We assessed the usefulness of reporting direct blood Gram stain results compared with the results of positive blood cultures in 482 episodes and monitored impact on selection of antimicrobial treatment. We found that the reporting groups “Staphylococcus spp,” “Pseudomonas spp and related organisms,” and “yeasts” identified in this way matched perfectly with later culture identification. When the report indicated Staphylococcus spp or Pseudomonas spp and related organisms, physicians started or changed antimicrobials suitable for these bacteria more frequently than when “other streptococci” and “family Enterobacteriaceae” were reported (P < .05). Incorrect recognition of Acinetobacter spp as Enterobacteriaceae family is still the most challenging problem in this context. Gram stain results that definitively identify Staphylococcus spp, Pseudomonas spp and related organisms, and yeasts reliably can be rapidly provided by clinical laboratories; this information has a significant impact on early selection of effective antimicrobials. Further investigation is needed to assess the clinical impact of reporting Gram stain results in bacteremia.
Clinically significant bacteremia is an important cause of serious morbidity and mortality. In a previous report of 843 episodes of positive blood cultures from 707 patients, bacteremia-associated mortality was 17.5%.1 In another report of 955 bacteremic episodes, mortality due to unidentified organisms was 44%, and even when the organisms were known, mortality was still as high as 25%.2
To reduce mortality and morbidity in bacteremic patients, it is necessary to initiate effective antimicrobial therapy as soon as possible. Many reports indicate a relationship between inadequate antimicrobial therapy and unsatisfactory outcome.3–8 Bacteremia frequently causes sepsis and septic shock. Because prompt institution of therapy active against the causative pathogen is one of the most important predictors of outcome, physicians must establish a system for rapid administration of a rationally chosen drug or combination of drugs.9
The clinical microbiology laboratory has an important role in the management of bacteremia. Detecting pathogenic microorganisms in blood cultures and testing antimicrobial susceptibility always assists in selecting the appropriate antimicrobial agent. The most important and primary test to perform on any positive blood culture is Gram stain,10 which is the most rapid and simplest test to characterize microorganisms. It is therefore highly likely that the information provided by the Gram stain will help to assess the adequacy of antimicrobial therapy selected after collecting blood culture specimens and before final identification of the microorganism.11 In recent reports, the impact of Gram stain results on patient mortality has been documented.12–15 On the other hand, there remains the possibility that Gram stain results do not match with the final identification of microorganisms. This would carry a risk leading to inadequate antimicrobial therapy and potentially affecting patients’ clinical course and mortality.
The aim of our study was to assess the advantages and disadvantages of reporting Gram stain results for selecting antimicrobial agents and correlating these results with the patients’ hospital courses.
Materials and Methods
Study Period and Patient Selection
For the 12-month period, April 1, 2005, through March 31, 2006, all inpatients with positive blood cultures clinically recognized as reflecting true pathogen presence were evaluated at Nihon University Itabashi Hospital, Tokyo, Japan. Microorganisms appearing in blood cultures were defined as true pathogens when it was necessary to start treating the patient with antimicrobials. If several blood culture sets were undertaken simultaneously, they were dealt with as 1 episode. Following the literature,1 if patients had positive blood culture results repeatedly over 48 hours following the previous positive results, they were included as new positive episodes.
Processing of Samples in the Microbiology Laboratory
For blood culture, we used culture bottles (BACTEC bottles 92F and 93F as 1 set or 94F for small children, Becton Dickinson Diagnostic Systems, Franklin Lakes, NJ) and automated systems (BACTEC 9240 and 9120, Becton Dickinson Diagnostic Systems), with continuous agitation. Microorganisms were examined by Gram stain (neo-B and M Wako, Wako Pure Chemical Industries, Osaka, Japan) using an uncentrifuged drop of blood taken from the culture bottle. This result was compared with results after subculture on standard culture media, with identification and antimicrobial susceptibility tests performed as appropriate for suspected microorganisms using standard procedures recommended by the Clinical and Laboratory Standards Institute. To identify bacterial species, a number of methods were used, including assessment of morphologic criteria and the appearance of Gram-stained colonies, biochemical characteristics with manual procedures or commercial kits (API series, bioMérieux, l’Etoile, France; and N-ID test SP-18, Nissui Pharmaceutical, Tokyo, Japan), or an automated identification system (VITEK 2 system, bioMérieux). For susceptibility tests, disk-diffusion methods, microdilution methods, and the VITEK 2 system were used.
