Abstract
BACKGROUND:
Wide variation in the care practices and survival rates of neonates born at peri-viable gestational ages of 22+0 – 24+6 weeks. This study elucidates the postnatal risk factors for morbidity/mortality, contrasts the care practices and short-term outcomes of this vulnerable group of preterm neonates from a single center with others.
METHODS:
Retrospective study of neonates born at 22+0 –24+6 weeks in a level 3 neonatal intensive care unit in UK, over a period of 4 years (2016–2019).
RESULTS:
94 neonates given active care were studied. Survival until discharge was 51.1%(22–23 wks –44%, 24 wks –59.1%) and survival with no major brain injury (MBI) [grade III/IV IVH, cystic periventricular leukomalacia] was 38.3%(22–23 wks –32%, 24 wks –45.4%). Of those who survived until discharge, 75%had no MBI (22–23 wks –72.7%, 24 wks –76.9%). Neonates requiring significant respiratory support within first 72 hours as well as needing rescue high frequency ventilation had significantly high risk of mortality or MBI [aOR –7.17 (2.24–25.79), p = 0.00; 4.76 (1.43–20.00), p = 0.01].
CONCLUSIONS:
Survival rate differed from other centres. MBI was low amongst survivors. Severe respiratory disease in the initial days was associated with a higher risk of death or MBI.
Keywords
Introduction
The survival of extremely preterm neonates born at gestational ages 22–24 weeks has improved considerably over the last three decades with significant variability in survival between different centers and countries [1–6]. Recent data from United Kingdom indicates that the survival of the neonates born at 22 weeks and who were offered active support has improved considerably over the past decade [7]. Watkins et al. (Iowa center, USA) reported encouraging results especially in relation to their long-term survival [8]. Though many networks have published their short- and long-term survival, very few have described the care practices and the course of stay of these neonates in detail [1]. Improving survival in this sub-group of neonates mandates improved evidence-based practices for which there is paucity in literature. We present our single center data together with our clinical practices and compare our results to published data.
Materials/subjects and methods
Study place/subjects/time period
This is a retrospective observational study evaluating neonates of gestational ages 22+0 to 24+6 weeks born at the John Radcliffe Hospital (JRH), Oxford, United Kingdom over a 4-year period from 1st January 2016–31st December 2019. Neonates who died in the delivery room were excluded from the analysis. The gestational age was determined by the best obstetric estimate based on the first trimester scan and the last menstrual period. The management protocol is provided in the supplement file.
Outcomes
The primary outcome of the study was survival to discharge with no major brain injury (MBI) [Grade III/IV intraventricular hemorrhage (IVH) [9], cystic periventricular leukomalacia (PVL)]. The sec-ondary outcomes were survival to discharge, respiratory outcomes non-invasive respiratory support (NRS) failure (within 72 hours and until disch-arge), duration of NRS, requirement and duration of invasive mechanical ventilation (IMV) (conven-tional and high frequency), bronchopulmonary dysplasia (BPD) [10], air leak, persistent pulmonary hypertension of the newborn (PPHN), pulmonary hemorrhage, home oxygen therapy, patent ductus arteriosus (PDA) requiring medical or surgical trea-tment, inhaled nitric oxide (iNO) use, sepsis related outcomes early onset (< 72 hours) and late onset sepsis (> 72 hours)/septic shock requiring inotr-opes/meningitis, ventilator associated pneumonia (VAP) [11], antibiotics use rate (AUR), nutritional outcomes necrotising enterocolitis (NEC) stage II or more [12], surgical NEC, osteopenia of prematurity (defined as serum phosphate < 4.1 mg/dl with alkaline phosphates > 800 IU/L and/or radiographic changes), extra uterine growth retardation (EUGR) at 36 weeks post menstrual age (PMA) [13], transfusion practices, retinopathy of prematurity (ROP) [14], ROP requiring treatment and duration of stay in the hospital.
