Free
Correspondence  |   February 2005
Water versus  Air Warmers
Author Notes
  • McGill University, Montreal General Hospital, Montreal, Quebec, Canada.
Article Information
Correspondence
Correspondence   |   February 2005
Water versus  Air Warmers
Anesthesiology 2 2005, Vol.102, 479-480. doi:
Anesthesiology 2 2005, Vol.102, 479-480. doi:
To the Editor:—
Taguchi et al.  1 stress the essential role of skin temperature and Newton’s Second Law of Thermodynamics in increasing core temperature (Tc) with skin warming systems, but they do not demonstrate this role and they confuse other issues.
To increase Tc directly, the Second Law requires that the warming system’s average cover temperature (Tcvr) > the mean skin temperature of multiple skin sites (Tmsk) > Tc. To demonstrate this, figure 1shows that in 12 post aortocoronary bypass patients rewarmed for 2 h under a 1.8 m2radiant ceiling, the change in Tc (ΔTc) occurred only when (Tmsk − Tc) > 0°C. If the temperature gradient is Tcvr > Tmsk < Tc, then heat passes from the cover to the skin but not to the core, and the warmer acts as an insulator, increasing Tc indirectly by retaining metabolic heat in the core.
Fig. 1. Twelve post aortocoronary bypass patients rewarmed for 2 h under a 1.8 m2radiant ceiling. X-axis: mean skin − tympanic core temperature (Tmsk − Tc) at 10-min intervals. Y-axis: the change in Tc from initial value (ΔTc). The 95% confidence interval (CI) is for all values of (Tmsk − Tc) when ΔTc < 0°C. The mean radiant temperature (Tmr) at the patient (  i.e.  , the driving temperature for radiant heat exchange) and the initial values of Tmsk and Tc are shown. Tmsk was averaged from 4 skin sites: chest, lateral upper arm, thigh and lateral calf. 
Fig. 1. Twelve post aortocoronary bypass patients rewarmed for 2 h under a 1.8 m2radiant ceiling. X-axis: mean skin − tympanic core temperature (Tmsk − Tc) at 10-min intervals. Y-axis: the change in Tc from initial value (ΔTc). The 95% confidence interval (CI) is for all values of (Tmsk − Tc) when ΔTc < 0°C. The mean radiant temperature (Tmr) at the patient (  i.e.  , the driving temperature for radiant heat exchange) and the initial values of Tmsk and Tc are shown. Tmsk was averaged from 4 skin sites: chest, lateral upper arm, thigh and lateral calf. 
Fig. 1. Twelve post aortocoronary bypass patients rewarmed for 2 h under a 1.8 m2radiant ceiling. X-axis: mean skin − tympanic core temperature (Tmsk − Tc) at 10-min intervals. Y-axis: the change in Tc from initial value (ΔTc). The 95% confidence interval (CI) is for all values of (Tmsk − Tc) when ΔTc < 0°C. The mean radiant temperature (Tmr) at the patient (  i.e.  , the driving temperature for radiant heat exchange) and the initial values of Tmsk and Tc are shown. Tmsk was averaged from 4 skin sites: chest, lateral upper arm, thigh and lateral calf. 
×
Instead of Tmsk the authors derive the peripheral temperature of the limbs (Tperif). If Tperif represents the buffer between Tcvr and Tc, then Tc did not increase directly because Tperif < Tc. Nor did Tc increase indirectly, for metabolic heat production was similar in both groups but ΔTc was not. What clinically relevant and measurable temperature gradient explains the increase in Tc?
Heat content differences between periphery and core do not reflect tissue insulation. The units of insulation (i.e.  , thermal resistance) are temperature gradient/heat transfer rate (°C/kCal/h or °C/W), not heat content (kCal or kJ). Tissue insulation is proportional to the temperature gradient required to cause heat transfer and increase Tc, which, in fig. 1, required a (Tmsk − Tc) of 1 ± 0.8°C (n = 94). In the authors’ study Tperif represented a heat sink and heat transfer to the core was “constrained” by the lack of an appropriate temperature gradient rather than by tissue insulation.
Peripheral heat content increased 2.1–2.5 times more than core because initially Tperif < Tc (water, −2.6°C; air, −2.1°C) and, therefore, exposed to the same Tcvr, the periphery warmed preferentially. Calculated from the change in heat content at the end of warming (ΔQ, their fig. 4), ΔTc =+2.9°C (water) and +1.7°C (air), as shown in their figure 3. Similarly, ΔTperif =+5.8°C (water) and +4°C (air). Therefore, their data implies a final (Tperif − Tc) of + 0.3°C (water) and +0.2°C (air), which is more consistent with the Second Law and the observed increase in Tc than the “−0.8°C” shown in their figure 5.
At the end of warming the change in heat content per unit area warmed was 1.25 times larger for water (185 kCal/m2) than for air (148 kCal/m2). Yet the authors write “. . . heat transfer (rate) per unit anterior area” was similar. This inconsistency results from the measurement method. Heat content is derived from temperature. Heat transfer rate per unit area, or “heat flux” (q”), is directly measured by heat flux transducers. Heat flux, q”=ΔT·h  , where ΔT is the cover to skin temperature gradient across the heat flux transducers and h  , from Newton’s Law of Cooling, is the heat transfer coefficient (i.e.  , the efficiency) of that particular system, usually expressed in W/m2·°C. For one conventional water mattress2 (and three others, including a suit similar to this study) h  was 1.5 times larger than that for four air warmers:3 40 W/m2·°C versus 26 W/m2·°C. The larger h  of water means that the same q” will produce a smaller ΔT than with air and, therefore, at the same Tcvr, Tmsk will be higher with water than with air—and heat content is derived from Tmsk. Although Tcvr was not measured, Tmsk was, and it would be instructive to know the Tmsk difference between the systems.
