J Clin Pathol: first published as 10.1136/jcp.36.7.823 on 1 July 1983. Downloaded from J Clin Pathol 1983;36:823-828

Early detection of bacterial growth in culture by impedance with a Bactometer model 32 ANNE BUCKLAND, STUART KESSOCK-PHILIP, SHOSHANA BASCOMB From the Department ofMedical Microbiology, St Mary's Hospital Medical School, London, W2 1PG

SUMMARY A Bactometer model 32 was evaluated for use in early detection of bacterial growth. Experiments with simulated cultures showed that 2 ml of broth introduced into the Bactometer module wells could detect 102 and 106 CFU/ml in 6 h and 2 h respectively. Both Brain Heart Infusion (BHI) and Fastidious Anaerobic broths supported good growth. Detection of nine of 10 organisms inoculated at approximately 106 CFU/ml in BHI were detected within 8 5 h. A culture of Bacteroides fragilis failed to grow under these conditions. Of 189 blood cultures, tested by incubation of 2 ml of BHI, 18 were positive by both conventional and Bactometer methods. False-positive or false-negative specimens were not observed using the Bactometer. Use of the Bactometer enables growth detection at least 12 h earlier than culture methods.

Rapid detection of septicaemia is of primary impor- Material and methods

tance in the diagnostic laboratory. Immediate Gram copyright. of blood specimens can detect approxi- CULTURE MEDIA mately 105 organisms/ml. However, culture methods Two blood culture media were used. Brain Heart are necessary to detect present at lower Infusion broth (BHI) + thymidine + sodium concentrations. Conventional methods rely on cul- polyanethyl sulphonate (SPS) and Fastidious turing of blood specimens and detection of growth Anaerobic broth (FAB) + thymidine, both supplied by subculturing and Gram staining requiring 18-24 by Lab M, Salford, Lancs. h for the first observation, negative cultures being http://jcp.bmj.com/ kept for seven days or more. More sensitive TEST CULTURES methods are therefore sought for faster detection of All strains were recent laboratory isolates identified positive blood cultures. Newer detection methods by starrdard technique.'3 available were reviewed by Bascomb.' Methods applied for detection of septicaemia include BLOOD CULTURES radiometric techniques2 measurement of bacterial Five ml of patient's blood was inoculated into 75 ml

ATP3 and monitoring of impedance changes.4-6 A of BHI and FAB bottles on the ward. The blood on September 29, 2021 by guest. Protected number of impedance measuring devices have been culture set was brought to the laboratory immedi- described.' The Bactometer7 is commercially avail- ately and incubated at 37°C. /8-lactamase (0.5 ml) able and has been used for detection of bacteriuria8 9 (Genzyme Biochemicals) was added to each bottle if and in the food industry'0 and sewage effluent the patient was receiving a /8-lactam . All monitoring."I Its use in detection of septicaemia has blood cultures were subcultured onto blood agar been limited.'24-6 We have evaluated the Bacto- anaerobically and chocolate agar in 7% CO2 after meter model 32 as to its suitability for monitoring 24 h, 48 h and seven days incubation at 37°C. Any growth of in blood cultures from the turbid bottle after 24 h incubation had a Gram film diagnostic laboratory. We also investigated the made and examined. effects of type and volume of medium, concentration of inoculum, and the applicability of VIABLE COUNTS the instrument to the variety of organisms likely to Miles and Misra counts'4 were performed on the be found in blood cultures. experimental broth cultures.

