Value of terminal subcultures for blood cultures monitored by BACTEC 9240

Diagnosis of bacteremia in whole-blood samples by use of a commercial universal 16S rRNA gene-based PCR and sequence analysis

In a prospective, multicenter study of 342 blood samples from 187 patients with systemic inflammatory response syndrome, sepsis, or neutropenic fever, a new commercial PCR test (SepsiTest; Molzym) was evaluated for rapid diagnosis of bacteremia. The test comprises a universal PCR from the 16S rRNA gene, with subsequent identification of bacteria from positive samples by sequence analysis of amplicons. Compared to blood culture (BC), the diagnostic sensitivity and specificity of the PCR were 87.0 and 85.8%, respectively. Considering the 34 BC-positive patients, 28 were also PCR positive in at least one of the samples, resulting in a patient-related sensitivity of 82.4%. The concordance of PCR and BC for both positive and negative samples was (47 + 247)/342, i.e., 86.0%. In total, 31 patients were PCR/sequencing positive and BC negative, in whom the PCR result was judged as possible or probable to true bacteremia in 25. In conclusion, the PCR approach facilitates the detection of bacteremia in blood samples within a few hours. Despite the indispensability of BC diagnostics, the rapid detection of bacteria by SepsiTest appears to be a valuable tool, allowing earlier pathogen-adapted antimicrobial therapy in critically ill patients.

Circulating biomarkers as surrogates for bloodstream infections

The use of biomarkers provides a novel approach to diagnosing infection, its severity and treatment response. Biomarkers, especially procalcitonin and, to a lesser extent, C-reactive protein and interleukin 8, can improve the diagnostic and prognostic assessment of bloodstream infections. Both strengths and weaknesses of biomarkers must be recognized for rational and safe use in clinical settings. Cut-off ranges must be chosen within the specific clinical context. Ultrasensitive assays for procalcitonin, capable of measuring low levels in healthy individuals, may help to identify even 'subclinical' inflammatory states before the development of clinically evident sepsis. For immunocompromised patients, the use of biomarkers could lead to an earlier and more targeted antimicrobial therapy for patients at risk of sepsis, whereas those patients with viral illness or a non-infectious aetiology of inflammation who maintain low levels of procalcitonin over time can be withheld from antibiotic exposure. The time has arrived to move beyond the observational reporting of 'promising' biomarkers. Specific cut-off ranges must be proposed and intervention studies conducted to tackle the existing vicious cycle of diagnostic uncertainty, antibiotic overuse and emerging multi-resistance.

Identification of human pathogens isolated from blood using microarray hybridisation and signal pattern recognition

Background

Pathogen identification in clinical routine is based on the cultivation of microbes with subsequent morphological and physiological characterisation lasting at least 24 hours. However, early and accurate identification is a crucial requisite for fast and optimally targeted antimicrobial treatment. Molecular biology based techniques allow fast identification, however discrimination of very closely related species remains still difficult.

Results

A molecular approach is presented for the rapid identification of pathogens combining PCR amplification with microarray detection. The DNA chip comprises oligonucleotide capture probes for 25 different pathogens including Gram positive cocci, the most frequently encountered genera of Enterobacteriaceae, non-fermenter and clinical relevant Candida species. The observed detection limits varied from 10 cells (e.g. E. coli) to 10⁵ cells (S. aureus) per mL artificially spiked blood. Thus the current low sensitivity for some species still represents a barrier for clinical application. Successful discrimination of closely related species was achieved by a signal pattern recognition approach based on the k-nearest-neighbour method. A prototype software providing this statistical evaluation was developed, allowing correct identification in 100 % of the cases at the genus and in 96.7 % at the species level (n = 241).

Conclusion

The newly developed molecular assay can be carried out within 6 hours in a research laboratory from pathogen isolation to species identification. From our results we conclude that DNA microarrays can be a useful tool for rapid identification of closely related pathogens particularly when the protocols are adapted to the special clinical scenarios.

Detection of fifteen species of Candida in an automated blood culture system

Candidemia was simulated with 15 Candida spp. by using an automated blood culture system. Candida growth was detected in 479/648 (74%) bottles: 211/216 (98%) aerobic bottles, 58/216 (27%) anaerobic bottles, and 210/216 (97%) mycology bottles. Only the growth of Candida lipolytica failed to be detected in all media.

