Skip Navigation LinksHome > December 2002 - Volume 24 - Issue 12 > Sputum Cultures: Low Yield in Community‐Acquired Pneumonia
Emergency Medicine News:
doi: 10.1097/01.EEM.0000292661.17147.0e
Journal Scan

Sputum Cultures: Low Yield in Community‐Acquired Pneumonia

Bukata, W. Richard MD

Free Access
Article Outline
Collapse Box

Author Information

Dr. Bukata is the medical director of the emergency department at San Gabriel Valley Medical Center in San Gabriel, CA, an associate clinical professor in the department of emergency medicine at Los Angeles County/USC Medical Center, and the editor of Emergency Medical Abstracts.

The guidelines of the American Thoracic Society (ATS) and the Infectious Disease Society of America (IDSA) recommend that certain tests be obtained in hospitalized patients with community-acquired pneumonia (CAP). It is important to acknowledge that some recommendations will have substantially more effect on outcomes than others, and it is often enlightening to see just how justifiable recommendations are.

Enter sputum testing. The literature on sputum testing in community-acquired pneumonia is painfully consistent. In short, true sputum specimens often are not able to be obtained, only a small subset will have a predominance of a single organism (usually pneumococcal pneumonia, the most common cause of community-acquired pneumonia anyway), and therapy is rarely altered by the results of sputum testing. In other words, it sounds like sputum tests are largely a waste of time despite the admonitions of the ATS and IDSA.

Figure. Rapid BNP Te...
Figure. Rapid BNP Te...
Image Tools

To affirm this position, the most recent study of the value of sputum testing in CAP is presented and, surprise, surprise, it comes to the same conclusions as do the large preponderance of similar studies — sputum studies are of limited yield and minimal impact. So the next time someone criticizes your care because you forgot to do the “all important” sputum culture, tell them to take a hike — a hike to the library so they can catch up on the facts.

I'm not saying not to do sputum cultures or actively go against the learned recommendation of the ATS and IDSA. I am saying to not have any delusions that sputum cultures matter, especially given that empiric treatment of CAP based on the likely causative organism is by far the most effective approach.

Back to Top | Article Outline

Applying Sputum as a Diagnostic Tool in Pneumonia: Limited Yield, Minimal Impact on Treatment Decisions

Ewig S, et al Chest 2002;121:1486

BACKGROUND: Initial sputum sampling has traditionally been recommended as a guide to treatment in patients with community-acquired pneumonia, although the utility of this practice has been widely challenged.

METHODS: This prospective German study evaluated the contribution of sputum sampling to the management of 116 immunocompetent patients hospitalized for CAP in a facility without a microbiology laboratory. Attempts were made to collect sputum samples from all patients. Collected samples were evaluated in the hospital laboratory for macroscopic and microscopic appearance and Gram staining, and samples fulfilling microscopic criteria were shipped to an off-site commercial laboratory for culturing. Empiric antibiotic therapy was initiated according to regional guidelines.

RESULTS: Sputum samples were obtained from only 36 percent of the patients, with failure to obtain samples being most often due to unproductive cough efforts, weakness and noncompliance; 69 percent of obtained samples were collected one day or more after admission. Of the 42 samples collected, only about half were “microscopically valid” (due to too few leukocytes or too many squamous epithelial cells observed), and among the 23 “valid” samples Gram stains were positive in five, negative in one, and yielded mixed flora in the remaining 17. On sputum cultures of microbiologically valid samples, 12 pathogens were identified in moderate to high amounts in 10 patients (9% of the total patient population). Culture results prompted targeted antibiotic therapy in only one patient, who had not responded to initial empiric antibiotics.

CONCLUSIONS: These findings confirm the low diagnostic yield of sputum sampling in immunocompetent patients with CAP.

Back to Top | Article Outline

Heliox: More Evidence of Efficacy

If you've been reading this column for any time at all, you know that I'm a heliox fan, and don't understand why every ED doesn't contain a tank of the stuff. It only can be helpful when used in the right situations, and can't be harmful as long as oxygenation is sufficient. There are not many modalities in medicine that are so safe. Finding the next paper reaffirmed, once more, that heliox ought be used a lot more than it is. Increases in FEV1 more than doubled when albuterol was administered via heliox nebulization when compared with use of oxygen in the setting of severe asthma. Why would we not use it? What will it take to bring this potentially important adjunct into our EDs? Is there a big effort on behalf of the air and oxygen lobbyists to relegate use of heliox to the radical fringe of emergency medicine?

Back to Top | Article Outline

The Utility of Albuterol Nebulized with Heliox During Acute Asthma Exacerbations

Kress JP, et al Am J Resp Crit Care Med 2002;165:1317

BACKGROUND: Heliox is a blend of helium and oxygen that has a lower density than oxygen, and has been found to improve airflow in the setting of airway obstruction. While it has no intrinsic bronchodilating qualities, its use as a delivery vehicle may enhance the airway deposition of bronchodilators.

