Neonatal Adverse Outcomes among Hospital Livebirths in Canada: A National Retrospective Study

While neonatal mortality has been in decline [1], the rate of severe neonatal morbidity has increased [2]. Several countries have developed hospital-based composite indicators to measure neonatal morbidity, and these have been used to track temporal trends or identify unexpected changes, leading to opportunities for system improvement or priority setting [3, 4].

The Canadian Perinatal Surveillance System historically used specific single measures of neonatal morbidity such as neonatal intubation or sepsis or the duration of newborn hospital stay [5]. Other indicators, including small-for-gestational age, birthweight, preterm birth, and neonatal mortality were also reported in Canada [6]. Applying the methodology developed in Australia [4, 7], the UK [8] and France [3], each reported their version of the composite measures of neonatal morbidity referred to as neonatal adverse outcome indicator (NAOI). Such NAOI are reported to be predictive of healthcare utilization and infant survival [2‒4, 7, 8]. The current study was undertaken by the Canadian Perinatal Surveillance System (CPSS) to formally create a NAOI for Canada, compare Canadian rates of NAOI to other nations that have used this indicator, and assess temporal trends and the relation between NAOI and mortality, neonatal intensive care unit (NICU) admission, and extended hospital stay.

All study data were obtained from the Canadian Institute of Health Information’s (CIHI) Discharge Abstract Database (DAD). The CIHI-DAD contains demographic, clinical, and administrative information for every hospital birth, including birthweight, gestational age at birth, length of stay, and discharge status as alive or deceased [9]. Almost all births in Canada occur in hospital (98%) [10]; therefore, the CIHI-DAD is the largest Canadian data source for maternal and infant health. All Canadian provinces and territories, except Quebec, report to CIHI-DAD [9]. Discharge diagnoses are coded using the Canadian version of the International Classification of Diseases (ICD-10-CA) and interventions/procedures using the Canadian Classification of Interventions (CCI) methodology.

This retrospective cross-sectional study included liveborn infants between 24 and 42 weeks’ gestation, at any hospital in Canada other than Quebec hospitals, from fiscal years 2013 to 2022 which include all births between April 1, 2013, and March 31, 2023. We excluded neonates born before 24 weeks and after 42 weeks’ gestation, livebirths following interruption of pregnancy, neonates with birthweight of <400 g and >7 kg, and those with missing data on gestational age (shown in online suppl. eFig. 1; for all online suppl. material, see https://doi.org/10.1159/000540559). Ethics approval was not required for this study as data have been anonymized and analyses were conducted under the surveillance mandate of the Public Health Agency of Canada.

The first NAOI developed in Australia consisted of infant birth characteristics, conditions/procedures grouped into diagnostic and intervention groups which included a birthweight <1,500 g, a gestational age 24–32 weeks, death before hospital discharge, birth trauma, chronic respiratory disease originating in the perinatal period, cerebral infarction, hypoxic ischemic encephalopathy (neonatal encephalopathy), intraventricular hemorrhage (grades 2, 3, or 4), necrotizing enterocolitis, primary atelectasis, respiratory failure, periventricular leukomalacia, pneumonia, respiratory distress syndrome, seizure, sepsis, any body cavity surgical procedure, receipt of any intravenous fluids, central venous or arterial catheter, pneumothorax treated with intercostal catheter, resuscitation, transfusion of blood or blood products or ventilatory support (mechanical ventilation and/or continuous positive airway pressure).

We identified conditions/procedures found in the NAOI developed in Australia and other published studies of neonatal morbidity, including those with composite indicators [2‒4, 7, 8, 11]. We assessed if there were any significant variations for each ICD-10-CA and CCI-10-CA code over time (i.e., due to coding changes) prior to inclusion of any code as part of the composite indicator.

