Community-associated Clostridium difficile infection among Veterans with spinal cord injury and disorder

Narrative Abstract

The impact of community-associated C. difficile infection (CA-CDI) on patients with spinal cord injuries and disorders (SCI/D) is not fully understood. We examined CA-CDI cases among SCI/D Veterans, comparing them with community-onset, healthcare facility-associated (CO-HCFA) cases. Generally, CA-CDI patients had less comorbidity, less severe CDI and lower likelihood of antibiotic exposure.

Introduction

Clostridium difficile infection (CDI) is frequently characterized as an infection acquired in healthcare settings, but a significant proportion of cases develops in the community.¹ Multiple studies have found over 20% of CDI cases to be community-associated.^2–3 Some findings also suggest that community-associated CDI (CA-CDI) cases may be increasing², although a recent, population-based study from the Centers for Disease Control and Prevention found that 94% of cases were healthcare-associated even though disease onset occurred outside of hospitals in 75% of cases.⁴ These data point to a need for comprehensive surveillance⁵ and deeper understanding of this segment of CDI.

The impact that increases in CA-CDI rates may have on patients with chronic and complex health needs, such as those with spinal cord injuries and disorders (SCI/D), is not fully understood; particularly since these patients often receive home-based healthcare outside of traditional clinical settings. Limited literature exists exploring the impact of community-associated infections in this population and we have previously demonstrated that community-associated infections may be a predictor of hospital-acquired infections.⁶

The objectives of this study were to describe the incidence and risk factors for CA-CDI among SCI/D patients and compare characteristics of CA-CDI with community-onset, healthcare facility-associated (CO-HCFA) cases.

Methods

Study Design and Participants

In this retrospective cohort study, eight years (January 1, 2002–December 31, 2009) of national Veterans Affairs (VA) medical data for Veterans with SCI/D were examined. Data from 104 VA healthcare facilities (HCF) with complete national data on SCI/D patients were included. This study was approved by the Hines VA Institutional Review Board.

Definitions

Classifications of CDI cases, settings, mode of acquisition and severity were defined by CDI surveillance recommendations⁵ and published clinical practice guidelines for CDI.⁷ We defined CO-HCFA CDI as cases occurring in VA HCFs within 48 hours of admission/visit, provided the patient had been discharged from a HCF within the previous four weeks.^5,7 We identified CA-CDI as cases occurring in VA HCFs within 48 hours of admission/visit, provided the patient had not been discharged from a HCF in the previous 12 weeks.^5,7 Indeterminate CDI cases were identified as occurring in VA HCFs within 48 hours of admission/visit, but had been discharged from a HCF within 4–12 weeks. Indeterminate cases were combined with CA-CDI cases in analyses due to similarities in characteristics.⁸ The severity of illness definition for severe-complicated illness was modified to accommodate typical low blood pressure in SCI/D patients; hypotension was defined for patients with paraplegia as systolic/diastolic <100/70 mm Hg and tetraplegia <90/60 mm Hg.⁹

Data Sources and Measures

Data were extracted from medical, pharmacy and laboratory records from the following VA databases. The VA Emerging Pathogens Initiative Database was used to obtain C. difficile testing information. Demographic data and medical diagnoses prior to CDI onset were derived from VA inpatient and outpatient medical SAS databases. Blood pressure data was acquired from VA Corporate Data Warehouse; creatinine and white blood cell count information was obtained from VA Decision Support System (DSS) laboratory data to develop the severity of illness variable. Antibiotic exposure in the six weeks prior to CDI was extracted from VA DSS inpatient and outpatient pharmacy data. Measures for SCI/D injury level, date of onset, and etiology were obtained from VA SCD Registry data.

Analysis

Incidence was calculated per 10,000 SCI/D patients treated each year instead of per 100,000 due to the size of the study population. Crude relationships between all variables and CA-CDI and CO-HCFA CDI were determined using unadjusted odds ratios. Cluster adjusted (within facility) logistic regression analyses were performed to further explore CA-CDI outcomes and compare with CO-HCFA cases. The final model included variables significant in bivariate analysis that remained significant in the model with p-value ≤ .05. Statistical analyses were conducted using SAS software version 9.3 and STATA MP Software version 12.1.