Statistical analyses
Continuous variables were expressed as mean with standard deviation (SD) as well as median with int-erquartile range (IQR). P-value was calculated using t-test/ANOVA for normally distributed parameters and Mann Whitney U-test/Kruskal Wallis for non-normally distributed parameters. Categorical data was compared using chi-square test and Fischer’s exact test. P-value of < 0.05 was considered significant. Multivariate logistic regression was used for calculating the adjusted odd’s ratio with 95%confidence interval [aOR (95%CI)] for explanatory analysis as well as for prediction models. For aOR calculation, the baseline characteristics such as gestational age, small for gestational age (SGA) status, sex, receipt of antenatal steroids and maternal chorioamnionitis were adjusted. The final model for prediction was selected based on comparisons using the partial likelihood ratios test. Time to event data was analyzed using Kaplan Meier plots as well by Cox proportional hazards regression analysis and expressed as adjusted hazard’s ratio (95%confidence interval) [aHR (95%CI)]. Proportional hazards assumption was tested using Schoenfeld residuals. Statistical analyses was performed using R software (R version –3.6.2).
Ethics
This work was reviewed and authorized by the Joint Research Office study classification group of the institution as per the Department of Health’s UK Policy Framework for Health and Social Care.
Results
A total of 94 neonates born at 22+0 –24+6 gestational age and admitted to NICU for active care were included in the study [22 wks (n = 4), 23 wks (n = 46), 24 wks (n = 44)]. The baseline characteristics of the enrolled neonates are summarized in Table 1.
Demographic details of the enrolled subjects
Demographic details of the enrolled subjects
*SGA –Small for gestational age defined as birth weight for gestational age less than 10th centile, pPROM –preterm premature rupture of membranes. #At least 1 dose of either betamethasone or dexamethasone received > / = 8 hours prior to delivery. +At least 2 doses of betamethasone or 4 doses of dexamethasone received > / = 24 hours but within 1 week before delivery.
The survival until discharge rate was 51.1%overall (22–23 wks –44%, 24 weeks –59.1%). 75%of the survivors were free from any apparent MBI (22-23 wks –72.7%, 24 wks –76.9%). 38.3%of the neonates survived with no MBI (22-23 wks –32%, 24 wks –45.4%). Neonates born at 22 weeks were given active resuscitation in this centre since 2019 and hence only four neonates of this gestation were included. Multiple births, requirement of inhaled nitric oxide, not receiving postnatal dexamethasone, air leak, IVH > / = grade 3 were significantly associated with mortality [aHR(95%CI) –2.13(1.09–4.17), 3.57(1.75–7.14), 9.10(3.44–24.06), 2.32(1.03–5.26), 2.27(1.15–4.54), respectively]. Those neonates who were not ready to be extubated due to high respiratory settings from IMV in the first 72 hours, not received postnatal dexamethasone and those who required rescue high frequency oscillation ventilation (HFOV) had significantly high risk of death or MBI [aOR –7.17 (2.24–25.79), p = 0.00; 5.26 (1.61–20.00), p = 0.00; 4.76 (1.43–20.00), p = 0.01 respectively].
Delivery room practices
Eighty six percent of the neonates were intubated in the delivery room, 75.5%of the neonates received rescue delivery room surfactant, one neonate received chest compression and 7.4%had severe perinatal asphyxia defined based on umbilical cord arterial blood gas pH (< 7.00) and base deficit (> 16). Only 21.1%of the neonates could be stabilized on non-invasive respiratory support (NRS).
Respiratory care
Surfactant use
Of the 88 (94.6%) neonates who received surfactant, 45 (48.4%, n = 94) received it in the NICU. Only 10.1%received surfactant through LISA (Less Invasive Surfactant Administration). The median (IQR) age of surfactant administration was 15 minutes (11–20) and the median number of doses was 1 (1–2) with no differences between the neonates who survived and who did not (p = 0.06).
Non-invasive respiratory support
A total of 35 infants (37.2%) were initiated on a NRS modality within the first 72 hours, of which 21 (22.5%) received a NRS modality (HHHFNC = 20.4%, BiPAP = 2.1%) within the first 24 hours. The incidence of NRS failure within the first 72 hours was 74.3%and it occurred at a mean (SD) of 19.7 (±16.1) hours. There was a statistically significant difference in the NRS failure rate between survivors and non survivors (70.4%vs 87.5%, p = < 0.01). Cox regression analysis after adjusting for the baseline characteristics revealed gestational age to be the only significant factor predicting NRS failure [aHR (95%CI) for 23 wks vs 22 wks –0.08 (0.00–0.91); 24 wks vs 22 wks –0.03 (0.00–0.43)]. The duration of NRS in the first 72 hours and during the hospital stay is given in Table 2.
