Results of the search

The literature search run in December 2019 identified 350 references (Figure 1). After screening, we included in the review four RCTs enrolling 317 infants (Akbarian Rad 2018; Eghbalian 2018; Sardar 2020; Stroustrup 2012). We excluded two studies (CTRI/2019/04/018661; Dehdashtian 2014), and one study is awaiting classification because the abstract and the full text were not available (Stroustrup 2010). We identified no ongoing trials.

Included studies

The four studies included in the review are described below (see also Table 1 and Characteristics of included studies).

Akbarian Rad 2018 enrolled late preterm, term, and post‐term neonates with TTN. The diagnosis of TTN was based on the presence of tachypnea (respiratory rate > 60 per minute) and at least one radiological sign suggestive of the diagnosis. If chest X‐ray was normal, the diagnosis of TTN was made if the Silverman Anderson score for respiratory distress was 4 or greater and the neonate was hospitalized within six hours of birth (Silverman 1956). The level of respiratory support provided by oxygen hood at admission did not differ significantly between the two groups (P = 0.147); however, the number of infants on oxygen hood in each group is not provided in the article.

Neonates in the restricted fluid group received 50 mL/kg if born at term or post‐term gestation, and 65 mL/kg if born at late preterm gestation. Neonates in the standard fluid group received 65 mL/kg if born at term or post‐term gestation, and 80 mL/kg if born at late preterm gestation. Fluid intake was increased by 20 mL/kg every subsequent day to reach a maximum of 150 mL/kg in term and post‐term neonates, and 180 mL/kg in late preterm neonates. An additional 10% of fluid was given if the baby was under a radiant warmer or was receiving phototherapy. Enteral feeds were added when the baby was "stable," and intravenous fluid intake was adjusted to target the planned total fluid intake. Mode of delivery, gestation category, and gender distributions were comparable in the two study groups. No information is provided about other baseline characteristics nor actual fluid intake in the two study groups. No information is provided about criteria for starting or stopping respiratory support. In this study, the primary outcome has not been defined or used for the sample size calculation. Outcomes reported include respiratory rate and distress score at 24 hours, duration of hospitalization (in days), duration of respiratory support (in hours), and level of respiratory support needed based on the need for the fraction of inspired oxygen (FiO₂).

Eghbalian 2018 enrolled neonates born at 37 to 41 weeks' gestation who were hospitalized with TTN within three days of birth. The diagnosis of TTN was based on the following criteria: tachypnea (respiratory rate > 60 per minute) continuing for at least 12 hours, prominent central pulmonary vessels, thickening of the interlobar fissure on chest X‐ray, and respiratory alkalosis in arterial blood gas analysis. Neonates with congenital malformations, other causes of respiratory distress, polycythemia, electrolyte imbalance suggestive of dehydration, or acute kidney injury were excluded. No information on the presence or absence of any respiratory support at the time of study entry is available.

Neonates assigned to the restricted fluid group received 40, 60, and 80 mL/kg of intravenous fluid on the first, second, and third days of birth, respectively. Neonates assigned to the standard fluid group received 60, 80, and 100 mL/kg of intravenous fluid on the first, second, and third days of birth, respectively. The intravenous fluid used in both groups was 10% dextrose to which 3 mEq/kg/d of sodium and 2 mEq/kg/d of potassium had been added from the second day of birth. An additional 20 mL/kg/d was given if the baby was under a radiant warmer. Gestation category and birth weight were comparable in the two study groups. Study authors indicate that the proportion of male infants was significantly higher in the restricted fluid group. However, the numbers for gender distribution in the baseline characteristics table do not sum to the total neonates enrolled in the study, indicating missing information or typographical error. No information about other baseline characteristics and actual fluid intake in the two study groups is provided. No information about criteria for starting or stopping respiratory support is provided. The primary outcome of the study was duration of hospitalization. Other outcomes included duration of oxygen therapy and need for and duration of different types of respiratory support.

Sardar 2020 enrolled neonates born at 34 to 41 weeks' gestation who weighed more than 1500 grams, had TTN, and needed CPAP as respiratory support within six hours of birth. The diagnosis was made if respiratory distress started within six hours of birth, with the chest‐X ray suggestive of at least one radiological sign of TTN. Neonates with congenital malformations, air leaks, hemodynamic instability, and alternative causes of respiratory distress were excluded. Infants with TTN for whom oxygen therapy has failed who were on nasal CPAP support within six hours of birth were enrolled; however, baseline distribution of the level of CPAP in both groups is missing from the article.

