13,786 (16.2)
      300-49931,045 (36.9)31,377 (36.9)
      ≥50028,505 (33.8)28,970 (34.1)
      NOTE. Values are n (%), mean ± SD, or as otherwise indicated.
      CDAD, Clostridium difficile-associated diarrhea; SNF, skilled nursing facility.
      P values indicate differences observed across all groups of each category.
      Table 2 presents select subgroup demographics. Although the mean age was slightly lower for patients with CDAD in the renal impairment and neoplasm subgroups and slightly higher for patients with CDAD in the IBD, immunocompromised, and concomitant antibiotic subgroups, these differences were not considered clinically meaningful. Throughout the subgroups, patients with CDAD had significantly higher rates of admission from and discharge to SNFs (P < .01 for all). Similarly, except for the neoplasm subgroup (P = .71), patients with CDAD had higher Charlson comorbidity index scores (P < .01).
      Table 2Select demographics by subgroup
      DescriptionCDADNon-CDADP value
      Total eligible discharges
       Renal impairment40,232 (100.0)32,731 (100.0)
       Neoplasm12,334 (100.0)10,834 (100.0)
       Immunocompromised4,632 (100.0)3,372 (100.0)
       Inflammatory bowel disease2,972 (100.0)1,551 (100.0)
       Concomitant antibiotic55,054 (100.0)52,524 (100.0)
      Age, y
       Renal impairment70.9 ± 15.071.3 ± 14.7<.01
       Neoplasm67.7 ± 14.168.6 ± 13.5<.01
       Immunocompromised61.5 ± 17.260.3 ± 16.7<.01
       Inflammatory bowel disease57.7 ± 20.752.8 ± 19.8<.01
       Concomitant antibiotic68.3 ± 16.767.7 ± 16.8<.01
      Female sex
       Renal impairment20,517 (51.0)16,409 (50.1).02
       Neoplasm6,019 (48.8)5,228 (48.3).42
       Immunocompromised2,321 (50.1)1,652 (49.0).32
       Inflammatory bowel disease1,602 (53.9)854 (55.1).44
       Concomitant antibiotic30,143 (54.8)28,602 (54.5).10
      SNF transfer admission source
       Renal impairment3,496 (8.7)1,229 (3.8)<.01
       Neoplasm521 (4.2)153 (1.4)<.01
       Immunocompromised226 (4.9)41 (1.2)<.01
       Inflammatory bowel disease120 (4.0)13 (0.8)<.01
       Concomitant antibiotic4,463 (8.1)1,772 (3.4)<.01
      SNF discharge status
       Renal impairment13,583 (33.8)7,345 (22.4)<.01
       Neoplasm2,660 (21.6)1,421 (13.1)<.01
       Immunocompromised1,107 (23.9)445 (13.2)<.01
       Inflammatory bowel disease518 (17.4)118 (7.6)<.01
       Concomitant antibiotic18,394 (33.4)10,911 (20.8)<.01
      Charlson comorbidity index, mean (median) ± SD
       Renal impairment3.52 (3.00) ± 2.393.19 (3.00) ± 2.30<.01
       Neoplasm4.87 (4.00) ± 3.514.85 (4.00) ± 3.41.71
       Immunocompromised4.10 (3.00) ± 3.433.87 (3.00) ± 3.36<.01
       Inflammatory bowel disease1.47 (1.00) ± 2.040.86 (0.00) ± 1.57<.01
       Concomitant antibiotic2.71 (2.00) ± 2.472.24 (2.00) ± 2.34<.01
      NOTE. Values are n (%), mean ± SD, or as otherwise indicated.
      CDAD, Clostridium difficile-associated diarrhea; SNF, skilled nursing facility.
      Compared with controls, patients with CDAD had significantly worse unadjusted outcomes (Table 3), with longer total LOS and higher rates of ICU admission and inpatient mortality. Unadjusted mean patient costs were 46.8% higher and unadjusted 30-day all-cause readmission rates were 8.4% higher for patients with CDAD than for patients without CDAD. Similar results were observed for unadjusted 60- and 90-day all-cause readmission (not presented). Unadjusted outcomes by subgroup were similar to and directionally the same as outcomes of the total population (Table 3).
