Home > Clinical Guides > The Management of Inadvertent... > RISK FACTORS FOR INADVERTENT...

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Nursing and Supportive Care (UK). The Management of Inadvertent Perioperative Hypothermia in Adults [Internet]. London: Royal College of Nursing (UK); 2008 Apr. (NICE Clinical Guidelines, No. 65.)

7RISK FACTORS FOR INADVERTENT PERIOPERATIVE HYPOTHERMIA

Clinical question

What risk factors contribute to perioperative hypothermia?

Background

Perioperative hypothermia develops in three characteristic phases: a rapid decrease in core temperature in the first hour due to core to peripheral redistribution of body heat – which is mediated by the use of volatile anaesthetic agents; a slow linear decrease in core temperature due to heat loss exceeding metabolic heat gain; a plateau in temperature in which vasoconstriction decreases heat loss from the skin. The pattern of development of these phases will be influenced by risk factors.

Numerous factors contribute to the risk of inadvertent perioperative hypothermia. Risk is perceived to depend on patient characteristics (such as age or BMI); surgery factors (such as magnitude of the procedure or whether body cavities are open); anaesthesia factors (such as type or duration of anaesthesia); perioperative pharmacological agents (such as premedication); environmental factors (e.g. theatre temperature) and any preventative measures (such as the use of forced air warming devices). Risk factors are not necessarily independent and combinations of risk factors may be important, for example, patient age may be a relevant factor only for long surgical procedures. Furthermore, for continuous variables, such as age, there may be thresholds above which inadvertent perioperative hypothermia (IPH) is more likely to occur.

It may also be important to distinguish between factors that make the patient more likely to become hypothermic, and those that put the patient at greater risk of prolonged hypothermia and/or at greater risk of complications from hypothermia because of their inability to recover quickly from the hypothermic state. Where possible we will examine the incidence of hypothermia and the rate of recovery.

The purpose of this review is to examine systematically the evidence base to elucidate the variety of factors that contribute to an individual’s risk of inadvertent perioperative hypothermia. This will provide a framework for targeting limited resources, if necessary, to those individuals at highest risk.

The risk factors review is split into two: one concerned with hypothermia risks associated with pharmacological agents used perioperatively for any purpose (Section 7.1), and the other covering all other risk factors (Section 7.2).

7.1. RISK FACTORS FOR IPH: PHARMACOLOGICAL AGENTS

SELECTION CRITERIA

Types of study design

Pharmacological agents as risk factors should be examined primarily in randomised trials because they are interventions.

Types of intervention

Any pharmacological agent used perioperatively. This includes, but is not restricted to, the following drug classes:

Premedications:

Reversal of benzodiazepines:

  • Benzodiazepine antagonists (e.g. flumenazil; used to reverse the effects of benzodiazepines and counter the unwanted effects of anaesthetics, in order to speed recovery of motor and cognitive function).

Muscle relaxants:

Reversal of muscle relaxants:

Induction of anaesthesia:

  • Barbiturate (e.g. thiopentone);
  • N-methyl-D-aspartate (NMDA) receptor antagonist (e.g. ketamine; used for induction of anaesthesia and analgesia).

General anaesthesia drugs:

  • General anaesthesia drugs (e.g. halothane, isoflurane, propofol).

Analgesia (for pain control):

  • Opioid (e.g. pethidine);
  • Other centrally-acting analgesics (e.g. tramadol, nefopam).

Control of nausea:

Types of comparison

The following comparisons were to be included:

  • Intervention versus placebo / no intervention;
  • Intervention 1 + intervention 2 versus intervention 2 alone;
  • Drug A versus drug B (both drugs in same class);
  • Duration 1 versus duration 2;
  • Dose 1 versus dose 2.

It was decided to combine the two types of comparison: (i) intervention versus placebo / no intervention and (ii) intervention 1 + intervention 2 versus intervention 2 alone, and examine this assumption using sensitivity analyses.

Outcomes

Studies were to be included if they reported either core temperature intra- or post-operatively, or the incidence of inadvertent perioperative hypothermia. Studies reporting only the incidence of shivering were excluded.

Stratification and subgroup analyses

We planned to stratify the studies by the following:

  • Classes of drugs;
  • Perioperative phase of intervention;
  • Trauma patients – elective and emergency surgery considered together initially.

The following subgroups were to be considered:

METHODS OF THE REVIEW

Search strategy for identification of studies

Searches were performed on the following core databases: MEDLINE, EMBASE, CINAHL and The Cochrane Library (1966 to current day with guidance from the GDG). Additional databases were not searched for this review. The search strategies are given in Appendix B.

The titles and abstracts from the search strategy were assessed. Thirty studies met the inclusion criteria for the review. The reference lists of the retrieved studies were inspected for further potential papers, but none were identified. The excluded studies are listed in Appendix E, along with reasons for exclusion.

DESCRIPTION OF STUDIES INCLUDED IN THE REVIEW

30 studies met the inclusion criteria for the review (Alfonsi 1998; Bilotta 2002; Buggy, abstract; Cheong 1998; Crozier 2004; Delauney 1991; De Witte 1995; De Witte 1998; Goto 1999; Grover 2002; Holdcroft 1978; Hong 2005; Horn 1997; Horn 1998; Ikeda 2001; Kelsaka 2006; Kimberger 2007; Kinoshita 2004; Mao 1998; Mathews 2002; Matsukawa 2001; Mizobe 2005; Piper 2002; Piper 2004; Powell 2000; Röhm 2005; Sagir 2007; Stapelfeldt 2005; Toyota 2004; Weinbroum 2001).

Two studies were conducted in the UK (Holdcroft 1978; Powell 2000); 15 were in the rest of Europe; six in Japan; one in Israel; one in Kuwait; one in India; one in Singapore; one in South Korea; one in Taiwan and one in the USA.

Seventeen studies had 50 patients or fewer in each comparison (Alfonsi 1998; Buggy, abstract; Delauney 1991; De Witte 1995; De Witte 1998; Goto 1999; Grover 2002; Holdcroft 1978; Horn 1997; Horn 1998; Ikeda 2001; Kimberger 2007; Kinoshita 2004; Matsukawa 2001; Mizobe 2005; Stapelfeldt 2005; Toyota 2004), one of which had fewer than 20 patients (Mizobe 2005). Six studies had more than 100 patients in total (Cheong 1998: 80 in each arm; Hong 2005: 30, 30 and 29 in the three arms; Mathews 2002: 50 in each of three arms; Piper 2004: 73 to 76 in each of the five arms; Sagir 2007: 40 in each of three arms; Weinbroum 2001: 34 to 50 in each of six arms. Eighteen studies had more than 2 arms, giving a total of 66 comparisons.

Population and details of surgery and anaesthesia

The mean age (where given) ranged across the studies from 28 to 73 years. Toyota (2004) included participants from the age of 13 to 52 years (median around 26 years). From the mean and standard deviation (mean 33 years; SD 13 years for ketamine group; mean 37 years, SD 16 for propofol group), one other RCT may have included some children (Ikeda 2001).

Surgery was carried out under general anaesthesia in 19 studies (Alfonsi 1998; Buggy, abstract; Cheong 1998; Crozier 2004; Delauney 1991; De Witte 1995; De Witte 1998; Goto 1999; Grover 2002; Holdcroft 1978; Horn 1997; Horn 1998; Ikeda 2001; Mathews 2002; Piper 2002; Piper 2004; Powell 2000; Röhm 2005; Stapelfeldt 2005; Toyota 2004; Weinbroum 2001); regional anaesthesia in five studies (Bilotta 2002; Hong 2005; Kelsaka 2006; Kinoshita 2004; Mao 1998; Sagir 2007); mixed general and epidural anaesthesia in one study (Mizobe 2005) and in two studies the anaesthesia type was unclear (Kimberger 2007; Matsukawa 2001).

Anaesthesia duration was more than 1 hour in 12 studies (Cheong 1998; Crozier 2004; De Witte 1998; Ikeda 2001; Kelsaka 2006; Mathews 2002; Piper 2002; Piper 2004; Röhm 2005; Stapelfeldt 2005; Toyota 2004; Weinbroum 2001); less than 1 hour in one study (De Witte 1995), and not stated in 17 studies (Alfonsi 1998; Bilotta 2002; Buggy; Delauney 1991; Goto 1999; Grover 2002; Holdcroft 1978; Hong 2005; Horn 1997; Horn 1998; Kimberger 2007; Kinoshita 2004; Mao 1998; Matsukawa 2001; Mizobe 2005; Powell 2000; Sagir 2007).

The types of surgery in the studies were orthopaedic (Alfonsi 1998; Bilotta 2002; Buggy, abstract; Kelsaka 2006; Toyota 2004); gynaecological (De Witte 1995; Grover 2002; Holdcroft 1978); ENT (Crozier 2004; Horn 1997; Horn 1998); neurosurgical (Kimberger 2007; Stapelfeldt 2005); urological (Mao 1998; Sagir 2007); abdominal (De Witte 1998; Goto 1999; Mizobe 2005); mixed (Cheong 1998; Ikeda 2001; Kinoshita 2004; Mathews 2002; Piper 2002; Piper 2004; Powell 2000; Röhm 2005; Weinbroum 2001) or unclear (Delauney 1991; Matsukawa 2001). There was one indirect study (Hong 2005) in which the patients were undergoing Caesarean section; the study was only considered if there were insufficient data for direct populations.

Surgery lasted less than 30 minutes in one study (Grover 2002); 30 to 60 minutes in 3 studies (De Witte 1998; Hong 2005; Horn 1997); 1 to 3 hours in 11 studies (Alfonsi 1998; Bilotta 2002; Buggy 1997, abstract; Delauney 1991; Horn 1998; Ikeda 2001; Kelsaka 2006; Piper 2002; Piper 2004; Röhm 2005; Weinbroum 2001); over 3 hours in one study (Stapelfeldt 2005) and was not stated in 14 studies (Cheong 1998; Crozier 2004; De Witte 1995; Goto 1999; Holdcroft 1978; Kimberger 2007; Kinoshita 2004; Mao 1998; Mathews 2002; Matsukawa 2001; Mizobe 2005; Powell 2000; Sagir 2007; Toyota 2004).

Ten studies recorded tympanic temperatures, six rectal, six oesophageal, two bladder, three aural canal and two nasopharyngeal.

Risk factors

The following pharmacological agents were examined; where applicable, we have indicated if the anaesthesia is not general, but have analysed the studies together regardless of type of anaesthesia.

A. Premedication
B. Reversal of benzodiazepines
C. Muscle relaxants
D. Reversal of muscle relaxants
E. Induction of anaesthesia
F. General anaesthesia drugs
G. Analgesia
H. Control of nausea

Other warming during the studies

Some studies used other methods to warm all the patients:

Two studies gave the patients warmed sheets (Horn 1997; Horn 1998), which is likely to have a negligible warming affect. In the other studies, patients received no active warming (Alfonsi 1998; Bilotta 2002; Buggy 1997; Cheong 1998; Delauney 1991; Goto 1999; Holdcroft 1978; Hong 2005; Kimberger 2007; Kinoshita 2004; Mao 1998; Matsukawa 2001; Mizobe 2005; Piper 2004; Powell 2000; Röhm 2005; Toyota 2004).

Interventions

The comparisons were also separated by the perioperative phase in which the pharmacological agent was given.

