Feature Story

Time to talk turkey in microbiology labs
Do’s and don’ts of managing specimen workups

July 2004
William Check, PhD

The abrupt decline in valuation of

One symposium focused on communication between clinicians and clinical microbiologists, showing that good communication can improve testing of respiratory, urine, blood, and wound specimens. In the ideal world, the clinician would also be a laboratorian, said Joan Barenfanger, MD, PhD, D(ABMM), director of microbiology at Pathology Associates of Central Illinois Ltd., Springfield, who organized this symposium. "The clinician would examine the patient, take a specimen, walk it down to the microbiology laboratory, and do a Gram stain," she says. "And the patient would benefit from one person knowing what is happening clinically and microbiologically."

That’s not feasible, of course, but it can be approached through better communication.

Testing of cerebrospinal fluid from a lumbar puncture is an example of how improved communication can be helpful. "Since this is an invasive procedure," Dr. Barenfanger says, "some clinicians think, Now that I have this fluid I am going to order everything. They automatically order cultures for Mycobacterium tuberculosis and fungi and PCR for herpes without seeing the glucose and protein levels." In fact, if the CSF doesn’t have inflammatory cells and the glucose and protein values and the white blood cell count are all normal, typically no unusual tests (such as the AFB and fungal cultures or herpes PCR) are needed.

Reporting results of Gram stains on respiratory samples can also be improved through better communication. Karen Carroll, MD, associate professor of pathology at Johns Hopkins University School of Medicine and director, Division of Medical Microbiology, Johns Hopkins Hospital, tested two ways of reporting findings on a Gram stain of sputum. First she gave residents a report listing all organisms seen with no interpretation: moderate polymorphonuclear leukocytes; moderate Gram-positive diplococci; light Gram-negative diplococci; light Gram-positive cocci in clusters; and light Gram-negative bacilli. One resident responded, "I would ignore all this as meaningless garbage and treat empirically."

Next, Dr. Carroll handed out a modified report picking out predominant morphotypes: polymorphonuclear leukocytes and moderate Gram-positive diplococci "suggestive of Streptococcus pneumoniae," along with "light mixed organisms suggestive of normal respiratory flora." A resident commented: "I would be much more likely to treat for Streptococcus pneumoniae based on this report."

"We assume that health care providers know how to interpret a detailed report," Dr. Carroll says. "But this is not always the case. Inexperienced health care providers will either treat everything you list or ignore it all."

Determining whether a positive blood culture represents true infection or contamination also requires two-way communication. "This decision represents one example of the inseparability of the clinician and the clinical microbiologist," says Melvin Weinstein, MD, chief of the Division of Infectious Diseases, Allergy, and Immunology and director of the microbiology laboratory at the Robert Wood Johnson Medical School, New Brunswick, NJ. "Each practitioner has part of the data needed to answer the question."

Communicating with infectious disease, or ID, physicians is especially important, Dr. Weinstein says. "They have a better understanding of the laboratory’s benefits and limitations and can act as liaison to other clinicians and administrators," he says.

To illustrate, he describes the case of an HIV-positive intravenous drug user with a lumbar spinal epidural abscess treated with vancomycin, gentamicin, and ceftazidime. An ID consultant recommended drainage, but the neurosurgeon refused because of the risk. So the ID physician said to stop ceftazidime and treat for possible methicillin-resistant Staphylococcus aureus, or MRSA. A spinal tap was done which grew coagulase-negative staphylococci (CoNS) in broth only—almost certainly a contaminant but interpreted by the non-ID clinicians as evidence of staphylococcal infection. An ID physician again recommended tissue diagnosis and the neurosurgeon again refused. At this point, a young ID consultant came on service and asked if the microbiology lab could analyze as many CoNS as were available from blood cultures and other sites over the several months of the man’s illness. Among seven isolates that the lab still retained, there was no consistent species, biotype, or antibiogram. Finally, the neurosurgeon did a bone biopsy, which grew Pseudomonas aeruginosa. The patient was put on therapy and gradually recovered. Says Dr. Weinstein, "We cured this patient because an astute ID clinician appropriately requested additional information from the lab."