The change in heat content/warmed area for water was 1.25 times that of air, not 1.5 times, as the ratio of their h  values suggests, because ΔT differed. The air warmer was set to “43°C.” The low heat capacity of air and environmental heat loss would degrade Tcvr to a relatively constant Tcvr <43°C. The water system was set to a target Tc of “37°C.” The high heat capacity and mass flow rate of water would prevent Tcvr environmental degradation. But the system’s servocontrol algorithms would produce a variable Tcvr: initially high (40°C) but decreasing as Tc increased. As Tcvr decreased, (Tcvr − Tmsk) also decreased until, finally, the water heat transfer rate equaled that of air (their fig. 1). Had ΔT been the same, the difference between the systems would have been greater.
It is improbable that an air warmer above the body and water mattress below would match the heat transfer of a water suit because above the body, the change in heat content/warmed area was less with air than with water and below the body, a contour-conforming water suit gives better skin contact, which is essential for conductive heat transfer, than a flat mattress.
This was a sophisticated study of thermoregulatory physiology rather than heat transfer physics. It did not report those fundamental differences in the heat transfer characteristics (i.e.  , Tcvr, Tmsk, and h  ) that must be known to safely and effectively exploit the full potential of any skin-warming system. The heat transfer rate, in isolation, is a description of what happened in a specific application. It is not an explanation for why it happened, nor is it suitable for universal application in any clinical situation—those answers lie in the system’s heat transfer characteristics.
Although anesthesiologists believe that air is a more effective heat exchanger than water,4 in lethal thermal environments (space and diving) water, not air, is used to maintain life because of its better thermal properties. This article and another5 will encourage further clinical evaluations of water-conditioned heat transfer systems. Those evaluations should be based on the physical principles that determine heat transfer and not on the limitations of current clinical practice.
McGill University, Montreal General Hospital, Montreal, Quebec, Canada.
References
Taguchi A, Ratnaraj J, Kabon B, Sharma N, Lenhardt R, Sessler DI, Kurz A: Effects of a circulating-water garment and forced-air warming on body heat content and core temperature. Anesthesiology 2004; 100:1058–64Taguchi, A Ratnaraj, J Kabon, B Sharma, N Lenhardt, R Sessler, DI Kurz, A
English MJM, Farmer C, Scott WAC: Heat loss in exposed volunteers. J Trauma 1990; 30:422–5English, MJM Farmer, C Scott, WAC
Perl T, Bräuer A, Timmermann A, Mielck F, Weyland W, Braun U: Differences among forced-air warming systems with upper body blankets are small. A randomized trial for heat transfer in volunteers. Acta Anesthesiol Scand 2003; 47:1159–64Perl, T Bräuer, A Timmermann, A Mielck, F Weyland, W Braun, U
Sessler DI, Moayeri A: Skin-surface warming: Heat flux and central temperature. Anesthesiology 1990; 73:218–24Sessler, DI Moayeri, A
Janicki PK, Higgins MS, Janssen J, Johnson RF, Beattie C: Comparison of two different temperature maintenance strategies during open abdominal surgery: Upper body forced-air warming versus whole body water garment. Anesthesiology 2001; 95:868–74Janicki, PK Higgins, MS Janssen, J Johnson, RF Beattie, C
Fig. 1. Twelve post aortocoronary bypass patients rewarmed for 2 h under a 1.8 m2radiant ceiling. X-axis: mean skin − tympanic core temperature (Tmsk − Tc) at 10-min intervals. Y-axis: the change in Tc from initial value (ΔTc). The 95% confidence interval (CI) is for all values of (Tmsk − Tc) when ΔTc < 0°C. The mean radiant temperature (Tmr) at the patient (  i.e.  , the driving temperature for radiant heat exchange) and the initial values of Tmsk and Tc are shown. Tmsk was averaged from 4 skin sites: chest, lateral upper arm, thigh and lateral calf. 
Fig. 1. Twelve post aortocoronary bypass patients rewarmed for 2 h under a 1.8 m2radiant ceiling. X-axis: mean skin − tympanic core temperature (Tmsk − Tc) at 10-min intervals. Y-axis: the change in Tc from initial value (ΔTc). The 95% confidence interval (CI) is for all values of (Tmsk − Tc) when ΔTc < 0°C. The mean radiant temperature (Tmr) at the patient (  i.e.  , the driving temperature for radiant heat exchange) and the initial values of Tmsk and Tc are shown. Tmsk was averaged from 4 skin sites: chest, lateral upper arm, thigh and lateral calf. 
Fig. 1. Twelve post aortocoronary bypass patients rewarmed for 2 h under a 1.8 m2radiant ceiling. X-axis: mean skin − tympanic core temperature (Tmsk − Tc) at 10-min intervals. Y-axis: the change in Tc from initial value (ΔTc). The 95% confidence interval (CI) is for all values of (Tmsk − Tc) when ΔTc < 0°C. The mean radiant temperature (Tmr) at the patient (  i.e.  , the driving temperature for radiant heat exchange) and the initial values of Tmsk and Tc are shown. Tmsk was averaged from 4 skin sites: chest, lateral upper arm, thigh and lateral calf. 
×