Based on expert panel review, four modifications were made to create the “NAOI Canadian method.” We added two conditions – neonatal intestinal perforation (ICD-10-CA P78.0) and retinopathy of prematurity (ICD-10-CA H35.1). Additionally, unlike NAOI from other countries [3, 4, 7, 8], we did not include an infant birthweight <1,500 g or <32 weeks’ gestation as part of the NAOI inclusion criteria. This was a consensus decision, as these birth characteristics predict neonatal outcomes. The final morbidities included in the NAOI comprised infant death, 14 groups of diagnostic codes, and seven groups of procedure codes (shown in online suppl. eTable 1). These conditions/procedures were identified in the index birth hospitalization record. If an infant had any single NAOI component, then they were coded as having NAOI. We examined the contribution of each condition and procedure to the overall NAOI rate.

We evaluated three neonatal outcomes: death, NICU admission, and an extended hospital stay. Mortality was defined as death during the index birth hospitalization, regardless of the duration of hospitalization. NICU admission was defined as admission of the neonate to an intensive care unit during the index birth hospitalization. Extended hospital stay for term infants (born ≥37 weeks) was defined as 5 days or longer, and for preterm births (born <37 weeks) was defined as more than 7 days after correcting to 37 weeks of post-menstrual age.

The rates of the composite NAOI and individual NAOI components were compared with those of France, Australia, and the UK [3, 4, 7, 8], for three gestational age groups (24–33, 34–36, 37–42 weeks).

We calculated the NAOI rate and the distribution of the number of NAOI components (0, 1, 2 3, 4, or ≥5) overall and by four gestational age groups (24–27, 28–31, 32–36, and 37–42 weeks). Rates and unadjusted rate ratios (RR) and 95% confidence intervals (CIs) for mortality, NICU admission, and extended length of stay were determined in relation to the number of NAOI components (0 [referent], 1, 2, 3, 4, or ≥5). When assessing risks of neonatal mortality, death was excluded from the NAOI components. Temporal trends in NAOI were assessed using a Cochrane-Armitage test. The alpha threshold for statistical significance was determined a priori, at a p value <0.01. Analyses were conducted using SAS EG version 7.1 (SAS Institute Inc. Cary, NC, USA).

The study comprised 2,821,671 infants born at 24–42 weeks’ gestation during the 10-year study period. There was a significant temporal trend of increase in NAOI from 7.3% in 2013 to 8.0% in 2022 (p < 0.01) (online suppl. eTable 2, shown in Fig. 1). Among the individual NAOI components, significant increases were seen for neonatal encephalopathy, respiratory distress syndrome, and ventilatory support (online suppl. eTable 2). Conversely, significant declining trends were seen for birth trauma, cerebral infarction, pneumonia, seizures, sepsis, receipt of intravenous fluids, and use of central venous or arterial catheter, and resuscitation (online suppl. eTable 2).

The NAOI rate was 98.4% for those born at 24–27 weeks’ gestation, 94.3% at 28–31 weeks, 27.5% at 32–36 weeks, and 5.1% at 37–42 weeks (Table 1). With lower gestational age, the number of NAOI components increased (Table 1). For example, only 0.1% of term infants had 4 NAOI components, and none had >5 NAOI components, whereas at 24–27 weeks, the respective proportions were 20.0% and 44.5% (Table 1).

Dominant NAOI components included ventilatory support, use of intravenous fluids, and a diagnosis of respiratory distress syndrome, especially among neonates born at 24–33 weeks’ gestation (online suppl. eTable 2). Infants born at <34 weeks also experienced a higher rate of intraventricular hemorrhage, necrotizing enterocolitis, seizures, sepsis, and active resuscitation compared to those born at >34 weeks’ gestation (online suppl. eTable 2).