The laboratory assumes clinicians know what tests to order and what specimens to collect and how. However, Dr. Carroll says, clinicians are not trained in noninvasive specimen collection, which is often delegated to nonphysician health care providers. And the task of completing test requests is delegated to clerical staff with at most a high school education. "It’s no wonder that what we get in the laboratory is not always what the clinician intended," Dr. Carroll says.

To improve specimen collection for Bordetella pertussis, the laboratory at Johns Hopkins made up a "pertussis package." It contains a calcium alginate swab to use for direct culture media plating and a nonalginate swab for PCR testing. (Calcium alginate inhibits PCR reactions.) In addition, there are glass slides to inoculate for direct fluorescent antibody testing (DFA). All collection devices have labels stating their purpose. After distributing these packages, compliance increased.

More broadly, clinicians need to know what specimens to collect for various tests: nasal swabs for bacterial culture when doing surveillance for MRSA, nasopharyngeal swabs and aspirates for viruses and fastidious pathogens such as Bordetella pertussis, throat swabs for beta-streptococci, induced sputum for Pneumocystis (appropriate only for those who are HIV-positive), and expectorated sputum (screened for acceptability) for general respiratory pathogens.

Bronchoalveolar lavage algorithms should be established in conjunction with pulmonologists to help diagnose ventilator-associated pneumonia and with oncologists and other physicians who treat immunocompromised patients to look for opportunistic pathogens (Mycobacteriumsp, Legionella, Pneumocystis jirovecii).

For ventilator-associated pneumonia, or VAP, the algorithm includes cytospin, Gram stain, quantitative bacterial cultures, Legionella culture, and cytology. "For some people the whole issue of quantitative bacterial cultures is still controversial," Dr. Carroll notes. "In my experience, quantitative bacterial culture is helpful in assessing whether the patient has bacterial pneumonia."

As an example of the possible value of taking quantitation into account, Dr. Carroll describes the case of a nonimmunocompromised patient in whom a postoperative bronchoscopy was done to remove mucus. In the alveolar wash on Gram stain were seen few polymorphonuclear leukocytes and few Gram-positive cocci. On culture, moderate normal respiratory flora and rare Pseudomonas aeruginosa were recovered. Dr. Carroll’s laboratory worked up the Pseudomonas isolate, including full antimicrobial susceptibility testing. She thinks the excessive workup may have led to overtreatment.

Turning to sputum samples, Dr. Carroll lists three things that drive ID physicians crazy. First is complete workup on all potential pathogens present without regard to quantity on the Gram stain. Guidelines for limiting what is reported are presented in the 2004 Cumitech 7B.

Second is identification of yeasts other than Cryptococcus. "Candida pneumonia is very rare and is a histopathological diagnosis," Dr. Carroll says. "You need to see hyphae invading tissue." Cryptococcus in sputum is probably significant and should be reported.

Third is the report that does not contain Gram stain, culture, and antimicrobial susceptibility testing results on the same screen.

Dr. Barenfanger highlights a case of a "look-alike" dilemma in respiratory microbiology. A sample contained many WBCs and many intra- and extracellular Gram-negative organisms suggestive of Moraxella catarrhalis, which a technologist reported. The next day Acinetobacter grew in culture. It was an embarrassing mistake, Dr. Barenfanger says, but the response was simple: "If you predict incorrectly, fix it." She notified the clinician, amended the written report, and educated the technologist. In this case the technologist’s error was understandable. The two organisms look similar and Moraxella is a more frequent cause of community-acquired pneumonia than Acinetobacter, so 99 percent of predictions of Moraxella are correct. Acinetobacter, on the other hand, is a more common cause of nosocomial pneumonia, particularly in a patient on a ventilator. "When Moraxella-like morphotypes are seen in a Gram stain and the patient has been hospitalized for more than three days," Dr. Barenfanger says, "consider this comment: ’Bacteria morphologically suggestive of Moraxella and consistent with Acinetobacter.’"