Neonates assigned to the restricted fluid group received 40, 60, and 80 mL/kg of intravenous fluid on the first, second, and third days of birth, respectively. Neonates assigned to the standard fluid group received 60, 80, and 100 mL/kg of intravenous fluid on the first, second, and third days of birth, respectively. The intravenous fluid used in both groups was 10% dextrose to which electrolytes were added from 48 hours of birth. Fluid status of enrolled neonates was monitored by daily measurement of serum sodium, body weight, urine output, and urine specific gravity. The planned daily increment in fluid intake was made only if these measurements indicated no dehydration or fluid overload. Neonates were excluded (four from each study group) from the study after enrollment and were not included in the analysis if they exhibited dehydration, fluid overload, or hypoglycemia. All important baseline variables were comparable in the two groups. However, neonates in the restricted fluid group were significantly less likely to have a low (≤ 7) Apgar score at five minutes of age. No information is provided about actual fluid intake in the two study groups. Information about criteria for starting or stopping respiratory support is provided. The primary outcome of the study was the duration of CPAP support. Other outcomes included incidence of CPAP failure, duration of the oxygen requirement, and incidence of common neonatal morbidities.

Stroustrup 2012 enrolled neonates born at 34 to 41 weeks' gestation who developed TTN within 12 hours of birth. The diagnosis of TTN was made in the presence of respiratory distress (flaring, grunting, and accessory muscle use with or without hypoxia) and chest X‐ray findings consistent with the diagnosis. Neonates with major congenital malformations or alternate respiratory diagnosis and those undergoing workup for sepsis and with air leaks were excluded. Infants whose oxygen saturation was ≤ 95% and/or who had hypercapnia were started on respiratory support (nasal CPAP, high‐flow nasal cannula therapy, or nasal cannula) before or at the time of enrollment, but the exact number of infants and their respective distribution of respiratory support are not known.

Neonates in the restricted fluid group received 40 mL/kg if born at term gestation and 60 mL/kg if born at preterm gestation. Neonates in the standard fluid group received 60 mL/kg if born at term gestation and 80 mL/kg if born at preterm gestation. Total fluid intake was calculated as a combination of intravenous and enteral fluid intake. Fluid intake was increased by 20 mL/kg every subsequent day to reach a maximum of 150 mL/kg or ad libitum feeding. Important baseline characteristics were comparable in the two study groups. However, a significantly higher proportion of neonates in the standard fluid intake group received antenatal steroids. In the subgroup of neonates with severe TTN (defined as the need for respiratory support for longer than 48 hours), mean gestation was significantly lower. No information is provided about the type of fluid administered in the two study groups. However, information about actual fluid intake in the two study groups during the first 72 hours of birth is provided. Criteria for starting respiratory support are provided. The mode of respiratory support and decisions about weaning and discontinuation of respiratory support were determined by the treating team. The primary outcome of the study was duration of respiratory support. Other study outcomes included time from birth to first enteral feed, duration of NICU stay, total cost of hospitalization, and component costs including physician, direct, and indirect costs.

Excluded studies

We excluded two studies: one because of the wrong indication (CTRI/2019/04/018661), and the other because of before‐after study design (Dehdashtian 2014).

Allocation

Two of the four studies included in the review had high risk of selection bias (Akbarian Rad 2018; Stroustrup 2012). Neonates were alternately assigned to receive restricted or standard fluid therapy in both these studies (Akbarian Rad 2018; Stroustrup 2012). The third study used a computer‐generated random number sequence for treatment assignment and serially numbered sealed opaque envelopes for allocation concealment (Sardar 2020). The remaining study used a table with block randomization for generating random number sequence (Eghbalian 2018); however, the method used to ensure allocation concealment is not reported.

Blinding

Two of the four studies included in the review had high risk of performance and detection bias (Sardar 2020; Stroustrup 2012). In both these studies, the healthcare teams providing clinical care and assessing the outcome were not blinded. In Akbarian Rad 2018, the healthcare team that provided clinical care and made decisions about starting, weaning, or stopping respiratory support was not blinded; therefore we judged this study to have high risk of performance bias. However, the nurses who recorded outcomes were blinded to the treatment assignment, hence the study was judged as having low risk of bias. In Eghbalian 2018, the decision to initiate or terminate respiratory support was made by healthcare workers who were blinded to the treatment assignment. Therefore, this study was judged to have low risk of performance bias. No information about the way the outcome was assessed is available; therefore we judged this study as having unclear risk of detection bias.

Incomplete outcome data

Two studies did not provide information about the flow of trial participants (Akbarian Rad 2018; Eghbalian 2018). From information available in the manuscript, it can be deduced that all enrolled participants had been included in the analysis. Therefore, these two studies were judged to be at low risk of attrition bias. In Sardar 2020 and Stroustrup 2012, participants were excluded after enrollment and randomization. Sardar 2020 excluded three and four participants from intervention and control groups, respectively, post randomization and did not include them in the analysis. Reasons for exclusion included development of complications (e.g. dehydration, hypoglycemia) in both groups, which may be related to the study intervention. However, reasons for exclusion are quite explicit, almost equal in both groups, and attrition is less than 10% among total enrolled participants; therefore this study is assessed to be at low risk of attrition bias. Stroustrup 2012 excluded participants from the analysis if an alternative cause of respiratory distress was found after enrollment and randomization. A total of two participants from the control group and one from the intervention group were excluded. Although this study practiced neither allocation concealment nor blinding of investigators, the decision of post‐randomization exclusion may be influenced by knowledge of the treatment assignment for both the participant excluded and the next participant in the randomization sequence. However, the reasons for exclusion are quite explicit, and attrition is less than 10% among total enrolled participants; therefore this study is assessed to be at low risk of attrition bias.