A. Premedication
1. Alpha2-adrenergic antagonist (e.g. clonidine; used as a premedication)
Intervention versus placebo / no intervention

Preoperative phase:

Intraoperative phase:

2. Benzodiazepines (e.g. midazolam; used as a premedication)
Intervention versus placebo / no intervention

Preoperative phase:

Intervention 1 + intervention 2 versus intervention 2 alone

Preoperative phase:

Intraoperative phase:

B. Reversal of benzodiazepines
1. Benzodiazepine antagonists
Intervention versus placebo / no intervention

Intraoperative phase:

C. Muscle relaxants
1. Anti-muscarinic agents
Intervention versus placebo / no intervention and

Preoperative phase:

Intervention 1 + intervention 2 versus intervention 2 alone

Preoperative phase:

D. Reversal of muscle relaxants
1. Cholinesterase inhibitor
Intervention versus placebo / no intervention

Intraoperative phase:

E. Drugs for induction of anaesthesia
1. N-methyl-D-aspartate (NMDA) receptor antagonist
Intervention versus placebo / no intervention

Intraoperative phase:

Intervention 1 + intervention 2 versus intervention 2 alone

Intraoperative phase

Comparison of two drugs in different classes

Intraoperative phase:

F. General anaesthesia drugs
Comparison of two drugs in the same class

Intraoperative phase:

Different doses of same drug

All phases:

G. Analgesia
1. Opioid (e.g. pethidine; used for pain control)
Intervention versus placebo / no intervention

Intraoperative phase:

Intervention 1 + intervention 2 versus intervention 2 alone

Intraoperative phase:

Comparison of two drugs in the same class (opioids)

Intraoperative phase:

Different doses of same drug

All phases:

2. Other centrally-acting analgesics (e.g. tramadol, nefopam; used for pain control)
Intervention versus placebo / no intervention

Preoperative phase

Intraoperative phase

Intervention 1 + intervention 2 versus intervention 2 alone

Preoperative phase

  • Tramadol plus glycopyrronium versus glycopyrronium only (De Witte 1995).
Comparison of two drugs in the same class

Intraoperative phase:

Different doses of same drug

Intraoperative phase:

  • Nefopam 0.2mg/kg versus nefopam 0.1mg/kg (Piper 2004);
  • Nefopam 0.2mg/kg versus nefopam 0.05mg/kg (Piper 2004);
  • Nefopam 0.1mg/kg versus nefopam 0.05mg/kg (Piper 2004).
H. Control of nausea
1. Serotonin-receptor antagonist (e.g. dolasetron, ondansetron)
Intervention versus placebo / no intervention

Intraoperative phase:

Intervention 1 + intervention 2 versus intervention 2 alone

Intraoperative phase:

METHODOLOGICAL QUALITY

The quality assessment for the included trials is shown in Appendix D. An adequate method of randomisation was reported in six studies (computer generated: Bilotta 2002; De Witte 1998; Kimberger 2007; Matsukawa 2001; Mizobe 2005, table of random numbers; Cheong 1998). The other studies did not state the method of randomisation.

Allocation concealment (variants on the sealed envelopes method) was reported in nine studies (Crozier 2004 (partial); Hong 2005 (partial); Kimberger 2007 (adequate); Mathews 2002 (partial); Mizobe 2005 (partial); Piper 2004 (partial); Powell 2000 (partial); Sagir 2007 (partial); Stapelfeldt 2005 (partial)). Allocation concealment was not reported or unclear in the other studies.

All studies but four reported that the outcome assessors and the patients were blinded to the interventions; blinding was not stated in Goto 1999; Holdcroft 1978; Ikeda 2001; Kinoshita 2004.

Ten studies (Bilotta 2002; Hong 2005; Kelsaka 2006; Kimberger 2007; Piper 2004; Röhm 2005; Sagir 2007; Stapelfeldt 2005; Toyota 2004) described an a-priori power calculation. These calculations suggested that the sample size should be 30 patients per group (Bilotta 2002); 27 (Hong 2005); 24 (Kelsaka 2006); 16 (Kimberger 2007); 27 (Piper 2002); 73 (Piper 2004); 27 (Röhm 2005); 40 (Sagir 2007); 17 (Stapelfeldt 2005) and 15 (Toyota 2004). All studies used an intention to treat analysis.

All studies included in the review demonstrated baseline comparability of the groups on characteristics such as age, gender, duration of surgery, and ambient air temperature. The comparability of baseline core temperatures is shown Figure 1. Delauney 1991, De Witte 1998, Holdcroft 1978, Horn 1998, Mao 1998, Mathews 2002 and Matsukawa 2001 did not report baseline core temperatures in the groups before the intervention. Figure 1 suggests that in four studies, baseline temperatures were significantly different between groups (Cheong 1998; Hong 2005; Powell 2000; Röhm 2005). However, core temperatures were described as ‘similar between the groups’ in Hong (2005) and Röhm (2005). The sizes of the differences in temperatures were 0.4ºC in Cheong (1998) and Röhm (2005); 0.2ºC in Hong (2005) and Powell (2000b), and 0.1°C in Crozier (2004). These differences in baseline were compared with the effect size, and only outcomes in which the baseline difference was less than 20% of the effect size were included in the analysis.

Figure 1. Baseline core temperatures.

Figure 1

Baseline core temperatures.

The risk of bias was assessed for each included study. Cheong (1998), Crozier (2004), Powell (2000b) and Röhm (2005) were treated with caution because of significant differences in baseline.

RESULTS

A. Premedication

1. Alpha2-adrenergic antagonist versus placebo
1.1. Intervention given in the preoperative phase

Mao (1998) compared clonidine 150μg, given orally 90 minutes before induction of spinal anaesthesia, with placebo (two starch tablets) in 100 patients. The ambient temperature was 22 to 23ºC. Mizobe (2005) compared clonidine versus placebo, given orally 30 minutes before entering the operating room, in patients having combined general plus epidural anaesthesia. Eight patients received 150μg clonidine, eight received 300μg clonidine, and eight received placebo. The ambient temperature was 24ºC.

a. Core temperature intraoperatively

The Mao (1998) study in 100 patients showed no significant difference in core temperature at 30 minutes after spinal anaesthesia. The confidence interval is fairly wide.

Figure 2. Core temperature.

Figure 2

Core temperature.

At 180 minutes, meta-analysis of the two Mizobe (2005) comparisons in 24 patients showed a significantly higher mean core temperature for the placebo group, with a fairly wide confidence interval: WMD −0.73°C (95%CI −1.03, −0.44).

1.2. Intervention given in the intraoperative phase at induction

Two studies gave clonidine or placebo at induction of anaesthesia (Buggy abstract: 150μg IV; Piper 2002: 3μg/kg IV).

The Buggy (1997) study gave few details. In the Piper (2002) study, the mean duration of surgery was 74.1 (SD 42.3) minutes in the clonidine group and 74.3 (SD 34.4) for the placebo group.

a. Core temperatures intraoperatively

The Buggy (1997) study recorded the temperature at 60 minutes intraoperatively in 60 patients. There was no significant difference between interventions; the confidence interval is fairly wide (Figure 3).

Figure 3. Intraoperative clonidine measured intra and postoperatively.

Figure 3

Intraoperative clonidine measured intra and postoperatively.

b. Core temperatures postoperatively

The Piper (2002b) study in 60 patients recorded the temperature at 15 and 60 minutes after extubation (Figure 3). There was no significant difference between interventions (Figure 3).

1.3. Intervention given in the intraoperative phase at the end of surgery

Six studies evaluated the effect of clonidine given at the end of surgery (for the prevention of shivering). Delauney (1991) compared clonidine versus control (isotonic saline) given at the end of surgery as an infusion over 20 minutes, before transfer to the recovery room. Rectal temperatures were measured in the recovery room (at some point during the first hour in recovery; exact time of measurement not stated).

Figure 4. Clonidine given at the end of surgery; postoperative temperatures.

Figure 4

Clonidine given at the end of surgery; postoperative temperatures.

Horn (1998) compared clonidine versus control (saline), given at the end of surgery; patients were extubated 5 minutes later. Patients were covered with warmed sheets during anaesthesia; ambient temperature was 23ºC.

Piper (2000) compared clonidine 3μg/kg with placebo, given at the end of surgery. The mean duration of surgery was 93.1 (SD 48.2) minutes in the clonidine group and 86.6 (SD 26.9) in the placebo group.

Piper (2001) compared clonidine 3μg/kg with placebo, given at the end of surgery. The mean duration of surgery was 91.0 (SD 52.1) minutes in the clonidine group and 77.9 (SD 34.9) in the placebo group.

Piper (2004) compared clonidine 1.5 μg/kg IV with placebo, given at the end of surgery.

Horn (1997) compared clonidine 3μg/kg with saline placebo 5 minutes before extubation. In two groups, these interventions were in addition to isoflurane anaesthesia, and in two further groups, clonidine or saline were combined with propofol anaesthesia. The temperature was measured 20 minutes after extubation.

a. Core temperatures postoperatively

Meta-analysis across these studies was carried out for temperatures measured 20 and 60 minutes post-extubation, in 60 and 267 patients respectively. Other results are given for single studies. There was no significant difference in postoperative temperatures at any time, and there was no significant heterogeneity in the meta-analyses.

2. Benzodiazepines versus placebo/no intervention
2.1. Intervention given in the preoperative phase

Three studies compared midazolam and placebo or no premedication in the preoperative phase; two of these gave midazolam in addition to other interventions (Kimberger 2007; Matsukawa 2001).

Toyota (2004) compared intramuscular midazolam versus no premedication. Room temperature was 24 to 25ºC; patients were covered with a single surgical drape.

Kimberger (2007) compared intravenous midazolam versus usual care and midazolam plus active warming versus active warming alone, with an outcome of core temperature preoperatively. Ambient temperatures at the start and end were around 19ºC.

Matsukawa (2001) compared midazolam plus atropine versus atropine alone with an outcome of change in core temperature preoperatively. Patients were ‘minimally clothed’ and covered with single layer cotton blanket; ambient temperature was 23 to 24ºC.

a. Core temperature preoperatively

Kimberger (2007) compared (a) midazolam 30μg/kg plus usual care versus usual care alone. This study also compared (b) midazolam 30μg/kg plus forced-air warming device versus forced air warming alone. Meta-analysis of the two comparisons in 40 patients showed no significant differences between groups, but heterogeneity across comparisons (I2=70%, p=0.07).

Matsukawa (2001) compared (a) 50μg/kg midazolam versus saline placebo and (b) 50μg/kg midazolam plus 10μg/kg atropine versus atropine alone, recording an outcome of change in core temperature preoperatively. Meta-analysis of the two comparisons in 40 patients showed a significantly lower mean core temperature for the midazolam group; WMD −0.36 (95% CI −0.47, −0.25).

Figure 5. Midazolam in the preoperative phase.

Figure 5

Midazolam in the preoperative phase.

Meta-analysis across the two subgroups showed significant heterogeneity between Matsukawa (2001) and Kimberger (2007) (I2=87%, p<0.0001), which may be a dose effect. This conclusion is supported by another Matsukawa (2001) RCT in volunteers [Matsukawa 1997 BJA 78: 396–399], which showed a dose effect: there was no significant difference in core temperatures at 30 minutes for 25μg/kg IM compared with no midazolam, but a significant difference for 75μg/kg IM when compared with either the 25μg/kg dose or the control group.

b. Core temperature intraoperatively

Toyota (2004) compared midazolam, either (a) 40μg/kg or (b) 80μg/kg IM, as premedication 30 minutes before induction of anaesthesia versus no premedication. Meta-analysis of the two comparisons in 60 patients showed a significantly higher mean core temperature at all times for the midazolam group, from 15 minutes intraoperatively (Figure 6).

Figure 6. Midazolam in preoperative phase, temperatures recorded intraoperatively.

Figure 6

Midazolam in preoperative phase, temperatures recorded intraoperatively.

2.2. Intervention given in the intraoperative phase

Grover (2002) compared 30μg/kg IV midazolam versus placebo, given at the end of the procedure (one minute before switching off halothane) in 40 women having brachytherapy for cervical cancer. The outcomes studied were the core temperature intraoperatively (i.e. before the intervention) and postoperatively.

a. Core temperature intraoperatively

There was no significant difference in core temperature at 15 and 20 minutes intraoperatively, but at 30 minutes, there was a small, significant difference, with a higher mean core temperature in the midazolam group (Figure 7).

Figure 7. Midazolam in the intraoperative phase.

Figure 7

Midazolam in the intraoperative phase.

b. Core temperature postoperatively

At 60 minutes postoperatively, there were significantly lower temperatures for the midazolam group (Figure 7).

B. Reversal of benzodiazepines versus placebo

1. Benzodiazepine antagonists
1.1. Intervention given in postoperative phase

Weinbroum (2001) compared flumenazil versus placebo IV (in 10ml volume, at a rate of 2ml/10sec) when the patients began to awaken, in three comparisons: a) using halothane; b) using enflurane and c) using isoflurane as the anaesthesia drug.

a. Postoperative temperatures

Outcomes measured were temperatures at 20, 30, 40 and 60 minutes postoperatively. Meta-analysis of the three comparisons in 261 patients showed significantly higher mean temperatures for the flumenazil group at all times postoperatively. The WMD ranged from −0.20 (95%CI −0.31, −0.08) for a control group temperature of 36.4 to 36.5°C at 40 minutes, to −0.27 (95%C −0.40, −0.15) for a control group temperature of 36.4°C at 30 minutes (Figure 8). There was no heterogeneity for any of the meta-analyses.

Figure 8. Flumenazil.

Figure 8

Flumenazil.

C. Muscle relaxants

1. Anti-muscarinic drugs versus placebo
1.1. Intervention given in the preoperative phase

Matsukawa (2001) compared (a) IM atropine (0.01mg/kg) versus saline placebo and (b) atropine (0.01mg/kg) plus midazolam (0.05mg/kg) versus midazolam (0.05mg/kg) in 40 patients. The outcome was the change in core temperature, compared with baseline, 30 minutes later, just before induction of anaesthesia.

a. Change in core temperature preoperatively

Meta-analysis of the two comparisons gave a significantly higher mean temperature for the atropine group, 30 minutes after the intervention was given. There was no heterogeneity (I2=0%, p=0.38). The WMD was 0.26°C (95%CI 0.15, 0.37).

