|Home | About | Journals | Submit | Contact Us | Français|
Accurate antimicrobial susceptibility testing (AST) and appropriate reporting of AST results for pathogens isolated from blood cultures are critical functions of the microbiology laboratory. We studied AST performance and reporting from positive blood cultures at hospital microbiology laboratories in Iowa. One hundred sixteen episodes of bacteremia from 14 participating hospitals were examined. We detected AST or identification errors for 18 episodes (16%) and judged reporting of AST results to be inappropriate for 38 episodes (33%). Further study is necessary to determine the impact of testing errors and suboptimal reporting of results on the management of bloodstream infection.
Bloodstream infections cause substantial morbidity and mortality, with up to one-quarter of affected patients dying as a result of their infection (14, 15). Early receipt of appropriate antimicrobial therapy has been demonstrated to improve the outcome of bloodstream infection (1, 7). Timely and accurate detection and reporting of bloodstream infection are therefore some of the most important functions of a clinical microbiology laboratory (2, 3, 5, 8, 11). In addition, antimicrobial susceptibility test (AST) results can affect both the clinician's choice of antimicrobial therapy and the patient's outcome (12, 13). If the laboratory reports results that are erroneous, patients with bloodstream infection may not receive optimal therapy. In addition, reporting all AST results for an organism, rather than reporting selectively (e.g., “cascading,” or only reporting the narrowest-spectrum agents to which the organism is susceptible), may encourage inappropriate antibiotic use. Thus, prompt detection of bloodstream infection, accurate microbial identification and susceptibility testing, and appropriate reporting of results are important patient safety issues (6).
We assessed the preanalytical, analytical, and postanalytical errors made by microbiology laboratories related to the detection, identification, AST, and reporting of results from positive blood cultures. The goal of the study was to identify key areas for improvement in order to guide the development and implementation of an educational intervention for laboratory personnel.
The microbiology laboratories of 14 Iowa hospitals participated in this study. The hospitals were all participating in the Emerging Infections and Epidemiology of Iowa Organisms study, a statewide antimicrobial resistance surveillance network (4). These hospitals were chosen based upon geographic location, population distribution, and hospital bed size. Hospital size ranged from 89 to 820 acute care beds (median of 175 beds) and included rural, rural-referral, and urban hospitals.
We studied organisms from five consecutive positive blood cultures from unique patients from whom gram-positive pathogens were isolated and five consecutive positive blood cultures from unique patients from whom gram-negative pathogens were isolated. All isolates were deemed clinically significant. A research assistant, infection control professional, or medical technologist recorded the number of blood cultures collected from each patient during the febrile episode.
Blood cultures were incubated and processed by the usual methods used by each laboratory. When a positive blood culture was identified, a subculture and copy of the final laboratory report were sent to the Centers for Disease Control and Prevention (CDC) laboratory. The CDC reidentified the isolate from the primary subculture and performed AST. A research assistant retrieved the AST results sent to the requesting physician from the hospital computer system or the patient's chart. Bacterial identification and antimicrobial susceptibility results from the laboratory were compared with those in the patient's chart and with results generated at the CDC.
The following parameters were examined: (i) accuracy of identifying bacterial isolates and of doing susceptibility testing; (ii) accuracy of the culture results and susceptibility test results in the hospital information system or on the patients' charts, and (iii) appropriateness of AST reporting according to National Committee for Clinical Laboratory Standards (NCCLS) guidelines (9, 10) and the investigators'expert opinion as the reference.
One hundred sixteen episodes of bacteremia were examined. Fever was present at the time of the episode in 69% of cases. Staphylococcus aureus and Escherichia coli were the most common organisms detected, together accounting for 49% of bloodstream isolates (Table (Table1).1). The number of blood cultures obtained for suspected sepsis was the only preanalytical error we examined, and the participating hospitals performed well on this measure. More than one blood culture was obtained in 104 cases (90%), which was in accordance with the hospitals' policies on blood cultures.
We detected at least one error in AST or instance of suboptimal reporting of AST results for 58 of the 116 episodes (50%). The most common problem identified was reporting results for drugs that should not be reported for a given organism (n = 28 [24% of cases]). In 16 cases (14%), we detected at least one AST error. In four cases (3%), we detected one very major error, including two cases in which an oxacillin (methicillin)-resistant S. aureus isolate was reported as susceptible (Table (Table2).2). In 10 cases (9%), laboratories did not report results for drugs that according to the NCCLS should be routinely provided for the organism.
The most frequent problems in reporting AST results that we identified included reporting carbapenem and other broad-spectrum antibiotic results for E. coli that are susceptible to ampicillin and narrow-spectrum cephalosporins (n = 14) and reporting nitrofurantoin results for blood culture isolates (n = 7). We considered reporting carbapenem and other broad-spectrum antimicrobial results for certain E. coli isolates to be inappropriate because report cascading is recommended to direct physicians to effective narrow-spectrum agents and away from agents that should be reserved for treating the most difficult infections. The laboratory plays a crucial role in helping clinicians choose antimicrobial agents for treating infections (2). Thus, laboratories that report results for every antimicrobial agent tested without censoring inappropriate results may encourage inappropriate antimicrobial use and compromise programs designed to promote judicious antimicrobial use. Reporting nitrofurantion results for blood culture isolates is not appropriate because nitrofurantoin should never be used to treat bacteremia.
It is apparent from these data that laboratories using commercial AST panels and automated instruments to generate their results (which most of these laboratories did) may not sufficiently edit the results to prevent inappropriate drugs from being reported in the patients' charts. For example, imipenem results were reported for ampicillin-susceptible E. coli, amikacin results were reported for gentamicin-susceptible Klebsiella pneumoniae, and nitrofurantoin results were reported for a variety of blood culture isolates.
In addition to the instances of inappropriate reporting, we identified some testing errors, primarily very major errors with staphylococci and oxacillin, and minor errors with gram-negative organisms and cephalosporins. The overall rate of very major errors was low, and most of the participating laboratories performed well. The most important error we identified—a very major error with staphylococci and oxacillin—occurred twice in a single laboratory. We identified only 3 identification errors among the over 50 gram-negative organisms tested, including the misidentification of E. coli as Citrobacter sp. and K. pneumoniae as E. coli. Further study is required to determine the effect of these errors in susceptibility testing and reporting on antimicrobial management and patient outcomes.
We did not observe any conflicts between the AST reports generated in the laboratory and the AST reports in the patients' charts. There are anecdotal reports of such problems, particularly when new laboratory information systems are implemented, but no such problems were identified in this study.
In summary, although most of the laboratories in this study performed well in the processing, testing, and reporting of blood culture results, we did find important areas for improvement. This report describes the initial phase of a multiphase study. We are in the process of developing and implementing a program to address the problems we identified, which will include an educational intervention to improve susceptibility testing and reporting of results. Areas of emphasis will include selection of appropriate antimicrobial agents for inclusion on an AST panel, identification of organism-agent combinations that may produce results in vitro that are not effective in vivo (e.g., nitrofurantoin for blood culture isolates), and cascading of reporting of results to optimize antimicrobial use. Subsequently, we will repeat this study to determine whether the educational program helped improve identification, susceptibility testing, and reporting of results for blood culture isolates at participating hospitals.
This study was supported by the Prevention Epicenters Program of the Centers for Disease Control and Prevention (UR8/CCU715091-03-3).
We thank Caroline Mohr, Patricia Shewmaker, and Sigrid McAllister for assistance with organism identification.