Dr. Barenfanger also suggests calling the physician or specifying in your report when you get a strong sense from the Gram stain that there has been aspiration pneumonia. "Unlike typical pneumonia, we won’t grow most of those organisms the next day, because aspiration pneumonia involves organisms we don’t routinely grow from sputum," she says. "And we shouldn’t even try to grow them." But the laboratory needs to inform the clinician, perhaps by saying in the report, "Variety of organisms highly suggestive of aspiration pneumonia; consider coverage for anaerobic bacteria as well as standard pathogens."

Adding a point to Dr. Carroll’s advocacy of semiquantitative cultures on bronchoalveolar lavages, Dr. Barenfanger recommends getting official approval for this procedure. "Be proactive," she says. "Ask for physician input in writing and get approval from the medical executive committee. And charge for it." She has instituted this policy at Memorial Medical Center, Springfield, Ill.

Some reports from the microbiology lab are of no help to clinicians and may lead to unnecessary treatment. Dr. Barenfanger lists several for respiratory specimens, starting with a report of a throat culture saying "No group A beta-strep, but MRSA present."

"We do throat cultures for only one reason," she says, "to look for group A beta-hemolytic strep. So the report should say ’No group A beta-strep.’ MRSA is not a pathogen in that site."

In another case, a clinician at Dr. Barenfanger’s hospital sent a throat culture to an outside laboratory. When he got a report listing "Many alpha-strep, Neisseria spp, moderate Micrococcus spp," he called Dr. Barenfanger and asked whether he should treat. She says, "Normal pharyngeal flora, such as Staphaureus, Streppneumoniae, and Haemophilus influenzae, should not be reported on routine throat cultures."

A different unintended result came from a respiratory specimen in which a lab reported "Many Neisseria meningitidis." A clinician ordered a lumbar puncture based solely on this report.

It is also not helpful to accede to clinician requests for meaningless tests. Often during flu season Dr. Barenfanger’s lab will get a request for "influenza serology," which is important only for epidemiological investigations. "What they really want is influenza culture or antigen test," Dr. Barenfanger says. Developing acceptability criteria for tests submitted for serologic testing can help avoid producing useless results (Crump JA, et al. J Clin Microbiol. 2004;42:881-883).

Some clinicians will order routine culture of sputum when they suspect Legionella. However, routine culture will not pick up Legionella, which is responsible for 29 percent of nosocomial respiratory infections. "Some literature says less than 15 percent of hospitals test sputum for Legionella, even though mortality is about 50 percent. I don’t know how to answer that problem," Dr. Barenfanger says.

Microbiological diagnosis of urinary tract infections has been changed in several ways, Dr. Carroll says, including changes in the criteria for defining significant bacteriuria, lab consolidation, emergence of resistant organisms, larger numbers of immunocompromised patients, and changes in technology. Technological advances now allow some patients who previously would have had chronic indwelling catheters or bladders draining into external bags to have functional bladders surgically created out of intestines and hooked up to the distal urethra, Dr. Carroll says. "The surgeon who does these doesn’t want us to report low colony counts."

At the other end of the spectrum are spinal-cord-injured patients who have neurogenic bladders. Some get indwelling catheters while others learn to catheterize themselves. But without functioning bladders they get recurring infections, often with multiple organisms. "So we may have to work up everything that grows from these patients," Dr. Carroll says.

In children up to two months of age or in febrile children up to two years who require prompt treatment, specimens should be collected via catheterization or suprapubic aspiration. For febrile children between two months and two years who do not require urgent treatment, a urine (U) bag is acceptable. If the U bag is positive by Gram stain, a sterile sample is collected for culture (American Academy of Pediatrics. Pediatrics. 1999;103:843).