Results

We identified 2,239 CDI cases in SCI/D patients across the study period from inpatient, outpatient and extended care data from the VA sites included for study. A total of 229 cases (10.2%) occurred in community settings; where 145 (63.3%) were CA-CDI and 84 (36.7%) were CO-HCFA cases. Among the CO-HCFA CDI cases, average time to diagnosis of CDI after discharge from a HCF was 15.3 days (SD=7.8). Twenty-eight CA-CDI cases were classified as indeterminate, ⁷ but were similar to CA-CDI cases in patient demographics, severity, comorbidities, and proportion receiving antibiotics. In comparing the analysis results with (a) indeterminate cases included with CA-CDI cases, and (b) with the indeterminate cases excluded (data not shown), we found no differences in the final adjusted model, except in macrolide exposure, which was no longer significant when indeterminate cases were excluded. Thus, indeterminate cases were included in the CA-CDI group. Overall, 6.5% of cases (145/2,239) were CA-CDI. The CA-CDI cohort median age was 59 years, 97.9% of cases were male, nearly 85% was white, and approximately 52% of the cohort had paraplegia; the CO-HCFA cohort was similar.

The incidence of CA-CDI in our cohort was 16.4 per 10,000 patients and fluctuated annually over the study period, from 15.2 to 21.3 per 10,000 patients, although not statistically significant (global chi-square p-value=0.12) (Figure 1). The incidence rate was typically higher in the CA-CDI cohort compared with CO-HCFA CDI (9.5 per 10,000 patients). CO-HCFA incidence rates varied over time with a moderate increase from 8.1 to 10.7 per 10,000 patients (p=0.05). The CA-CDI cohort had less severe/severe complicated cases than the CO-HCFA group (28.3% vs. 42.9%, OR=0.53, 95% CI 0.30–0.92).

Figure 1.

Incidence rates per 10,000 patients, of community-associated C. difficile infection (CA-CDI) and community-onset, healthcare facility-associated (CO-HCFA) CDI among SCI/D Veterans, 2002–2009. CA-CDI incidence was 16.4 per 10,000 over study period (global chi-square p-value=0.12).

Table 1 describes unadjusted and multivariable analyses assessing the association between characteristics and CDI onset (CA-CDI or CO-HCFA). The prevalence of several comorbidities was higher in the CO-HCFA group than the CA-CDI group. In logistic regression analyses, the odds of aminoglycoside, third generation cephalosporins, extended spectrum penicillins, fluoroquinolones, macrolides, and PPI exposure was lower in CA-CDI than CO-HCFA CDI patients. Additionally, the odds of having a pressure ulcer in the 30 days preceding CDI onset was lower among CA-CDI cases.

Table 1.

Comparing CA-CDI and CO-HCFA CDI outcomes: Bivariate analysis of risk factors and multivariable logistic regression adjusted odds ratios.

