No new eligible trials were found. Thus, there is no change to the conclusion that maintaining an abdominal skin temperature at 36C in low birth weight babies in incubators reduces the neonatal death rate compared to setting a constant incubator air temperature of 31.8C.
In low birth weight infants, to determine the effect on death and other important clinical outcomes of targeting body temperature rather than air temperature as the end-point of control of incubator heating.
Standard search strategy of the Cochrane Neonatal Review Group. Searches were made of the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 3, 2004) and MEDLINE, 1966 to June 2004.
Randomized or quasi-randomized trials which test the effects of having the heat output of the incubator servo-controlled from body temperature compared with setting a constant incubator air temperature.
Trial methodologic quality was systematically assessed. Outcome measures included death, timing of death, cause of death, and other clinical outcomes. Categorical outcomes were analyzed using relative risk and risk difference. Meta-analysis assumed a fixed effect model.
Two eligible trials were found. In total, they included 283 babies and 112 deaths. Compared to setting a constant incubator air temperature of 31.8C, servo-control of abdominal skin temperature at 36C reduces the neonatal death rate among low birth weight infants: relative risk 0.72 (95% CI 0.54, 0.97); risk difference -12.7% (95% CI -1.6, -23.9). This effect is even greater among VLBW infants.
During at least the first week after birth, low birth weight babies should be provided with a carefully regulated thermal environment that is near the thermoneutral point. For LBW babies in incubators, this can be achieved by adjusting incubator temperature to maintain an anterior abdominal skin temperature of at least 36C, using either servo-control or frequent manual adjustment of incubator air temperature.
Survival of low birth weight infants is increased by keeping them warm. The question of "how warm" has been addressed in randomized trials which compared a higher vs lower incubator air temperature. Silverman 1958 compared the effects on survival of a warmer condition (air temperature 31.7C, range 31.1-32.2) versus a cooler condition (air temperature 28.9C, range 28.3-29.4). Survival of low birth weight infants was enhanced in the warmer condition. Jolly 1962 found similar results.
The warmer conditions tested in these trials were still below the range of "thermoneutral" air temperatures for newborn infants, which Bruck 1961 determined to be 32-34C. Moreover, the body temperatures of the infants comprising Silverman's warmer group, especially of those under 1500 g birth weight, were lower than the body temperatures which Silverman 1966 found to be associated with minimal rates of oxygen consumption. Thus, it remained possible that even warmer or more finely regulated thermal conditions might further improve survival of low birth weight infants.
It became possible to test this hypothesis with the advent of a new incubator which incorporated radiant heat lamps whose output could be servo-controlled from the infant's skin temperature.
In low birth weight infants, to determine the effects on death and other important clinical outcomes of targeting body temperature rather than air temperature as the end-point of control of incubator heating.
Randomized or quasi-randomized controlled trials which test the effects of incubator heating servo-controlled from body temperature compared with control of incubator air temperature.
Low birth weight infants, enrolled in the early neonatal period.
Servo-control of incubator heating, using the baby's body temperature as the end-point of thermal control, compared with air temperature control.
Death
Death by postnatal age period
Death by birth weight group
Other important clinical outcomes
This review used the search strategy of the Cochrane Neonatal Collaborative Review Group. Searches were made of the Oxford Database of Perinatal Trials, the Cochrane Neonatal Group's Neonatal Trials Register, the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 3, 2004) and MEDLINE. MEDLINE was searched for the period 1966 to 1997, and the search was updated in November 2001 and in June 2004, using the terms: exp temperature or body temperature or skin temperature or body temperature regulation, or oxygen consumption or exp energy intake, or incubator, infant or neonatal nursing or radiant (tw) and warm: (tw) and clinical trial and infant, premature or infant, low birth weight. Forward searches of Science Citation Index for citations of the two included trials, Buetow 1964 and Day 1964, were done for the period 1965 through 1998, and this was updated in June 2004.
The standard method was used for conducting a systematic review, as described in the Cochrane Collaboration Handbook. Each included trial was assessed for blinding of randomization, blinding of the intervention, completeness of followup, and blinding of the outcome assessment. Categorical outcomes were analyzed using relative risk and risk difference, with their 95% confidence intervals. A fixed effect model was assumed for meta-analysis.
Two trials were found which met inclusion criteria: Buetow 1964 and Day 1964. The clinical details for each trial are given in the table, Characteristics of Included Studies.
Day 1964 was a randomized trial, whereas Buetow 1964 was quasi-randomized in that allocation to the experimental group depended on the availability of the servo-controlled incubator at the time of the baby's admission. In addition, when twins were admitted to Buetow's trial, the first-born was systematically allocated to the servo-controlled incubator if it was available, and the second-born therefore entered the control group. For the trial as a whole, this resulted in imbalance between the two exposure groups for second-born twins, and they were excluded from the analysis in the report of Buetow and in this review. (For further details of trial quality and quality assessments, see table, Characteristics of Included Studies.)
