br. j. anaesth.-2012-walsh-63-71

9
Multicentre cohort study of red blood cell use for revision hip arthroplasty and factors associated with greater risk of allogeneic blood transfusion T. S. Walsh 1 * , J. Palmer 2 , D. Watson 2 , K. Biggin 2 , M. Seretny 1 , H. Davidson 2 , M. Harkness 2 and A. Hay 1 1 Anaesthetics and Critical Care, Edinburgh Royal Infirmary, Little France Crescent, Edinburgh EH16 2SA, UK 2 Clinical Effectiveness Group, Scottish National Blood Transfusion Service, 21 Ellen’s Glen Road, Edinburgh EH17 7QT, UK * Corresponding author. E-mail: [email protected] Editor’s key points Factors associated with blood transfusion in patients undergoing revision hip arthroplasty were studied. Data from 210 patients from 11 hospitals were analysed. Preoperative haemoglobin, body weight, intraoperative loss, and use of cell saver independently predicted the need for transfusion. Importantly, the use of cell saver reduced the need for allogeneic blood transfusion. Background. Revision hip arthroplasty (RHA) is associated with high rates of allogeneic blood transfusion (ABT). We aimed to determine factors associated with ABT in patients undergoing RHA in Scottish hospitals, with particular focus on perioperative cell salvage (PCS). Methods. A prospective observational cohort study of RHA procedures performed in 11 hospitals over 7 months was performed. We recorded predefined patient, surgery- related, and blood conservation factors that may influence perioperative ABT, together with postoperative haemoglobin (Hb) data and ABTs to day 7. We explored factors with strongest independent association with ABT during the perioperative period using multiple regression analysis. Results. Two hundred and ten cases were studied, of whom 58% received ABTs (mean 1.8 units), most of which (52%) occurred on the day of surgery. Eighty-eight (42%) patients received PCS, of whom 68 had red cells re-infused [mean re-infusion volume 312 ml (1st, 3rd quartile: 260, 363 ml)]. In unadjusted comparisons, patients receiving PCS had lower intraoperative (9% vs 40%) and total (55% vs 63%) exposure to ABTs. The mean (95% confidence interval) theatre blood loss was 1013 (899–1128) ml and was higher for combined femoral/acetabular revision and femoral revision than other categories. The mean postoperative Hb transfusion trigger was 80 g litre 21 . In multivariable models, preoperative Hb [odds ratio (OR) 0.35; P,0.001], patient weight (OR 0.96; P¼0.004), operating theatre blood loss (OR 1.002; P,0.001), and re-infusion of PCS blood (OR 0.31; P¼0.02) were independent predictors of ABT exposure. Conclusions. PCS is an effective blood conservation strategy for RHA, especially for patients with preoperative anaemia, low body weight, or both. Keywords: blood transfusion; revision hip arthroplasty Accepted for publication: 24 August 2011 Orthopaedic surgery utilizes a high proportion of total red blood cell (RBC) supplies. 1 For primary arthroplasty procedures, RBC use is typically greatest for hip arthroplasty compared with other procedures, 2 but wide variation in RBC transfusion require- ments has been described between centres and individual surgeons. 3 4 Despite the publication of clinical transfusion guidelines, there appears to be widespread variation in beliefs regarding safe transfusion practice, including between anaes- thetists. 5 As the number of patients undergoing arthroplasty increases, and larger numbers of patients live longer, the number of revision hip arthroplasty (RHA) procedures per- formed has increased significantly in most health-care systems. Published studies indicate that patients undergoing RHA frequently require RBC transfusions at higher rates than the other high-volume orthopaedic procedures. 26 There has been increased interest in perioperative blood conservation for orthopaedic surgery in recent years, and lower RBC use for surgery is increasingly considered a quality marker. 67 Drivers for blood conservation include the greater cost of RBCs, reduced donation rates, concerns regarding the risks of receiving allogeneic RBCs, and patient preference to avoid receiving blood transfusions. Blood con- servation strategies include preoperative autologous donation, preoperative erythropoietin therapy, perioperative red cell salvage, and antifibrinolytic drugs. 67 Of these, sys- tematic reviews suggest that preoperative autologous donation is of limited value, 8 and although erythropoietin is clinically effective, its cost-effectiveness and safety have not been conclusively established. 6 7 There are few high- quality randomized trials of cell salvage or antifibrinolytic British Journal of Anaesthesia 108 (1): 63–71 (2012) Advance Access publication 27 October 2011 . doi:10.1093/bja/aer326 & The Author [2011]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected] by guest on April 12, 2015 http://bja.oxfordjournals.org/ Downloaded from

Upload: catia-lopes

Post on 13-Nov-2015

214 views

Category:

Documents


0 download

DESCRIPTION

CEC

TRANSCRIPT

  • Multicentre cohort study of red blood cell use for revisionhip arthroplasty and factors associated with greaterrisk of allogeneic blood transfusionT. S. Walsh1*, J. Palmer2, D. Watson2, K. Biggin2, M. Seretny1, H. Davidson2, M. Harkness2 and A. Hay11 Anaesthetics and Critical Care, Edinburgh Royal Infirmary, Little France Crescent, Edinburgh EH16 2SA, UK2 Clinical Effectiveness Group, Scottish National Blood Transfusion Service, 21 Ellens Glen Road, Edinburgh EH17 7QT, UK

    * Corresponding author. E-mail: [email protected]

    Editors key points

    Factors associated withblood transfusion inpatients undergoingrevision hip arthroplastywere studied.

