Introduction

Antibiotic resistance increases the likelihood of initial inadequate therapy, which corresponds to an increased mortality, length of stay and health-care costs.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 Antibiograms are facility-specific aggregated reports of bacterial susceptibilities and are recommended as a component of antimicrobial stewardship programs.15 Clinically, the antibiogram is used as a tool to aid providers in making evidence-based decisions when choosing the most appropriate agent for initial empiric antibiotic treatment because they are reflective of the susceptibility of organisms isolated at that facility.16 Thus antibiograms specific to certain patient care areas may allow for identification of resistance issues specific to that population and allow for focused antimicrobial stewardship and infection-control efforts.15, 17, 18 However, less than half of academic medical centers have unit-specific antibiograms.19 As inappropriate initial empiric antibiotic treatment is associated with increased morbidity and mortality, the lack of unit-specific antibiograms may negatively affect clinical outcomes because clinicians have less information available to guide treatment selection.15 Further, previous studies comparing unit-specific antibiograms to hospital-wide antibiograms found significant differences in the susceptibility rates across different units,20, 21 suggesting that use of unit-specific or population-stratified antibiograms might be appropriate.22

Individuals with spinal cord injury or disorder (SCI/D) are at high risk for infections and antimicrobial resistance compared to the general patient population secondary to factors, such as frequent hospitalization, previous antibiotic use, frequent and chronic use of invasive devices and development of pressure ulcers.23, 24 However, data on bacterial resistance in the SCI/D population is scare and use of antibiograms in hospital units caring for these patients is limited.25 The lack of SCI-specific antimicrobial susceptibility data poses a significant limitation to the optimal management of infectious diseases in SCI/D patients, because resistance patterns differ based on patient history, care area and past exposure to health-care services and antimicrobials.26, 27 This leads to potential significant bias in using hospital antibiograms to guide treatment decisions, particularly for complicated patients with SCI/D. Thus the purpose of this project was to evaluate the prevalence and resistance patterns of bacteria isolated from patients with SCI/D as compared to a general patient population and determine whether an SCI/D-specific antibiogram would be valuable for facilities treating these patients.

Materials and methods

Study design, setting and population

A retrospective cross-sectional analysis of microbiology laboratory reports from 1 October 2012 to 30 September 2013 for all cultures (inpatient and outpatient) obtained at a 471-bed Midwestern VA facility were evaluated. The SCI/D cohort at the study facility was identified using an ongoing registry of Veterans with SCI/D maintained by the VA Allocation Resource Center (ARC), which is used to allocate resources to VA medical centers.28 The SCI/D group was compared with the remaining non-SCI/D patients with cultures during the study period. Patient location (inpatient vs outpatient) was defined as the patient location when the culture was taken. This study was reviewed and approved by the local institutional review board.

Microbiology data and definitions

Microbiology data were obtained from the VA’s Corporate Data Warehouse, which includes data on medical encounters occurring in the VA health-care system. Microbiology records were queried to extract all cultures for aerobic and anaerobic bacteria, regardless of the source (for example, blood, urine, sputum). Only the first isolate cultured from a single patient was included unless the susceptibility pattern changed. Consistent with current recommendations for comparing susceptibility results, only those bacteria isolated in ⩾30 non-surveillance cultures were included in this evaluation.29 Extended spectrum beta-lactamase (ESBL) producing Enterobacteriaceae and Klebsiella-producing carbapenemases (KPC) were conducted using broth microdilution and the modified Hodge test, respectively, using standardized Clinical and Laboratory Standards Institute methodology.30

Multi-drug resistant Gram-negatives were defined as isolates with intermediate or resistant susceptibilities to ⩾3 of the following classes/agents: aminoglycosides, ampicillin, aztreonam, antipseudomonal penicillins in combination with a Beta-lactamase inhibitor, carbapenems, cephalosporins, cephamycins, chloramphenicol, colistin, fluoroquinolones, fosfomycin, penicillins in combination with a Beta-lactamase inhibitor, sulfamethoxazole/trimethoprim, tetracyclines, and tigecycline.30 Carbapenem resistance was evaluated for all organisms based on testing resistant to any agent in the carbapenem class. Carbapenem-resistant Enterobacteriaceae (CRE) was based on agent minimum inhibitory concentrations and organism isolated according to definitions by the Clinical and Laboratory Standards Institute.31 For Escherichia sp., Klebsiella sp., Enterobacter sp., Salmonella sp., Shigella sp., Serratia sp. or Citrobacter sp., CRE was defined as intermediate or resistant minimum inhibitory concentrations for the carbapenem class (exception ertapenem=resistant only) and resistant to all tested third-generation cephalosporins. Proteus sp., Providencia sp. and Morganella sp. were similarly defined with the exception of displaying nonsusceptible minimum inhibitory concentrations to at least two of the following: imipenem, meropenem, and/or doripenem.

