Meta-analysis of Inhaled Nitric Oxide for Hypoxaemic Respiratory Failure in Preterm Infants
YU Zhang-bin,HAN Shu-ping,GUO Xi-rong
DOI: https://doi.org/10.3969/j.issn.1673-5501.2007.05.002
2007-01-01
Abstract:Objective Inhaled nitric oxide(iNO) has been used in preterm infants with hypoxic respiratory failure (HRF). However, there is controversy about its therapeutic effect. In this article, the studies of iNO for HRF were summarized and the effect of iNO for HRF was assessed in preterm infants using meta-analysis method. Methods Medical electronic databases and other sources were searched without language restriction. According to inclusion and exclusion cretria, articles were evaluated (Shekelle PG, et al). RCTs were assessed according to the Juni assessment. Software Rev Man 4.2 was used to assess control trials, and other studies were reviewed and descrbied. Results Nine RCTs with 6 blind trials and 3 non-blind trials were selected from 883 articles. Publication bias was not found through funnel polt analysis. There were 3 RCTs that compared the oxygen use and the duration of ventilation, the length of hospitalization, including 259 patients in iNO and 259 in the control group. Compared to the control group, iNO could reduce the oxygen use (WMD=-10.24,95%CI:-19.62,-0.86,P0.05); There was no difference in the duration of ventilation between iNO and the control group (WMD=-4.87,95%CI:-10.68,0.94,P0.05); Compared to the control group, iNO could reduce the length of hospitalization (WMD=-8.31,95%CI:-16.23,-0.38,P0.05). There were 6 RCTs that compared the death before discharge, including 1084 patients in iNO and 1096 in the control group. There was no difference in the deaths before discharge between iNO and the control group (OR=0.95,95%CI:0.77,1.17,P0.05). There were 5 RCTs that compared the incidence of BPD among survivors at 36 weeks, including 618 patients in iNO and 621 in the control group. Compared to the control group, iNO could reduce the incidence of BPD (OR=0.75,95%CI:0.58,0.95,P0.05). There was no difference in the incidence of intraventricular hemorrhage (all grades) between iNO and the control group (OR=0.79,95%CI:0.62,1.00,P=0.05). There were 4 RCTs that compared the incidence of intraventricular hemorrhage (grade 3 or 4), including 767 patients in iNO and 784 in the control group. Compared to the control group, iNO could reduce the incidence of intraventricular hemorrhage (grade 3 or 4) (OR=0.84,95%CI:0.61,1.16,P0.05). There were 5 RCTs that compared the intraventricular hemorrhage (grade 3 or 4) or periventicular leukomacia, including 1021 patients in iNO and 1012 in the control. There was no difference in the incidence of intraventricular hemorrhage (grade 3 or 4) or periventicular leukomacia between iNO and the control group (OR=0.95,95%CI:0.68,1.22,P0.05). There were 2 RCTs that compared the incidence of threshold ROP, including 608 patients in iNO and 605 in the control group. There was no difference in the incidence of threshold ROP between iNO and the control group (OR=0.98,95% CI:0.72,1.34,P0.05). There were 4 RCTs that compared the incidence of ROP, including 474 patients in iNO and 463 in the control group. There was no difference in the incidence of ROP between iNO and the control group (OR=1.08,95%CI:0.75,1.57,P0.05). There was 1 RCT that compared the incidence of ROP requiring surgery, including 294 patients in iNO and 288 in the control group. There was no differencein the incidence of ROP requiring surgery between iNO and the control group(OR=1.05,95%CI:0.72,1.53, P0.05). There were 3 RCTs that compared the incidence of cerebral palsy, including 168 patients in iNO and 184 in the control group. There was no difference in the incidence of cerebral palsy between iNO and the control group (OR=1.08,95%CI:0.75,1.57,P0.05). There were 2 RCTs that compared the incidence of Bayley MDI2SD, including 156 patients in iNO and 110 in the control group. There was no difference in the incidence of Bayley MDI2SD between iNO and the control group (OR=0.84,95%CI:0.50,1.42,P0.05). There were 2 RCTs that compared the incidence of Bayley PDI2SD, including 155 patients in iNO and 167 in the control group. There was no difference in the incidence of Bayley PDI2SD between iNO and the routine therapy group (OR=0.94,95%CI:0.57,1.58,P0.05). Conclusions The currently published evidence from randomised trials does not support the routine use of iNO in preterm infants with HRF. Further multicentre and large-scale RCTs are still needed to be performed to evaluate the potential benefit of iNO as routine therapy in preterm infants with HRF, and these trials need to be designed to evaluate not only the short term effects, but should be powered to evaluate neurondevelopmental outcome at a minimum of two years of age.