There was a curvilinear positive relation between NAOI and each outcome (Table 2). For example, relative to no NAOI, the RR for neonatal death was 8.5 (95% CI 7.6–9.5) with 1, 118.1 (95% CI 108.4–128.4) with 3, and 395.3 (95% CI 367.2–425.0) with 5+ NAOI components (Table 2). For NICU admission, the respective RR were 6.7 (95% CI 6.6–6.7), 11.2 (95% CI 10.9–11.3), and 11.9 (95% CI 11.6–12.2), and for extended hospital stay, the RR were 6.6 (95% CI 6.4–6.7), 12.2 (95% CI 11.7–12.7), and 26.4 (95% CI 25.2–27.5), respectively (Table 2).

Across the four countries, there was marked variability in reported rates of certain components of NAOI, particularly among the 24–33 weeks’ GA neonates. For example, receipt of intravenous fluids was 2.1% in the UK and >60% in Canada. Any body cavity surgery was lowest in Canada with a prevalence of 1.5% compared to 7.9% in the UK. These variabilities were minimal among the 34–36 and 37–42 neonates. Overall, Canada had a higher prevalence of NAOI (7.6%) compared to France (4.9%), the UK (5.4%), and Australia (5.4%) (Table 3).

Based on the adapted version of NAOI for Canada, 7.6% of infants born in hospital experienced NAOI. The NAOI rate rose with degree of prematurity and over time. As the number of NAOI components increased, so did the associated risk of newborn death, NICU admission, and extended hospital length of stay in hospital.

We observed an overall higher prevalence of NAOI in Canada (7.6%) compared to France (4.8%) or the UK and Australia (5.4% respectively) [3, 7, 8]. This was unlikely to be explained by the four changes made to the NAOI codes for Canada, which altered the overall NAOI rate by only 0.1%. Overall, differences in the individual NAOI components between countries were minimal in term neonates, but large differences were observed among the most preterm infants. Among the preterm newborns, Canada had lower rates of central venous catheter insertion than Australia or the UK, but higher use of intravenous fluids (Table 3). The lower rate of central venous catheter use could be due to a change in coding directive in 2015 [12], whereas the higher use of intravenous fluids is likely to reflect over-coding in our data because no dosing information was used to establish its inclusion as part of NAOI. Rates of ventilatory support among newborns at 24–33 weeks were lower in Canada than elsewhere, but this could be explained by the fact that the Canadian data were from a more recent era, during which ventilation practices have evolved with more neonates being managed using noninvasive respiratory support [13]. The overall rate of newborn resuscitation was also higher in the comparator countries than in Canada.

While other countries reported lower rates of NAOI, their mortality rates tended to be higher. This may be due to potential differences in study era, improvements in the provision of antenatal and newborn clinical care, and changes in baseline population risk. Furthermore, we excluded a small number of infants born with signs of life after a termination of pregnancy; since they are not usually viable, their inclusion would otherwise artifactually inflate the mortality rate. The current study had a sample size two to five times that of previously published NAOI studies and used more recent data than previous studies which ended in 2015. Importantly, data were not linked to vital statistics death registries, so death after the index birth hospital discharge date was not accounted for. NAOI seems to occur in Canada at a comparable rate to other countries using the international method versus the Canadian method, suggesting reasonably high concurrent validity with prior studies of NAOI as a composite indicator [3, 4, 7, 8].

A novel finding was the curvilinear relation between the number of individual NAOI components and the likelihood of experiencing the three adverse outcomes. For example, infants with two NAOI indicators were 49 times more likely to experience mortality than those without NAOI, while infants with at least five NAOI features were almost 400 times more likely to die. An inverse relation was also seen between the number of NAOI components and gestational age at birth. Certainly, a future study might clarify whether the number of NAOI components offers greater clinical prediction of infant death than gestational age at birth.

Two key distinguishing features of our study include marked variability in reported rates of certain components of NAOI between countries. We believe this is due to variability in interpretation of ICD codes at coding level. For example, receipt of intravenous fluids for 24–33 weeks’ GA neonates was 2.1% in the UK and >60% in Canada. Hence, there is a need to harmonize the definitions and codes for creating a NAOI, including periodical updates by new coding directives [12]. The NAOI should be also subjected to clinical expert review, to ensure its comprehensiveness in future iterations.