Statistical analysis

Antimicrobial susceptibility results of individual isolates were compiled into a standardized SCI/D-specific antibiogram and compared to a compiled non-SCI/D antibiogram using Chi-square tests and Fisher’s Exact test as appropriate. Comparison of susceptibilities by location were similarly conducted within each group. Demographics of persons were compared using Student’s t-test or Chi-square. SAS 9.3 (SAS, Inc, Cary, NC,USA) was used for data and statistical analyses. A P-value<0.05 was considered significant.

Results

During the study period, there were a total of 5131 unique cultures for 3135 unique patients. Veterans with SCI/D had 23.0% (n=1179) of unique cultures for 249 unique patients. There was a higher mean number of cultures per patient with SCI/D as compared to the overall hospital population (SCI/D=4.7 cultures/patient, overall=1.4 cultures/patient, P-value<0.0001). Overall, those with SCI/D were younger and more likely to be male than those without SCI/D (Table 1). Compared to those without SCI/D, Veterans with SCI/D had a higher Charlson Index (a measure of severity of illness) but were less likely to have cerebrovascular disease, chronic pulmonary disease, heart failure or renal disease (P⩽0.0423 for all). Although those with SCI/D were more likely to be hospitalized within the past 30 and 90 days (P⩽0.0001), the frequency of surgical procedures within the past 30 and 90 days did not differ between groups (P=NS).

Table 1 Demographics of patients with positive cultures
Full size table

Frequencies of pathogens isolated in the two groups differed. The most frequent pathogens identified in Veterans with SCI/D were Enterococcus sp. (15.2%), Escherichia coli (14.5%), Pseudomonas aeruginosa (12.2%), Enterococcus faecalis (11.0%), Klebsiella pneumoniae (9.3%) and Proteus mirabilis (9.0%) (Table 2). All of these organisms, except for E. coli, were significantly more likely to be isolated in those with SCI/D compared to those without SCI/D. Compared to those without SCI/D, Staphylococcus aureus, coagulase-negative Staphylococcus, Streptococcus agalactiae and viridans Streptococcus were less likely to be isolated in those with SCI/D.

Table 2 Distribution and frequency of bacteria identified in SCI vs facility-wide antibiograms
Full size table

Key antimicrobial agents and their susceptibilities are reported for select organisms in Table 3. Susceptibilities were generally poorer in those with SCI/D. S. aureus isolates in patients with SCI/D were less susceptible to oxacillin, clindamycin, erythromycin and tetracycline than in isolates from non-SCI/D patients. E. faecalis was less susceptible to gentamicin (for synergy) in SCI/D patients, but there were no differences when stratified by vancomycin susceptibility. As compared to non-SCI/D, methicillin-susceptible S. aureus (MSSA)was less susceptible to clindamycin and erythromycin in SCI/D (P⩽0.04 for both). Although SCI/D isolates of methicillin-resistant S. aureus (MRSA) were less susceptible to clindamycin, they were more susceptible to tetracyclines (P⩽0.04 for both). For every antibiotic tested, K. pneumoniae was significantly less susceptible in patients with SCI/D (P<0.05; Table 3). Similarly, E. coli had lower susceptibilities for all agents tested except for carbapenems. P. mirabilis also demonstrated lower susceptibilities in SCI/D for Beta-lactams, quinolones, gentamicin and tobramycin (P<0.05). The P. aeruginosa isolates from SCI/D were less susceptible to cefepime, ciprofloxacin, gentamicin, imipenem and tobramycin as compared to the isolates in those without SCI/D (P⩽0.02 for all).

