Abstract
Objectives:
To describe the burden of extended-spectrum beta-lactamase (ESBL) Enterobacteriaceae in Veterans with spinal cord injury/disorder (SCI/D), identify risk factors for ESBL acquisition, and assess impact on clinical outcomes.
Design:
Retrospective case-case-control
Patients and setting:
Veterans with SCI/D and utilization at a Veteran’s Affairs medical center from January 1, 2012-December 31, 2013.
Methods:
Cases had a positive culture for ESBL Klebsiella pneumoniae, Escherichia coli, or Proteus mirabilis and were matched to patients with non-ESBL organisms by organism, facility, and level of care and to uninfected controls by facility and level of care. Inpatients were also matched by time at risk. Univariable and multivariable matched models assessed for differences in risk factors and outcomes.
Results:
492 cases (62.6% outpatients) were matched 1:1 with both comparison groups. Recent prior use of 3rd/4th generation cephalosporins and fluoroquinolones were independently associated with ESBL compared to non-ESBL [adjusted odds ratio (aOR) 3.86, 95% confidence interval (CI) 2.06-7.25, p<0.001 and aOR 2.61, 95% CI 1.77-3.84, p<0.001, respectively] and control (aOR 3.31, 95% CI 1.56-7.06, p=0.002 and aOR 2.10, 95% CI 1.29-3.43, p=0.003, respectively) groups. Although there were no mortality differences, the ESBL group had longer post-culture length of stay (LOS) than the non-ESBL group (incidence rate ratio 1.36, 95% CI 1.13-1.63, p=0.001).
Conclusions:
All SCI/D patients with ESBL were more likely to have recent exposure to fluoroquinolones and 3rd/4th generation cephalosporins and hospitalized patients were more likely to have increased post-culture LOS. Programs targeted toward reduced antibiotic use in SCI/D patients may prevent subsequent ESBL acquisition.
Introduction
The prevalence of infections caused by multidrug-resistant organisms (MDROs) has been steadily increasing in both healthcare and community settings.1–3 In particular, Enterobacteriaceae that produce extended-spectrum β-lactamase (ESBL) enzymes have rapidly proliferated4 and now account for 18.6% of gram-negative organisms isolated from patients in U.S. intensive care units (ICUs).1 Infection with ESBL bacteria and the associated delay in effective antimicrobial therapy often leads to increased length of stay (LOS), higher healthcare costs, and increased mortality.5,6
Patients with spinal cord injury/disorder (SCI/D) have an increased risk of infection compared to the general patient population due to frequent healthcare contact, comorbidities, and use of invasive medical devices.7–11 Furthermore, rehabilitation hospitals and long-term care facilities (LTCF) have become important reservoirs for MDROs.12,13 Given the frequency with which SCI/D patients require admission to these types of facilities, the threat of infection with MDROs remains a significant burden in this population.
Studies describing the prevalence, risk factors for acquisition, and outcomes of MDRO infections in general acute care facilities may not adequately reflect the SCI/D population.14 A few small studies have shown high rates of MDROs, including ESBL, in SCI/D patients in rehabilitation hospitals.15–17 However, data are limited on the burden and outcomes of infection or colonization with multidrug resistant gram-negative organisms (MDRGNO) among patients with chronic SCI/D, a population likely to have greater risk compared to those acutely injured due to repeated healthcare exposures. Likewise, little data exist on the prevalence of MDRGNOs among SCI/D patients across a range of healthcare settings.
In this study, we investigated the prevalence of ESBL-producing Enterobacteriaceae, identified risk factors for ESBL acquisition, and examined clinical outcomes in a large population of Veterans with SCI/D. We used a case-case-control design with three comparison groups: 1) ESBL cases; 2) patients with non-ESBL organisms; and, 3) uninfected controls. This allowed us to analyze predictors specific to acquisition of the ESBL resistant phenotype, rather than just acquisition of Enterobacteriaceae.
Methods
Study setting and design
This was a retrospective case-case-control study involving adult SCI/D patients treated at VA medical facilities between January 1, 2012 and December 31, 2013. Clinical data was collected retrospectively to January 1, 2011 for risk factor analysis. The cohort was drawn from a cumulative list of Veterans with SCI/D maintained by the VA Allocation Resource Center.18 Veterans with multiple sclerosis, amyotrophic lateral sclerosis, and Guillain-Barre syndrome were excluded because the VA SCI/D system of care focuses on individuals with stable non-progressive spinal cord neurological deficits. National VA datasets were used to identify cases and controls and collect clinical data.
Cases had a positive culture for ESBL-producing Escherichia coli, Klebsiella pneumoniae, or Proteus mirabilis. These organisms were chosen because they are among the most common gram-negative bacteria isolated from SCI/D patients and frequently produce ESBLs.19–21 Cases included patients with cultures performed in any healthcare setting (inpatient, outpatient, rehabilitation, LTCF) from any site except rectal screening cultures. Thus, both infected and silently colonized patients were included. Patients with cultures positive for > 1 organism were included if at least one organism was identified as E. coli, K. pneumoniae, or P. mirabilis. Only the first positive culture was included if patients had more than one ESBL isolate identified during the study period.