    Data from 210 patientsfrom 11 hospitals wereanalysed.

    Preoperativehaemoglobin, bodyweight, intraoperativeloss, and use of cell saverindependently predictedthe need for transfusion.

    Importantly, the use ofcell saver reduced theneed for allogeneic bloodtransfusion.

    Background. Revision hip arthroplasty (RHA) is associated with high rates of allogeneicblood transfusion (ABT). We aimed to determine factors associated with ABT in patientsundergoing RHA in Scottish hospitals, with particular focus on perioperative cellsalvage (PCS).

    Methods. A prospective observational cohort study of RHA procedures performed in11 hospitals over 7 months was performed. We recorded predefined patient, surgery-related, and blood conservation factors that may influence perioperative ABT, togetherwith postoperative haemoglobin (Hb) data and ABTs to day 7. We explored factors withstrongest independent association with ABT during the perioperative period usingmultiple regression analysis.

    Results. Two hundred and ten cases were studied, of whom 58% received ABTs (mean 1.8units), most of which (52%) occurred on the day of surgery. Eighty-eight (42%) patientsreceived PCS, of whom 68 had red cells re-infused [mean re-infusion volume 312 ml (1st,3rd quartile: 260, 363 ml)]. In unadjusted comparisons, patients receiving PCS had lowerintraoperative (9% vs 40%) and total (55% vs 63%) exposure to ABTs. The mean (95%confidence interval) theatre blood loss was 1013 (8991128) ml and was higher forcombined femoral/acetabular revision and femoral revision than other categories. Themean postoperative Hb transfusion trigger was 80 g litre21. In multivariable models,preoperative Hb [odds ratio (OR) 0.35; P,0.001], patient weight (OR 0.96; P0.004),operating theatre blood loss (OR 1.002; P,0.001), and re-infusion of PCS blood (OR 0.31;P0.02) were independent predictors of ABT exposure.Conclusions. PCS is an effective blood conservation strategy for RHA, especially for patientswith preoperative anaemia, low body weight, or both.

    Keywords: blood transfusion; revision hip arthroplasty

    Accepted for publication: 24 August 2011

    Orthopaedic surgery utilizes a high proportion of total redblood cell (RBC) supplies.1 For primary arthroplasty procedures,RBC use is typically greatest for hip arthroplasty comparedwithotherprocedures,2 butwidevariation inRBCtransfusion require-ments has been described between centres and individualsurgeons.3 4 Despite the publication of clinical transfusionguidelines, there appears to be widespread variation in beliefsregarding safe transfusion practice, including between anaes-thetists.5 As the number of patients undergoing arthroplastyincreases, and larger numbers of patients live longer, thenumber of revision hip arthroplasty (RHA) procedures per-formed has increased significantly in most health-caresystems. Published studies indicate that patients undergoingRHA frequently require RBC transfusions at higher rates thanthe other high-volume orthopaedic procedures.2 6

    There has been increased interest in perioperative bloodconservation for orthopaedic surgery in recent years, andlower RBC use for surgery is increasingly considered aquality marker.6 7 Drivers for blood conservation include thegreater cost of RBCs, reduced donation rates, concernsregarding the risks of receiving allogeneic RBCs, and patientpreference to avoid receiving blood transfusions. Blood con-servation strategies include preoperative autologousdonation, preoperative erythropoietin therapy, perioperativered cell salvage, and antifibrinolytic drugs.6 7 Of these, sys-tematic reviews suggest that preoperative autologousdonation is of limited value,8 and although erythropoietin isclinically effective, its cost-effectiveness and safety havenot been conclusively established.6 7 There are few high-quality randomized trials of cell salvage or antifibrinolytic

    British Journal of Anaesthesia 108 (1): 6371 (2012)Advance Access publication 27 October 2011 . doi:10.1093/bja/aer326

    & The Author [2011]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved.For Permissions, please email: [email protected]

    by guest on April 12, 2015

    http://bja.oxfordjournals.org/D

    ownloaded from

  • drugs. Systematic reviews of available evidence suggest thatthese technologies are effective for some patients under-going orthopaedic surgery, but interpretation is difficultbecause heterogeneous procedures were included, andmany studies were carried out before restrictive haemoglobin(Hb) transfusion triggers were adopted based on evidence-based guidelines.6 911

    The primary aim of this study was to explore the factorsassociated with the risk of allogeneic RBC transfusion for asingle surgical procedure with a high-risk of perioperativeRBC transfusion, RHA, in a representative sample of Scottishhospitals. We were particularly interested in the impact ofperioperative cell salvage (PCS) on the risk of receiving RBCtransfusions.

    MethodsWe used data from a previously published survey of bloodconservation in revision hip surgery,12 together with surgicalprocedure data from the Scottish Transfusion EpidemiologyDatabase (which covers all National Health Service hospitalsites in Scotland), to identify all hospitals in Scotland thatundertake revision hip surgery. No minimum number of pro-cedures was set, and some hospitals undertook ,10 pro-cedures per year. We identified 24 hospitals using thisprocess. We used available procedural data to group hospi-tals into three groups: .100 procedures per year (n2),30100 procedures per year (n10), and ,30 proceduresper year (n12).