Table 3 Differences in antimicrobial susceptibility for select organisms
Full size table

Over half (57.1%; n=2928) of the unique cultures were associated with the inpatient setting; 27.1% (n=794) were for Veterans with SCI/D. In all, 42.9% (n=2203) of cultures were taken in the outpatient setting; 17.5% (n=385) in Veterans with SCI/D. In both groups, coagulase-negative Staphylococcus, S. aureus, Enterococcus sp., P. aeruginosa, K. pneumoniae, P. mirabilis, Providencia stuartii and Acinetobacter baumannii were more frequently isolated in the inpatient setting (Table 4). As compared to isolates in those with SCI/D in the inpatient setting, non-SCI/D outpatient cultures had higher frequencies of E. coli. Enterobacter cloacae was isolated more frequently in outpatients with SCI/D, whereas this organism was more frequent in the inpatient setting in those without SCI/D.

Table 4 Distribution and frequency of isolates by culture locationa stratified by SCI/D and non-SCI/D
Full size table

Differences were identified in the susceptibility of organisms isolated by location (Table 5). Although there was no difference in the frequency of MRSA or methicillin-resistant coagulase-negative Staphylococcus in inpatient and outpatients, there was a greater frequency of vancomycin resistant Enterococci (VRE) isolates cultured in the inpatient setting, although this difference was not significant in those with SCI/D (SCI/D group: inpatient=11.1%, outpatient=5.0%, P=not significant; non-SCI/D group: inpatient=10.1%, outpatient=2.1%, P<0.01). For both groups, inpatient E. coli isolates were significantly less susceptible for aztreonam, cefepime, ceftriaxone, ciprofloxacin and piperacillin/tazobactam (P⩽0.04). For patients without SCI/D, E. coli inpatient isolates were less susceptible to cefazolin, ampicillin/sulbactam, gentamicin and tobramycin (P⩽0.03). Interestingly, there were no common differences in susceptibilities by culture location for K. pneumoniae. For inpatient SCI/D patients, K. pneumoniae isolates, ciprofloxacin, piperacillin/tazobactam, tobramycin and amikacin were significantly less susceptible (P<0.05). In patients without SCI/D, K. pneumoniae was less susceptible to cefazolin and gentamicin. P. aeruginosa isolates cultured in the inpatient setting were less susceptible to piperacillin/tazobactam for both groups (P<0.04). Additionally, non-SCI/D inpatient isolates were less susceptible to cefepime, imipenem and tobramycin for P. aeruginosa (P<0.03).

Table 5 Differences in select antimicrobial susceptibilities for select organisms by culture locationa stratified by SCI/D and non-SCI/D
Full size table

Significantly more ESBL-producing organisms were isolated in patients with SCI/D (P<0.0001). E. coli produced ESBLs most frequently overall and occurred significantly more frequently in the non-SCI/D group (P<0.001; Table 6). Conversely, P. mirabilis and K. pneumoniae were more frequently ESBL-producers in Veterans with SCI/D. Resistance to at least one agent in the carbapenem class occurred in 3.6% of unique isolates. The SCI/D patients had significantly more isolates with carbapenem resistance (7.6% of all SCI/D isolates) as compared to those without SCI/D (2.4% of all non-SCI/D isolates, P<0.0001). Although there were few isolates meeting the definition of CRE (n=21), CREs were more common in the SCI/D group (2.4% vs 0.5%; P<0.0001). Multi-drug resistance occurred in 36.6% of all Gram-negative isolates in the sample: 60.7% of SCI/D and 28.0% of non-SCI/D cultures (P=0.0001). For multi-drug resistance isolates, susceptibilities exceeded 80% only for amikacin, ertapenem and imipenem. With the exception of ampicillin/sulbactam, susceptibilities for multi-drug resistant isolates were generally poorer in Veterans with SCI/D (Table 6).