Reporting the Results of Microbiologic Examinations
For positive blood culture bottles, results of the Gram stain were reported immediately to the physicians by telephone using categories such as gram-positive cocci, gram-positive rods, gram-negative cocci, gram-negative rods, fungi, and multiple microorganisms. Written reports were sent to an online reference system after definitive identification had been made and susceptibilities were obtained.
After positive episodes were recognized as reflecting true pathogen infection by physicians, patients’ medical charts were reviewed and clinical data for each episode extracted, including when the blood culture specimen was collected, age and sex of the patient, hospital department, infected site causing bacteremia, WBC count, and predisposing conditions.
Impact of Reporting the Results of Gram Stains on Antimicrobial Therapy
Antimicrobial therapy for each episode was considered by physicians when the blood culture became positive and Gram stain results had been reported and when the results of identification and susceptibility had been reported. Potential reasons for changing antimicrobial therapy were assessed in each reporting category and compared with the actual rate of changing antimicrobial agents.
Accuracy of Gram Stain Results and Incorrectly Identified Episodes
After the results of culture identification and susceptibility testing had become available, the frequency of isolated microorganisms and accuracy of Gram stain were assessed. Correct identification was defined as when the results of the latter perfectly matched culture identification, whereas incorrect results were defined as when microorganisms expected to be present according to the Gram stain were not cultured or when unexpected microorganisms were cultured. For the incorrectly identified episodes, details of microbiologic features, change of antimicrobial therapy, and clinical data were evaluated.
For each microorganism, mortality and duration of hospital stay of episodes with incorrect Gram stain results were compared with the data for other episodes with correct results.
For univariate analyses, statistical significance was calculated by using the χ2 test or Fisher exact probability test. For comparisons of distributions, the Mann-Whitney U test was used. All statistical analyses were performed with Statmate II (ATMS, Tokyo, Japan).
During the study period, 482 blood culture–positive episodes in 374 patients were clinically recognized as true bacteremia. The age range of patients was 0 to 96 years (median, 67 years). Of the patients, 291 were male and 191 female Table 1.
Clinical data pertaining to infectious episodes are also shown in Table 1. The Department of Gastroenterology and Hepatology, the Department of Hematology and Rheumatology, and the Department of Emergency and Critical Care Medicine were the 3 departments with the highest frequency of true bacteremia. The most common sites of infection were the bloodstream infection (catheter-related) followed by undetermined and gastrointestinal/biliary tract. Malignant diseases, ongoing chemotherapy, and diabetes mellitus were frequent predisposing conditions.
Impact of Reporting the Results of Gram Stain on Antimicrobial Therapy
Suspected microorganisms as assessed by Gram stain of blood cultures are shown in Table 2. The most common was Staphylococcus spp, followed by the Enterobacteriaceae family and other streptococci. Multiple microorganisms were also frequent. Table 2 also shows changes in antimicrobial therapy after the results of the Gram stain were reported. “Adjust spectrum” means that antimicrobials selected for empiric therapy did not cover microorganisms detected by Gram stain; more effective antimicrobials were then substituted (eg, ceftazidime changed to vancomycin to cover Staphylococcus spp).