    We planned to collect data prospectively from half of thehospitals (n12). For each of the three groups, 50% of thehospitals were chosen at random as study sites. These siteswere approached to participate in data collection; all sitesapproached agreed to collaborate. The study was reviewedby the Scottish Multicenter ethics committee, who classifiedit as audit/service evaluation. The need for individual patientconsent was waived.

    As this was a descriptive study, no formal sample size cal-culation was undertaken. The steering group agreed by con-sensus to a recruitment target of 200 cases. Assuming atransfusion rate of 5060% of cases, we expected toobtain data for 100120 patients in whom transfusionoccurred, which would allow regression modelling for1012 variables assuming one variable per 10 events ofinterest (exposure to allogeneic RBC transfusion). Based onknown case numbers in the centres, we calculated that7 months of recruitment at eligible sites would acquire 200cases assuming 100% data collection.

    Data collection

    The data fields were agreed a priori by the steering groupwith additional input from collaborating orthopaedic sur-geons and anaesthetists.

    Baseline data collection comprised relevant patientcharacteristics (gender, age), co-morbidities (hypertension,ischaemic heart disease, chronic obstructive pulmonarydisease, cerebrovascular disease, rheumatoid arthritis,

    diabetes mellitus, chronic anaemia), medication (aspirin, clo-pidogrel, warfarin, heparin, oral iron, erythropoietin), ortho-paedic history (underlying diagnosis, reason for arthroplastyrevision), and preoperative Hb. Day of surgery data comprisedthe operation performed, anaesthetic technique (general,general plus regional, regional alone), use of perioperativecell salvage, re-infusion volume (for salvaged patients), anti-fibrinolytic use, RBC transfusion [in the operating theatre,total on day of surgery], and estimated blood loss. Hb datacomprised all routinely measured Hb values up to day 7after operation. RBC use data comprised RBC units transfusedin the operating theatre, and all RBC transfusions up to day 7post-surgery. The timing of all RBC transfusions was recordedand related to measured Hb values. The use of other bloodproducts was also recorded.

    Baseline data and day of surgery data were collected byanaesthetists, surgeons or both and operating theatrenursing staff. Postoperative data were collected by transfu-sion practitioners or other dedicated staff. All data wererecorded on a case record file. Data were checked by amember of the study team and any queries resolvedagainst source data.

    Analysis

    Descriptive statistics were used to document baseline patientcharacteristics and day of surgery data. Three RBC transfu-sion outcomes were calculated: first, RBC transfusions inthe operating theatre; secondly, RBC transfusions from dis-charge from the operating theatre to 7 days after operation(or time of hospital discharge if ,7 days); thirdly, theentire perioperative period including the operating theatreto 7 days after operation. Hb data were used to describe pre-operative anaemia and rates of anaemia for postoperativedays 17. For all RBC transfusions on days 17, we calculatedthe preceding Hb value as a measure of the transfusiontrigger used.

    We dichotomized the cohort according to the use of PCSand described RBC use for these subgroups.

    Multivariable analysis

    We chose exposure to allogeneic RBCs from day 0 (day ofsurgery) to day 7 post-surgery as a binary outcome for logis-tic regression analysis. The factors included in our regressionmodelling are listed in Table 1. We developed two models:first, we examined the preoperative, surgical, and anaes-thetic factors associated with allogeneic RBC transfusion(excluding subsequent theatre blood loss and use of PCS).This model was intended to potentially inform decisionsregarding when blood conservation technologies are likelyto be most effective; secondly, we included estimated oper-ating theatre blood loss and the use of PCS to the model toexplore the relative contribution of salvage to avoidance ofallogeneic RBC transfusion.

    For all models, backward conditional binary logisticregression was used, excluding any variables with P.0.1.Statistical significance was taken as P0.05. Associations

    BJA Walsh et al.

    64

    by guest on April 12, 2015

    http://bja.oxfordjournals.org/D

    ownloaded from

  • are presented as odds ratio (OR) with 95% confidenceintervals (95% CI).

    ResultsDuring the study period, hospital records showed that therewere 210 RHAs in the 12 hospitals. Data were available forall cases. As predicted, there was wide variation in caseload,and one site did not undertake an RHA during the studyperiod (Table 2). Data were therefore only evaluated from11 hospitals.

    Patient characteristics and baseline data are shown inTable 3. The mean (95% CI) Hb concentration was higher

    for males than females [136 (133140) vs 125 (122127) glitre21], and a higher proportion of females had a preopera-tive Hb concentration of ,100 g litre21 (females 7.0% vsmales 3.1%). None of the patients underwent preoperativeautologous red cell donation or erythropoietin therapy asthese are not routinely offered in Scotland before orthopae-dic surgery.

    The surgery performed is shown in Table 4. Anaesthetictechniques used were as follows: general anaesthesiaalone 16 cases; combined general and regional technique93 cases; regional anaesthesia alone 107 cases (spinal 70;epidural 2; combined spinal/epidural 27; type: 8 not stated).Eight patients received aprotinin (procedures undertakenbefore withdrawal of the drug) and 28 received tranexamicacid; 25 of these 36 cases were at a single site, whichtreated 25 of 44 cases with antifibrinolytic drugs.

