Few data exist on complications attributable to Clostridium difficile infection (CDI) in the United States in an endemic setting. In surveys from Canada and the United Kingdom, attributable mortality rates ranged from 2.2% to 7.7%.Reference Gravel, Miller and Simor 1 , Reference Walker, Eyre and Wyllie 2 Higher mortality rates have been associated with infections due to the more virulent NAP1 strain.Reference See, Mu and Cohen 3 There are also few data on rates of admission to an intensive care unit (ICU) because of CDI or rates of colectomy for the disease. In a US survey where approximately 28% of strains were NAP1, 2% of patients were admitted to an ICU and 0.3% underwent colectomy for CDI.Reference See, Mu and Cohen 3 In the Canadian survey where NAP1 constituted about 30% of infecting strains, 2% of patients were admitted to an ICU and 1% underwent colectomy for CDI.Reference Gravel, Miller and Simor 1 A European survey where NAP1 constituted only 5% of isolates reported an attributable ICU admission rate of 0.2%, a colectomy rate of 0.7%, and a mortality rate of 2%.Reference Bauer, Notermans and van Benthem 4
The Veterans Health Administration of the US Department of Veterans Affairs (VA) implemented an initiative in July 2012 to decrease CDI in acute care facilities nationwide. This initiative stresses infection prevention and control strategies, environmental management, and a culture transformation where infection control becomes everyone’s responsibility. Data on community-onset and healthcare facility–associated CDI cases as well as complications including admission to an ICU, colectomy, or death attributable to CDI are monitored and reported monthly.
Our objective was to evaluate CDI complications reported nationwide over the first 36 months of the initiative.
METHODS
Multidrug-Resistant Organism (MDRO) Prevention Coordinators at each of the 127 VA acute care facilities nationwide entered data monthly from July 1, 2012, through June 30, 2015, into a database maintained by the VA Inpatient Evaluation Center in Cincinnati, Ohio, as previously described.Reference Evans, Simbartl, Kralovic, Jain and Roselle 5 Surveillance definitions have been described in detail elsewhere.Reference Evans, Simbartl, Kralovic, Jain and Roselle 5 Of note, a clinically confirmed hospital-onset healthcare facility–associated CDI case was defined as one with diarrhea or histopathologic or colonoscopic evidence of pseudomembranous colitis.Reference Cohen, Gerding and Johnson 6 These cases were collected to estimate the number of individuals who actually had an illness consistent with CDI because a positive laboratory result can sometimes be obtained from the stool of an asymptomatic patient. ICU admissions, colectomies, and deaths from all causes within 30 days of the index positive CDI LabID Event and those attributable to CDI during the same interval were reported to the Inpatient Evaluation Center each month. Deaths were reviewed to confirm the judgment that they were definitely attributable to CDI by communication with healthcare providers caring for the patient and medical chart review. Medical chart reviews were performed by 2 individuals in the MDRO Prevention Office (J.L.W. and M.E.E.) using VistA and CAPRI, 2 components of the VA’s national electronic health record that allow unlimited access to the same medical record available to providers at a patient’s local facility. The type of diagnostic test (nucleic acid amplification test [NAAT] or toxin A/B enzyme immunoassay [EIA]) used each month by each facility was obtained from the Inpatient Evaluation Center database.
The process of reviewing deidentified national data sets was reviewed by the Cincinnati VA Medical Center Institutional Review Board (IRB#05-6-29-2).
RESULTS
During the analysis period, there were 1,525,433 admissions, 8,458,336 patient-days, and 6,871 clinically confirmed hospital-onset healthcare facility–associated CDI in VA acute care facilities nationwide. There were 247 clinically confirmed hospital-onset healthcare facility–associated CDI cases with an ICU admission, colectomy, or death deemed attributable to CDI, and 1,821 cases with one of these complications occurring as the result of all causes, including CDI, within the 30-day interval after the CDI LabID Event (Table 1). A NAAT was used to make the diagnosis in 21 (78%) of the 27 deaths and toxin A/B EIA in the remainder.
