Portal de Periódicos da CAPES

A reduction in Type and Screen: preoperative prediction of RBC transfusions in surgery procedures with intermediate transfusion risks †

2001; Elsevier BV; Volume: 87; Issue: 2 Linguagem: Inglês

10.1093/bja/87.2.250

1471-6771

Wilton A. van Klei, Karel G.M. Moons, A. T. Rheineck Leyssius, J. T. A. Knape, C. L. G. Rutten, D. E. Grobbee,

Trauma, Hemostasis, Coagulopathy, Resuscitation

In many patients, a ‘type and screen’ procedure is routinely performed before surgery. However, most patients are not transfused after all. Can we predict, which surgical patients will and will not be transfused, to reduce the number of these investigations? We studied 1482 consecutive surgical patients with intermediate risk for transfusion. Multivariate logistic regression modelling and the area under the Receiver Operating Characteristic curve (ROC area) were used to quantify how well age, gender, surgical procedure, emergency or elective surgery and anaesthetic technique predicted transfusion, and whether the preoperative haemoglobin concentration had added predictive value. Gender, age ≥70 yr, and type of surgery were independent predictors of transfusion, with a ROC area of 0.75 (95% CI: 0.72–0.79). Validating this model with an easily used prediction rule in a second patient population yielded a ROC area of 0.70 (95% CI: 0.63–0.77). With this rule type and screen could correctly be withheld in 35% of these patients. In the remaining 65% of the patients, a further reduction in type and screen investigations of 15% could be achieved using the preoperative haemoglobin concentration. Using a simple prediction rule, preoperative type and screen investigations in patients who have to undergo surgery procedures with intermediate transfusion risk can be avoided in about 50%. This may reduce patient burden and hospital costs (on average: 3 million US$ per 100 000 procedures). In many patients, a ‘type and screen’ procedure is routinely performed before surgery. However, most patients are not transfused after all. Can we predict, which surgical patients will and will not be transfused, to reduce the number of these investigations? We studied 1482 consecutive surgical patients with intermediate risk for transfusion. Multivariate logistic regression modelling and the area under the Receiver Operating Characteristic curve (ROC area) were used to quantify how well age, gender, surgical procedure, emergency or elective surgery and anaesthetic technique predicted transfusion, and whether the preoperative haemoglobin concentration had added predictive value. Gender, age ≥70 yr, and type of surgery were independent predictors of transfusion, with a ROC area of 0.75 (95% CI: 0.72–0.79). Validating this model with an easily used prediction rule in a second patient population yielded a ROC area of 0.70 (95% CI: 0.63–0.77). With this rule type and screen could correctly be withheld in 35% of these patients. In the remaining 65% of the patients, a further reduction in type and screen investigations of 15% could be achieved using the preoperative haemoglobin concentration. Using a simple prediction rule, preoperative type and screen investigations in patients who have to undergo surgery procedures with intermediate transfusion risk can be avoided in about 50%. This may reduce patient burden and hospital costs (on average: 3 million US$ per 100 000 procedures). Transfusion of blood (red blood cells or RBC) is sometimes necessary in patients having surgery. A ‘type and screen’ is done preoperatively to prevent complications from incompatibility between donor and recipient or the existence of irregular antibodies. This procedure is much cheaper than full cross matching, and gives the same immuno-haematological safety.1Atrah HI Galea G Urbaniak SJ The sustained impact of a group and screen and maximum surgical blood ordering schedule policy on the transfusion practice in gynaecology and obstetrics.Clin Lab Haem. 1995; 17: 177-181PubMed Google Scholar, 2Riedler GF “Type and screen”: die immunhamatologische Sicherheit ist gewahrleistet und die Kosten werden gesenkt. [English summary].Schweiz Med Wochenschr. 1996; 126: 1946-1951PubMed Google Scholar, 3Ng SP Blood transfusion requirements for abdominal hysterectomy: 3-year experience in a district hospital (1993–1995).Aust N Z J Obstet Gynaecol. 