report Blood culture-current state andfuture prospects* · shock have shown evidence of unresolved sepsis at necropsy. This series of presentations provides a concise review of current

J Clin Pathol 1983;36:963-977

Symposium report

Blood culture-current state and future prospects*Edited by JC GOULD*, BI DUERDENt

From the *Central Microbiological Laboratory, Western General Hospital, Edinburgh EH4 2XU and thetDepartment ofMedical Microbiology, University ofSheffield Medical School, Sheffield SIO 2RX

Introduction JC GOULD

Blood culture results are important to the clinicianin the management of individual patients and therelevant methods and techniques present consider-able problems for laboratory workers. In this meet-ing, a distinguished group of laboratory and clinicalexperts discussed the technical problems in thelaboratory and referred to clinical attitudes to theneed for, and performance of blood culture.Dr Duerden, in discussing the clinical significance

of bacteraemia, pointed out that a great majority ofblood cultures were taken to obtain laboratoryconfirmation of suspected clinical bacteraemia andnot for the reassurance of a negative result. Henoted that differences in the proportion of positiveresults found in different hospitals has not beenrelated to significantly different outcome for thepatients. He elaborated on the general interpreta-tion of laboratory results of blood culture in differ-ent groups of patients and emphasised that moreattention should be paid to the significance of asingle positive culture in the individual patient. DrMunro continued this theme from the clinician'spoint of view, illustrated by a number of individualclinical experiences. In his opinion there was a greatneed for close liaison between clinical and laborat-ory staff and he emphasised the value of blood cul-ture results in medical audit.

Routine diagnostic laboratories have becomeaccustomed to carrying out blood cultures in a moreor less conventional manner, and the techniques thatappear to give the most consistent results werereviewed by Dr Shanson. The range of organisms tobe covered and their nutritive and environmentalrequirements for growth are so variable that atten-

*A review based upon papers presented in "Blood Culture", aJoint Symposium of the Royal College of Pathologists and theAssociation of Clinical Pathologists, at the Barbican Centre, Lon-don, on October 3, 1982

Accepted for publication 25 May 1983

tion must be given to covering as wide a range ofconditions as possible. This is particularly true whenthe nature of the infecting organism is not known, orif there is continuing clinical suspicion of bac-teraemia despite previously negative blood cultures.Of particular importance are infections due toanaerobic bacteria, many of which are associatedwith bacteraemia. Dr Watt described the steps to betaken in the laboratory to increase the chances ofisolation of obligate anaerobes from blood cultures.Improvement in patient management, particularly

of immunocompromised hosts in whom the clinicalsigns of bacteraemia may be masked or absent, hasincreased interest in rapid methods for the detectionof bacteraemia. Dr Johnston dealt in detail withthose techniques designed to detect bacterial pro-ducts, concentration methods that are not yet estab-lished and innovative "automated" methods that arenot necessarily more rapid than conventional tech-niques. The two important new methods dependupon radiometry and impedance-conductance andare technically more convenient in a laboratory butprovide only marginal clinical benefit.

All laboratory workers have experienced theproblem of contamination of blood cultures; undernormal conditions the contamination rate may be>10%. Not only does this waste effort, but theresults are confusing to the clinician. Professor Col-lee detailed the main sources of contamination andemphasised procedures to reduce the risks to aminimum; these involve both the clinical andlaboratory staff at different stages in the procedureof sampling blood and their importance cannot bestressed too strongly. Clearly, results of blood cul-ture can be no better than the care taken in vene-puncture.

Finally, Professor Ledingham discussed the ques-tion of the diagnosis, treatment and pathogenesis ofthe bacteriogenic shock syndrome which may occurwith or without bacteraemia. Although resuscitationof patients with bacteriogenic shock utilises non-specific methods, he pointed out that the definitive

963

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

964

diagnosis of the nature and sites of the underlyingsepsis is important for the success of specific surgicaland antimicrobial therapy. Most patients who ulti-mately succumb after surviving the acute episode ofshock have shown evidence of unresolved sepsis atnecropsy.

This series of presentations provides a concisereview of current blood culture procedures, alongwith their relevance to the clinical diagnosis andtreatment of infection. This collection of contem-porary opinion should therefore be of value to allthose concerned with the diagnosis and managementof patients in whom bacterial invasion of the bloodstream is suspected.

Why take blood cultures?-the clinical significanceof bacteraemia BI DUERDEN

Most blood cultures are taken to detectbacteraemia-only a small minority are taken forthe reassurance of a negative result. Blood should besterile and it is assumed generally that a positiveblood culture result is highly significant; it indicatesa life-threatening situation and requires immediateaction. However, this may be a rather simplisticapproach.

Bacteraemia may be a dramatic event but the dif-ferences between the numbers and proportions ofpositive results found in different hospitals are notreflected in significantly different outcomes for thepatients. In hospitals with a highly developed bloodculture system, more positive results are found.Some are important but others may be relativelyminor complications of underlying conditions, and,in hospitals where a less vigorous search is made,most of these patients recover uneventfully. Else-where, enthusiastic sampling of moribund patientscan boost the ratio of positive findings with terminalbacteraemias. With recent advances in blood culturemethods, microbiologists and clinicians shouldexamine the clinical situations in which bacteraemiamay be detected because their interpretation of theresult depends upon the type of patient, any under-lying conditions and the organism.' 2

The bacteraemia itself may threaten life; a ful-minating septicaemia-for example, overwhelmingGram-negative bacillary sepsis or meningococcalsepticaemia, is lethal. Alternatively it may be thespecific presentation of the illness; bacteraemia isthe identifiable aspect of the disease in-for exam-

ple, typhoid and infective endocarditis. This intro-duces the concept of PUO; fever is one of the mainresponses to infection so that bacteraemia and PUOare inevitably linked. The positive blood cultureresult may be the only finding that explains the feverand provides the diagnosis. In all of these condi-

Gould, Duerden

tions, the diagnosis and management of the patientdepend upon the reliability of the blood culture sys-tem. The significance of the result is beyond doubt.

In many infections, however, bacteraemia is animportant part of the underlying disease but can bealmost an incidental finding. Bacteraemia is com-mon in pyelonephritis, pneumococcal disease,cholecystitis, osteomyelitis, epiglottitis and mening-itis and may be the life-threatening aspect of thedisease. Moreover, in many of these conditions theblood culture provides the most reliable bacteriolog-ical diagnosis. The detection of a bacteraemia mayalso be the key to an underlying disease; the isola-tion of Bacteroides fragilis or a mixture of B ftagilisand Escherichia coli may indicate intra-abdominalsepsis and there is an association between a Bac-teroides bacteraemia, as the initial finding, and car-cinoma of the colon.3The interpretation of the laboratory findings is

usually straightforward in these well defined groupsbut a large proportion of blood culture investiga-tions are on patients in whom the evaluation of apositive result requires consideration of their gen-eral condition. The positive blood culture may be anincidental, although not necessarily insignificant,finding. Many patients have a transient bacteraemiaafter intestinal, genitourinary or oral surgery. Ana-haemolytic-streptococcal bacteraemia may followeven minor dental manipulations and sigmoidoscopyor cystoscopy can lead to a coliform bacteraemia,but most of these are insignificant; they are clearedwithin minutes with no lasting ill effects.34 Theenthusiasm with which post-operative bacteraemiasare sought is one reason for differences between thepatterns of blood culture results in different hospi-tals. The number of positive results will increase ifcultures are taken from all postoperative patients.This has a particularly noticeable effect upon thenumber of Bacteroides spp, or mixture of E coli andBacteroides spp, isolated from blood cultures. How-ever, the differences in detection rates do not neces-sarily affect the outcome for the patients. Thosepatients who develop significant sepsis with bac-teraemia after-for example, abdominal surgery,are clearly unwell and blood cultures are collectedbecause of their clinical condition rather than as aroutine. Operations on sites where there is pre-existing infection may also lead to bacteraemia withorganisms of proven virulence and the bacteraemias,perhaps with metastatic abscesses, are significantproblems. However, improved surgical techniquesand prophylactic chemotherapy have reduced post-operative sepsis in abdominal and gynaecologicalsurgery-and several centres have reported a fall inthe incidence of postoperative bacteraemia.' 5 Simi-larly, operations on infected sites are often per-

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

Blood culture-current state and fiuture prospectsformed with antibiotic cover to prevent the spreadof infection via the blood stream.Other "incidental" but significant bacteraemias

are those secondary to serious underlying diseases.In compromised patients the interpretation of a

positive blood culture depends upon the generalstate of the patient and the organism. Modern inten-sive care creates many opportunities for infection.Bacteraemia is common in These patients and may

kill them, but the assessment of a positive bloodculture is not easy. Opportunist pathogens suchas Serratia, Enterobacter, Acinetobacter andcoagulase-negative staphylococci are common buttheir significance may vary. In one patient, changinga colonised vascular cannula will stop the bac-teraemia and solve the problem but the isolation ofthe same organism from another patient may indi-cate a life-threatening septicaemia. The bacteraemiamay also be a terminal event, not necessarily signify-ing that the bacteraemia kills the patient but that it isan inevitable part of the breakdown of normal bodyfunctions.

