Self‐warming blanket versus forced‐air warming blanket during total knee arthroplasty under spinal anaesthesia: A randomised non‐inferiority trial

Arthroplasty patients are at high risk of hypothermia. Pre‐warming with forced air has been shown to reduce the incidence of intraoperative hypothermia. There is, however, a lack of evidence that pre‐warming with a self‐warming (SW) blanket can reduce the incidence of perioperative hypothermia. This study aims to evaluate the effectiveness of an SW blanket and a forced‐air warming (FAW) blanket peri‐operatively. We hypothesised that the SW blanket is inferior to the FAW blanket.


Editorial Comment
Clinical efficacy of two different patient warming apparati during spinal anaesthesia and knee arthroplasty was assessed. The authors chose non-inferiority as the analysis form, but due to necessary protocol deviations, the results were mixed.

| INTRODUCTION
Arthroplasty patients are at high risk of developing intraoperative hypothermia 1 and resultant well-known adverse effects. [2][3][4][5] The incidence of inadvertent hypothermia can be reduced by pre-operative warming. 6 The most widely used warming method is the forced-air warming (FAW), which has been shown to be effective in perioperative settings. 7,8 An active single-use self-warming (SW) blanket (Barrier ® EasyWarm ® , Mölnlycke Health Care AB, Gothenburg, Sweden, model 629900) is an option for warming patients peri-operatively. 9 The SW blanket consists of 12 warming pads (13 Â 10 cm) containing iron powder which oxidises when exposed to air and heat is produced. The warming pads reach an average temperature of 40 C within 30 min, with a maximum temperature of not more than 43 C. The temperature of the pads is maintained for up to 10 h.
The SW blanket offers advantages over the FAW method in certain important aspects. For example, the SW blanket does not need additional equipment or electric current, it is noiseless, does not generate air turbulence and is readily available. The superiority of the SW blanket over standard hospital clothing has been reported elsewhere. 10 However, other randomised controlled trials which have compared the peri-operative use of the SW blanket with the FAW blanket under spinal anaesthesia (SA) or general anaesthesia (GA) have yielded conflicting results. [11][12][13] The aim of this trial was to investigate whether the SW blanket is non-inferior to the FAW blanket in maintaining normothermia in patients undergoing elective unilateral total knee arthroplasty (TKA).
We hypothesised that the SW blanket is inferior to the upper-body FAW blanket.

| Participants and randomisation
The study population comprised patients who were scheduled for primary unilateral TKA under SA in Coxa Hospital for Joint Replacement, Tampere, Finland. Inclusion criteria were age 40-90 years, American Society of Anesthesiologists ASA physical status I-III and body mass index from 25 to 40. 14 Exclusion criteria were GA or inability to give written consent. Assessment for eligibility was verified in advance (S-LL). After arriving at the lounge of the surgical ward on the day of the surgery, patients were recruited (S-LL, JK, AA) and equally and randomly allocated to have either an SW blanket (SW group) or an upper-body FAW blanket (3M™ Bair Hugger™, Arizant Healthcare Inc, Eden Prairie, MN, USA; model 62200; FAW group) according to a sealed randomisation envelope which was opened after consent was obtained. A computer-generated list of random numbers was used.

| Study design
In accordance with the hospital's standard of care, patients were given paracetamol 1 g and cetirizine 10 mg per os as pre-medication. In addition, two thirds of the patients were given extended-release oxycodone/naloxone 5/2.5 mg or 10/5 mg per os. Patients waited for the surgery in the lounge where they were picked up to the preoperative holding area. Before SA induction, patients were pre-warmed in supine position for 30 min in a preoperative holding area. After pre-warming, standard monitoring (non-invasive blood pressure, electrocardiography and pulse oximetry) was applied, intravenous access was opened, and SA was induced with isobaric bupivacaine

| Warming and temperature control
The SW blanket was opened and unfolded at least 30 min before use. It was placed on the patient's entire body for pre-warming.
The upper-body FAW blanket was set to 38 C and placed longitudinally on the patient for pre-warming. To ensure sterile conditions, active warming was discontinued during induction of SA. The FAW was turned off during transfer of the patient to the OR and the RR. Both blankets were applied on the upper-body and arms, and a warm cotton blanket was placed over the non-surgical limb in the OR. The FAW was turned on at 38 C after surgical draping. The patient's head was left uncovered. Intravenous fluids were taken from the warming cabinet (Termaks AS, Bergen, Norway, model B 9420) set to 37.5 C, but the fluids were not warmed in the OR. Hypothermia was defined as core temperature below 36.0 C, as indicated in the NICE guideline. 16 If intraoperative hypothermia existed, the patient, regardless of the study group, was warmed with an upper-body FAW blanket set to 43 C. Postoperatively, FAW was initiated when a patient shivered or felt cold, or the measured core temperature was below 36.0 C. Active warming was discontinued at any time when the core temperature exceeded 37.0 C or the patient complained of discomfort.

| Statistical analysis
The primary outcome measure was the core temperature value on admission to the RR. Secondary outcomes were core and peripheral temperature changes peri-operatively, the usability and convenience of the warming methods, related costs and postoperative complications.
The non-inferiority design was chosen with the widely used FAW method as an active control. 17 We predetermined an inferiority margin to be 0.2 C, since the margin had to be narrower than usually used 0.5 C in superiority trials 12 and further, 0.2 C was the accuracy of the ZHF thermometer. 18 For the sample size calculation, a SD of 0.4 C was chosen 10 with an alpha error of 0.05 and a power of 0.9.
The sample size of 69 patients per group was calculated. To allow for dropouts, 75 patients were enrolled for each study group.
Initially, all patients were allocated the warming blanket according to randomisation and there were no changes between groups. The results were analysed using intention-to-treat analysis. To evaluate the thermal redistribution, we applied Ramanathan's formula, 15 where four peripheral temperatures are used to calculate mean skin temperature (MST): To evaluate body heat content, mean body temperature was calculated by using Burton's formula 19 : Temperatures were measured as degrees Celsius. A p < .05 was considered statistically significant.