For all analyses, ESBL cases were matched in a 1:1 ratio with two comparison groups: 1.) SCI/D patients with non-ESBL E. coli, K. pneumoniae, and P. mirabilis; and 2.) uninfected SCI/D controls. Controls were identified by the absence of cultures positive for Enterobacteriaceae and the absence of an International Classification of Disease-Clinical Modification, 9th revision (ICD9-CM) code for an infection. All case patients were matched to group 1 (non-ESBL) by organism, facility, and level of care (outpatient vs. inpatient vs. LTCF), and to group 2 (control) by facility and level of care. Additionally, inpatients and residents of LTCFs were matched by time at risk, which was defined as the number of days from admission to the index positive culture date and was matched within ± 60 days. This flexibility in matching was required because SCI/D patients frequently have long inpatient, rehabilitation, and LTCF stays. The institutional review board at the Edward Hines, Jr. VA Hospital approved this study.
Clinical and microbiology data collection
Patient demographics, characteristics, and comorbid conditions were collected from national VA datasets, including the Veterans’ Health Administration (VHA) Corporate Data Warehouse (CDW). These datasets were also used to gather information on healthcare and antibiotic exposures and clinical outcomes. The modified Charlson comorbidity index at the time of culture for ESBL and non-ESBL groups and at the index outpatient encounter or admission date for controls was also calculated.22
To determine ESBL cases, we obtained information from the CDW on all bacterial cultures for which antibiotic susceptibility testing was performed. Given that each local VA laboratory may have entered results of ESBL testing into the electronic medical record in slightly different places, we used both microbiology and general laboratory domains of the CDW to identify positive ESBL results. Medical charts were reviewed from a random subset of 100 patients in each group to validate the administrative data with regard to accurate assignment of patients into ESBL, non-ESBL, and control groups.
Statistical analysis
Paired t-tests and Wilcoxon sign-rank tests were used to compare continuous variables and univariable logistic regression was used for categorical variables to identify risk factors for ESBL and non-ESBL organisms. Bonferroni correction for multiple comparisons was applied, with p < 0.017 considered significant. Variables with a p-value < 0.05 or a 95% CI not including 1 were considered for inclusion in multivariable matched logistic regression analyses. Separate analyses were conducted comparing each group to the others. Variables were added with stepwise selection and the variance inflation factor was used to assess for multicollinearity. Multivariable adjusted logistic regression models were created to examine the association between ESBL and clinical outcomes with the exception of post-culture LOS, which was assessed with a multivariable adjusted negative binomial model. For all multivariable models, the individual clinical variables that make up the modified Charlson score were used to examine associations between individual comorbidities and outcomes. To avoid being overly conservative and increasing Type II error, correction for multiple comparisons was not applied to multivariable analyses, and p < 0.05 was considered statistically significant. Statistical analyses were carried out using SAS, version 9.4 (SAS Institute) and Stata, version 12.1 (Stata Corp LP).
Results
A total of 19,665 Veterans with SCI/D were eligible for inclusion, of which 13,862 (70.5%) had bacterial cultures performed during the study period. Among these, 7,067 (51.0%) had a positive culture for E. coli, K. pneumoniae, and/or P. mirabilis with 745 (10.5%) patients having ESBL-producing organisms. We successfully matched 492 of 745 (66.0%) ESBL cases to both comparison groups for a final cohort of 1,476 patients. Within this cohort, there were 924 (62.6%) outpatients, 528 (35.8%) inpatients, and 24 (1.6%) in rehabilitation units or LTCFs. Seventy-five percent (n=1,107) of visits or admissions were at a VA SCI specialty center. For the ESBL and non-ESBL groups, urine was the most common culture site (n=791, 80.4%) followed by blood (n=159, 16.2%), respiratory (n=6, 0.6%), and other sites (n=28, 2.8%). Among ESBL cases, 245 (49.8%) had positive cultures for E. coli, 208 (42.3%) for K. pneumoniae, 17 (3.5%) for P. mirabilis, and 22 (4.5%) for > 1 organism.
Table 1 summarizes demographic and clinical characteristics and the results of the univariable analysis. Patients in the ESBL group had higher median Charlson comorbidity index scores. Prior healthcare exposures including hospital admission, surgery, ICU admission, and mechanical ventilation were associated with ESBL but not non-ESBL organisms. Fluoroquinolones were the most common prior antibiotic exposure for all three groups, and exposures to a number of antibiotics were associated with increased odds of ESBL (Table 1).
Table 1.
Univariable analysis of risk factors for acquisition of non-ESBL and ESBL-producing Enterobacteriaceae.
