Spectrum bias in the evaluation of diagnostic tests: lessons from the rapid dipstick test for urinary tract infection

OBJECTIVE

To determine if the leukocyte esterase and bacterial nitrite rapid dipstick test for urinary tract infection (UTI) is susceptible to spectrum bias (when a diagnostic test has different sensitivities or specificities in patients with different clinical manifestations of the disease for which the test is intended).

DESIGN

Cross-sectional study.

PATIENTS

A total of 366 consecutive adult patients in whom clinicians performed urinalysis to diagnose or exclude UTI.

SETTING

An urban emergency department and walk-in clinic.

MEASUREMENTS

After the patient encounter, but before dipstick test or culture was done, clinicians recorded the signs and symptoms that were the basis for suspecting UTI and for performing a urinalysis and an estimate of the probability of UTI based on the clinical evaluation. For all patients who received urinalysis, dipstick tests and culture were done in the clinical microbiology laboratory by medical technologists blinded to clinical evaluation. Sensitivity for the dipstick was calculated using a positive result in either leukocyte esterase or bacterial nitrite, or both, as the criterion for a positive dipstick, and greater than 10(5) CFU/mL for a positive culture.

RESULTS

In the 107 patients with a high (greater than 50%) prior probability of UTI, who had many characteristic UTI symptoms, the sensitivity of the test was excellent (0.92; 95% CI, 0.82 to 0.98). In the 259 patients with a low (less than or equal to 50%) prior probability of UTI, the sensitivity of the test was poor (0.56; CI, 0.03 to 0.79).

CONCLUSIONS

The leukocyte esterase and bacterial nitrite dipstick test for UTI is susceptible to spectrum bias, which may be responsible for differences in the test's sensitivity reported in previous studies. As a more general principle, diagnostic tests may have different sensitivities or specificities in different parts of the clinical spectrum of the disease they purport to identify or exclude, but studies evaluating such tests rarely report sensitivity and specificity in subgroups defined by clinical symptoms. When diagnostic tests are evaluated, information about symptoms in the patients recruited for study should be included, and analyses should be done within appropriate clinical subgroups so that clinicians may decide if reported sensitivities and specificities are applicable to their patients.

Molecular diagnosis of Ureaplasma urealyticum septic arthritis in a patient with hypogammaglobulinemia

OBJECTIVE

We report a hypogammaglobulinemic patient with a destructive oligoarticular arthritis, whose synovial fluid cultures were repeatedly sterile.

METHODS AND RESULTS

We identified a Ureaplasma urealyticum infection in his affected joints, using a polymerase chain reaction (PCR) assay.

CONCLUSION

The PCR technique promises to be extremely valuable in the rapid and specific diagnosis of infectious arthritis.

In vitro activity of azithromycin against clinical isolates of Legionella species

The activities of azithromycin, erythromycin, and ciprofloxacin against 21 Legionella isolates were measured by an agar dilution method and in macrophages. The MICs for 90% of strains tested were 2.0, 1.0, and 0.5 micrograms/ml for azithromycin, erythromycin, and ciprofloxacin, respectively. Azithromycin and ciprofloxacin were both bactericidal in the macrophage system, but erythromycin was bacteriostatic.

Rifampin resistance of Legionella pneumophila is not increased during therapy for experimental Legionnaires disease: study of rifampin resistance using a guinea pig model of Legionnaires disease

Isolates of Legionella pneumophila serogroup 1, obtained from guinea pigs with experimentally induced Legionnaires disease, were tested for rifampin resistance. Thirteen isolates were from animals treated with rifampin alone, four isolates were from animals treated with saline, and three isolates each were from animals treated with erythromycin or erythromycin plus rifampin; all of these isolates were derived from the same parent strain, F889. Most of the isolates were obtained from rifampin-treated animals that survived infection but had persistence of bacteria in their lungs at necropsy. No differences in rifampin agar dilution MICs were detected for the 23 isolates and parent strain that were tested. None of the 13 isolates from animals treated with rifampin alone had a high number of resistant organisms detected by using a rifampin gradient plate assay. Thirteen isolates plus the parent strain were tested by using a quantitative method of determining resistance frequency. Considerable heterogeneity among isolates was observed, but there was no evidence of increased resistance for any treatment group. The range of rifampin resistance frequencies was 10(-7) to 10(-8). No evidence for rifampin-induced resistance of L. pneumophila was found in this study.

In vitro activity of sparfloxacin (CI-978; AT-4140) for clinical Legionella isolates, pharmacokinetics in guinea pigs, and use to treat guinea pigs with L. pneumophila pneumonia

The activities of sparfloxacin, ciprofloxacin, and erythromycin for 21 clinical Legionella isolates were determined by agar and broth dilution susceptibility testing and by growth inhibition assays in guinea pig alveolar macrophages (sparfloxacin and ciprofloxacin). All three antimicrobial agents had roughly equivalent activities when buffered charcoal yeast extract agar medium supplemented with 0.1% alpha-ketoglutarate was used as the test medium; the MICs for 90% of strains were 1.0 micrograms/ml for erythromycin and sparfloxacin and 0.5 microgram/ml for ciprofloxacin. Buffered charcoal yeast extract medium supplemented with 0.1% alpha-ketoglutarate inhibited the activities of all the antimicrobial agents tested, as judged by the susceptibility of a control Staphylococcus aureus strain. Broth macrodilution MICs for two L. pneumophila strains in buffered yeast extract supplemented with 0.1% alpha-ketoglutarate were less than or equal to 0.03 microgram/ml for sparfloxacin, 0.06 microgram/ml for ciprofloxacin, and 0.25 microgram/ml for erythromycin; only erythromycin was inhibited by this medium. Ciprofloxacin and sparfloxacin (both 0.25 microgram/ml) reduced bacterial counts of two L. pneumophila strains grown in guinea pig alveolar macrophages by 2 log10 CFU/ml, but regrowth occurred over a 3-day period. Sparfloxacin, but not ciprofloxacin (both 1 microgram/ml), caused a 3- to 4-day postantibiotic effect. Pharmacokinetic and therapy studies of sparfloxacin were performed in guinea pigs with L. pneumophila pneumonia. For the pharmacokinetic study, sparfloxacin was given (10 mg/kg of body weight) to infected guinea pigs by the intraperitoneal route; peak levels in serum and lung were 2.6 micrograms/ml and 1.6 micrograms/g, respectively, at 1 h, with a terminal-phase half-life of elimination from serum of 5 h.(ABSTRACT TRUNCATED AT 250 WORDS)