Duration of respiratory support and respiratory morbidities in survivors versus non-survivors
Duration of respiratory support and respiratory morbidities in survivors versus non-survivors
*p-values in bold are significant. + Conventional mechanical ventilation. # High frequency oscillation ventilation. ∧ ANOVA used for multiple mean comparisons.
There was a statistically significant difference in the duration of any IMV (conventional and HFOV) between the different gestational ages (p = 0.03) (Table 2). The requirement of rescue HFOV was significantly higher amongst non survivors (56.3%vs 82.6%, p = < 0.01). Respiratory morbidities and drugs for prevention/treatment of BPD are shown in Table 2.
Cardiovascular morbidity
Seventy percent of the studied neonates required either medical or surgical therapy for PDA. Intravenous/ oral paracetamol was the first line drug used in the majority of the neonates (62.1%). 16.7%of the total neonates required ligation of the PDA after failed medical therapy. The mean (SD) age of ligation was 46.7 (± 24.5) days. The details of iNO use is given in Table 3.
Details of non-respiratory care practices and morbidities in survivors versus non-survivors
Details of non-respiratory care practices and morbidities in survivors versus non-survivors
*p-values in bold are significant. ∧p-values are given for outcomes where clinically significant comparisons are possible.
About 42.5%of the neonates had blood culture proven sepsis. Of these only 4.2%were early onset (< 72 hours). E.coli was the predominant organism in early onset sepsis (EONS) and Coagulase negative staphylococcus (CONS) was the predominant organism amongst the neonates with late onset neonatal sepsis (LONS). Among the survivors, the incidence of LONS was 50%. The mean antibiotic use rate (AUR) was 28.3%(survivors –22.8%; non survivors –71.7%). The other aspects related to sepsis are given in Table 3.
Nutrition and NEC
The incidence of NEC II or above was 32.9%and that of surgical NEC/perforation was 23.4%. The median (IQR) postnatal and post menstrual age of diagnosis of NEC II or above was 9 days (6.3–15.7 days) and 25.3 weeks (24.5–26.1 weeks).The risk of EUGR was higher in neonates who were of female sex [aOR = 5.0 (1.33–25.00)], received dexamethasone [aOR –5.55 (1.58–21.98)], and diuretics [aOR –15.02 (4.10 –70.67)]. Data related to nutrition, blood products transfusion practices and acute kidney injury are given in Table 3.
Neurological and visual outcomes
IVH of any extent was diagnosed in 70.2%of the neonates, 18.1%(survivors –10.4%) with grade III and 22.3%(survivors –6.2%) with grade IV IVH. The incidence of cystic PVL was 10.4%among the neonates who survived until discharge. While ROP was diagnosed in 75%of those who survived, 35.4%of the total survivors required therapy for the same.
The total duration of stay and causes of mortality are provided in Table 3.
Discussion
In this single centre retrospective study, the overall survival with no MBI was 38.3%(22–23 wks –32%, 24 wks–45.4%). Of those who survived until discharge, 75%did not have any MBI (22–23 wks –72.7%, 24 wks –76.9%). The survival rates of neonates in this study are similar to that reported in the recent systematic review and meta-analysis (2019) [2]. However, they are lower than the ones reported from Sweden (EXPRESS-2, 2019), Japan (NRNJ, 2017), Germany (single centre, Cologne, 2016), USA (single centre, Iowa, 2020) and higher than those from UK (EPICure-2, 2012), France (EPIPAGE, 2015) [5–8, 15–17].
There are many reasons for different survival rates between centres. A recent study had shown that centres that have a comprehensive approach in giving active care for these vulnerable infants have better survival than those that follow a selective approach [18]. Differing care practices also contribute to the variability in survival rates. Watkins et al. in their Iowa centre study had reported some practices which are different from our centre such as caesarian section for fetal compromise, predominantly inborn deliveries, provision of a separate NICU with experienced staff and use of HFOV as a primary mode of respiratory support [8]. Isayama et al. had reported certain other unique practices that are followed in Japan which are contrasting to ours such as monitoring internal cerebral vein velocity pattern in acute phase to prevent IVH, circulatory management guided by certain functional echocardiography parameters such as end-systolic wall stress, preference for invasive MV in the initial postnatal period to prevent intraventricular hemorrhage, use of gloves/masks/gowns for routine care practices for prevention of nosocomial infection as well as serial monitoring of CRP values to detect, treat bacterial infection and ubiquitous sedation for all ventilated neonates [19]. Some of these practices have been proven to be of doubtful benefit in relatively higher gestational age neonates [20–23]. Clearly, the general dearth of clinical studies on neonates of these low gestational age groups mandate further studies to explore many of these aspects.