Usefulness of a Molecular Strategy for the Detection of Bacterial DNA in Patients with Severe Sepsis Undergoing Continuous Renal Replacement Therapy

INTRODUCTION

Sepsis is a major cause of morbidity and mortality in critically ill patients. Sepsis is associated with cell necrosis and apoptosis. Circulating plasma levels of DNA have been found in conditions associated with cell death, including sepsis, pregnancy, stroke, myocardial infarction and trauma. Plasma DNA can also derive from bacteria. We have recently implemented a method to detect bacterial DNA and, in the present study, we validated this technique comparing it to standard blood culture in terms of diagnostic efficacy.

METHODS

We examined a cohort of 9 critically ill patients with a diagnosis of severe sepsis and acute renal failure requiring continuous renal replacement therapy (CRRT). We analyzed bacterial DNA in blood, hemofilters, and ultrafiltrate (UF) by polymerase chain reaction amplification of 16S rRNA gene sequence analysis. Standard blood cultures were performed for all patients.

RESULTS

The blood cultures from 2 of the 9 (22%) patients were positive. However, bacterial DNA was identified in the blood of 6 patients (67%), including the 2 septic patients with positive blood cultures. In 9 (100%) patients bacterial DNA was found on the filter blood side, whereas in 7 (78%) subjects it was found in the dialysate compartment of the hemofilters. Bacterial DNA was never detected in the UF.

CONCLUSIONS

Using the 16S rRNA gene, the detection of bacterial DNA in blood and adsorbed within the filter could be a useful screening tool in clinically septic, blood culture-negative patients undergoing CRRT. However, the identification of the etiologic agent is not feasible with this technique because specific primers for the defined bacteria must be used to further identify the suspected pathogenic organisms.

Detection of candidaemia in high risk patients: can yield of blood cultures be improved by blind subculture?

Rapid detection of candidaemia is crucial for timely antifungal chemotherapy. However, the sensitivity of automated blood culture (BC) systems has been questioned. Blind subculture might increase detection rate and possibly also reduce time to detection of candidaemia. This retrospective study aimed to evaluate the efficacy of blind subcultures in patients deemed at high risk of candidaemia. BCs were processed by the BacT/Alert BC system, and during a 5-y period (1998-2003) subculture on the third d of incubation was performed for patients selected by clinical and microbiological assessment. A total of 79,165 BCs were drawn during the study period. 2154 BCs from 285 patients were selected for subculture. 103 (4.8%) BCs from 52 patients were yeast positive; 71 were detected positive prior to the planned subculture, 25 were positive on subculture, and 7 were negative on subculture, but became positive during further incubation. The 25 BCs positive on subculture originated from 14 patients, 11 of whom had already been diagnosed with candidaemia during the previous 14 d. Thus, a primary diagnosis of candidaemia was obtained by subculture in only 3 (1.1%) of the 285 patients selected. In conclusion, in our clinical setting blind subculture did not materially increase the detection of candidaemia, but helped to document persistent infection in a subset of cases.

Effect of inoculum size on detection of Candida growth by the BACTEC 9240 automated blood culture system using aerobic and anaerobic media

Simulated candidemia was produced with 20 Candida isolates at three inoculum sizes (100, 10, and 1 CFU/ml of blood). Growth detection was better with larger inocula. The time to growth detection was shorter with larger inocula. Inoculum size does effect Candida growth detection and time to detection in BACTEC 9240 automated systems.

Bactec 9240 blood culture system: to preincubate at 35 degrees C or not?

Bactec Plus blood culture bottles were preincubated at 35 degrees C or at room temperature before entry into the Bactec 9240 instrument to determine the influence of preincubation temperature and time. Of 463 positive blood culture sets, 956 bottles were positive, of which the instrument detected 92.1%. Of 76 positive bottles undetected by the instrument, 68 were preincubated at 35 degrees C and eight at room temperature. The median entry delay and instrument detection times were 17.9 and 7.2 h for preincubated bottles, and 16.4 and 13.4 h for bottles held at room temperature. Short entry delay and inspection before entry into the instrument are necessary if preincubation at 35 degrees C is used.