Inappropriate testing for diarrheal diseases in the hospital

To assess the degree to which routine stool cultures, ova and parasite examinations, and Clostridium difficile toxin assays may be inappropriately ordered on hospitalized patients, we conducted a retrospective study to determine the relative yield of these tests on specimens collected from outpatients and inpatients as a function of time after admission. During a 3-year period, only 1 of 191 positive stool cultures and none of the 90 ova and parasite examinations with positive results were from the group of patients who had stool specimens submitted after 3 days of hospitalization. Analysis of laboratory work load for a 1-year period showed that specimens from this patient group contributed nearly 50% of the more than 3000 specimens received each year. In contrast, approximately 25% (range, 17% to 33%) of samples, regardless of admission status, were positive for C difficile toxin. Eliminating routine stool cultures and ova and parasite examinations on hospitalized patients would significantly reduce hospital and patient costs without altering patient care. Nationwide, such a policy might achieve a cost savings of +20 to +30 million per year.

WIN 57273 is bactericidal for Legionella pneumophila grown in alveolar macrophages

The in vitro antimicrobial activity of WIN 57273, a new quinolone antimicrobial agent, was determined for 21 Legionella strains, using broth macrodilution and agar dilution testing methods; ciprofloxacin and erythromycin were tested as well. Three different buffered yeast extract media were used for the agar dilution studies, two of which were made with starch rather than charcoal. Broth macrodilution susceptibility testing was performed with buffered yeast extract broth and two Legionella pneumophila strains. Antimicrobial inhibition of L. pneumophila growth in guinea pig alveolar macrophages was also studied, using a method able to detect bacterial killing. The MICs for 90% of the 21 strains of Legionella spp. grown on buffered charcoal yeast extract medium were 0.125 microgram/ml for WIN 57273, 0.25 microgram/ml for ciprofloxacin, and 1.0 micrograms/ml for erythromycin. These MICs were falsely high, because of inhibition of drug activity by the medium used. Use of less drug-antagonistic, starch-containing media did not support good growth of the test strains. The broth macrodilution MICs for two strains of L. pneumophila serogroup 1 were less than or equal to 0.03 microgram/ml for WIN 57273 and ciprofloxacin and 0.125 microgram/ml for erythromycin. WIN 57273, ciprofloxacin, and erythromycin all inhibited growth of L. pneumophila in guinea pig alveolar macrophages at concentrations of 1 microgram/ml, but only WIN 57273 prevented regrowth or killed L. pneumophila after removal of extracellular antimicrobial agent.

Evaluation of the Merifluor-Legionella immunofluorescent reagent for identifying and detecting 21 Legionella species

We evaluated a 33-valent polyclonal indirect immunofluorescent-reagent kit (Merifluor-Legionella; Meridian Diagnostics Inc., Cincinnati, Ohio) made for the detection of Legionella species by testing bacterial isolates, seeded sputum, and negative sputum samples. Use of the reagent according to the directions of the manufacturer gave false-negative staining of homologous culture isolates due to a prozone phenomenon; this was solved by diluting test strain suspensions. After this change in testing protocol was made, the reagent gave bright fluorescent staining with 31 of the 33 Legionella strains with which it supposedly reacts. Strongly reacting Legionella strains included the type strains of L. pneumophila serogroups 1 to 10, L. longbeachae serogroups 1 and 2, and serogroup 1 of L. anisa, L. bozemanii, L. cherrii, L. dumoffii, L. gormanii, L. hackeliae, L. jamestowniensis, L. jordanis, L. maceachernii, L. micadedi, L. oakridgensis, L. rubrilucens, L. sainthelensi, L. spiritensis, L. steigerwaltii, and L. wadsworthii. Type strains of L. erythra and L. feeleii fluoresced only dimly with the reagent. Of 10 non-Legionella bacteria known to cross-stain with other polyvalent antisera, 5 also cross-reacted with the Merifluor reagent; these included 3 Bacteroides fragilis and 2 Pseudomonas fluorescens strains. The lower limit of detection of L. pneumophila serogroup 1 in seeded sputum was about 5 x 10(4) to 5 x 10(5) cells per ml. None of 21 randomly collected sputum specimens tested contained fluorescing legionellalike organisms, but 6 specimens did contain brightly fluorescing bacteria atypical in morphology for Legionella species. The Merifluor-Legionella kit appears to perform as well as other polyclonal immunofluorescent reagents used for detection of Legionella species. Because of the cross-reactions observed, which are common to all polyclonal reagents, utilization of this reagent for either bacterial identification or detection must be performed in combination with culture.