Figure 9. Atropine.

Figure 9

Atropine.

The lack of heterogeneity suggested it was valid to combine the two types of comparison. In the absence of midazolam, the core temperature of patients given both atropine and placebo increased, and it is assumed that atropine is actively increasing the temperature rather than just preventing cooling.

b. Intraoperative core temperature (before extubation)

De Witte (1995) compared glycopyrronium versus placebo in 22 patients, as premedication 60 minutes before induction of anaesthesia. There was no significant difference between interventions, although the confidence interval is fairly wide .

Figure 10. Glycopyrronium.

Figure 10

Glycopyrronium.

D. Reversal of muscle relaxants

1. Cholinesterase inhibitor versus placebo
1.1. Intervention given in the preoperative phase

Horn (1998) compared physostigmine versus placebo (saline), given at the end of anaesthesia; patients were extubated 5 minutes later and core temperature measured 15 minutes after that. The ambient temperature was 23ºC.

Figure 11. Physostigmine postoperatively.

Figure 11

Physostigmine postoperatively.

Röhm (2005) compared physostigmine versus placebo, given intravenously over 15 minutes at the start of skin closure. Patients were covered with sheets during anaesthesia. Outcomes were temperatures 15 and 60 minutes after arrival in PACU.

a. Core temperature postoperatively

The Röhm (2005) study had a large baseline difference (0.4°C), which was larger than the effect size, so this study was not included in the analysis. The remaining study (Horn 1998a) in 30 patients showed no significant difference between interventions, but the confidence interval was fairly wide.

E. Induction of anaesthesia

1. N-methyl-D-aspartate (NMDA) receptor antagonist versus placebo
1.1. Intervention given in the preoperative phase

Two studies compared the effects of ketamine and placebo:

Sagir (2007) compared 0.5mg ketamine versus saline placebo during regional anaesthesia, in 80 patients. The theatre temperature was 24ºC; irrigation and IV fluids were pre-heated to 37ºC; patients were covered with 1 layer of cotton blanket. The outcome was the final core temperature at 60 minutes.

Kinoshita (2004) compared ketamine and saline in 20 patients, at a rate of 0.3mg/kg/h, given at induction, together with propofol. The theatre temperature was 25°C and warmed IV fluids were also given.

Sagir (2007) also compared 0.25mg ketamine plus 1.5mg granisetron versus 3mg granisetron during regional anaesthesia. This comparison did not correspond to an investigation of the added effect of ketamine because the amounts of granisetron were not the same in the two groups. This comparison was therefore not included.

a. Core temperatures intraoperatively

Figure 12 shows the two studies at different intraoperative times. There was a significant difference in core temperature from about 30 minutes, with the placebo group being warmer. The confidence intervals were fairly wide, apart from at the final temperature in the Sagir (2007) study.

Figure 12. Ketamine.

Figure 12

Ketamine.

Figure 13. Ketamine in the intraoperative phase.

Figure 13

Ketamine in the intraoperative phase.

1.2. Intervention given in the intraoperative phase

Ikeda (2001) compared ketamine plus propofol versus propofol alone during general anaesthesia in 20 patients.

a. Core temperature intraoperatively

The core temperature decreased significantly less in the ketamine group (0.5ºC versus 0.9ºC) at 60 minutes after the start of the infusion.

F. General anaesthesia drugs

1. Anaesthesia drug 1 versus drug 2
1.1. Intervention given in the preoperative phase

Sahin (2002) compared isoflurane versus propofol in 20 patients. The confidence intervals were too wide at all time points to determine if there was a difference between isoflurane and propofol (Figure 14). All patients received dextrose-free crystalloids and colloids at room temperature; ambient temperature was 21ºC (SD 1).

Figure 14. General anaesthesia drugs given in the preoperative phase.

Figure 14

General anaesthesia drugs given in the preoperative phase. NB: Scale −4 to +4

1.2. Intervention given in the intraoperative phase

Goto (1999) compared (a) xenon 43% plus isoflurane 0.5% (n=13) versus isoflurane 1.2% (n=13), and (b) nitrous oxide 63% plus isoflurane 0.5% (n=12) versus the same control group of isoflurane 1.2%. The outcome was the lowest core temperature intraoperatively.

Figure 15. General anaesthesia drugs given in the intraoperative phase.

Figure 15

General anaesthesia drugs given in the intraoperative phase.

a. Lowest core temperature intraoperatively

There was no significant difference in the lowest core temperature between xenon plus isoflurane and isoflurane, but the lowest core temperature for nitrous oxide plus isoflurane was higher than for isoflurane alone.

2. Different doses of halothane
2.1. Intervention given in the preoperative phase

Holdcroft (1978) assessed halothane 0.5% versus halothane 1% in 15 patients, given preoperatively.

Figure 16. Doses of halothane.

Figure 16

Doses of halothane.

a. Core temperature intraoperatively

There was no significant difference in core temperature at 1, 2 or 3 hours, although the confidence interval was wide at one hour and fairly wide at two hours.

G. Analgesia

1. Opioid versus placebo

Two studies compared pethidine and placebo for patients given general anaesthesia (Horn 1998; Piper 2000), and two received regional anaesthesia (Kelsaka 2006; Hong 2005). These studies were combined where appropriate.

1.1. Interventions given in the preoperative phase

Kelsaka (2006) compared pethidine with saline placebo in 50 patients, given immediately before spinal anaesthesia for patients undergoing elective orthopaedic surgery. Lactated Ringer’s solution, warmed to 37ºC, was infused at 10ml/kg/hr for 30 min before surgery; ambient temperature was 21 to 22ºC; patients were covered with one layer of surgical drape intraoperatively and one cotton blanket post-operatively.

Figure 17. Pethidine preoperatively.

Figure 17

Pethidine preoperatively.

a. Lowest intraoperative temperature

The outcome was the change in rectal temperature (i.e. the lowest rectal temperature minus the preoperative rectal temperature). There was no significant difference between groups.

1.2. Interventions given in the intraoperative phase
1.2.1. Pethidine

Two studies compared pethidine versus control (saline), given at the end of surgery (Horn 1998; Piper 2000) in 90 patients. Patients were extubated and the core temperature measured 15 and 60 minutes after that. One additional study (Hong 2005) compared 10mg pethidine plus 0.5% bupivacaine versus bupivacaine alone for regional anaesthesia for elective Caesarean section. This indirect study was not considered further.

Figure 18. Pethidine – core temperatures postoperatively.

Figure 18

Pethidine – core temperatures postoperatively.

a. Postoperative core temperatures

Meta-analysis of Horn (1998) and Piper (2000) in 90 patients at 15 minutes, and results from the Piper (2000) study of 60 patients at 60 minutes post-extubation, showed no significant differences in core temperature, between pethidine and placebo. There was no heterogeneity.

1.2.2. Morphine

Hong (2005) compared three groups in women undergoing combined spinal-epidural anaesthesia for elective Caesarean sections: the women received 0.1mg morphine (Hong a), or 0.2mg morphine (Hong b), each in addition to 0.5% bupivacaine versus bupivacaine alone. This is an indirect population, and there was a baseline difference for each of these comparisons, which was not small compared with the effect size. Therefore the results are not reported.

2. Opioid dose 1 versus dose 2
2.1. Intervention given in the intraoperative phase

Hong (2005) compared 0.1mg morphine with 0.2mg morphine, each in addition to 0.5% bupivacaine, for combined spinal-epidural anaesthesia for elective Caesarean section (indirect population). This comparison had a large difference in baseline, so results were not considered further.

3. Opioid type 1 versus type 2
3.1. Intervention given in the intraoperative phase
3.1.1. Morphine versus pethidine

Hong (2005) compared 0.1mg morphine (Hong e), or 0.2mg morphine (Hong f), with 10mg pethidine, each in addition to 0.5% bupivacaine, for combined spinal-epidural anaesthesia for elective Caesarean section (indirect population), measured at 60 minutes. Meta-analysis of the two comparisons in 90 patients showed no significant differences in temperatures between the groups.

Figure 19. Morphine versus pethidine in indirect population.

Figure 19

Morphine versus pethidine in indirect population.

3.1.2. Remifentanil versus alfentanil

Crozier (2004) compared remifentanil versus alfentanil during elective ENT surgery in 98 patients. All patients were actively warmed during the procedure with forced air warming; the opioid infusion rate could be varied according to clinical need. The study had a baseline difference of −0.1°C and this was comparable with the difference in effect size, so conclusions were not drawn.

4. Other centrally-acting analgesics (for pain control) versus placebo / no intervention
4.1. Intervention given in the preoperative phase

De Witte (1995) compared tramadol plus glycopyrronium versus glycopyrronium alone in 21 patients, given as premedication 60 minutes before induction of anaesthesia. The outcome was the core temperature before extubation. There was no significant difference between interventions, although the confidence interval was fairly wide.

Figure 20. Tramadol given preoperatively.

Figure 20

Tramadol given preoperatively.

4.2. Intervention given at the start of the intraoperative phase
4.2.1. Nefopam

Bilotta (2002) compared nefopam with placebo in 60 patients, given immediately before epidural or subarachnoid anaesthesia. The theatre temperature was 22ºC (SD 1).

a. Core temperatures intraoperatively

The outcomes studied were the core temperature at 15, 30, 60 and 90 minutes intraoperatively. There was no significant difference between interventions until 90 minutes, after which time the placebo group was warmer by 0.30°C (MD −0.30°C (95%CI −0.57, −0.03); the confidence interval was fairly wide at 90 minutes. This is shown in Figure 21.

Figure 21. Nefopam and tramadol.

Figure 21

Nefopam and tramadol.

4.2.2. Tramadol

Bilotta (2002) compared tramadol with placebo in 60 patients, given immediately before epidural or subarachnoid anaesthesia. The theatre temperature was 22ºC (SD 1).

a. Core temperatures intraoperatively

The outcomes studied were the core temperature at 15, 30, 60 and 90 minutes intraoperatively. There was a significant difference from 30 minutes intraoperatively, with the placebo group being warmer by up to 0.50°C. The confidence intervals were fairly wide at 30 and 90 minutes (Figure 21).

4.3. Intervention given at the end of the intraoperative phase
4.3.1. Nefopam

Piper (2004) compared nefopam at doses of 0.2mg/kg, 0.1mg/kg, and 0.05mg/kg with placebo, given at the end of surgery. The outcomes studied were the core temperature at 15 and 60 minutes after extubation.

Röhm (2005) compared nefopam with placebo, given intravenously over 15 minutes at the start of skin closure. Outcomes were temperatures at 15 and 60 minutes after arrival in PACU.

a. Core temperatures postoperatively

Meta-analysis of the four comparisons in 356 patients showed a significantly higher mean core temperature for the placebo group at 60 minutes after arrival in PACU: WMD −0.21 (95%CI −0.33, −0.09), for a control group temperature range of 36.0 to 36.2°C. There was no heterogeneity.

Figure 22. Nefopam.

Figure 22

Nefopam.

4.3.2. Tramadol

Two studies compared the effects of tramadol and placebo, given at the beginning of wound closure (Matthews 2002; de Witte 1998).

Mathews (2002) compared tramadol at either 2mg/kg or 1mg/kg versus saline control in 100 patients. The ambient temperature was 21.2 to 24.9ºC.

De Witte (1998) compared tramadol and placebo. The mean duration of surgery was 56 (SD 16) minutes in the tramadol group and 61 (SD 16) minutes for placebo.

a. Incidence of IPH postoperatively

Meta-analysis of the two comparisons in the Mathews (2002) study, in 100 patients, showed no significant difference in the incidence of IPH (less than 36.0°C), but the confidence interval is fairly wide.

Figure 23. Tramadol – incidence of IPH.

Figure 23

Tramadol – incidence of IPH.

b. Core temperature at extubation

One study (de Witte 1998) recorded the core temperature at extubation in 40 patients. There was no significant difference between interventions, but the confidence interval is fairly wide.

Figure 24. Tramadol – core temperature.

Figure 24

Tramadol – core temperature.

5. Centrally acting analgesia dose 1 versus dose 2
5.1. Intervention given in the intraoperative phase
5.1.1. Nefopam

Mathews (2002) compared tramadol 2mg/kg with 1mg/kg, given at the beginning of wound closure, in 100 patients.

a. Incidence of hypothermia

The outcome recorded was the number of patients with a core temperature below 36ºC. There was no significant difference between doses.

Figure 25. Tramadol dose comparison.

Figure 25

Tramadol dose comparison.

