Participants
Through a departmental quality review of transfusion practices, patients undergoing spine surgery on multiple spinal segments and cytoreductive surgery were identified; these surgeries were more likely to be associated with a sufficiently large blood loss volume to trigger anemia. The inclusion criteria were patients aged from 18 to 80 years who were scheduled for spine surgery or cytoreductive surgery, for whom the estimated blood loss was more than 15% of their total blood volume and who required an arterial catheter as a part of the standard care procedures for continuous arterial pressure monitoring and intermittent blood analysis. The exclusion criteria were an American Society of Anesthesiologists physical status score > 3, an inability to use their upper extremities for SpHb monitoring, preoperative anemia (male laboratory vein Hb < 12 g/dl and female < 11 g/dl), hepatic insufficiency (phosphatase alkaline, aspartate aminotransferase, alanine aminotransferase > 2 times the normal values), impaired renal function (serum creatinine > 1.5 mg/dl), coagulation disorders (activated partial thromboplastin time > 1.5 times the normal value or taking antiplatelet/anticoagulation drug) and pregnancy.
Interventions
SpHb monitoring group
In the SpHb monitoring group, an adhesive sensor (R2-25a), connected to the Radical-7® Pulse CO-Oximeter (software version V7740, Masimo Corp., Irvine, CA), was placed on the proximal third of the nail bed of the second, third, or fourth finger of the hand on the side opposite the arterial catheter before the induction of general anesthesia. The adhesive portion of the sensor was applied according to the directions for use and was covered with opaque shields to prevent optical interference. If the perfusion index, which is an indicator of localized perfusion, was < 1%, then the sensor position was recalibrated by switching the monitor off and on.
The baseline SpHb was recorded after the SpHb was stable for at least 15 min following induction of anesthesia. To obtain a time-matched invasive Hb concentration, a blood sample was drawn through the radial arterial catheter placed in the wrist contralateral to the SpHb sensor. Then, the CoOxHb was obtained using a CO-Oximeter (Radiometer ABL800; Radiometer, Copenhagen, Denmark), which is the routine method for intraoperative Hb measurement in our hospital. An SpHb level 1 g/dl lower than the baseline was set as the threshold for the alarm. When the SpHb monitor sounded the alarm or the surgery ended, both the SpHb and CoOxHb were recorded simultaneously. When clinicians deemed it necessary to take a Hb measurement but the pulse CO-Oximetry did not generate an alert, blood was drawn and tested to ensure the patient’s safety.
Standard care group
In the standard care group, the first blood sample was drawn from the radial arterial catheter and tested for CoOxHb at 20 min following the induction of anesthesia as the baseline. Other Hb measurements during surgery were ordered at the discretion of the clinicians.
Intraoperative and postoperative transfusion decisions in both groups were made by the clinicians considering both the CoOxHb value and the clinical situation.
Primary and secondary outcomes
The hypothesis of the study was that SpHb monitoring could more appropriately estimate the timing for further Hb measurement than clinicians during surgery without creating a delay in treatment. The primary outcomes were the positive predictive values (PPVs) of SpHb monitoring and the clinicians’ perception of the decrease in CoOxHb. A true-positive finding was defined as the ability to detect a decrease in CoOxHb of 1 g/dl or CoOxHb < 10 g/dl, whereas a false-positive finding was defined as an inability to detect stable CoOxHb (Fig. 1).
The secondary outcomes included a delay in treatment, the absolute accuracy and precision of SpHb monitoring compared with the reference method (CO-oximeter), and the accuracy of the trend in SpHb compared with changes in CoOxHb. A delay in treatment was defined as a decrease in CoOxHb to 7 g/dl, which was the intraoperative transfusion threshold in China.
Sample size
The sample size calculation was performed using internet-based software (http://www.sample-size.net/) based on the comparison of two independent proportions using exact test. In our hospital, the PPV of the clinicians’ perception of a substantial decline (> 1 g/dl) in CoOxHb was estimated to be less than 50%. Based on the results of the pilot trial, the study was designed to raise the PPV from 50 to 90% with SpHb monitoring, assuming a power of 0.8 and a two-sided significance level (α) of 0.05. We calculated a required sample size of 20 per group. Considering that the incidence of the loss of a sufficient quantity of blood to lead to anemia during cytoreductive or spine surgeries was approximately 25%, we planned to enroll 160 patients in total (80 per group).
Randomization, allocation and blinding
The trial was randomized (1:1) created using a website-generated (https://www.random.org/coins/) allocation list. Group allocations were concealed in sequentially numbered opaque envelopes, which were opened after the patients entered the operating rooms. After identifying the random number, the research nurse provided SpHb monitoring for the patient or offered standard care following the protocol.
The investigator who performed the statistical analysis was blinded to the group allocation, whereas the research nurse and anesthesiologists were not blinded.
Statistical methods
Continuous variables are expressed as the means ± standard deviation (SD) when normally distributed and as interquartile ranges (25th to 75th percentile) when not normally distributed. The distribution of variables was checked using visual inspection of histogram. Variables were compared using the t test or Mann-Whitney U test when appropriate. Categorical variables are presented as numbers and percentages (%) and were compared using Fisher’s exact test.
In line with the primary objective, PPVs were presented as percentages and compared by Fisher’s exact test.
For the SpHb accuracy analysis, the Bland-Altman test for data pairs of SpHb and CoOxHb was used to compare bias (mean error), precision (standard deviation, SD) and limits of agreement (LOA) using MedCalc (version 18.11, MedCalc Software, Acacialaan, Ostend, Belgium). The agreement between the two assays was presented by data plotting.
For the SpHb trend accuracy analysis, a four-quadrant plot was generated to evaluate clinically significant directional changes [delta (Δ) Hb > 1 g/dl] and a regression analysis was performed to calculate correlation coefficients (r) and 95% CIs were reported for all data using R (version 3.5.1, The R Foundation, Welthandelsplatz, Vienna, Austria). The coefficient from the model could be interpreted as each unit increase of SpHb was associated with CoOxHb. A two-sided p-value less than 0.05 was considered statistically significant. Statistical analysis was carried out using IBM SPSS for Windows (version 22.0, IBM Corporation, Armonk, New York, USA).