For Gram stains indicating Staphylococcus spp, some changes in antimicrobial therapy in 107 (60.1%) of 178 episodes were instituted. In particular, in 20 (11.2%) of these episodes, no antimicrobial therapy had been given before Gram stain results were reported, but antimicrobial therapy was started thereafter. In contrast, when the results of the Gram stain were Streptococcus pneumoniae or other streptococci, antimicrobial therapy was changed in 4 (33%) of 12 and 15 (43%) of 35 episodes, respectively. The frequency of changing antimicrobials in episodes of Staphylococcus spp was significantly greater than in episodes of other streptococci (P < .05), but there was no difference between Staphylococcus spp and S pneumoniae.
When the results of Gram stain were Enterobacteriaceae family, antimicrobial therapy was changed in 45 (30.6%) of 147 episodes. In particular, patients with 10 (6.8%) of these episodes had not received antimicrobial therapy before the staining results were reported. In contrast, when the results were Pseudomonas spp and related organisms, the episodes for which antimicrobial therapy was changed numbered 16 (59%) of 27. Changing antimicrobials in episodes of Pseudomonas spp and related organisms was thus significantly more frequent than in episodes of Enterobacteriaceae family (P < .05).
Patients with all 21 episodes reported as yeasts had already received some antimicrobial therapy, but after Gram stain results were reported, additional or changed antifungal agents were instituted in 12 episodes. This frequency would probably have been even higher had 2 patients not already died by the time the Gram stain result became available.
In 38 episodes, multiple microorganisms were recognized by Gram stain, and for 18 (47%), some changes of antimicrobial therapy were required. Here, choice of antimicrobials depended on the site of infection and other clinical considerations.
Accuracy of Gram Stain Results and Incorrect Episodes
Table 3 shows the accuracy of Gram stain results. The results for the Staphylococcus spp, S pneumoniae, other streptococci, Corynebacterium spp, Neisseria/Moraxella spp, Pseudomonas spp and related organisms, and yeasts groups were in perfect agreement with culture identification results.
However, in 12 episodes, the results of Gram stain were not consistent with the results of culture identification. One episode in the undetermined gram-positive cocci group was finally identified as Acinetobacter baumannii. Of 9 episodes in the Enterobacteriaceae family group, 6 were finally identified as Acinetobacter spp, and 3 contained Pseudomonas aeruginosa. There were 2 episodes in the multiple microorganisms group. One episode recognized as Staphylococcus spp and Pseudomonas spp and related organisms was identified as only P aeruginosa, and another recognized as Bacillus spp and Pseudomonas spp and related organisms was identified as Clostridium perfringens and P aeruginosa. In 9 episodes (75%), antimicrobials were changed after culture identification and antimicrobial susceptibility data became available. Especially in 7 episodes, the choice of antimicrobials was inappropriate because of incorrect Gram stain results. Details of the microbiologic features and change of antimicrobials are shown in Table 4.
Clinical characteristics of the patients affected in the incorrectly identified episodes are shown in Table 5. In 7 of 12 episodes, there were difficulties in identifying the site of infection. The pharynx was secondarily a frequent site. Patients frequently were from the Department of Gastroenterology and Hepatology, the Department of Otolaryngology, and the Department of Hematology and Rheumatology. Nine episodes were in patients who had malignant diseases, including 6 receiving chemotherapy. Liver cirrhosis was recognized in 3 patients. Leukocytopenia was only seen in 1 episode.
Of 12 episodes with incorrectly identified species, 2 yielded 2 bacterial genera at the same time. Patients with 6 (50%) of these 12 episodes died despite antimicrobial therapy; duration of hospital stay was 23 to 418 days (median, 77 days). Table 6 also shows the outcome and duration of hospital stay for episodes with incorrect Gram stain results, summarized and compared with episodes with correct results for each microorganism. Even though no statistically significant differences in mortality and hospital days were observed between incorrectly and correctly identified groups, the comparison of each microorganism is described here in detail.
In 7 episodes with incorrect identification by Gram stain, 2 patients died (29%). However, only 1 patient with a correctly identified episode also died (100%).
In 3 episodes with incorrect Gram stain results, 2 patients died (67%), whereas among 24 correctly identified episodes, 10 died (42%).