1997; 37: 452-457Crossref PubMed Scopus (11) Google Scholar, 4Rehm JP Otto PS West WW et al.Hospital-wide educational program decreases red blood cell transfusions.J Surg Res. 1998; 75: 183-186Abstract Full Text PDF PubMed Scopus (21) Google Scholar Generally, physicians type and screen patients who might need a perioperative transfusion (commonly based on past experience with the surgical procedure as single predictor). However, most patients who are typed and screened before surgery will not require a transfusion, which means unnecessary patient burden and costs. It would be efficient to further classify patients according to their risk of transfusion using objective and easy obtainable information. Various prediction rules have been developed (especially in orthopaedic surgery), but a laboratory value (preoperative haemoglobin concentration or haematocrit) was always included.5Nuttall GA Santrach PJ Oliver-WC J et al.The predictors of red cell transfusions in total hip arthroplasties.Transfusion. 1996; 36: 144-149Crossref PubMed Scopus (136) Google Scholar However, it would be even more efficient if the same predictive accuracy could be obtained without the need for laboratory tests.6Ong AH Sim KM Boey SK Preoperative prediction of intra and postoperative red blood cell transfusion in surgical patients.Ann Acad Med Singapore. 1997; 26: 430-434PubMed Google Scholar, 7Keating EM Meding JB Faris PM Ritter MA Predictors of transfusion risk in elective knee surgery.Clin Orthop. 1998; : 50-59Crossref PubMed Scopus (144) Google Scholar, 8Larocque BJ Gilbert K Brien WF Prospective validation of a point score system for predicting blood transfusion following hip or knee replacement.Transfusion. 1998; 38: 932-937Crossref PubMed Scopus (43) Google Scholar, 9Bierbaum BE Callaghan JJ Galante JO Rubash HE Tooms RE Welch RB An analysis of blood management in patients having a total hip or knee arthroplasty.J Bone Joint Surg Am. 1999; 81: 2-10Crossref PubMed Scopus (792) Google Scholar We developed and validated a rule based on patient and surgery characteristics, to predict blood transfusion in patients undergoing surgery with intermediate transfusion risk (1–30%). Subsequently, we evaluated how knowing the preoperative haemoglobin concentration could increase the predictive accuracy of this prediction rule. We studied 1482 patients (aged 18–98 yr) with intermediate transfusion risk (‘type and screen patients’), undergoing surgery in the Twenteborg Hospital in The Netherlands, in 1998. This hospital is a 638-bed non-university hospital in which neurosurgery and cardiac surgery are not performed. The classification of type and screen patients was based on the current transfusion guide. This divides patients into three surgical groups according to expert opinion. Group A patients have low expected risk for transfusion (0–1%; e.g. arthroscopy or ear surgery), group B patients have intermediate risk for transfusion (1% to approximately 30%; e.g. cholescystectomy or hysterectomy) and group C are high risk patients (more than 30%; e.g. aortic surgery). In group A patients, type and screen is never done (78% of all patients). Patients belonging to group B are always typed and screened, but blood is not stored (16%). Group C patients are always typed and screened and blood is stored (6%). Of all patients in group A, nearly 2% received transfusions. In group B and C the incidence of transfusion was 19% and 43%, respectively. This study evaluates only group B patients (‘type and screen patients’). None of the 1482 patients donated autologous blood preoperatively. The outcome was defined as any allogeneic RBC transfusion (defined as transfusion of one or more units packed cells) on the day of surgery or the first postoperative day. The transfusion decision was made by individual clinicians (anaesthesiologists and surgeons). A rigid protocol was not in use, but in general blood was given when the haemoglobin level was below 10 g dl−1 (6.0 mmol litre−1). Age, gender, surgery procedures, whether it was an emergency operation (yes/no), the anaesthetic technique and the preoperative haemoglobin level were evaluated as potential predictors. As 39 different surgical procedures were used, they were allocated into 5 categories based on actual risk (occurrence) of transfusion: Group 1 contained only laparoscopic cholecystectomy (transfusion incidence <5%); Group 2 mastectomy and transurethral resection of tumour (TURT) or prostate (TURP) (transfusion incidence 5–9%); Group 3 open cholescystectomy, vaginal hysterectomy, Caesarean section, surgery for urine incontinence or vaginal prolapse (10–19%); Group 4 non-cardiac thoracic surgery (e.g. lobectomy), vascular (arterial) surgery (e.g. femoro-popliteal bypass), prostate enucleation and endometrial cancer surgery (20–29%); Group 5 abdominal and vaginal hysterectomy, hip fracture