Direct comparison of the BACTEC 9240 and BacT/ALERT 3D automated blood culture systems for candida growth detection

A direct comparison of two automated blood culture systems was conducted to compare their ability to detect Candida growth. The systems evaluated were the BACTEC 9240 (Bactec) and BacT/ALERT 3D (BacT). The aerobic, anaerobic, and mycology media for each system were evaluated: Bactec Plus Aerobic/F, Plus Anaerobic/F, and Myco/F Lytic bottles, respectively, and BacT FA, SN, and MB bottles, respectively. Each blood culture bottle was inoculated with fresh blood from healthy donors. Fifty isolates of Candida spp. were used. The six different blood culture bottles were each inoculated with 1000 yeasts per bottle and then incubated in the corresponding automated system. The BacT detected growth of 90% (135 of 150) of Candida pathogens, while Bactec detected 66% (100 of 150). Growth was detected in all BacT and Bactec mycology bottles, all BacT aerobic bottles, and by terminal subculture of all bottles. Sixty-five of 300 (22%) bottles had no growth detected; 50 from the Bactec (5 aerobic and 45 anaerobic) and 15 from the BacT (all anaerobic). Terminal subculture of "negative" bottles demonstrated viable yeast growth from all 65 bottles, representing 65 false-negatives. The mean time to growth detection in the BacT system was 25.62 h while the Bactec was 27.30 h (P < 0.01). Both automated blood culture systems detected all episodes of simulated candidemia when specialized mycology media were used. However, when only standard aerobic and anaerobic media were used, the BacT performed better than the Bactec in overall growth detection, time to growth detection, and number of false-negatives.

Detection of simulated candidemia by the BACTEC 9240 system with plus aerobic/F and anaerobic/F blood culture bottles

We studied the ability of the BACTEC 9240 automated blood culture system to detect simulated candidemia, including both Candida albicans and non-albicans Candida species. Simulated blood cultures were produced using 50 Candida isolates and BACTEC Plus Aerobic/F and Anaerobic/F blood culture bottles. Ten milliliters of blood and a suspension of each isolate containing 1,000 CFU were introduced into each bottle and then incubated at 35 degrees C in the BACTEC 9240 system. The system detected growth in 56 of 100 bottles. Four isolates did not have growth detected in either bottle after 21 days of incubation, resulting in four missed episodes of candidemia. If the blood culture bottles had been incubated for 5 days, an additional episode of candidemia would have remained undetected. If the bottles had been incubated for only 3 days, another episode would have been missed, resulting in up to six missed episodes of candidemia (four Candida glabrata isolates, one C. albicans isolate, and one Candida rugosa isolate). Terminal subculture of bottles without detected growth recovered yeast in 93% (41 of 44) of the bottles, representing 41 false negatives. In bottles where growth was detected, the time to detection was approximately 24 h. However, the mean time to growth detection for C. glabrata isolates in anaerobic medium was 22.14 +/- 2.47 h, but it was 120.89 +/- 35.33 h in aerobic medium (P < 0.001). The BACTEC 9240 system detected growth of most Candida isolates; however, the delayed time to detection of C. glabrata is clinically significant. Given the high rate of false negatives, terminal subcultures may be helpful in certain situations.

Evaluation of the VITAL (bioMérieux) automated blood culture system using blind subculture

OBJECTIVE

This study was performed to determine the ability of the VITAL system to detect and allow recovery of microorganisms that are difficult to grow, such as Brucella spp., yeasts, or anaerobes, as well as to determine the need for blind subcultures after the incubation period.

METHODS

A prospective evaluation of the system was performed, and 8247 blood culture bottles were processed. The standard was blind subculture from all the bottles after 5 days of incubation.

RESULTS

There were 3.2% false-positive and 0.6% false-negative results (72% of clinical importance). The system sensitivity for yeasts was 41%. The mean time for detection of Neisseria meningitidis was 31.9 +/- 2.8 h, for Brucella spp. 119.7 +/- 2 h, and for yeast 51.5 +/- 27.8 h.

CONCLUSIONS

The VITAL system poses has serious difficulties in the detection of N. meningitidis, Brucella spp., yeast and methicillin- and aminoglycoside-resistant Staphylococcus aureus (MARSA). The low system sensitivity for yeast detection makes the blind subculture necessary after the incubation period.