      Table 3Unadjusted outcomes of patients with CDAD versus patients without CDAD
      OutcomeCDADNon-CDADP valuePercentage difference
      Length of stay, dMean ± SDMedian (IQ range)Mean ± SDMedian (IQ range)
      All eligible14.4 ± 18.310 (5-17)8.7 ± 15.66 (3-10)<.0165.5
      Renal impairment14.76 ± 12.4611 (6-19)9.37 ± 9.266 (4-12)<.0157.5
      Neoplasm15.30 ± 12.6712 (7-20)9.97 ± 9.267 (4-13)<.0153.5
      Immunocompromised18.78 ± 14.5915 (6-19)12.68 ± 11.729 (4-12)<.0148.1
      Inflammatory bowel disease12.45 ± 11.868 (4-16)7.17 ± 7.825 (3-8)<.0173.6
      Concomitant antibiotic15.13 ± 12.6711 (7-20)9.44 ± 9.386 (4-12)<.0160.3
      ICU admissionn%n%Percentage-point difference
      All eligible30,94236.726,14730.8<.015.9
      Renal impairment19,41948.313,39840.9<.017.3
      Neoplasm4,44536.03,45631.9<.014.1
      Immunocompromised1,67536.295028.2<.018.0
      Inflammatory bowel disease1,07336.124315.7<.0120.4
      Concomitant antibiotic24,21244.019,00336.2<.017.8
      Inpatient mortalityn%n%Percentage-point difference
      All eligible8,55610.26,7437.9<.012.2
      Renal impairment6,55016.34,34513.3<.013.0
      Neoplasm1,71413.91,34512.4<.011.5
      Immunocompromised55211.93259.6<.012.3
      Inflammatory bowel disease38513.0473.0<.019.9
      Concomitant antibiotic6,46511.74,7659.1<.012.7
      Total inpatient costsMean ± SDMedian (IQ range)Mean ± SDMedian (IQ range)Percentage difference
      All eligible$27,408 ± $30,664$16,353 ($8,269-33,598)$18,676 ± $24,369$10,119 ($5,401-$20,992)<.0146.8
      Renal impairment$32,552 ± $33,504$20,565 ($10,644-$41,236)$22,329 ± $27,579$12,529 ($6,408-$25,958)<.0145.8
      Neoplasm$33,246 ± $33,908$20,934 ($10,773-$42,907)$23,579 ± $13,096$12,868 ($7,174-$28,184)<.0141.0
      Immunocompromised$43,078 ± $41,102$27,655 ($13,050-$59,987)$32,914 ± $37,001$18,093 ($7,987-$44,244)<.0130.9
      Inflammatory bowel disease$27,387 ± $32,440$14,787 ($7,089-$34,898)$14,334 ± $20,341$7,720 ($4,541-$15,131)<.0191.1
      Concomitant antibiotic$33,072 ± $33,754$20,954 ($10,897-$42,097)$22,484 ± $27,720$12,461 ($6,430-$26,134)<.0147.1
      30-d all-cause readmissionn%n%Percentage-point difference
      All eligible17,53923.211,53614.8<.018.4
      Renal impairment8,41525.04,71816.6<.018.4
      Neoplasm2,70725.51,87819.8<.015.7
      Immunocompromised1,23930.472323.7<.016.6
      Inflammatory bowel disease54221.023515.6<.015.3
      Concomitant antibiotic11,30523.37,06614.8<.018.5
      CDAD, Clostridium difficile-associated diarrhea; ICU, intensive care unit; IQ, interquartile; SNF, skilled nursing facility.
      After adjusting for risk factors, modeled results continued to show significant differences (Table 4). Patients with CDAD had increased odds of inpatient mortality, longer total LOS, longer ICU LOS, increased total patient costs, and increased odds of 30-, 60-, and 90-day all-cause readmission compared with patients without CDAD (P < .01 for all). These results were statistically significant overall and in the analyzed subgroups (P < .01), with the exception of odds of inpatient mortality for patients with neoplasms (P = .14) or immunocompromised status (P = .26). Although directionally the same, these results were not statistically significant.
      Table 4Adjusted outcomes of patients with CDAD versus patients without CDAD
      Inpatient mortalityOdds ratioLower
      95% confidence interval of odds ratio.
      Upper
      95% confidence interval of odds ratio.