IMPEDANCE MEASUREMENTS Accepted for publication 26 January 1983 The Bactometer consists of a thermostatically con- 823 J Clin Pathol: first published as 10.1136/jcp.36.7.823 on 1 July 1983. Downloaded from 824 Buckland, Kessock-Philip, Bascomb trolled monitoring unit. Four plastic modules (100 repeatedly traverse the channel width in either x 75 mm) each having eight pairs of rectangular direction, they then flatten out but still tend to drift wells can be accommodated in the . Each slightly. This response, seen in all the blood cultures well contains a pair of electrodes that are connected examined, is thought not to be due to growth but to the monitoring unit via metal strips on the under- rather to the erythrocytes settling to the bottom of side of the module. The metal strips converge in a the well. tab that is plugged into the incubator. The periods in which growth responses were The modules come in a sterile plastic envelope, observed varied from 5 to 18 h, and in two cultures, the wells of which are covered by a plastic film, Clostridum perfringens and epider- through which broth can be introduced aseptically midis two growth periods were observed, with a with a syringe and needle. After filling, another period of several hours of no growth in between. sterile adhesive film supplied with each module, is All but one of the organisms selected showed placed over the wells to seal the needle holes. growth within 8-5 h. The times for appearance of Each well can hold 2'5 ml of broth but, to avoid detectable growth are shown in Table 1. The failure overfilling, 2-0 ml was used. The lower row of wells, to detect growth of Bacteroides is probably due to filled with sterile broth, were used as reference, the the poor anaerobic conditions in the small volume of upper row of wells were filled with inoculated broth broth. or media taken from inoculated blood cultures. All eight wells of the module need not be filled EFFECT OF INOCULUM SIZE ON SPEED OF before monitoring commences, further cultures can DETECTION be added as and when they arrive in the laboratory. Dilutions of , Haemophilus influen- Monitoring of impedance can also be achieved by zae and were prepared in inserting Bactometer electrodes into conventional BHI to give initial concentrations of approximately growth vessels. These are reusable autoclavable 1 x 102, 1 x 104 and 1 x 106 CFU/ml. Viable electrodes that are incorporated into a screw cap counts"4 were performed on these broths to establish that can replace the cap of the blood culture bottle. the exact concentrations. All wells were monitored copyright. This enables the bottles to be connected to an exten- for 6 h. As can be seen from Fig. 2 there is a good sion socket inside an ordinary incubator which correlation between inoculum size and detection relays information back to the Bactometer. Each time; 107 CFU/ml were detected in 1.5 h, 102 CFU/ blood culture, bottle requires a reference bottle of ml in 6 h. sterile broth with inserted electrodes. The electrodes can be reused about 30 times. EFFECT OF CULTURE MEDIA ON SPEED OF

The impedance changes are shown as a trace on a DETECTION http://jcp.bmj.com/ chart recorder. Each trace is a representation of a Eight blood cultures were chosen to show differ- growth curve. To fit 32 curves onto one width of a ences between the FAB and BHI media. They were chart recorder paper, each curve is divided into monitored for 48 h and their traces examined. Six segments 8 mm wide (Fig. la). The rate at which the blood cultures showed no response, two showed a trace traverses a channel width gives an indication of growth response. The duration and the shape of the the rate of growth (Fig. ib). growth curves were the same for both media.

Bacterial growth is represented by the trace on September 29, 2021 by guest. Protected traversing a channel width from left to right whereas DETECTION OF GROWTH IN BLOOD CULTURE produce the opposite effect. SPECIMENS One hundred and eighty nine blood cultures were Results examined during a period of four weeks. A 2-0 ml aliquot was drawn from the BHI bottle with a RANGE OF ORGANISMS DETECTABLE BY THE. syringe and needle for examination on the Bactome- BACTOMETER ter as described above. These cultures were moni- Cultures of 10 organisms likely to be found in blood tored for 48 h before being discarded. Complete (Table 1) were inoculated into BHI to give an agreement was found between growth and impe- approximate concentration of 1 x 10 CFU/ml. Two dance methods. Eighteen specimens were found to ml of each broth were transferred to the wells in the be positive by both methods. The remaining 171 top row of the module and 2-0 ml of sterile BHI into were found to be negative by both. The list of the reference wells. The modules were monitored on species isolated is given in Table 3. The majority of the Bactometer for 48 h for signs of growth. cultures were detected within 3 h. One specimen As can be seen from Fig. ic there is an initial containing pyogenes was detected period, ranging from 0-5 to 3 h, in which the traces after 4-5 h, two specimens containing Staphylococ- J Clin Pathol: first published as 10.1136/jcp.36.7.823 on 1 July 1983. Downloaded from Early detection ofbacterial growth in blood culture by impedance monitoring with a Bactometer model 32 825 (a) 7 6 5-- 3 __.__ 2 .

6 ------,~r---- ~ ~ ~ - ..

...... *...... -.

7 2 ~~~.....

2 ._ ...__.------.- _..--..

4 \ _r="___t._...-. 3 .1.;....