Direct identification of bacteria from positive blood cultures by amplification and sequencing of the 16S rRNA gene: evaluation of BACTEC 9240 instrument true-positive and false-positive results

In a previous study which evaluated the BACTEC 9240 automated blood culture system (Becton Dickinson Diagnostic Instrument Systems, Sparks, Md.), we noted a 1.3% "instrument false-positive" rate. That is, the BACTEC system signaled that a bottle (BACTEC Plus Aerobic/F bottle or BACTEC Anaerobic Lytic/10 bottle) culture was positive but a Gram stain was negative and there was no growth of bacteria or yeasts on subculture to chocolate agar. Furthermore, from the same sample of blood, cultures for fungi using the Isolator blood culture system (Wampole Laboratories, Cranbury, N.J.) were negative for growth. For the present study, we evaluated 76 instrument false-positive samples for the presence of 16S ribosomal DNA using the MicroSeq 500 kit (PE Biosystems, Foster City, Calif.). These samples also were negative for fungi by the Isolator method. This kit has a PCR module and sequencing module for the amplification and sequencing of the 16S RNA gene and provides a database for sequence alignment and identification of bacteria. To optimize the assay, we evaluated the effect of adding 0.5% bovine serum albumin to the sample from blood culture bottles and found that it decreased the effects of inhibitors on the PCR. Two control groups of blood culture specimens were also evaluated. One group (n = 45) were "instrument true positives"; the instrument signaled positive, and subsequent Gram stains were positive and subcultures on chocolate agar grew bacteria. The other group (n = 20) were "instrument true negatives"; the instrument signaled negative, the Gram stain was negative, and subcultures on chocolate agar and from the Isolator tube on fungal media showed no growth. None of the 76 instrument false-positive samples had evidence for 16S rRNA gene sequences. All of the instrument true-positive samples and all of the instrument true-negative specimens were positive and negative, respectively, using the MicroSeq 500 kit. Total peripheral white blood cell counts were statistically significantly higher for patients who had instrument false-positive results than for patients who had instrument true-positive or true-negative results (P = 0.001). We conclude that instrument false positives signaled by the BACTEC 9240 system are not due to bacteria in the blood culture samples.

Detection of Candida spp. in blood cultures using nucleic acid sequence-based amplification (NASBA)

Candida spp. are the main causes of fungal infections in immunocompromised patients. It is known, that the routinely used automated blood culture systems may fail to detect yeasts. We therefore investigated, whether Nucleic Acid Sequence-Based Amplification (NASBA) can be used to improve the detection rate of Candida spp. in blood cultures. Culture-positive as well as negative blood cultures from patients with a proven candidaemia were analyzed, and the results of BacT/Alert monitoring were compared with the results of NASBA-based detection of yeast RNA. With the NASBA-assay, the number of positive blood cultures increased from 21% to 34%. The NASBA-assay may confirm the diagnosis and demonstrate the need for prolonged treatment. In addition it may shorten the time to detection. In summary, using NASBA for the detection of yeast RNA in blood cultures, we have shown for the first time that it is possible to improve the detection rate of yeasts in blood cultures by using amplification technology.

Failure of an automated blood culture system to detect nonfermentative gram-negative bacteria

During a 1-year study we observed that both aerobic and anaerobic blood culture bottles from patients were negative by the BacT/Alert system during a 7-day incubation period. However, upon subcultivation of negative bottles, growth of Pseudomonas aeruginosa was detectable. In an attempt to explain this observation, aerobic BacT/Alert Fan bottles were seeded with a defined inoculum (0.5 McFarland standard; 1 ml) of Escherichia coli, Klebsiella pneumoniae, Serratia marcescens, P. aeruginosa, Stenotrophomonas maltophilia, or Acinetobacter baumannii. Half of the inoculated bottles were loaded into the BacT/Alert system immediately, and the remainder were preincubated for 4, 8, 16, and 24 h at 36 degrees C. With preincubation all bottles seeded with the Enterobacteriaceae signaled positive during the next 1.5 h. Organisms in bottles seeded with the nonfermentative species P. aeruginosa and A. baumannii remained undetected by the BacT/Alert system for 7 days. S. maltophilia was detected if the preincubation time was equal or less than 8 h. Without preincubation all bottles seeded with the Enterobacteriaceae or nonfermentative species signaled positive. Since nonfermentative species seem to enter a state of bacteriostasis within the preincubation period, we reasoned that an unknown factor is consumed. Accordingly, a smaller inoculum should allow the detection of nonfermentative species, even after preincubation, and serial dilutions of P. aeruginosa were detected in preincubated bottles. In this case preincubated bottles signaled positive faster than bottles without preincubation. We conclude that all bottles from clinical settings should be subcultured prior to loading to avoid false negatives. An alternative may be preincubation at room temperature.