In patients with prosthetic implants-for example,CSF shunts for hydrocephalus or artificial heart val-ves, there is a high risk that the implants will becomecolonised, often with organisms such as coagulase-negative staphylococci or coryneforms leading to a

persistent low-grade bacteraemia. Endocarditis on a

prosthetic valve is a very serious condition and inpatients with CSF shunts, infection causes valve mal-function, neurological emergencies and immunecomplex nephritis. The initial effects may be mild,diffuse and difficult to assess-a "silent" bac-teraemia. Blood cultures are taken when suchpatients are in any way unwell, but it is difficult tointerpret the significance of a coagulase-negativestaphylococcus in a culture from a vaguely unwellchild, with difficult veins and a high risk of contami-nation, who has so many medical problems thathealth and illness are only relative terms.

Similar problems arise in immunocompromisedpatients with immunodeficiency and neutropenia as

a result of malignant disease or its chemotherapy.Bacteraemia is a common cause of death. Theorganisms involved include recognised pathogensand opportunist species and the bacteraemia may bealmost completely silent, but a diagnosis is impor-tant because a silent bacteraemia may be rapidlyfatal.6

The significance of a blood culture result dependsto a variable extent upon the organism isolated.25The isolation of Salmonella typhi is undonbtedlyimportant, or of a f3-haemolytic streptococcus, or

E coli from a patient with acute pyelonephritis andbacteraemic shock. Isolating, identifying and testingthe antibiotic sensitivity of these organisms is essen-

965

tial, but there is less concern about the same E colifrom a patient shortly after sigmoidoscopy. Similarlyan a-haemolytic streptococcus from a patient withendocarditis is the cardinal laboratory result onwhich survival may depend whereas the same organ-ism in a blood culture taken shortly after a dentalextraction from a healthy person signifies little. Theisolation of Staphylococcus aureus, the meningococ-cus, Haemophilus influenzae, or the pneumococcushas immediate significance, but the isolation of acoagulase-negative staphylococcus, a coryneform,or an unusual Gram-negative bacillus from a com-promised patient is more difficult to interpret.

This contrast highlights the dilemma in determin-ing a policy for blood cultures. The microbiologistaims to recover any potential pathogen from cul-tures taken by the clinician at appropriate timesfrom potentially infected patients. Improvements inthe system, greater co-operation and more emphasison blood cultures should produce more positiveresults, but some may be the less significant findings.This is not a reason for neglecting to improve thesystem, but it may dilute the impact of a positiveresult. Microbiologists and clinicians should developa critical approach to blood culture results so thatthey neither over-react to the less significant, norlose the impact of the important one. Most bloodcultures will be sterile and a positive result is thesignal for immediate action, but tempered by a care-ful assessment of its significance.

The blood culture curious clinician JF MUNRODepartment of Medicine, Eastern General Hospital,Edinburgh

Blood cultures are performed primarily for theirimmediate diagnostic or therapeutic value. Addi-tionally, information may be obtained which mayprove helpful in the management of subsequentpatients. The analysis of blood culture results is alsoa useful part of medical audit.

DIAGNOSTIC VALUE-POSITIVE BLOODCULTURESThe diagnostic value of blood cultures is greatest inthose patients in whom the cause of their illness isuncertain. Occasionally, such patients are sufferingfrom a rare condition, such as brucellosis, which theclinician might overlook. Sometimes the problem ischaracteristically difficult to diagnose-for example,infective endocarditis. More frequently, however,the final diagnosis turns out to be a fairly commoncondition presenting atypically.

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

966

Case illustrationA 58-year-old man who had worked with animals allhis life had been investigated previously because of"shivering turns." No explanation was found. In thepast he had a myocardial infarct and was known tohave renal calculi. At the time of his admission,there was some derangement in liver function testsand a coliform organism was isolated from his blood.Subsequent investigations showed a stone in hiscommon bile duct although he had never had attacksof biliary colic.

This case also illustrated that septicaemia per se isnot a diagnosis. The cause of the septicaemia andthe source of the organism should always be soughtand the possibility of blood-borne dissemination ofinfection considered. Sometimes, the source ofinfection may remain uncertain.

Case illustrationA 62-year-old housewife was admitted with increas-ing polyuria of several weeks duration and increas-ing confusion and abdominal pain for two days. Shewas pyrexial and had the features of severe diabeticketoacidosis with generalised abdominal tenderness.She had a Gram-negative septicaemia and a highserum amylase, which suggested but did not confirm,the possibility of pancreatitis. Further investigationsrevealed an E coli urinary tract infection and an IVPshowed chronic pylonephritic changes in her leftkidney. Moreover, she had gall stones and a dilatedcommon bile duct.

NEGATIVE BLOOD CULTURESA substantial number of patients who appear, clini-cally, to be bacteraemic do not have positive bloodcultures. Only some of these episodes are attribut-able to prior administration of antibiotics. Althoughnegative blood cultures cannot exclude a bac-teraemic episode, they should alert the clinician toother possibilities.

Case illustrationA 54-year-old housewife experienced a transientischaemic attack four days before admission. Shehad a further cerebrovascular episode on the day ofadmission. She was found to have multiple splinterhaemorrhages and haematuria; she was assumed tohave infective endocarditis and treated accordingly.However, several sets of blood cultures obtainedbefore chemotherapy was started gave negativeresults and this, combined with a failure to improve,suggested the possibility of non-infective endocar-ditis. Further investigations revealed the presence ofa left ovarian tumour that had already seededthroughout the peritoneum.

Gould, Duerden

THERAPEUTIC VALUEThe detection of a relatively unexpected bac-teraemia is not only of diagnostic but also oftherapeutic significance. Frequently, however, posi-tive blood cultures merely confirm the clinical diag-nosis of severe iAfection. In these circumstances, thetherapeutic value is restricted by the delay in bac-teriological confirmation but may be of great valuein a number of specific situations.1 The isolation and identification of an organismprovides the clinician with the opportunity to nar-row the spectrum of antibiotic therapy. However, itis important to appreciate that sometimes, particu-larly with intra-abdominal sepsis, antibiotic treat-ment should not be restricted to the only organismidentified.2 Treatment may need to be changed when theidentity of the infecting organism was unsuspected.

Case illustrationAn elderly man was admitted to hospital with asevere chest infection. By chance, two weeks previ-ously his general practitioner obtained a chest x-raywhich was normal. The x-ray on admission showed anarrow, walled lung abscess. The clinical assumptionof staphylococcal infection had to be revised when aKlebsiella species was isolated from the blood.

3 When the organism has an unusual pattern ofantibiotic resistance, it may be necessary to changethe initial therapy. This is particularly likely inpatients who develop a septicaemia while in hospi-tal.4 Identification of the organism and determinationof MICs and MBCs of suitable antibiotics are essen-tial when it is very important to achieve appropriatetherapeutic levels-for example, in the managementof infective endocarditis.

MEDICAL AUDITBacteriological information is helpful to the clinicianwho has to apply a "best guess" when initiatingtherapy before sensitivities are available. Moreover,previous experience may influence medical practice,particularly in those fields in which therapeuticdevelopments have altered the pattern ofmedicine-for example, renal transplantation andhaematological malignancies. During a 5-yearevaluation of blood cultures in a general medicalunit, patients were divided into four groups-PUO,infected, at risk, and both infected and at risk-onthe basis of the clinical findings at the time of per-forming the initial blood culture. Of the latter, rela-tively small group of 101 patients, 30 were sep-ticaemic. Diabetes mellitus provides an example ofincreased risk. Diabetic metabolic decompensation

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

Blood culture-current state and future prospects

may be caused by infection but it also predisposespatients to infection. Of 211 consecutive patientsadmitted to a general medical unit with severemetabolic decompensation, 41% were infected.Nine patients died, of whom only three were underthe age of 60 yr; all three were septicaemic onadmission. As a consequence of this study, it hasnow become unit policy to initiate chemotherapy inpatients with severe ketoacidosis as soon as theappropriate bacteriological samples, including bloodcultures, have been obtained.The diagnostic, therapeutic and audit value of

performing blood cultures in patients suffering froma severe infection or presenting with a diagnosticchallenge depends to a great extent upon the close-ness of the liaison between the clinical and laborat-ory staff.