5.1.2. Nefopam

Piper (2004) compared nefopam at doses of 0.2mg/kg, 0.1mg/kg, and 0.05mg/kg, given at the end of surgery, with about 75 patients in each arm.

a. Core temperatures postoperatively

Piper (2004) recorded the core temperatures at 15 and 60 minutes post extubation. There were no significant differences between doses (Figure 26).

Figure 26. Nefopam dose comparison.

Figure 26

Nefopam dose comparison.

6. Centrally acting analgesia type 1 versus type 2
6.1. Intervention given at the start of the intraoperative phase
6.1.1. Nefopam versus tramadol

Bilotta (2002) compared nefopam with tramadol, given immediately before epidural or subarachnoid anaesthesia in 60 patients.

a. Core temperatures intraoperatively

The outcomes studied were the core temperature at 15, 30, 60 and 90 minutes intraoperatively. Patients receiving nefopam were significantly warmer than those receiving tramadol after 60 and 90 minutes; mean difference at 60 minutes: 0.40°C (95%CI 0.22, 0.58), for a tramadol temperature of 35.6°C.

Figure 27. Nefopam versus tramadol.

Figure 27

Nefopam versus tramadol.

H. Control of nausea

1. Serotonin receptor antagonists versus placebo

Two studies examined these drugs during general anaesthesia (Powell 2000; Piper 2002) and two during regional anaesthesia (Sagir 2007; Kelsaka 2006). We combined the studies across types of anaesthesia.

Piper (2002) compared 12.5mg dolasetron versus placebo, given after induction of general anaesthesia, in 60 patients, and recorded the temperature at the end of surgery (mean duration of surgery 70.2 (SD 32.5) minutes for dolasetron group and 74.3 (SD 34.4) for controls) and 15 and 60 minutes after extubation.

Powell (2000) compared ondansetron 4mg or 8mg, given at induction, versus saline control, in 55 patients, and recorded the temperature at 30, 60 and 90 minutes after induction. The duration of anaesthesia administration was 38 minutes (SD 12 to 18).

Kelsaka (2006) compared 8mg IV ondansetron with saline placebo, given immediately before spinal anaesthesia in 50 patients undergoing elective orthopaedic surgery. The outcome was the change in rectal temperature (i.e. the lowest rectal temperature recorded during the operation minus the preoperative rectal temperature). Patients received warmed IV fluids.

Sagir (2007) compared (a) granisetron (3mg) versus placebo and (b) granisetron (1.5mg) plus ketamine (0.25mg) versus ketamine (0.5 mg) alone during regional anaesthesia, in 120 patients. The duration of anaesthesia/surgery was not stated. The comparison of the combination versus ketamine alone was excluded from the analysis because it did not have the same amount of ketamine in each arm.

a. Core temperature intraoperatively

Two studies (Powell 2000, in 82 patients; Kelsaka 2006, in 50 patients) recorded the core temperature intraoperatively, at 30 minutes and lowest intraoperative temperatures respectively. There was no significant difference at either time or dose, although the confidence intervals were fairly wide.

Figure 28. Serotonin receptor antagonists.

Figure 28

Serotonin receptor antagonists.

b. Core temperature at the end of surgery

Two studies recorded the core temperature at the end of surgery. There was no significant difference for the Piper (2002) study comparing dolasetron with placebo in general anaesthesia, but there was a large effect for granisetron versus placebo in regional anaesthesia, with granisetron treated patients being warmer: MD 0.60°C (95%CI 0.36, 0.84) in 60 patients.

c. Core temperature postoperatively

One study reported postoperative temperatures (Piper 2002) (Figure 29) in 60 patients. There was no significant difference between dolasetron and placebo.

Figure 29. Serotonin receptor antagonists (end of surgery).

Figure 29

Serotonin receptor antagonists (end of surgery).

2. Serotonin receptor antagonist dose 1 versus dose 2
2.1. Intervention given in the preoperative phase
2.1.1. Ondansetron dose comparison

Powell (2000) assessed ondansetron 4mg versus ondansetron 8mg in 54 patients.

a. Core temperatures intraoperatively

There was no significant difference between interventions at 30 minutes intraoperatively, but the confidence interval was fairly wide

Figure 30. Ondansetron dose comparison.

Figure 30

Ondansetron dose comparison.

7.2. Risk factors for IPH – non-pharmacological

SELECTION CRITERIA

Selection criteria were as outlined in the general methods section apart from the types of risk factor and outcomes described below.

Types of risk factor

Any property reported to be a risk factor for IPH was to be considered, including the following a-priori ones predicted by the GDG:

  • Age
  • Length of preoperative starvation
  • Temperature of patient at the beginning of the preoperative phase
  • Temperature of patient at first anaesthetic intervention
  • ASA grade
  • Pre-existing medical conditions (diabetes mellitus, thyroid disease, corticosteroid disease, cardiac disease)
  • Type of surgery: according to the grade defined in the NICE Preoperative Test guideline
  • Magnitude of surgery (major, intermediate, minor)
  • Site of surgery: open body cavity or other
  • Duration of anaesthesia
  • Duration of surgery
  • Urgency of operation: urgent, emergency, elective
  • Environmental factors: temperature, humidity (pre-, intra-, and post-operative)
  • Irrigation fluids: warmed/unwarmed
  • Infused fluids: warmed/unwarmed, by volume infused.

Type of outcome measure

As noted in the general methods section, ideally, the incidence of hypothermia should be determined for patients who were not warmed, but studies in which some or all of the patients were warmed could also be included. The GDG considered that risk factors may be different in warmed patients. Preferably patient warming would be included as a variable in multivariate analyses.

SEARCH STRATEGY

Searches were performed on the following core databases: MEDLINE, EMBASE, CINAHL and The Cochrane Library (1966 to current day with guidance from the GDG). Additional databases were not searched for this review. The search strategies are given in Appendix B. Twenty-five studies met the inclusion criteria for the review. The reference lists of the retrieved studies were inspected for further potential papers.

METHODOLOGICAL QUALITY OF STUDIES

The methodological quality of studies was assessed according to the type of study design. In evaluating the literature, RCTs and cohort studies were selected to be the best available evidence source for this review, and were quality assessed separately.

Both RCTs and cohort studies were assessed according to the criteria given in the general methods section

CHARACTERISTICS OF CLINICAL STUDIES INCLUDED IN THE REVIEW (APPENDIX C)

We included 25 studies, for which full data extraction was carried out. Although there were additional studies available, we did not believe their results would materially affect the review and therefore decided to truncate it at this point. In most of the remaining studies multivariate analyses had not been carried out or the study design was inferior.

The 25 included studies examined had different study designs:

One of the RCTs had an ANCOVA multivariate analysis that covered risk factors other than the randomised comparison (Frank 1992).

The study sizes ranged from 13 (Steinbrook 1997) to 101 for the RCTs, and 22 (Morris 1971) to 18,759 (Lau 2001) for the cohort studies. The case-control study included 400 patients, 200 cases of patients with core temperatures less than 35.0°C and 200 with temperatures over 36.0°C.

Two studies were carried out in the UK (Closs 1986; Hind 1994); one in each of Austria, Italy, Finland and Portugal; eight were in North America; one in Mexico; five in Japan; two in Thailand; one in China (Hong Kong); one in Egypt and one in Australia.

A range of procedures was undertaken.

Three studies stated they included patients receiving emergency surgery (Baker 1995; Lau 2001 (31% elective); Flores Maldonado 1997 (35%)). Two studies had patients with elective surgery only (Hind 1994; Kurz 1995). The rest did not state if the surgery was elective or emergency.

The studies covered a range of types of anaesthesia:

All studies but four (Baker 1995; Closs 1986; Kasai 2002, case control; Steinbrook 1997) reported the duration of surgery and/or anaesthesia. Full details are given in Table 1.

Table 1. Duration of surgery/anaesthesia.

Table 1

Duration of surgery/anaesthesia.

Three studies included some children: Flores Maldonado (1997) ranged from 5 to 90 years (mean 42); Lau (2001) had 13% of the patients under 15 years; and Kongsayreepong (2003) had a range of 15 to 93 years (children ≤ 14 years were excluded from the analysis for this study). The GDG was concerned that large numbers of children may have been included in the Flores Maldonado (1997) study.

All studies but five (Baker 1995; Closs 1986; Lau 2001; Roberts 1994; Vorrakitpokatorn 2006) reported the theatre temperature.

Eleven of the studies recorded the core temperature using a tympanic membrane thermometer (Abelha 2005; El Gamal 2000; Flores Maldonado 2007; Frank 1994; Frank 2000; Kasai 2002; Kitamura 2000; Kongsayreepong 2003; Nakajima 2002; Nguyen 2000; Vorrakitpokatorn 2006); one recorded temperature in the pulmonary artery (Baker 1995); two in the bladder (Danelli 2002; Stewart 1998); six in the oesophagus (Hind 1994; Kurz 1995; Mizobe 2005; Morris 1971; Roberts 1994; Steinbrook 1997); one in the rectum (Yamakage 2000); one sublingually using a reliable method (Frank 1992); one recorded aural or nasopharyngeal temperatures (Hendolin 1982) and one recorded aural temperatures, but not in the intra and immediate postoperative phases (Closs 1986). One study (Lau 2001) did not state the measurement site.

The studies varied in their use of warming mechanisms:

Risk factors investigated by the cohort studies (multivariate analyses) or RCTs

The following risk factors have been investigated in the included studies:

Patient characteristics

Anaesthesia factors

  • Duration of anaesthesia
  • Type of anaesthesia
  • Anaesthesia: end expiratory pressure
  • Height of spinal block

Surgery factors

  • Urgency of operation: urgent, emergency, elective
  • Type of surgery: according to NICE preoperative test guideline grade (none classified in this way)
  • Magnitude of surgery (major, intermediate, minor)
  • Laparoscopic/open surgery
  • Duration of surgery
  • Patient position intraoperatively

Other risk factors

Environmental factors

  • Theatre temperature.

Outcomes

The studies measured the following outcomes:

Seven studies measured the incidence of IPH. The studies differed in their definitions of hypothermia:

Kongsayreepong (2003) also recorded the incidence of core temperatures less than 35.5°C and less than 35.0°C, and noted that multivariate analyses using these alternative definitions gave results consistent with those for a definition of less than 36.0°C.

The studies also differed in the phase of measurement: all but two (Flores Maldonado 1997; Kasai 2002, case control) measured the incidence in PACU or ICU; these exceptions measured the occurrence intraoperatively.

Three studies (El-Gamal 2000; Frank 2000; Morris 1971) carried out multivariate analyses for the core temperature and five RCTs (Frank 1994; Mizobe 2005; Nakajima 2002; Nguyen 2000; Steinbrook 1997) recorded the mean difference between interventions, in core temperature at different times.

METHODOLOGICAL QUALITY OF INCLUDED STUDIES

The methodological quality was assessed separately for the cohort studies and RCTs and details for each study design are given in Appendices C and D. Further details of the criteria are given in the general section.

RCTs

Three studies reported the method of randomisation and this was adequate in each case (Danelli 2002, random number sequence; Mizobe 2005, computer generated; Steinbrook 1997, coin toss). The other studies did not state the method.

Two studies reported a method of allocation concealment, in each case the method was partially adequate (Mizobe 2005, sequentially numbered envelopes; Nguyen 2000, sealed envelopes). The other studies did not state the method.

Blinding of the outcome assessors was carried out in two studies (Kinoshita 2004; Mizobe 2005), possibly carried out in two studies (Danelli 2002; Frank 1994) and definitely not carried out, or highly unlikely, in one study (Nguyen 2000). The other studies did not state the blinding.

All studies but one (Mizobe 2005) reported that all the patients were analysed. For these other studies there was less than 20% missing data. There was no difference in the extent of missing data between groups (where reported). Two studies carried out a power calculation (Danelli 2002; Nguyen 2000).

Baseline comparability was demonstrated in most of the studies. Two studies (Frank 1992; Frank 1994) were not comparable for the volume of crystalloid used (greater for general anaesthesia). However, this factor was taken into account in the analysis in the former. One other study (Steinbrook 1997) was not comparable at baseline for age, weight, intraoperative fluids (may not be significant difference). One study (Danelli 2002) had a significantly longer duration of surgery in the laparoscopic group (mean difference 1.1h). The GDG regarded the Steinbrook (1997) study to have potential for bias, but the other studies were considered acceptable.

Overall, only one study (Steinbrook 1997) was considered to have potential for bias on the basis of conventional quality assessment.