    Perioperative cell salvage data

    Perioperative cell salvage was utilized in five of the 11 hospi-tals, all of which were in the medium to high case volumegroups (Table 2). A description of cell salvage use and trans-fusion outcomes is shown in Figure 1. In total, 88 (42%) ofcases had a cell saver set up for surgery, and red cells werere-infused in 68 cases. For the 68 cases in which red cellre-infusion took place, the mean (95% CI) re-infusedvolume was 312 (260362) ml per case. The distribution ofre-infusion volumes is shown in Figure 2. Unadjusted com-parison between patients who did not receive salvaged redcells and patients who received red cell salvage showedthat those receiving cell salvage were less likely to requireallogeneic RBC transfusion during surgery (9% vs 40%) andduring the combined intra-/postoperative period (55% vs63%). Preoperative Hb concentrations for patients whoreceived cell salvage were similar to those in whom it wasnot used. Cell salvage was more frequently used for com-bined femoral/acetabular revisions than single componentor other surgical procedures (Table 4).

    Table 1 Factors included in the multivariable regressionmodelling as predictors of exposure to allogeneic RBCs. Themethod of inclusion is noted in brackets as: Cont, continuous/ordinal; Cat, categoric

    Preoperative factors Age (Cont)

    Weight (Cont)

    Preoperative Hb concentration (Cont)

    ASA risk classification (Cont)

    Gender (Cat)

    Preoperative heparin/aspirin therapy (Cat)

    Infected hip (Cat)

    Surgical/anaesthetic factors

    Anaesthetic technique (regional techniqueincluded) (Cat)

    Surgical procedure (4 categories:acetabular only, femoral only, combined,other) (Cat)

    Hospital site at which surgery undertaken(Cat)

    Cell salvage set up (whether re-infusionoccurred or not) (Cat)

    Cell salvage used and red cell re-infusionoccurred (Cat)

    Procedure relatedfactors

    Estimated operating theatre blood loss(Cont)

    Table 2 Summary of activity and case numbers for the 12 study sites. RBC, red blood cell

    Site Patients Patients transfused [n (%)] Transfusion events RBC units transfused [n (rate per case)] Cell salvage cases [n (%)]

    A 71 29 (41) 37 76 (1.1) 34 (48)

    B 18 11 (61) 18 35 (1.9) 15 (83)

    C 26 19 (73) 25 69 (2.7) 12 (46)

    D 44 31 (70) 41 83 (1.9) 22 (50)

    E 18 5 (28) 7 13 (0.7) 0

    F 4 3 (75) 4 19 (4.8) 0

    G 10 10 (100) 20 42 (2.0) 0

    H 11 8 (73) 12 26 (2.4) 5 (45)

    I 2 2 (100) 3 11 (5.5) 0

    J 5 4 (80) 7 12 (2.4) 0

    K 1 0 (0) 0 0 (0) 0

    L 0 N/A N/A N/A N/A

    Total 210 120 (57) 174 386 (1.8) 88 (42)

    Blood transfusion in revision hip arthroplasty BJA

    65

    by guest on April 12, 2015

    http://bja.oxfordjournals.org/D

    ownloaded from

  • Estimated operating theatre blood loss

    The overall mean (95% CI) estimated theatre blood loss was1013 (8991128) ml. The estimated blood loss was higherfor combined femoral/acetabular revision and femoral

    revision than the other categories (Table 4). The estimatedblood loss was relatively low for the 20 patients in whomcell salvage was set up but no red cell re-infusion occurred(median 450 ml; 1st quartile 350 ml; 3rd quartile 644 ml),suggesting that salvage may not have been feasible forthese cases.

    Allogeneic RBC transfusion

    Overall, 57.7% of patients received an RBC transfusion in theperioperative period (day of surgery up to 7 days after oper-ation). In relation to surgery, 3.8% of patients received pre-operative transfusions, 29.5% were transfused duringsurgery, and 44.7% were transfused after operation.Overall, the mean (95% CI) red cell use per case (includingnon-transfused patients) was 1.8 units (1.522.08; total386 units). The majority of RBCs (52.1%) were transfusedon the day of surgery, and 96% of RBCs were transfused bythe end of the third postoperative day. For transfusedpatients, the majority (61.5%) received their first RBC transfu-sion on the day of surgery, and 98% received their first trans-fusion by postoperative day 3.

    The median postoperative pre-transfusion Hb from day 1through to day 7 post-surgery was 8082 g litre21, and81% of postoperative transfusions were preceded by an Hbvalue of ,90 g litre21. The median (1st, 3rd quartile;range) last recorded Hb concentration at day 7 post-surgerywas 100 g litre21 (93, 109; 74145). There was no differencein day 7 Hb concentration between males and females orbetween patients who did or did not receive cell salvage.

