Challenges in Preparation of Cumulative Antibiogram Reports for Community Hospitals

Abstract

Knowledge of local antimicrobial resistance is critical for management of infectious diseases. Community hospitals' compliance with Clinical and Laboratory Standards Institute (CLSI) guidance for creation of cumulative antibiograms is uncertain. This descriptive cohort study of antibiogram reporting practices included community hospitals enrolled in the Duke Infection Control Outreach Network. Cumulative antibiograms from 2012 were reviewed for criteria on reporting practices and compliance with CLSI guidelines. Microbiology personnel were sent a voluntary, electronic survey on antibiogram preparation practices. Data were compiled using descriptive statistics. Thirty-two of 37 (86%) hospitals provided antibiograms; 26 of 37 (70%) also provided survey responses. Twelve (38%) antibiograms specified methods used for compiling data and exclusion of duplicates. Eight (25%) reported only species with >30 isolates. Of the 24 that did not follow the 30-isolate rule, 3 (13%) included footnotes to indicate impaired statistical validity. Twenty (63%) reported at least 1 pathogen-drug combination not recommended for primary or supplemental testing per CLSI. Thirteen (41%) separately reported methicillin-resistant and -susceptible Staphylococcus aureus. Complete compliance with CLSI guidelines was observed in only 3 (9%) antibiograms. Survey respondents' self-assessment of full or partial compliance with CLSI guidelines was 50% and 15%, respectively; 33% reported uncertainty with CLSI guidelines. Full adherence to CLSI guidelines for hospital antibiograms was uncommon. Uncertainty about CLSI guidelines was common. Alternate strategies, such as regional antibiograms using pooled data and educational outreach efforts, are needed to provide reliable and appropriate susceptibility estimates for community hospitals.

INTRODUCTION

Facility-specific cumulative antibiograms serve several important purposes in the care of patients with infectious diseases. For clinicians, knowledge of local drug resistance rates improves the selection of empirical antibiotics prior to return of culture and susceptibility results. In addition, cumulative susceptibility data are used to track changes in resistance over time, perform surveillance for emergence of drug-resistant organisms, and identify areas for intervention by hospital infection prevention and antimicrobial stewardship programs (1). For example, the Centers for Disease Control and Prevention includes two functions of the cumulative antibiogram as “core” elements of hospital antimicrobial stewardship programs: (i) tracking antimicrobial resistance and (ii) regular reporting of information on antibiotic resistance to relevant hospital staff (2). Cumulative antibiogram preparation and distribution are considered an essential function of the clinical microbiology laboratory (3). Antibiogram data can also improve hospital antibiotic formulary decisions and local protocols such as surgical prophylaxis or empirical treatment guidelines.

The Clinical and Laboratory Standards Institute (CLSI) first provided guidelines for preparation of cumulative antibiograms in 2002 and revised them in 2009 and 2014 (3, 4). However, published data on adherence to these guidelines in community hospitals are not available. Further, adherence to guidelines may be particularly important in small, community hospitals where local access to expertise in infectious diseases and microbiology is often limited. The aims of this study were to (i) describe reporting practices of cumulative antimicrobial susceptibility data in a cohort of community hospitals in the southeastern United States, (ii) determine adherence to CLSI guideline recommendations, and (iii) describe perceptions from antibiogram preparers on compliance with guidelines and the impact of cumulative antibiograms on program- or facility-level decision-making.

MATERIALS AND METHODS

We performed a descriptive analysis of antibiogram reporting practices in community hospitals enrolled in the Duke Infection Control Outreach Network (DICON). DICON is a collaborative network of community hospitals in the southeastern United States that share surveillance data on health care-associated infection, educational materials, and consultative services for their infection prevention programs (5). We requested cumulative antibiograms that included data from calendar year 2012 from the 37 acute care hospitals participating in DICON starting in January 2013. Of those facilities that voluntarily provided antibiograms, microbiology laboratory directors were sent a voluntary, electronic survey on antibiogram preparation knowledge and practices. The directors were asked to delegate the survey response to the individual responsible for preparing the facility cumulative antibiogram. Surveys were completed in April-May 2014. Surveys were designed and distributed using Qualtrics (Provo, UT).

