Journal of Applied Physiology
Vol. 82, No. 3, pp. 1008-1017, March 1997
CONTROL OF BREATHING, CIRCULATION, AND TEMPERATURE

Dew-point hygrometry system for measurement of evaporative water loss in infants

Ronald L. Ariagno¹, Steven F. Glotzbach², Roger B. Baldwin¹, David M. Rector³, Susan M. Bowley⁴, and Robert J. Moffat⁵

¹ Department of Pediatrics, Stanford University School of Medicine, Stanford, California 94305-5119; ² Program in Human Biology, Stanford University, Stanford, California 94305-2160; ³ Brain Research Institute, Universty of California, Los Angeles, California 90095; ⁴ National Aeronautics and Space Administration Ames Research Center, Moffett Field, California 94035-1000; and ⁵ Department of Mechanical Engineering, Stanford University, Stanford, California 94305-3030

ABSTRACT

INTRODUCTION

METHODS

RESULTS

DISCUSSION

APPENDIX

ACKNOWLEDGEMENTS

FOOTNOTES

REFERENCES

ABSTRACT

Ariagno, Ronald L., Steven F. Glotzbach, Roger B. Baldwin, David M. Rector, Susan M. Bowley, and Robert J. Moffat. Dew-point hygrometry system for measurement of evaporative water loss in infants. J. Appl. Physiol. 82(3): 1008-1017, 1997.Evaporation of water from the skin is an important mechanism in thermal homeostasis. Resistance hygrometry, in which the water vapor pressure gradient above the skin surface is calculated, has been the measurement method of choice in the majority of pediatric investigations. However, resistance hygrometry is influenced by changes in ambient conditions such as relative humidity, surface temperature, and convection currents. We have developed a ventilated capsule method that minimized these potential sources of measurement error and that allowed second-by-second, long-term, continuous measurements of evaporative water loss in sleeping infants. Air with a controlled reference humidity (dew-point temperature = 0°C) is delivered to a small, lightweight skin capsule and mixed with the vapor on the surface of the skin. The dew point of the resulting mixture is measured by using a chilled mirror dew-point hygrometer. The system indicates leaks, is mobile, and is accurate within 2%, as determined by gravimetric calibration. Examples from a recording of a 13-wk-old full-term infant obtained by using the system give evaporative water loss rates of ~0.02 mgH₂O ^· cm^{2 ·} min¹ for normothermic baseline conditions and values up to 0.4 mgH₂O ^· cm^{2 ·} min¹ when the subject was being warmed. The system is effective for clinical investigations that require dynamic measurements of water loss.

evaporation; water loss; sweating; sleep; infant

INTRODUCTION

EVAPORATIVE WATER LOSS (EWL) refers to the total water loss through the skin, including both "sensible" (i.e., sweating) and "insensible" (i.e., vapor diffusion through the epidermis) water loss. Transepidermal water loss (TEWL) is the portion of EWL that moves through the skin by diffusion, as contrasted with liquid water ejected from the sweat glands.

EWL is an important component of heat exchange in the newborn infant because of the permeability of the stratum corneum and the infant's relatively large surface area-to-volume ratio. EWL is a major cause of water loss and energy loss in premature infants during the first week after birth (6, 12, 14, 23, 26, 31). In older patients, accurate measurement of EWL is important to studies of burn damage and is also useful in the assessment of skin barrier function in a variety of dermatological applications (27, 29). Some studies in adult humans (15, 16, 21) have reported decreased sweat rates in rapid-eye-movement (REM) sleep, which supports the contention that inhibition of thermoregulatory mechanisms occurs during REM compared with non-REM (NREM) sleep or awake states. However, Amoros et al. (1) reported that cessation of sweating during REM was not supported in their study. Recent studies have reported that local sweating rate may be similar during REM and NREM states in premature infants during warming (4) and that evaporative water loss in term infants, aged between 2 days and 3 mo, is not different in REM and NREM states (3). Excessive sweating has been reported in siblings of sudden infant death syndrome (SIDS), and some SIDS victims had an elevated temperature at death (2, 8, 9, 13, 17, 24, 28). The higher proportion of REM in infants, compared with adults, makes the effect of sleep state on thermoregulatory capabilities important in these investigations. Studies of thermoregulation and EWL in the infant may be important to investigate abnormalities associated with SIDS.