surgery, revision knee prosthesis, leg amputation, gastro-enterostomy, colon-resection and radical abdominal hysterectomy (30% or more). Anaesthetic technique was defined as a dichotomous variable: a single form of anaesthesia (general, regional or local) compared with a combination anaesthesia (general anaesthesia combined with epidural analgesia). Although in principle a potential predictor, we decided not to include the identity of the surgeon and anaesthesiologist in the model, as they are hard to extrapolate to other hospitals and the aim was to derive an easy and widely applicable prediction rule. The hospital ethics committee approved the study. All data were collected retrospectively from the hospital information system. There were no missing data on any of the predictor or outcome variables, except for the haemoglobin concentration. In 152 patients (10%) it was not measured preoperatively. In the present study, two data sets were randomly selected from all 1482 patients: a derivation set of approximately 75% (1151 patients) and a validation set of approximately 25% (331 patients). SPSS release 9.0 for Windows was used in the analysis (Windows NT 4.0, DELL computer). In the derivation set the association between each predictor and transfusion outcome was quantified using univariable logistic regression modelling. This type of analysis is an alternative to using chi-square tests and gives similar results. In the analysis, surgery was included as four indicator variables (groups 2 to 5) with group 1 as the reference. As the incidence of transfusion in patients aged 18 to 69 was between 10% and 20% in each decade, whereas in patients aged over 70 the incidence increased more rapidly, age was included in the model after dichotomization at 70. After univariate analyses, multivariable logistic regression modelling was applied in order to obtain a prediction model including the independent predictors of transfusion outcome only. This was done by a two-step approach. As age, gender, type of surgery (again included as four indicators), elective surgery and anaesthetic procedure are much easier to obtain, we first evaluated whether these had independent value in the prediction of perioperative transfusion. In this, the interaction between type of surgery and anaesthetic technique was evaluated as well, since both are closely related (regional anaesthesia may reduce blood loss). Subsequently, the added predictive value of the preoperative haemoglobin concentration was evaluated. The full model was reduced by manually (i.e. not automatically) deleting non-significant variables. Predictors with odds ratios that differed significantly from one, defined as odds ratio with P-value 0.20) indicates that there is no difference between both probabilities, which means good fit of a model.11Hosmer DW Lemeshow S Applied Logistic Regression. J. Wiley and sons, New York1989: 140-145Google Scholar The ability of the model to discriminate between patients with and without transfusion was quantified by using the area under the Receiver Operating Characteristic curve (ROC area).10Harrel FE Lee KL Mark DB Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors.Stat Med. 1996; 15: 361-387Crossref PubMed Scopus (7039) Google Scholar, 11Hosmer DW Lemeshow S Applied Logistic Regression. J. Wiley and sons, New York1989: 140-145Google Scholar, 12Hanley JA McNeil BJ The meaning and use of the area under a receiver operating characteristic curve.Radiology. 1982; 143: 29-36Crossref PubMed Scopus (16025) Google Scholar The ROC area can range from 0.5 (no more predictive than a coin flip) to 1.0 (perfect discrimination). A value over 0.7 can be interpreted as reasonable or fair, and over 0.8 as good.13Weinstein MC Fineberg HV Clinical Decision Analysis. WB Saunders, Philadelphia1980Google Scholar Differences in ROC area were used to quantify the difference in discriminative ability between full and reduced models taking into account the correlation between the models as they were based on the same cases.14Hanley JA McNeil BJ A method of comparing the areas under receiver operating characteristic curves derived from the same cases.Radiology. 1983; 148: 839-843Crossref PubMed Scopus (5991) Google Scholar The performance of the rule was tested in the validation set and the resulting ROC area was compared with the derivation set. A ROC area reflects the overall added value of a model and does not directly indicate its clinical value.15Moons KGM Stijnen T Michel BC et al.Application of treatment thresholds to Diagnostic-test Evaluation: An alternative to the comparison of areas under the Receiver Operating Characteristic Curves.Med Decis Making. 