    Factors associated with requiring allogeneicRBC transfusion

    In the first regression model (Table 5A), which included onlypreoperative factors and the type of surgery and anaesthe-sia, the univariate analysis showed an association betweengreater risk of requiring a perioperative RBC transfusion andlower preoperative Hb, lower patient weight, and femalegender. In multivariable analysis, the only significant inde-pendent associations with higher risk of blood transfusionwere with lower preoperative Hb and patient weight,although the surgical procedure approached significance pri-marily as a result of the difference between acetabular repla-cements alone vs combined acetabular/femoralreplacement. In the second regression model, which

    Table 3 Patient characteristics, co-morbidities, diagnoses,indications for surgery, and pre-surgery Hb concentration.*Patients could have .1 reason for surgery

    Age [mean (SD, range)] 71.2 (10.1, 3492)

    Gender (% male) 46

    Co-morbidity [number (%)]

    Hypertension 94 (45)

    Ischaemic heart disease 23 (11)

    COPD 21 (10)

    Cerebrovascular disease 10 (5)

    Preoperative medication [number of patients (number of patientswho were instructed to stop medication before surgery)]

    Aspirin 78 (15)

    Clopidogrel 2 (2)

    Warfarin 8 (7)

    Oral iron 9 (0)

    Erythropoietin 0

    Underlying diagnosis

    Osteoarthritis 143

    Rheumatoid arthritis 17

    Avascular necrosis 6

    Post-traumatic arthritis 5

    Congential dislocation 5

    Hip fracture 26

    Other 18

    Indication for revision hip arthroplasty*

    Component loosening 109

    Acetabular migration 30

    Infection 28

    Femoral component substance 9

    Dislocation/subluxation 38

    Pain 72

    Implant fracture 15

    Other (any) 10

    Preoperative Hb [mean (95% CI, range),g litre21]

    130 (128132, 88186)

    Table 4 Surgical procedure performed, use of bone grafting, numbers of patients in whom PCSwas set up, estimated blood loss in the operatingtheatre, and proportion of patients transfused between day of surgery and day 7 post-surgery. *Bone grafting data incomplete for 18 cases

    Component forrevision (n)

    Number (%) of casesincluding bone grafting

    Proportion of patients in whomcell salvage used (%)

    Estimated blood loss [mean(95% CI)]

    Proportion of patientstransfused (%)

    Femoral (14) 3 (21) 21 1151 (6621640) 57

    Acetabular (31) 11 (38) 16 611 (367856) 42

    Combined (157) 64 (41) 49 1096 (9601233) 61

    Other (8) 0 38 625 (417833) 62

    Total (210) 78 (41)* 42 1013 (8991128) 58

    BJA Walsh et al.

    66

    by guest on April 12, 2015

    http://bja.oxfordjournals.org/D

    ownloaded from

  • included estimated operating theatre blood loss and cellsalvage-related variables (Table 5B), the univariate analysisfound associations between higher risk of receiving RBCtransfusions and lower preoperative Hb and patient weight,female gender, and higher operating theatre blood loss. Inthe multivariable model, independent associations werefound between higher risk of blood transfusions and lowerpreoperative Hb and patient weight and higher operatingtheatre blood loss. There was also an independent associ-ation between receiving RBC re-infusion from the cell saverand a lower blood transfusion rate. There was a trendtowards an association between the case being undertakenin a centre using cell salvage and a lower risk of RBC transfu-sion, but this did not reach statistical significance.

    DiscussionIn a multicentre prospective cohort study of RHA, we haveshown that 58% of patients required ABTs during the peri-operative period, receiving a mean (95% CI) of 1.8 red cellunits (1.522.08) per operated case. In multivariable analysisof predefined factors that may be associated with a greaterrisk of ABT, a lower preoperative Hb and patient weight hadthe strongest independent associations with greater transfu-sion risk. After inclusion of potential confounders, includingintraoperative factors, re-infusion of salvaged RBCs was inde-pendently associated with lower transfusion risk. The mean(95% CI) re-infusion RBC volume was 312 (260362) ml,which equates to approximately 1 unit of allogeneic RBCs.

    All cases*210 patients(11 sites)

    No cell salvage116 patients

    Transfusionlocation

    Patientstransfused

    RBCunits

    Mean RBC/patienttransfused

    Total** 73 247 3.4Intraoperativeonly 22 47 2.1

    Postoperativeonly 26 69 2.7

    Intra- and postoperative 24 127 5.3

    Cell salvage,#88 patients

    Transfusion location

    Patientstransfused

    RBCunits

    Mean RBC/patienttransfused

    Total 48 134 2.8

    Intraoperativeonly 5 10 2.0

    Postoperativeonly 32 78 2.4

    Intra- and postoperative 11 46 4.2

    No salvaged blood re-infused20 patients

    Transfusion location

    Patients transfused

    RBCunits

    Mean RBC/patienttransfused

    Total 10 20 2.0

    Intraoperativeonly 1 1 1.3

    Postoperativeonly 9 19 2.1

    Intra- and postoperative 0 0 0.0

    Salvaged blood re-infused64 patients

    Transfusion location

    Patients transfused

    RBCunits

    Mean RBC/patienttransfused

    Total 35 108 3.1

    Intraoperativeonly 3 6 2.0

    Postoperativeonly 21 56 2.7

    Intra- and postoperative 11 46 4.2

    Fig 1 Description of study cohort according to the use of cell salvage and associated use of red cells at different stages of the perioperativeperiod. RBC, red blood cell. *Six of 210 (2.9%) patients had unknown cell salvage status. **One of 73 patients was transfused with 4 allogeneicred cell units but timing in relation to surgery was not documented. Four of 88 patients had an unknown re-infusion volume. #Eight (3.8%)patients received postoperative cell salvage in four centres.