Considerations of contamination in relation to bloodculture JG COLLEEDepartment ofBacteriology, University MedicalSchool, Teviot Place, Edinburgh EH8 9AG

Contamination of a blood culture at some stage ofsampling or subsequent processing is a great wasterof time and effort and it may put the patient at risk.In a recent Scottish survey, almost 14% of 3000 setsof blood cultures were contaminated. Perhaps we dotoo many investigations with specimens of poorquality and not enough with specimens of high qual-ity. In blood culture work, a replicated finding isoften assumed to be confirmatory, but this must notbe based on a consistent error that allows the literalreproduction of a mistake.

Whilst an extended series of blood cultureattempts may be justified in some clinical circum-stances, most cases of bacteraemia can be detectedby the first or second set of bottles if the methodol-ogy is good.7 Accordingly, our plea to the cliniciansmust be to take special care to start with one or twoclean specimens. Thereafter, the law of diminishingreturns applies; but it must not be inferred that theclinician can then proceed to be less careful.

SOURCES OF CONTAMINATIONThe skin of the patient or the operator is the mostlikely source of contaminants. As various skin com-mensal organisms or transient skin contaminantsmay occur as opportunist invaders in debilitated orcompromised hosts, their presence in blood culturesmay be significant on occasion and cannot be invari-ably dismissed. This is why adequate cleansing anddisinfection of the venepuncture site are so impor-tant. First, the operator's hands must be clean. Someauthorities hold that the operator should wearsterile gloves, and this has much to commend it.

967

Then the venepuncture site must be disinfected.It is essential to stress that soap and water and

quatemary ammonium compounds-for example,benzalkonium chloride or cetrimide in aqueous solu-tion, are good cleansing agents but are, at best, slowdisinfectants. Moreover, some aqueous preparationsheld in hospital wards may be heavily contaminatedwith bacteria. Small disposable packs should beused.

After cleansing, skin should be disinfected as fol-lows. A preparation containing 70% alcohol andwater should be applied quite vigorously with one ortwo sterile swabs to the venepuncture site to achievea quick reduction in the numbers of the vegetativebacteria. A little time should be allowed for this"quick kill" stage-until the alcohol has dried. Theimportant constituent of the skin disinfectant is 70%ethyl alcohol or isopropyl alcohol (which is cheaper)in water; chlorhexidine 1%, iodine 2% or aniodophor 2% may be added to the alcohol toenhance its disinfectant action.The volume of blood to be taken is discussed later

in this review. The bulk-blood-culture approachwhereby multiple small amounts are dispensed fromone grand venepuncture into a batch of several bot-tles vitiates any confirmatory value of replicate isola-tions. As the practice increases the chances of con-tamination and reduces the chances of detection ofbacteraemia, it should be avoided. At this stage theoperator should not be afraid to put all of his eggsinto one or two good microbiological baskets, but ifone syringe is to provide material for bacteriology,clinical chemistry and haematology, the microbiol-ogy is prejudiced. The quality of the sampling isvitally important and our clinicians must be dis-criminating.Changing the sampling needle The sampling needleshould be discarded and another sterile needle fittedbefore the sample is injected into each bottle. Theoperator must here bear in mind the hazard ofacquiring hepatitis B virus from a needle stickinjury. The top of the rubber liner should not beswabbed if it is already sterile (but see Shanson'sfindings8). My colleagues and I prefer crimped pro-tective aluminium foil caps, as contaminants some-times grow in a film of moisture trapped betweenplastic ViskapsR and the bottle caps.

Specimens from special care units for neonatesand from intensive care areas for adults are oftencontaminated. Patients in these units are at specialrisk from bacteraemia and septicaemia, and bloodculture is a crucial aid to their management. Withneonates, the method of dropping blood into anopen bottle from a fine needle inserted into a verysmall vein has evolved because of the tendency offragile veins to collapse when a syringe is used. As

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

968

the blood culture bottle cap is removed and the sys-tem is then very vulnerable to airbome contamina-tion, bacteriologists deplore the compromise whilstunderstanding the technical difficulties. There is anurgent need for new miniaturised soft plastic equip-ment with a sterile, closed system for neonates andinfants.

If the inguinal vessels are used, either in the adultor the infant, there is an increased risk of contamina-tion. If the doctor turns in despair to tapping longcentral venous lines or peripheral venous catheters,shunts or three-way taps for the sample, mic-robiologists in turn look despairingly at the ensuingcollection of isolates that may be of doubtfulsignificance. Under these circumstances, the medicalmicrobiologist must see the patient and cliniciansand must attempt to relate the findings to the clinicalcircumstances.

LABORATORY CONSIDERATIONSSubcultures need not be done in a laminar flowcabinet, but still air in a quiet, spacious room helpsto reduce contamination. If a wire loop is used forsubculturing from the bottle, it should be longenough to ensure that the handle does not enter theneck of the bottle. Some workers frown on this pro-cedure and insist on a separate syringe and needlefor each subculture.

In our experience at Edinburgh with an early sub-culture step at 4-6 h in our routine blood culturesystem, we seem to be getting results that make iteasier to distinguish between contamination and asignificant isolate. This study was generated by mycolleague Mr G McInnes after consideration ofresults communicated to us by Dr Leela Ganguli atManchester.There are some miscellaneous contaminant situa-

tions that merit consideration. (i) Organisms seen instained smears taken directly from the blood culturebottle must not be contributed from neglected solu-tions of stains that sometimes contain yeast-likeforms; (ii) if an early centrifugation step is includedto increase the chance of seeing organisms in stainedsmears, it should be remembered that some ingre-dients of culture media contain dead but morpholog-ically intact bacteria that can confuse the issue; (iii)viable organisms may be contributed from a heat-sensitive ingredient added as a supplement to theblood culture medium after the autoclaving stage.Quality control of blood culture media is esentialand a record of all contaminants should be kept inthe laboratory.A challenge to conventional concepts of the steril-

ity of normal blood has been declared by the work ofBisset and Bartlett9 and others who believe thatL-forms and revertants of Bacillus spp are regularly

Gould, Duerden

associated with the red cells of normal people. Thechallenge has gone unanswered for too long anddemands further, independent study. Some workershave noted that media containing thiol compoundsmay become toxic after storage. This aspect ofchemical contamination of a medium merits consid-eration along with aspects of microbial contamina-tion and calls for a careful distribution and recallsystem for media on the wards combined with qual-ity control in the laboratory.

Anerobic blood cultures B WATTBacteriology Laboratory, City Hospital, GreenbankDrive, Edinburgh

Anaerobic bacteria can be associated with life-threatening episodes of bacteraemia and sep-ticaemia. A recent survey of cases of bacteraemia inEngland and Wales showed that 6-5% of all caseswere due to anaerobic organisms, mainly Bac-teroides spp.s There is, therefore, a need for diagnos-tic laboratories to be able to recover anaerobes con-sistently from blood cultures.Many laboratories may questidn their ability to

recover anaerobes or may seek to compare theirperformance with that of other centres. There havebeen many comparative studies'0-14 but no one sys-tem, or single medium emerges as a clearly consis-tent favourite from this extensive literature. There isno such thing as a "standard diagnostic laboratory",and comparison of different isolation rates from dif-ferent centres may be very misleading because sev-eral factors other than technical expertise mayinfluence the results-notably the type of hospital(hence the type of patient) and the extent ofprophylaxis with agents such as metronidazole. Thepurpose of this short paper is not to confuse thereader by a "conducted tour" of the literature, butrather to suggest some guidelines that may helpindividual laboratories to improve their proceduresto suit their own requirements.The main factors governing the success or failure

of attempts to recover anaerobes from blood cul-tures include clinical awareness, sampling techni-ques, transport of specimens to the laboratory, thetype of blood culture system used, and the bloodculture procedure. Clinical awareness is crucial inthe diagnosis of anaerobic septicaemia; apart fromcases due to Clostridium perfringens, most cases donot have any characteristic clinical features, so clini-cians must be aware of the possibility in order thatsamples can be processed correctly. Other aspectssuch as a careful sampling technique and rapidtransport of specimens to the laboratory need to beborne in mind. It should always be remembered that

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

Blood culture-current state and future prospects

anaerobic bacteriology can demonstrate the pres-ence of an anaerobe in a blood sample but cannotprove the absence of anaerobes from such a sample.