1997; 17: 447-454Crossref PubMed Scopus (56) Google Scholar 16Moons KGM van Es G Michel BC Buller HR Habbema JDF Grobbee DE Redundancy of Single Diagnostic Test Evaluation.Epidemiology. 1999; 10: 276-281Crossref PubMed Scopus (70) Google Scholar Therefore, in the validation set, we estimated the absolute number of correctly predicted transfused and not transfused patients for the various risk scores of the rule. Table 1 shows the comparison of patient characteristics of the derivation and validation set. There were no major differences between the two sets. The transfusion rates for the derivation and validation set were 18.1% (n=208) and 20.8% (n=69), respectively.Table 1Patient characteristics of derivation and validation set. Values are numbers and percentages between parentheses. # Values are mean and standard deviation between parentheses. * Procedures are listed in the textDerivation setValidation setn=1151n=331Mean age (years)transfused patients62 (21.3)#62 (20.3)non-transfused patients56 (18.3)58 (18.8)Age (%)18–69 yr790 (69)218 (66)≥70 yr361 (31)113 (34)Gender (%)male404 (35)119 (36)female747 (65)212 (64)Anaesthetic technique (%)single type anaesthesia1052 (91)307 (93)combined anaesthesia99 (9)24 (7)Surgery procedures*(%)group 1121 (11)23 (7)group 2295 (26)94 (28)group 3356 (31)93 (28)group 494 (8)24 (7)group 5285 (25)97 (29)Type of surgery (%)elective787 (68)226 (68)emergency364 (32)105 (32)Transfusion (%)208 (18)69 (21) Open table in a new tab In the univariate analysis (Table 2) all variables were significantly associated with transfusion. The odds ratios of the four indicators for surgery (groups 2 to 5) indicate the relative risk of transfusion for that group, compared to the reference group 1 (e.g. procedures within group 3 have a 4.1 times higher risk of transfusion than those of group 1). After entering age, gender, surgery procedure and emergency surgery into a multivariate logistic model, all were independently associated with transfusion (Table 3), except emergency surgery (OR 1.26; 95% CI: 0.84–1.88). The ROC area of this first model was 0.75 (95% CI: 0.71–0.79). As further exclusion of variables from this model significantly reduced the ROC area, the model with dichotomized age, gender and surgery procedure was defined as the final prediction model. Addition of anaesthetic technique (including the interaction terms with surgery procedure) to this model showed no added value in the prediction of transfusion: the ROC area remained 0.75. (For the estimation of the added value of the preoperative haemoglobin concentration see below.) Its ROC area in the validation set was 0.71 (95% CI: 0.64–0.78). The model's estimated risks of transfusion were comparable to the observed risks, which indicated a good model fit (the P-value of the Hosmer and Lemeshow test was 0.98).Table 2Association of each variable with the incidence of transfusion. # Reference category. * Surgery procedures were included as four indicator variables with group 1 as the reference category. ¶ mean. † OR per g dl−1 increase in haemoglobin concentration. OR=odds ratio; 95% CI=95% confidence interval; LLR=log likelihood ratio testDeterminantTransfusedNot transfusedOR (95% CI)P-value (LLR)Age (%)18–69 years110 (14)680 (86)#≥70 years98 (27)263 (73)2.3 (1.7–3.1)<0.001Gender (%)male49 (12)355 (88)#female159 (21)588 (79)2.0 (1.4–2.8)<0.001Anaesthetic technique (%)single type anaesthesia167 (16)885 (84)#combined anaesthesia39 (39)60 (61)3.4 (2.2–5.3)<0.001Surgery procedures* (%)group 15 (4)116 (96)#group 218 (6)277 (94)1.5 (0.5–4.2)0.425group 353 (15)303 (85)4.1 (1.6–10.4)0.002group 425 (27)69 (73)8.4 (3.1–23.0)<0.001group 5107 (38)178 (62)13.9 (5.5–35.2)<0.001Type of surgery (%)elective112 (14)675 (86)#emergency96 (26)268 (74)2.2 (1.6–2.9)<0.001Preoperative haemoglobin (g dl−1)13.4¶11.5¶0.4 (0.3–0.5)†<0.001 Open table in a new tab Table 3Association of each variable with the incidence of transfusion in the multivariate logistic model. * Procedures per group are listed in the text; group 1 is the reference group. # The score of each predictor was obtained by dividing the corresponding regression coefficient by the smallest coefficient (0.524) and rounded to the nearest integer. 