This review assessed the wrapping of premature newborn infants in transparent wrap to prevent the loss of body heat. The authors concluded that skin wrap prevented a significant amount of heat loss, although the evidence so far did not show a significant reduction in mortality. The review presented insufficient information to be confident that the results from the included studies are reliable.

The main objective was to determine the effectiveness of occlusive skin wrapping to prevent heat loss in premature infants.

MEDLINE (1996 to 2003), EMBASE (1980 to 2003), Web of Science and EBM Reviews (Issue 1, 2003) were searched. The search terms were not reported, although the search strategies were said to be available on request. Study authors and topic experts were contacted and reference lists in relevant articles examined for additional studies. No limitations were placed on publication status.

Study designs of evaluations included in the review

Primary studies of any design appeared to be eligible for inclusion.

Specific interventions included in the review

Studies of occlusive transparent skin wrap applied immediately after delivery compared with standard care were eligible for inclusion. Wrapping in conjunction with a radiant heater was implicit. The types of wrap used in the included studies were polyethylene, polyurethane, polyvinyl chloride and thin plastic wrap. The mean transfer time from delivery room to the neonatal intensive care unit (ICU), where reported, ranged from 7 (standard deviation, SD=10) to 16.4 (SD=4) minutes in the included randomised trials.

Participants included in the review

Studies in premature neonates born at less than 36 weeks' gestation were eligible for inclusion. Gestation duration in the included studies varied from less than 33 weeks to less than 28 weeks; one study included participants on the basis of weight less than 1,000 g.

Outcomes assessed in the review

The primary outcome was temperature at admission to the neonatal ICU. The secondary outcomes included death, incidence of hypothermia, respiratory and neurological problems, and length of hospital stay. Complications and adverse events were also assessed. Rectal temperature was used in most of the included studies, otherwise it was axillary or not specified. The definition of hypothermia was available for three of the included studies: defined as less than 35.5 degrees C or less than 36.4 degrees C. Respiratory outcomes reported in the review were the fraction of inspired oxygen (FiO2) and duration of oxygen therapy. The main neurological problem reported was major brain pathology. The main adverse event reported was hyperthermia.

How were decisions on the relevance of primary studies made?

Two reviewers applied the inclusion criteria independently.

Randomised controlled trials (RCTs) were assigned a quality score according to the Jadad scale. Given that double-blinding was not feasible, only the criteria relating to the description of randomisation and withdrawals or drop-outs were assessed. The maximum possible score was 3. Allocation concealment was assessed separately. Two reviewers assessed study quality independently. The validity of the observational studies was not assessed.

A standard form was used to extract the data. One reviewer extracted the data and a second reviewer checked the accuracy of it. Temperature data extracted were the mean and SD or median and range. The number of deaths, incidence of hypothermia and the number of participants in the treatment and control groups in each study were extracted. Other outcome data were extracted as reported. Missing data were sought from investigators, as were individual patient data (IPD) on admission temperature, treatment, gestational age, delivery room temperature, maternal temperature, transfer time from delivery room to ICU and route of temperature measurement.

How were the studies combined?

Meta-analysis was used to obtain pooled estimates of the weighted mean difference (WMD) in admission temperature, the relative risk (RR) of death and the incidence of hypothermia in wrapped compared with non-wrapped infants. The World Health Organization definition of hypothermia (less than 36.5 degrees C) was assumed for two of the RCTs for which IPD were obtained, in order to include them in the meta-analysis of the incidence of hypothermia. Pooled estimates were reported with 95% confidence intervals (CIs), calculated using a random-effects model. Pooled RR values were expressed as the number-needed-to-treat. Pooled estimates of the incidence of hyperthermia were expressed as the risk difference, indicating that in some studies the outcome did not occur in either the treatment or control group.

IPD were used to perform a multiple linear regression analysis of admission temperature. The explanatory variables included in the model were treatment (wrap, no wrap), gestational age (23 to 33 weeks), delivery room temperature (18 to 31.1 degrees C) and maternal temperature (35.6 to 39.4 degrees C). The model was adjusted for study design (RCT, historical controlled trial) as a fixed-effect.