METHODS: In this prospective, controlled study from the University of Chicago Medical Center, 45 adults with severe acute exacerbations of asthma were randomized to receive albuterol nebulized with oxygen or with an 80:20 heliox mixture. A semi-closed delivery system was utilized to ensure delivery of high concentrations of heliox to the airways. Outcome parameters were assessed 15 minutes after each of three administrations of nebulized albuterol (0.5 ml of 0.5% albuterol in 2.5 ml of normal saline).

RESULTS: The mean baseline FEV1 was 32% of predicted in the heliox group and 33% of predicted in controls. When compared with controls, subjects in the heliox group exhibited greater increases in FEV1 after each albuterol treatment (median percent change 32% vs. 15% after the first treatment, 52% vs. 23% after the second treatment, and 65% vs. 27% after the third treatment). The heliox group exhibited a six percent mean increase in heart rate after the third albuterol treatment, compared with a 3% decrease in controls, which was felt to provide indirect evidence of increased albuterol deposition. There were no differences between the groups in hospital admission rates (about one-fourth of the patients in each treatment group).

CONCLUSIONS: In this study, improvement in FEV1 in patients with severe acute asthma was enhanced by administration of nebulized albuterol with 80:20 heliox via a semi-closed delivery system.

Back to Top | Article Outline

Elusive Positive Blood Cultures

It seems that one way for a child to move to the front of the line of patients waiting to be seen is to have a febrile seizure. This always gets everyone's attention. The following study performed at Children's Hospital of Philadelphia looked at the risk of bacteremia in children with febrile seizures, and hopefully to no one's surprise, found that it was not any higher than would be expected in similarly febrile children without seizures.

Only eight of 379 blood cultures were positive, with Streptococcus pneumoniae being the organism in seven (of which six were of serotypes now covered by the pneumococcal vaccine, which wasn't in use at the time of this study). This study is consistent with other more recent papers that indicate that children with simple febrile seizures are at no greater risk for serious bacterial illness than those with similar fevers without seizures. The tendency to do more testing in these children than those without seizures should be aggressively resisted (to my way of thinking, the fact that a third of the children had CBCs done reflects muddled thinking that shouldn't be occurring at a teaching hospital).

Back to Top | Article Outline

Low Risk of Bacteremia in Children with Febrile Seizures

Shah SS, et al Arch Ped Adol Med 2002;156:469

BACKGROUND: Rates of bacteremia in children with febrile seizures were reported to range between 2.7 and 6.7 percent in studies conducted before widespread use of H. influenzae type b immunization. Since that time, overall rates of occult bacteremia in well-appearing febrile children without an apparent focus of infection have been reported to be 1.6 to 2.6 percent, but the risk in those with febrile seizures has not been clearly established.

METHODS: In this study from the Children's Hospital of Philadelphia, a retrospective chart review was performed for 379 children aged 2 to 24 months who presented to the ED with a febrile seizure, had blood cultures drawn, and were discharged home from the ED. The study encompassed the period 1993 to 1996, after widespread use of H. influenzae type b vaccine but before availability of the pneumococcal conjugate vaccination.

RESULTS: Blood cultures yielded pathogenic organisms in 2.1 percent. S. pneumoniae was isolated in seven of the eight children with positive blood cultures (in six of these, the causative serotype was included in the currently available pneumococcal conjugate vaccine). The likelihood of a positive blood culture was unrelated to the patient's age, gender, temperature, white blood cell count (measured in 36% of the patients), or use of oral antibiotics prior to the ED visit. On reevaluation in the ED, six of the eight children with positive blood cultures were afebrile; repeat blood cultures were drawn in five and were negative. Two patients were hospitalized for persistent fever, but repeat blood cultures were negative, and neither had a diagnosis of meningitis or other serious outcome.

CONCLUSIONS: The risk of occult bacteremia does not appear to be increased among children aged 2 to 24 months who present with febrile seizures.

Back to Top | Article Outline

Fever Without a Source

The following paper is an even better example of over-testing in a place that ought to know better. My concern is that residents start believing that they are being taught state-of-the-art care because they are in so-called “high-powered” teaching programs when, in fact, in many cases the management of patients is really not optimal. In this study of 2,641 children with a fever without source, almost half had blood cultures drawn! Amazing! Of these, three percent were positive (84% being S. pneumoniae). Two children in the entire study developed serious bacterial infections, and recovered with treatment. What results would we expect now that pneumococcal vaccine is ubiquitous? No doubt, less than one percent positives.