95% CI=95% confidence interval; OR=odds ratioDeterminantRegression coefficient (95% CI)OR (95% CI)P-valueScore#Gender (woman)0.629 (0.20; 1.06)1.9 (1.2;2.9)0.0041Age ≥700.546 (0.18; 0.90)1.7 (1.2;2.5)0.0031Surgery procedure*group 20.524 (–0.52; 1.06)1.7 (0.6;4.8)0.3241group 31.291 (0.35; 2.23)3.6 (1.4;9.3)0.0072group 42.287 (1.26; 3.32)9.8 (3.5;27.8)< 0.0014group 52.386 (1.45; 3.33)10.9 (4.2;27.9)< 0.0015Intercept (constant)−3.701 (–4.67;–2.73)< 0.001 Open table in a new tab This final model was transformed into an easily used scoring rule by dividing each regression coefficient by the smallest coefficient (0.524) and rounded to the nearest integer (last column of Table 3): 1*gender + 1*age≥70 + (1, 2, 4 or 5)*surgery procedure. Being a woman counts for 1 point, age ≥70 for 1 point, and surgery procedure for 1, 2, 4 or 5 points (for groups 2, 3, 4 and 5, respectively). Such a scoring rule can be considered as one overall predictor test, including several predictor variables. The score can be considered as its (test) result and can be estimated for each patient by assigning the points for each predictor present and adding these points. For instance, a 72-year-old man who will undergo a colon-resection, receives a score of 6 (0 + 1 + 5). In both datasets the score ranged from 0 to 7. The ROC area of the transformed prediction rule was 0.75 (95% CI: 0.71–0.78) and 0.70 (95% CI: 0.63–0.77) in the derivation and validation set, respectively (Fig. 1). This prediction rule can be used preoperatively to distinguish patients who will and will not be transfused and therefore should and should not be typed and screened. Table 4 shows the actual number of transfused and not transfused patients across score categories (and across corresponding risk of transfusion as estimated by the untransformed model, i.e. second column of Table 3), after the rule was applied to the validation set.Table 4Distribution of transfused and not transfused patients in the validation set, according to the score of the rule (and to the corresponding risk of transfusion). Values are presented as absolute numbers and as percentages of the ‘Total’ column between parentheses.# Categories of the score estimated from the (transformed) scoring rule (Table 3). *Risk or probability of transfusion estimated by the untransformed prediction model (Table 3) that correspond to the score from the first row: Risk=1/(1 + exp –(–3.701 + 0.629*gender +0.546*age ≥70 + 0.524*group2 + 1.291*group3 + 2.287*group4 + 2.386*group5)). n=number of subjects per score (risk) category.Score by the rule#≤23 and 4≥5TotalRisk of transfusion* (%)≤1011–20≥21Transfused11 (16)19 (26)39 (58)69 (100)Not transfused104 (40)86 (32)72 (28)262 (100)n115 (35)105 (31)111 (34)331 (100) Open table in a new tab From Table 4 one can directly obtain the predictive value for transfusion per score category (reading the table vertically). For example, of all 115 patients with score ≤2 (or risk of transfusion ≤10%), 104 patients were indeed not transfused, yielding a negative predictive value of 90%. In the group of patients with score ≥5, 39 of the 111 were indeed transfused; a positive predictive value of 35%. Table 4 also enables to estimate the sensitivity and specificity at different score thresholds (reading the table horizontally). For example, introducing a threshold at 2, a score ≤2 will be considered as test negative and a score >2 will be considered as test positive. This means that, according to the rule, a test negative patient will not be transfused and does not need to be typed and screened, whether a test positive patient will be transfused and needs to be typed and screened. Using this threshold of ≤2 (or transfusion risk ≤10%), the specificity was 40% (104/262) with 60% unnecessary type and screen procedures, whereas the sensitivity was 84% (19+39/69) with 11 (16%) missed transfused patients. This 16% of patients who needed transfusion and who would not have been tested, was only 2% less than using a model with type of surgery as a single predictor. Because age and gender are easily obtainable, we decided to leave them in the model. The sensitivity and specificity of all possible score thresholds can be obtained from the ROC curve in Figure 1. We wished to reduce the number of unnecessary type and screen procedures (i.e. to obtain a high specificity). We tested whether the preoperative haemoglobin concentration (preopHb), when added to the former prediction model (Table 3), contributed useful information. Adding preopHb to the previous model of Table 3, the ROC area increased from 0.71 to 0.80 (95% CI: 0.74–0.86) in the validation set. In absolute numbers the percentage of missed transfused patients decreased from 16% to 12%. We reasoned that it would therefore be inefficient to include preopHb in the initial prediction model, as it led to only a small decrease (4%) in missed transfusions at the expense of a haemoglobin