In vitro activity of lomefloxacin (NY-198 or SC 47111), ciprofloxacin, and erythromycin against 100 clinical Legionella strains

Agar dilution minimum inhibitory concentrations (MICs) of lomefloxacin (LO), ciprofloxacin (CI), and erythromycin (ER) were determined for 100 clinical isolates of Legionella using buffered charcoal yeast extract medium supplemented with alpha-ketoglutarate (BCYEa). The Legionella strains tested included 84 L. pneumophila, 2 L. micdadei, 6 L. dumoffii, 4 L. longbeachae, and one each of L. bozemanii, L. hackeliae, L. wadsworthii, and L. maceachernii. Geometric mean MICs microgram/ml were 0.56 for LO, 0.50 for CI, and 0.25 for ER. Ninety percent MICs were 1.0 for LO and CI, and 0.5 for ER. All Legionella strains except one (L. hackeliae) were inhibited by 1.0 microgram/ml of LO or CI; this strain had an Er MIC of 1.0 microgram/ml and was inhibited by 2.0 micrograms/ml of LO or CI. Control strains of Eschershia coli and Staphylococcus aureus were also tested on both BCYEa and Mueller-Hinton agar (MHA) to determine BCYEa-mediated inhibition of the antimicrobials. All three antimicrobials were inactivated in varying degrees by BCYEa. BCYEa:MHA MICs of the S. aureus control strain were 4:1 for LO and ER, and greater than 4:1 for CI. BCYEa: MHA MICs of the E. coli control strain were 4:1 for LO and CI, and unmeasurable for ER. Both LO and CI have good in vitro activity against legionella, probably greater than that measured in this study because of antimicrobial inactivation by BCYEa. Since LO is concentrated by phagocytic cells, like ER and CI, it is likely that it will be effective in the treatment of Legionnaires' disease.

Use of DNA probes for the diagnosis of infections caused by Mycoplasma pneumoniae and legionellae--a review

Specific DNA probes have been made for both M. pneumoniae and Legionella species. Dot blot methods have been used in research laboratories to test culture isolates of both organisms, and also to test animal tissues with a L. pneumophila-specific probe. Commercial kits are also available for direct specimen testing for these two organisms. The commercial kits are made by a single manufacturer, Gen-Probe, Inc. (San Diego, CA), and use a novel in-solution rapid hybridization assay, using 125I-labeled cDNA to rRNAs of the organisms. The Gen-Probe M. pneumoniae probe appears to be 80% to 100% sensitive, and 97% to 100% specific, based on analysis of two clinical studies using positive culture as the diagnostic criterion. The Gen-Probe legionella probe appears to be 33% to 71% sensitive (mean 57%), and 98.9% to 99.7% specific (mean 99.7%), based on analysis of four prospective clinical studies, using positive culture as the definition of disease, with a total sample size of 3,243 patients, 49 of which were culture-positive. Both Gen-Probe direct tests appear to be clinically useful, although the poor performance of the legionella test in one major university laboratory, and the expense of performing these tests, mandate that thorough evaluations be carried out in each laboratory anticipating using the test. Culture must always be performed for legionella whether or not the DNA probe test is used. It is likely that the use of the M. pneumoniae kit would greatly speed diagnosis, but whether this would alter medical practice or result in lower morbidity and health care costs is unknown.

Retrospective evaluation of the Du Pont radioimmunoassay kit for detection of Legionella pneumophila serogroup 1 antigenuria in humans

We used the Du Pont radioimmunoassay kit for soluble Legionella pneumophila serogroup 1 antigenuria (Du Pont Co., Wilmington, Del.) to test 422 urine samples from patients with and without Legionnaires disease (LD). The urine specimens were collected from 23 patients with culture-proven LD and from 346 patients without LD. L. pneumophila serogroup 1 was isolated from 14 patients with culture-proven LD, and other L. pneumophila serogroups or other Legionella species were isolated from 9 patients; 58 urine specimens were tested from these 23 patients. The non-LD group was composed of 75 bacteremic patients (35 gram-negative and 40 gram-positive bacteremias), 7 patients with candidemia, 48 patients with non-LD pneumonia, 90 patients with gram-negative bacteriuria (greater than 10(5) CFU/ml), 23 patients with gram-positive bacteriuria (greater than 10(5) CFU/ml), 14 patients with candiduria (greater than 10(5) CFU/ml), and 89 outpatients with negative urine cultures. All tests were performed in duplicate, including positive and negative controls. Sample results with values greater than or equal to 3.0 times those of the negative controls were considered positive for L. pneumophila serogroup 1 antigenuria. The average sample-to-negative ratios were 19.1 for the L. pneumophila serogroup 1 specimens, and 1.0 for both the non-serogroup 1 legionella group and the non-LD specimens. All but one of the patients who were culture positive for L. pneumophila serogroup 1 had at least one specimen positive for serogroup 1 antigenuria; none of the non-L. pneumophila serogroup 1 patients had a positive urine test. The test was highly specific (100%) and sensitive (93%) for the detection of L. pneumophila serogroup 1 antigenuria. Concentrations of urine by vacuum evaporation increased test sensitivity without apparently affecting specificity.

Legionella pneumophila serogroup Lansing 3 isolated from a patient with fatal pneumonia, and descriptions of L. pneumophila subsp. pneumophila subsp. nov., L. pneumophila subsp. fraseri subsp. nov., and L. pneumophila subsp. pascullei subsp. nov

Previous DNA relatedness and enzyme electrophoretic mobility studies indicated heterogeneity among strains of Legionella pneumophila serogroups 1, 4, 5, and Lansing 3 (a new, as yet unnumbered serogroup). In this study 60 L. pneumophila strains were studied by DNA hybridization (hydroxyapatite method) to assess their genomic relatedness. These strains were also studied biochemically and serologically to determine whether they formed one or more phenotypic groups. DNA relatedness studies identified three groups. DNA group 1 contained the type strain Philadelphia 1 and strains from serogroups 1 through 14 of L. pneumophila. The average relatedness of DNA group 1 strains was 88% at 60 degrees C with 1.1% divergence in related sequences and 85% at 75 degrees C. DNA group 2 contained strain Los Angeles 1, the reference strain of serogroup 4, and strains of serogroups 1, 4, 5, and Lansing 3, an unnumbered serogroup. Average relatedness of DNA group 2 strains was 84% at 60 degrees C with 0.7% divergence and 87% at 75 degrees C. Reciprocal relatedness of DNA groups 1 and 2 was approximately 67% at 60 degrees C with 6.0% divergence and 48% at 75 degrees C. DNA group 3 strains were in serogroup 5. They were 98% related at 60 degrees C with 0.5% divergence and 97% related at 75 degrees C. Reciprocal relatedness of DNA group 3 and DNA group 1 was approximately 74% at 60 degrees C with 5.3% divergence and 43% at 75 degrees C, and reciprocal relatedness of DNA groups 3 and 2 was 66% at 60 degrees C with 5.7% divergence and 55% at 75 degrees C. The DNA groups could not be separated biochemically or serologically or by cell wall fatty acid and isoprenoid quinone composition. Three subspecies of L. pneumophila are proposed to accommodate the three DNA groups: L. pneumophila subsp. pneumophila subsp. nov. for DNA group 1, L. pneumophila subsp. fraseri subsp. nov. for DNA group 2, and pneumophila subsp. pascullei subsp. nov. for DNA group 3.