Infection With Multiple Microorganisms
Forty patients had infections with multiple microorganisms according to culture identification. Of 4 patients with incorrectly identified episodes (cases 9–12 in Tables 4 and 5), 2 died (50%). Among 36 correctly identified episodes, 11 patients died (31%).
Clinical microbiology laboratories have a key role in the management of bacteremia because they are the first to be in a position to register infection from the positive blood culture results. Appropriate early antimicrobial therapy improves the clinical course and reduces mortality in patients with bacteremia.3–8
Gram stain is the simplest and most rapid test to assess microorganisms directly from positive blood culture bottles and is important for starting adequate antimicrobial agents as soon as possible.11 Recent studies have examined the impact of reporting Gram stain results on patient mortality.2,12–15 The American Society for Microbiology (ASM) indicates that the single most important first test to be performed on any positive blood culture is Gram stain and states that the terminology used to report such results should be as descriptive as possible.10 Thus, we reported the Gram stain results as shown in Table 2 in a manner intended to be as useful and easily understandable as possible for clinicians.
However, the assessment of Gram stain results and microscopic examination is quite subjective, being influenced by medical technologists’ skill and experience. Thus, Gram stain results are occasionally inconsistent with the final identification of microorganisms, which represents a risk leading to inadequate antimicrobial therapy and can affect the clinical course and mortality. In our clinical laboratory, Gram stain morphologic examination of direct smears follows the recommendations of the ASM,16 and results are reviewed by supervisory personnel for quality control. So in our study, as shown in Table 3, except for 12 mismatched episodes, Gram stain results completely matched the corresponding culture identification.
We found that when the Gram stain result was Staphylococcus spp, physicians changed or started antimicrobial therapy suitable for Staphylococcus spp significantly more frequently than for the other streptococci group. Staphylococcus spp were most frequently isolated from blood cultures, and all episodes were identified as Staphylococcus spp correctly. In the present study, Staphylococcus spp were the main causative microorganisms of catheter-related systemic infections (137 episodes [79.2%]). Among 107 episodes with a change of antimicrobials after Gram stain results became available, antimicrobials active against methicillin-resistant Staphylococcus aureus had been started in 100 (93.5%). Furthermore, in 89 (89.0%) of 100 episodes, the same antimicrobials were continued after culture identification and susceptibility tests. In only 11 cases could the use of anti–methicillin-resistant S aureus agents be discontinued because the causative microorganisms were not resistant to methicillin. Thus, it is clearly safe and advantageous to report definitive Staphylococcus spp to initiate treatment effective for one of the most important health care–associated infections.
In the gram-negative rods series, when Pseudomonas spp and related organisms were identified, physicians changed or started antimicrobial therapy significantly more frequently than when the Enterobacteriaceae family was reported. This is important for selecting antipseudomonal agents because the administration of appropriate antimicrobial therapy is essential for a good outcome in P aeruginosa bacteremia.17 In our study, in 16 episodes (59%) of Pseudomonas spp and related organisms, it was necessary to change antimicrobials started for empiric therapy, although the infected sites were not identified in 6 episodes. Only 2 episodes were in patients from the Department of Hematology and Rheumatology, and none showed leukocytopenia. These results might cause difficulties in anticipating infection with Pseudomonas spp and related organisms at the time of selecting empiric therapy. All episodes in the Pseudomonas spp and related organisms group were correctly identified as nonfermentative gram-negative rods, and continued administration of antipseudomonal agents after culture identification and susceptibility tests was required. However, bacteria in 4 episodes were resistant to antimicrobials selected after the Gram stain result was reported. Therefore, one solution for appropriate therapy would be to use multiple antipseudomonal agents after Gram stain by referring to the antibiogram of each hospital. In our clinical laboratory, Pseudomonas spp and related organisms are recognized as gram-negative thin bacilli, medium length to long with rounded to pointy ends, following the criteria in the ASM manual.16 As a result, the sensitivity and specificity values for the Gram stain for distinguishing nonfermentative gram-negative rods from all gram-negative rods defined by culture were 71.1% and 98.9%, respectively, and the positive and negative predictive values were 94.1% and 93.0%, respectively (unpublished data). Even though the recognition of Pseudomonas spp and related organisms by Gram stain has not been standardized, we conclude that it is helpful and safe to report results definitively to treat patients with bacteremia caused by nonfermentative gram-negative rods appropriately and rapidly. The problem of prediction of Acinetobacter spp will be discussed later.