BLOOD CULTURE SYSTEMSAny laboratory that intends to set up a blood culturesystem for anaerobes has to decide upon the follow-ing: (i) number of bottles; (ii) size of the bottle(s);(iii) volume of culture media in each bottle; (iv)culture medium; and (v) incubation environment.Some laboratories use a single bottle, with a

"compromise" medium for culture of aerobes andanaerobes, but most laboratories use a dual systemwith one bottle for aerobes and one for anaerobesand some authorities recommend a three-bottle sys-tem.8 Whatever the size of bottle chosen, it is impor-tant to ensure that the volume of fluid is such thatthe addition of a suitable volume of blood leaveslittle or no gas space above the culture. The bottleshould be incubated with the cap loosened or with aplugged vent in an anaerobic atmosphere thatincludes 10% CO2 to allow good recovery ofanaerobes.'5 Anaerobic cabinets may prove to bevery useful as anaerobic incubators for such sam-ples. In many laboratories, samples are either incu-bated in a CO2 incubator, placing reliance on theculture medium itself to maintain suitable conditionsfor the growth of anaerobes, or samples are col-lected into vacuum systems that may not provideany carbon dioxide in the incubation environment.Neither system will allow optimal recovery ofanaerobes.Examples of some of the blood culture media

used commonly for anaerobes are show in Table 1,together with approximate costs. In deciding on asuitable medium, it is important to bear in mind: (i)a suitable medium gives both good recovery and sus-tained viability of anaerobes; (ii) comparativestudies of blood culture media using laboratorymodels may not have much predictive value forresults with real samples from real patients; (iii) therelative costs of different media may varyconsiderably-but so may the relative ease of prep-aration, e.g., cooked-meat broth medium gives verygood results and appears to be reasonablyinexpensive-but it is time-consuming to prepare.

Table 1 Relative costs ofsome commonly-used bloodculture media

Brewers thioglycollate (Southern Group) £33.00*Thiol broth (Difco) £52-80Fastidious anaerobe broth (Lab M) £31-00Cooked-meat broth £22-29

*Costs exclusive of VAT and carriage.From Ganguli et all'

969

WANTED AND UNWANTED CONSTITUENTS OFBLOOD CULTURE MEDIAHowever good the culture medium and the incuba-tion conditions, the recovery of anaerobes may stillbe jeopardised by the presence of inhibitory sub-stances in media. The most important is probablyoxygen, hence the need for rapid transport of sam-ples of the laboratory, for media containing reducingsystems and for an anaerobic incubation environ-ment. Many patients are receiving antibiotic therapywhen blood cultures are taken and many sampleswill, therefore, contain antibacterials and/or theirmetabolites which may inhibit the growth ofanaerobes. At present it is not possible to inactivateall such agents in routine tests; this is especiallyimportant in the case of metronidazole.Some culture media contain thiol compounds or

thioglycollate, which may be toxic for anaerobes.'6However, their continuing widespread use wouldsuggest that problems of toxicity are not perceivedto be important in practice. Moreover, substancesthat are added to culture media to enhance recoveryof some organisms have been considered to beinhibitory to others-for example, liquoid (sodiumpolyethanolsulphonate) is widely used in culturemedia but inhibits the growth of anaerobic cocci,notably Peptostreptococcus anaerobius, in vitro. Yetthere is little evidence from "in-use" studies thatthis substance, when added to blood culture media,affects recovery of this organism from patients-indeed a recent study showed that better recoverywas obtained from media containing liquoid thatfrom media containing a supposedly "non-toxiC'derivative, sodium amylosulphate.'7

DETECTION OF GROWTH, SUBCULTURE ANDINTERPRETATIONOnce a culture has been incubated, the laboratorymust be able to detect bacterial growth. The special-ised techniques such as gas-liquid chromotogra-phy,14CO2 detection and impedance measurementswill be considered by Dr Johnston. Although help-ful, such techniques can only show that anaerobesare present; they cannot identify the organism con-cerned. Most laboratories rely upon examination ofcultures by Gram-stained smears and subculture tosolid media. Subculture techniques may be"blind"-that is, after fixed periods of incubationsuch as 24 or 48 h-or may be performed if culturesappear turbid on visual inspection. If the lattercourse is adopted, care must be taken, because, onthe one hand, some media such as cooked-meatbroth are innately turbid whereas, on the other, sus-pensions of some anaerobes (notably anaerobiccocci) may appear relatively non-turbid. Subcultureshould always be performed onto fresh blood agar

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

970

media, and the seeded plates incubated in carefulanaerobic conditions for at least 48 h.Once isolated, the identity of an anaerobe and its

sensitivity pattern can be determined-but theinterpretation of the results may be more difficult.There is a tendency to regard the isolation of every

obligate anaerobe from blood cultures as significant,but the skin at the site of venepuncture may have a

predominantly anaerobic flora. This is especially so

in debilitated patients, in whom venepuncture may

be difficult and samples may be taken from sitessuch as the femoral vein. Just as for aerobic isolates,microbiologists must ask the question "pathogen or

contaminant?" and seek the answer through fullliaison with clinical colleagues.

Culture of blood for anaerobes is a complex sub-ject with large and confusing literature. The multi-plicity of types of media and culture systemdescribed can bewilder and perplex-but attentionto basic principles, coupled with careful techniquesis more important that undertaking elaborate trialsof new media. The patients' interests are best servedwhen an "anaerobe-conscious" clinician works inclose co-operation with an "anaerobically-competent laboratory".

Conventional blood culture techniquesDC SHANSONDepartment of Clinical Microbiology, St Stephen'sHospital and Westminster Medical School, London

The object of good, conventional blood culturetechniques is to detect, rapidly and reliably, the pre-sence of micro-organisms in the blood. It is an

advantage to include detailed written instructionswith each blood culture set to assist the optimal col-lection of blood cultures. At least two sets of bloodcultures should be collected before the start of anti-biotic treatment but it is rarely worthwhile to collectmore than three sets.'8 A routine blood culture setshould include, preferably, at least two differentbroths for direct inoculation of blood at the beolside.Occasionally, the use of additional media isindicated,-for example, when gonococcal bac-teraemia is suspected; in this instance, broths with-out liquoid could be included for incubation in anatmosphere of 10% CO2 in air because liquoid mayinhibit some strains of gonococci.'9

VOLUME OF BLOOD CULTUREDPublished studies have recommended increasingvolumes of blood for each set of cultures fromadults; these vary from 10 Ml,20 and 15 ml2' to atleast 20 m122 in the most recent publications. Theculture of a large volume of blood is essential for the

Gould, Duerden

optimal isolation of Gram-negative bacilli such asEscherichia coli. A large volume of blood is particu-larly necessary when a low-grade bacteraemia issuspected which is characteristic of patients withstreptococcal endocarditis.

ADDITION OF SODIUM POLYETHANOLSULPHONATE ("'LIQUOID") AND DILUTION OFBLOODClinical trials have demonstrated the value of includ-ing liquoid routinely in broth used for the isolationof strict aerobes and facultative anaerobes. The con-centration of liquoid recommended varies between0-025 %-0'05%. Only a few aerobic organisms areinhibited by these concentrations but they includesome strains of Neisseria spp. Liquoid is not recom-mended for inclusion in the broth used primarily forthe isolation of strict anaerobes because occasionalstrains of anaerobic cocci are inhibited.'022 23Sodium amylosulphate is less inhibitory than liquoidfor anaerobic cocci24 but it is less efficient thanliquoid in neutralising the antibacterial effects offresh human blood.

Dilution of blood by a factor of at least 1 in 30removes the antibacterial effects of fresh bloodagainst most bacteria without using liquoid,25 and a1 in 17 dilution appears satisfactory for the isolationof anaerobes from thioglycollate media.'0 However,the dilution of blood in the broth required for theoptimal recovery of bacteria varies for differenttypes of organisms even when liquoid is included. Arecent study suggests that significantly better isola-tion of Streptococcus pneumoniae is obtained with a1 in 5 dilution of blood in liquoid broth than with a 1in 10 dilution, but that more isolations of E coli areobtained with a 1 in 10 dilution than with a 1 in 5dilution of blood in liquoid broth.26 It is suggestedthat the bottles in each blood culture set shouldinclude at least two different dilutions of blood inthe broths.