However, in terms of possible confounders, there are some features that may influence the results of the risk factors analyses even though these features were held constant or were likely to be distributed equally across groups:

  • In one study all patients had forced air warming (Nguyen 2000). The GDG considered that other risk factors may depend on whether the patient is warmed. In another study (Steinbrook 1997) patients were selectively warmed if their temperatures fell below 35.0°C, which may have confounded the study
  • One RCT had a high theatre temperature, 24 to 26°C (Nakajima 2002).

The Frank (1994) study, which randomised patients to general and epidural anaesthesia, reported non-randomised within-trial subgroups of older and younger patients (cut at 62 years, the median). We decided not to consider the subgroup comparison of older and younger patients, but the post-hoc subgroup analysis of general versus epidural for each of the age groups was considered acceptable. This is not ideal, because we are unclear about the distribution of baseline characteristics across the general and epidural groups within the two age subgroups, but the randomisation was at least partly retained.

Cohort studies

No study was considered to be truly representative of the population (i.e. all procedures under general or regional anaesthesia in adults).

Fifteen studies were considered to be somewhat representative of the community:

Two studies were considered to be a selected group:

In all studies, the non exposed cohort was drawn from the same community as the exposed cohort. All studies but two recorded the temperature at an adequate site. Closs (1986) recorded aural temperatures on the ward and Frank (1992) used a sublingual recording, but the method was detailed. All studies were prospective apart from Roberts (1994) and the case-control study.

All studies but three reported that all the patients were followed up. One (Closs 1986) did not say; Lau (2001) reported that 2159/20918 (10%) of patients had missing data; Kongsayreepong (2003) reported that 10/194 (5%) patients were deliberately excluded from analysis because they were children under 14 years or they were hyperthermic.

Five studies stated that the patients were not hypothermic preoperatively (El-Gamal 2000; Frank 1992; Morris 1971; Roberts 1994; Yamakage 2000); in two studies (Abelha 2005; Kongsayreepong 2003) some of the patients were hypothermic (<36.0°C by GDG definition) at the start of surgery: the patients in Abelha (2005) had a range of 35.0 to 38.6°C and mean 36.37°C; however these patients were not hypothermic according to the authors’ definition (less than 35.0°C). The patients in Kongsayreepong (2003) had a range of 34.5 to 39.3°C (although hyperthermic patients were excluded from the analysis) and mean 37.0°C (authors’ definition less than 36.0°C); 49% patients were warmed intraoperatively however. The rest of the studies did not say if the patients were hypothermic at the start of the intraoperative phase.

Confounders taken into account

We considered whether the studies took account of particular confounders, either in the study design or the multivariate analysis. The GDG had identified, by consensus, four risk factors to be important: age, ASA grade, type of anaesthesia, and duration of anaesthesia/surgery or magnitude of surgery.

Three studies were comparable at baseline apart from the study risk factor (El-Gamal 2000; Kitamura 2000; Morris 1971).

  • El-Gamal (2000) (n=40) selected two cohorts of different ages, and held constant the ASA grade (I–II), the type of surgery (lower extremity orthopaedic) and the type of anaesthesia (general). The groups were also comparable at baseline for BMI, duration of surgery, IV fluid volume and preoperative core temperature. Overall 4/4 important confounders were taken into account. It is noted that the ratio of events:covariates is too small (4) for the dichotomous outcome.
  • Kitamura (2000) (n=27) investigated the effect of diabetes, in older and younger age groups. The four groups were comparable for BMI, IV fluid rate, duration of surgery, ambient temperature. The type of anaesthesia was constant (general). However, the diastolic arterial blood pressure was significantly different for diabetes with and without neuropathy. The GDG did not consider this to be an important difference. Overall 3/4 important confounders were taken into account.
  • Morris (1971) (n=22) investigated the effect of theatre temperature in subgroup analyses. There was no significant difference in age or site of operation between lower and higher temperature theatres. Duration of surgery was constant (all over 2 hours) as was the type of anaesthesia (general). Overall 2 to 3 of 4 important confounders were taken into account.

Four studies had all or most of the important confounders taken into account in the multivariate analysis (Abelha 2005; Frank 1992; Lau 2001; Vorrakitpokatorn 2006).

  • In Lau (2001) (n=18,759), the multivariate analysis included age, ASA grade and type of anaesthesia. The duration of surgery was held partially constant – operations were selected if they lasted longer than 2 hours. Overall 3 or 4/4 important confounders were taken into account. There were 111 events for 4 covariates, i.e. ratio of 28, which is acceptable.
  • In Vorrakitpokatorn (2006) (n=128), the multivariate analysis included age and duration of surgery. The type of anaesthesia was held constant (general). Overall 3/4 important confounders were taken into account (ASA grade was missing). There were 72 events for 4 covariates, i.e. a ratio of 18, which is acceptable.
  • In Abelha (2005) (n=185), the multivariate analysis reported results for magnitude of surgery and SAPS II. It was also adjusted for anaesthesia type and anaesthesia duration. The SAPS II score (Simplified Acute Physiology Score) is used to predict death and is assigned after 24 hours of ICU admission. The score is derived from 12 physiologic variables, age and underlying disease variables (AIDS, metastatic cancer and haematologic malignancy). Thus, at least indirectly, this study does include all 4 important variables.
  • Frank (1992) (n=97) was an RCT that also had multivariate analysis. This study had different types of analgesia for the two types of anaesthesia: the general anaesthesia group had morphine PCA and the epidural group had fentanyl. The GDG considered this difference to be acceptable. The study had 3/4 important risk factors.

Two studies were considered to be fairly acceptable - the multivariate analysis only had between 8 and 10 events per covariate (Kongsayreepong 2003; Flores Maldonado 1997).

  • Kongsayreepong (2003) (n=184) included in the multivariate analysis: age, ASA grade, magnitude of surgery, type of anaesthesia and duration of surgery, i.e. 4/4 important confounders taken into account, but the ratio of events to covariates was 105/12 = 9
  • Flores Maldonado (1997) (n=130) included in their multivariate analysis age, duration of surgery, magnitude of surgery, and type of anaesthesia, i.e. 3/4 important confounders taken into account, but the ratio of events to covariates was 53/7 = 8.

Five studies were considered to be possibly confounded because not enough of the important factors were included in the analysis (Baker 1995; Hind 1994; Kurz 1995; Closs 1986; Yamakage 2000).

  • Hind (1994) (n=30) carried out two multivariate analyses on the same data.
  • The first of these analyses (Hind 1994a) included age and kept constant the type of anaesthesia (general). Surgery duration was excluded from the analysis on the basis of univariate analysis. This meant that only 2/4 important confounders were taken into account. This study also had too many variables in total for the number of patients (30/6 = 5).
  • The second analysis (Hind 1994b) included none of the important factors, but kept constant the type of anaesthesia (general). Surgery duration was excluded from the analysis on the basis of univariate analysis. This meant that only 1/4 important confounders were taken into account.
  • In addition, the Hind (1994) study reported many correlations between 'independent' variables, i.e. confounding. For example, between age and theatre temperature or body fat or IV fluids or blood loss. Body fat also correlated with theatre temperature. The authors commented that the age-theatre temperature correlation was possibly due to the fact that older patients were put first on the operating list, which was when the theatre was colder.
  • Baker (1995) (n=56) included age and type of surgery of the important factors (i.e. 2/4 confounders taken into account). This study also had a large number of other variables in the multivariate analysis, so that the number of patients per covariate was 56/13 = 4.
  • Closs (1986) (n=31) was only adjusted for age in the analysis, i.e. 1/4 important risk factors. In addition, no data were recorded during the intraoperative and immediate postoperative periods.
  • In Kurz (1995) (n=40), the multivariate analysis included none of the important variables. The type of anaesthesia was constant (general); the patients had colon surgery and the mean duration was 3.8 hours (SD 1.3). The type of surgery was reported to be comparable for different size patients. Thus, account was taken of 2 of 4 important factors.
  • In Yamakage (2000) (n=60), the type of anaesthesia was held constant (general) and the surgery type was fairly specific (on lumbar vertebrae) and had a duration of approximately 3 hours. Age was partly adjusted in the body fat calculator. Thus account was taken of 2 to 3 of 4 important factors.

Three studies did not have enough events or patients for the number of variables included in the multivariate analysis (Hind 1994a, see above; Baker 1995, see above; Frank 2000). The Frank (2000) study had 44 patients for 6 covariates, i.e. 7 patients per covariate, which is slightly low.

The remaining two cohort studies were considered to be confounded: Roberts (1994) used a subgroup analysis, but confounders were not allowed for and were not comparable at baseline for duration of surgery. In Stewart (1998), all patients having open surgery had combined general/epidural anaesthesia, but all receiving laparoscopic surgery had general anaesthesia, leading to confounding.

Other factors:

  • The Stewart (1998) study reported that all the patients were given forced air warming; Abelha (2005) reported that 44% of patients were given forced air warming devices, but this was taken into account in the multivariate analysis; Kongsayreepong (2003) reported that 49% of patients were given forced air warming devices and this was assessed by univariate analysis and then not included in the multivariate analysis; the case control study gave the patients a circulating water mattress and warmed fluids (Kasai 2002).
  • As mentioned earlier, the GDG was concerned that large numbers of children may have been included in the Flores Maldonado (1997) study.
  • One study (El-Gamal 2000) had high theatre temperatures (24 to 26°C).

Overall the GDG decided that five studies were confounded: Roberts (1994), Stewart (1998), as above; Baker (1995), because of the type of surgery and low ratio of events to covariates; and the Closs (1986) and Hind (1994b) studies, each of which had only one of the four important factors. These studies were not considered in the analyses. The case control study was also considered to have greater potential for bias, and was not included further.

Four studies were treated cautiously, three because there were only 2/4 important factors included (Hind 1994a; Kurz 1995; Yamakage 2000). The Hind (1994a) study also had too many variables in total for the number of patients (30/6 = 5) and the Frank (2000) study had a ratio of 44/6 (=7). The presence of warming devices in about half of the patients in Kongsayreepong (2003) study without adjustment in the multivariate analysis was also taken into consideration, as was the Flores Maldonado (1997) study because it included children. All these studies at higher risk of bias were considered in sensitivity analyses.

RESULTS (see Appendix F for more details)

The results for the different risk factors are given in Appendix F. We consider below the effects of different risk factors on the incidence of hypothermia or the core temperature.

A. PATIENT RELATED RISK FACTORS

1. Age

Meta-analysis was not possible in many instances because the risk factor comparators were different (Figure 1). However, it was possible to combine two studies that had less than 40 years as a comparator (Kongsayreepong 2003; El-Gamal 2000) (Figure 2).

Figure 1. Age – incidence of IPH in ICU/PACU.

Figure 1

Age – incidence of IPH in ICU/PACU.

Figure 2. Age older cohort versus younger (under 40 years) cohort (not overlapping) – incidence of IPH in ICU/PACU.

Figure 2

Age older cohort versus younger (under 40 years) cohort (not overlapping) – incidence of IPH in ICU/PACU.

a. Incidence of IPH intraoperatively

One study (Flores Maldonado 1997) reported the effect of age on the incidence of IPH (temperature less than 36.0°C) intraoperatively. The multivariate analysis in 130 patients gave no numerical data for this risk factor, simply reporting that the effect was non significant for age as a continuous variable (mean 42 years, SD 20 years). Anaesthesia was general or regional and the theatre temperature was 22.9°C.

b. Incidence of IPH in PACU or ICU

Four cohort studies (Kongsayreepong 2003 [n=184; temperature less than 36.0°C]; El-Gamal 2000 [n=40; temperature less than 36.0°C]; Lau 2001 [n=18,759; temperature less than 35.0°C]; Vorrakitpokatorn 2006 [n=128; temperature less than 35.0°C]) investigated the effect of age on the incidence of IPH postoperatively. Each study considered age as a categorical variable. The incidence of IPH did not appear to be affected by adult age, but, in the large Lau 2001 study (18,758 patients), older adults (over 65 years), in comparison with children under 15, had significantly more patients with a core temperature below 35°C. The meta-analysis (Figure 2) of two studies in 224 patients compared older cohorts (over 65 or over 70 years) with a younger cohort (under 40). There was no statistically significant difference between cohorts in the number of patients with temperatures below 36.0°C, but the younger group was favoured. There was no heterogeneity (I2 = 0%).

El-Gamal (2000) had a theatre temperature greater than 24°C; Kongsayreepong (2003) had a temperature of 20 to 21°C and the others did not say. The confidence intervals were generally wide, which gives uncertainty to the conclusions.

c. Core temperature

Two cohort studies, El-Gamal (2000) in 40 patients and Kitamura (2000) in 36 patients reported the effect of age on core temperature at various times after the start of general anaesthesia (the mean duration of surgery was 1.7 to 1.8 h and 3.1 to 3.3 h respectively). The El-Gamal (2000) study included two cohorts of patients aged 60 to 75 years and 20 to 40 years, and the Kitamura (2000) study divided the cohort into older (60 years and older) and younger (less than 60 years) patients. The results are shown in Figure 3. There is no significant difference between age groups, until 3 hours after the start of surgery and on arrival in PACU, where the younger group had significantly higher temperatures (WMD: 3 hours: −0.30°C (95%CI −0.54, −0.06); PACU: −0.30°C (95%CI −0.58, −0.02)), however, the confidence intervals are fairly wide or wide. At shorter durations, the younger cohort is favoured.

