| Backgroud and objectivePerioperative bleeding remainsa challenge forsurgery. Low hemoglobin levelshad proven to berelated to operative mortality.Early diagnosis of severe anemia and transfusiontherapy are important to prevent anemia-induced tissuehypoxia, organ failure, and mortality.However, there is an increased risk imparted by the increased transfusion. Allogeneic blood transfusion always related to hyperkalaemia, impairment ofcoagulation,acute lung injury, postoperative infections, andthromboticevents.Besides,even autologous blood transfusions will lead toimmunosuppressive effects.Traditional hemoglobin measurement requires blood draws, which is time-consuming andcan only provide intermittentmeasurement. In contrast,the multiwavelength pulse co-oximeter Masimo Radical-7can provide real-time andcontinuous estimation of hemoglobin concentration noninvasively, which approach a early diagnosis of anemia and avoid over-transfusion during surgeryand thus exhibits a sufficient capacityfor detecting the site of acute bleeding.The ability of this new technique to evaluate hemoglobin concentration and guide transfusion has recently beendemonstrated in a series of clinical trials with varied conclusions about accuracy andprecision.Those validation studies involved diversesoftware versions, patient populationand types of surgery. Whether the factors described above will be possible confounders in noninvasive hemoglobin measurement isstill unknown, and there is no clear understanding of how confounders causes changesof noninvasive hemoglobin measurement.Furthermore, we observedsome evidence that anelevated bilirubinlevel may influence the reading of some old model of multiwavelength oximeters, and no studyhas investigated the accuracy of noninvasive hemoglobin measured by this new pulse co-oximeter Masimo Radical-7over a wide range of bilirubin.In this study we hypothesized that elevated bilirubin level, different types of surgery and the updated software version were possible confounders and then conducted a three phases trail to investigate the confounding effects. Phase1was designed to evaluate the impact of elevated bilirubin level on intraoperative monitoring of noninvasive hemoglobin by using Masimo Radical-7software version V7.6.0.4.Phase2was designed to assess the contribution of different surgical procedure and anesthesia method in intraoperative monitoring of noninvasive hemoglobin by using updated software version V7.8.0.1. Results from Phase land Phase2would support the assessments in Phase3, whichtested the accuracy of noninvasive hemoglobin measured by two different software versions mentioned above.Methods, Results and Conclusions1. Accuracy of intraoperative noninvasive hemoglobin monitoring influenced by elevated bilirubin levelMethods:Patients undergoing abdominal surgery with elevated bilirubin level have recruited as jaundice group, patients undergoing liver surgery without elevated bilirubin level have recruited as control group.All patientsreceivedinvasive arterial pressure monitoring and the probes of Masimo Radical-7were fixed on the second finger on the same side as arterial catheter. The intraoperative monitoring of noninvasive hemoglobin by using software version V7.6.0.4. When noninvasive hemoglobin value were recorded, time matched blood samples were taken to ABL-800Fleximmediately for blood gas analysis.The accuracy of noninvasive hemoglobin of two groups will be assessed respectively by comparingthe time matched data pairs.The concordance between the two measurements were evaluated by themodified Bland-Altman method with non-constant bias. Matching on propensity scores was used to balance jaundice groupacross measured covariates.Results:Bland-Altman analysis presented a bias of1.03g/d for data collected from jaundice group withthe95%limits of agreementof-1.39-3.45g/dl.A non-constant biaswith negative tendencywere also found in this group. The control group yielded a bias of1.08g/dl with95%limits of agreement-1.88-4.05g/dl and presentedno significantlynon-constant bias. The propensity score matched cohort yielded a bias of1.21g/dl with95%limits of agreement-1.37-3.80g/dl and presented a non-constant biaswith positive tendency. The increased bilirubin level did not seem to alter the noninvasive to invasive hemoglobin bias.Conclusions:Elevated bilirubin level seems hardly effected the overall accuracy and precision ofnoninvasive hemoglobin measurement. The presence of hyperbilirubinemia may generate a decreased measurement error along with the increased hemoglobin concentration.2. Comparison of the accuracy of noninvasive hemoglobin monitoring in patients undergoing liver and spine surgery Methods:The liver surgery group enrolled patients undergoing liver surgery without elevated bilirubin level, the spine surgery group enrolled patients undergoing complex surgery without elevated bilirubin level.Method of data collection adopted inthis study phasewas the same asthe first study phase.The intraoperativemonitoring of noninvasive hemoglobin by using updated software version V7.8.0.1.A correction constant was available for the new version V7.8.0.1to adjust the baseline of hemoglobin agreements. Only the data pairs after adjusting were included in the final analysis.The statistic method was the same as that mentioned above. Result:The bias of liver surgery group was0.30g/dl, the corresponding95%limits of agreement were-1.18-1.78g/dl. This group presented a non-constant biaswith positive tendency.For the spine surgery group the bias was0.20g/dl withthe corresponding95%limits of agreement were-1.12-1.52g/dl. No significantlynon-constant biaswas found in this group.The difference of PI values between two groups was statistically significant (P<0.05by two independent samples t test).Conclusions:The diverse surgical processes and anesthesia method adopted in liver and spine surgery have been identified had little influence on the overall measurement error of noninvasive hemoglobin measurement. However, the noninvasive hemoglobin monitor in liver surgery will more likely overestimate the true hemoglobin concentration. This may be associate with the frequent occurrence of PI during liver surgery.3. Comparison of the accuracy of noninvasive hemoglobin monitoring measuredby two different Masimo Radical-7software versionsMethod:The data of two groups for each of the first and second study phases (monitored by software version V7.6.0.4or V7.8.0.1respectively) cannot be joined together to discuss the impact of updated software version because it has been proved that elevated bilirubin level and spine surgerygenerated avariation of bias and95%limits of agreement by comparing with the result of liver surgery.The statistic method was the same as the first study phase.Result:Thepropensity score matched cohort from software version V7.6.0.4presentedbias and95%limits of agreement were0.93g/dl and-1.72-3.58g/dl respectively.Forsoftware version V7.8.0.1group the correspondingbias and95%limits of agreement were0.30g/dl and-1.18-1.78g/d1.Non-constantbias with a very similar tendency to software version V7.8.0.1group also appeared in the propensity score matched cohort. The ratiofor noninvasive hemoglobin as a right guidance of transfusion was84.4%for software version V7.8.0.1group and59.4%for the propensity score matched cohort.Conclusions:Theearlier software version V7.6.0.4have shown a much more greater bias than the similar studies carried out in American and Europe, but itsaccuracy and precision were very closeto a study carried out in the East Asian region.This factsuggested that the measurement error of non-invasive hemoglobin concentration might be influenced by ethnic differences. Our results demonstrate the updated software version V7.8.0.1greatly improved the accuracy of noninvasive hemoglobin and was accurate enough to management transfusion. Removingthe individual differencesby correction system may be the possible reason of this improvement. |
PDF Full Text Request