Patients with otitis media were excluded, but virtually every other paper on the subject of fever without source in children includes those with otitis media. I heartily support this latter position. A small abscess behind the tympanic membrane doesn't cause fever of any consequence; no other abscesses do, so why attribute fevers to otitis media? It is my position that significant fever in this setting is caused by another process (typically a viral infection), and that the otitis media is a concomitant problem but not the source of the fever. This being the case, I believe that physicians should not attribute fever to otitis media and need to find another diagnosis responsible for the fever.

The paper did demonstrate that not giving antibiotics routinely to children having blood cultures had no apparent effect on outcomes, although to be fair, physicians could give them if they found a focus of infection, suspected serious bacterial infection, or the child had a positive blood culture. In addition, a huge percentage of their patients got blood cultures, many more than I think would be clinically indicated.

My personal practice is to use clinical judgment regarding the use of blood cultures, and virtually never do a CBC. However, when I do a blood culture, I cover the patient with antibiotics for a day until culture results are back. It seems to make sense to me to take the safe road, given that I thought the child was sick enough to get the test (which I don't frequently) and given that I don't know what the outcome will be. If the culture is positive, I'll be glad I did give antibiotics. Now don't write me a bunch of letters about this practice. I do blood cultures very infrequently, and as a result, I'm giving antibiotics to only a fraction of the patients compared with most physicians.

Back to Top | Article Outline

Risk of Serious Bacterial Infection in Children with Fever Without a Source in the Post-Haemophilus Influenzae Era When Antibiotics are Reserved for Culture-Proven Bacteremia

Bandyopadhyay S, et al Arch Ped Adol Med 2002;156:512

METHODS: In this study from the Children's Hospital of Wisconsin, the authors retrospectively reviewed data for 2,641 immunocompetent children aged 2 to 36 months with fever without a source (rectal temperature 39°C or higher; those with otitis media were excluded), who were treated between 1995 and 2000. Guidelines recommended testing for urinary tract infection, selective chest radiography, and blood culturing for a temperature of 40.9°C or higher and at the physician's discretion at lower temperatures. Antibiotics were recommended only for a bacterial focus of infection, suspected serious bacterial infection, or positive blood cultures. A continuous blood culture monitoring system was utilized, with a nurse-led telephone follow-up system.

RESULTS: Blood cultures were performed in 1,202 children and were positive in three percent (84% S. pneumoniae). Contaminants were isolated in an additional 3.2 percent. The mean interval to blood culture positivity was 17.5 hours. Serious bacterial infection (meningitis, septic shock, or death due to infection) occurred in two patients (0.08%), both of whom recovered with treatment. Among the 1,439 patients not having initial blood cultures, five percent returned with persistent fever without a source. Blood cultures were performed in 34 percent and were invariably negative. The number of children with fever without a source needed-to-treat with empiric antibiotics to prevent one serious bacterial infection was 1,250.

CONCLUSIONS: In this study, the risk of serious bacterial infection was not increased by a protocol reserving antibiotics for culture-proven bacteremia in children aged 2 to 36 months with fever without a source.

Back to Top | Article Outline

Communication in the ED

The next paper is absolutely fascinating. It measured communication loads in the ED, and found large numbers of communications taking place, lots of interruptions, and lots of simultaneous conversations regarding different subjects. It's a wonder we don't make more mistakes. Or is it? Unfortunately, I don't think their proposed solutions are very practical and I'm not so sure that we make all that many mistakes in the ED in the first place.

I'm very skeptical concerning the return on investment of efforts to decrease opportunities to stop miscommunication. For example, the Leapfrog Group is mandating that all hospitals to which it sends its patients throughout the United States have computerized physician order entry, a very expensive proposition that I think is entirely untested with regard to return on investment when it comes to safety. Clearly medical errors occur; however, I think there are good studies that have successfully challenged the magnitude of the problem, and although I think we can always do better, I'm not aware of any errors in our department over the past 15 years that have truly hurt patients.

Back to Top | Article Outline

Communication Loads on Clinical Staff in the Emergency Department

Coiera EW, et al Med J Australia 2002;176:415

BACKGROUND: Although communication difficulties between health care professionals waste valuable time and are a factor contributing to medical error, communication interactions in the ED have not been adequately studied.

METHODS: The authors of this Australian study observed six nurses and six physicians in two emergency departments to record and analyze communication.

RESULTS: During a total of 35 hours and 13 minutes of observation, 1,286 communication events were documented (averaging 36.5 events per person per hour). About 80 percent of the events for both nurses and physicians involved communication with ED staff. Overlapping communication, involving two or more simultaneous conversations accounted for 14.6 percent of the physicians' communication time and 6.7 percent of the nurses' communication time. Interruptions by others accounted for 30.6 percent of observed communication events (11.15 interruptions per hour). Face-to-face communication was the method employed in 82 percent of the communication events, and 94.8 percent of events involved the exchange of information.