Nosocomial Legionnaires' disease: a global perspective

Nosocomial Legionnaires' disease is a worldwide problem. The lack of prospective surveys using sensitive diagnostic means, such as culture, has resulted in ignorance about the exact magnitude of the problem except at a handful of individual hospitals. Contaminated hospital hot water systems, cooling towers, and non-sterile tap water used for respiratory therapy may all cause disease. Culture diagnosis is the preferred method of case ascertainment, and provides bacterial isolates which can be used to implicate specific environmental sites using molecular epidemiologic techniques. Environmental culture surveys, done in the absence of known disease, are probably not indicated, but there remains concern that wards housing very high risk patients should be legionella-free. Use of monoclonal antibody typing of environmental isolates of L. pneumophila serogroup 1 may be useful in this regard, as it can possibly detect strains most likely to cause disease. Control of nosocomial disease requires centralized management and consultation by expert engineers, epidemiologists, and microbiologists.

The laboratory diagnosis of Legionnaires' disease

Legionnaires' disease can be diagnosed by culture, by direct detection of the bacterium or its products by using immunologic means, with a DNA probe, and by serologic means. Culture diagnosis is the most specific and sensitive test. Legionellae can be isolated from sputum samples by using selective techniques. Antibody detection is more suited for epidemiologic purposes than for use in individual cases. Immunofluorescent microscopy is a useful and rapid means of diagnosis. Alternative rapid and specific tests are urinary antigen detection and the use of a DNA probe. Culture must always be performed for optimal sensitivity and for epidemiologic purposes.

Legionella micdadei and Legionella dumoffii monoclonal antibodies for laboratory diagnosis of Legionella infections

Two different monoclonal antibodies directed against Legionella micdadei and L. dumoffii (Genetic Systems Corp., Seattle, Wash.) were evaluated for their specificity and ability to detect L. micdadei and L. dumoffii in human and animal clinical samples and bacterial isolates in an indirect immunofluorescence assay. All three frozen sputum samples and all three Formalin-fixed sputum and liver samples from patients with culture-documented L. micdadei pneumonia were positive when tested with the L. micdadei monoclonal antibody. A Formalin-preserved lung sample from a patient with culture-documented L. dumoffii pneumonia was positive with its homologous monoclonal antibody. No cross-staining reactions were found with either monoclonal antibody on any of 25 human sputum samples tested from patients without Legionella infections. A total of 66 Legionella strains and 56 non-Legionella strains including 22 Pseudomonas strains and 34 other bacterial strains were studied. No cross-staining reactions were found except in Staphylococcus aureus Cowan 1 ATCC 12598. The lower limit of detection in seeded sputum samples was about 7 X 10(4) cells per ml for both monoclonal antibodies. Lung and tracheal lavage specimens from L. micdadei- or L. dumoffii-infected guinea pigs showed specific staining only with their respective monoclonal antibodies. The monoclonal antibodies stained homologous bacteria slightly less intensely than did the polyclonal antisera, but the signal-to-noise ratio was considerably higher for the monoclonal antibodies. No differences in sensitivity of staining of clinical specimens or bacterial isolates were noted between the monoclonal antibodies and the polyclonal reagents for L. micdadei and L. dumoffii (Centers for Disease Control, Atlanta, Ga., and BioDx, Denville, N.J. These monoclonal antibodies ae sensitive and specific, making them good candidates for laboratory diagnostic purposes.

Influence of growth temperature on virulence of Legionella pneumophila

The effect of growth temperature on the virulence of a strain of broth-grown serogroup 1 Legionella pneumophila (Wadsworth F889) was examined by growing the bacterium at different temperatures and then infecting guinea pigs (by intratracheal injection) and guinea pig alveolar macrophages. The 50% lethal dose for guinea pigs infected with 25 degrees C-grown F889 was log10 5.0 CFU and that for 41 degrees C-grown F889 was log10 5.7 CFU, or a fivefold difference. Guinea pig alveolar macrophages were infected in quadruplicate with log10 3.8 CFU of F889 cells grown at either 25 or 41 degrees C. Counts of F889 in the alveolar macrophages infected with 25 degrees C-grown bacteria were 40% greater after 1 day of incubation (P = 2 X 10(-4)) than were counts in the alveolar macrophage suspensions inoculated with 41 degrees C-grown bacteria. However, the counts were not significantly different after 3 days of incubation. Examination of cover slip cultures of guinea pig alveolar macrophages infected with 25 degrees C-grown or 41 degrees C-grown bacteria showed that the bacteria grown at the lower temperature were twice as likely to be macrophage-associated after 1 h of incubation than were the bacteria grown at the higher temperature. Growth at the lower temperature was also associated with a change in reactivity with monoclonal antibodies, but not with a change in plasmid content. Thus, environmental temperature may play an important role in modulating the virulence of L. pneumophila, possibly by affecting bacterial adherence to host cells.