	CA-CDI n=145 Frequency (%)	CO-HCFA CDI n=84 Frequency (%)	Unadjusted Odds Ratio (95% CI)	Adjusted Odds Ratio (95% CI)
Demographics
Age
18–49	27 (18.6)	9 (10.7)	Ref
50–64	70 (48.3)	41 (48.8)	0.57 (0.24, 1.33)
65+	48 (33.1)	34 (40.5)	0.47 (0.2, 1.13)
Gender
Male	142 (97.9)	82 (97.6)	1.15 (0.19, 7.05)
Female	3 (2.1)	2 (2.4)	Ref
Race
White	123 (84.8)	67 (79.8)	Ref
Black	15 (10.3)	14 (16.7)	0.58 (0.27, 1.28)
Other	7 (4.8)	3 (3.6)	1.27 (0.32, 5.08)
Region
Northeast	28 (19.3)	20 (23.8)	Ref
Midwest	23 (15.9)	26 (31)	0.63 (0.28, 1.41)
South	51 (35.2)	22 (26.2)	1.66 (0.77, 3.54)
West	43 (29.7)	16 (19)	1.92 (0.85, 4.32)
Duration of Injury
< 5 years	39 (26.9)	20 (23.8)	Ref
5–11 years	21 (14.5)	11 (13.1)	0.98 (0.4, 2.43)
11–16 years	8 (5.5)	5 (6)	0.82 (0.24, 2.84)
≥ 16 years	64 (44.1)	37 (44)	0.89 (0.45, 1.74)
Unknown	13 (9)	11 (13.1)	0.61 (0.23, 1.59)
Level of Injury
Tetraplegia	69 (47.6)	31 (36.9)	Ref
Paraplegia	76 (52.4)	51 (60.7)	0.67 (0.39, 1.16)
Unknown	0 (0)	2 (2.4)	0.09 (0, 1.94)
Comorbidities with at least 10% prevalence a
Had within 30 days:
Any Cardiovascular Disease/Disorder b	7 (4.8)	14 (16.7)	0.25 (0.10, 0.66)
COPD	7 (4.8)	21 (25.0)	0.15 (0.06, 0.38)
Diabetes	15 (10.3)	20 (23.8)	0.37 (0.18, 0.77)
Pressure Ulcer	24 (16.6)	28 (33.3)	0.4 (0.21, 0.75)	0.24 (0.08, 0.71)
Received diagnosis within study period:
Cancer	37 (25.5)	19 (22.6)	1.17 (0.62, 2.21)
COPD	44 (30.3)	36 (42.9)	0.58 (0.33, 1.02)
Any Cardiovascular Disease/Disorder b	54 (37.2)	38 (45.2)	0.72 (0.42, 1.24)
Diabetes	40 (27.6)	25 (29.8)	0.9 (0.5, 1.63)
Liver disease	24 (16.6)	14 (16.7)	0.99 (0.48, 2.04)
Pressure ulcer	79 (54.5)	49 (58.3)	0.85 (0.5, 1.47)
Renal disease	22 (15.2)	12 (14.3)	1.07 (0.5, 2.3)
Gastrointestinal Ulcer	14 (9.7)	11 (13.1)	0.71 (0.31, 1.64)
Medication exposure c
Antibiotics in last 6 weeks prior to CDI
No	67 (46.2)	6 (7.1)	Ref
Yes	78 (53.8)	78 (92.9)	0.09 (0.04, 0.22)
Aminoglycosides	2 (1.4)	11 (13.1)	0.09 (0.02, 0.43)
Ampicillin	3 (2.1)	10 (11.9)	0.16 (0.04, 0.59)
Amino-penicillin	17 (11.7)	12 (14.3)	0.8 (0.36, 1.76)	0.05 (0.01, 0.43)
Carbapenems	3 (2.1)	12 (14.3)	0.13 (0.03, 0.46)
Clindamycin	5 (3.4)	12 (14.3)	0.21 (0.07, 0.63)
Extended spectrum penicillins	11 (7.6)	34 (40.5)	0.12 (0.06, 0.26)	0.11 (0.03, 0.37)
Fluoroquinolones	38 (26.2)	45 (53.6)	0.31 (0.17, 0.54)	0.33 (0.14, 0.79)
Third generation cephalosporins	6 (4.1)	23 (27.4)	0.11 (0.04, 0.3)	0.20 (0.06, 0.71)
Macrolides	1 (0.7)	13 (15.5)	0.04 (0.0, 0.3)	0.02 (0.001, 0.44)
PPI	48 (33.1)	48 (57.1)	0.37 (0.21, 0.65)	0.33 (0.12, 0.87)
H-2 Blockers	13 (9)	16 (19)	0.42 (0.19, 0.92)
PPI or H-2 Blockers	58 (40)	56 (66.7)	0.33 (0.19, 0.58)

^aCompared to those without at least 10% prevalence.

^bIncludes congestive heart failure, cerebrovascular accident, myocardial infarction, and atherosclerotic peripheral vascular disease.

^cCompared to those without exposure to that particular medication.

Discussion

Veterans with SCI/D represent a unique population segment with complex, chronic needs. This is evident in that the proportion of CDI cases are predominately HCFA in this population; this is contrary to recent data showing that onset of 75% of HCFA CDI cases was outside of hospitals.⁴ However, our findings on the proportion of CA-CDI cases, risk factors and infection severity were generally consistent with other studies exploring CA-CDI in non-SCI/D populations.^2–4 Patients with CA-CDI had less comorbidity and fewer cases of severe CDI, compared to CO-HCFA cases. Pressure ulcers, while common in both groups, were more frequent among CO-HCFA cases, consistent with prior research.⁶ Further examination is needed to understand the relationship between CA-CDI and other comorbidities.

Patients with CA-CDI were less likely to be exposed to antibiotics compared to CO-HCFA cases. Prior studies have found that over 30% of CA-CDI patients had no recent exposure to antibiotics.^{2, 10} We were unable to identify recent antibiotic exposure in 46.2% of our CA-CDI sample. Thus, additional risk factors may exist for the SCI/D community. These data, while consistent with CA-CDI studies, are surprising considering the strong historical association of antibiotics with CDI. The role of antibiotic exposure as a CA-CDI risk factor requires further study.

Study limitations include use of a laboratory-based definition to confirm CDI diagnosis rather than a clinical diagnosis of CDI symptoms. Additionally, analyses included only VA laboratory data and VA-prescribed antibiotics. We did not account for care or antibiotic treatment received outside of VA facilities. Consequently, this may have underestimated CDI rates and antibiotic exposure.

Effectively preventing CDI in communities requires a thorough understanding of the infection’s impact. This study helps build our knowledge base, underscoring the need for further research on the effects of CA-CDI on the SCI/D population.