The relationships between body temperature, sleep, and insensible loss of weight have been the subject of scientific investigations for over 50 years (7). Determinations of EWL in infants have been made by a number of techniques that have been discussed in several publications (10, 20, 25).

Our research required a system less subject to environmentally induced artifacts, one which could provide accurate, time-resolved, local measurements of EWL, with a time constant sufficiently low that transient events with a period of <1 min could be detected. Such an instrument would be advantageous for studies examining dynamic change in EWL. The ventilated-capsule system developed in our laboratory addresses these requirements. The purpose of this communication is to describe the system and to present bench and clinical performance data.

METHODS

Figure 1 is a schematic depicting the key system components: 1) air preconditioner (chiller bath) to establish a consistent, known input condition; 2) flowmeter upstream of the vapor pickup site as one of two components necessary for leak detection; 3) pickup capsule over the measurement site; 4) dew-point temperature (T_dp) sensor; 5) flowmeter downstream of the sensor as the second component of leak detection; 6) controlled suction; and 7) hydrophobic tubing between the preconditioning and T_dp measurement components to maintain the integrity of the sample. Other parameters (skin, rectal, room temperatures, etc.) can be monitored simultaneously.

Inlet-stabilizing bath. The need for an inlet-stabilizing bath was clearly demonstrated by a series of trials in which significant scatter was observed due to variations in inlet dew point (as much as 1°C) during a measurement period. Chemical dryers (anhydrous calcium sulfate chambers) were tried, but these dried the air too much. Dual T_dp sensors were tested, but the passive chiller bath was more easily maintained.

An ice-slurry, chiller bath apparatus (Fig. 2) preconditions the air to 0°C, providing a stable reference T_dp in the incoming air. Copper tubing (~4.25 m × 1/4-in. ID) was rolled into a coil (10 cm high × 14 cm in diameter), with the low end bent to a gentle right angle to pass through one hole of a two-hole stopper. The stopper covers a petri dish (15 × 55 mm), which has a layer of water at the bottom. The second hole accepts a piece of thick-wall Teflon tubing (1/4 in. diameter × 55 mm length), which mates to corrugated Teflon tubing. This apparatus is placed into an ice-water slurry [some ice is held in position below the dish by an aluminum-wire mesh (1.5 × 1.5 mm) lattice]. Air is drawn through the chilled coil, over the water in the dish and exits to a flowmeter. This arrangement precipitates moisture when the ambient T_dp is >0°C and adds moisture if ambient T_dp is <0°C.

Interconnecting tubing. Teflon tubing was used between the stabilizing bath and the dew-point sensor to avoid absorption or desorption of water in the walls of the tubes. Repeatability tests and hysteresis tests conducted to investigate the system's performance under clinical conditions showed that the Tygon tubing, originally used, was a source of error. Absorption and desorption greatly increased stabilizing time (due to the time required to "load" or "unload" the tubing), and results were sensitive to changes in room temperature or thermal irradiation of the tubing. As long as the room air temperature is above the measured T_dp of the sample, neither insulation nor heating of the tubing is necessary.

Upstream flow measurement. It is essential to eliminate leaks in advance in the system before the sample reaches the T_dp-measuring chamber. The system operates subatmospherically, so leakage would bring room air into the system. Leakage can be detected by noting a discrepency between the upstream and the downstream flowmeters. A digital flowmeter measures flow upstream from the capsule, and the resultant value is sampled by a data-acquisition system at 0.1 Hz and stored to hard disk. Core body (rectal), skin, and other temperatures are also sampled at 0.1 Hz and stored. The resultant data arrays are downloaded to a desktop computer for statistical and graphic analysis. The flowmeter has a special calibration showing accuracy of ±0.4% in the flow range usually used (0-100 ml/min).