MEDIA FOR AEROBESThe media used most commonly for the isolation ofaerobes from blood cultures include nutrient broth,glucose broth, tryptone soya broth, brain-heart infu-sion broth, and diphasic media. The data availablefrom clinical trials suggest that there are few majordifferences between the results obtained with thesemedia provided that the bottles do not contain avacuum at the time of inoculation. If the media arenot vented there may be poor recovery of strictlyaerobic organisms including Pseudomonas aerugin-osa and Candida albicans.27 Glucose 0-1% is oftenadded to nutrient broth for blood cultures. If muchhigher concentrations of glucose are used, the organ-isms may be killed soon after the start of growth by

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

Blood culture-current state and future prospects

the acid produced by fermentation. However, thereare no published reports that have investigated thevalue of adding glucose 0 1% to broth for bloodculture. In a recent clinical study of patients under-going dental extractions there was no significantadvantage in adding glucose 0a 1% to nutrient brothfor the detection of bacteraemia due to viridansstreptococci.28 Diphasic media save a great deal oflabour during subcultures and are also associatedwith a reduced laboratory contamination rate duringprolonged incubation periods such as those requiredfor the isolation of Brucella spp. Nonetheless, thebroths of diphasic media require subculture on toblood agar at least once during the first 48 h ofincubation as some strains of streptococci,haemophili and neisseriae might not produceeasily-seen colonies on the agar in the diphasic bot-tle.2' If only a small volume of blood is available forculture it is probably best to inoculate a broth thatwill support the growth of both aerobes andanaerobes such as fastidious anaerobe broth.

AEROBIC CO2 ATMOSPHERE FOR BROTHCULTUREIdeally, one of the aerobic bottles in the blood cul-ture set should be incubated in air with CO2 10%

and the CO2-10-ml-liquoid nutrient broth men-tioned in the ACP Broadsheet is recommended.2'Many of these bottles can be conveniently incubatedwith the caps loose in a CO2 incubator. I have seenextremely carboxyphilic strains of Haemophilusaphrophilus and Str mutans isolated from thisCO2-broth long before other broths in the setbecame positive.

MEDIA FOR ANAEROBESBlood culture media vary considerably in their abil-ity to yield the early growth of strict anaerobes.'0Cooked-meat blood culture media that were com-

971

mercially available in Britain in 1973 were associ-ated with the late isolation of non-sporinganaerobes, although the viability of the organismswas good once growth had started.0'0" In contrastcertain thioglycollate media, including Brewer'sthioglycollate from Southern Group Laboratoriesgave more isolations of anaerobes at an early stageof incubation. The most abundant growth of Bac-teroides fragilis occurred with USP thioglycollate butthe organisms were quickly killed after growth hadstarted in this medium. A mixture of glucose digestbroth and cooked meat supported the early growthand sustained viability of anaerobes.'° During thelast few years there has been much improvement inthe range of good anaerobic media available forblood culture, including fresh home-madecooked-meat mixed with brain-heart infusionbroth,'3 fastidious anaerobe broth,'4 home-madethioglycollate medium containing Panmede2 andbrain-heart infusion broth supplemented with cys-teine.30 Recently, we have shown that brain-heartinfusion broth supplemented with both cysteine andPanmede is excellent for the growth of fastidiousnon-sporing anaerobes in simulated blood cul-tures.3' A recent study showed that the performanceof anaerobic media deteriorated with storage,32 andit is important to ensure that media are not kept onthe wards for many weeks before use.

RECOMMENDED ROUTINE BLOOD CULTURE SETAn example of a routine blood culture set thatsatisfies many of the requirements discussed above isshown in Table 2. The total volume of blood cul-tured from an adult in this set is 15 ml; there is a

range of dilutions of blood of from 1 in 3 to 1 in 18;the broth used for the isolation of strict anaerobesdoes not include liquoid, and the CO2 broth is forstrict anaerobes and carboxyphilic organisms.

Table 2 A recommended blood culture set

Volume of Liquoid Volume of blood Incubation Organismsmedium (%o) added (dilution)

1 Carbon Nutrient 10 ml 0-05 5 ml (1 in 3) Air plus Strict aerobes*dioxide brotht 10% carbon and facultativebottle dioxide anaerobes*

2 Aerobic Glucose (0.1%) 50 ml 0-05 5 ml (1 in 11) Air Strict aerobesbottle broth and facultative(diphasic anaerobesmedium)

3 Anaerobic Fastidious 75 ml - 5 ml (1 in 16) Cap tight Strict anaerobesbottle anaerobe broth and facultative

anaerobes

*Including carboxyphilic organisms such as Brucella abortus and Haemophilus aphrophilus.tPrepared according to the method given in ACP Broadsheet 81.

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

972

REMOVAL OF ANTIBACTERIAL EFFECTS DUE TOANTIBIOTICSWhen the patient is receiving antibiotics at the timeof collection of the blood culture, a useful reductionof the antibacterial effect will be obtained by dilut-ing the blood by a factor of at least 1 in 10. Broadspectrum 83-lactamase may be added to one of thebroths in the set when appropriate, but this carriesan increased risk of contamination33 and the,8-lactamase should be tested for sterility simultane-ously in another broth not containing blood. Mostauthorities do not recommend the routine additionof sucrose to isolate L-forms. Some constituents ofmedia-for example, Thiol, may antagonise certainantibiotics.'2 34 Conflicting results have beenobtained in recent studies with an ion-resin devicefor the removal of antimicrobials which has beenintroduced in the USA.35

DURATION OF INCUBATION OF THE BROTHSCommon pathogens such as E coli and Staphylococ-cus aureus are usually isolated from the blood ofpatients with septicaemia during the first 2 or 3 daysof incubationn of the broth. In practice, it is notworthwhile incubating the broths for longer thanone week when the clinical diagnosis is sep-

ticaemia.36 If the patient has PUO or clinical evi-dence of infective endocarditis, the broths should beincubated for at least three weeks because occasion-ally important pathogens such as H aphrophilus,Brucella abortus and Eikenella corrodens are iso-lated only after prolonged incubation. When brucel-losis is strongly suspected clinically, diphasic mediashould be incubated in a CO2 atmosphere for up tosix weeks.

DETECTION OF A POSITIVE BLOOD CULTUREThe three main methods for the rapid detection ofpositive blood cultures remain daily macroscopicinspection of the broths, immediate Gram-stainedsmears of suspect broths that look positive macros-

copically and routine subcultures of all blood cul-tures within 24 h of collection.' Work is in progressto evaluate the usefulness of routine subculturesafter incubation for 6 h. Gas-liquid chromatographyhas been disappointing for detecting positive cul-tures earlier than the conventional methods. Othertests for the recognition of specific microbial anti-gens such as the CIE test for Klebsiella antigen," or

rapid agglutination tests for detecting salmonellae ina broth, are only occasionally applicable.When colonies are seen in the broth culture, it is

preferable to transfer some of these with a sterilePasteur pipette to a sterile screw-capped vial forpreparation of Gram-stained smears, subculturesand other investigations. Anaerobic broths under

Gould, Duerden

vacuum are best subcultured with a syringe and nee-dle to help maintain the vacuum. Whichevermethods are used for subculture every effort mustbe made to avoid laboratory contamination (seeProfessor Collee's paper). A second routine subcul-ture is sometimes advocated after 48 h and this ispracticable when diphasic media are used. A finalsubculture is recommended before the broths arediscarded although few further positive cultures aredetected at this stage.

Fresh blood agar is recommended for subcultureand all the plates should be incubated in an atmos-phere containing 10% CO2 for at least 48 h. In prac-tice, anaerobes are rarely isolated today, possiblydue to the widespread clinical use of drugs such asmetronidazole, and it has been suggested that onlyaerobic incubation is necessary for routine subcul-ture plates. However, an anaerobic plate is also use-ful for interpretation of the occasional query aboutplate contamination that may arise.