measurement in all patients. Nevertheless, using the preopHb additionally after the application of the rule, i.e. only measuring the preoperative haemoglobin level among those patients with score >2, a further reduction in the number of unnecessary type and screen procedures was achievable. Of the 216 patients with score >2 (Table 4), 31 were excluded due to missing values on preopHb, leaving 185. Although we evaluated preopHb as a continuous as well as a dichotomous predictor variable, we decided to use the dichotomised form (at 14 g dl−1) to enhance applicability, as there was no difference in predictive accuracy. Table 5 shows the results and can be read in the same way as Table 4. By withholding type and screen procedures in all patients with a preopHb level ≥14.0 g dl−1, a further reduction in type and screen investigations of 24% could be achieved at the expense of another five missed transfused patients. Other haemoglobin thresholds yielded worse results.Table 5Distribution of transfused and not transfused patients according to the preoperative haemoglobin concentration in the patients from Table 4 with score >2. Values are presented as absolute numbers and as percentages of the ‘Total’ column between parentheses. * Preoperative haemoglobin concentration in g dl−1. n=number of subjects per haemoglobin categoryHb (g dl−1)* 2 units. #TUR=transurethral resection of prostate or tumour. *PreopHb= preoperative haemoglobin concentrationSurgery procedurePatients (n)Patients (n) with >2 units transfused (no. of units RBC)¶After application of the scoring rule: TUR prostate/tumour#175 (3; 3; 4; 4; 6) cholecystectomy (laparoscopically/open)102 (8; 10) mastectomy with lymph node dissection80After additional application of preopHb*: abdominal hysterectomy60 hip fracture surgery40 lobectomy (lung)31 (5) peripheral artery surgery31 (4) colon resection10 prostate adenoma enucleation21 (5) revision knee prosthesis10 Open table in a new tab To reduce the number of unnecessary preoperative type and screen procedures, we defined an easily applicable scoring rule containing three simple variables (gender, age, and surgery procedure) to predict transfusion in patients undergoing surgical procedures with intermediate transfusion risk. Some comments are necessary. First, the prediction rules were based on data from one particular hospital. It is known that there are large differences in blood use between hospitals.17Carson JL Spence RK Poses RM Bonavita G Severity of anaemia and operative mortality and morbidity.Lancet. 1988; 340: 727-729Abstract Scopus (301) Google Scholar, 18The Sanguis study group Use of blood products for elective surgery in 43 European hospitals.Trans Med. 1994; 4: 251-268Crossref PubMed Scopus (152) Google Scholar, 19Carson JL Duff A Poses RM et al.Effect of anaemia and cardiovascular disease on surgical mortality and morbidity.Lancet. 1996; 348: 1055-1060Abstract Full Text Full Text PDF PubMed Scopus (790) Google Scholar, 20Baele P Beguin C Waterloos H et al.The Belgium BIOMED Study about transfusion for surgery.Acta Anaesthesiol Belg. 1998; 49: 243-303PubMed Google Scholar, 21Capraro L Transfusion practices in primary total joint replacements in Finland.Vox Sang. 1998; 75: 1-6Crossref PubMed Google Scholar, 22Surgenor DM Churchill WH Wallace EL et al.The specific hospital significantly affects red cell and component transfusion practice in coronary artery bypass graft surgery: a study of five hospitals.Transfusion. 1998; 38: 122-133Crossref PubMed Scopus (66) Google Scholar, 23Carson JL Duff A Berlin JA et al.Perioperative blood transfusion and postoperative mortality.JAMA. 1998; 279: 199-205Crossref PubMed Scopus (361) Google Scholar Although we have tried to show the robustness of the rule by testing it in a second dataset of our hospital, further research has to be done to validate the rule in other hospitals. Second, the transfusion trigger was a haemoglobin level of 10 g dl−1 as was recommended formerly.24Adams RC Lundy JS Anesthesia in cases of poor surgical risk. Some suggestions for decreasing the risk.Surg Gynecol Obstet. 1942; 74: 1011-1019Google Scholar, 25Kowalyshyn TJ Prager D Young J A review of the present status of preoperative hemoglobin requirements.Anesth Analg. 1972; 51: 75Crossref PubMed Google Scholar, 26Finch CA Lenfant C Oxygen transport in man.N Engl J Med. 1972; 286: 407-415Crossref PubMed Scopus (240) Google Scholar Currently, fewer patients are transfused as RBC transfusion is now based on the patient's risk for developing inadequate tissue oxygenation, which decreases the transfusion trigger to a haemoglo