Figure 3. Age - older cohort versus younger (under 40 years) cohort (not overlapping) – core temperatures.

Figure 3

Age - older cohort versus younger (under 40 years) cohort (not overlapping) – core temperatures.

A third study (RCT with multivariate analysis), Frank (2000), reported that, for patients aged 47 to 67 years, age had a statistically significant effect on core temperature in PACU. Treating age as a continuous variable, gave a ‘b’ coefficient of 0.03°C/year (p=0.01). The mean duration of surgery was 1.5 hours.

d. Change in core temperature

Two cohort studies (Frank 1992; Hind 1994a), in 97 and 30 patients respectively, carried out multivariate analyses for the change in core temperature intraoperatively. For Frank (1992), this was the difference between the ‘first postoperative temperature’ and the preoperative temperature. For Hind (1994), the change in oesophageal temperature was reported but it was not stated when this was measured. The durations of surgery were over 4 hours for Frank (1992) and 1 to 2 hours for Hind (1994). Both studies reported the unstandardised ‘b’ coefficients and Hind (1994) also reported the standardised ß coefficient. Meta-analysis showed a statistically significantly larger decrease in temperature for older patients, with no heterogeneity (I2=0%); mean −0.07°C/year (95%CI −0.11, −0.03) (Figure 4). We note, however, that the Hind (1994) study had methodological imperfections.

Figure 4. Age – continuous variable – change in core temperature.

Figure 4

Age – continuous variable – change in core temperature.

e. Rate of change of temperature in intraoperative phase

One cohort study (Kitamura 2000), in 36 patients, recorded the rate of change of core temperature before and after vasoconstriction and found no significant difference between older (≥60 years) and younger (less than 60) patients at either time.

Figure 5. Age over or equal to 60 years versus under 60 years – rate of change of temperature.

Figure 5

Age over or equal to 60 years versus under 60 years – rate of change of temperature.

f. Time for rewarming to 36.0°C

One study (Frank 1992) reported a borderline significant decrease in the time for rewarming to 36°C for younger patients. The standardised ß coefficient was 0.111 hours per year (p ≤ 0.05).

Conclusions for age as a risk factor

The evidence suggests that age is not an important risk factor for the incidence of hypothermia either intraoperatively or postoperatively, although the data on core temperature suggests that older people (over 60 years) have lower temperatures after 3 hours of surgery and in PACU. There does not appear to be a sensible cut-off point above which adult patients are at higher risk of perioperative hypothermia, although 60 years is a possibility.

There is some evidence that older patients take longer to rewarm to 36°C postoperatively. The GDG noted that some consequences of hypothermia are more severe for older people, especially morbid cardiac events.

2. Gender
a. Incidence of IPH intraoperatively

One cohort study (Flores Maldonado 1997) in 130 patients showed no significant effect of gender on the incidence of IPH (temperature less than 36.0°C) using multivariate analysis, but no numerical data were given (see Appendix F). This study may have had a less representative population (some children included).

3. ASA grade
a. Incidence of IPH in PACU or ICU

Two cohort studies (Kongsayreepong 2003; Lau 2001), in 184 and 18,759 patients respectively, investigated the effect of ASA grade on the incidence of IPH in PACU or ICU, using multivariate analysis. Lau (2001) subdivided the patients into categories I, II, III, IV, V and Kongsayreepong (2003) into I, II and higher than II. We carried out meta-analyses using either ASA III versus ASA I, ASA IV versus ASA I, or ASA V versus ASA I for the Lau (2001) study in combination with the Kongsayreepong (2003) comparison, ASA above II versus ASA I (Figure 6). The proportion of patients in the Kongsayreepong (2003) study in the higher ASA groups was not given. We note that the Kongsayreepong (2003) study defined IPH as temperature below 36.0°C, whereas the Lau (2001) study used below 35.0°C. The former also reported that 49% of the patients had forced air warming, which was not taken into account in the multivariate analysis.

Figure 6. Effect of ASA grade – incidence of IPH.

Figure 6

Effect of ASA grade – incidence of IPH.

Meta-analysis of ASA II versus ASA I showed a statistically significant difference favouring ASA I (OR 1.97 (95%CI 1.19, 3.24) with no heterogeneity (I2=0%), which suggests the difference in the definition of hypothermia may not be important (and the Kongsayreepong (2003) study suggested that the results in their study were consistent regardless of the definition). There are also statistically significant differences at higher ASA grades compared with ASA I, increasing, in the Lau (2001) study, with ASA grade. There is, however, some heterogeneity for the combination of ASA III versus ASA I with ASA II+ versus ASA I. This could be because the ASA II+ in Kongsayreepong (2003) was closer to ASA IV and V (although patients with these grades are rarer); it could possibly be related to the definition of hypothermia, or some other factor. It is notable that Lau (2001) shows a similar odds ratio for both ASA II and ASA III in comparison with ASA I.

To obtain an indication of the effect of any ASA grade above II for the Lau (2001) study, we calculated a weighted odds ratio (using log odds) and a weighted standard error, and combined these statistics in a meta-analysis with the Kongsayreepong (2003) study. This gave an odds ratio of 2.68 (95%CI 1.40, 5.12), with some heterogeneity (I2=56%, p=0.13).

Conclusion for ASA as a risk factor

ASA grade greater than ASA I is a risk factor for perioperative hypothermia, and the risk increases with ASA grade.

4. Body fat/body weight/height

Five cohort studies (Frank 2000; Hind 1994a; Kongsayreepong 2003; Kurz 1995; Yamakage 2000) investigated the effect of body fat or body weight, either on the incidence of IPH or on core temperatures. Both body fat and body weight were treated as continuous variables. One study investigated the effect of height (Kurz 1995). No studies investigated body mass index (BMI).

Meta-analysis was not carried out, either because of a lack of data – some studies reported only whether or not the factor was significant (Kurz 1995; Frank 2000; Hind 1994a; Yamakage 2000 for some outcomes) – or because of different outcome measures. We note that the Kurz (1995), Hind (1994) and Yamakage (2000) studies are possibly confounded because they used only 2 out of the 4 important risk factors in the multivariate analyses, and the Hind (1994) study also reported correlations between body fat and age (with an unexpected negative correlation), and body fat and theatre temperature.

Appendix F summarises all the results.

Kongsayreepong (2003) reported a mean weight of 57.2kg (SD 12) and a range of 30 to 91kg, which suggests children were included.

Kurz (1995) reported a mean height of 169 cm (SD 7), range 152 to 180 cm; and a mean weight of 73 kg (SD 20), range of 40 to 110 kg; the body fat ranged from 15 to 49%.

Frank (2000) reported a mean weight of 88kg (SD 20) and range 70 to 120 kg; the body fat mean was 27% (SD 7), with a range of 13 to 39%.

Hind (1994) reported a mean body fat content of 23.7% (SD 5.6); range 15 to 39.4%.

Yamakage (2000) reported a mean height of 159 cm (SD 7); and weight 63 kg (SD 8).

a. Incidence of IPH in ICU

One cohort study (Kongsayreepong) in 184 patients showed a small statistically significant effect of body weight on the incidence of IPH (temperature less than 36.0°C) in ICU, using multivariate analysis; OR 0.94 (95%CI 0.89, 0.98), with less hypothermia for a higher body weight.

b. Core temperature

The Kurz (1995) study in 40 patients reported no significant effect of body weight on change in core temperature over the first hour of surgery (no numerical data given), but there was a statistically significant effect identified with body fat (0.016°C/%, p<0.01) and with body weight divided by surface area (0.033°C.m2/kg). Yamakage reported that there was no statistically significant effect of body fat on the change in core temperature at 1 hour (p=0.054), however no numerical data were given.

At 2 hours, the Yamakage (2000) study in 60 patients reported a statistically significant effect of body fat on change in core temperature (0.03°C/%; p<0.0001) but Hind (1994) (n=30) found no significant effect of body fat on the change in core temperature intraoperatively (time not stated or data given). The latter study also reported correlations between body fat and age, and body fat and theatre temperature, and had more than one methodological limitation.

Frank (2000) (n=44) reported no significant effect of body fat or body weight on the core temperature in PACU (p=0.14).

Kurz (1995) (n=40) reported no significant effect of height on change in core temperature over the first hour of surgery (data not given). This study was possibly confounded because the authors used only 2 out of the 4 important risk factors in the multivariate analysis.

Conclusions for body fat/weight and height as a risk factor

Increased body weight may have a small protective effect on the incidence of perioperative hypothermia in ICU. The evidence for body weight and body fat intraoperatively is inconsistent. There is no significant effect of height on IPH in a poorer quality study.

5. Comorbidities – diabetes

Two cohort studies investigated diabetes as a risk factor for IPH (Kongsayreepong 2003 (n=184); Kitamura 2000 (n=27)). The Kitamura (2000) study divided the cohort into diabetics (with and without neuropathy) and controls; the groups in the comparisons considered below were comparable at baseline for characteristics other than those under study. Kongsayreepong (2003) carried out a multivariate analysis which included the risk factor, history of diabetic neuropathy.

a. Incidence of IPH in ICU

Kongsayreepong (2003) investigated the effect of a history of diabetic neuropathy compared with no history on the incidence of IPH in ICU (temperature less than 36.0°C) and found no significant difference; OR 0.86 (95%CI 0.24, 3.14); 14% of patients were reported to have diabetic neuropathy.

b. Core temperature

Kitamura (2000) reported the core temperature intraoperatively, for groups of patients with diabetes and no neuropathy versus those without diabetes. There were no significant differences between groups at any time, although the confidence intervals are fairly wide.

Figure 7. Effect of diabetes – core temperature.

Figure 7

Effect of diabetes – core temperature.

Kitamura (2000) reported the core temperature intraoperatively, for groups of patients with diabetes, with and without neuropathy. There were no significant differences between groups until three hours, at which time the neuropathy group had significantly lower core temperatures; mean difference: −0.49°C (95%CI −0.76, −0.22). The confidence intervals are fairly wide.

Figure 8. Effect of diabetic neuropathy – core temperature.

Figure 8

Effect of diabetic neuropathy – core temperature.

Conclusion for diabetes

Diabetes without neuropathy is not a risk factor for IPH, but patients with diabetic neuropathy have significantly lower core temperatures than diabetic patients without neuropathy after three hours of surgery.

6. Patient temperature preoperatively
a. Incidence of IPH in ICU

Two cohort studies (Kongsayreepong 2003; Abelha 2005) included patient preoperative temperature in the multivariate analyses of incidence of IPH in ICU (Abelha (2005) defined hypothermia as temperatures below 35.0°C; Kongsayreepong (2003) used below 36.0°C). The mean core temperature initially in Abelha (2005) was 36.37°C (SD 0.49), range 35.00 to 38.60; in Kongsayreepong (2003) it was 37.0°C (SD 0.7) range 34.5 to 39.3 (although hyperthermic patients were excluded from the analysis). We note also that warming devices were used in both studies, but only Abelha (2005) took this into account in the multivariate regression analysis. The studies did not report the perioperative stage in which warming devices were used.

Meta-analysis of 369 patients found a statistically significant effect of preoperative temperature (Figure 9); OR 0.31 (95%CI 0.17, 0.55), with a remarkably high homogeneity (I2=0%, p=0.96), despite differences in the definition of IPH.

Figure 9. Effect of patient preoperative temperature – incidence of IPH in ICU.

Figure 9

Effect of patient preoperative temperature – incidence of IPH in ICU.

GDG consensus was that patients arriving in the holding area with temperatures below 36.0°C should not undergo surgery until their temperature has been raised, except in an emergency.

Conclusion

A low preoperative temperature is a significant risk factor for IPH.

B. ANAESTHESIA RISK FACTORS

1. Type of anaesthesia

Eight studies investigated the effect of type of anaesthesia (Abelha 2005; Flores Maldonado 1997; Frank 1992; Frank 1994; Hendolin 1982; Kongsayreepong 2003; Lau 2001; Steinbrook 1997). Four of these were RCTs (Frank 1992; Frank 1994; Hendolin 1982; Steinbrook 1997) and the others were cohort studies. In the latter, different approaches were taken to the analysis: Lau (2001) compared, separately, regional anaesthesia or combined anaesthesia versus general anaesthesia (reference); Abelha (2005) compared, separately, general anaesthesia or combined anaesthesia versus regional anaesthesia (reference). Flores Maldonado (1997) considered spinal, epidural and general anaesthesia as separate categories and Kongsayreepong (2003) included categories of general, regional and combined anaesthesia. In the latter two studies, this meant that, for example, spinal was compared with the remaining categories (general and epidural).

