Authors: Yusuke Iizuka, Koichi Yoshinaga, Takeshi Nakatomi, Kyosuke Takahashi, Kyoko Yoshida, Masamitsu Sanui
Categories: Original Article, Capillary refill time, general anesthesia, operating room, peripheral perfusion index
Source: Saudi Journal of Anaesthesia
Capillary refill time (CRT) is the gold standard for evaluating peripheral organ perfusion; however, intraoperative CRT measurement is rarely used because it cannot be conducted continuously, and it is difficult to perform during general anesthesia. The peripheral perfusion index (PI) is another noninvasive method for evaluating peripheral perfusion. The PI can easily and continuously evaluate peripheral perfusion and could be an alternative to CRT for use during general anesthesia. This study aimed to determine the cutoff PI value for low peripheral perfusion status (prolonged CRT) by exploring the relationship between CRT and the PI during general anesthesia.
We enrolled 127 surgical patients. CRT and the PI were measured in a hemodynamically stable state during general anesthesia. A CRT >3 s indicated a low perfusion status.
Prolonged CRT was observed in 27 patients. The median PI values in the non-prolonged and prolonged CRT groups were 5.0 (3.3–7.9) and 1.5 (1.2–1.9), respectively. There was a strong negative correlation between the PI and CRT (r = −0.706). The area under the receiver operating characteristic curve generated for the PI was 0.989 (95% confidence interval, 0.976–1.0). The cutoff PI value for detecting a prolonged CRT was 1.8.
A PI <1.8 could accurately predict a low perfusion status during general anesthesia in the operating room. A PI <1.8 could be used to alert the possibility of a low perfusion status in the operating room.
University Hospital Medical Information Network (UMIN000043707; retrospectively registered on March 22, 2021, https://center6.umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno = R000049905).
Keywords: Capillary refill time, general anesthesia, operating room, peripheral perfusion index
It is important to maintain adequate organ perfusion during general anesthesia. Currently, the management of sufficient perfusion involves fluid therapy to avoid hypovolemia and blood pressure control to maintain sufficient organ perfusion pressure. However, these management strategies do not involve the direct evaluation of organ perfusion. There are several methods for evaluating organ perfusion, with the capillary refill time (CRT) being the gold standard for evaluating peripheral organ perfusion.[1] CRT is a noninvasive parameter that is easy to evaluate because of the predetermined cutoff values; therefore, it is widely used, especially in critical care. CRT measurement during general anesthesia is rarely applied, possibly because it cannot be used continuously and it is difficult to perform intraoperatively when the peripheral areas of patients, including the fingers, are covered. Therefore, there is need for an appropriate index of organ perfusion during general anesthesia other than CRT.
The peripheral perfusion index (PI) is another noninvasive method for evaluating peripheral perfusion. The PI is derived from the plethysmographic signal obtained through pulse oximetry. The plethysmographic signal has two components, i.e., a pulsatile component that reflects changes in the finger blood volume during one cardiac cycle, which may depend on changes in stroke volume, and a non-pulsatile component that is related to light absorbed by tissues, including connective tissue, bone, and venous and capillary blood.[2] The PI can easily evaluate peripheral perfusion continuously and could comprise an alternative to CRT during general anesthesia; however, the relationship between CRT and the PI and the optimal cutoff PI value to detect a low perfusion status have not been determined. However, an appropriate cutoff PI value for detecting prolonged CRT continuous PI monitoring could be used to alert low systemic organ perfusion.
This study aimed to determine the cutoff PI value for low peripheral perfusion status (CRT >3 s) by exploring the relationship between CRT and the PI during general anesthesia.
This single-center prospective study was approved by the Research Ethics Committee of Saitama medical center Jichi medical university (S20-125, 17/12/2020). Because of the noninvasive nature of the study, the requirement for written informed consent was waived by the Research Ethics Committee of Saitama medical center Jichi medical university. This study was conducted in accordance with the Declaration of Helsinki and registered in the University Hospital Medical Information Network (UMIN000043707, 22/3/2021). We enrolled patients scheduled to undergo surgery in our institution. The inclusion criteria were age >20 years, scheduled elective surgery, or urgent surgery. The exclusion criteria were necrosis of the extremities, septic shock, or surgery with a brachial plexus block.