Definitions for adherence to CLSI guidelines were determined by six criteria defined in a binary fashion and based on recommendations from CLSI document M39-A3 (2009) (1, 3). Adherence to each criterion was determined upon visual inspection of the antibiogram reports; thus, some recommendations from CLSI could not be objectively assessed (e.g., reporting only final, verified results). We considered antibiograms that included a statement indicating removal of duplicate isolates from the same patient to be adherent to the “first isolate per patient” recommendation. We considered the report to be adherent to the recommendation for “time period” if it included at least 1 year of data. Reports that did not indicate duplicate removal or time period were considered not adherent. Antibiograms were considered adherent to the “30-isolate” recommendation if at least one of two criteria were met: (i) report included only species with at least 30 isolates tested or (ii) report provided a statement to explain impaired interpretability when the number of isolates per species was less than 30. Antibiograms that included only pathogen-drug test combinations recommended for routine or supplemental reporting in Table 1 in CLSI M100-S23 (2013) were considered compliant with the “routine testing” recommendation (6). Reports that separated methicillin-sensitive from methicillin-resistant Staphylococcus aureus were deemed adherent. Antibiograms that reported susceptibilities for Streptococcus pneumoniae were deemed adherent if percent susceptible was reported at both meningitis and nonmeningitis breakpoints. We defined “full” adherence as fulfilling all six recommendations. Antibiograms that did not report S. pneumoniae due to the 30-isolate rule were deemed to be in full adherence if compliant with the other 5 recommendations.

Survey responses and adherence to CLSI guidance were analyzed using descriptive statistics. All analyses were completed using Microsoft Excel (2010). The study was deemed exempt research by the Duke University Health System Institutional Review Board.

RESULTS

Cumulative antibiogram report analysis.

Thirty-two of 37 (86%) acute care hospitals provided antibiograms (Table 1). The formats of the 32 antibiograms varied considerably. The number of species with percent susceptibility reported included a median 11 (interquartile range [IQR], 7.75 to 13) Gram-negative species and 5 (IQR, 5 to 6) Gram-positive species. Only a single facility reported Candida species susceptibilities. The median number of isolates per facility for common species was as follows (IQRs in parentheses): Escherichia coli, 914 (768 to 1,429); Pseudomonas aeruginosa, 129 (88 to 184); S. aureus, 341 (206 to 615). Four (13%) antibiograms provided unit-specific (e.g., intensive care unit) susceptibility estimates. Three (9%) antibiograms separately reported inpatient and outpatient data, 7 (22%) reported combined inpatient and outpatient data, 1 included inpatient data only, and 1 included data from inpatients plus the emergency department. The majority (n = 20, 63%) of antibiograms failed to specify whether the reported susceptibility data included inpatient and/or outpatient specimens. Eight (25%) reports separately presented an antibiogram compiled from urine specimens only. Extended-spectrum beta-lactamase-producing pathogens were separately reported in 15 (47%) antibiograms.

TABLE 1.

Response rates and characteristics of participating hospitals, Duke Infection Control Outreach Network, 2013

Hospital characteristic	n	%
DICON acute care hospitals invited to participate	37
Provided 2012 antibiogram	32	86
Provided survey response	26	70
Respondent hospital characteristicsa
Location
North Carolina	22	69
Virginia	4	13
South Carolina	3	9
Georgia	2	6
Florida	1	3
Ownership
Not for profit	23	72
Government, nonfederal	6	19
For profit	3	9

^aRespondent hospitals contained a median of 211 beds (IQR, 129 to 279).