      P value
      All eligible1.131.091.17<.01
      Renal impairment1.141.091.19<.01
      Neoplasm1.060.981.15.14
      Immunocompromised1.090.941.27.26
      Inflammatory bowel disease2.571.843.59<.01
      Concomitant antibiotic1.131.081.18<.01
      Total length of stay, dCDADNon-CDADPercentage differenceP value
      All eligible13.28.555.3<.01
      Renal impairment14.59.454.3<.01
      Neoplasm13.68.952.8<.01
      Immunocompromised17.812.048.3<.01
      Inflammatory bowel disease10.46.950.7<.01
      Concomitant antibiotic14.99.556.8<.01
      ICU length of stay, dCDADNon-CDADPercentage differenceP value
      All eligible8.36.625.8<.01
      Renal impairment10.38.521.2<.01
      Neoplasm4.73.727.0<.01
      Immunocompromised6.65.520.0<.01
      Inflammatory bowel disease7.86.127.9<.01
      Concomitant antibiotic5.04.122.0<.01
      Total patient costCDADNon-CDADPercentage differenceP value
      All eligible$25,804$18,51839.3<.01
      Renal impairment$31,263$22,32140.1<.01
      Neoplasm$24,694$17,71939.4<.01
      Immunocompromised$33,064$24,37235.7<.01
      Inflammatory bowel disease$19,667$14,14139.1<.01
      Concomitant antibiotic$29,581$21,03640.6<.01
      30-d all-cause readmissionOdds ratioLower
      95% confidence interval of odds ratio.
      Upper
      95% confidence interval of odds ratio.
      P value
      All eligible1.771.731.82<.01
      Renal impairment1.661.591.73<.01
      Neoplasm1.451.361.56<.01
      Immunocompromised1.451.301.62<.01
      Inflammatory bowel disease1.331.121.59<.01
      Concomitant antibiotic1.701.641.76<.01
      60-d all-cause readmissionOdds ratioLower
      95% confidence interval of odds ratio.
      Upper
      95% confidence interval of odds ratio.
      P value
      All eligible1.831.791.87<.01
      Renal impairment1.711.651.77<.01
      Neoplasm1.491.401.58<.01
      Immunocompromised1.531.381.70<.01
      Inflammatory bowel disease1.411.211.65<.01
      Concomitant antibiotic1.731.681.78<.01
      90-d all-cause readmissionOdds ratioLower
      95% confidence interval of odds ratio.
      Upper
      95% confidence interval of odds ratio.
      P value
      All eligible1.831.791.87<.01
      Renal impairment1.711.651.77<.01
      Neoplasm1.481.391.57<.01
      Immunocompromised1.581.431.75<.01
      Inflammatory bowel disease1.471.271.71<.01
      Concomitant antibiotic1.731.681.78<.01
      CDAD, Clostridium difficile-associated diarrhea; ICU, intensive care unit.
      95% confidence interval of odds ratio.
      Costs of care for inpatient discharges attributable to CDAD were derived by subtracting the adjusted total costs for non-CDAD patients from those of patients with CDAD. For the eligible population, CDAD-attributable costs were $7,286. CDAD-attributable costs for subgroups are as follows: renal impairment, $8,942; neoplasm, $6,975; immunocompromised status, $8,692; IBD, $5,526; and use of concomitant antibiotics, $8,545.

      Conclusions

      The burden of CDAD on resource utilization and total cost to acute care hospitals is significant. This study extends previous research by evaluating a database of all patients treated at 477 U.S. acute care hospitals during a 3-year period. The incremental costs of CDAD observed here, both overall and in individual high-risk subgroups, may provide useful information for cost-benefit analyses of new treatment regimens for CDAD.
      The present data are consistent with other studies that found increased LOS, total patient costs, and risk of readmission for patients with CDAD.
      • Dubberke E.R.
      • Olsen M.A.
      Burden of Clostridium difficile on the healthcare system.
      The increase in LOS with CDAD was 4.7 days, and the total cost attributable was $7,286. These results are broadly similar to those of Kyne et al
      • Kyne L.
      • Hamel M.B.
      • Polavaram R.
      • Kelly C.P.
      Health care costs and mortality associated with nosocomial diarrhea due to Clostridium difficile.
      (attributable LOS, 3.6 days; attributable cost, $3,669) and Song et al
      • Song X.
      • Bartlett J.G.
      • Speck K.
      • Naegeli A.
      • Carroll K.
      • Perl T.M.
      Rising economic impact of Clostridium difficile-associated disease in adult hospitalized patient population.