6 copyright. 2 .Y .. 6~~~~~~~~~~-.------

00 http://jcp.bmj.com/ recording of32~ cutrs>.,^.*.,t is diie 1into^.fou<-,horzota--.section,. __eac secton i.s lablle_ fro~~~~~~~~(b)0.to 7 representing.,,..th eight culre4sin one mou. (b) Th cetr trac show a grwt curv rsoe. It deostae a shr setln dow period.

~~~~~~~~~~~~~~~~~~(c) on September 29, 2021 by guest. Protected 3 . . ,_

followed by a stationary phase and growth response. The lines get closer together as growth accelerates into the logarithmic phase. The other traces show no growth respon-se. (c) The traces offour sterile blood cultures show how the sealing down periods can vary in shape and duration. *For clarity ofreproduction the Bactometer chart grid has been erased. J Clin Pathol: first published as 10.1136/jcp.36.7.823 on 1 July 1983. Downloaded from 826 Buckland, Kessock-Philip, Bascomb Table 1 Taxa detected by Bactometer Time for 1011. inital detection (h) Staphylococcus aureus 1 Staphylococcus epidermidis '/2 ½/2 1091 Streptococcus pyogenes 23/4 Streptococcus mitior 8Y2 Escherichia coli 1 /2 LL Haemophilus influenzae 31/2 A ( 108. 5 Clostridium perfringens 3Y2 Bacteroides fragilis (failed to grow)

Table 2 Effect ofbroth volume on time for initial detection c0 106\ ofgrowth (h) Volume of broth (ml) 0~~~~~~~~~ 0.5 1-0 15 2-0 2-5 c106 \ Staphylococcus aureus 4Y/2 4½/2 4 31/4 31/4 Streptococcus pyogenes 41/2 33/4 3¼/4 3 23/4 Haemophilus influenzae 4 4 4 3½2 3½2 102. cus epidermidis and Staphylococcus aureus were detected after 8 h, and two specimens containing Staphylococcus epidermidis and Streptococcus mitior were detected after 8*5 h...... 10. copyright. 1 2 3 4 5 6 7 Discussion Time (h) Fig. 2 Relation between initial microbial concentration Monitoring of growth by the Bactometer can be and impedance response detection time. Two ml of done in the special 2.5 ml capacity wells of the suspensions ofE coli (A), Staph aureus (a) and growth modules, or by inserting reuseable sterile Haemophilus influenzae (a) in BHI were inoculated into Bactometer 32 wells and monitored for impedance change.

electrodes into conventional culture containers. To http://jcp.bmj.com/ avoid any danger of contaminating patients' speci- Microbial concentrations determined by viable counts. Fitted line y = 8-157 - I*0148x and 95% confidence limits; mens we used the growth modules only, in this pre- liminary study. Reading of the Bactometer charts is r2 = 0-87908. easily acquired though the initial period of settling Moreover, complete agreement was found between of blood cells can sometimes confuse the interpreta- conventional and impedance detection of sep- tion. It might be useful to investigate further the use ticaemia with 189 blood specimens. Most infected of non-ionic detergent for of the blood cells cultures being detected by the Bactometer method prior to incubation as suggested by Makin and Cor- within 8-5 h, this being at least 10-36 h earlier than on September 29, 2021 by guest. Protected kill'5 to avoid such problems. conventional methods, thereby providing a valuable Both media used for blood cultures in our rapid service to the patient. laboratory were suitable for use in the Bactometer. In simulated cultures nine of 10 different organ- In simulated cultures, presence of 102 CFU/ml could isms tested were detected within 10 h of incubation. be detected in about 6 h, that of 106 in 2 h. The culture of Bacteroides fragilis failed to show Table 3 Bactometer 32 detection time oforganisms found in blood cultures Isolates Detection time (h) Total number ofcultures 05 1 1-5 2 2-5 4-5 8 8-5 Staphylococcus epidermidis 2 2 1 1 2 - 2 1 11 Streptococcus aureus - 2 - 1 -- - - 3 Streptococcus pneumoniae 1 ------1 Streptococcus pyogenes - - - - - 1 - - 1 Streptococcus mitior ------1 1 Escherichia coli - - 1 - - - - - 1 J Clin Pathol: first published as 10.1136/jcp.36.7.823 on 1 July 1983. Downloaded from Early detection ofbacterial growth in blood culture by impedance monitoring with a Bactometer model 32 827 change in impedance during 48 h of monitoring, tant factor. In Table 4 we have shown the costs of probably owing to insufficient anaerobiosis in the Bactometer against that of the other instrument growth well. Over-laying with liquid paraffin might available for processing blood specimens, the Bactec produce better conditions for anaerobic growth, but system.2 As can be seen the Bactometcr is very would probably prove difficult to introduce into the much cheaper than the Bactec, whereasgthe running wells through the plastic strip. A better solution costs per