ISOLATION OF DIFFICULT ORGANISMS ANDFURTHER INVESTIGATIONSIf a broth is suspected of being positive and theroutine subculture is negative, further proceduresmay be indicated for the isolation of fastidiousorganisms-for example, pyridoxal dependentstrains of viridans streptococci, Mycoplasmahominis, Campylobacter jejuni and Legionellapneumophila. The need for appropriate further sub-cultures depends upon the results of Gram-stainedsmears of the broth and the clinical circumstances.If, for example, the patient has the clinical signs ofendocarditis and Gram-positive cocci are seen in thebroth, the blood agar subculture plates should havea streak of S aureus applied after inoculation of thebroth so that satelliting streptococci can be isolatednear the streak after incubation. Rapid identificationand antibiotic sensitivity tests on the isolates areessential. These tests are discussed in detail else-where.' 213438

INTERPRETATION OF ISOLATESOccasionally, the clinical significance of an isolate isuncertain but, as a general rule, any organism that isisolated from two or more different blood culturesets should be considered significant. Some organ-isms previously considered to be contaminants-forexample, S epidermidis, have been recognisedincreasingly as causes of septicaemia in patientsreceiving intravenous therapy. Viridans streptococcimay have been mistaken occasionally for diph-theroids in the Gram-stained smears of broths incases of endocarditis. An occasional isolate of a viri-dans streptococcus may be difficult to interpret. Asimple serum fluorescent antibody test using the

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

Blood culture-current state and future prospects

patient's streptococcus will almost always show atitre of 400 or greater when endocarditis is pres-ent.39 A titre of <100 is found with this test whenthe blood culture isolate represents only a transientbacteraemia from the mouth.Whenever a positive blood culture is detected the

clinical microbiologist will discuss the findingsimmediately with colleagues, preferably at the bed-side of the patient. In this way, most isolates can besatisfactorily interpreted.

Rapid methods for the diagnosis ofbacteraenia HH JOHNSTONPublic Health Laboratory, John Radcliffe Hospital,Oxford, OX3 9DU

Several rapid and automated techniques have beenused for the early detection of bacteraemia. Thesecomprise: (i) immediate tests for bacterial antigen orother chemical indicator substances; (ii) concentra-tion techniques to increase the sensitivity of existingmethods; and (iii) the use of a detector system forautomated or semiautomated monitoring oflaboratory culture systems.

IMMEDIATE TESTSThe main merit of tests for bacterial products isspeed, but their sensitivity and specificity are invari-ably lower than those of conventional culture. Clini-cal evaluation has concentrated on three types oftest-Limulus Amoebocyte Lysate (LAL), GasLiquid Chromatography (GLC) and Counter-current Immunoelectrophoresis (CIE).LAL is a very sensitive test for endotoxin and is

highly specific. Sensitivity is reduced by inhibitorysubstances in serum and plasma. In clinical use,results are conflicting and the sensitivity varies bet-ween 75% and 10% when compared with positiveculture from Gram-negative organisms. False posi-tive results are also seen and the prognostic value ofa positive LAL result is uncertain.GLC is a means of detecting volatile products of

bacterial metabolism or portions of the bacteriathemselves in blood. Clinical experience is limitedbut, according to some reports, 100% detection ofbacteraemias caused by Streptococcus pneumoniae,Mycobacterium tuberculosis and certain Gram-negative bacilli can be achieved. Other workers havehad less success but most reports agree that yeastinfections are readily detected. The test is per-formed on serum and is rapid. Further extensiveinvestigation will be needed to establish whetherGLC is suitable for the routine laboratory.CIE offers the possibility of detecting and ident-

ifying organisms causing bacteraemia. Unlike the

973

other rapid techniques, its usefulness is limited tothe detection of specific micro-organisms. Workwith this technique has concentrated uponHaemophilus influenzae, Str pneumoniae and Neis-seria meningitidis. Sensitivity of CIE is low andrarely approaches 50%. Important limiting factorsare the requirement for an antiserum of high titreand the lack of specificity of many bacterial antigens.

CONCENTRATION TECHNIQUESIn normal laboratory practice, substances that mayinhibit bacterial growth are either specifically neut-ralised or diluted out. The recent introduction of anion-exchange resin cocktail for the removal of anti-microbials is an example of such an approach. Filtra-tion or centrifugation offer the possibility of remov-ing antimicrobial agents and, at the same time, con-centrating organisms sufficiently to permit immedi-ate or early detection. In the face of such obviousattractions, it must be assumed that most workersregard these methods as inconvenient or cumber-some and it is self-evident that concentration shouldbe carried out as soon as possible after collection.

Filtration involves the use of membranes thatretain bacteria. There is an extensive literature,much of it concerned with technical problems thatinclude problems in filter design and optimisation ofthe lysing system. Blood cells are removed either bydifferential filtration, with variable success, or bylysis, which is often toxic for bacteria. Clogging offilters results in long processing times for small vol-umes of blood. I found that the filtration equipmentfor blood culture marketed by the market leader inmembrane filtration was of no practical use for thesereasons. However, recent reports suggest that cur-rent techniques are more successful. The crucial fac-tors appear to be the choice of lytic agent and care-ful control of pH. "Triton X 100", "Rhozyme" and"Varidase" have been used; "Triton X 100" is toxicfor bacteria and must be used only at low concentra-tions.

Lysis-centrifugation is a complicated procedure inwhich anticoagulant is added, the erythrocytes arelysed and the bacteria and leucocytes are cen-trifuged on a density layer of sucrose-gelatin. Inevit-ably, contamination is a problem.

It is difficult to assess the effectiveness of concent-ration procedures. They appear to work in somehands, but objective analysis casts doubts on claimsthat in the clinical laboratory these methods areeither more sensitive or more rapid that a carefullyoperated conventional culture technique.

AUTOMATED METHODSIt is incorrect to describe the currently available

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

974

automated techniques for blood culture monitoringas rapid, because all depend upon the detection ofbacterial growth which is not accelerated by the pro-cedure. Several detector systems have been investi-gated with a view to automation. Some-for exam-ple, bioluminescence and microcalorimetry, appearto have no relevance in the clinical laboratory, buttwo methods require serious consideration-radiometry and impedence-conductance.Radiometry involves the radiometric detection of'4CO2 produced by bacterial metabolism of 14Clabelled substrates and is available commercially asthe BactecR system. The instrument samples headspace gas above a broth culture to which thepatient's blood has been added. Any labelled gasevolved is deemed to be due to bacterial growth andthe culture is declared "provisionally positive" whena predetermined 14C value (the "growth index") isachieved. Cultures are incubated independently ofthe detection instrument and have to be placed in itfor reading. The instrument is a substitute for thedaily visual inspection performed in most clinicallaboratories. Results with the Bactec appear to beequivalent to those obtained by conventionalmethods. In a recent study we found that the rate ofpositivity detected by Bactec (7%) was equal to thatdetected by routine culture for 2000 blood speci-mens. No organism was missed by the Bactec systemand there was an enhanced recovery of non-sporinganaerobes and yeasts. The use of the instrument wasjustified economically and it was popular withlaboratory staff. There are some controversialaspects of the use of Bactec, particularly therequirement for blind subcultures. Published evi-dence suggests that it is necessary but the results ofthe current study do not support the recommenda-tion. There is no evidence to support claims ofincreased speed or sensitivity for this system in com-parison with conventional culture, but its conveni-ence is undeniable.Impedance-conductance Growing bacteria producealterations in the flow of alternating current in theculture medium as a result of metabolism of subs-trates. Workers in the UK and USA have shownthat electrical measurements with sensitive instru-ments indicate bacterial growth and can be used todetect growth in blood-culture systems. The mainproblems have been the sensitivity and stability ofthe instrumentation. However, these have beenlargely overcome and practical, commercially pro-duced systems will soon be available. Cultures willbe on-line and continuously monitored which shouldprovide the earliest growth detection in a systemthat has the required sensitivity. It is to be hopedthat the requirement for blind subculture will beeliminated.

Gould, Duerden

CONCLUSIONSRapid methods and automation already offeracceptable alternatives to conventional culture butthe advantages are marginal. The greatest benefit isin convenience, a matter of considerable significancein a stressed routine laboratory. Rapid methods areonly relevant in specific situations such as epidemics,but automated systems are of more general interest.

COMMENTIn the course of opening the general discussion, DrCED Taylor (Public Health Laboratory, Cam-bridge) reported favourably on a trial being per-formed in Cambridge of equipment manufacturedby Malthus Instruments (now part of the JohnsonMatthey group) for measuring electrical conductiv-ity in blood cultures. As well as allowing rapid detec-tion of bacterial growth, the use of this equipmentobviates the need for frequent subculturing of bloodcultures. The equipment also has potential value forrapid bacterial identification and antimicrobial sus-ceptibility testing.

