With programs of quality control, quality assurance, and quality improvement in place, analytical errors in the clinical microbiology have been greatly reduced during the past 20 years. Clinical microbiology testing has also improved because of significant progress in testing methodologies, technical improvements, and availability of tests with better sensitivity and specific- ity. The frequency of analytical errors in the clinical laboratory is now quite low. Total quality management programs now in use in most clinical laboratories are designed to prevent and minimize analytical errors. By implementing and integrating training, competency assessment, proficiency testing, quality control, and quality assurance in the clinical microbiology labora- tory, many potential analytical errors are prevented or detected and corrected before reporting, reducing the number of errors that impact patient results. With the availability of standards and guidelines for most of the testing performed by clinical microbiology laborato- ries, analytic errors have been reduced to a very low level. When errors do occur during the analytic phase of microbiology testing, they tend to occur as a result of biological variability exhibited by microorganisms.
Automated systems for identification and susceptibil- ity testing are also subject to occasional errors, espe- cially with testing of unusual organisms, fastidious organisms, or organisms with unusual resistance to
antimicrobial agents. Some of the more commonly seen analytical errors that can occur in the clinical microbiology laboratory are discussed next.
Gram Stains Errors
The Gram stain is one of the most important meth- ods used in the clinical microbiology laboratory. It is used for initial assessment of specimens prior to cul- ture and is an important step in the workup and identification of pathogens isolated from clinical speci- mens. Gram stains guide workup pathways and iden- tification of organisms; thus, they are extremely important to the accurate processing and workup of specimens. Gram stain errors can occur for several rea- sons: Organisms may not exhibit an expected Gram stain morphology, or technical errors performing the Gram stain may occur:
Technical errors in preparation of the smear and performing the Gram stain: Proper preparation of the Gram smear must be performed to ensure that the smear is not too thick or too thin. Thin smears can result in reduction of sensitivity, especially when organisms are rare in the specimen, and are prone to over-decolorization, causing Gram-positive organisms to appear Gram-negative. Thick smears are prone to under-decolorization, can reduce specificity, cause Gram-negative organisms to appear Gram-positive, and may retain artifacts during the staining, making reading difficult.
Inherent variability in organism Gram characteristics:
Certain organisms do not always exhibit expected Gram stain characteristics; occasionally, Gram- positive organisms will appear Gram-negative, and Gram-negative organisms retain crystal violet during the destaining step and appear Gram- positive (Table 20.1). Two genera of Gram-positive organisms that are often isolated in blood cultures, Clostridiumspp. andBacillusspp., may appear Gram-negative in smears prepared from culture- positive bottles.Acinetobacter, a Gram-negative coccobacillus, may resist destaining and appear Gram-positive in a Gram smear. Medical
technologists should be aware of this possibility and use other features of the organism (size,
morphology, and other biochemical characteristics) to assist in reporting. Alternative methods of determining the Gram characteristic (e.g., Gram- Sure reagent) may be used to test organisms isolated in pure culture.
Analytical Errors in Interpretation of Gram Stains Performed on Positive Blood Cultures
There are several types of errors that can potentially occur when Gram stains are performed on bottles that signal positive in automated systems. The two most common sources of analytical errors are incorrect reporting of the Gram characteristic (Gram positives reported as Gram negatives and Gram negatives reported as Gram positives). These were primarily due
TABLE 20.1 Organisms That May Exhibit Unusual Gram Stain Characteristics and Suggestions for Identifying Aberrant Results Organism Usual Gram Morphology Possible Aberrant Result Clues to Possible Aberrant Result
Bacillusspp. Gram-positive bacilli Gram negative Use colony morphology with Gram stain
result
Bacillusspp. are often larger than most gram-negative bacilli
Clostridiumspp. Gram-positive bacilli Gram negative Clostridia are anaerobes
Organism size is very large Often in short chains Acinetobacterspp. Gram-negative coccobacilli or
bacilli
Gram positive Coccobacillary shape is unusual for gram- positive organisms such as streptococci or staphylococci
Campylobacter Faintly staining Gram negative; small coccobacilli with characteristic shape
No organisms seen Use fuchsin counterstain Perform an acridine orange stain
Legionella,Bordetella Faintly staining Gram negative
No organisms seen Use fuchsin counterstain
Brucellaspp. Small Gram-negative cocci or coccobacilli
Gram positive Use colony characteristics and growth characteristics to verify
319
ANALYTICAL ERRORS IN THE MICROBIOLOGY LABORATORY
to the staining characteristics of the organisms as described previously[13]:
Error in not recognizing more than one organism present in a positive smear: The incidence of bacteremia with more than one organism is fairly low. In a study published in 2003, the rate of polymicrobial sepsis was 4.7%[14]. There is potential for failure to recognize the presence of a second morphotype when Gram stains of positive blood cultures with multiple morphotypes present are read. Positive blood cultures with mixed morphologies (e.g., Gram-positive cocci and Gram-negative bacilli) are occasionally misread as single morphologies present. Occasionally, errors can occur when two morphologies occur and one organism is
predominant. It may be difficult to distinguish Gram-positive cocci in clusters, pairs, or short chains when these organisms are mixed.