We applied standard monitoring that comprised 3- or 5-lead electrocardiography, noninvasive blood pressure monitoring, and pulse oximetry. Propofol, remifentanil, and rocuronium were used for anesthesia induction and intubation. Anesthesia was maintained using propofol or remimazolam or sevoflurane or desflurane and remifentanil. The anesthesiologists were allowed to administer fluid and intermittent or continuous vasopressors (phenylephrine or noradrenaline) to maintain mean arterial pressure above 65 mmHg. A forced-air warming device covering the fingers was used for all fingers during the surgery. The PI was measured using Radical-7 (Masimo Corp., Irvine, CA, USA) at the index finger and set to the “long-term” method, which displays the averaged PI value over 30 s. Measurements were performed between anesthesia induction and end of general anesthesia during a hemodynamically stable state (defined as changes in mean arterial pressure <10% over 5 min and systolic arterial pressure lower than 180 mmHg). After measuring the heart rate, systolic/mean/diastolic blood pressure, and PI, we measured the CRT using the middle finger in the same hand where the PI was recorded. We compressed the ventral surface of the middle finger distal phalanx using a 10-ml syringe (TERUMO Corp., Hatagaya, Tokyo, Japan) filled with 10 ml of air compressed to fit a 7-ml volume and maintained the pressure for 10 s [Figure 1].[3] After releasing the pressure, the skin's return to normal color was recorded using the video camera of an iPhone 7^TM^ (Apple Inc., Cupertino, CA, USA). IY and YK, who were blinded to the PI value, determined the CRT by carefully observing the video. A CRT >3 s indicated low perfusion.[4]
Figure 1 Method for compressing the ventral surface of the middle finger distal phalanx using a 10-ml syringe filled with 10 ml of air compressed to fit a 7-ml volume
Data are expressed as median (25–75% interquartile range) or mean (standard deviation) as appropriate. One hundred and twenty patients were required to demonstrate that a low PI value could predict low perfusion with an area under the receiver operating characteristic (ROC) curve ≥0.75 (type I error of 5% and type II error of 20%) in a situation where 10% of the patients would have low perfusion. The Mann–Whitney U test and Student's t test were used to compare patients with or without low perfusion according to the normality of the data distribution. Spearman's rank correlation was used to evaluate the relationship between CRT and the PI. ROC curves were constructed for the PI's ability to detect prolonged CRT maximizing sensitivity and specificity using Youden's index. We determined the inter-observer reproducibility for CRT using Bland–Altman plots, described as mean bias. Statistical significance was set at P < 0.05. Statistical analyses were performed using EZR version 3.6.3[5] and SPSS version 26.
Between January 2021 and July 2021, we enrolled 127 patients. Table 1 shows the patient characteristics. Twenty-seven patients showed a CRT >3 s. There were no significant differences in the measured parameters, except for heart rate, between the groups with and without prolonged CRT. The median PI values in the groups with and without prolonged CRT were 5.0 (3.3–7.9), and 1.5 (1.2–1.9), respectively [Table 2]. There was a strong negative correlation between the PI and CRT (r = −0.706) [Figure 2]. Table 3 shows the ability of the PI to detect prolonged CRT. The area under the ROC curve generated for the PI was 0. 989 (95% CI, 0.976–1.0) [Figure 3]. The cutoff PI value for detecting prolonged CRT was 1.8 (sensitivity 1.0; specificity, 0.94). Bland–Altman analysis showed fair concordance between the CRT estimation performed by the two investigators, with a mean bias of 0.37 and limits of agreement between − 1.4 and 2.1 [Figure 4]. In the prolonged CRT group, some values were out of the limits of agreement.
Figure 2 Scatter plots of the peripheral perfusion index and capillary refill time
Figure 3 Receiver operating characteristic curve for the perfusion index and prolonged capillary refill time
Figure 4 Bland–Altman plot for the measurements of both investigators. There is a mean bias of 0.37 with limits of agreement between − 1.4 and 2.1
In this study, we observed a strong negative correlation between CRT and the PI; moreover, we found that a PI value <1.8 could accurately predict low perfusion during general anesthesia. To our knowledge, this was the first study to evaluate the relationship between intraoperative CRT and the PI.
Few studies have investigated the relationship between CRT and the PI outside the operating room. Lima et al.[6] reported that the PI could strongly predict prolonged CRT in critically ill patients, with a cutoff PI value of 1.4. Our results showed that a cutoff PI value of 1.8 could predict prolonged CRT. This difference could be attributed to the influence of general anesthesia. Anesthetic agents and standard air warming devices for preventing intraoperative hypothermia induce vasodilation, which increases the PI. Specifically, studies have reported intraoperative PI values of 6–9 with standard volatile anesthetic agents and preoperative PI values of approximately 1.5. Among-study differences in vasodilation in the study populations might affect the cutoff values.[7] As suggested by our findings, the cutoff PI value for predicting low perfusion could differ between operating rooms and intensive care units.