| No. (%)a | Matched OR (95%CI) P value* |
|||||
|---|---|---|---|---|---|---|
| Variable | Control | Non-ESBL | ESBL | Non-ESBL vs. Control | ESBL vs. Control | ESBL vs. Non-ESBL |
| Demographics | ||||||
| Age, years, mean (SD) | 61.2 (13.7) | 60.9 (13.5) | 61.5 (14.4) | p=0.80 | p=0.71 | p=0.53. |
| Sex, female | 19 (3.9) | 13 (2.6) | 10 (2.0) | 0.67 (0.33-1.38) p=0.28 |
0.50 (0.22-1.11) p=0.09 |
0.77 (0.34-1.75) p=0.53 |
| Comorbidities | ||||||
| SCI levelb | ||||||
| Tetraplegia | 199 (40.4) | 228 (46.3) | 233 (47.4) | Reference | Reference | Reference |
| Paraplegia | 154 (31.3) | 229 (46.5) | 217 (44.1) | 1.30 (0.97-1.74) p=0.07 |
1.22 (0.91-1.63) p=0.19 |
0.92 (0.71-1.20) p=0.55 |
| SCI onsetb | ||||||
| Non-traumatic | 135 (26.9) | 119 (23.8) | 136 (27.1) | Reference | Reference | Reference |
| Traumatic | 228 (46.3) | 325 (66.1) | 311 (63.2) |
1.70 (1.23-2.35)
p=0.001 |
1.32 (0.96-1.81) p=0.09 |
0.83 (0.61-1.13) p=0.23 |
| SCI extentb | ||||||
| Incomplete | 265 (52.9) | 269 (53.7) | 215 (42.9) | Reference | Reference | Reference |
| Complete | 69 (14) | 183 (37.2) | 225 (45.7) |
2.78 (1.94-3.99)
p<0.001 |
4.31 (2.94-6.31)
p<0.001 |
1.50 (1.16-1.96)
p=0.002 |
| SCI durationb | ||||||
| 0-10 years | 174 (35.4) | 182 (37.0) | 201 (40.9) | Reference | Reference | Reference |
| 11-20 years | 51 (10.4) | 77 (15.7) | 80 (16.3) | 1.39 (0.89-2.17) p=0.15 |
1.36 (0.89-2.06) p=0.15 |
0.94 (0.65-1.36) p=0.74 |
| >20 years | 118 (24.0) | 184 (37.4) | 156 (31.7) | 1.49 (1.07-2.06) p=0.02 |
1.09 (0.79-1.50) p=0.61 |
0.77 (0.58-1.03) p=0.08 |
| Charlson comorbidity index, median (range) | 2 (0-20) | 2 (0-12) | 3 (0-14) | p=0.06 | p<0.001 | p<0.001 |
| Gastrostomy or jejunostomy | 7 (1.4) | 8 (1.6) | 18 (3.7) | 1.14 (0.41-3.15) p=0.80 |
2.57 (1.07-6.16) p=0.03 |
2.43 (1.01-5.86) p=0.05 |
| Chronic kidney disease | 46 (9.3) | 38 (7.7) | 60 (12.2) | 0.81 (0.52-1.27) p=0.36 |
1.34 (0.9-2.01) p=0.16 |
1.65 (1.08-2.52) p=0.02 |
| Chronic liver disease | 39 (7.9) | 23 (4.7) | 37 (7.5) | 0.58 (0.34-0.98) p=0.04 |
0.94 (0.59-1.52) p=0.81 |
1.64 (0.96-2.78) p=0.07 |
| AIDS | 9 (1.8) | 1 (0.2) | 5 (1.0) | 0.11 (0.01-0.88) p=0.04 |
0.56 (0.19-1.66) p=0.29 |
5.00 (0.58-42.80) p=0.14 |
| Malignancy or tumor | 57 (11.6) | 44 (8.9) | 49 (10) | 0.74 (0.48-1.13) p=0.17 |
0.84 (0.56-1.27) p=0.41 |
1.12 (0.74-1.71) p=0.59 |
| CHF | 36 (7.3) | 32 (6.5) | 37 (7.5) | 0.88 (0.53-1.45) p=0.61 |
1.03 (0.65-1.64) p=0.91 |
1.18 (0.71-1.95) p=0.52 |
| Diabetes | 117 (23.8) | 138 (28.0) | 180 (36.6) | 1.24 (0.94-1.65) p=0.13 |
1.89 (1.42-2.52)
p<0.001 |
1.54 (1.16-2.05)
p=0.003 |
| Cerebrovascular disease | 39 (7.9) | 51 (10.4) | 41 (8.3) | 1.32 (0.86-2.01) p=0.20 |
1.06 (0.66-1.69) p=0.81 |
0.78 (0.51-1.21) p=0.27 |
| Peripheral vascular disease | 39 (7.9) | 38 (7.7) | 75 (15.2) | 0.97 (0.61-1.55) p=0.91 |
2.16 (1.41-3.31)
p<0.001 |
2.28 (1.47-3.52)
p<0.001 |
| COPD | 101 (20.5) | 69 (14.0) | 96 (19.5) |
0.64 (0.46-0.89)
p=0.01 |
0.93 (0.67-1.29) p=0.68 |
1.51 (1.07-2.14) p=0.02 |
| Pressure ulcer | 71 (14.4) | 162 (32.9) | 268 (54.5) |
2.82 (2.04-3.89)
p<0.001 |
6.79 (4.74-9.74)
p<0.001 |
2.47 (1.88-3.25)
p<0.001 |
| Healthcare exposures in past 90 days | ||||||
| Hospital admission | 108 (22.0) | 96 (19.5) | 170 (34.6) | 0.85 (0.61-1.17) p=0.32 |
1.87 (1.4-2.5)
p<0.001 |
2.42 (1.75-3.35)
p<0.001 |
| LTCF or rehabilitation stay | 15 (3.0) | 7 (1.4) | 16 (3.3) | 0.38 (0.14-1.08) p=0.07 |
1.07 (0.53-2.16) p=0.86 |
2.5 (0.97-6.44) p=0.06 |
| Surgery | 34 (6.9) | 25 (5.1) | 55 (11.2) | 0.69 (0.39-1.22) p=0.20 |
1.75 (1.1-2.78)
p=0.02 |
2.76 (1.59-4.81)
p<0.001 |
| GU procedurec | 17 (3.5) | 22 (4.5) | 29 (5.9) | 1.31 (0.68-2.52) p=0.41 |
1.80 (0.96-3.38) p=0.07 |
1.41 (0.76-2.63) p=0.28 |
| ICU admission | 23 (4.7) | 20 (4.1) | 44 (8.9) | 0.85 (0.45-1.62) p=0.62 |