How were differences between studies investigated?

In the meta-analysis, RCTs were grouped separately from studies with historical controls and the chi-squared and I-squared statistics were used to assess statistical heterogeneity within the subgroups and across all studies.

Eight studies were included. Three were RCTs and five were studies that used historical controls. The total number of participants appeared to be 1,007 (200 in the RCTs and 807 in the non-randomised studies).

The RCTs had a Jadad quality score of 1 or 2 out of a possible 3. The adequacy of allocation concealment was not reported.

A meta-analysis showed that admission temperature was statistically significantly higher among wrapped infants in 3 RCTs (200 participants); the WMD was 0.63 degrees C (95% CI: 0.38, 0.87, p<0.00001). There was no statistically significant heterogeneity between the studies. A meta-analysis of 5 non-randomised studies (807 participants) showed a larger difference (WMD 0.96, 95% CI: 0.66, 1.27, p<0.00001) with significant heterogeneity between the studies. A regression analysis using data from the RCTs and 2 non-randomised studies showed statistically significant associations between higher admission temperatures and wrapped infants, older infants, warmer mothers and warmer delivery rooms. A significant negative interaction between treatment and gestational age suggested that wrapping was less effective in older infants.

Mortality: a meta-analysis showed no significant difference in mortality in the 3 RCTs or 4 non-randomised studies that reported mortality. Heterogeneity was not statistically significant.

Incidence of hypothermia: a meta-analysis of 3 RCTs showed a statistically significant lower risk of hypothermia in wrapped infants (RR 0.51, 95% CI: 0.37, 0.71), which gave a number-needed-to-treat of 2.9 (95% CI: 2.3, 4.9). Heterogeneity was not reported. The non-randomised studies were highly heterogeneous and, overall, showed no significant difference.

Respiratory problems: the results of the included studies were not reported in full. One study that compared 11 wrapped infants to historical controls reported that the highest FiO2 was significantly lower amongst wrapped infants.

Neurological problems: in one RCT the occurrence of major brain injury was similar amongst wrapped and non-wrapped infants.

Length of stay: in one RCT the length of hospital stay was not significantly different between wrapped and non-wrapped infants.

Adverse events: a meta-analysis showed no significant difference in the risk difference for the incidence of hyperthermia in 3 RCTs or in 2 studies with historical controls. In one RCT none of the participants experienced infection, skin maceration or interference with resuscitation.

Occlusive skin wrap prevented heat loss in premature infants.

The review addressed a clear question. Several sources were searched for relevant studies although it was not possible to assess the adequacy of the search strategy from the report. The review was not restricted to published studies, but it was unclear if non-English language studies were eligible for inclusion. Steps were taken to minimise reviewer bias in the study selection process and the method used to minimise errors in the data extraction was reasonable. The assessment and reporting of the quality of the included studies was inadequate. The tool used to assess the quality of the RCTs was unhelpful, and the failure to report on allocation concealment means that the potential for selection bias (which can lead to overestimation of treatment effects) in the individual studies is unknown. No attempt was made to assess the potential for bias in the non-randomised studies, so the reliability of their results is unknown.

The characteristics of the individual included studies were not reported in sufficient detail to enable an independent judgment of how appropriate it was to pool the data. Although statistical heterogeneity was assessed, there was no attempt to explain it and the lack of study details precluded independent assessment. Overall, the review left too many unanswered questions about the reliability of the results from the individual studies to be confident that the conclusion is reliable.

Practice: The authors indicated that recommendations on the use of wrap should be included in guidelines on neonatal resuscitation.

Research: The authors stated that an RCT is underway to determine the effect of occlusive wrap on mortality in premature infants; secondary outcomes include temperature on introduction into the ICU and neurodevelopmental follow-up. See Web Address at end of abstract (accessed 01/05/2007).

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.