TABLE 1 Complications of Clinically Confirmed Hospital-Onset Healthcare Facility–Associated Clostridium difficile Infection Occurring Within 30 Days of Diagnosis in Acute Care Veterans Affairs Facilities July 1, 2012–June 30, 2015

NOTE. ICU, intensive care unit.
a Deaths confirmed by medical chart review and consultation with healthcare providers.
b Complication from any cause.
c Per 100 cases.
DISCUSSION
CDI-attributable ICU admission and colectomy rates in VA facilities were similar to those reported from non-VA facilities but the attributable mortality rate of 0.4% was lower. This may be because previous reports of mortality were in the context of outbreaks, especially those due to the NAP1 strain that has been associated with more severe disease.Reference See, Mu and Cohen 3 In one analysis, the reported mortality among patients infected with the NAP1 strain was 5% compared with 1.8% among those infected with other strains.Reference See, Mu and Cohen 3 It is unknown what the prevalence of the NAP1 strain was among veterans during the analysis period, although the strain has been reported to constitute approximately 28% of isolates in other settings in the United States.Reference See, Mu and Cohen 3 Another possibility for the lower mortality rate in VA patients was the increasing use of NAAT rather than toxin A/B EIA for making a diagnosis during the analysis period. In July 2012, 41% of VA clinical laboratories were using an EIA and 56% a NAAT. By June 2015, approximately 12% were using EIA and 84% using a NAAT. Others have reported that CDI cases diagnosed by toxin assays have a higher mortality than those detected by NAAT, presumably because the latter detects C. difficile in colonized patients who develop diarrhea for other reasons.Reference Planche, Davies and Coen 7 In the VA, however, 80% of our deaths had the CDI diagnosis made using NAAT.
Of note, the ratio of deaths attributable to CDI to deaths due to all causes among patients with a positive CDI LabID Event within the previous 30 days was 1:27. Others have reported the difficulty of determining how much CDI contributes to patient morality because many infected individuals have comorbidities associated with an abbreviated survival.Reference Gilca, Frenette, Theriault, Fortin and Villeneuve 8 Our data, however, suggest that CDI mortality rates may be overestimated unless experienced personnel carefully evaluate each death for attribution to CDI.
This analysis has a number of limitations, including the lack of data on how prevalent the NAP1 and other hypervirulent C. difficile strains are in the VA population and reliance upon reporting of complications by MDRO Prevention Coordinators at each facility each month. The latter may have been offset by aggressive and accurate follow-up of each death through telephone conversations and review of the actual patient medical record by specific individuals in the MDRO Prevention Office using the VA’s national electronic health record. Strengths of this analysis include the large number of patients monitored from multiple centers throughout the United States and the contemporary 3-year time frame of the analysis. Although some clusters or outbreaks may have occurred in VA facilities during the analysis period, we monitor CDI data monthly and have not observed any sustained increases in CDI case rates in any specific facilities. Thus, unlike in some reports, we believe the complications observed in the VA system were those of an endemic rather than an epidemic setting. Although CDI affects all age groups, it affects older persons, such as veterans, more than younger persons.Reference Lessa, Mu and Bamberg 9 Thus, this analysis may give an accurate picture of complications attributable to CDI in the patients most commonly involved. Attribution of death to CDI is difficultReference Gilca, Frenette, Theriault, Fortin and Villeneuve 8 but simply using all-cause mortality may lead to an overestimation of the impact of CDI on a patient population. We chose to make this difficult determination by medical chart review and in consultation with healthcare providers who took care of each patient.
ACKNOWLEDGMENTS
We thank the VA Under Secretary for Health, the Deputy Under Secretary for Health for Policy and Services, the CDI Data Workgroup, the VA MRSA/MDRO Taskforce, the MRSA/MDRO Prevention Coordinators, infection prevention and control professionals, infectious diseases specialists, and clinical laboratory personnel at each facility for support of the MDRO Prevention Initiatives and their hard work and dedication toward improving the healthcare of America’s veterans. Thanks also to Linda Flarida of the MDRO Prevention Office for help with data validation and cleaning.
Financial support. None reported.
Potential conflicts of interest. All authors report no conflicts of interest relevant to this article.