Retrospective study of Gen-Probe rapid diagnostic system for detection of legionellae in frozen clinical respiratory tract samples

The Gen-Probe Rapid Diagnostic System for legionellae, which uses 125I-labeled cDNA directed against the rRNAs of legionellae, was evaluated for its ability to detect members of the genus by using clinical specimens which had been frozen at -70 degrees C for 2 to 8 years. Culture and direct immunofluorescence (DFA) results obtained at the time of specimen collection were used to categorize samples. The specimens tested were 112 samples culture positive for legionellae and 230 samples negative on culture and DFA tests. They were tested in a blinded and randomized fashion. Results were expressed in terms of the ratio of counts per minute of the sample to the counts per minute of the provided negative control. A ratio of greater than or equal to 4.0 was picked for optimal specificity. Of the 112 previously positive specimens, 63 (57%) were positive by the probe assay, and of the 230 previously negative samples, 228 (99.1%) were negative. The 51 discrepant specimens were reexamined by culture and DFA testing if adequate amounts remained; this was possible for 34 specimens. On repeat culture, 22 of 33 previously culture-positive samples yielded legionellae and 11 were negative. Ten of the positive repeat cultures yielded two or fewer colonies per plate. One probe-positive but previously culture-negative sample was overgrown by contaminants on repeat culture. Reanalysis of data after exclusion of the 17 unavailable, 11 repeat culture-negative, and 1 unevaluable specimen gave a probe sensitivity of 74% and specificity of 100%. The Gen-Probe test is therefore specific and is of useful sensitivity.

Comparison of blood culture methods for recovery of Legionella pneumophila from the blood of guinea pigs with experimental infection

Blood was cultured from guinea pigs with experimental Legionella pneumophila serogroup 1 pneumonia, using four different methods. A 0.03-ml amount was spread onto each of several plates of buffered charcoal-yeast extract supplemented with alpha-ketoglutarate (BCYE) (direct plate); 1.5 ml each was inoculated into a BCYE agar-yeast extract broth bottle (biphasic), a pediatric Isolator tube (E. I. du Pont de Nemours & Co., Inc., Wilmington, Del.), and a glass tube containing 0.025% sodium polyanethanolsulfonate. Blood processed in the Isolator tube was plated on BCYE, as was the buffy coat blood fraction, which was obtained by centrifugation of the tube containing sodium polyanethanolsulfonate and blood. Observations were made of the number of positive cultures, the time to detection of positive cultures, and the absolute bacterial concentrations. Each system was equally sensitive in detecting bacteremia. The biphasic method required 5 days for cultures to become positive, whereas the other systems required 2 to 3 days to detect all positive cultures (P = 1.3 X 10(-5) by Friedman group statistic, and P less than 10(-5) for comparison of the biphasic and other methods). The direct plating method demonstrated the best quantitative recovery of L. pneumophila in comparison to the other methods tested (P = 2.0 X 10(-5) by analysis of variance group statistic and P less than 0.05 for comparison between each of the methods). Quantitative recovery by the Isolator method was intermediate between the direct plating and buffy coat methods. The biphasic and Isolator blood culture methods performed poorly in comparison to the other methods, indicating the need for caution in choosing blood culture methods for Legionella isolation.

Laboratory diagnosis of infections caused by legionellae

Legionnaires' disease can be diagnosed by detection of antibody to the organism or by direct detection of the bacterium and/or its products using immunological means, culture or a DNA probe. Culture isolation, still the most specific and sensitive test, can be done with sputum samples if proper selective techniques are used. Antibody detection is more suited for epidemiological purposes than individual cases. It has been overused to the exclusion of more specific and rapid methods, such as culture and other means of bacterial detection. Immunofluorescent microscopy remains an important tool in diagnosis, although urinary antigen detection and DNA probes are promising alternative tests. For optimal sensitivity, culture must be performed, regardless of which test is used.

Nosocomial outbreak of Legionnaires' disease: molecular epidemiology and disease control measures

Molecular laboratory techniques were used to study the epidemiology of an outbreak of nosocomial Legionnaires' disease. All patient isolates were Legionella pneumophila serogroup 1 and showed identical plasmid profiles and reactions with serogroup-specific monoclonal antibodies. L pneumophila was also cultured from four of five cooling tower water samples; however, the isolate from only one tower was serogroup 1 of the same subtype as patient isolates. Since the cases were temporally clustered and epidemiologically associated with exposure to cooling tower aerosols, the single cooling tower implicated by molecular analysis was the most likely source of the outbreak. Chlorination of cooling tower ponds has eradicated the epidemic strain. Since potable water also harbored the infecting organism and was the probable source for cooling tower contamination, decontamination of the hospital water system was also undertaken. Superchlorination of hot water holding tanks to 17 ppm on a weekly basis has effectively eradicated L pneumophila from the potable water system and appears to be a reasonable, simple, and relatively inexpensive alternative to previously described methods of control.

Control of Legionella in hospitals

Legionellas are present in water distribution systems and cooling towers of many hospitals. No firm data are available regarding the need for prophylactic disinfection of these contaminated systems. Disinfection of water systems and cooling towers is an accepted and effective means of ending nosocomial Legionnaires' disease, but it should be performed in conjunction with good epidemiological and microbiological studies to pinpoint environmental reservoirs and disseminators. Chlorination, pasteurization, or both, are the only means of disinfection found to be effective in disease outbreaks. Prospective surveillance of immuno compromised patients with pneumonia is probably the most effective means to determine if a hospital is a source of Legionnaires' disease and, therefore, requires further investigations and disinfection.