Table 6 Gram-negative resistance
Full size table

Discussion

Increasing antimicrobial resistance in complex infections can result in inadequate empiric prescribing. Antibiograms are commonly used tools to guide empiric selection of antimicrobial agents based on bacterial susceptibility patterns at a specific hospital. The lack of rapid and accurate diagnostic tests for pathogen identification and antimicrobial susceptibility is a challenge for all providers, especially for those who care for populations such as SCI/D, with high rates of bacterial resistance. Although the use of an antibiogram to guide empirical antimicrobial treatment has been well documented in the hospital setting,32 the applicability of an antibiogram to SCI/D patients has previously not been explored. As our results demonstrate, not only were the organisms isolated different, but susceptibilities were also generally poorer in Veterans with SCI/D. This difference was especially meaningful in Gram-negative organisms. In addition, inpatient isolates in patients without SCI/D were generally less susceptible where few differences were identified by culture location within the SCI/D group. Patients with SCI/D have multiple risk factors for resistant organisms,23, 24 and our data reflect this increased risk. Carbapenem resistance, MRSA, ESBLs, CREs and multi-drug resistant Gram-negatives were significantly more frequent in Veterans with SCI/D.

Antibiograms are a resource to trend bacterial susceptibility over time, assist with formulary decisions, identify antimicrobial overuse and detect outbreaks and lapse of infection-prevention strategies. Use of local susceptibility data, which is contained in an antibiogram, is also recommended to be used when formulating an antimicrobial stewardship program.15, 32 Although antibiograms are a valuable tool, it has been reported that half of SCI/D providers do not use the facility antibiogram to guide empiric treatment regimens.25 The reasons for this lack of use have not been explored, but it is possible that SCI/D providers are aware that bacterial susceptibility patterns in the overall facility-wide antibiogram do not reflect those observed in patients with SCI/D. Providing a customized antibiogram composed of only SCI/D isolates may improve empiric antibiotic prescribing as the information contained in the antibiogram better reflects those pathogens causing infection.

Comparison of bacterial susceptibilities patterns to develop an SCI-specific antibiogram is not without limitations. The microbiology laboratory reports that were collected and analyzed in this study represent all non-surveillance cultures from all sources obtained from patients regardless of the hospital location. Cultures from different sources may have differing susceptibilities, and consideration of the source of the infection is critical to therapy selection (for example, urine vs blood). Our definition of multi-drug resistance did not account for intrinsic resistance and thus may be an overestimate of high-level resistance for each Gram-negative organism. In addition, the cultures and susceptibilities represented in an antibiogram are highly dependent on the geographic location and the population served. Therefore, our results of bacterial susceptibilities may not apply to the entire SCI/D population. Although the results presented herein are representative of the Veteran population at our facility, they may not be generalizable to other patients seen in other facilities. General limitations of antibiograms include inability to differentiate colonization from infection and lack of patient-specific characteristics and isolate source.

To our knowledge, this is the first report comparing the results of an antibiogram prepared specifically from cultures obtained from patients with SCI/D to cultures obtained from a non-SCI/D population. Although previous studies have described patterns in susceptibility of urinary pathogens in the SCI/D setting,33, 34, 35 the overall prevalence of bacteria responsible for causing disease and the corresponding susceptibilities in this patient population are still largely unknown. In addition, the changing epidemiology of infection and bacterial resistance requires more recent data. In the results presented herein, the frequency of isolation and corresponding susceptibilities of several of the most commonly isolated bacteria from SCI/D patients revealed significant differences compared with the non-SCI/D antibiogram, for which susceptibilities were generally decreased. Thus, reliance on the hospital antibiogram for selecting initial antimicrobial therapy for SCI/D patients could potentially lead to inadequate empiric treatment. It has been found in previous studies that over one-third of patients with SCI/D received inadequate empiric treatment with an antibiotic.36, 37 The availability of more representative data from patients with SCI/D, in the form of an SCI-specific antibiogram, would not only be highly useful in guiding selection of an empiric antimicrobial agent, but may also result in the preservation of antimicrobials through prudent and judicious selection. In addition, the availability of antibiograms specific to certain care areas could offer an avenue for expansion of antimicrobial stewardship efforts customized to the population, including education and improved empirical antimicrobial prescribing.

Different pathogens with poorer susceptibilities are causing infections in patients with SCI/D. However, in a population known to be colonized with multi-drug resistant organisms, future work should be conducted to determine the frequency of colonization vs infection. Thus antimicrobial stewardship is important in this population. Developing stewardship resources, such as unit-specific antibiograms, may be useful in increasing appropriateness of empiric antibiotic treatment.

Data archiving

There were no data to deposit.