      (attributable LOS, 5.5 days; attributable cost, $6,326). Both of these studies were conducted at single institutions, whereas the current study used recent, geographically diverse data from several hundred hospitals, presumably providing a more generalizable estimate of current conditions.
      The subgroup analysis of vulnerable clinical populations with known increased risk of infection suggested that the effect of CDAD on the reported outcomes is consistent throughout the populations. The effect of CDAD on inpatient mortality and readmission was similar to the overall population analysis. In all subgroups, total inpatient costs were higher for patients with CDAD than for controls, confirming previous findings.
      • Campbell R.
      • Dean B.
      • Nathanson B.
      • Haidar T.
      • Strauss M.
      • Thomas S.
      Length of stay and hospital costs among high-risk patients with hospital-origin Clostridium difficile-associated diarrhea.
      • Quimbo R.A.
      • Palli S.R.
      • Singer J.
      • Strauss M.E.
      • Thomas S.M.
      Burden of Clostridium difficile-associated diarrhea among hospitalized patients at high risk of recurrent infection.
      • Tabak Y.P.
      • Zilberberg M.D.
      • Johannes R.S.
      • Sun X.
      • McDonald L.C.
      Attributable burden of hospital-onset Clostridium difficile infection: a propensity score matching study.
      • Ananthakrishnan A.N.
      • McGinley E.L.
      • Binion D.G.
      Excess hospitalisation burden associated with Clostridium difficile in patients with inflammatory bowel disease.
      The CDAD-attributable cost was slightly higher for patients with renal impairment ($8,942), immunocompromised status ($8,692), and concomitant antibiotic exposure ($8,545), compared with the overall population.
      Patients with CDAD had significantly higher 30-day readmission rates than controls in the overall population and in each high-risk subgroup studied. Comparing all-cause readmission rates observed here with previous studies is difficult because of varying definitions used for identifying the CDAD population and differences in sample size, time periods, and definition of readmission. Despite these potential confounding variables, the rates of readmission reported here are comparable with those found in recent studies.
      • Dubberke E.R.
      • Butler A.M.
      • Reske K.A.
      • Agniel D.
      • Olsen M.A.
      • D'Angelo G.
      • et al.
      Attributable outcomes of endemic Clostridium difficile-associated disease in nonsurgical patients.
      • Dubberke E.R.
      • Olsen M.A.
      Burden of Clostridium difficile on the healthcare system.
      • Aitken S.L.
      • Joseph T.B.
      • Shah D.N.
      • Lasco T.M.
      • Palmer H.R.
      • DuPont H.L.
      • et al.
      Healthcare resource utilization for recurrent Clostridium difficile infection in a large university hospital in Houston, Texas.
      • Collins C.E.
      • Ayturk M.D.
      • Flahive J.M.
      • Emhoff T.A.
      • Anderson Jr., F.A.
      • Santry H.P.
      Epidemiology and outcomes of community-acquired Clostridium difficile infections in Medicare beneficiaries.
      Although we have not specifically focused on CDAD recurrence, it is likely a contributor to the increased 30-day all-cause readmission rates and should be considered by hospitals for more accurate estimations of potential future costs. Indeed, a recent study demonstrated that approximately 50% of patients with recurrence were rehospitalized within 3 months.
      • Aitken S.L.
      • Joseph T.B.
      • Shah D.N.
      • Lasco T.M.
      • Palmer H.R.
      • DuPont H.L.
      • et al.
      Healthcare resource utilization for recurrent Clostridium difficile infection in a large university hospital in Houston, Texas.
      Moreover, it seems that using the appropriate initial treatment for CDAD should be a priority for preventing downstream resource utilization and readmission, specifically in vulnerable patients.
      This study adds to previous research in several ways particularly important to acute care hospitals. First, it provides an analysis using data from a large number of geographically diverse hospitals. Previous large-database analyses relied on Medicare data, the National Hospital Discharge Survey, or state databases, whereas other cohort studies relied on retrospective or prospective data from small numbers of acute care facilities. Each approach limits the generalizability of results. Second, we used an easily reproduced definition of CDAD. Use of ICD-9 coding to identify CDAD cases has been shown to have good concordance with cases identified by C difficile toxin assays.
      • Dubberke E.R.
      • Reske K.A.
      • McDonald L.C.
      • Fraser V.J.
      ICD-9 codes and surveillance for Clostridium difficile-associated disease.