Adherence to CLSI guidance was low. Only 3 (9%) of 32 antibiograms achieved full adherence to the six CLSI recommendations (Table 2). Twelve (38%) antibiograms specified methods used to compile data and exclude duplicates. Eight (25%) antibiograms reported only species with >30 isolates; 3 (13%) of the 24 that did not follow the 30-isolate rule included a footnote to indicate impaired ability to interpret susceptibility estimates. Twenty (63%) antibiograms reported at least 1 pathogen-drug combination not recommended for primary or supplemental testing by the CLSI (Table 3). The most frequently reported pathogen-drug test combination not recommended for routine reporting was Enterococcus species and tigecycline (10 [31%] antibiograms). Clinically inappropriate pathogen-drug combinations that represent serious errors were reported in 13 (41%) antibiograms (Table 3) (e.g., aztreonam and S. aureus). P. aeruginosa had the most commonly reported pathogen-drug combination serious errors (e.g., ceftriaxone). Thirteen (41%) antibiograms separately reported methicillin-resistant and methicillin-susceptible S. aureus. Approximately half (12 [52%]) of the 23 facilities that reported susceptibilities for S. pneumoniae correctly included both meningitis and nonmeningitis breakpoints.

TABLE 2.

Adherence to CLSI guidelines for creation of facility cumulative antibiogram^c

Criterion	n	Total no.	%
Statement that duplicate isolates were excluded from report	12	32	38
Reported at least 1 yr of data	30	32	94
Reported data only when isolate n was >30 or included footnote to indicate impaired interpretation due to small no.	12	32	52
Separately reported MSSA and MRSAd	13	32	41
Provided meningitis vs. nonmeningitis susceptibilities for S. pneumoniaea	12	23	52
Reported only pathogen-drug test combinations that are recommended for routine or supplemental reportingb	12	32	38
Full compliance with CLSI guidance (all six criteria above were met)	3	32	9

^aReports that did not report any S. pneumoniae susceptibilities due to small isolate numbers were deemed compliant with this element in assessment of full adherence to CLSI guidelines.

^bAntibiograms that included only pathogen-drug test combinations recommended for routine or supplemental reporting in Table 1 in CLSI M100-S23 (2013) were considered compliant (6).

^cSee reference 3.

^dMSSA, methicillin-susceptible Staphylococcus aureus; MRSA, methicillin-resistant Staphylococcus aureus.

TABLE 3.

Number of antibiograms reporting pathogen-drug test combinations that are not recommended by CLSI for routine or supplemental reporting

^aNumbers indicate the number of facility antibiograms (of 32 participating facilities) reporting this pathogen-drug test combination as percent susceptible on their cumulative antibiogram. Overall, 20 (64%) of 32 antibiograms contained at least one pathogen-drug combination not recommended for routine or supplemental reporting. Gray shading indicates combinations considered to be serious errors. Thirteen (41%) of 32 antibiograms contained at least one serious error. Abbreviations: Amp-sulbactam, ampicillin-sulbactam; Pip-Tazo, piperacillin-tazobactam; TMP/SMX, trimethoprim-sulfamethoxazole; Quin-dalfo, quinupristin-dalfopristin.

Survey results.

Twenty-six (70%) of 32 facilities that submitted antibiogram reports provided voluntary survey responses. The majority of survey respondents were microbiology laboratory staff members; approximately half were senior technologists or team leaders (Table 4). Vitek 2 and MicroScan were the most commonly utilized automated antimicrobial susceptibility testing platforms. Individuals most frequently involved in antibiogram preparation included microbiology technologists (19 [90%]) and clinical pharmacists (10 [48%]); infectious disease-trained pharmacists (5 [24%]) and infectious disease physicians (2 [10%]) were less commonly involved. Approximately a third (6 [30%]) of respondents indicated that no committee formally or routinely reviewed the antibiogram report at their facility. The majority (16 [84%]) of respondents indicated that they were unaware or uncertain of any change in facility-level decisions that had occurred as a result of antibiogram results. Survey respondents' self-assessment of full or partial compliance with CLSI guidelines was 50% and 15%, respectively. Over a third (7 [35%]) of respondents reported uncertainty or unfamiliarity with CLSI guidance.

TABLE 4.