Paleoepidemiologic investigation of Legionnaires disease at Wadsworth Veterans Administration Hospital by using three typing methods for comparison of legionellae from clinical and environmental sources

Multilocus enzyme electrophoresis, monoclonal antibody typing for Legionella pneumophila serogroup 1, and plasmid analysis were used to type 89 L. pneumophila strains isolated from nosocomial cases of Legionnaires disease at the Veterans Administration Wadsworth Medical Center (VAWMC) and from the hospital environment. Twelve L. pneumophila clinical isolates, obtained from patients at non-VAWMC hospitals, were also typed by the same methods to determine typing specificity. Seventy-nine percent of 33 VAWMC L. pneumophila serogroup 1 clinical isolates and 70% of 23 environmental isolates were found in only one of the five monoclonal subgroups. Similar clustering was found for the other two typing methods, with excellent correlation between all methods. Enzyme electrophoretic typing divided the isolates into the greatest number of distinct groups, resulting in the identification of 10 different L. pneumophila types and 5 types not belonging to L. pneumophila, which probably constitute an undescribed Legionella species; 7 clinical and 34 environmental VAWMC isolates and 2 non-VAWMC clinical isolates were found to be members of the new species. Twelve different plasmid patterns were found; 95% of VAWMC clinical isolates contained plasmids. Major VAWMC epidemic-bacterial types were common in the hospital potable-water distribution system and cooling towers. Strains of L. pneumophila which persisted after disinfection of contaminated environmental sites were of a different type from the prechlorination strains. All three typing methods were useful in the epidemiologic analysis of the VAWMC outbreak.

Comparison of cross-staining reactions by Pseudomonas spp. and fluorescein-labeled polyclonal and monoclonal antibodies directed against Legionella pneumophila

Commercially prepared polyclonal antisera to Legionella pneumophila are known to cross-react with organisms of the genus Pseudomonas. To determine whether a commercially available monoclonal antibody reagent specific for L. pneumophila would also cross-react with pseudomonads, a two-laboratory study was undertaken to test both monoclonal and polyclonal reagents against 33 isolates of Pseudomonas spp., including 25 Pseudomonas aeruginosa, 4 P. putida, 2 P. maltophilia, 1 P. fluorescens, and 1 P. alcaligenes. Four antisera were tested; polyclonal anti-legionella antisera pools A and B (Centers for Disease Control [CDC], Atlanta, (Ga.), polyclonal 1-6 antisera (BioDx, Inc., Denville, N.J.), and a monoclonal antibody reagent produced by Genetic Systems Corp., Seattle, Wash. All reagents were labeled with fluorescein. Cross-staining reactions were found with the BioDx L. pneumophila antisera and 10 isolates of Pseudomonas. Four of these isolates demonstrated cross-staining with CDC pool A. When tested with individual serotype-specific reagents (CDC), three of four cross-reacted with L. pneumophila serotype 1 antisera; the fourth cross-reacted with serotype 3. No cross-staining reactions were noted with the monoclonal reagent and any of the pseudomonads tested, demonstrating that the Genetic Systems Corp. monoclonal reagent is the most specific of the four reagents tested.

Cross-reactive urinary antigens among patients infected with Legionella pneumophila serogroups 1 and 4 and the Leiden 1 strain

A solid-phase radioimmunoassay was developed to detect antigens of Legionella pneumophila serogroup 4. The assay detected antigen in urine from four of seven patients with pneumonia caused by L. pneumophila serogroup 4, two of two patients with L. pneumophila Leiden 1 strain (proposed new serogroup), and 26 of 55 patients with L. pneumophila serogroup 1 (proven by culture or direct fluorescent-antibody assay). The antigen of two patients with serogroup 4, one with Leiden 1 strain, and 49 with serogroup 1 infections could be detected in a previously described assay for antigens of L. pneumophila serogroup 1. Serogroup specificity of the assays could be demonstrated if serial dilutions of urine were tested. None of 347 urine specimens used as controls were positive in the radioimmunoassay. We conclude that antigens are excreted in the urine of at least some patients with Legionnaires' disease caused by serogroup 4 and Leiden 1 strain. The urinary antigens in patients with serogroup 4 and Leiden 1 strain infections, although different from those in patients with serogroup 1 infections, serologically contain cross-reactive components.

Clinical utility of a monoclonal direct fluorescent reagent specific for Legionella pneumophila: comparative study with other reagents

Twenty-four lower respiratory tract samples taken from patients with culture-confirmed Legionella pneumophila infection were examined with three different direct immunofluorescent antisera to L. pneumophila, as were 29 samples from similar sources taken from patients without Legionnaires disease. The reagents studied were Genetic Systems Corp. (GS) monoclonal L. pneumophila conjugate, which reacts with all known serogroups of L. pneumophila, BioDx polyvalent L. pneumophila serogroups 1 through 6 conjugate, and Centers for Disease Control polyvalent pool A L. pneumophila serogroups 1 through 4 conjugate. The specimens had been frozen at -70 degrees C for 0.5 to 5 years. Randomization was used in coding the samples, which were stained and read by an independent observer. All three conjugates correctly identified all positive and negative samples. No difference was noted among the conjugates in the absolute numbers of fluorescent L. pneumophila bacteria per sample. The GS conjugate had a much cleaner background than did the other two reagents. Mean staining intensity scores were 3.4, 3.9, and 3.7 for the GS, BioDx, and Centers for Disease Control conjugates, respectively. This study demonstrates that the diagnostic efficiency of all three conjugates is equivalent. Since the GS conjugate is easier to read, does not cross-react with non-L. pneumophila bacteria, and reacts with serogroups 1 through 10 of L. pneumophila, it appears to be preferable for use in diagnostic testing on nonhistopathologically processed specimens.

A new biotype of Legionella dumoffii

A new biotype of L. dumoffii was isolated from lung and sputum samples of an immunosuppressed patient with pneumonia. This strain differs from other described strains of L. dumoffii in that it fails to produce browning of tyrosine-containing buffered yeast extract medium.