1.1. Regional versus general anaesthesia
a. Incidence of IPH intraoperatively

Two studies compared regional and general anaesthesia as risk factors for the incidence of IPH intraoperatively (Flores Maldonado 1997 (n=130); Hendolin 1982 (n=38)). Flores Maldonado (1997) reported that there was no significant difference in the incidence of IPH (temperature below 36.0°C) between general anaesthesia and spinal or epidural anaesthesia, but no numerical data were given. Hendolin (1982) was a small RCT that compared general versus epidural anaesthesia in 38 patients, and recorded the incidence of hypothermia according to two definitions, less than 36.0°C and less than 35.0°C (figure 10). There was no significant difference when the definition less than 36.0°C was applied, but for a temperature below 35.0°C, there was a statistically significant difference favouring epidural anaesthesia, although the confidence interval is very wide.

Figure 10. Regional versus general anaesthesia.

Figure 10

Regional versus general anaesthesia.

b. Incidence of IPH in PACU or ICU

Two studies compared regional versus general anaesthesia as risk factors for the incidence of IPH in PACU or ICU (Abelha 2005; Lau 2001). Both studies defined IPH as less than 35.0°C. Abelha (2005) reported that the type of anaesthesia was adjusted for in the multivariate analysis, but no results were given. It is assumed this was not significant.

The Lau (2001) study in 18,759 patients, however, found a statistically significant odds ratio for the incidence of IPH below 35.0°C, favouring regional anaesthesia; OR 0.22 (95%CI 0.07, 0.70), although the confidence interval is wide.

Figure 11. Regional versus general anaesthesia.

Figure 11

Regional versus general anaesthesia.

c. Core temperature intraoperatively (Figure 14)

One RCT in 30 patients compared general with epidural anaesthesia and recorded core temperatures at various times intraoperatively (Frank 1994). Fluids were warmed for both groups. The study showed a statistically significant difference 30 minutes after induction of anaesthesia, with the epidural groups being warmer, but thereafter there was no significant difference between groups. The confidence intervals were fairly wide or wide. At 30 minutes the mean difference was 0.37°C (95%CI 0.09, 0.65), for a general anaesthesia group temperature of 35.8°C.

Figure 12. Regional versus general – core temperature intra- and postoperatively.

Figure 12

Regional versus general – core temperature intra- and postoperatively.

Figure 13. Combined versus not combined anaesthesia – core temperature intra- and postoperatively.

Figure 13

Combined versus not combined anaesthesia – core temperature intra- and postoperatively.

Figure 14. Duration of surgery above and below 2 hours – incidence of hypothermia.

Figure 14

Duration of surgery above and below 2 hours – incidence of hypothermia.

Overall, it is unclear whether regional anaesthesia constitutes less of a risk than general anaesthesia. This is emphasised by the evidence from the small Hendolin (1982) study that indicates that conclusions depend on the definition of IPH. We were therefore reluctant to take the results from the Lau (2001) study for the incidence of IPH (temperature less than 35.0°C) in PACU, also taking into consideration the Abelha (2005) study (temperature less than 35.0°C, not significant) and the intraoperative incidence of IPH (temperature less than 36.0°C, not significant) from the Flores Maldonado (1997) study. We have therefore erred on the side of caution and concluded that the risk of IPH has not been shown to differ between general and regional anaesthesia.

1.2. Combined versus not combined

Two studies analysed the effect of combined (both general and regional) anaesthesia versus not combined. Kongsayreepong (2003) compared combined anaesthesia with general and regional separately in 184 patients and Lau (2001) compared combined with general anaesthesia in 18,759 patients.

a. Incidence of IPH in PACU or ICU

Kongsayreepong (2003) found a statistically significant odds ratio for the incidence of IPH in ICU (temperature less than 36.0°C), favouring general and regional anaesthesia; OR 3.39 (95%CI 1.05, 10.91), although the confidence interval was wide.

Lau (2001) found a statistically significant odds ratio for the incidence of IPH in PACU (temperature less than 35.0°C), favouring regional anaesthesia; OR 2.77 (95%CI 1.69, 4.55).

Meta-analysis of the two studies in 18,943 patients gave a statistically significant odds ratio of 2.86 (95%CI 1.81, 4.51), favouring non-combined anaesthesia, with no heterogeneity (I2=0%, p=0.76).

Conclusions for type of anaesthesia

The following conclusions were drawn:

  1. Two studies showed that there was no significant difference for general versus regional anaesthesia in the incidence of IPH (temperature less than 36.0°C) intraoperatively, but in a small study (n=38) there was a statistically significant difference favouring epidural anaesthesia for temperatures less than 35.0°C. The confidence interval was very wide in the latter.
  2. One RCT in 30 patients showed a significant difference for general versus epidural anaesthesia in core temperature at 30 minutes intraoperatively, favouring epidural anaesthesia, but the confidence interval was fairly wide. There were no significant differences at 15 minutes or one hour or in PACU.
  3. Two studies compared the incidence of IPH (temperature less than 35.0°C) in PACU for general versus regional anaesthesia. One of these appeared to report there was no significant difference, but the other, very large study reported significantly less IPH for regional anaesthesia.
  4. Meta-analysis of two studies (one very large) showed the incidence of IPH in ICU or PACU was significantly higher for combined general and regional anaesthesia compared with general or regional anaesthesia separately. The definition of hypothermia did not seem to be important.
2. Duration of anaesthesia and duration of surgery

Six studies investigated the effect of the duration of anaesthesia or the duration of surgery on the incidence of hypothermia or changes in temperature (Abelha 2005 (n=185); Flores Maldonado 1999 (n=130); Frank 1992 (n=97); Frank 2000 (n=44); Kongsayreepong 2003 (n=184); Vorrakitpokatorn 2006 (n=128)). The studies investigated duration in different ways, either as a continuous variable, or as groups dichotomised at a threshold value. One study split the patients at 3 hours of anaesthesia (Abelha 2005) and two at 2 hours (Kongsayreepong 2003; Vorrakitpokatorn 2006). None of the studies considered 1 hour as a suitable cut-off point.

a. Incidence of hypothermia intraoperatively

One study (Flores Maldonado 1999) investigated the effect of duration of surgery as a continuous variable (mean duration 83 minutes, SD 59) for IPH (temperature less than 36.0°C) in 130 patients. The authors stated there was no significant effect, but numerical data were not given.

b. Incidence of hypothermia in ICU

One study (Abelha 2005) in 185 patients investigated the effect of the duration of anaesthesia on the incidence of IPH (temperature less than 35.0°C) in ICU in 185 patients. The study reported that the duration of anaesthesia, as subdivided into above and below 3 hours, was analysed in a multivariate analysis, but no results were given. It is assumed not to be significant. The range of anaesthesia time was 44 minutes to 11 hours.

Two studies recorded the effect of duration of surgery as a risk factor for the incidence of IPH in PACU or ICU. Kongsayreepong (2003) (temperature less than 36.0°C) and Vorrakitpokatorn (2006) (temperature less than 35.0°C) both investigated the duration of surgery, as subdivided into above and below 2 hours. The studies differed as follows:

There was a statistically significant effect for Kongsayreepong (2003) favouring shorter times, but no significant difference for Vorrakitpokatorn (2006). In the meta-analysis of the two studies, there was significant heterogeneity (I2=85%, p=0.01), and the confidence intervals are wide.

Overall the GDG concluded that the Kongsayreepong (2003) study was more reliable because of the greater range of operation durations and the definition of hypothermia, however there may have been confounding because of patient warming.

c. Change in core temperature intraoperatively

One study (Frank 1992) in 97 patients investigated the effect of time in the theatre, as a continuous variable, for mean durations of 4.4 to 6.6 h. The authors reported that there was no significant effect, but no data were given.

d. Core temperature in PACU

One study (Frank 2000) in 44 patients investigated the effect of duration of surgery as a continuous variable, for a range of surgery of 65 to 155 minutes. The authors reported that there was no significant effect (p=0.22), but no data were given.

e. Time to rewarm to 36°C

One cohort study (Frank 1992) in 97 patients reported the time to rewarm the patients to 36°C. The authors reported that there was no significant effect of duration of surgery as a continuous variable, for mean durations of 4.4 to 6.6 hours, but no data were given.

Conclusions

The view of the GDG was that the likely cut-off point for duration of anaesthesia would be one hour, but few studies had short term operations. The exceptions were Flores Maldonado (1999) and Kongsayreepong (2003). Therefore, most of the studies were considered unsuited to investigating duration of anaesthesia/surgery as a risk factor.

The Flores Maldonado (1999) study, in 130 patients, showed no significant effect of duration of anaesthesia, as a continuous variable on the incidence of IPH (temperature less than 36.0°C) intraoperatively (mean 83 minutes, SD 59).

The Kongsayreepong (2003) study, in 184 patients showed a significant effect of duration of surgery above and below 2 hours, on the incidence of IPH (temperature less than 36.0°C) in ICU (range 0.25 to 10.25 h).

3. Height of spinal block

One small cohort study (Frank 2000, n=44) reported a statistically significant difference in the effect of the height of the spinal block, but no data were given for the multivariate regression analysis; the p values was reported to be p=0.002. The outcome measured was core temperature in PACU for height of block as a categorical variable in the range T3 to T8, with a high level of blockade giving low core temperatures. We note that the Frank (2000) study had too many variables in total for the number of patients (44/6 = 7), so this is treated as weak evidence.

4. Positive end expiratory pressure (PEEP)

One study (Mizobe 2005) compared a positive end expiratory pressure (PEEP) at 10cm H2O versus zero end expiratory pressure (ZEEP) in 16 patients undergoing lower abdominal surgery.

There was no significant difference between 10 cm H2O PEEP and ZEEP at 20 and 40 minutes, but significantly higher core temperatures at 1 to 3 hours for patients given PEEP. This study is small, however, and the evidence is insufficient to make recommendations.

Figure 15. Positive end expiratory pressure versus zero pressure – core temperature.

Figure 15

Positive end expiratory pressure versus zero pressure – core temperature. NB: Scale −4 to +4°C

C. SURGERY RISK FACTORS

1. Magnitude of surgery

Three cohort studies (Abelha 2005 (n=185); Flores Maldonado 1997 (n=130); Kongsayreepong 2003 (n=184)) investigated the effect of magnitude of surgery on the incidence of IPH. One of the studies (Flores Maldonado 1997) divided operations into major and minor (but only defined ‘major’). In the other two studies a third category, intermediate, was defined. Operations were divided by the authors into:

1.1. Major versus minor

The three studies had different definitions of hypothermia, and recorded the incidence at different stages.

a. Incidence of hypothermia intraoperatively

One study (Flores Maldonado 1997) recorded the incidence of IPH (temperature less than 36.0°C) intraoperatively in 130 patients. There was a statistically significant effect of magnitude of surgery, with major surgery giving rise to a higher incidence of IPH.

b. Incidence of hypothermia in ICU

Two studies recorded the incidence of IPH in ICU (Abelha 2005, temperature less than 35.0°C; Kongsayreepong 2003, temperature less than 36.0°C). Meta-analysis of the two studies in 369 patients, showed a statistically significant effect, with major surgery giving rise to a higher incidence of IPH. There was significant heterogeneity, however (I2=74%, p=0.05). Each study was significant individually.

Figure 16a. Magnitude of surgery, major versus minor – incidence of hypothermia.

Figure 16a

Magnitude of surgery, major versus minor – incidence of hypothermia.

The GDG decided that the odds ratio in Kongsayreepong (2003) was unexpectedly high and so decided to carry out a meta-analysis of the other two studies, despite the differences between them in time of measurement, definition of hypothermia and possible differences in the definition of minor surgery. This meta-analysis gave an odds ratio of 3.20 (95%CI 1.68, 6.07), with no heterogeneity (I2=0%, p=0.62). This probably erred on the side of caution.

Figure 16b. Sensitivity analysis for magnitude of surgery, Kongsayreepong excluded.

Figure 16b

Sensitivity analysis for magnitude of surgery, Kongsayreepong excluded.

1.2. Intermediate versus minor

Two studies compared intermediate and minor surgery (Abelha 2005; Kongsayreepong 2003). The studies had different definitions of hypothermia.

a. Incidence of hypothermia in ICU

Two studies recorded the incidence of IPH in ICU (Abelha 2005, temperature less than 35.0°C; Kongsayreepong 2003, temperature less than 36.0°C). Meta-analysis of the two studies in 369 patients showed a statistically significant effect, with intermediate surgery giving rise to a higher incidence of IPH; OR 4.31 (95%CI 2.03, 9.13). There was no heterogeneity (I2=0%, p=0.47).