2.11 (1.22-3.63)
p=0.008 |
2.50 (1.40-4.46)
p=0.002 |
| Mechanical ventilation | 10 (2.0) | 5 (1.0) | 22 (4.5) | 0.50 (0.17-1.46) p=0.21 |
2.33 (1.07-5.09) p=0.03 |
5.25 (1.8-15.29)
p=0.002 |
| Medication exposures in past 90 days | ||||||
| Any antibiotic | 148 (30.1) | 185 (37.6) | 287 (58.3) |
1.40 (1.07-1.82)
p=0.01 |
3.21 (2.42-4.25)
p<0.001 |
2.44 (1.85-3.21)
p<0.001 |
| Chronic steroidsd | 6 (1.2) | 2 (0.4) | 6 (1.2) | 0.33 (0.07-1.65) p=0.18 |
1.00 (0.32-3.10) p=1.00 |
3.00 (0.61-14.86) p=0.18 |
| Penicillins | 24 (4.9) | 36 (7.3) | 57 (11.6) | 1.55 (0.90-2.64) p=0.11 |
2.57 (1.55-4.26)
p<0.001 |
1.72 (1.09-2.72) p=0.02 |
| Extended-spectrum penicillins | 23 (4.7) | 26 (5.3) | 56 (11.4) | 1.14 (0.64-2.02) p=0.66 |
2.65 (1.58-4.43)
p<0.001 |
2.50 (1.49-4.20)
p=0.001 |
| 1st/2nd gen cephalosporins | 27 (5.5) | 35 (7.1) | 36 (7.3) | 1.35 (0.79-2.31) p=0.28 |
1.35 (0.81-2.24) p=0.25 |
1.03 (0.64-1.66) p=0.90 |
| 3rd/4th gen cephalosporins | 19 (3.8) | 17 (3.4) | 80 (16) | 0.89 (0.47-1.72) p=0.73 |
5.07 (2.91-8.81)
p<0.001 |
5.50 (3.11-9.72)
P<0.001 |
| Carbapenems | 7 (1.4) | 6 (1.2) | 28 (5.7) | 0.86 (0.29-2.55) p=0.78 |
4.50 (1.86-10.9)
p=0.001 |
6.50 (2.27-18.62)
p<0.001 |
| Macrolides | 13 (2.6) | 13 (2.6) | 13 (2.6) | 1.00 (0.46-2.16) p=1.00 |
1.00 (0.46-2.16) p=1.00 |
1.00 (0.46-2.16) p=1.00 |
| Tetracyclines | 9 (1.8) | 5 (1) | 18 (3.6) | 0.56 (0.19-1.66) p=0.29 |
2.00 (0.90-4.45) p=0.09 |
3.60 (1.34-9.70)
p=0.01 |
| Aminoglycosides | 7 (1.4) | 8 (1.6) | 14 (2.8) | 1.14 (0.41-3.15) p=0.80 |
2.00 (0.75-5.33) p=0.17 |
1.77 (0.74-4.27) p=0.20 |
| Fluoroquinolones | 59 (12.0) | 52 (10.6) | 142 (28.9) | 0.87 (0.58-1.29) p=0.48 |
3.18 (2.21-4.58)
p<0.001 |
3.31 (2.31-4.73)
p<0.001 |
| Vancomycin | 38 (7.7) | 28 (5.7) | 78 (15.9) | 0.70 (0.41-1.19) p=0.18 |
2.38 (1.54-3.67)
p<0.001 |
3.94 (2.31-6.71)
p<0.001 |
| Clindamycin | 6 (1.2) | 8 (1.6) | 8 (1.6) | 1.33 (0.46-3.84) p=0.59 |
1.33 (0.46-3.84) p=0.59 |
1.00 (0.35-2.85) p=1.00 |
| Nitrofurans | 16 (3.3) | 22 (4.5) | 43 (8.7) | 1.38 (0.72-2.62) p=0.33 |
2.80 (1.55-5.05)
p=0.001 |
2.31 (1.29-4.16)
p=0.005 |
| Sulfonamides | 30 (6.1) | 35 (7.1) | 56 (11.4) | 1.20 (0.71-2.04) p=0.50 |
2.04 (1.26-3.29)
p=0.003 |
1.66 (1.07-2.57) p=0.02 |
| Metronidazole | 6 (1.2) | 14 (2.8) | 36 (7.3) | 2.33 (0.90-6.07) p=0.08 |
7.00 (2.74-17.87)
p<0.001 |
2.70 (1.43-5.06)
p=0.002 |
| Methenamine | 4 (0.8) | 26 (5.3) | 18 (3.7) |
6.50 (2.27-18.62)
p<0.001 |
4.50 (1.52-13.30)
p=0.007 |
0.67 (0.35-1.25) p=0.21 |
| Rifamycin | 4 (0.8) | 2 (0.4) | 2 (0.4) | 0.50 (0.09-2.73) p=0.42 |
0.50 (0.09-2.73) p=0.42 |
1.00 (0.14-7.10) p=1.00 |
ESBL, extended-spectrum beta-lactamase; OR, odds ratio; CI, confidence interval; SD, standard deviation; SCI, spinal cord injury; AIDS, acquired immune deficiency syndrome; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; LTCF, long term care facility; GU, genitourinary; ICU, intensive care unit; gen, generation;
All data displayed are number (%) unless otherwise indicated
Some data missing for this variable
Minimally invasive or non-invasive GU procedures
Defined as ≥ 85 days of use in the prior 90 days
Paired t-test was used for age and Wilcoxon sign-rank test was used for Charlson score
Matched multivariable analyses for acquisition of ESBL are displayed in Table 2. When compared with uninfected controls, the following variables were significant independent predictors of ESBL: diabetes, complete SCI, pressure ulcer, prior use of 3rd/4th generation cephalosporins, and prior use of fluoroquinolones. However, diabetes, complete SCI, and pressure ulcer were also associated with increased odds of non-ESBL organisms. Therefore, the only independent predictors of ESBL, but not non-ESBL, organisms were recent prior use of 3rd/4th generation cephalosporins and fluoroquinolones.