Assessment of the FAN anaerobic bottle for culture of continuous ambulatory peritoneal dialysis fluid using the BacT/Alert system

The purpose of this study was to determine if the newly available FAN anaerobic bottle (FANAN) alone would be comparable to the combination of the FAN aerobic (FANAE) plus the standard BacT/Alert anaerobic (REGAN) bottles for culture of continuous ambulatory peritoneal dialysis (CAPD) fluid from patients with CAPD peritonitis. CAPD fluid (10 mL) was injected into each bottle, which was then monitored by the BacT/Alert instrument by using a 7-day protocol. Aerobic and anaerobic terminal subculture were performed on all bottles before they were classified as being culture negative. There were 181 effluents received that were suitable for analysis. Growth was detected in 76 (42%) effluents by at least one method. FANAE was the single best medium detecting 84/96 (88%) of all organisms whereas the FANAN and REGAN each detected 69/96 (72%). The combination of FANAE and REGAN bottles detected 92/96 (96%) isolates, which was significantly better than the FANAN or FANAE alone for isolate recovery (p < 0.001). The isolates that were missed by the FANAN but that were recovered by either FANAE or REGAN were all facultative anaerobes commonly detected in CAPD fluids. Terminal subculture revealed otherwise undetected pathogens in 3.9% of positive effluents, usually Pseudomonas aeruginosa. Based on our data, FANAE was the single best bottle for detection of CAPD peritonitis and, in combination with an anaerobic bottle, detected growth from the most effluents. FANAN alone could not substitute for the FANAE/REGAN combination. Although terminal subculture remains controversial, we recommend routine aerobic subculture to ensure that no P. aeruginosa isolates are missed.

Controlled clinical laboratory comparison of two supplemented aerobic and anaerobic media used in automated blood culture systems to detect bloodstream infections

A 20-ml blood sample was collected from adult patients with suspected bloodstream infections and distributed equally into the four volume-controlled bottles of a blood culture set consisting of aerobic and anaerobic BACTEC Plus/F bottles and aerobic and anaerobic BacT/Alert FAN bottles. All bottles were incubated in their respective instruments for a standard 5-day protocol or until the instruments signalled positivity. Samples in all bottles with negative results by these instruments were terminally subcultured. A total of 8,390 blood culture sets were obtained during the study period, of which 4,402 (52.5%) met the study criteria. Of these, 946 (21.5%) were positive either by instrument signal or by additional terminal subculture of all negative bottles and yielded growth of microorganisms. Five hundred eighty-nine (13.4%) blood culture sets were considered to have recovered 663 clinically significant organisms. When both the BACTEC and the BacT/Alert systems were used, 465 positive sets were detected; BACTEC alone detected 52 positive sets and BacT/Alert alone detected 72 (P = 0.09). No differences were found between the two systems in microbial recovery rate from blood cultures obtained from patients on antibiotic therapy. Significantly more members of the family Enterobacteriaceae (P < 0.01) were detected from patients without antimicrobial therapy by BacT/Alert than by BACTEC. The false-negative rates were 0.20% for BACTEC and 0.32% for BacT/Alert. A significantly higher false-positive rate was found for BACTEC (P < 0.0001). Both systems were comparable for the time to detection of microorganisms. However, gram-positive bacteria were detected faster by BACTEC and Enterobacteriaceae were detected faster on average by BacT/Alert. We concluded that both systems are comparable in their abilities to recover aerobic and anaerobic organisms from blood cultures and a terminal subculture might not be necessary for either of the two systems. The increased positivity rate when using an anaerobic bottle in a two-bottle blood culture set is due to the additional blood volume rather than to the use of an anaerobic medium.

Detection of fungemia from blood cultures using the BACTEC 9240 instrument

From January 1995 to November 1996, 14,623 resin-containing blood culture vials were tested with the BACTEC fluorescent series instrument. A total of 1560 microorganisms were recovered. 48 of the microorganisms were fungi. We could demonstrate the ability of the BACTEC 9240 to detect cases of fungemia with Candida species and moulds such as Aspergillus fumigatus (4 cases) and Fusarium solani (1 case).

Multicenter clinical comparison of resin-containing bottles with standard aerobic and anaerobic bottles for culture of microorganisms from blood

In a study comparing the Bactec 9240 (Plus Aerobic/F and Anaerobic/F bottles, containing resins; Becton Dickinson, USA) and the Vital (standard aerobic and anaerobic bottles, with no additives; bioMérieux, France) blood culture systems, 6456 sets of four bottles of 9660 blood cultures submitted were evaluated. There were 531 clinically significant isolates from 795 positive blood cultures. Of the 531 positive blood cultures, 355 were positive in both systems, 141 with the Bactec 9240 alone, and 30 with the Vital alone (p < 0.001); five were not detected by either system. The average time to detection of positive cultures for the matched sets was 10.65 h and 18.41 h by the Bactec 9240 system and the Vital system, respectively. The false-positive rate per bottle was 0.65% in the Bactec 9240 and 0.71% in the Vital. The rate of false-negative pairs (i.e., major errors) was very low (0.12% for the Bactec 9240, 0.19% for the Vital) and not significantly different between the two systems. The striking differences in recovery of microorganisms may be due to the presence of resins in the Bactec medium. However, the observed superiority of the Bactec 9240, even for patients not receiving antibiotics, suggests that resins adsorb other inhibitors present in patients' blood.