For adults, the requisitions should allow the clinician to specify the type of sample submitted. "Culture if" options for ambulatory patients are helpful. "Some family practitioners don’t want to do dipstick urines in the clinic," Dr. Carroll says. "They send us a specimen to screen for pyuria or bacteriuria. If either is present, we can move on to culture."

Specimen collection in adults has typically specified a midstream, clean-catch, early-morning sample, which Dr. Carroll concedes is optimal. But it’s time-consuming, and the instructions can be embarrassing. "Many studies support discontinuation of midstream clean-catch samples," she says. The patient just needs to be told not to touch the cup to his or her body.

Criteria for defining significant bacteriuria differ among populations. For a symptomatic woman, it requires B310² coliforms/mL or B310⁵ non-coliforms/mL. For symptomatic men or catheterized patients, the threshold is B310³ CFU/mL, while for asymptomatic individuals it is B310⁵ CFU/mL.

For susceptibility testing, Dr. Carroll says clinicians want urine-specific antibiograms and identification of vancomycin-resistant enterococci, MRSA, and extended-spectrum beta-lactamase-producing Enterobacteriaceae in nosocomial cases.

Gerri Hall, PhD, staff microbiologist in the Department of Clinical Pathology, Section of Clinical Microbiology, Cleveland Clinic Foundation, discussed screening urine samples, which can lighten the workload and get negative results to clinicians faster. Tools for urine screening include Gram stain, dipstick, UTIScreen (Coral Diagnostics), Cellenium (Trek Diagnostics), and others. Microscopic screening is helpful because 90 percent of patients with UTIs have pyuria. Gram stain is rapid, reliable, and inexpensive and has high sensitivity when correlated with colony count greater than 10⁵ CFU/mL.

At the Cleveland Clinic, Dr. Hall uses UTIScreen followed by culture of positives on normal catheter urines, mid-void urines, suprapubic catheters, VB1, and ureterostomy and nephrostomy tubes. Culture is done without screening on several types of samples, such as grossly bloody urine or urines obtained in surgical procedures.

Culture is not always necessary in women with dysuria and pyuria. Cultures are indicated for patients with complicated or uncertain clinical features, those with a UTI in the past three weeks (possible relapse), persons with symptoms for more than seven days, recent hospitalization or catheterization (which indicates possible nosocomial infection), or pregnancy or diabetes.

In addition to standard culture, Dr. Hall is evaluating Chromagar orientation media, a modification of chromogenic agar for identifying and quantifying most common UTI pathogens. It can help see a mixed culture (for example, E. coli, enterococcus, and Proteus) more easily than blood agar plates or MacConkey agar. "It’s my impression that Chrom agar provides quick ID with few chemicals and may be cost-effective on screened specimens," she says. It requires more than 16 hours of incubation (D’Souza HA, et al. J Clin Microbiol. 2004; 42: 60-64).

When deciding which isolates to work up, good communication between laboratory and clinicians is mandatory. At the Cleveland Clinic, Dr. Hall works up one or two organisms present at greater than 104/mL. For three or more organisms, one needs to weigh the value of working up versus reporting "Mixed flora, call lab if further work needed." When more than one organism is present, rules for workup vary depending on whether one is predominant. "This algorithm requires physician buy-in," Dr. Hall says.

Deciding when to do antimicrobial susceptibility testing on urine cultures is challenging. Probably not all isolates should get AST. One can go by physician request, as long as it’s reasonable. Algorithms based on colony counts and specific pathogens are helpful. "In general we do a susceptibility test if the isolate is worked up—if we give it a name," Dr. Hall says.

Communication between laboratory and clinician is necessary when deciding whether to treat a given blood isolate: One-third to nearly one-half of all positive blood cultures represent contamination. Clinicians know about fever, leukocytosis, and imaging studies; laboratorians know the identity of the organism. For instance, Propionibacterium acnes is nearly always a contaminant.