Occasionally, errors occur when Gram-negative bacilli or coccobacilli are overlooked in smears that also have Gram-positive organisms.
Error in not recognizing the presence of organisms in blood cultures that signal positive but no organisms are seen on the Gram stain: Blood cultures that are incubated in continuously monitored instruments may signal positive but appear negative when the culture medium is stained. There are two
scenarios in which this may occur: false-positive signaling by the instrument (no organisms are present but the instrument sensor detects CO2, giving a positive signal) and the presence of organisms that are difficult to visualize in a Gram stain. The laboratory should have methods in place to determine which of these is occurring when it has signal positive bottles that are Gram stain negative.
Certain organisms do not Gram stain well, includ- ing Legionella, Campylobacter, Bartonella, and Brucella.
Alternate staining procedures should be used, depend- ing on whether the Gram stain is being performed on direct clinical specimens or on isolated colonies grow- ing on solid media. Carbol fuchsin stains can be used when organisms are not staining well with the Gram stain. Gram stain errors can occur with fastidious or difficult-to-stain organisms growing in blood culture bottles. These bottles may signal positive, but when the smear is examined by a Gram stain, they are nega- tive. Occasionally, blood cultures that are incubated in a continuously monitoring system will signal positive and the Gram stain shows no visible organisms. In this case, an alternate staining method such as acridine orange should be used to determine if organisms are present. Anaerobes may not stain properly when exposed to oxygen[15].
CASE REPORT: GRAM STAINS NEGATIVE, CULTURE POSITIVE BLOOD BOTTLES: Four blood culture bot- tles signaled positive in the automated blood culture instrument. Gram staining of the culture medium was performed; no organisms were noted on the Gram smear. The blood culture medium was subcultured to solid media, and the bottles were returned to the instrument. The following day, bacterial colonies were evident on the subcultured media. The organism was identified asCampylobacter jejuniby sequencing.
Negative Gram stains of blood culture media that signal positive in automated instruments occasionally occur. In this case, C. jejuni was growing in the blood cultures but was not recognized in the Gram stain when the bottles signaled positive; recognition that an organism was present was not achieved until the sub- cultures were growing. Staining of the blood culture medium using an alternative method such as acridine orange can reveal the presence of organisms, allowing for a report to be generated.
Analytical Error: False-Positive Reports of Growth in Liquid Broth Cultures—Reporting the Presence of Organisms in Culture-Negative Broth Cultures
Liquid culture media such as thioglycollate broth is commonly used in addition to solid media as a pri- mary isolation media to allow for recovery of anae- robes and organisms that may be present in low numbers in a specimen. The liquid cultures are usually examined visually, and if turbidity is present, a smear is made for Gram staining. A common occurrence is to observe organisms present in the Gram stain (e.g., Gram-positive bacilli, Gram-positive cocci, or Gram- negative bacilli) and to report these. On subculture, the organisms do not grow. These false-positive reports of organisms present a challenge to the laboratory.
Liquid nutrient media such as thioglycollate broth may be sterilized by methods other than filtration;
organisms present in the original material are nonvia- ble, but they may still be present in the liquid media and may appear in the Gram stain. Liquid nutrient media should not be used for initial processing of spe- cimens (e.g., tissues) because the primary Gram stain report may be a false positive. Tissues should be ground in sterile saline, and then a portion of the ground specimen should be transferred to the liquid thioglycollate medium. Also, a Gram stain report from the broth media culture should not be made; subcul- ture should be performed, and if organisms are recov- ered, they can be reported.