    Blood transfusion in revision hip arthroplasty BJA

    67

    by guest on April 12, 2015

    http://bja.oxfordjournals.org/D

    ownloaded from

  • We collected data from a representative sample of hospi-tals including centres with a wide range of surgical volumeswhich, together with the completeness of our data, suggest alow risk of bias and high generalizability. We defined our vari-ables of interest a priori by consensus based on an earliernational survey,12 review of the literature, and the expertopinion of surgeons and anaesthetists undertaking RHAsurgery. Although residual confounding by unmeasured vari-ables cannot be excluded, we consider it unlikely that wefailed to measure important factors that may have alteredthe associations we observed. Among the patients studied,121 were transfused (58%); this number of events of inter-est justified the inclusion of 13 variables in the multivariableanalysis (10 events of interest per factor evaluated),suggesting a relatively low risk of type II error in the associ-ations we observed.

    A criticism of previous randomized trials and studies ofblood conservation technologies in surgery has been the het-erogeneity of procedures undertaken on the includedpatients, which limits the validity of findings to specific surgi-cal procedures. In addition, lack of control of other poten-tially important factors such as perioperative transfusiontrigger and surgical technique make interpretation of the evi-dence difficult, especially comparing recent with olderstudies. These factors were noted in systematic reviews ofcell salvage and the use of antifibrinolytics in orthopaedicsurgery.610 13 14 We chose to study only RHA, because thisprocedure is associated with a high risk of ABT. Studying asingle procedure minimized the confounding effect of thesurgery itself on the associations found and allowed us toinclude different subtypes of the procedure in the analysis.

    For the univariate analysis, this suggested that acetabularrevision alone was associated with significantly lower bloodloss than femoral or combined procedures. We collecteddata on the postoperative transfusion triggers used, whichwere relatively restrictive and consistent with recentevidence-based guidelines,15 16 providing evidence that ourfindings are externally valid to other populations usingevidence-based transfusion thresholds in the perioperativeperiod.

    Our data indicate that preoperative anaemia and patientweight are strong predictors of the need for ABT in revisionhip surgery, which confirms the finding of previous studiesin orthopaedic surgery.2 6 An association with femalegender has also been shown previously, but in our regressionmodel, this was not significant in the multivariable model,most likely because of the interaction with patient weightwhich correlates with circulating blood volume. Femalepatients also tend to have lower preoperative Hb concen-tration than males. The introduction of intraoperativefactors in the second regression model found a strong associ-ation between the estimated blood loss and greater ABTs,which was unsurprising. Importantly, however, re-infusionof salvaged RBCs remained as a significant predictor oflower ABT risk, suggesting that the successful use of cellsalvage was clinically useful and important. The cost-effectiveness of technologies such as PCS will vary accordingto a range of factors including individual patient character-istics, the surgery performed, and the baseline rates of ABT.The additional patient benefit of transfusion avoidance isalso relevant. An association between ABT and increasedperioperative complications has been found in previous

    >900800900700800600700500600400500300400200300100200

  • studies of orthopaedic surgery, although proving causation isdifficult in cohort studies.6 The gold standard for assessingthe clinical- and cost-effectiveness of cell salvage for RHAwould be a randomized trial, but the high uptake of cellsalvage in routine practice means that clinical equipoise isoften lacking among surgeons and anaesthetists, especiallythose already using the technology. Although our cohortstudy provides a lower evidence quality than a randomizedtrial, it provides strong supportive evidence for a clinicallyimportant ABT-sparing effect in RHA as a component ofthe current blood conservation strategies. Our data justifythe routine use of intraoperative cell salvage for patients

    with preoperative anaemia and low body weight, andstrongly support the use for all patients having combinedfemoral/acetabular procedures or other procedures forwhich the surgeon anticipates a high blood loss. The meanre-infusion volume of 312 ml, equivalent to 1 allogeneicRBC unit, suggests likely cost-effectiveness measuredagainst the cost of disposables alone even if the other clinicalbenefits of ABT avoidance and the reduced demands onblood services are not taken into account. However, ourobservation that 25% of cases in which cell salvage wasset up did not have autologous blood re-infused emphasizesthe importance of appropriate case selection. Our findings

    Table 5 Regression analysis exploring univariable and multivariable associations between risk factors and the probability of perioperative RBCtransfusion (as a binary outcome). Panel A shows associations when only preoperative variables and the surgical and anaesthetic techniquewere included. Panel B shows associations when estimated operating theatre blood loss (continuous variable), surgery carried out in a centreusing cell salvage, whether cell salvage was set up for the case, and whether RBC re-infusion from cell saver occurred (all binary variables) wereincluded. For the ORs, the changes are relative to the variable being absent in most cases, and for procedure are relative to acetabularreplacement alone (index procedure). For data dealt with as continuous variables, the unit of change used is indicated. The ORs indicate changesin probability rather than changes in clinical rates. *Analysis based on 194 cases with complete data for all fields (16 cases missing 1 variable).Based on 182 cases with complete data for all fields (28 cases missing 1 variable)