| RESULTS
The data were collected between November 2018 and November 2019. The follow-up lasted until May 2020. During the study period, 155 patients were screened for eligibility. Of these, five declined to participate in the study. Enrolled patients were randomised to either the SW group (n = 75) or the FAW group (n = 75). Data of three patients were excluded from the final analyses ( Figure 1).
Both study groups had similar patient characteristics (Table 1).
Moreover, there was no difference between groups in the length of the waiting, pre-warming, surgery and transition times ( Table 2).
The ambient temperature did not differ between the study groups on admission to the OR. However, intraoperative ambient temperature differed statistically significantly between groups. The other relevant peri-operative data were the same in each group (Table 3).
Similar core temperature changes were observed in both groups ( Figure 2). Core temperature did not change in initially normothermic patients but rose to more than 36 C in all, but one hypothermic patient during pre-warming. Core temperature remained unchanged after SA induction but decreased after admission to the OR. Intraoperative hypothermia was common: 45 patients (61%) in the SW group and 36 patients (49%) in the FAW group became hypothermic intra-operatively and needed rescue warming with the upper-body FAW set to 43 C (Table 3).
Hypothermia developed during the first 30 min in the OR and per-

| DISCUSSION
To the best of our knowledge, this is the first study comparing the SW blanket and the FAW blanket used in a similar manner during the pre-warming and intraoperative periods. The results of our study reveal that the SW blanket and the upper-body FAW blanket set to 38 C both were capable of increasing MBT and warming up initially hypothermic patients during pre-warming. Despite active pre-warming, intraoperative hypothermia was unexpectedly common in both groups, and rescue warming with the FAW method was needed in 61% of patients in the SW group and in 49% of patients in the FAW group. The primary outcome of the study, the core temperature value on admission to the RR, did not differ between groups.
SA is known to predispose patients to hypothermia. 20,21 However, the induction of SA has been shown to decrease core temperature less than the induction of GA. 22 In our study, the core temperature remained unchanged in both groups after SA induction. This might have resulted from reduced heat redistribution due to effective pre-warming and from the moderate amount of isobaric bupivacaine limiting the level of spinal block.
The high incidence of hypothermia observed in the present study might have been the result of the over 20 min interruption between preoperative and intraoperative warming. 23 This was most evident in the FAW group, as active warming was discontinued during transfer to the OR and during surgical preparation. In contrast, patients in the SW group had the blanket without interruption. The patients in our study became only slightly hypothermic. Our findings are therefore in accordance with those of previous studies that pre-warming does not prevent intraoperative hypothermia completely, but reduces its severity. 24,25 Intraoperative hypothermia might also be caused by the ambient temperature being colder than recommended, 16 as the core temperature drop was only observed in the OR. However, when patients are warmed with forced air, the ambient temperature has been shown to have a negligible effect on core temperature. 26  On the other hand, using the SW blanket for pre-warming during waiting and transferring to the OR seems to keep the core temperature more balanced than the FAW, which has to be interrupted for the course of transfer and surgical draping. Also, the same FAW blanket seldom lasts from the OR to the RR due to staining and tearing.
In the present study, active warming was well tolerated, although a few patients in each group felt excessively warm regardless of the core temperature value. Subjective thermal perception is largely determined by skin temperature. 27 Therefore, the FAW device was set to 38 C during pre-warming to avoid discomfort and sweating. In addition, the medium set temperature has been shown to be as effective as the high set temperature in increasing the heat content of peripheral tissues. 28 It has been argued that the FAW may increase the risk of surgical site infection (SSI). 29,30 However, hypothermia itself is known to be a risk factor for SSI. 31  There are some significant limitations in our study that should be considered. The choice of primary outcome seemed to be unsuitable since intraoperative core temperature change and the duration of hypothermia have been used to assess the thermal condition or the hypothermia burden of surgical patients rather than a single temperature value. 33 Further, the incidence of intraoperative hypothermia was unexpectedly high and according to the study protocol, all hypothermic patients, regardless of study group, were warmed with the upper-body FAW blanket set to 43 C. Thus, core temperature on admission to the RR does not reflect the warming effect of the SW blanket and the upper-body FAW blanket set to 38 C. Moreover, our clinical practise resulted in long interruptions between pre-warming and intraoperative warming, leading to temperature decrease especially in the FAW group. Although the schedule of the day was predetermined, the actualised course of a single patient depended on the time schedules of the operative unit.
In this study, non-inferiority of the SW blanket was not demonstrated when comparing intraoperatively normothermic patients, that is patients who only were warmed either with the SW or FAW blanket, as the lower CI À0.23 was outside the predetermined inferiority margin. In future, to assess the non-inferiority of the SW blanket to the FAW method more reliably, the threshold core temperature for inadvertent hypothermia should be defined as 35.5 C, since according to the latest findings this temperature has been shown to be sufficient for surgical patients. 34 In conclusion, based on our results, both methods were effective in raising MBT during pre-warming. However, although the SW blanket was statistically non-inferior to the FAW method, hypothermia was more frequent in the SW group, requiring rescue warming as we strictly held to the NICE guideline.