specimen using the growth modules inocu- might be to insert the Bactomatic electrodes into an lated from the conventional broth are similar. anaerobic culture bottle. Detection of growth of At present it would not be safe to replace conven- Eubacterium limosum and of Fusobacterium nuc- tional culturing methods by the Bactometer impe- leatum has been reported.7 dance monitoring, but the instrument might be used In preliminary work with the Bactometer before concomitantly with the conventional method, cul- the study began, a problem was encountered with a tures being kept for 24 h only. In this way heavily blood culture that had been incubated over the infected specimens would be detected within the weekend before being monitored. It yielded large working day and treatment of patients could be numbers of bacteria on subculture but showed no started or amended sooner. Cultures not detected change of impedance. 'This was probably due to the within 24 h would, for the present, have to be hand- fact that nutrients in the culture media were led by conventional methods. exhausted during the weekend and metabolism could not therefore be detected by the Bactometer. We thank BCJ Levett of Bactomatic Inc, for advice However, the bacteria present could grow when and loan of equipment used in this evaluation. supplied with new nutrients by subculturing. The Bactometer procedure is therefore not suitable for References cultures can 'Bascomb S. Application of automation to the general and specific such specimens. These heavily infected detection of bacteria. Lab Practice 1981;30:461-4. be quickly detected by Gram staining those bottles 2 DeBlanc HJ Jr, Deland F, Wagner HN Jr. Automated which show turbidity. Alternatively, such specimens radiometric detection of bacteria in 2967 blood cultures. Appl should be subcultured into fresh broth before Bac- Microbiol 1981;22:846-9. tometer monitoring. 3Schrock CG, Barza MJ, Deming JW, Picciolo GL, Chapelle EW, copyright. Weinstein L. Rapid detection of bacterial growth in blood cul- The model 32 Bactometer system has one serious tures by means of adenosine triphosphate determination. In: limitation, the number of cultures it can deal with at Abstracts, American Society for Microbiology. Annual Meet- one time (32) is too small. As impedance is much ing, 1976 Paper No C71:37. affected it is to incubate a Khan W, Friedman G, Rodriquez W, Controni G, Ross S. Rapid by temperature essential detection and of bacteria in blood and cerebrospinal reference cell for each culture. If each specimen is fluids in children by the electrical impedance method. In: tested in both aerobic and anaerobic broth, this Johnston HH, Newsom SWB, eds. 2nd International sym- http://jcp.bmj.com/ reduces the number of specimens to 16. Moreover, posium on rapid methods and automation in microbiology. if to two it allows Oxford: Learned Information, 1976:13-4. they need to be kept for up days, 'Throm R, Strauss R, Specter S, Friedman H. Automated blood for only eight blood specimen investigations per day. culture testing using radioisotope and electrical impedance In our laboratory serving a group of hospitals with a monitoring equipment. In: Johnston HH, Newsom SWB, eds. total of 547 beds, an average of 14 blood cultures 2nd International symposiurn on rapid methods and automa- A with ton in microbiology. Oxford: Learned Information, are processed per day. machine greater 1976:21-2. capacity would be necessary in busy laboratories. 6 Specter S, Throm R, Strauss R, Friedman H. Rapid detection of on September 29, 2021 by guest. Protected The capital outlay for an instrument is an impor- bacterial growth in blood samples by a continuous-monitoring electrical impedance apparatus. J Clin Microbiol 1977;6:489-93. Table 4 Comparison ofpurchase and running costs of Cady P, Dufour SW, Shaw J, Kraeger SJ. Electrical impedance Bactometer 32 and Bactec 460 measurements: rapid method for detecting and monitoring micro-organisms. J Clin Microbiol 1978;7:265-72. Bactometer Bactec Zafari Y, Martin WJ. Comparison of the Bactometer* microbial (£) (£) monitoring system with conventional methods for detection of micro-organisms in urine specimens. J Clin Microbiol Capital outlay 9900-00 16 604-00 Cost per blood culture 1977;5:545-7. bottle 0-35 0-76 Cady P, Dufour SW, Lawless P, Nunke B, Kraeger SJ. Cost of module per test Impedimetric screening for . J Clin Microbiol (2 wells) 0-45 1978;7:273-8. Cost of electrode 1-00 '° Martins SB, Selby MJ. Evaluation of a rapid method for the Cost per patient 1-60* 1-52 quantitative