Table 2.
Multivariable conditional logistic regression analysis of risk factors for acquisition of non-ESBL and ESBL-producing Enterobacteriaceae.
| Non-ESBL vs. Control | ESBL vs. Control | ESBL vs. Non-ESBL | ||||
|---|---|---|---|---|---|---|
| aOR (95% CI) | P | aOR (95% CI) | P | aOR (95% CI) | P | |
| Variable | ||||||
| Diabetes | 1.35 (0.95-1.90) | 0.09 | 1.63 (1.10-2.42) | 0.02 | 1.26 (0.91-1.74) | 0.17 |
| Complete SCI | 2.69 (1.83-3.94) | <0.001 | 3.15 (2.02-4.91) | <0.001 | 1.34 (0.99-1.82) | 0.06 |
| Pressure ulcer | 2.52 (1.72-3.70) | <0.001 | 4.15 (2.77-6.21) | <0.001 | 2.07 (1.53-2.80) | <0.001 |
| Exposures in the prior 90 days | ||||||
| Mechanical ventilation | 0.12 (0.03-0.47) | 0.003 | 1.81 (0.59-5.60) | 0.30 | 3.46 (0.97-12.35) | 0.06 |
| 3rd/4th gen cephalosporins | 0.65 (0.29-1.45) | 0.29 | 3.31 (1.56-7.06) | 0.002 | 3.86 (2.06-7.25) | <0.001 |
| Fluoroquinolones | 0.75 (0.46-1.23) | 0.26 | 2.10 (1.29-3.43) | 0.003 | 2.61 (1.77-3.84) | <0.001 |
ESBL, extended-spectrum beta-lactamase; aOR, adjusted odds ratio; CI, confidence interval; SCI, spinal cord injury
The univariable analysis of clinical outcomes stratified by comparison group is shown in Table 3. Patients in both the non-ESBL and ESBL groups had lower 30-day mortality and greater odds of hospital readmission within 90 days than controls but there was no difference between non-ESBL and ESBL groups. Inpatients with ESBL had longer post-culture LOS than the non-ESBL group. In adjusted multivariable models, ESBL was not an independent predictor of increased 30-day mortality (Table 4). Similarly, patients in the ESBL group did not have increased 1-year mortality compared with the control group [adjusted odds ratio (aOR) 0.45, 95% confidence interval (CI) 0.28-0.71] or the non-ESBL group (aOR 0.87, 95% CI 0.54-1.39). Inpatients with ESBL had greater odds of hospital readmission within 90 days as compared to the control group (aOR 3.73, 95% CI 2.20-6.32) and the non-ESBL group (aOR 1.56, 95% CI 0.99-2.47), although this did not reach statistical significance. Finally, ESBL was significantly associated with increased post-culture LOS among inpatients (Table 5).
Table 3.
Clinical outcomes stratified by comparison group.
| No. (%) | OR (95% CI) P value |
|||||
|---|---|---|---|---|---|---|
| Outcome | Control | Non-ESBL | ESBL | Non-ESBL vs. Control | ESBL vs. Control | ESBL vs. Non-ESBL |
| 30-day mortality | 27 (5.5) | 11 (2.2) | 11 (2.2) | 0.36 (0.17-0.77) p=0.009 |
0.41 (0.20-0.82) p=0.01 |
1.00 (0.42-2.40) p=1.00 |
| 1-year mortality | 72 (14.6) | 46 (9.3) | 56 (11.4) | 0.58 (0.39-0.88) p=0.009 |
0.74 (0.50-1.08) p=0.12 |
1.26 (0.83-1.93) p=0.28 |
| Post-culture LOS, days, median (range)a | -- | 11 (0-419) | 22 (0-985) | -- | -- | p=0.001 |
| 90-day hospital readmissiona | 28 (5.7) | 50 (10.2) | 66 (13.4) | 2.00 (1.20-3.34) p=0.008 |
3.38 (1.93-5.90) p<0.001 |
1.44 (0.94-2.21) p=0.09 |
OR, odds ratio; CI, confidence interval; ESBL, extended-spectrum beta-lactamase; LOS, length of stay
For inpatients only
Table 4.
Multivariable logistic regression analysis of variables associated with 30-day mortality.
| Variable | aOR (95% CI) | P value |
|---|---|---|
| ESBL (ref: control group) | 0.14 (0.05-0.40) | <0.001 |
| ESBL (ref: non-ESBL group) | 0.42 (0.14-1.23) | 0.11 |
| Patient location (ref: outpatient) | ||
| Inpatient | 2.82 (1.20-6.59) | 0.02 |
| LTCF | 3.10 (0.43-22.22) | 0.26 |
| Culture source (ref: urine) | ||
| Blood | 4.63 (1.88-11.42) | <0.001 |
| Other | 2.50 (0.92-6.82) | 0.07 |
| Traumatic SCI | 0.28 (0.11-0.70) | 0.006 |
| Malignancy or tumor | 7.22 (3.40-15.34) | <0.001 |
| Renal disease | 2.44 (1.09-5.47) | 0.03 |
| Pressure ulcer | 3.30 (1.49-7.31) | 0.003 |
| Exposures in prior 90 days | ||
| Mechanical ventilation | 9.68 (2.89-32.35) | <0.001 |
| Sulfonamides | 0.06 (0.01-0.39) | 0.004 |
| Vancomycin | 3.50 (1.42-8.65) | 0.007 |
aOR, adjusted odds ratio; CI, confidence interval; ESBL, extended-spectrum beta-lactamase; LTCF, long-term care facility; SCI, spinal cord injury
Table 5.