Second, we identified an increasing rate of NAOI in Canada in recent years. After 2019, this increase was primarily driven by an increase in the use of ventilatory support and “other respiratory conditions,” the latter of which includes primary atelectasis and respiratory failure. The use of noninvasive respiratory support has been steadily increasing, while the use of invasive ventilation has declined in the management of newborns with respiratory distress in recent years [14, 15]. However, that alone cannot explain the recent increase. Previous international data are only available until 2015, whereas Canadian data are up to 2022 and therefore more contemporaneous. We could not compare our increasing rate with other countries in last 4–5 years due to lack of data, and we propose and ask international communities to update and report their results. In the meantime, we will investigate further within the Canadian context as to reasons for this increase.

This study supports the use of hospitalization data for generating and for monitoring NAOI, and its large sample size is representative of almost all Canadian newborns since 98% of babies are born in hospital [16]. The absence of data from Quebec is unlikely to influence the precision or accuracy of NAOI estimates. For the purposes of this study, adjustment for confounding was not warranted, as it could complicate international comparisons. Moreover, it remains to be determined which maternal, newborn, or country-based factors confound this relationship. Another strength of our study is its contemporaneous nature with data from births in 2022 included.

This Canadian version of NAOI offers a contemporary, comprehensive, and valid way to monitor trends in neonatal morbidity and mortality using available hospitalization data. Infants with NAOI are at higher risk of adverse outcomes when compared to those without NAOI and in a curvilinear dose-response manner. The comparability of the Canadian NAOI to those published in other countries suggests that Canada can contribute to existing international efforts to establish a common measure of neonatal morbidity. Further studies are warranted to understand reasons for increasing rates of NAOI in Canada in recent years.

The first author would like to thank Dr. Susie Dzakpasu for her invaluable feedback on the draft manuscript.

The data analyzed in this study were provided to the Public Health Agency of Canada (Agency) by the Canadian Institute of Health Information (CIHI) to fulfill the Agency’s national health surveillance mandate. All researchers with access to the data were employees of the Public Health Agency of Canada. This is secondary data analysis using administrative data; therefore, consent was not needed for this study. The CIHI data analyzed in this study were collected under the authority of jurisdictional privacy or health information legislation and provided to the Agency through a data-sharing agreement. Data disclosed to third parties for research, like in this study, is in the form of de-identified record-level data. Additionally, as per Article 2.2 of the Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans, this research is exempt from ethics review or from requiring participant consent as it relied on de-identified, secondary data analysis.

The authors have no conflicts of interest to declare.

Joel G. Ray receives funding from the PSI Foundation, and Nathalie Auger is the recipient of a clinical research award from the Fonds de recherche du Québec-Santé (#296785). Both authors declare that their funding had no role in the study design, execution, analysis, manuscript conception, planning, writing, and decision to publish this study.

C.R.M.N., P.S.S., J.G.R., N.A., J.L., and A.M. contributed to the conception of the study. C.R.M.N., P.S.S., and J.G.R. carried out the study and developed statistical methods. C.R.M.N. performed all statistical analysis and drafted and revised the manuscript based on comments from other authors. C.R.M.N., J.G.R., P.S.S., N.A., J.L., A.M., P.M., and M.H. contributed to the interpretation of the data, critically reviewed all drafts of the manuscript, and approved the final version submitted for publication.

The data that support the findings of this study were made available from the Canadian Institute for Health Information to fulfill the surveillance mandate of the Public Health Agency of Canada. The data analyzed for this study are not publicly available as they were provided by a third party through a data-sharing agreement. Applications for accessing the data can be made by completing a request form through the Canadian Institute for Health Information’s website at https://www.cihi.ca/en/data-inquiry-form.