Liquid blood culture media can also have nonviable organisms present; if the bottles signal in an auto- mated instrument, organisms may be present on the Gram stain, but they are nonviable. These organisms
represent artifacts, but they are very problematic and can be confusing.
Methods to Reduce Gram Stain Errors
Gram stain errors can be minimized by ensuring ongoing quality assurance programs, including that routine competency assessment programs are in place in the laboratory. A good way to provide ongoing edu- cation is for the laboratory to include challenging organisms in regular plate rounds, allowing everyone to review the Gram stains after final identification of the organism. For anaerobic bacteria, preparation of the smear in anaerobic conditions and prompt fixing of the smear can help ensure that organisms will dis- play the proper Gram staining characteristic[15].
Misidentification of Organisms
Incorrect identification of microorganisms remains one of the more common errors that occur in the clinical microbiology laboratory. Phenotypic identification using biochemical testing is still the most common method by which microorganisms are identified, but there are limitations to the use of biochemical identifica- tion of bacteria. Fastidious bacteria and anaerobic bacte- ria may not grow well in these systems and often are not correctly identified. Automated systems perform well for most nonfastidious Gram-negative organisms and the most commonly isolated Gram-positive cocci such as staphylococci and enterococci. Automated sys- tems have limitations for certain groups of organisms (e.g., Gram-positive bacilli) and thus alternative meth- ods must be available for laboratories to use when they encounter organisms that are not accurately identified in an automated system. There are several different rea- sons for misidentification of organisms, and as with many other types of errors in the laboratory, organism misidentification is often a result of more than one error in the identification process.
Incorrect Algorithm or Identification Pathway Identification of organisms that have been isolated in culture is performed using many tests and para- meters in stepwise algorithms. Different groups of organisms may require different testing methodolo- gies, especially for fastidious organisms, for organism identification.
Commercial/Automated Identification System Errors
Commercial identification systems have limitations for several groups of organisms. The error rates for identification of nonfastidious Gram-negative bacteria, staphylococci, and enterococci are quite low. Fastidious
Gram-negative organisms are often misidentified by automated systems. Organisms may not grow ade- quately in the panels used in these identification sys- tems, and databases may not contain all organisms.
Examples of organisms that can be misidentified in commercial and automated systems include Brucella spp. and Francisella tularensis. Francisella is frequently misidentified asHaemophilus influenzae orActinobacillus spp. on automated systems. Brucella spp. have been misidentified asOchrobactrum anthropi,Oligella ureolyti- ca, or Psychrobacter [16 18]. Laboratories should be aware of the potential for misidentification of these organisms, and suspectedFrancisella,Brucella, and simi- lar organisms should not be tested on an automated system. Alternative methods such as DNA sequencing or mass spectrometry should be considered. This approach can not only prevent misidentification but also will ensure that the laboratory is safely handling highly pathogenic organisms that can be easily aerosol- ized. Commercial identification systems may also have limited databases, and strains of organisms with rare phenotypes may be misidentified.
CASE REPORT: BORDETELLA HOLMESII MISDENTIFIED AS ACINETOBACTER LWOFFI: In this report, four cases of bacteremia in asplenic children were described [19]. Blood cultures from all four chil- dren were positive; and Gram-negative bacilli were recovered from the cultures. The organisms were iden- tified as Acinetobacter lwoffi (99% probability) using a Vitek2 Gram-negative card. Key reactions that were used to distinguish betweenAcinetobacter andBordetella were catalase activity and growth on MacConkey agar.
These isolates were catalase negative and did not grow on MacConkey—reactions that are not consistent with Acinetobacter. DNA sequencing of the 16 S rRNA gene was used to identify these four isolates as Bordetella hol- mesii. Investigation showed that B. holmesii was not in the database for the Vitek2 instrument.
Unexpected Positive or Negative Reactions in Biochemical Methods Used to Identify Organisms
Most organisms have rare strains that do not exhibit expected biochemical reactions; when these identifica- tion characteristics are used in algorithms, an incorrect identification can result. Examples of this phenomenon areStaphylococcus aureus that are negative in the rapid latex agglutination test [20] and metabolically inactive strains of Escherichia coli that are misidentified as Shigella[21].