Multivariable negative binomial regression analysis of variables associated with increased post-culture hospital length of stay for inpatients.
| Variable | IRR (95% CI) | P value |
|---|---|---|
| ESBL (ref: non-ESBL) | 1.36 (1.13-1.63) | 0.001 |
| Patient seen at SCI center | 1.60 (1.19-2.13) | 0.002 |
| Exposures in prior 90 days | ||
| Mechanical ventilation | 1.86 (1.26-2.73) | 0.002 |
| Vancomycin | 1.29 (0.99-1.68) | 0.06 |
| Macrolides | 0.49 (0.27-0.86) | 0.01 |
IRR, incidence rate ratio; CI, confidence interval; ESBL, extended-spectrum beta-lactamase; SCI, spinal cord injury
Chart reviews were conducted on a random subset of 100 patients from each group to estimate the validity of the clinical and administrative data in assigning patients to comparison groups. No patients in the non-ESBL group actually had ESBL organisms but one patient in the ESBL group had a non-ESBL organism. Six out of 100 (6.0%) uninfected control patients were actually infected based on presence of signs and/or symptoms identified via chart review.
Discussion
This study represents a large and comprehensive evaluation of risks for and clinical outcomes of ESBL acquisition in SCI/D patients. This population deserves specific attention due to increased comorbidities and frequent use of healthcare services and indwelling devices--factors that broadly increase risk for many types of MDROs.23 The VHA treats more than 26,000 patients with SCI/D every year in multiple care settings24, making VA data a robust resource for studying this population.
Few studies have characterized MDRGNO epidemiology in SCI/D patients. Prior work by our group showed that hospital-acquired infections in SCI/D patients are frequently caused by methicillin-resistant Staphylococcus aureus, vancomycin-resistant Enterococcus faecium, and Pseudomonas, suggesting common risks for many MDROs in this population.7 Other studies have been limited by small size,17,25–27 inclusion of only urinary isolates,16,21,25,27 and imprecise definitions of antimicrobial resistance.17,26 Our study included all cultures collected from a large SCI/D population and identified laboratory-confirmed ESBL cases. We found that prior use of fluoroquinolones and 3rd/4th generation cephalosporins were significant independent predictors of ESBL but not non-ESBL organisms. Although prior antibiotic use is a well-recognized risk factor for ESBL,2,12,28,29 some studies have had conflicting results. Many of these conflicting studies did not examine individual antibiotic classes and, thus, may have missed associations with specific antibiotics not reflected in overall antibiotic exposure.30–32
Frequency and type of antibiotic use in SCI/D patients is likely different from that observed in general patient populations. We have previously identified increased antibiotic prescribing for SCI/D patients in the Emergency Department, and higher prescribing for patients seen at non-SCI centers.33 These patients’ multiple comorbidities may create more opportunities for inappropriate antibiotic use, especially among providers less familiar with SCI/D. One circumstance may be treatment of asymptomatic catheter-associated bacteriuria, a practice not recommended due to lack of efficacy and emergence of resistance.34 Waites et al. found that administration of ciprofloxacin for 10 days in men with SCI/D who performed intermittent urinary catheterization resulted in a subsequent increase in resistant staphylococci, enterococci, and Acinetobacter spp.35 Clearly, these results, and those from our study, suggest that fluoroquinolone use strongly contribute to subsequent colonization and infection with MDROs in SCI/D patients.
In contrast to prior studies,6 we did not identify increased mortality in our ESBL group compared with controls. Our chart reviews estimated that 6% of patients classified into the control group were actually infected. This estimate is consistent with prior literature evaluating the validity of observational administrative data36,37, and suggests that our criteria for classifying patients performed well but not perfectly. Erroneously including enough infected patients in the control group may have contributed to a higher observed mortality than expected. Furthermore, our cohort included patients with ESBL isolated from any site and those who were colonized as well as infected, both factors that may have lowered the mortality observed in our ESBL group. Interestingly, we did observe increased post-culture hospital LOS for inpatients with ESBL, a finding previously reported in a small single-center study5 but not validated in a larger population until now.
Our study has a number of important limitations. First, it was subject to selection bias, particularly from the exclusion of ESBL patients who could not be matched. Despite this, our final cohort included a diverse population with a range of SCI severity, who sought care from multiple different settings, and who visited both SCI specialty centers and non-SCI centers. Second, we did not collect data on antibiotic treatment and, thus, could not analyze adequacy or timeliness of antibiotic therapy in regards to clinical outcomes. And finally, data were missing for some of the SCI characteristics such as level and extent of injury, which may have introduced bias into analyses including these variables.
In conclusion, we used a national cohort of Veterans to demonstrate that colonization and infection with ESBL organisms is common in SCI/D patients, particularly in the urine. Prior use of 3rd/4th generation cephalosporins and fluoroquinolones independently increases the odds of ESBL, but not non-ESBL organisms. Interventions that reduce inappropriate 3rd/4th generation cephalosporin and fluoroquinolone use will be particularly effective in decreasing ESBL acquisition in this population. Furthermore, given that ESBL was associated with longer post-culture hospital LOS, these same interventions are likely to have clinical and financial benefits on a broader, system-wide scale.