      Inclusion of patients treated with fidaxomicin, metronidazole, or vancomycin helps identify individuals in an empirical manner. Finally, our method for cost calculation was to use patient-care costs reported directly from hospital chargemasters, rather than billed charges or cost-to-charge ratios.
      Our study has several limitations. First, the method for identifying CDAD used only ICD-9-CM and antibiotic exposure data and did not require a positive C difficile toxin assay. Although the concordance between the 2 is believed to be high, the size of the CDAD population may have been overestimated. Second, our study only considered hospital costs and not physician or treatment costs beyond the index hospitalization. Therefore, our costs were not an estimate of the total cost of CDAD to the health system. Third, readmission rates were calculated based on readmission to the same hospital. Admission to a different hospital and mortality outside of the hospital would not have been identified by the Premier database; therefore, readmission rates may have been underestimated; however, as shown previously, most patients return to the same hospital for continuing care.
      • Murphy C.R.
      • Avery T.R.
      • Dubberke E.R.
      • Huang S.S.
      Frequent hospital readmissions for Clostridium difficile infection and the impact on estimates of hospital-associated C. difficile burden.
      Because the CDAD population had higher rates of discharge to SNFs, hospices, and other acute care facilities, this limitation may have systematically underestimated the CDAD-related readmission risk. Fourth, our risk adjustment methods relied on patient data, including comorbidities at the time of discharge. Previous studies have observed that mortality associated with CDAD is usually associated with underlying disease.
      • Kyne L.
      • Hamel M.B.
      • Polavaram R.
      • Kelly C.P.
      Health care costs and mortality associated with nosocomial diarrhea due to Clostridium difficile.
      • Olson M.M.
      • Shanholtzer C.J.
      • Lee Jr., J.T.
      • Gerding D.N.
      Ten years of prospective Clostridium difficile-associated disease surveillance and treatment at the Minneapolis VA Medical Center, 1982–1991.
      Because we were unable to adjust for disease severity at admission, this should be considered when interpreting inpatient mortality risk. Finally, it is unknown whether patients acquired C difficile in the community or the hospital; however, most patients were admitted from home or a SNF.
      • Khanna S.
      • Pardi D.S.
      The growing incidence and severity of Clostridium difficile infection in inpatient and outpatient settings.
      The impact of disease origin on acute care LOS, hospital costs, and readmission in patients with CDAD, overall and in high-risk subgroups, could be addressed in future studies.
      In summary, after adjustment for risk factors, patients with CDAD had increased odds of inpatient mortality, longer total and ICU LOS, increased total patient costs, and increased odds of 30-, 60- and 90-day all-cause readmission compared with patients without CDAD, overall and in the analyzed high-risk subgroups. These results emphasize the continuing burden CDAD imparts on U.S. hospitals. Efforts focused on preventing initial CDAD episodes, and targeted therapy to prevent recurrences for vulnerable patients, are essential to decrease this burden.

      Acknowledgment

      Medical writing assistance was provided by Dan Rigotti, PhD, of StemScientific, Lyndhurst, NJ, an Ashfield Company, part of UDG Healthcare, plc. This assistance was funded by Merck & Co., Inc., Kenilworth, NJ.

      References

        • Khanna S.
        • Pardi D.S.
        The growing incidence and severity of Clostridium difficile infection in inpatient and outpatient settings.
        Expert Rev Gastroenterol Hepatol. 2010; 4: 409-416
      1. Lucado J, Gould C, Elixhauser A. Clostridium difficile infections (CDI) in hospital stays, 2009. Healthcare Cost and Utilization Project (HCUP) Statistical Brief #124. Available from: http://www.hcup-us.ahrq.gov/reports/statbriefs/sb124.pdf. Accessed August 14, 2014.

        • Song X.
        • Bartlett J.G.
        • Speck K.
        • Naegeli A.
        • Carroll K.
        • Perl T.M.
        Rising economic impact of Clostridium difficile-associated disease in adult hospitalized patient population.
        Infect Control Hosp Epidemiol. 2008; 29: 823-828
        • Peery A.F.
        • Dellon E.S.
        • Lund J.
        • Crockett S.D.
        • McGowan C.E.
        • Bulsiewicz W.J.
        • et al.
        Burden of gastrointestinal disease in the United States: 2012 update.