The Gram stain results for the yeasts group also matched with culture identification perfectly. Although not statistically significant, mortality from candidemia was reported as 19% to 24% despite many kinds of new antifungal agents18; thus, reporting yeasts as the result of Gram stain should be helpful for selecting suitable antifungal agents immediately. Therefore, it is recommended that clinical laboratories report Gram stain results clearly in these categories.
Regarding the episodes with incorrect Gram stain results shown in Table 5, the most common misrecognition was of nonfermentative gram-negative rods as Enterobacteriaceae family. In particular, incorrect recognition of Acinetobacter spp is a major problem. Acinetobacter spp were strictly aerobic, nonfermentative, gram-negative coccobacillary rods, which sometimes stain like gram-positive cocci from positive blood culture bottles.19 In a previous study, 5 of 13 cases of bacteremia with Acinetobacter spp were recognized as gram-positive bacteria.20 In contrast, another study reported that it was easy to recognize Acinetobacter spp as gram-negative rods.21 In our study, there were 8 episodes of bacteremia caused by Acinetobacter spp. Of the 8 episodes, 7 were recognized as gram-negative bacteria, but only 2 were recognized as Pseudomonas spp and related organisms. Thus, each clinical laboratory needs to know the appearance of Acinetobacter spp in its own Gram stain procedure. Other episodes yielding multiple microorganisms also had these nonfermentative gram-negative rods. To recognize these properly when performing Gram stain, it may be helpful to know that they can grow only in aerobic culture bottles or to obtain information on the antimicrobial agents previously used.
In addition, considering the clinical characteristics of episodes with incorrect Gram stain results, samples from patients in particular departments (ie, hematology and rheumatology, gastroenterology and hepatology, and otolaryngology) and with cirrhosis, malignant diseases, and chemotherapy should be assessed with special caution. For example, we found that 3 of 18 patients from the Department of Otolaryngology had bacteremia of Acinetobacter spp (17%). This was not a statistically significantly higher prevalence than for patients in the Department of Hematology and Rheumatology (3/70 [4%]) or the Department of Gastroenterology and Hepatology (3/78 [4%]), but this information would be helpful in determining the reporting category for patients from the Department of Otolaryngology. Nevertheless, reporting Gram stain results equivocally as “family Enterobacteriaceae” should be avoided in these clinical conditions.
It has been reported that rapid detection of systemic infections shortens the length of hospital stay.22 The limitation of this study is that the number of these episodes was so small that the influence of incorrect Gram stain results or bacterial resistance to antimicrobials on mortality and hospital days was difficult to assess. Further investigation of larger populations for longer periods is needed to fully assess the impact of Gram stain results on mortality and hospital days.
Reporting Gram stain results of positive blood cultures facilitates the correction of empiric antimicrobial therapy before the results of culture identification and susceptibility testing become available. Episodes in the categories Staphylococcus spp, Pseudomonas spp and related organisms, and yeasts showed perfect agreement with later culture identification. Selected antimicrobials were frequently changed owing to Gram stain results in the former 3 categories. Therefore, it is recommended that clinical laboratories definitively report Gram stain results for these microorganisms, which can have a significant impact on early selection of optimal antimicrobial agents. Recognition of Acinetobacter spp by Gram stain remains a challenging problem, however. Further investigations are required to assess the clinical impact of reporting Gram stain results in bacteremia.
We acknowledge the assistance provided by Aleemuzzaman Sheikh, MD, for critical reading of the manuscript as a native English speaker.