Legionella anisa: a new species of Legionella isolated from potable waters and a cooling tower

Between March 1980 and June 1981, five strains of Legionella-like organisms were isolated from water. Four were recovered from potable water collected from hospitals in Chicago, Ill., and Los Angeles, Calif., during outbreaks of nosocomial legionellosis. The fifth strain was isolated from water collected from an industrial cooling tower in Jamestown, N.Y. The strains exhibited biochemical reactions typical of Legionella species and were gram-negative motile rods which grew on buffered charcoal-yeast extract agar but not on blood agar, required cysteine, and were catalase positive, urease negative, nitrate negative, hippurate negative, and nonfermentative. All strains were positive for oxidase and beta-lactamase and produced a brown, diffusible pigment. Of the five strains, four exhibited blue-white autofluorescence under long-wavelength UV light. The fatty-acid composition and ubiquinone content of these strains were consistent with those of other Legionella species. Direct fluorescent-antibody examination of the five strains with conjugates to previously described Legionella species demonstrated no cross-reactions except with the conjugates to L. longbeachae serogroup 2 and L. bozemanii serogroup 2. Four strains gave a 4+ reaction to the L. longbeachae serogroup 2 conjugate and the fifth strain gave a 1+ reaction. Each of the five strains gave a 4+ reaction with the conjugate to L. bozemanii serogroup 2. DNAs from the five strains were highly related (84 to 99%) and showed 5 to 57% relatedness to other Legionella species. These strains constitute a new species in the genus Legionella, and the name Legionella anisa sp. nov. is proposed. The type strain of L. anisa is WA-316-C3 (ATCC 35292).

Antimicrobial therapy of experimentally induced Legionnaires' disease in guinea pigs

A guinea pig model of Legionnaires' disease was produced by intratracheal inoculation of Legionella pneumophila. The bacterial inoculum given reproducibly resulted in a 95% fatality rate in untreated animals. These animals had lung histopathologic and bacteriologic findings almost identical to those found in human Legionnaires' disease. Administration of cefoxitin and gentamicin, which inhibited the infecting strain in low antibiotic concentration in vitro, was no more effective than saline; this occurred despite adequate serum and lung levels of the drugs. Erythromycin, rifampin, doxycycline, and cotrimoxazole significantly reduced fatality rates. The combinations of erythromycin with rifampin and doxycycline with rifampin were not significantly better or worse than use of either drug alone. In animals treated with any regimen containing rifampin, a dramatically higher rate of bacterial killing was observed than that observed in those animals treated with erythromycin or doxycycline alone. Also, animals treated with any regimen containing rifampin had significantly less late lung histologic evidence of pneumonia than did those treated with other agents. No rifampin-induced resistance was found. This animal model reflects prior clinical findings, provides experimental grounds for therapy with drugs other than erythromycin, and suggests that therapy of Legionnaires' disease should probably include rifampin in most cases.

Legionella pneumophila serogroup 9: a cause of human pneumonia

A new serogroup of Legionella pneumophila was isolated from bronchoscopic washings and an open-lung biopsy specimen of a patient from California with pneumonia. A serologically identical isolate was obtained from tap water of a hospital ward in the Netherlands, and a fatal case of pneumonia in a patient from Virginia was shown retrospectively to have been caused by this new organism. The type strain of what is now serogroup 9 of L. pneumophila is IN-23-G1-C2 (American Type Culture Collection no. 35289). Disease caused by L. pneumophila serogroup 9 is not apparently different from that caused by other L. pneumophila serogroups.

Oropharyngeal colonization with Legionella pneumophila

A total of 186 volunteers, including 40 hospital patients, participated in a cross-sectional survey of oropharyngeal colonization with Legionella pneumophila. Colonization was defined as the appearance of any L. pneumophila organisms on culture or a positive direct fluorescent-antibody (FA) test or both in the absence of signs or symptoms of pneumonia. The direct FA tests were performed on throat swabs, using a polyvalent conjugate directed against L. pneumophila serogroups I through IV. Throat swabs were cultured for L. pneumophila on a selective medium. Blood specimens were tested for antibody, using an indirect FA test and heat-killed polyvalent antigen for L. pneumophila serogroups I through IV. Eight people, none of whom had pneumonia or fever, had positive direct FA tests; no subject had a positive culture for L. pneumophila. Whether the positive direct FA results represent colonization cannot be stated with assurance. In any case, the results suggest that colonization occurs infrequently.

Culture diagnosis of Legionella infections

Culture diagnosis of Legionella infections is feasible in most instances. Using selective culture techniques, sputum can be used to diagnose these infections specifically, and with a degree of sensitivity equal to or greater than currently used immunological means of diagnosis. Every laboratory performing diagnostic testing for this group of infections should perform cultures to achieve optimal yield.

Legionella wadsworthii species nova: a cause of human pneumonia

A new species of Legionella that caused pneumonia in a patient with chronic lymphocytic leukemia has been characterized. The Legionella wadsworthii species nova is proposed for this bacterium. The type strain is Wadsworth 81-716A (American Type Culture Collection 33877). The new species differs phenotypically from L. pneumophila in that the predominant cellular fatty acid is methyl-12-methyltetradecanoic acid (a-15:0) rather than methyl-14-methylpentadecanoic acid (i-16:0), and in its failure to react with L. pneumophila antiserum. The clinical illness caused by L. wadsworthii was not apparently different from that seen with other legionella infections, except for prolonged fever and slow resolution of pulmonary infiltrates.

Enhancement of recovery of Legionella pneumophila from contaminated respiratory tract specimens by heat

Heating of Legionella pneumophila and other Legionella spp. was studied to determine whether this technique could be used as a selective technique with contaminated clinical specimens. Studies of 13 different strains of Legionella spp. showed heterogeneous heat survival; heating at 60 degrees C for 1 to 2 min did not affect the survival of the majority of strains. Heating of four Pseudomonas aeruginosa strains at 60 degrees C for 2 min reduced bacterial counts by 98% or greater. Enterococci were heat tolerant, with virtually no inhibition under the same conditions. No inoculum effect was noted for any of the organisms tested. Heating of eight contaminated clinical specimens before plating on buffered charcoal-yeast extract medium reduced the numbers of contaminants on most plates but increased by only one the number of specimens yielding L. pneumophila. Plating the same specimens on selective media with or without heat pretreatment yielded L. pneumophila in every case. Heating of clinical specimens at 60 degrees C for 1 to 2 min before plating may occasionally increase the recovery of L. pneumophila from contaminated specimens, but this technique should not be generally used.