Survey responses regarding practices for preparation of cumulative antibiogram

Response	No. of responses (total, 26)	% of respondentsa
Survey respondent role
Senior technologist or team leader	12	48
Medical laboratory or microbiology technologist	4	16
Director or manager of microbiology	4	16
Microbiologist	3	12
Pharmacist	1	4
Infection preventionist	1	4
No response	1
Site of microbiology lab
Onsite	24	96
Referral lab	1	4
No response	1
Automated antimicrobial susceptibility test system
Vitek 2	13	52
MicroScan	10	40
Vitek	2	8
No response	1
Antimicrobial stewardship program
Formal program present	16	76
No formal program	5	24
No response	5
Individual involved in antibiogram prepna
Microbiology technologist	19	90
Clinical pharmacist	10	48
Infectious disease pharmacist	5	24
Pharmacy director	3	14
Infectious disease physician	2	10
No response	5
Hospital group or committee that formally reviews the antibiograma
Pharmacy and therapeutics	8	40
Infection prevention committee	5	25
Antimicrobial stewardship	5	25
Not formally reviewed	6	30
Other group in the pharmacy department	3	15
No response	6
Method of antibiogram disseminationa
Pocket card	10	48
Facility intranet	9	43
Hard copies in patient care areas	4	19
Integrated into facility handbooks	1	5
Uncertain	2	10
No response	5
Which of following changes occurred as result of antibiogram reviewa
Revision of order sets and clinical guidelines for infectious diseases	1	5
Preoperative prophylaxis choices adjusted	0	0
Formulary additions, deletions, or revisions	3	16
Uncertain of any resulting change	16	84
Other	1	5
No response	7
Perceived compliance with CLSI M39 (2009)
Fully compliant	10	50
Partially compliant	3	15
Not compliant	0	0
Uncertain or unfamiliar with guidance	7	35
No response	6

^a“Select all that apply” questions allowed multiple responses possible from the same respondent. Thus, totals and percent may exceed 26 and 100%, respectively.

DISCUSSION

This study highlights limited adherence to CLSI guidelines for cumulative antibiograms, unfamiliarity with these guidelines, and the perceived limited effect that antibiogram reports have on facility-level decision making in our community hospital cohort. The incidence of antibiotic-resistant pathogens varies geographically. Therefore, clinicians must understand rates of resistance in local populations in order to best manage empirical treatment of infectious diseases. The facility antibiogram is the primary tool that provides this important information. However, the facility antibiogram may prove to be an unstandardized, ignored, and at times clinically inappropriate representation of antibiotic resistance.

Our study demonstrated that fewer than 1 in 10 hospitals had full compliance with CLSI guidelines for cumulative antibiograms. Antibiogram reports were heterogeneous in both format and approach. Our study also revealed problems with a lack of documentation of the method of antibiogram preparation (e.g., specifying time period, outpatient versus inpatient, and duplicate removal). Prior published surveys of antibiogram preparation practices have also shown variability in report format and limited uptake of CLSI guidelines (7–9). Several prior investigators have used self-reported, voluntary surveys to assess compliance with specific CLSI recommendations (7, 8, 10). A survey of laboratory directors at 494 U.S. acute care hospitals in 2004 reported 60% of responders in compliance with annually compiling, updating, and distributing a facility antibiogram (8). The self-reported compliance rates in a 2009 survey of pharmacy directors in the University Health Consortium were favorable (10). Respondents reported publishing at least annually (98%), eliminating duplicates (89%), not including surveillance cultures (83%), and including at least 30 isolates for each organism (64%) (10). Studies that assessed compliance by direct inspection of antibiograms showed lower adherence than those that used self-reported compliance (9, 11).

Direct inspection of antibiograms in our study revealed not only low guideline compliance but also that some facility antibiograms contained serious errors of clinically inappropriate pathogen-drug combinations. Similarly, Zapantis et al. examined 209 antibiograms and found that a number of reports included inappropriate pathogen-drug combinations (e.g., Klebsiella pneumoniae and ampicillin) or unlikely percent susceptibility results (11). A longitudinal study evaluated the effect of statewide educational outreach efforts by reviewing cumulative antibiogram reports from 86 hospitals in Michigan (9). Serious errors in antibiograms were defined as improbable or impossible percent susceptibility results or the reporting of misleading or inappropriate pathogen-drug combinations. The percentage of antibiograms with serious errors decreased over time (59% to 19%). Serious errors in antibiograms could result in medical errors if antibiograms are used to make clinical decisions. These serious errors suggest that antibiograms were not thoroughly reviewed for accuracy and clinical relevance. Further, these errors may potentially represent a limited knowledge of clinical microbiology among antibiogram preparers and reviewers.