    Variable Univariable analysis Multivariable analysis*,

    OR (95% CI) P-value OR (95% CI) P-value

    A

    Preoperative Hb concentration (change per increase in Hb in g dl21) 0.47 (0.360.60) ,0.001 0.44 (0.330.58) ,0.001

    Weight (change per kg weight increase) 0.98 (0.960.99) 0.005 0.98 (0.961.00) 0.032

    Female gender (female relative to male) 1.85 (1.063.22) 0.03 NS

    Surgical procedure

    Acetabular Index procedure 0.28 Index procedure 0.051

    Femoral 1.85 (0.526.62) 4.92 (0.9126.61)

    Combined 2.18 (1.004.76) 4.13 (1.5512.03)

    Other 2.31 (0.4711.4) 4.26 (0.5036.08)

    Preoperative aspirin/heparin therapy 1.50 (0.832.73) 0.18 NS

    Age (change per year increase) 1.02 (0.991.05) 0.17 NS

    ASA grade (change per grade increase in ASA grade) 1.43 (0.932.19) 0.10 NS

    Infection 1.96 (0.824.68) 0.13 NS

    Regional anaesthesia 1.42 (0.513.96) 0.50 NS

    B

    Preoperative Hb concentration (change per increase in Hb in g dl21) 0.47 (0.360.60) ,0.001 0.35 (0.230.51) ,0.001

    Weight (change per kg weight increase) 0.98 (0.960.99) 0.005 0.96 (0.930.99) 0.004

    Female gender (female relative to male) 1.85 (1.063.22) 0.03 NS

    Surgical procedure

    Acetabular Index procedure

    Femoral 1.85 (0.526.62) 0.28 NS

    Combined 2.18 (1.004.76)

    Other 2.31 (0.4711.4)

    Preoperative aspirin/heparin therapy 1.50 (0.832.73) 0.18 NS

    Age (change per year increase) 1.02 (0.991.05) 0.17 NS

    ASA grade (change per grade increase in ASA grade) 1.43 (0.932.19) 0.10 NS

    Infection 1.96 (0.824.68) 0.13 NS

    Regional anaesthesia 1.42 (0.513.96) 0.50 NS

    Operating theatre blood loss (change per ml increase in blood loss) 1.002 (1.0011.002) ,0.001 1.002 (1.0021.003) ,0.001

    Hospital using cell salvage 1.10 (0.552.22) 0.79 0.31 (0.081.17) 0.08

    Cell salvage set up for case 0.78 (0.451.36) 0.38 NS

    Salvaged red cells re-infused 0.82 (0.451.48) 0.51 0.31 (0.110.82) 0.02

    Blood transfusion in revision hip arthroplasty BJA

    69

    by guest on April 12, 2015

    http://bja.oxfordjournals.org/D

    ownloaded from

  • are consistent with a retrospective case-matched study,which found a large reduction in ABT rate for patientsmanaged with PCS (on average 4 red cell units per case).17

    The smaller effect seen in our cohort was probably a reflec-tion of the lower overall transfusion rates, which may beexplained by improved surgical techniques and more restric-tive transfusion triggers.

    The use of antifibrinolytics was low in our cohort, andmost use occurred in a single centre, which made assess-ment of their association with allogeneic RBC transfusion dif-ficult to assess. Previous systematic reviews have suggestedthat antifibrinolytic agents reduce blood loss and possiblyRBC requirements in orthopaedic surgery, but most studiesare potentially confounded by the same factors notedabove for evaluations of cell salvage.7 11 14 In addition, inter-action between antifibrinolytics and cell salvage has notbeen evaluated or the safety profile, especially for thrombo-tic events, conclusively established.14

    In conclusion, for RHA, we have confirmed the associationbetween preoperative anaemia and patient weight and therisk of perioperative ABTs. After adjusting for other potentiallyimportant confounding factors, we have demonstrated thatintraoperative cell salvage decreases the chance of requiringABT during the perioperative period and typically results inre-infusion of about 1 unit of autologous red cells.

    Declaration of interestNone declared.

    FundingThis study was supported by the Clinical Effectiveness groupof the Scottish National Blood Transfusion Service.

    Appendix

    CollaboratorsThe following individuals and hospitals contributed to thecollection of data for this study:

    Royal Infirmary of Edinburgh, NHS LothianLead: Dr Vicki Clark, Consultant AnaesthetistDr Karen Stevenson, Ms Catrina Hendry, andMs Catherine Innes

    Victoria Hospital, Kirkcaldy, NHS FifeLead: Dr Peter Curry, Consultant AnaesthetistDr Matthew Royds, Ms Karen Ritchie, and Ms KirstyDalrymple

    Ninewells Hospital, NHS TaysideLead: Dr Matthew Checketts, Consultant AnaesthetistAnaesthetic colleagues and Orthopaedic Theatrenursing staff, Ms Sheila Milne, Ms Sarah Crooks, andMs Eleanor Hazra

    Glasgow Royal Infirmary, NHS Greater Glasgow and ClydeLead: Dr Stephen Hickey, Consultant AnaesthetistMs Anna Hufana, Ms Lorna Sinclair, Ms MargaretMcGarvey, and Ms Hazel Wisniewski