Acknowledgements:
We would like to thank Scott Miskevics for his assistance with programming and data analysis. This work was supported by The Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Rehabilitation Research and Development Service Merit Review Award (B-1583-P), Health Services Research and Development Service Post-Doctoral Fellowship Award (TPR 42-005), and the Spinal Cord Injury Quality Enhancement Research Initiative (SCI-98-001). Dr. Nasia Safdar is additionally supported by a VA-funded Patient Safety Center of Inquiry. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the U.S. government. All authors report no conflicts of interest or financial disclosures relevant to this article.
References:
- 1.Sader HS, Farrell DJ, Flamm RK, Jones RN. Antimicrobial susceptibility of gram-negative organisms isolated from patients hospitalized in intensive care units in United States and European hospitals (2009-2011). Diagn Microbiol Infect Dis 2014;78:443–448. [DOI] [PubMed] [Google Scholar]
- 2.Ben-Ami R, Rodriguez-Bano J, Arslan H, et al. A multinational survey of risk factors for infection with extended-spectrum beta-lactamase-producing Enterobacteriaceae in nonhospitalized patients. Clin Infect Dis 2009;49:682–690. [DOI] [PubMed] [Google Scholar]
- 3.Sievert DM, Ricks P, Edwards JR, et al. Antimicrobial-resistant pathogens associated with healthcare-associated infections: Summary of data reported to the National Healthcare Safety Network at the Centers for Disease Control and Prevention, 2009-2010. Infect Control Hosp Epidemiol 2013;34:1–14. [DOI] [PubMed] [Google Scholar]
- 4.Kallen AJ, Hidron AI, Patel J, Srinivasan A. Multidrug resistance among gram-negative pathogens that caused healthcare-associated infections reported to the National Healthcare Safety Network, 2006-2008. Infect Control Hosp Epidemiol 2010;31:528–531. [DOI] [PubMed] [Google Scholar]
- 5.Lee SY, Kotapati S, Kuti JL, Nightingale CH, Nicolau DP. Impact of extended-spectrum beta-lactamase-producing Escherichia coli and Klebsiella species on clinical outcomes and hospital costs: A matched cohort study. Infect Control Hosp Epidemiol 2006;27:1226–1232. [DOI] [PubMed] [Google Scholar]
- 6.Schwaber MJ, Carmeli Y. Mortality and delay in effective therapy associated with extended-spectrum beta-lactamase production in Enterobacteriaceae bacteraemia: A systematic review and meta-analysis. J Antimicrob Chemother 2007;60:913–920. [DOI] [PubMed] [Google Scholar]
- 7.Evans CT, LaVela SL, Weaver FM, et al. Epidemiology of hospital-acquired infections in Veterans with spinal cord injury and disorder. Infect Control Hosp Epidemiol 2008;29:234–242. [DOI] [PubMed] [Google Scholar]
- 8.Annual statistical report – complete public version. National Spinal Cord Injury Statistical Center, University of Alabama at Birmingham. https://http://www.nscisc.uab.edu/PublicDocuments/reports/pdf/2014NSCISCAnnualStatisticalReportCompletePublicVersion.pdf. Published 2014. Accessed September 24, 2015. [Google Scholar]
- 9.Nicolle LE, Buffet L, Alfieri N, Tate R. Nosocomial infections on a rehabilitation unit in an acute care hospital. Infect Control Hosp Epidemiol 1988;9:553–558. [DOI] [PubMed] [Google Scholar]
- 10.Mylotte JM, Graham R, Kahler L, Young BL, Goodnough S. Impact of nosocomial infection on length of stay and functional improvement among patients admitted to an acute rehabilitation unit. Infect Control Hosp Epidemiol 2001;22:83–87. [DOI] [PubMed] [Google Scholar]
- 11.LaVela SL, Evans CT, Miskevics S, Parada JP, Priebe M, Weaver FM. Long-term outcomes from nosocomial infections in persons with spinal cord injuries and disorders. Am J Infect Control 2007;35:393–400. [DOI] [PubMed] [Google Scholar]
- 12.Lim CJ, Cheng AC, Kennon J, et al. Prevalence of multidrug-resistant organisms and risk factors for carriage in long-term care facilities: A nested case-control study. J Antimicrob Chemother 2014;69:1972–1980. [DOI] [PubMed] [Google Scholar]
- 13.Ludden C, Cormican M, Vellinga A, Johnson JR, Austin B, Morris D. Colonisation with ESBL-producing and carbapenemase-producing Enterobacteriaceae, vancomycin-resistant enterococci, and meticillin-resistant Staphylococcus aureus in a long-term care facility over one year. BMC Infect Dis 2015;15:168. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.LaVela SL, Smith B, Weaver FM, Miskevics SA. Geographical proximity and health care utilization in veterans with SCI&D in the USA. Soc Sci Med 2004;59:2387–2399. [DOI] [PubMed] [Google Scholar]
- 15.Mylotte JM, Kahler L, Graham R, Young L, Goodnough S. Prospective surveillance for antibiotic-resistant organisms in patients with spinal cord injury admitted to an acute rehabilitation unit. Am J Infect Control 2000;28:291–297. [DOI] [PubMed] [Google Scholar]
- 16.Yoon SB, Lee BS, Lee KD, Hwang SI, Lee HJ, Han ZA. Comparison of bacterial strains and antibiotic susceptibilities in urinary isolates of spinal cord injury patients from the community and hospital. Spinal Cord 2014;52:298–301. [DOI] [PubMed] [Google Scholar]