estimation of coliforms in meat by impedimetric Cost per patient 1-52t procedures. Appl Environ Microbiol 1980;39:518-24. *Based on one bottle and two wells for each medium. "Silverman MP, Munoz EF. Automated electrical impedance tBased on two bottles of each medium and 1/30th of cost of a technique for rapid enumeration of fecal coliforms in effluents reusable electrode. from sewage treatment plants. Appl Environ Microbiol Prices are exclusive of VAT. 1979;37:521-6. J Clin Pathol: first published as 10.1136/jcp.36.7.823 on 1 July 1983. Downloaded from 828 Buckland, Kessock-Philip, Bascomb 12 Hadley WK. The use of electrical impedance measurements for the blood. J Hyg 1938;38:732-49. detection of in blood culture. In: Johnston "S Makin T, Corkili JE. The detection of bacteria in blood cultures HH, Newsom SWB, eds. 2nd International symposium on by impedance monitoring-a pilot study. Med Lab World rapid methods and automation in microbiology. Oxford: 1978:409-15. Learned Information, 1976:14. 13 Cowan ST. Cowan and Steel's manual for the identicration of medical bacteria. 2nd ed. Cambridge: Cambridge University Requests for reprints to: Dr S Bascomb, Department of Press, 1974. , St Mary's Hospital Medical School, 14 Miles AA, Misra SS. The estimation of the bactericidal power of London W2 1PG, England. The June 1983 issue THE JUNE 1983 ISSUE CONTAINS THE FOLLOWING PAPERS Screening hospital patients for uterine cervical Phenomenon of resistance to Augmentin associated cancer ELIZABETH HUDSON, SHEILA HEWERTSON, with sensitivity to ampicillin: occurrence and C JANSZ, H GORDON explanation W BRUMFITT, JMT HAMILTON-MILLER, Histopathology of papilloma virus of the SHIRLEY DIXSON, RA GARGAN, ANN GOODING cervix uteri: the history, taxonomy, nomenclature Diagnosis of urogenital gonorrhoea by detecting and reporting of koilocytic dysplasias S FLETCHER gonococcal antigen with a solid phase Diagnosis of amniotic fluid embolism using an immunoassay (GonozymeTM) D DANIELSSON, antiserum to human keratin IWC GARLAND, WD H MOI, L FORSLIN THOMPSON Effect of adding a papain digest of ox liver to brain Epithelial markers in the diagnosis of naso- heart infusion cysteine broth on the recovery of pharyngeal carcinoma: an immunocytochemical non-sporing anaerobes from simulated blood study BA GUSTERSON, DIANA P MITCHELL, MJ WAR- cultures DC SHANSON, JANET PRATT

BURTON, RL CARTER Problems of infection after bone marrowcopyright. Granulomatous disease in the vermiform appen- transplantation J GRAHAM WATSON dix DC ALLEN, JD BIGGART Increased mean platelet volume in septicaemia i An ultrastructural and immunocytochemical study VAN DER LELIE, AEG KR VON DEM BORNE of a renal carcinoma secreting inactive renin GBM A fluorimetric method for red blood cell sorbitol LINDOP, IAR MORE, BRENDA LECKIE dehydrogenase activity G VACA, P ZtIGA, C Comparison of the haematoxylin basic fuchsin picric MEDINA, R ALONSO, G GONZALEZ-QUIROGA, RI http://jcp.bmj.com/ acid method and the fluorescence of haematoxylin ORflZ-DE-LUNA, JM CANTO and eosin stained sections for the identification of A fluorimetric method for the measurement of early myocardial infarction HK AL-RUFAIE, RA pyridoxal and pyridoxal phosphate in human plasma FLORIO, EGJ OLSEN and leucocytes, and its application to patients with Liver histopathology of the hepatitis A virus sideroblastic marrows GILLIAN P SMITH, DIANA infection: a comparison with hepatitis type B and SAMSON, TJ PETERS non-A, non-B P KRYGER, P CHRISTOFFERSEN Isolation of collagen stimulating factors from on September 29, 2021 by guest. Protected Preliminary report of an association between healing wounds RP O'HARE, A FALLON, JF virus and achalasia DB JONES, JF BRADLEY, J BURNS, J O'D MCGEE MAYBERRY, J RHODES, JULIA MUNRO Serum malondialdehyde-like material (MDA-LM) Coxsackie B virus infection in coronary care unit in acute myocardial infarction J AZNAR, M TERESA patients D O'NEILL, JD MCARTHUR, JA KENNEDY, G SANTOS, JUANA VALLES, J SALA CLEMENTS Use of total cholesterol/albumin ratio as an Examination of operation specimens from patients alternative to high density lipoprotein cholesterol with spinal tuberculosis for tubercle bacilli BW measurement AA NANJI, SUSEELA REDDY ALLEN, DA MITCHISON, JANET DARBYSHIRE, WWK Letters to the Editor CHEW, M GABRIEL Book reviews Attempts to isolate Campylobacter jejuni from various body sites EP WRIGHT Some new titles Copies are still available and may be obtained from the PUBLISHING MANAGER. BRITISH MEDICAL ASSOCIATION, TAVISTOCK SQUARE, LONDON WC1H 9JR. price £500, including postage