    Raigmore Hospital, NHS HighlandLead: Dr Alexander Hunter, Consultant AnaesthetistAnaesthetic colleagues, Angus MacLennan, KevinBarnett, and Mr James McAllan

    Royal Alexandra Hospital, NHS Greater Glasgow and ClydeLead: Mr Alan Crombie, Consultant Orthopaedic SurgeonDr Fred Davies, Ms Rhona Gardner, and Ms Tina Watson

    Victoria Infirmary, NHS Greater Glasgow and ClydeLead: Dr Deepa Singh, Consultant AnaesthetistAnaesthetic colleague, Ms Johan Fleming

    Stirling Royal InfirmaryLead: Dr Mark Worsley, Consultant AnaesthetistAnaesthetic colleagues, Ms Caroline Izatt, andMr Rodeney Denoven

    Borders General HospitalLead: Dr Nigel LearyAnaesthetic colleague and Orthopaedic Theatre nursingstaff, Ms Susan Cottrell

    Wishaw General HospitalLead: Dr Seamus Thomson, Consultant AnaesthetistAnaesthetic colleague, Ms Heather Daniels

    Dr Grays Hospital, ElginLead: Mr David Anderson, Consultant OrthopaedicSurgeonDr Colin McFarlane, Ms Catrina Hendry

    References1 Wells AW, Mounter PJ, Chapman CE, Stainsby D, Wallis JP. Where

    does blood go? Prospective observational study of red cell trans-fusion in north England. Br Med J 2002; 325: 803

    2 Saleh E, McClelland DB, Hay A, Semple D, Walsh TS. Prevalence ofanaemia before major joint arthroplasty and the potential impactof preoperative investigation and correction on perioperativeblood transfusions. Br J Anaesth 2007; 99: 8018

    3 Audet AM, Andrzejewski C, Popovsky MA. Red blood cell transfu-sion practices in patients undergoing orthopaedic surgery: amulti-institutional analysis. Orthopedics 1998; 21: 8518

    4 Surgenor DM, Wallace EL, Churchill WH, Hao SH, Chapman RH,Poss R. Red cell transfusions in total knee and total hip replace-ment surgery. Transfusion 1991; 31: 5317

    5 Turgeon AF, Fergusson DA, Doucette S, et al. Red blood cell trans-fusion practices amongst Canadian anesthesiologists: a survey.Can J Anaesth 2006; 53: 34452

    6 Spahn DR. Anemia and patient blood management in hip andknee surgery: a systematic review of the literature. Anesthesiol-ogy 2010; 113: 48295

    7 Munoz M, Garcia-Erce JA, Villar I, Thomas D. Blood conservationstrategies in major orthopaedic surgery: efficacy, safety and Euro-pean regulations. Vox Sang 2009; 96: 113

    8 Boulton FE, James V. Guidelines for policies on alternatives toallogeneic blood transfusion. 1. Predeposit autologous blooddonation and transfusion. Transfusion Med. 2007; 17: 35465

    9 Carless PA, Henry DA, Moxey AJ, OConnell D, Brown T,Fergusson DA. Cell salvage for minimising perioperative allo-geneic blood transfusion. Cochrane Database Syst Rev. 2010; 3:CD001888

    10 Ashworth A, Klein AA. Cell salvage as part of a blood conservationstrategy in anaesthesia. Br J Anaesth. 2010; 105: 40116

    BJA Walsh et al.

    70

    by guest on April 12, 2015

    http://bja.oxfordjournals.org/D

    ownloaded from

  • 11 Zufferey P, Merquiol F, Laporte S, et al. Do antifibrinolytics reduceallogeneic blood transfusion in orthopedic surgery. Anesthesiol-ogy 2006; 105: 103446

    12 Harkness M, Palmer J, Watson D, Walsh TS. A questionnaire basedsurvey of perioperative blood conservation practice for revisionhip arthroplasty in Scotland. Transfusion Med 2008; 18: 296301

    13 Davies L, Brown TJ, Haynes S, Payne K, Elliott RA, McCollum C.Cost-effectiveness of cell salvage and alternative methods ofminimising perioperative allogeneic blood transfusion: a sys-tematic review and economic model. Health Technol Assessment2006; 10: 11114

    14 Kagoma YK, Crowther MA, Douketis J, Bhandari M, Eikelboom J,Lim W. Use of antifibrinolytic therapy to reduce transfusion in

    patients undergoing orthopedic surgery: a systematic review ofrandomized trials. Thromb Res 2009; 123: 68796

    15 American Society of Anesthesiologists Task Force on PerioperativeBlood Transfusion and Adjuvant Therapies. Practice guidelines forperioperative blood transfusion and adjuvant therapies. Anesthe-siology 2006; 105: 198208

    16 Blood Transfusion and the Anaesthetist. Association of Anaesthe-tists of Great Britain and Ireland. Available from http://www.aagbi.org/publications/guidelines/docs/red_cell_08.pdf (accessed 2November 2010)

    17 Bridgens JP, Evans CR, Dobson PMS, Hamer AJ. Intraoperative redblood-cell salvage in revision hip surgery. J Bone Joint Surg 2007;89: 2705

    Blood transfusion in revision hip arthroplasty BJA

    71

    by guest on April 12, 2015

    http://bja.oxfordjournals.org/D

    ownloaded from