- 17.Slim E, Smit CA, Bos AJ, Peerbooms PG. Nosocomial transmission of highly resistant microorganisms on a spinal cord rehabilitation ward. J Spinal Cord Med 2009;32:422–427. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Smith BM, Evans CT, Ullrich P, et al. Using VA data for research in persons with spinal cord injuries and disorders: Lessons from SCI QUERI. J Rehabil Res Dev 2010;47:679–688. [DOI] [PubMed] [Google Scholar]
- 19.Vasoo S, Barreto JN, Tosh PK. Emerging issues in gram-negative bacterial resistance: An update for the practicing clinician. Mayo Clin Proc 2015;90:395–403. [DOI] [PubMed] [Google Scholar]
- 20.Hawser SP, Badal RE, Bouchillon SK, et al. Susceptibility of gram-negative aerobic bacilli from intra-abdominal pathogens to antimicrobial agents collected in the United States during 2011. J Infect 2014;68:71–76. [DOI] [PubMed] [Google Scholar]
- 21.Waites KB, Chen Y, DeVivo MJ, Canupp KC, Moser SA. Antimicrobial resistance in gram-negative bacteria isolated from the urinary tract in community-residing persons with spinal cord injury. Arch Phys Med Rehabil 2000;81:764–769. [DOI] [PubMed] [Google Scholar]
- 22.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 1987;40:373–383. [DOI] [PubMed] [Google Scholar]
- 23.Safdar N, Maki DG. The commonality of risk factors for nosocomial colonization and infection with antimicrobial-resistant Staphylococcus aureus, Enterococcus, gram-negative bacilli, Clostridium difficile, and Candida. Ann Intern Med 2002;136:834–844. [DOI] [PubMed] [Google Scholar]
- 24.VA research on spinal cord injury. U.S. Department of Veteran’s Affairs, Office of Research and Development. http://www.research.va.gov/pubs/docs/va_factsheets/sci.pdf. Published 2015. Accessed August 12, 2015. [Google Scholar]
- 25.Martins CF, Bronzatto E, Neto JM, Magalhaes GS, D’Anconna CA, Cliquet A Jr. Urinary tract infection analysis in a spinal cord injured population undergoing rehabilitation--how to treat? Spinal Cord 2013;51:193–195. [DOI] [PubMed] [Google Scholar]
- 26.Girard R, Mazoyer MA, Plauchu MM, Rode G. High prevalence of nosocomial infections in rehabilitation units accounted for by urinary tract infections in patients with spinal cord injury. J Hosp Infect 2006;62:473–479. [DOI] [PubMed] [Google Scholar]
- 27.Cremet L, Bemer P, Rome J, et al. Outbreak caused by Proteus mirabilis isolates producing weakly expressed tem-derived extended-spectrum beta-lactamase in spinal cord injury patients with recurrent bacteriuria. Scand J Infect Dis 2011;43:957–961. [DOI] [PubMed] [Google Scholar]
- 28.Chopra T, Marchaim D, Johnson PC, et al. Risk factors for bloodstream infection caused by extended-spectrum beta-lactamase-producing Escherichia coli and Klebsiella pneumoniae: A focus on antimicrobials including cefepime. Am J Infect Control 2015;43:719–723. [DOI] [PubMed] [Google Scholar]
- 29.Rodriguez-Bano J, Picon E, Gijon P, et al. Community-onset bacteremia due to extended-spectrum beta-lactamase-producing Escherichia coli: Risk factors and prognosis. Clin Infect Dis 2010;50:40–48. [DOI] [PubMed] [Google Scholar]
- 30.Marchaim D, Chopra T, Bhargava A, et al. Recent exposure to antimicrobials and carbapenem-resistant Enterobacteriaceae: The role of antimicrobial stewardship. Infect Control Hosp Epidemiol 2012;33:817–830. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Denis B, Lafaurie M, Donay JL, et al. Prevalence, risk factors, and impact on clinical outcome of extended-spectrum beta-lactamase-producing Escherichia coli bacteraemia: A five-year study. Int J Infect Dis 2015. [DOI] [PubMed] [Google Scholar]
- 32.Van Aken S, Lund N, Ahl J, Odenholt I, Tham J. Risk factors, outcome and impact of empirical antimicrobial treatment in extended-spectrum beta-lactamase-producing Escherichia coli bacteraemia. Scand J Infect Dis 2014;46:753–762. [DOI] [PubMed] [Google Scholar]
- 33.Evans CT, Rogers TJ, Chin A, et al. Antibiotic prescribing trends in the emergency department for Veterans with spinal cord injury and disorder 2002–2007. J Spinal Cord Med 2013;36:492–498. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Nicolle LE, Bradley S, Colgan R, et al. Infectious Diseases Society of America guidelines for the diagnosis and treatment of asymptomatic bacteriuria in adults. Clin Infect Dis 2005;40:643–654. [DOI] [PubMed] [Google Scholar]
- 35.Waites KB, Canupp KC, Brookings ES, DeVivo MJ. Effect of oral ciprofloxacin on bacterial flora of perineum, urethra, and lower urinary tract in men with spinal cord injury. J Spinal Cord Med 1999;22:192–198. [DOI] [PubMed] [Google Scholar]
- 36.Barber C, Lacaille D, Fortin PR. Systematic review of validation studies of the use of administrative data to identify serious infections. Arthritis Care Res (Hoboken) 2013;65:1343–1357. [DOI] [PubMed] [Google Scholar]
- 37.Drahos J, Vanwormer JJ, Greenlee RT, Landgren O, Koshiol J. Accuracy of ICD-9-CM codes in identifying infections of pneumonia and herpes simplex virus in administrative data. Ann Epidemiol 2013;23:291–293. [DOI] [PMC free article] [PubMed] [Google Scholar]