Rapid detection of Brucella melitensis from blood cultures by a commercial system

To assess the capability of the Peds Plus medium of the Bactec 9240 blood culture system to recover Brucella melitensis within the routine seven-day protocol used by most clinical microbiology laboratories, inoculated blood culture bottles were monitored by the Bactec 9240 instrument for four weeks, and blind subcultures were performed once a week. A total of 2579 blood cultures were drawn, 42 (1.6%) of which were positive for Brucella melitensis. Forty-one of the 42 (97.6%) positive cultures were detected by the Bactec 9240 instrument within two to six days; a single positive culture was missed by the instrument and detected by blind subculture performed on day 7.

Update on detection of bacteremia and fungemia

The presence of microorganisms in a patient's blood is a critical determinant of the severity of the patient's illness. Equally important, the laboratory isolation and identification of a microorganism present in blood determine the etiologic agent of infection, especially when the site of infection is localized and difficult to access. This review addresses the pathophysiology and clinical characteristics of bacteremia, fungemia, and sepsis; diagnostic strategies and critical factors in the detection of positive blood cultures; characteristics of manual and instrument approaches to bacteremia detection; approaches for isolating specific microorganisms associated with positive blood cultures; and rapid methods for the identification of microorganisms in blood cultures.

Guidelines for the investigation of invasive fungal infections in haematological malignancy and solid organ transplantation. British Society for Medical Mycology

Invasive fungal infections are increasing in incidence and now affect as many as 50% of neutropenic/bone marrow transplant patients and 5 to 20% of solid organ transplant recipients. Unfortunately, many of the diagnostic tests available have a low sensitivity. The guidelines presented here have been produced by a working party of the British Society for Medical Mycology in an attempt to optimise the use of these tests. The yield of fungi from blood cultures can be increased by ensuring that at least 20 ml of blood are taken for aerobic culture, by using more than one method of blood culture, and by employing terminal subculture if continuous monitoring systems are used with a five-day incubation protocol. Skin lesions in febrile neutropenic patients should be biopsied and cultured for fungi. The detection of galactomannan in blood or urine is of value in diagnosing invasive aspergillosis only if tests are performed at least twice weekly in high-risk patients. Antigen detection tests for invasive candidiasis are less valuable. Computed tomography scanning is particularly valuable in diagnosing invasive pulmonary fungal infection when the chest radiograph is negative or shows only minimal changes. Bronchoalveolar lavage is most useful in patients with diffuse changes on computed tomography scan. The major advances in the diagnosis of invasive fungal infection in patients with haematological malignancy or solid organ transplantation have been in the use of imaging techniques, rather than in the development of new mycological methods in the routine laboratory.

Assessment of Difco ESP 384 blood culture system by terminal subcultures: failure to detect Cryptococcus neoformans in clinical specimens

Terminal subcultures were performed on 1,162 5-day negative blood culture sets which had been monitored by the Difco ESP 384, a continuous-monitor blood culture system. Of these, 16 (1.4%) had growth upon terminal subculture. The isolates not detected by the Difco ESP 384 were Cryptococcus neoformans (eight isolates), Candida albicans (one isolate), Staphylococcus aureus (two isolates) coagulase-negative staphylococcus (three isolates), Bacillus sp. (one isolate), and Corynebacterium sp. (one isolate). Acridine orange staining was performed on 200 randomly selected negative blood culture sets from the study group. Of these, two sets were positive and grew out C. neoformans, as did the terminal subculture. A review of patient's medical records indicated that many of these false-negative isolates were clinically insignificant. The Difco ESP 384 failed to detect 1.4% of the isolates in this study, 50% of which were C. neoformans, indicating a deficiency in the detection mechanism of the system. Further studies demonstrated that while these isolates (C. neoformans) grew in the Difco media, the system did not detect this growth when the standard 5-day protocol was used.