Efficacy of ozone in eradication of Legionella pneumophila from hospital plumbing fixtures

The effect of ozonation of supply water for one wing of an unoccupied hospital building which had positive cultures for Legionella pneumophila from multiple potable water fixtures was studied in a prospective, controlled fashion. Mean ozone residual concentrations of 0.79 mg/liter eradicated L. pneumophila from the fixtures, but so did nonozonated water in the control wing fixtures. The efficacy of the nonozonated water was most probably due to a mechanical flushing effect and to an unexpected rise in the chlorine content of the supply water. Determination of the in vitro activity of ozone against L. pneumophila did not predict the efficacy of its eradication from water fixtures treated with ozone.

Comparative study of selective media for isolation of Legionella pneumophila from potable water

A total of 100 water samples, 95% of which were taken from hospital potable water fixtures, were cultured on three different media used for the isolation of Legionella pneumophila. The media used were buffered charcoal-yeast extract medium (BCYE alpha medium), BCYE alpha medium with antimicrobial agents (BMPA alpha medium), and BCYE alpha medium with antimicrobial agents, glycine, and differential dyes (MWY medium). An acid wash procedure was also used for specimens plated on BCYE alpha and BMPA alpha media. A total of 24 samples were culture positive for L. pneumophila by one or more techniques. MWY medium detected 92% of positive cultures, BCYE alpha medium with the acid wash detected 83% of positive cultures, BMPA alpha medium detected 79% of positive cultures. BCYE alpha medium detected 71% of positive cultures, and BMPA alpha medium with the acid wash detected 62% of positive cultures. MWY medium was the best medium for isolating L. pneumophila from potable water specimens and can probably be depended upon as the sole medium for this type of testing.

Susceptibility of Legionella spp. to mycinamicin I and II and other macrolide antibiotics: effects of media composition and origin of organisms

Thirty-three strains of Legionella spp., 29 of which were L. pneumophila, were tested for their susceptibilities to erythromycin (EM), rosaramicin, tylosin, mycinamicin I (Sch-27897), and mycinamicin II (Sch-27896). Testing was performed using an agar dilution method with two different types of media: buffered charcoal yeast extract medium supplemented with 0.1% alpha-ketoglutarate (BCYE alpha) and filter-sterilized yeast extract medium with 0.1% alpha-ketoglutarate (BYE alpha). The minimal inhibitory concentrations (MICs) of the drugs tested relative to the MICs of erythromycin were: rosaramicin, MIC approximately equal to 0.2 EM MIC; tylosin, MIC approximately equal to 2 EM MIC; mycinamicin I, MIC approximately equal to 0.5 EM MIC; and mycinamicin II, MIC approximately equal to EM MIC. Both types of media caused equivalent partial inactivation of the macrolides which was apparently due entirely to pH effect. MICs on BCYE alpha were one to five times more than those observed on BYE alpha; this may be due to poorer growth on BYE alpha.

Isolation of Legionella pneumophila from hospital potable water specimens: comparison of direct plating with guinea pig inoculation

Eighty-five potable water samples were analyzed for the presence of Legionella pneumophila by direct plating and by guinea pig inoculation methods. Direct plating methods utilized antimicrobial-containing media with and without acid treatment of specimens before inoculation. Thirty-three specimens were culture positive for L. pneumophila by one or both techniques. A total of 14 specimens were positive for L. pneumophila by direct plating only, and 2 specimens were positive for L. pneumophila by guinea pig inoculation only. The sensitivity of direct plating is significantly greater than that of guinea pig inoculation when isolating L. pneumophila from potable water (P less than 0.01).

Rapid diagnosis of Legionnaires' disease by urinary antigen detection. Comparison of ELISA and radioimmunoassay

An enzyme-linked immunosorbent assay was developed to detect urinary antigen excreted by patients with Legionnaires' disease. Of 47 patients tested, antigen was detected in 39. Of these 39 specimens, 35 gave clearly positive results by visual analysis; four others required spectrophotometric verification of positive results. Antigen was not detected in any of 178 urine specimens from patients with other pulmonary, bacteremic or urinary tract infections after a quick and simple confirmatory test. The assay required more time to perform than a previously described radioimmunoassay but was of equivalent sensitivity and specificity and did not require expensive equipment or contract with radioactive reagents. We conclude that enzyme-linked immunosorbent assay is a rapid, sensitive and specific means for rapidly diagnosing legionnaires' disease that can be performed in clinical laboratories unwilling or unable to use radioisotopes.

Rapid diagnosis of Legionnaires' disease

An enzyme linked immunosorbent assay was developed to detect urinary antigen excreted by patients with Legionnaires' disease. Of 47 patients tested, antigen was detected in 39. Antigen was not detected in any of 178 urine specimens from patients with other pulmonary, bacteremic, or urinary tract infections after performance of a quick and simple confirmatory test. The assay required more time to perform than a previously described radioimmunoassay but was of equivalent sensitivity and specificity and did not require expensive equipment of contact with radioactive reagents. We conclude that enzyme linked immunosorbent assay is a rapid, sensitive, and specific means for rapidly diagnosing Legionnaires' disease which can be performed in clinical laboratories unwilling or unable to use radioisotopes.

Comparison of liquid growth media for Legionella pneumophila

Ten liquid media were compared under standard conditions for their ability to support the growth of Legionella pneumophila. Modified gonococcal-ferric cysteine broth (without sodium chloride) supplemented with 1% yeast extract yielded the best overall growth of the one strain of L. pneumophila examined. Growth rates were independent of pH changes which occurred during incubation. The growth rates of 10 different strains of L.pneumophila were compared in this medium. There appeared to be little difference in the growth rates of strains passaged frequently or infrequently, or between environmental and clinical isolates. Moderate aeration resulted in a faster growth rate and in approximately a 1 log10 higher final cell concentration as compared to a static broth culture. These experiments demonstrate that there are moderate to marked differences among the various media described in the literature and that no liquid medium yet developed supports rapid growth of L. pneumophila incubated without shaking.