Bactenogenic shock I MCA LEDINGHAMDepartment of Surgery, Western Infirmary, Glasgow

The possibility of bacteriogenic shock should beconsidered whenever cardiovascular instabilityoccurs in a patient with major infection. The fullydeveloped syndrome leaves little doubt as to diag-nosis and the attendant mortality is high. Treatment,if it is to be successful, must be instituted in thecourse of the condition, often before the diagnosiscan be confirmed. Fortunately, acute resuscitation ofa patient suffering from any form of shock isinfluenced more by the nature of the associatedphysiological disturbance than by specific aetiologi-cal factors. Success of subsequent treatment, how-ever, is largely dependent on detection of the under-lying cause.These observations are particularly relevant in

bacteriogenic shock, the immediate management ofwhich may necessitate the simultaneous correctionof hypovolaemia, hypoxaemia and cardiac failure inaddition to the administration of antibiotics to com-bat bacteraemia. Thereafater, rapid identification ofthe source of sepsis is essential and, in manyinstances, surgical intervention will be required todeal with a focus of infection. The ultimate fate ofthe patient is determined by the clinical course ofthe underlying pathological process, the integrity ofthe host defence mechanisms and the presence ofassociated disease.

PATHOPHYSIOLOGICAL MECHANISMSThe pathophysiological basis of bacteriogenic shock

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

Blood culture-current state and future prospects

is a complex mixture of factors, only one of which isthe presence of infection. Shock occurs in associa-tion with both Gram-positive and Gram-negativeinfections-less frequently in the former (approx 1in 10 patients) than in the latter (approx 1 in 3patients). In the past it was believed possible to dis-criminate between Gram-positive and Gram-negative septic shock on clinical grounds, but recentevidence makes it clear that this is untrue. In a studyof 59 patients, Wiles et al40 were unable to detect adifference in any physiological variable betweeninfections with different organism groups, or bet-ween specific organisms. After initial resuscitation,all patients exhibited a hyperdynamic cardiovascularresponse with abnormal vascular tone; somemyocardial depression was evident in most patients.Mortality was >70% in both Gram-positive andGram-negative septicaemia. They concluded thatwhile the pathogenesis of the septic responseremained unclear, it did not seem to be peculiar to aspecific micro-organism. This supports the conten-tion that the fate of patients suffering from bac-teriogenic shock is determined by the speed andadequacy of initial resuscitation and, thereafter, bythe nature of the underlying disease process.

In the early stages of bacteriogenic shock (thehyperdynamic phase), cardiac output is often highand the skin warm and dry. Urine output may beraised and some patients may exhibit an inappropri-ate polyuria at the expense of effective circulatingplasma volume. These effects are related to the pre-sence in the circulation of pyrogens and to therelease by bacterial products of plasma kinins, his-tamine and prostaglandins of the E series.4' Theseagents may later contribute (by direct effects oncapillary permeability) to the transudation of fluidfrom capillaries. The primary respiratory response ishyperventilation, mediated by both central andperipheral mechanisms, and arterial blood gasanalysis may reveal only some reduction in Pco2.Metabolic acidosis in bacteriogenic shock is morevariable than in other forms of shock and, therefore,less reliable as a prognostic indicator.42

In the later stages of bacteriogenic shock (thehypodynamic phase), cardiac output falls and thereis marked hypotension, peripheral vasoconstrictionand oliguria. These effects are almost certainlycaused by the release of catecholamines andangiotensin II. More recently discovered substancessuch as thromboxane A2, derived from damageplatelets, may augment this reaction particularly inthe pulmonary circulation43 and contribute to adeterioration in pulmonary gas exchange with aprogressive fall in arterial P02. Metabolic distur-bances appear earlier in bacteriogenic shock than inother forms of shock. Characteristically, the

975

arteriovenous oxygen difference falls and, in somepatients, oxygen consumption may be very low."The most likely explanation for this observation isimpaired cell metabolism. This metabolic defectinterferes with capillary and cell membrane functionand prevents the cells from utilising oxygen andenergy substrates properly to form ATP. Lipolysisoccurs with high circulating levels of free fatty acidsand serum triglycerides.45 In muscle there is a blockto the intracellular oxidation of glucose, whichtogether with an inability to utilise ketone bodiesleads to increased catabolism of branched chainamino acids. The disturbance in carbohydratemetabolism may be the result of inhibition of certainreactions in the citric acid cycle which may accountfor the lower levels of lactate in bacteriogenic shockthan would otherwise be expected. Hypophos-phataemia and hypocalcaemia are also common.Many of the haemodynamic and metabolic effects

that have been described may be attributable to cir-culating endotoxin-the complex lipopolysaccharidecoating of Gram-negative bacteria-and in particu-lar to the lipid-A moiety. Endotoxin causes aggrega-tion of platelets by direct damage to the cell wall,and of white cells by an indirect action via activatedcomplement. Platelet aggregation leads to the for-mation of thrombi in the microcirculation at a ratethat may swamp fibrinolytic and reticuloendothelialphagocytic mechanisms. White cell aggregation mayalso cause plugging of the microvasculature withdamage of the capillary endothelium resulting fromdirect contact with toxic oxygen radicals, such assuperoxide anion and hydrogen peroxide, releasedfrom the leucocytes.4647 These microcirculatory dis-turbances may be observed widely throughout thebody and may result in focal necrosis in the lung,kidneys, liver and other organs.

In the lung, damage to the capillary endotheliumleads to the development of protein-rich pulmonaryoedema which, if not arrested early, may progress tothe fully developed adult respiratory distress syn-drome. In the kidney, endotoxaemia may cause analready reduced glomerular filtration rate to fallfurther and precipitate acute renal failure. If thesource of sepsis can be eradicated, recovery is poss-ible.

Jaundice may occur in the presence of extrahepa-tic Gram-negative infection. The jaundice is choles-tatic, typically reflected in elevation of conjugatedserum bilirubin and alkaline phosphatase, withmodest increases in serum transaminases. The his-tological appearances in the liver are those of inter-cellular and intracanalicular bile stasis with little orno evidence of parenchymal damage. The observedcholestasis in sepsis may be caused by the inhibitoryeffect of endotoxin on the active sodium transport

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

976

mechanisms of hepatocytes.4 Failure of liver func-tion abnormalities to return to normal within a fewdays of a septic insult often indicates a persistingsource of major infection.495s

Attention has once again begun to centre on therole of the hepatic Kupffer cells in the removal ofbacteria and their breakdown products, includingendotoxin, from the circulation. The activity of thesecells is inhibited during an episode of bacteriogenicshock5' and the success of treatment may well bedetermined by the degree of functional recoverywhich ensues after the phase of resuscitation.

PATIENTS AND CLINICAL COURSEMost patients who develop bacteriogenic shockbecome infected while in hospital, often followingoperations on the gastrointestinal or genitourinarysystems and particularly when surgery has been per-formed under emergency or semielective conditions.Major trauma and burns carry an increased risk ofGram-negative sepsis and bacteriogenic shock5253and the transplant patient is also more than usuallyliable to develop these complications. The non-surgical causes of bacteriogenic shock includepneumonia, endocarditis, meningitis and der-matological infections. Advanced age, diabetes,malnutrition, uraemia, malignancy and immunologi-cal defects are common associated findings, as is

treatment with chemotherapeutic and corticosteroiddrugs.

Mortality is high amongst patients who have suf-fered an episode of bacteriogenic shock but mayvary from 15% in otherwise healthy young women

after septic abortion54 to 84% in elderly patientswith faecal peritonitis.55 Improvement in techniquesof resuscitation has reduced the incidence of deathsduring the period of shock. More often, the patientsurvives the acute episode and appears to be improv-ing only to develop, perhaps several days later, a

complication such as respiratory failure, renal fail-ure or bleeding which sets the scene for multipleorgan failure. At necropsy most of these patientshave evidence of unresolved sepsis.

References

'Washington JA II. The detection of sepiicaemia. West PalmBeach, FA: CRC Press, 1979.

2 Kreger BE, Craven DE, Carling PC, McCabe WR. Gram-negative bacteremia III. Reassessment of etiology, epidemiol-ogy and ecology in 612 patients. Am J Med 1980;68:332-43.

3 Finegold SM. Anaerobic bacteria in human disease. New York:Academic Press, 1977:182-201.

4 Rogosa M, Hampp EG, Nevin TA, Wagner HN, Driscoll EJ,Baer PN. Blood sampling and cultural studies in the detectionof post-operative bacteremias. JAm Dent Assoc 1960;60:171.

Young SEJ Bacteraemia 1975-1980: a survey of cases reportedto the PHLS Communicable Disease Surveillance Centre. JInfect 1982;5:19-26.

Gould, Duerden6 Kreger BE, Craven DE, McCabe WR. Gram-negative bac-

teremia IV. Re-evaluation of clinical features and treatment.Am J Med 1980;68:344-55.

Washington JA II, Brewer ND. Collection and handling ofspecimens. In: Washington JA. Laboratory procedures in clini-cal microbiology New York: Springer-Verlag, 1981:24-5.