Shorter time to detection and higher number of bottles positive in a set have been taken as indicators of true infection. While there is a correlation on a population basis, Dr. Barenfanger says, "work by Dr. Weinstein has shown that doesn’t help with individual patients" (Weinstein MP, et al. Clin Inf Dis. 1997;24:584-602).

Some blood isolates are highly likely to be significant: S. aureus, S. pneumoniae, E. coli, Klebsiella pneumoniae, P. aeruginosa, and enterococci; probably not significant are viridans streptococci and CoNS. "Some experts disagree about viridans strep," Dr. Barenfanger says. "It is only significant if two out of two blood cultures are positive." Whether one of two blood cultures positive for CoNS is important depends on the situation. In a patient with no line in, it is probably unimportant. If only the line is positive, that is also probably not significant.

Finally, Dr. Barenfanger issued a call for help to LIS vendors. With electronic systems, it is extremely difficult to report several organisms from multiple sites along with antibiotic susceptibility data in an easily understood format. "We could do it by hand, but that increases error rates," she says.

For example, Dr. Weinstein and others have shown that virtually all episodes of bacteremia and fungemia can be detected by two to three blood cultures of adequate volume. Sensitivity is therefore good, but specificity can be problematic. As Dr. Barenfanger had indicated, CoNS pose a particular problem. "Coagulase-negative staph are the bane of our existence," Dr. Weinstein says. "They are exceedingly common but most often contaminants." True bacteremia caused by CoNS is most often seen in people with central venous catheters, prosthetic heart valves, or vascular grafts.

Also difficult is predicting which blood cultures will be positive—85 to 90 percent are negative. "We don’t currently have predictors of bacteremia that raise the pretest probability of getting a positive result," Dr. Weinstein says. "The most sensitive predictor is fever, but fever occurs in many conditions without bacteremia."

In his 1997 article, Dr. Weinstein found several prognostic factors for outcome of bacteremias: age >70, nosocomial episode, primary source in lungs or bowel, underlying malignancy or AIDS, inappropriate antibiotic therapy, and body temperature. Similar data were reported by Diekema (J Clin Microbiol. 2003;41: 3655-3660).

Appropriateness of therapeutic intervention was evaluated at three time points—before culture results (empiric), after a positive blood culture, and after antimicrobial susceptibility testing. Mortality ranged from a low of 10 percent when therapy at all three time points was judged appropriate to more than 30 percent when two or more time points were inappropriate. About three-fourths of episodes had appropriate therapy throughout.

About half of all therapeutic interventions by physicians occur within four hours after the blood culture is obtained, which is empiric therapy. Laboratories can influence treatment at two places: by phone notification of a positive blood culture and Gram stain and by reporting of antimicrobial susceptibility testing results. Analysis of 509 bloodstream infections showed that notification of a positive result and Gram stain had greater influence (Munson EL, et al. J Clin Microbiol. 2003;41: 495-497). Both types of information are clinically important, Dr. Weinstein notes. His interpretation: "Earlier information gives you more impact."

Dr. Hall calls wounds "the most difficult topic" to talk about in clinical microbiology. And the worst collection device for wounds is the swab, she says: "If we work up swabs from wounds, we are doing a disservice to clinician and patients." She has been asked, "How do I transport this swab for a ’rule out anaerobes in this wound’?" Her answer: "Do not send a swab for anaerobes. Aspirates or tissues or body fluids are acceptable specimens."

Many other questions about swabs also demand a "no" answer:

• Should I send a swab for nasal culture? No, except for surveillance of specific pathogens.
• Should I send a swab of donor tissue for a transplant? No.
• Should I send an ulcer swab? No.
• Should I send a swab of the ear for otitis? No.