CONCLUSIONS: Communication loads are high in the ED, and are often associated with informal exchange of information and interruptions. This pattern may negatively influence the quality of informational exchange and clinical outcomes. The authors suggest that communication in the ED may be improved by training to increase awareness of the costs of interruption, the use of asynchronous communication methods such as e-mail and voice mail, and the utilization of communal vehicles such as message boards.

Back to Top | Article Outline

Alteplase for Stroke

The following paper is extraordinarily important. I'll let it speak for itself.

Back to Top | Article Outline

Alteplase for Stroke: Money and Optimistic Claims Buttress The “Brain Attack” Campaign

Lenzer J Br Med J 2002;324:723

Awareness of manufacturer funding of published clinical trials has prompted both clinicians and the lay public to look to non-profit organizations for unbiased recommendations. The author, a medical journalist, scrutinizes the American Heart Association (AHA) recommendation for the use of alteplase for stroke. This recommendation is essentially based on the results of a single trial (NINDS), and ignores studies reporting increased rates of cerebral hemorrhage and mortality in patients treated with alteplase.

It also ignores the refusal of the NINDS investigators (and the manufacturer, Genentech) to make their raw data available for analysis, as well as some troubling methodologic elements of funded trials in addition to the negative results of a community-wide study of stroke thrombolysis from Cleveland.

National emergency medicine organizations in the U.S. and Canada have found that there is insufficient evidence to support widespread use of thrombolysis for stroke, or to consider it as a standard of care. Despite the lack of supporting evidence, the “brain attack” campaign of the AHA in the 1990s touted the value of alteplase for stroke.

It has since come to light that Genentech contributed $11 million to the AHA over the past decade, and funded construction of its national headquarters. A nine-person panel convened by the AHA to examine its recommendations in support of stroke thrombolysis had a single dissenter (Jerome R. Hoffman, MD), who was subsequently removed from the list of panelists. Six of the remaining eight who favored alteplase for stroke had financial ties to Genentech. The author notes that industry funding of nonprofit organizations may compromise their objectivity, and issues a call for impartiality and independence in the formulation of expert guidelines.

Finally, with flu season upon us, some of us will be incorporating flu tests into our practice. The following paper compared four “quickie” flu tests, and concluded that the Directigen, Quickvue and Biostar tests were similarly efficient, with the Z-Stat Flu test being not as good.

The question to be answered, however, is what role these tests should have, if any. Studies in children have shown that testing resulted in less antibiotic prescribing in those children with positive tests (a laudible outcome, to say the least).

In this setting, physicians had more confidence that the child had a viral illness, and in my opinion, felt more comfortable not diagnosing the ubiquitous ear infection as an excuse to give antibiotics to febrile children. Given that only half the cases were positive, it is clear that use of clinical judgment alone is problematic, especially if you are thinking of empirically prescribing a lot of expensive “flu pills” such as oseltamivir to buy more rapid recovery by about a day.

Back to Top | Article Outline

Evaluation of Diagnostic Tests for Influenza in a Pediatric Practice

Rodriguez WJ, et al Ped Infect Dis J 2002;21(3):193

BACKGROUND: Zanamivir (Relenza) and oseltamivir (Tamiflu) are effective against both influenza A and B, particularly if started within 48 hours after symptom onset. Early diagnosis of influenza is necessary to ensure their efficient use.

METHODS: The authors from the FDA and INOVA-Fairfax Hospital in Falls Church, VA, compared the accuracy of four rapid diagnostic tests (cost range, $14-$20) in 116 children with influenza symptoms presenting during the 1999–2000 flu season. Specimens were evaluated with the four rapid tests (Z Stat Flu [ZymeTx], Directigen Flu A [Becton Dickinson], Quickvue [Quidel] and Biostar OIA [Biostar]), and were subjected to viral culture (used as the gold standard). The Directigen test detected only influenza A, while the other three rapid tests detected influenza A and B. The Directigen test was run immediately, while the other tests were run on samples stored 12 to 24 hours after collection.

RESULTS: Viral cultures yielded influenza A in 49 percent of the specimens and other viruses in 8.6 percent. The Z Stat flu test had the lowest sensitivity (72% vs. 93–95% for the other rapid tests). Specificities of the four tests ranged between 76 and 84%, and the positive predictive values was 80 to 86%. The negative predictive value was lowest for the Z Stat Flu test (75%) and highest for the Directigen test (94%), and was 92 to 93% for the other two tests. Rates of overall efficiency were 90 percent for the Directigen Flu A, 88 percent and 86 percent for the Biostar OIA and the Quickvue, and 77 percent for the Z Stat Flu.

CONCLUSIONS: The Z Stat flu test had a lower sensitivity and negative predictive value than the other rapid influenza tests in this pediatric population. The authors note that the results might have been influenced by the predominance of influenza A in these patients and delays in test performance.

© 2002 Lippincott Williams & Wilkins, Inc.

Login

Article Tools

Images

Share