Blood specimens collected for culture by using the high-volume resin-based BACTEC system over an 18-month period at the Seattle Veterans Administration Center were examined in this study. Of 7,783 cultures obtained, 624 were classified as true positives. Patients'in this group had between 20 and 60 ml of blood drawn per culture and separated into 10-ml aliquots for incubation. Analysis of the results stratified by cultured volume and time interval between specimen collection accorded yield advantage to culture volume at the maximal amounts tested. No advantage was observed with any particular interval of collection. Increasing cultured volume from 20 to 40 ml increased yield by 19%o. Increasing cultured volume from 40 to 60 ml increased yield by an additional 10lo. The same effect was seen whether cultures were drawn simultaneously or serially within 24 h. These observations support other reports demonstrating increased yield with increased cultured blood volume. However, they demonstrate increases in yield at volumes much higher than previously considered. In conclusion, this study demonstrates that high-volume blood cultures drawn serially or simultaneously return the best yields.

During a 24-month period, 5,625 blood culture specimens were collected at the Seattle Veterans Administration Medical Center in 20-ml volumes and divided into separate 10-ml aliquots. The two aliquots were processed as duplicate sets (set 1, set 2) by the BACTEC system (Johnston Laboratories, Inc., Towson, Md.). Specimens (5 ml) from each set were inoculated into aerobic (6B) and anaerobic (7C/7D) vials. A total of 434 significantly positive blood cultures were found. In 342 of these positive cultures, yielding 379 isolates (112 members of the family Enterobacteriaceae, 104 staphylococci, 87 streptococci, 27 anaerobes, 20 yeasts, 14 pseudomonads, and 15 miscellaneous organisms), there was adequate specimen volume to fill all four vials. The utilization of set 1 would have resulted only in the failure to detect 65 of 379 (17.2%) significant isolates, 52 of 342 (15.2%) positive cultures, and 20 of 198 (10.1%) bacteremic episodes. There were no significant differences in the recovery of individual species in sets 1 and 2. Although the range of isolates recovered by the aerobic and anaerobic vials of each set differed, the percent yield of total isolates was similar, indicating total isolate yield was predominantly a function of specimen volume. The addition of set 2 most dramatically increased the recovery of Escherichia coli (30%), yeasts (33%), and anaerobes (42%).

We evaluated the ability of the autoSCAN-W/A (MicroScan Division, Baxter Healthcare Corporation, West Sacramento, Calif.), in conjunction with the dried colorimetric Neg ID type 2 panel (DCP) and new rapid fluorometric Neg ID panel (RFP), to identify non-glucose-fermenting gram-negative bacilli by challenging the system with 310 previously identified reference strains. Of these 310 isolates, 286 organisms were in the DCP data base and 269 were in the RFP data base. Use of the DCP panels resulted in 118 (41.3%) correct and 64 (22.4%) incorrect first choice identifications at .85% probability, 61 (21.3%) low-probability identifications, and 43 (15.0%) reports of unidentified organisms. The RFP system reported 135 (50.1%) correct and 25 (9.3%) incorrect identifications at .85% probability and 109 (40.5%) low-probability identifications. Unidentified isolates (DCP system only) and isolates producing low-probability first choice identifications (both systems) required supplementary biochemical testing. Over half (37 of 64 [57.8%]) of the DCP misidentifications were due to four commonly isolated, saccharolytic organisms (Alcaligenes xylosoxidans subsp. xylosoxidans, Pseudomonas putida, Pseudomonas fluorescens, and Xanthomonas maltophilia), while 7 of 25 (28%) of misidentifications in the RFP system were due to P. fluorescens. Of note, the RFP system identified non-glucose-fermenting gram-negative bacilli within 2 h of panel inoculation, allowing additional conventional biochemical tests to be set up the same day on low-probability isolates, whereas only 13.5% of the DCPs could be read at 18 h, with the remainder requiring 42 h of incubation before reading. When organism identifications were recalculated with the updated RFP data base and revised software, only 8.1% of all 310 isolates were misidentified at .85% probability while 77.1% of the isolates were now correctly reported at this same high probability.

During a 6-month study we critically evaluated the accuracy of the AutoMicrobic system Gram-Negative Identification Card (Vitek Systems, Inc., Hazelwood, Mo.) in identifying glucose-nonfermenting gram-negative bacilli by testing 419 selected isolates in parallel with a conventional reference method. Of 356 isolates included in the AutoMicrobic system profile, a total of 307 (86.2%) were correctly identified, 36 (10.1%) were not identified, and 13 (3.7%) were misidentified. Fifty-eight of 63 (92%) isolates not included in the profile were correctly reported as "unidentified organisms." Overall, if the first-choice identification was always accepted, only 18 (4.3%) isolates would have been incorrectly reported. When first-choice identifications appended with the special message "questionable biopattern" were rejected, and organisms were screened for characteristic odor and antimicrobial susceptibility before final acceptance of the AutoMicrobic system report, the number of misidentifications was reduced to 5 (1.2%). The average time to identification with the AutoMicrobic system Gram-Negative Identification Card was 15 h. This compares favorably with the 65 h required by the reference method.

We studied the ability of a new DNA probe-based assay system to correctly identify isolates of the thermophilic campylobacters Campylobacter jejuni, C. coli, and C. laridis grown in vitro. We examined 424 organisms, including 214 Campylobacter isolates and 210 other aerobic and anaerobic isolates. The probe assay, which uses a new homogeneous system in which ail reactions take place within a single tube, demonstrated 100% accuracy, producing neither false-positive nor false-negative results. The assay does not, however, distinguish among C. jejuni, C. coli, and C. laridis.