Selective reporting

All four studies included in this review report outcomes related to the initiation and duration of various modes of respiratory support (Akbarian Rad 2018; Eghbalian 2018; Sardar 2020; Stroustrup 2012). Three of the included studies were registered at one of the available clinical trial registers (Akbarian Rad 2018; Sardar 2020; Stroustrup 2012). We were able to access the registered protocol for these three registered studies (Akbarian Rad 2018; Sardar 2020; Stroustrup 2012). All four studies have been judged to be at low risk of reporting bias.

Other potential sources of bias

We have not identified any other potential source of bias in any of the four studies included in the review (Akbarian Rad 2018Eghbalian 2018Sardar 2020; Stroustrup 2012). No external funding has been declared by study authors in the three studies (Akbarian Rad 2018; Eghbalian 2018; Sardar 2020). Stroustrup 2012 was funded by a grant from the National Institutes of Health, in the USA.

Primary outcome

Duration of supplemental oxygen therapy

Three trials reported this outcome (Akbarian Rad 2018; Eghbalian 2018; Sardar 2020).

Akbarian Rad 2018 reported only a measure of central tendency (no information whether mean or median) and not of dispersion (i.e. standard deviation) for the duration of oxygen therapy (21.35 hours versus 31 hours; P = 0.048). Therefore, the data from this study could not be used to calculate the pooled effect size. Pooled data from the other two studies ‐ Eghbalian 2018 and Sardar 2020 ‐ show no significant difference in duration of oxygen therapy between the two groups (mean difference [MD] ‐12.95 hours, 95% confidence interval [CI] ‐32.82 to 6.92 hours) (Analysis 1.1; Figure 4). We judged the certainty of evidence for this outcome to be very low due to risks of selection and detection bias. We downgraded the evidence due to inconsistent results (heterogeneity: Chi² = 43.34, df = 1 [P = 0.20]; I² = 98%, random‐effects model) and imprecision due to suboptimal information size, as only two small trials contributed to the pooled effect size. This high heterogeneity (I² = 98%) is the result of grossly different mean values for the duration of supplemental oxygen therapy in the two studies (three to six hours in Eghbalian 2018 versus 29.5 to 40 hours in Sardar 2020). Moreover, for Eghbalian 2018, we have converted units of the duration of oxygen therapy from days to hours, so as to make data uniform for the meta‐analysis.

Figure 4

---

---

Figure 4. Restricted versus standard fluid management, Outcome 1.1: Duration of supplemental oxygen therapy

Secondary outcomes

Need for invasive ventilation

Three trials reported this outcome (Akbarian Rad 2018; Eghbalian 2018; Sardar 2020). The pooled incidence was comparable in two groups (RR 0.73, 95% CI 0.24 to 2.23; RD ‐0.02; 95% CI: ‐0.07 to 0.04) (Analysis 1.6; Figure 5). We judged the certainty of evidence for this outcome to be very low due to risk of detection bias and upper and lower bounds of the 95% CI of the pooled risk ratio reaching points of clinically significant reduction or increase in the outcome.  

Figure 5

---

---

Figure 5. Restricted versus standard fluid management, Outcome 1.6: Need for invasive ventilation

Subgroup analysis

We were unable to conduct any of the planned subgroup analyses as the review included only four trials with limited information.

Three out of four studies included both term and late preterm infants (Akbarian Rad 2018; Sardar 2020; Stroustrup 2012). However, except for one study (Sardar 2020), others have not separately reported data on late preterm infants. Therefore, the subgroup analysis based on gestational age could not be performed.

All infants were randomized and started treatment within 12 hours after birth. Therefore, the subgroup analysis based on time of the start of treatment (≤ 24 hours versus > 24 hours) could not be done.

Infants were receiving different types of respiratory support at the time of enrollment and its exact distribution is not reported. Sardar 2020 enrolled only those infants for whom oxygen therapy had failed who were on nasal CPAP support. Thus, the subgroup analysis based on type of respiratory support was not possible.

Infants in the fluid‐restricted group received 15 to 20 mL/kg/d less fluid than the control group for varying durations across all four studies. This restriction varied from ~ 9% (15 mL/kg/d/ 70 mL/kg/d) to 33% (20 mL/kg/d/60 mL/kg/d) on different days of life, depending on the absolute volume of total fluid required. Hence, no subgroup analysis based on the degree of fluid restriction (≤ 20% versus > 20%) could be performed.

Sensitivity analysis

We could not perform sensitivity analysis due to the limited number of included trials. None of the included trials was free of bias to be included in the sensitivity analysis under the category "adequate methodology" as defined in the review protocol.