Death.
Among all babies enrolled, both Buetow 1964 and Day 1964
found that death rate tended to be lower in the group servo-controlled at
36C, but in neither trial did the difference reach statistical significance.
However, the meta-analysis shows that servo-control at 36C substantially
and statistically significantly lowers death rate, relative risk 0.72 (95%
CI 0.54, 0.97); risk difference -12.7% (95% CI -1.6, -23.9).
Death (very low birth weight).
An analysis was done which was restricted to the lowest birth weight
infants (Buetow, 1000-1500 g; Day 800-1599 g). Day found the death rate to
be significantly reduced in these infants: relative risk 0.37 (95% CI 0.17,
0.83), whereas Buetow found a trend in the same direction which was not statistically
significant. The meta-analysis shows that servo-control at 36C substantially
and statistically significantly lowers death rate: relative risk 0.66 (95%
CI 0.48, 0.90); risk difference -17.8% (95% CI -5.2, -30.3).
Time of death.
Buetow 1964 used life table statistics to
analyze the effect of treatment group on time of death. It was noted that
the greatest reductions in mortality occurred between the 2nd and 14th day
of life.
Cause of death.
In both Buetow 1964 and Day 1964, the distribution of principal findings at necropsy was not different between the two exposure groups.
Other outcomes.
Buetow 1964 assessed infants' thermal stability
after the servo-controlled incubator had been discontinued at 96 hours. Compared
to the control infants, significantly more of the experimental group were
able to maintain axillary temperatures of 35C or greater during the 5th,
6th and 7th days of life.
The two trials which qualified for inclusion in this review were closely similar as regards the participants who were enrolled and in the thermal exposures which were tested. The meta-analysis of the results shows a very important reduction in death rate of low birth weight infants attributable to servo-control of abdominal skin temperature at 36C. The effect is particularly important in very low birth weight babies during the early neonatal period. The mechanism of the thermal effect in reducing death rates was not uncovered in these trials. The causes of death were not different between the two exposure groups.
No adverse biologic consequences of the close-to-thermoneutral exposure achieved by servo-control at 36C were noted in these trials. Buetow 1964 assessed temperature regulation of infants after their removal at 96 hours from the servo-controlled condition and found that they were actually better than the control infants at maintaining axillary temperatures of >35C during days 5-7. However, it should be noted that Glass 1968 and Glass 1969 showed that more prolonged (2-week) exposure of VLBW infants to a servo-controlled thermoneutral environment resulted in a reduction in their subsequent thermal stability in the face of a mild cold stress. The early and late effects of providing graded cold stimulation during the late neonatal period (ie after the time when the survival advantage of thermoneutrality has been shown) should be addressed in future trials. In particular, the possible effect of postnatal thermal stimulation of brown fat metabolism on later diet-induced thermogenesis should be investigated.
The particular incubators which were tested in these trials, which supplied supplementary heat using servo-controlled radiant heat lamps mounted on top of the incubator hood, are no longer being manufactured. Accidental dislodgement of the skin thermister used in servo-control can result in accidental overheating of the baby - a particularly serious problem when the servo-controlled heat source is a powerful radiant heater. Presently available incubators which incorporate servo-control use convective heating. Since there is no evidence that the route of heat transfer is itself critical in achieving thermoneutrality, the survival advantage shown with the particular incubators used in these trials can be expected to occur also with servo-controlled convective heating.
It should be noted that the set-point of 36C abdominal skin temperature which was chosen in these trials may be still slightly below the temperature required for strict thermoneutrality. Silverman 1966 estimated that minimal rates of oxygen consumption in VLBW infants occur when abdominal skin temperature is regulated at between 36-37C. When servo-control incubators are used in present-day care of VLBW infants in the first days of postnatal life, an abdominal skin temperature of 36.5C is recommended as the set-point to achieve thermoneutrality.
The possible benefits of "therapeutic" hypothermia in specific clinical circumstances are not ruled out by the trials reviewed here. In particular, the role of brain cooling after serious asphyxial brain insult should be investigated in properly controlled research.
Servo-control of abdominal skin temperature at 36C reduces neonatal death rate in LBW infants, compared with air temperature control at 31.8C. During at least the first week after birth, LBW babies should be provided with a carefully regulated thermal environment. An abdominal skin temperature of 36C may be slightly below the temperature required for strict thermoneutrality in LBW babies. If thermoneutrality is the goal, incubator heating for such infants should be adjusted to maintain abdominal skin temperature at approximately 36.5C, using either servo-control or frequent manual adjustment.
The early and late effects of providing graded cold stimulation during the late neonatal period (ie after the time when the survival advantage of thermoneutrality has been shown) should be addressed in future trials.