Pickup capsule. The flowmeter is connected to a pickup capsule (Fig. 3) with convoluted and corrugated Teflon fluorinated ethylene propylene (FEP) tubing with connectors of thick-wall Teflon tubing. Plastic luer fittings on the tubing mate to the manifolds of the pickup capsule or to a 30-cm Teflon coil, which is used periodically to check baseline chiller bath conditioning. The pickup capsule was milled from clear acrylic, is held together with screws, and is affixed to the skin by double-sided tape and medical-grade adhesive dressing. No force or pressure was therefore necessary. To prevent leaks, care needed to be taken not to distort the skin of the subject when affixing and maintaining the capsule. Air enters the capsule through four holes (0.1-cm diameter) on one side of the collecting chamber, passes over the skin site, then exits through four holes (0.1-cm diameter) in the facing side. This arrangement, designed for turbulence, produces a well-mixed sample. The pickup capsule weighs 9.3 g; has external dimensions of 3.5 × 2.8 × 1.2 cm; covers an exposed skin area 1.97 × 0.97 cm, yielding a collection area of 1.91 cm²; and has a chamber volume of 1.55 cm³. There is sufficient wall thickness (0.4 cm) on the collecting chamber to allow it to be comfortably attached to the subject. With a flow of 100 ml/min, the average air velocity across the skin inside the chamber is 0.04 m/s.

2

1

RESULTS

System characteristics were quantified by a number of tests and analyses that investigated the steady-state calibration accuracy and response to an abrupt change in input.

Calibration accuracy. A gravimetric calibration procedure was used to determine the accuracy of the system. A device with a constant air-water interface ("wet block") was constructed to assess system performance. The wet block design is shown schematically in Fig. 4. The pickup capsule is held in place on an aluminum plate stage over a water reservoir within an acrylic block during a test run. The aluminum stage, with an intervening rubber gasket, covers the top of a cavity in the block and is secured by screws. The stage has an opening that is filled with cotton and that has a tail extending into the water reservoir. The cotton provides a wet surface, level with the stage surface, which is maintained by water wicked from the reservoir. The surface of the stage is smooth and provides a leak-free surface for the pickup capsule. This arrangement allows continuous evaporation to take place from the top of the wick without substantially changing the air-water interface, in contrast to evaporation from a drop of water or wetted piece of paper. The block was weighed on an electronic analytic balance just before and just after test periods. Typical weight loss was ~0.065 g. The wet block was at room temperature (~20°C), yielding EWL_c rates of ~0.15 mgH₂O ^· cm^{2 ·} min¹, which is within the expected range of rates for measurements in patients (0.017-1 mgH₂O ^· cm^{2 ·} min¹). Data were recorded over test periods of ~210 min, downloaded to a desktop computer, and integration of the second-to-second EWL_c rates yielded "system EWL_c," i.e., total calculated EWL_c system water loss. These values were compared with the "weighed EWL"; i.e., the weight loss of the wet block during the measurement period.

Representative results. The wet-block test data in the following Table 1 are those achieved with the final design of the EWL measurement system. Note the repeatability over long intertest intervals.

Table 1. Testing results Date Weighed EWL, g System EWLc, g %Weighed  1 Jun '95 0.0672 0.0669 99.55 22 Mar '95 0.0697 0.0693 99.43 31 Jan '94 0.0635 0.0627 98.74 29 Jan '94 0.0605 0.0593 98.02 27 Jan '94 0.0645 0.0643 99.69 Accuracy, expressed as %weight change (n = 5, mean ± SD) was 99.09 ± 0.70%. EWL, evaporative water loss; EWLc, capsule EWL.

Based on these data sets, the measurement uncertainty of the system is within the theoretical estimation (see Appendix).

Response characteristics. The transient response of the system depends on the void volume within the system and the thermal time constant of the sensor. The void volume depends on the volumes of the pickup capsule and T_dp measuring chamber and on the length and diameter of the connecting tubing. With 1 m of tubing, the system's flow time constant (volume divided by flow rate) is ~20 s. The humidity sensor's characteristic time is ~15 s. Overall, the system requires ~80 s to complete its response to a step change (Fig. 5A). Figure 5B is a y-axis magnification of the baseline test condition. Figure 5, C-E, shows the results of bench testing to elucidate the response characteristics of the system. Multiple sequential inflation and deflation cycles of a water droplet under the pickup capsule were done at different cycle lengths (20-, 40-, and 60-s durations). The detection of periodicities appears to be valid for cycles on the order of tens of seconds.

2

1

2

1

2

1

In vivo examples. Figures 6 through 8 show examples from a recording from a 13-wk-old healthy male full-term infant in a study evaluating the relationship between sleep state and EWL_c. Figure 6 is from a representative 3.5-h recording showing the 0°C T_dp condition at the start and the end of the session. Figures 7 and 8 show additional time-related clinical examples of EWL_c rates measured with this system. Figure 7 shows a rapid increase in EWL_c after the subject was exposed to 20 min of warming. Figure 8 shows cyclical patterns in the same subject. Further analysis and interpretation of these clinical phenomena will follow in a future publication. We believe the variations in EWL_c are real and that the system allows the examination of dynamic changes in EWL_c rate.