The current study is unique in that it combines direct inspection of antibiogram reports with survey responses, which provides insight into perceptions from antibiogram preparers in addition to evaluations of guideline compliance. A third of survey respondents reported uncertainty or unfamiliarity with CLSI guidance. The majority of respondents indicated that they were unaware of any change in facility policy or decision making that had occurred as a result of antibiogram data. A large proportion reported that no formal review of antibiogram data occurred. These responses plus the presence of serious clinical errors in study antibiograms are further evidence of the absence of formal, multidisciplinary review.

This study is focused on antibiogram preparation practices in smaller, community hospitals in the southeastern United States. We have observed that such hospitals face specific barriers to effective antibiogram preparation, such as limited microbiology or informatics personnel dedicated to the task, lack of resources to purchase proprietary CLSI documents, lack of support from clinicians and/or multidisciplinary teams, and small numbers of isolates that impair the interpretability of cumulative data. Several hospitals in our cohort do not have infectious disease specialists on staff. Specific efforts to address problems with production of high-quality antibiograms and to promote antibiogram use for facility decision making are needed in order to provide reliable, local susceptibility estimates and to track the emergence of drug resistance in community hospitals. Educational outreach on the elements of CLSI guidance and provision of CLSI documents by state health departments may have some benefit, as demonstrated by investigators in Michigan (9). Also, regular review and feedback from clinical providers are essential for both the accuracy and the usability of the antibiogram document, as well as for distribution of the information for clinical use. To specifically address the problems of small isolate numbers, pooled data from multiple facilities or longer cumulative data collection periods (e.g., including more than 1 year of data) may provide more reliable estimates of drug resistance for small facilities. In our study, there was variation in what patient populations were included in different institutions' antibiograms (e.g., inpatient versus outpatient), which made pooling of antibiogram data in order to compare regional trends impossible to accomplish.

We believe that CLSI should expand and highlight specific guidance for small facilities in their recommendations for cumulative antibiogram preparation. Specifically, guidelines should address the commonly encountered problem of small isolate numbers and encourage standardization in order to assist public health authorities in obtaining comparable data between institutions. In addition, there is substantial opportunity to improve reporting through electronic medical record vendors, which could streamline the process of data collection and analysis, use methods that are both CLSI compliant and reliable, and produce standardized data that can be pooled between facilities. We recently established the Duke Antimicrobial Stewardship Outreach Network to support the development of antimicrobial stewardship programs in community hospitals in our region (http://dason.medicine.duke.edu/home). We have initiated review and feedback of annual antibiograms as part of our consultation services as a result of this study.

This study has limitations. While it is limited to community hospitals within DICON, review of the literature suggests that problems identified in our study are also present in other geographic areas. The survey portion of our study is subject to recall and response bias due to its voluntary design and lack of independent verification of responses. Preparation of the antibiogram report and the survey response date were separated by approximately 1 year, which could have further contributed to recall bias. Data for cumulative antibiograms may have been directly obtained from automated testing instruments instead of laboratory information systems. Therefore, some of the data used in antibiograms may have included unverified results. We could not investigate this specifically, but it may contribute significantly to some of the errors in reporting of pathogen-drug combinations and problems with compliance with CLSI guidance. We also did not determine the brand of electronic laboratory information system used at each hospital, which may impact the ability to adhere to CLSI guidance. We considered lack of documentation regarding removal of duplicates to indicate nonadherence. If this recommendation was followed without documentation on the report, adherence to this criterion may have been misclassified. Despite these limitations, we believe that this study highlights problems that need attention from both clinicians and public health advocates, especially as bacterial resistance to existing antibiotics continues to worsen (12, 13).

In conclusion, full compliance with CLSI guidance for hospital antibiograms was uncommon in our cohort of community hospitals. A survey of antibiogram preparers revealed uncertainty about CLSI guidance and little perceived effect from use of facility antibiogram reports. Alternate strategies, such as using pooled regional data and educational outreach, are greatly needed in order to improve the quality of antibiograms, emphasize the urgency of emerging antibiotic resistance in local settings, and provide reliable and appropriate susceptibility data for clinicians practicing in community hospitals.