Dr. Hall notes that wound specimens can be triaged by a quality (Q) score proposed initially by Raymond Bartlett, MD, and written about recently by Susan Sharp, PhD, now director of microbiology at Kaiser Permanente, Portland, Ore. (Sharp S. Clinical Microbiology Newsletter. 1999;21:118-120). A sample is acceptable if WBCs exceed epithelials on Gram stain; if epithelials exceed WBCs, do not process and note the reason on the chart. Epithelial cells and WBCs are given a score from 0 to 3 depending on their frequency, and the Q score is derived from these two numbers. Workup depends on the Q score, as described in Dr. Sharp’s paper.

"Consider rejecting requests for broth cultures on wound specimens," Dr. Hall advises. "They often grow coagulase-negative staph and don’t add to the information obtained from a good Gram stain" (Morris AJ, et al. J Clin Microbiol. 1996; 34: 355-357).

For reporting, Dr. Hall is trying to switch from giving an MIC number to interpreting the result as "sensitive" or "resistant." Dr. Barenfanger, too, has recently stopped reporting MICs. "Even though infectious disease physicians wanted that information, the vast majority of physicians do not use it correctly," she says. For example, confronted with an MIC of 20-40 for TMP/SMX, a family practice resident asked, "How much drug do I have to give to get that level?"

Ann Robinson, PhD, D(ABMM), director of microbiology and virology at Sacred Heart Medical Center and Pathology Associates Medical Laboratories, Spokane, Wash., provided tips for managing clinically irrelevant test orders. She spoke at the second symposium, which marked the 30th anniversary of the publication of Dr. Bartlett’s textbook, Medical Microbiology: Cost, Quality and Clinical Relevance.

"Full" throat cultures are undesirable because they identify colonizing microorganisms. "Group A streptococci are the most common bacterial cause of pharyngitis and the only cause of pharyngitis for which antimicrobial therapy is clearly indicated," she says. If a clinician insists on complete workup of organisms from a nasopharyngeal swab and you isolate S. pneumoniae, H. influenzae, M. catarrhalis, or S. aureus, report "mixed flora including [organism(s) isolated]." You can add a prevalence comment pertinent to the organism isolated, such as: "As many as 75 percent of healthy individuals harbor S. pneumoniae in the upper respiratory tract."

If nasopharyngeal culture from a patient under age 12 grows S. pneumoniae, H. influenzae, or M. catarrhalis, add this comment: "Nasopharyngeal culture should not be used for microbiological diagnosis of otitis media because of the high prevalence of asymptomatic carriage of potential pathogens. Tympanocentesis is the recommended procedure for collecting specimens" for otitis media.

Requesting Legionella direct fluorescent antibody without culture is a suboptimal order. Attach a comment giving the inferior sensitivity (25-75 percent) and specificity (95-99 percent) of DFA and recommending culture, showing its higher numbers (80-99 percent and 100 percent).

Requests for environmental cultures are frequently not indicated, Dr. Robinson says. The CDC does not recommend routine, undirected culturing of air, water, or surfaces in health care facilities, nor are there guidelines for interpreting the results of such testing. Microbiologic culturing for quality assurance purposes is appropriate for monitoring of sterilization processes and monthly water and dialysate cultures in hemodialysis units.

Assessing the quality of lower respiratory tract specimens by Gram stain identifies superficially contaminated specimens and increases diagnostic utility of culture. Evidence supports the utility of Gram stain of sputum smears relative to culture in several situations, Dr. Robinson showed: pneumonia in general, community-acquired pneumonia caused by Acinetobacter, pneumococcal pneumonia, and H. influenzae pneumonia. A specimen is considered to be of good quality if it contains an abundance of neutrophils relative to the number of squamous epithelial cells per 100x field.

Poor-quality specimens may be reported by saying: "Squamous cells in the specimen indicate the presence of superficial material that may contain contaminating or colonizing bacteria unrelated to infection." Suggest re-collection but hold the specimen.

The goal in clinical microbiology, Dr. Robinson sums up, is to "restrict testing to high-quality samples that will improve the clinical value of information." And clear communication between the lab and the pertinent clinical services, she adds, is central to improving quality and containing cost.

William Check is a medical writer in Wilmette, Ill.