DISCUSSION

The rate of TEWL from skin depends on four primary factors: 1) the subcutaneous skin temperature; 2) the permeability of the skin to diffusion of water vapor; 3) the mass transfer coefficient between the skin and the ambient air; and 4) the concentration of water vapor in the ambient air. Any measurement system that affects these parameters will likely alter TEWL and, hence, alter EWL. Ideally, the measured EWL should represent the undisturbed value of EWL. The subcutaneous temperature and the skin permeability are patient and situation specific and may, in fact, be the objective of a measurement series. The mass transfer coefficient depends on the mean velocity and turbulence intensity in the airflow around the skin. The ambient humidity is part of the patient-care environment. The clinical examples provided in this report are estimates of EWL_c from chest and back sites; however, estimates for the total body could be obtained by sampling from multiple sites (11). Estimates of the boundary layer characteristics in infant and adult patients suggest that the TEWL of healthy full-term infants and of adults with healthy skin is only slightly affected by air velocity. The overall resistance to water loss is the sum of two resistances: that of the stratum corneum and that of the boundary layer above the skin (inversely proportional to the mass transfer coefficient). For healthy full-term infants and adults with healthy skin, substantially all of the overall resistance to TEWL lies in the stratum corneum, with the boundary layer resistance contributing only 5-10% of the total. The mass transfer coefficient varies approximately with the square root of the air velocity. A change of 50% in the air velocity would decrease the mass transfer coefficient by only 25%. Thus an increase in the air velocity by 50% would change the TEWL by <2.5%. The other component of EWL, active sweating, represents water loss by direct evaporation from liquid water on the surface of the skin. Whereas the evaporation rate of an existing droplet is directly proportional to the mass transfer coefficient, the rate at which the droplets are produced is set by the thermoregulatory mechanisms of the patient.

In this communication, we have described a system that can measure EWL with an accuracy of ±2%, based on gravimetric calibration, and has the capability of detecting periodicities on the order of tens of seconds associated with sweating. The system is not affected by environmental changes (e.g., temperature, humidity, and air currents) and has the advantage of measuring EWL_c during steady-state and dynamic changes in evaporative water loss rate with minimal disruption to the subject. The modifications described are the result of attempting to use existing techniques and finding significant limitations for the study of steady-state and dynamic changes in EWL during multihour studies of infants. In pediatrics, many clinical investigations have utilized resistance hygrometry (Evaporimeter) (20, 22, 25, 29). The Evaporimeter uses the water vapor concentration gradient within a hollow cylindrical placed perpendicular to the skin surface to determine evaporative water loss. This measurement condition changes the local evaporation rate by introducing extra resistance due to the height of the cylinder. This technique was investigated by Wheldon and Monteith (30), who concluded that the Evaporimeter would underestimate "true" water loss by 10% when indicating 20 gH₂O ^· m^{2 ·} h¹ and by 60% or more at rates of ~80 gH₂O ^· m^{2 ·} h¹. Blichmann and Serup (5) quote the manufacturer's accuracy statement as "±15% or 2 gH₂O ^· m^{2 ·} h¹, whichever is larger" and report tests showing the "technical accuracy" (in vitro calibration under carefully controlled conditions) to be ±3.3%. Their in vivo results showed reproducibility, same location on the same subject, between ±6.3 and ±14.5%, depending on the time between successive measurements. This in vivo experience was corroborated by the studies of Smit et al. (27), who reported 13.5% (same day) and 15.1% (consecutive days) intraindividual coefficients of variation.

A recent technique developed by Graichen et al. (10) uses a Peltier module device (Bi-Tronics Dewpoint Sensor, model BI-102). This technique held the expectation for rapid and accurate measurements, but after using it for some months we found that it fell short of our needs. In this method, the sample capsule receives ambient air from the environment adjacent to the capsule, resulting in sampling site vapor mixed with vapor of unknown humidity composition. Nilsson (20) described the problems that must be solved if a ventilated chamber system was to work: 1) precise measurement of the humidity of the inlet and exit streams; 2) thoughtful selection of the flow rate, to match the normal mass transfer coefficient; 3) susceptibility to leakage; and 4) possible physiological interferences if too much force was applied to prevent leaks. The problem of obtaining a precise measurement of the humidity of the BI-102 inlet air was addressed by instituting a 2-m piece of tubing that allowed us to pull presample vapor from a more removed site. The instrument was unusable in conditions in which the dew point is <2°C, an ambient condition that can occur in our environment. It also had no capability for detection of leaks in the system, and the cable connecting the Peltier module to the main console had substantial torque, which made maintenance of an adequate leak-free measurement condition difficult.