Shanson DC. Microbiology in clinical practice. London: Wright,1982:93-4.

9 Bisset KA, Bartlett R. The isolation and characters of L-formsand reversions of Bacillus licheniformis var endoparasiticus(Benedek) associated with the erythrocytes of clinically nor-mal persons. J Med Microbiol 1978;11:335-49.

'° Shanson DC. An experimental assessment of different anaerobicblood culture methods. J Clin Pathol 1974;27:273-9.

Forgan-Smith WR, Darrell JH. A comparison of media used invitro to isolate non-sporing Gram-negative anaerobes fromblood. J Clin Pathol 1974;27:280-3.

12 Szarwatkowski MV. A comparison of three readily availabletypes of anaerobic blood culture media. Med Lab Sci1976;23:5-12.

13 Collee JG, Duerden BI, Brown R. Recovery of anaerobic bac-teria from small inocula: a model for blood culture studies. JClin Pathol 1977;30:609-14.

14 Ganguli LA, Turton LG, Tiliotson GS. Evaluation of FastidiousAnaerobe Broth as a blood culture medium. J Clin Pathol1982;35:458-61.

Is Watt B. The influence of carbon dioxide on the growth of obli-gate anaerobes on solid media. J Med Microbiol 1973;6:307-13.

16 Dye M. Anaerobic bacteria: a study of some problems in theirisolation. University of Aberdeen: PhD thesis, 1975.

7 Tenney JH, Reller LB, Wang W-LL, Cox RL, Mirret S. Com-parative evaluation of supplemented peptone broth withsodium polyanetholesulfonate and trypticase soy broth withsodium amylosulfate for detection of septicaemia. J ClinPathol 1982;16:107-10.

18 Washington JA II. Blood cultures. Principles and techniques.Mayo Clin Proc 1975;50:91-8.

19 Staneck JL, Vincent S. Inhibition of Neisseria gonorrhoeae bysodium polyanetholesulfonate. J Clin Microbiol1981;13:463-7.

20 Hall MM, Ilstrup DM, Washington JA II. Effect of volume ofblood cultured on detection of bacteraemia. J Clin Microbiol1976;3:643-5.

21 Stokes EJ. Blood culture technique. ACP Broadsheet 81, 1974.22 Hoare RD. The suitability of 'liquoid' for use in blood culture

media, with particular reference to anaerobic streptococci. JPathol Bacteriol 1939;48:573-7.

23 Graves MH, Morello JA, Kocka FE. Sodium polyanethol sulfo-nate sensitivity of anaerobic cocci. Appl Microbiol1974;27: 1131-3.

24 Kocka FE, Arthur EJ, Searcy RL. Comparative effects of twosulfated polyanions used in blood culture on anaerobic cocci.Am J Clin Pathol 1974;61:25-7.

2S Roome AP, Tozer RA. Effect of dilution on the growth of bac-teria from blood cultures. J Clin Pathol 1968;21:719-21.

26 Auckenthaler R, Ilstrup DM, Washington JA II. Comparison ofrecovery of organisms from blood cultures diluted 10%(volume/volume) and 20% (volume/volume). J Clin Microbiol1982;15:860-4.

27 Blazevic DJ, Stemper JE, Matsen JM. Effect of aerobic andanaerobic atmospheres on isolation of organisms from bloodcultures.J Clin Microbiol 1975;1:154.

26 Shanson DC, Akash S, Harris M. Clinical comparison of glucosebroth with nutrient broth blood cultures for the detection of'Strep. viridans' bacteraemia. J Med Microbiol 1983;16:(in press).

29 Hunt GH, Price EH. Comparison of a homemade blood culturebroth containing a papain digest of liver, with four commer-cially available media for the isolation of anaerobes from simu-

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from

Blood culture-current state and future prospects

lated paediatric blood cultures.J Clin Pathol 1982;35: 1142-9.30 Shanson DC, Singh J. Effect of adding cysteine to brain-heart

infusion broth on the isolation of Bacteroides fragilis fromexperimental blood cultures. J Clin Pathol 1981 ;34:221-3.

31 Shanson DC, Pratt J. The effect of adding a papain digest of oxliver to brain-heart infusion cysteine broth on the recovery ofnon-sporing anaerobes from simulated blood cultures. J ClinPathol 1983; (in press).

32 Barr JG. A cooked-meat blood culture medium; shelf-life andclinical evaluation compared with other systems. J Infect1980;2:247-58.

33 Norden CW. Pseudosepticaemia. Ann Intern Med 1969;71:789.34 Shanson DC. Blood culture techniques. In: William JD, ed.

Modern topics in infection London: Heineman 1978:20.35 Appleman MD, Swinney RS, Heseltine PNR. Evaluation of the

antibiotic removal device. J Clin Microbiol 1982;15:278-81.36 Effersoe P. The importance of the duration of incubation in the

investigation of blood cultures. Acta Pathol Microbiol Scand1965;65:129.

Simpson RA, Speller DCE. Detection of bacteraemia by coun-tercurrent immunoelectrophoresis. Lancet 1977;i:206.

3 Washington JA II. The role of the microbiology laboratory in thediagnosis and antimicrobial treatment of infective endocar-ditis. Mayo Clin Proc 1982;57:22-32.

3 Shanson DC, Hince C. An immunofluorescent method fordetecting antibodies against viridans streptococci in Strep-tococcus viridans endocarditis. J Clin Pathol 1978;31:292.

40 Wiles JB, Cerra FB, Siegel JH, Border HR. The systemic septicresponse: does the organism matter? Crit Care Med1980;8:55-60.

Ledingham I McA, Routh GS. The pathophysiology of shock. BrJ Hosp Med 1979;2:472-82.

42 Burns HJG, Cowan BN, Ledingham I McA. Metabolic acidosisin the critically ill. In: Metabolic acidosis London: Pitman,Ciba Found Symp 1982;87:293-300.

43 Parratt JR, Coker SJ, Hughes B, et al. In: McConn R, ed. Thepossible role of prostaglandins and thromboxanes in the pul-monary consequences of experimental endotoxin shock andclinical sepsis. Raven Press: (in press).

"Wilson RF, Gibson BS. The use of arterial-central venousoxygen differences to calculate cardiac output and oxygen con-

977

sumption in critically ill surgical patients. Surgery1978;84:362-9.

45 Wardle N. Shock: Bacteraemic and endotoxic shock. Br J HospMed 1979;1:223-31.

4Hammerschmidt DE, Weaver LI, Hudson LD, Craddock PR,Jacob HS. Association of complement activation and elevatedplasma C5a with adult respiratory distress syndrome. Lancet1980;i:947-9.

4HJacob HS, Craddock PR, Hammerschmidt DE, Moldow CF.Complement-induced granulocyte aggregation: an unexpectedmechanism of disease. N Engl J Med 1980;302:789-94.

"Utili R, Abernethy CO, Zimmerman MJ. Studies on the effectsof E coli endotoxin and canalicular bile formation in the iso-lated perfused rat liver. J Lab Clin Med 1977;89:471-82.

"Norton L, Moore G, Eiseman B. Liver failure in the post-operative patient: The role of sepsis and immunologic defi-ciency. Surgery 1975;78:6-13.

5 Royle GT, Kettlewell MGW. Liver function tests in surgicalinfection and malnutrition. Ann Surg 1980;192:192-4.

5' Scovill WA, Saba TM, Blumenstock FA, Bernard H, Powers SR.Opsonic a2 surface binding glycoprotein therapy during sepsis.Ann Surg 1978;188:521-9.

52 Miller RM, Polakavetz SH, Hornick RB, Cowley RA. Analysisof infection acquired by the severely injured patient. SurgGynecol Obstet 1973;137:7-10.

53 Heimbach DM. Burn shock. In: Ledingham I McA. Shock: clini-cal and experimental aspects Amsterdam: Elsevier,1976:211-30.

Morris JA. Bacteremic shock in obstetrics. In Marcus SL, MarcusCC, eds. Advances in obstetrics and gynecology, vol 1, Balti-more: Wilkins and Wilkins, 1967.

Neely WA, Berry DW, Rushton FW, Hardy JD. Septic shock.Clinical physiological and pathological survey of 244 patients.Ann Surg 1971;173:657-66.

Request for reprints to: Professor BI Duerden, Depart-ment of Medical Microbiology, The University of SheffieldMedical School, Beech Hill Road, Sheffield S1O 2RX,England.

copyright. on M

ay 10, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.36.9.963 on 1 S

eptember 1983. D

ownloaded from