None
Study | Methods | Participants | Interventions | Outcomes | Notes | Allocation concealment |
Buetow 1964 | Quasi-random allocation (dependent on availability of servo-controlled incubator). Blinding of allocation - no. Blinding of intervention - no. Complete follow-up - no. Blinding of outcome assessment - not relevant for death (all causes), can't tell for other outcomes. | 181 very low birth weight newborn infants, birth weight range 1000-1500 g. | Experimental group: servo-control of abdominal skin temperature at 36C using radiant heat, from admission for first 96 hours; then standard incubator providing ambient T 31.1-32.2C. Control group: standard incubator providing ambient T 31.1-32.2C. | Deaths from admission to day 28. Deaths by postnatal age interval. Cause of death. Temperature regulation after removal from servo-controlled incubator. | Second-born twins excluded from analysis: 5 of 94 experimental and 18 of 87 control infants were excluded on this basis. | C |
Day 1964 | Randomized trial. Blinding of allocation - yes (for first part of study); no (for second of pair in last part of study). Blinding of intervention - no. Completeness of follow-up - can't tell. Blinding of outcome assessment - not relevant for death (all causes); can't tell for other outcomes. | 125 low birth weight newborn infants weighing 600-1999 g. Later in the study admission was limited to those weighing 750-1749 g. | Experimental group: servo-control of abdominal skin temperature at 36C using radiant heating, incubator air temperature 31.8C (range 31.1-32.2C); trial conditions maintained from admission until 1750 g or, if <1750 g BW, until BW regained. Control group: incubator air temperature 31.8C (range 31.1-32.2C). | Death until hospital discharge. Duration of life among deaths. Cause of death. Axillary temperature range. | In first part of trial, randomization was done using table of random numbers, sealed envelopes; in second part of trial, coin toss was used, with next infant assigned to other group. Infants dying within 12 hours of birth were removed from the trial; number and numbers by group are not given. | A |
Study | Reason for exclusion |
Horns 2002 | No clinical outcomes assessed. |
Perlstein 1976 | Controls identified retrospectively. |
Buetow KC, Klein SW. Effect of maintenance of "normal" skin temperature on survival of infants of low birth weight. Pediatrics 1964;34:163-70.
Day 1964 {published data only}
Day RL, Caliguiri L, Kamenski C, Ehrlich F. Body temperature and survival of premature infants. Pediatrics 1964;34:171-81.
Horns KM. Comparison of two microenvironments and nurse caregiving on thermal stability of ELBW infants. Advances in Neonatal Care 2002;2:149-60.
Perlstein 1976 {published data only}
Perlstein PH, Edwards NK, Atherton HD, Sutherland JM. Computer-assisted newborn intensive care. Pediatrics 1976;57:494-502.
* indicates the primary reference for the study
Bruck K. Temperature regulation in the newborn infant. Biology of the Neonate 1961;3:65-119.
Glass L, Silverman WA, Sinclair JC. Effect of the thermal environment on cold resistance and growth of small infants after the first week of life. Pediatrics 1968;41:1034-46.
Glass L, Silverman WA, Sinclair JC. Relationship of thermal environment and caloric intake to growth and resting metabolism in the late neonatal period. Biology of the Neonate 1969;14:324-40.
Jolly H, Molyneux P, Newell DJ. A controlled study of the effect of temperature on premature babies. Journal of Pediatrics 1962;60:889-94.
Silverman WA, Fertig JW, Berger AP. The influence of the thermal environment on the survival of newly born premature infants. Pediatrics 1958;22:876-86.
Silverman WA, Sinclair JC, Agate FJ. The oxygen cost of minor changes in heat balance of small newborn infants. Acta Paediatrica Scandinavica 1966;55:294-300.
Sinclair JC. Management of the thermal environment. In: Sinclair JC, Bracken MB, editor(s). Effective Care of the Newborn Infant. Oxford: Oxford University Press, 1992:40-58.
Sinclair JC. Servo-control for maintaining abdominal skin temperature at 36C in low birthweight infants (Cochrane Review). In: The Cochrane Library, Issue 1, 2000. Oxford: Update Software.
Sinclair JC. Servo-control for maintaining abdominal skin temperature at 36C in low birth weight infants (Cochrane Review). In: The Cochrane Library, Issue 1, 2002. Oxford: Update Software.
01.01 Death
01.02 Death (very low birth weight)
Comparison or outcome | Studies | Participants | Statistical method | Effect size |
---|---|---|---|---|
01 Servo-control of abdominal skin temperature vs air control | ||||
01 Death | 2 | 283 | RR (fixed), 95% CI | 0.72 [0.54, 0.97] |
02 Death (very low birth weight) | 2 | 228 | RR (fixed), 95% CI | 0.66 [0.48, 0.90] |
This review is published as a Cochrane review in The
Cochrane Library, Issue 4, 2004 (see http://www.thecochranelibrary.com/ for information).
Cochrane reviews are regularly updated as new evidence emerges and in response
to comments and criticisms, and The Cochrane Library should be consulted
for the most recent version of the Review. |