In our system, two flowmeters are used to detect leaks, allowing corrective action and documentation of nonoptimal measurement conditions. The air inlet to the pickup capsule is a set of small jets, which increases mixing inside the pickup capsule to simulate a "well-stirred reactor." This ensures that mass transfer from the skin can be evaluated between the surface concentration at the skin and the mixed mean concentration in the capsule. With the assumption of a well-stirred reactor, the mass transfer is related to the difference between the mixed mean concentration in the chamber and the surface concentration. The mixed mean concentration in the pickup capsule is the value that is measured by the T_dp sensor. With the above problems addressed, significant improvements in precision and accuracy in the measurement of TEWL were possible.

Advantages of our system include 1) continuous long-term recording capability; 2) excellent accuracy and repeatability in dynamic testing; 3) excellent precision; 4) minimal susceptibility to environmental perturbation compared with other devices, e.g., Evaporimeter and Dew-Point Hygrometer; 5) calibration checks at beginning and end of data acquisition; 6) leak detection; 7) maintenance of integrity of sample quasi-bolus; and 8) standardization of measurement conditions. Some limitations of our system are as follows: 1) additional control systems would be necessary to run the system in high EWL_c rate conditions occurring in a cool testing situation (i.e., over ~1.167 mgH₂O ^· cm^{2 ·} min¹ at ~75°F); 2) tubing to patient could be restrictive (or could be dislodged) in an ambulatory or vigorously active patient; 3) slight delay (~80 s for a step change) in determinations; and 4) at present, nonambient measurement conditions.

In summary, the accuracy of this system, based on gravimetric calibration, is ±2% and stable to environment disturbance (e.g., changes in temperature, humidity, and air currents). The system has the advantage of examining steady-state and dynamic changes in evaporative water loss rate with minimal disturbance of the subject. Therefore, it is very suitable for studies on the development of the thermoregulatory effector responses and the effect of sleep development or sleep state on those responses.

ACKNOWLEDGEMENTS

We thank Don Doucet (General Eastern Instruments) for suggestions on sample preconditioning; Wolfgang Jung (Dept. of Physics, Stanford Univ.) and John Wallace (National Aeronautics and Space Administration Ames Research Center) for assistance in design and manufacture of components; and Margaret Boeddiker (Dept. of Pediatrics, Stanford Univ.) for assistance with this project.

FOOTNOTES

Address for reprint requests: R. L. Ariagno, Dept. of Pediatrics, Stanford Univ. School of Medicine, Stanford, CA 94305-5119.

Received 15 November 1995; accepted in final form 28 October 1996.

APPENDIX

Equations for Calculation of EWL (_{s 2})

2

1

The value of SH₁ is set by the equation of state of water and the design of the inlet conditioning tank. Its value is known a priori and will be considered to have negligible uncertainty.

The equations used treat the mixture density at the flowmeter location as a constant, equal to that of air at standard conditions of temperature and pressure. Because flowmeters are essentially in equilibrium with the room air temperature, this seems a reasonable assumption.

Uncertainty

The remaining factor is SH₂/SH₂. We estimate this from the uncertainty in the measured T_dp temperature (±0.02°C), by using a polynomial curve fit to the relationship between SH and T_dp. Based on that curve fit with an uncertainty of ±0.2°C in the measured T_dp (a conservative estimate).

Then

Tdp	SH	SH/SH
0	6 × 106	0.016
10	10 × 106	0.013
20	16 × 106	0.011

The overall uncertainty in _{s 2} is, therefore, a slight function of the operating T_dp

Tdp 2	s 2 /s 2
0	3.3%
10	3.1%
20	3.0%

These estimates reflect uncertainty in the measurement but do not assess whether the act of measurement altered the patient's behavior. The uncertainty reflects the effect on the calculated _{s 2} associated with the measurements of T_dp, flow rate, and area.

REFERENCES