Figure 17. Magnitude of surgery, intermediate versus minor.

Figure 17

Magnitude of surgery, intermediate versus minor.

2. Urgency of surgery – elective or emergency

One cohort study (Kongsayreepong 2003 (n=184)) investigated the effect of urgency of surgery on the incidence of IPH (temperature less than 36.0°C) in ICU. There was no significant difference between elective and emergency surgery.

Figure 18. Urgency of surgery, emergency versus elective – incidence of hypothermia.

Figure 18

Urgency of surgery, emergency versus elective – incidence of hypothermia.

3. Type of surgery

Two RCTs (Nguyen 2001 [n=101]; Danelli 2002 [n=44]) compared laparoscopic and open procedures, for gastric bypass and colorectal surgery respectively. Both studies reported significantly longer durations of surgery for the laparoscopic procedure (64 minutes median difference for Danelli and 30 minutes mean difference for Nguyen). Danelli (2002) gave all patients warmed fluids; Nguyen (2001) reported that all patients had forced air warming, but fluids were not warmed.

Danelli (2002) reported median and range core temperatures, but stated that there was no significant difference between the two interventions at any time intraoperatively or postoperatively. There was no signfiicnt difference in core temperature intraoperatively for Nguyen (2001), but there were significantly higher temperatures in PACU for the open procedure (Figure 19). For the incidence of hypothermia, the confidence intervals were too wide to determine if there is a difference (Figure 20).

Figure 19. Type of surgery, laparoscopy versus open procedure – core temperature.

Figure 19

Type of surgery, laparoscopy versus open procedure – core temperature.

Figure 20. Type of surgery, laparoscopy versus open procedure – incidence of IPH.

Figure 20

Type of surgery, laparoscopy versus open procedure – incidence of IPH.

4. Patient position

One small RCT (Nakajima 2002) investigated the effect of patient position during surgery. The patients were randomly assigned to one of four positions: supine (n = 8); 15° to 20° head-down tilt (Trendelenburg position, n = 8); leg-up (lithotomy position, n = 8); leg-up combined with head-down tilt (n = 8). The designated positions were initiated 10 min after the induction of general anaesthesia and were maintained for 3 hours. There was no significant difference in core temperature between the Trendelenburg and supine positions at any time, although the confidence interval was fairly wide. There were significantly higher core temperatures at 2 and 3 hours for leg-up and leg-up with head-down tilt, in comparison with the supine position, however, the confidence intervals were fairly wide. The GDG considered that the small numbers in each comparison precluded drawing conclusions.

Figure 21. Position of patient in surgery – core temperature.

Figure 21

Position of patient in surgery – core temperature. NB: Scale −4 to +4°C

D. OTHER RISK FACTORS

1. Intravenous fluid infusion

Three cohort studies investigated the effect of intravenous fluids. Two investigated the incidence of hypothermia in ICU: Kongsayreepong (2003), temperature less than 36.0°C, n=184; Abelha (2005), crystalloid, temperature less than 35.0°C, n=185. Neither study stated if the fluids were warmed, with Abelha (2005) reporting that the number receiving warmed fluids was unknown. For Kongsayreepong (2003) the volume of fluids given was 0.1 to 11.2 litres and the volume was dichotomised into above and below 4 litres. The confidence interval was fairly wide. This study shows that fluid volume above and below 4 litres did not have a significant effect on the incidence of hypothermia.

Abelha (2005) reported a range of crystalloid fluid volumes from 0.2 to 10.5 litres, with a mean of 2.9 litres, and crystalloid volume was analysed as a continuous variable. This was found to have a statistically significant effect, with lower volumes giving less hypothermia in ICU: OR 1.4 (95%CI 1.1, 1.7). The study also included volume of colloid and this was found to be non-significant in univariate analyses.

A third study (Hind 1994a, n=30) investigated the effect of room temperature IV fluids, as a continuous variable, on the change in intraoperative temperature. The patients received 0.14 to 1.25 litres over one to two hours, and reported no significant effect. We note that this study had some methodological limitations and also reported an interaction of IV fluid volume and age.

2. Irrigation fluids

One study (Vorrakitpokatorn 2006) in 128 patients reported a large significant effect of room temperature irrigation fluid, above and below 20 litres, on the incidence of hypothermia in PACU (temperature less than 35.0°C). This was a large effect, in which lower volumes of irrigation fluids resulted in less hypothermia: OR 7.42 (95%CI 2.13, 25.94). The confidence interval was fairly wide.

Figure 22. Fluid volume – incidence of hypothermia in PACU.

Figure 22

Fluid volume – incidence of hypothermia in PACU.

3. Blood transfusion

Two cohort studies investigated the effect of blood transfusion versus no transfusion on the incidence of hypothermia; Flores Maldonado (1997) gave 13 of 130 patients blood at 4°C and Vorrakitpokatorn (2006) gave 16% of the 128 patients blood (8% had two units), but warming was not stated. Flores Maldonado (1997) found a statistically significant difference in the incidence of core temperatures below 36.0°C, but Vorrakitpokatorn (2006) found no significant difference in the incidence of temperatures below 35.0°C. The GDG thought it likely that the blood was warmed in the Vorrakitpokatorn (2006) study.

One RCT analysed by multiple regression (Frank 1992), in 97 patients, gave 0.7 to 1.2 units of warmed blood and found no significant difference in the change in intraoperative temperature or in the time to rewarm to 36.0°C, for blood transfusion treated as a continuous variable. We note that this study used sublingual temperature measurements.

Figure 23. Blood transfusion – incidence of hypothermia in PACU.

Figure 23

Blood transfusion – incidence of hypothermia in PACU.

Conclusions – fluids and blood

For intravenous fluids in the Kongsayreepong (2003) study, we considered the 4 litre threshold to be too high to be representative of the effect of fluids, and we noted that there were methodological limitations in the Hind (1994) study. The remaining study (Abelha 2005) gave weak evidence that volume of IV fluids may a risk factor for hypothermia, but the effect was fairly small. The lack of information on whether the fluids were warmed was a limitation.

There was acceptable evidence to show that a volume of more than 20 litres of unwarmed irrigation fluids was a significant risk factor for IPH.

There was acceptable evidence to show that transfusion of unwarmed blood (4°C) significantly increases the risk of IPH intraoperatively. Other studies investigating this risk factor did not state if the blood was warmed, so it was unclear whether their conclusions of no significant effect were reliable.

E. ENVIRONMENTAL RISK FACTORS

1. Theatre temperature

Six studies investigated the effect of theatre temperature on the incidence of IPH or on the core temperature (Flores Maldonado 1997; Frank 2000; Hind 1994a; Kongsayreepong 2003; Frank 1992; Morris 1971). Hind (1994) was treated with caution because only two of four risk factors were included in the multivariate analysis and the study also had too many variables in total for the number of patients (30/6 = 5).

a. Incidence of IPH intraoperatively

One study (Flores Maldonado 1997) in 130 patients reported the effect of theatre temperature, as a continuous variable, on the incidence of IPH intraoperatively (temperature less than 36.0°C). This showed a large statistically significant effect of theatre temperature for a mean of 22.9°C (SD 1.2) in patients undergoing either general or regional anaesthesia; OR 0.61 (95%CI 0.42, 0.89).

b. Incidence of IPH in ICU

One study (Kongsayreepong 2003) in 184 patients undergoing combined, general or regional anaesthesia, for a theatre temperature of mean 19.5 to 20.6°C (SD 1.8), reported the incidence of IPH in ICU (temperature less than 36.0°C), and showed an almost identical odds ratio to that obtained intraoperatively (Figure 24), statistically significantly in favour of warmer theatres.

Figure 24. Effect of theatre temperature – incidence of IPH intraoperatively and in ICU.

Figure 24

Effect of theatre temperature – incidence of IPH intraoperatively and in ICU.

c. Core temperature intraoperatively

One small cohort study (Morris 1971), in 22 patients undergoing general anaesthesia, compared the effect of theatre temperature in two groups: cool theatre (18 to 21°C) and warm theatre (21 to 24°C). There was a statistically significant effect at all times (Figure 25). The control group was hypothermic at one hour in the cooler theatre.

Figure 25. Effect of theatre temperature – core temperature intraoperatively and in ICU.

Figure 25

Effect of theatre temperature – core temperature intraoperatively and in ICU.

d. Change in temperature intraoperatively

Two studies reported the effect of theatre temperature on the change in temperature intraoperatively.

Frank (1992) (n=97) compared warm (24.5°C) and cool theatres (21.3°C) in patients undergoing either general or epidural anaesthesia and reported no statistically significant effect of theatre temperature on the difference between the ‘first postoperative temperature’ and the preoperative temperature (p=0.07). The forest plot demonstrates the confidence interval is wide, but warmer theatre temperatures are favoured (Figure 26).

Figure 26. Effect of theatre temperature – change in core temperature intraoperatively.

Figure 26

Effect of theatre temperature – change in core temperature intraoperatively.

Hind (1994a), in 30 patients undergoing general anaesthesia, reported a statistically significant effect (p<0.001) of theatre temperature for a mean of 21.3°C (SD 1.2); range 19.6 to 23.3. We note that this study reported correlations between age and theatre temperature, which the authors attributed to older patients being in the theatre at the start of the list when the theatre was at its coldest. Hind (1994a) was also of poorer quality.

e. Core temperature in PACU

Another cohort study (Frank 2000) in 44 patients, reported that, in a multiple regression analysis, there was no statistically significant effect (p=0.70) of theatre temperature for a mean of 20.9°C (SD 0.13), with a range of 18.7 to 22.9°C. No other numerical data were given. This study only included patients receiving spinal anaesthesia.

f. Time to rewarm to 36.0°C

One study (Frank 1992) in 97 patients reported no significant effect of theatre temperature on rewarming patients in warm (24.5°C) versus cool theatres (21.3°C) in patients undergoing either general or epidural anaesthesia).

2. Interaction between theatre temperature and type of anaesthesia
a. Change in core temperature

One study (Frank 1992) in 97 patients included interaction terms in the multivariate analysis, and reported a statistically significant effect of a combination of theatre temperature and type of anaesthesia on the change in temperature intraoperatively between the ‘first postoperative temperature’ and the preoperative temperature. There was a greater decrease in temperature for general anaesthesia versus epidural in a colder theatre (21.3°C), than in a warmer theatre (24.5°C). This is illustrated in Figure 27. We note that these are not randomised groups. There is a statistically significant difference in the colder theatre, favouring epidural anaesthesia, but there is no significant difference at warmer temperatures. The confidence intervals are wide.

Figure 27. Epidural versus general anaesthesia for theatre temperature subgroups.

Figure 27

Epidural versus general anaesthesia for theatre temperature subgroups.

These subgroup results support the observation found for the Frank (2000) study in spinal anaesthesia, in which there was no effect of theatre temperature (for a range of 18.7 to 22.9°C).

3. Interaction between theatre temperature and age
a. Change in core temperature

One study (Frank 1992) in 97 patients reported no significant effect of a combination of theatre temperature and age on the change in temperature intraoperatively between the ‘first postoperative temperature’ and the preoperative temperature.

Conclusions

The evidence suggests that:

  • In patients undergoing general (mainly) or combined or regional anaesthesia, an increase in theatre temperature is protective of patients becoming hypothermic, both intraoperatively and in ICU.
  • In patients undergoing general anaesthesia, one small study (n=22) reported that increased core temperatures are obtained intraoperatively in a warmer theatre (24°C versus 21°C).
  • In patients undergoing spinal anaesthesia, one study reported no significant effect of theatre temperature in the range 18.7 to 22.9°C.
  • One moderately sized study (n=97) reported there is an interaction between type of anaesthesia and theatre temperature, such that there is a smaller effect of theatre temperature for epidural compared with general anaesthesia
  • There does not appear to be a threshold above which further increases in theatre temperature have no effect.
3. Humidity

One study (Hind 1994a), in 30 patients, investigated the effect of theatre humidity in the range 50 to 65%, and found that this was not significantly correlated with the core temperature, so this risk factor was excluded from the multivariate analysis. We note that Hind (1994a) is poorer quality.

Copyright © 2008, National Collaborating Centre for Nursing and Supportive Care.
Cover of The Management of Inadvertent Perioperative Hypothermia in Adults
The Management of Inadvertent Perioperative Hypothermia in Adults [Internet].
NICE Clinical Guidelines, No. 65.
National Collaborating Centre for Nursing and Supportive Care (UK).

Download

NICE (National Institute for Health and Care Excellence)

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...