Misidentification of Rapidly Growing Mycobacteria
The rapidly growing mycobacteria commonly iso- lated from specimens such as blood, sputum, or tissues 321
ANALYTICAL ERRORS IN THE MICROBIOLOGY LABORATORY
(e.g., Mycobacterium abscessus, Mycobacterium chelonae, and Mycobacterium fortuitum) may be misidentified as diphtheroids. These species may grow in routine media such as liquid broth blood culture media or on solid agars used for routine bacteriology. These organ- isms can grow rapidly enough that they are recovered in routine culture. They may not be recognized as acid-fast organisms because they stain quite well with a Gram stain, and they may not always exhibit branch- ing or beaded morphologies associated with rapid growers. If one of these organisms is recovered on rou- tine culture, and the Gram stain shows Gram-positive bacilli, the organism may be incorrectly reported as diphtheroids, with no further workup.
Misidentification ofListeria monocytogenes Listeria monocytogeneshas occasionally been misiden- tified as Streptococcus agalactiae(group B Streptococcus).
Listeriacan exhibit coccobacillary morphology in Gram stains. Both of these organisms are β-hemolytic on blood agar plates, and colony morphology is similar.
Most identification algorithms call for a catalase test to be used on hemolytic colonies; occasionally, Listeria will not exhibit a strong catalase reaction. False- positive agglutination of Listeria in grouping reagents used to identify theβ-hemolytic streptococci can rarely occur. Accurate identification of Listeria and group B Streptococcus is very important because both of these organisms can cause neonatal meningitis, and antimi- crobial therapy differs significantly.
Misidentification ofBurkholderia cepacia
Burkholderia cepacia is an important organism when identified in patients with cystic fibrosis, and a report of B. cepacia is highly significant. Accurate identifica- tion ofB. cepacia, Stenotrophomonas maltophilia, and the related glucose-nonfermenting organisms is especially important in this setting. Laboratories should be aware of the possibility of misidentification of organisms, especially if they are using automated phenotypic identification systems. Burkholderia pseudomallei, rarely identified in the United States, has also been misidenti- fied asB. cepacia[22]. Isolates with a presumptive iden- tification of B. cepacia should be confirmed by an alternate method before release of the identification to the clinician.
Prevention of Misidentification Errors
Understanding the limitations of the various meth- ods of organism identification is the best way to pre- vent misidentifications from occurring. Well-designed algorithms for workup and identification of organisms can ensure that the appropriate method for identifica- tion is selected. A laboratory should have several methods available to identify microorganisms. Newer
methods such as DNA sequencing and mass spectrom- etry are much more accurate than biochemical meth- ods, and they are currently available to laboratories.
Not only can use of the newer methods result in more accurate identification but also results are often available much more quickly than with conventional methods. Careful review of results and correlating instrument reports with other data on the culture (source of isolate, growth on selective media, and man- ual biochemical test results) should be performed for every isolate.
Failure to Isolate a Pathogen: Negative Culture Results
False-negative cultures occur for many reasons.
Organisms may be rare in clinical specimens, and if the volume of material is low, culture results may be nega- tive. This is often the case with joint aspirates, where the volume of material submitted for culture is usually extremely small. False-negative cultures may also occur if a fastidious or difficult-to-culture organism is causing the infection. Examples includeLegionella,Brucella, and Kingella. For specimens such as sterile body fluids, which may contain rare or difficult-to-isolate patho- gens, the laboratory may recommend using a more sen- sitive method of culture such as lysis centrifugation or placing the specimen in a blood culture bottle.
The timing of specimen collection has significant influence on the culture results, especially when anti- microbial agents are going to be used. This is critical for specimens such as CSF for diagnosis of bacterial meningitis. The CSF may be sterilized within 30 min of administration of antimicrobial agents, resulting in false-negative cultures. Obtaining a specimen prior to administration of antimicrobial agents is critical to ensure the isolation and identification of the pathogen.
Whenever possible, the CSF should be obtained before antimicrobials are given.
Antimicrobial Susceptibility Testing Errors
Antimicrobial susceptibility testing (AST) is one of the most important functions performed by the clinical microbiology laboratory. With improvements in testing methods and quality control and quality assurance programs, AST errors are infrequent and are usually detected and corrected before results are released to the clinician. Although laboratory methods are designed to accurately detect most resistance mechan- isms, microorganisms are constantly evolving and in the setting of antimicrobial usage, resistance mechan- isms emerge. Resistance is complex, and organisms have many methods by which they can acquire or