Recently, a retrospective observational study by Agerskov et al.[8] reported that an intraoperative low PI (mean PI ≤1.5) was associated with severe postoperative complications or death (odds 1.65 (95% CI 1.20–2.27; P = 0.002) in acute surgical patients. They also reported that each 15-min increase in intraoperative time spend with a low PI was associated with severe postoperative complications or death. These results highlight the importance of continuous PI monitoring to avoid complications; however, they defined thresholds for low values of PPI at 1.5 empirically with no reference standard; the appropriate cutoff value to warn of low perfusion during general anesthesia remains unknown because of the low predictive ability of postoperative complications (a cutoff of mean intraoperative PI 1.5 yielded a sensitivity of 0.72 and specificity of 0.37).[8]
Our results have several potential implications regarding intraoperative management. A low perfusion status in the early postoperative period is associated with perioperative complications.[9,10,11] Maintaining an adequate perfusion status from the intraoperative to the postoperative period can decrease avoidable complications. The PI can be continuously evaluated, and a PI <1.8 could suggest an intraoperative low perfusion status. There are serious limitations to the current strategy of maintaining organ perfusion by evaluating fluid responsiveness and maintaining mean arterial pressure using vasopressors.[12] Using fluid responsiveness to guide fluid therapy can be influenced by the vasodilation degree; therefore, volume overload may occur during general anesthesia that induces vasodilation. However, there are no good indicators for adjusting the vasopressor dose to correct the vasodilation. Indicators for guiding the use of vasopressors are required for maintaining organ perfusion. A cutoff PI value >1.8 may be a good indicator to guide intraoperative hemodynamic management. If the PI is maintained above 1.8 during general anesthesia with hypotension, a vasopressor, rather than a fluid bolus, can be initially used. If the PI decreases below 1.8 during anesthesia with hypotension, fluid administration may be initially performed. Future studies should examine the advantages of a strategy combining the current management and the PI as a perfusion parameter.
Our study had several limitations. First, CRT and the PI do not directly evaluate organ perfusion. There remain no established indicators of systemic perfusion, and we can only infer systemic perfusion from peripheral perfusion. We selected CRT as the gold standard for evaluating peripheral perfusion; however, prolonged CRT may not represent inadequate organ perfusion during general anesthesia. Second, we did not record the temperature of the fingers. A low temperature might cause decrease in the PI not attributable to low systemic perfusion.[2] However, a forced-air warming device covering the fingers was used for all fingers. Extremely low digital temperatures are very rare. Third, the PI could be affected by surgical stimuli, anesthetic depth, and use of vasopressors, which are not related to systemic hypoperfusion.[2] However, there was no significant between-group difference in almost all hemodynamic parameters, including the bispectral index, which is the index of anesthetic depth, and the proportion of use of continuous vasopressors. This could be attributed to the efforts of anesthesiologists to maintain adequate mean arterial pressure (above 65 mmHg) and anesthetic depth in all patients. There appears to be a minimal possibility of extreme vasoconstriction inducing a low PI owing to inadequate anesthetic depth. Only the heart rate was significantly higher in the prolonged CRT group than in the non-prolonged CRT group, which could be attributed to the small stroke volume common in the hypoperfusion status. However, the cause of this between-group difference in heart rate is unknown since we did not measure the stroke volume. Fourth, the PI and CRT were measured in patients at different body positions; however, most patients underwent surgery in the supine position. In our study, the PI and CRT might have been affected by the body position rather than by organ hypoperfusion.[13] Future studies should examine the relationship between organ and peripheral perfusion at several body positions. Finally, we found that some values were out of the limits of agreement when evaluating inter-observer reproducibility in the prolonged CRT group. In fact, the mean bias was 0.134 in non-prolonged CRT group; however, the mean bias was 1.267 in the prolonged CRT group. This disagreement may be attributed to the difficulty to define the point where the color of the ventral surface of the finger becomes normalized, especially in the prolonged CRT group. The effect of the minor disagreement in the values was not negligible, although the mean bias was small in the non-prolonged CRT group. In our analysis for the CRT definition, one investigator detected 28 values as CRT >3 s, and the other detected 17 values as CRT >3 s, although 27 values were considered to indicate a low perfusion status based on the averaged CRT. However, the high sensitivity of the PI to detect low perfusion may be advantageous because it should be used as an alarm to inform the possibility of intraoperative hypoperfusion; 1.8 may be a reasonable value because 1.8 is the cutoff value where the sensitivity decreased from 100% for both investigators. The effect of the discrepancy in CRT between the evaluators might not have significantly affected the results.
Our findings indicated a strong relationship between CRT and the PI and showed that a PI <1.8 could accurately predict low perfusion during anesthesia. A PI <1.8 could be used to alert the possibility of a low perfusion status in the operating room.
CRT: capillary refill time; PI: peripheral perfusion index; ROC: receiver operating characteristic.
A peripheral perfusion index <1.8 could accurately predict prolonged capillary refill time (the area under the receiver operating curves generated for PI was 0. 989). A peripheral perfusion index <1.8 suggests prolonged capillary refill time and could indicate low perfusion status in the operating rooms.
Nil.
There are no conflicts of interest.
We would like to thank Editage (www.editage.com) for English language editing.