In our cohort, severe respiratory disease in the initial postnatal life as indicated by not ready to be extubated in the first 72 hours and requirement of HFOV was associated with an increased risk of mortality or MBI. Laughon et al. had reported similar findings in neonates of 23–30 weeks’ gestation, with requirement of higher FiO2 and IMV (conventional as well as HFOV) being independently associated with the combined outcome of mortality or BPD [24]. Use of postnatal dexamethasone had a survival advantage which was in agreement with the meta-analysis by Doyle et al. which reported that the use of dexamethasone in neonates whose baseline risk of BPD was more than 60%resulted in a decreased incidence of death or cerebral palsy [25]. However, it is emphasized that this association between postnatal dexamethasone and survival could be due to more neonates who survived till fourth week of life and eventually receiving it, when compared to those who died early and were not eligible for postnatal dexamethasone. The NRS failure rate within 72 hours was 74.3%which was much higher than reported by the single center study from Cologne, Germany (50%) which uses NRS as a preferential initial mode wherever appropriate similar to our center [16]. This might be explained by the use of HFNC as a primary mode of respiratory support in our center compared to CPAP that was used in Cologne.
The incidence of blood culture proven late onset neonatal sepsis in our cohort was 38.3%overall and 50%among the survivors which was comparable to the NICHD-NRN data (40.7%) [1]. The mean AUR was 28.3%for the neonates included in the study. Not that many studies have reported the antibiotic exposure in this vulnerable group. Schulman et al. in their study of AUR amongst all neonates admitted to NICUs from California had reported a mean AUR of 21.8%[26].
The incidence of surgical NEC was 23.4%which was higher than reported by the Japan NRNJ (10.1%) [15]. Most of the nutritional practices in the present study NICU such as use of human milk, early use of minimal enteral feeds and probiotics are similar to that reported by Isayama et al. from Japan. AUR and transfusion practices were not reported in the Japan NRNJ report making further conclusions difficult. Finally, NEC is a multifactorial disease with genetic disposition as well which might explain the variability in the incidence of NEC between centers. The onset of NEC was also much earlier in our study [9 days (6.3–15.7 days)] compared to the time of onset reported in the literature. Sharma et al. had shown an inverse association between the post menstrual age and day of onset of NEC with mean age of onset of 22 (± 11) days among neonates of 23–26 weeks gestational age [27]. The gestational age range studied was different and the sample size was much lesser in the study by Sharma et al.
The rates of severe ROP requiring intervention in this study (35.4%) was similar to some of the published reports (Japan NRNJ –43.5%, NICHD-NRN–36.3%) but different from the Iowa centre study (11.3%) [1, 15]. These could be explained by the differing practices in relation to NRS, invasive mechanical ventilation, blood component transfusion, threshold at which ROP intervention is done as well as the varying prevalence of sepsis, all of which determine incidences of severe ROP mandating treatment [28].
This study has its limitations. It was retrospective in nature. It had not reported on the still birth rates, delivery room deaths and long-term neurodevelopmental outcomes of the studied neonates. A relatively small sample size resulted in imprecise estimates for many of the calculated aOR as well as aHR. The main strength of this study is that it had reported in detail the various care practices related to the neonates born at 22–24 weeks and contrasted them with centres reporting better outcomes. Also, it had tried to explore the various postnatal risk factors for mortality and severe short-term neurological outcomes which has rarely been addressed before.
Conclusion
The survival rates amongst the neonates born at peri-viable gestations of 22–24 weeks were better than those reported by the EPICure 2 and EPIPAGE-2 studies but lower than those reported from Japan, Sweden and some centres from USA, Germany. Differences in care practices (for which there is paucity of evidence) between various centres might explain the variability in survival. Most neonates who survived till discharge had no MBI. Severe respiratory disease in the initial days as indicated by not ready to be extubated within the first 72 hours of life and requirement of rescue HFOV was associated with a higher risk of death or MBI.
Ethics
This work was reviewed and authorized by the Joint Research Office study classification group of the institution as per the Department of Health’s UK Policy Framework for Health and Social Care.
Disclosure statements
Financial Disclosure Statement: The authors have no potential conflicts of interest. The research was not funded.
Human research statement
This work was reviewed and authorized by the ethics committee.