        Gastroenterology. 2012; 143: 1179-1187
      2. Centers for Disease Control and Prevention. Antibiotic resistance threats in the United States, 2013. Available from: http://www.cdc.gov/drugresistance/threat-report-2013/pdf/ar-threats-2013-508.pdf. Accessed April 9, 2014.

        • Hensgens M.P.
        • Goorhuis A.
        • Dekkers O.M.
        • van Benthem B.H.
        • Kuijper E.J.
        All-cause and disease-specific mortality in hospitalized patients with Clostridium difficile infection: a multicenter cohort study.
        Clin Infect Dis. 2013; 56: 1108-1116
        • Dubberke E.R.
        • Butler A.M.
        • Reske K.A.
        • Agniel D.
        • Olsen M.A.
        • D'Angelo G.
        • et al.
        Attributable outcomes of endemic Clostridium difficile-associated disease in nonsurgical patients.
        Emerg Infect Dis. 2008; 14: 1031-1038
        • Kyne L.
        • Hamel M.B.
        • Polavaram R.
        • Kelly C.P.
        Health care costs and mortality associated with nosocomial diarrhea due to Clostridium difficile.
        Clin Infect Dis. 2002; 34: 346-353
        • Campbell R.
        • Dean B.
        • Nathanson B.
        • Haidar T.
        • Strauss M.
        • Thomas S.
        Length of stay and hospital costs among high-risk patients with hospital-origin Clostridium difficile-associated diarrhea.
        J Med Econ. 2013; 16: 440-448
        • Quimbo R.A.
        • Palli S.R.
        • Singer J.
        • Strauss M.E.
        • Thomas S.M.
        Burden of Clostridium difficile-associated diarrhea among hospitalized patients at high risk of recurrent infection.
        J Clin Outcomes Manag. 2013; 20: 544-554
        • Dubberke E.R.
        • Olsen M.A.
        Burden of Clostridium difficile on the healthcare system.
        Clin Infect Dis. 2012; 55: S88-S92
        • Gabriel L.
        • Beriot-Mathiot A.
        Hospitalization stay and costs attributable to Clostridium difficile infection: a critical review.
        J Hosp Infect. 2014; 88: 12-21
        • D'Agostino Jr., R.B.
        Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group.
        Stat Med. 1998; 17: 2265-2281
        • Tabak Y.P.
        • Zilberberg M.D.
        • Johannes R.S.
        • Sun X.
        • McDonald L.C.
        Attributable burden of hospital-onset Clostridium difficile infection: a propensity score matching study.
        Infect Control Hosp Epidemiol. 2013; 34: 588-596
        • Ananthakrishnan A.N.
        • McGinley E.L.
        • Binion D.G.
        Excess hospitalisation burden associated with Clostridium difficile in patients with inflammatory bowel disease.
        Gut. 2008; 57: 205-210
        • Aitken S.L.
        • Joseph T.B.
        • Shah D.N.
        • Lasco T.M.
        • Palmer H.R.
        • DuPont H.L.
        • et al.
        Healthcare resource utilization for recurrent Clostridium difficile infection in a large university hospital in Houston, Texas.
        PLoS One. 2014; 9: e102848
        • Collins C.E.
        • Ayturk M.D.
        • Flahive J.M.
        • Emhoff T.A.
        • Anderson Jr., F.A.
        • Santry H.P.
        Epidemiology and outcomes of community-acquired Clostridium difficile infections in Medicare beneficiaries.
        J Am Coll Surg. 2014; 218: 1141-1147
        • Dubberke E.R.
        • Reske K.A.
        • McDonald L.C.
        • Fraser V.J.
        ICD-9 codes and surveillance for Clostridium difficile-associated disease.
        Emerg Infect Dis. 2006; 12: 1576-1579
        • Murphy C.R.
        • Avery T.R.
        • Dubberke E.R.
        • Huang S.S.
        Frequent hospital readmissions for Clostridium difficile infection and the impact on estimates of hospital-associated C. difficile burden.
        Infect Control Hosp Epidemiol. 2012; 33: 20-28
        • Olson M.M.
        • Shanholtzer C.J.
        • Lee Jr., J.T.
        • Gerding D.N.
        Ten years of prospective Clostridium difficile-associated disease surveillance and treatment at the Minneapolis VA Medical Center, 1982–1991.
        Infect Control Hosp Epidemiol. 1994; 15: 371-381