Discussion
The findings of this study indicate a huge regional variation in percentage of births registered as live born for periviable babies and had a great impact on regional ranking of NMR. Our hypothesis was not supported, that is, the percentage of registration of live born for periviable babies in cities was not higher than the percentage in counties. However, metropolitans (such as Taipei City and Kaohsiung City) with relatively higher percentage and larger number of periviable babies which have been registered as live born would have greater change in ranking of NMR.
Series of studies in Canada have indicated provincial variation in registration of live born for birth weight <500 g and associated with variation in ranking of provincial infant mortality rates.5–10 For example, the registration of live births less than 500 g and less than 24 weeks of gestation was more meticulous in Alberta than elsewhere in Canada, which resulted in an increase in infant mortality rate in Alberta.10
Two UK studies indicated higher proportion of very preterm infants registered as live born and transferred to NICU in Trent, UK, compared with those in Nord Pas-de-Calais, France, and New South Wales/the Australian Capital Territory, Australia, was the main reason for higher NMR in Trent. The authors reminded that information about very preterm babies (not usually included in routine statistics) is vital to avoid inappropriate interpretation of international perinatal and infant data.11 12
A study in the USA investigated the state-level variations in the classification of live birth or fetal death would be reflected in the fetal and infant mortality rates for this birth weight group. The findings of that study suggested that six states were more likely to classify outcomes as a live birth/infant death, while 14 states were more likely to classify as a fetal death, when compared with a large reference state. The authors concluded that fetal death and early infant death outcomes reported for babies less than 500 g reflect differential classification, thus influencing the validity of these vital statistics data at the state level.13 Another US study further indicated that the birth hospital was an important predictor of whether the death was classified as a fetal or infant death. Among the 31 hospitals selected for study, there was a nearly 15-fold variation in the probability of events being classified as early neonatal versus fetal death.14
A recent US study assessed county-level variation in death event reporting at 17–20 weeks of gestation and associations with county infant mortality rates. Of 2391 counties studied, the percentage of deaths at 17–20 weeks reported as fetal ranged from 0% to 100% (mean 63.7%) and every 1 point increase in this percentage was associated with a 0.02 point decrease in county infant mortality rate. The authors suggested that the variation in the reporting of previable gestation deaths likely reflects a combination of legislative policy and hospital practices, cultural norms in attitudes towards pregnancy termination and neonatal resuscitation, as well as social, economic and religious views.15
A study of 147 primary care trusts (PCT) in England indicated wide between-PCT variation in percentages of births <24 weeks of gestation registered as live born with 90% central range from 26.3% to 79.5%. Excluding births <24 weeks led to significant changes in infant mortality rankings of PCTs, with a median worsening of 12 places for PCTs with low rates of live born preterm births <24 weeks of gestation compared with a median improvement of four ranks for those with higher live birth registration rates.16
The findings of this study are consistent with the findings of study in England. We noted prominent improvement in ranking of NMR after excluding live births <500 g in thee largest metropolitans (Taipei City and Kaohsiung City). The two cities had relatively higher percentage and larger number of births <500 g registered as live born and consequently had a greater impact on ranking of NMR after excluding the live birth <500 g. We also identified several cities/counties showing relatively large increase in number of reporting live births <500 g in 2015 or 2016. Future research is needed to explore the possible reasons of the increase.
One possible explanation that our hypothesis was not supported was that many pregnant women with possible periviable births who resided in counties (eg, Nantou County, Miaoli County or Changhua County) nearby Taichung City (figure 1) might go to level 3 medical centres with NICUs in Taichung City for deliveries. These births were still registered in the counties in which these women resided. Further studies are needed to examine the variation in birth hospitals in Taiwan.
To better interpret the findings of international comparisons of infant and NMR in the fact of regional variations in reporting live births <22 weeks’ gestation or <500 g birth weight across countries, the OECD Stat presents both criteria, that is, ‘no minimum threshold of gestation period or birth weight’ and ‘minimum threshold of 22 weeks or (500 g birth weight)’.23 The findings of this study also reveal large variation in percentage of reporting live births <500 g across cities/counties in Taiwan. We thus recommend presenting the infant and NMR using both criteria and let the readers have more information in interpretation of the findings of the comparisons.
Several limitations should be noted while interpreting the findings of this study. First, this is an ecological study, we did not take into account the associated factors affecting the city/county NMR. Second, the information on birth hospital that delivered the periviable babies was not available. We thus could not determine how many births <500 g registered as live born in counties were actually delivered in level 3 medical centres in the cities. We also could not detect if the behaviour of registration of live born for births <500 g in some cities/counties was concentrated in particular hospitals. Third, this is not a survey study, we could not understand the real reasons in some healthcare providers having higher percentage of registration of live born for births <500 g. Fourth, the regional variation in registration of live born would be influenced by the regional variation in determination of stillbirth. However, the information on the stillbirth on parents at earlier gestations or whether the stillbirth was due to termination of pregnancy in each city/county was not available in Birth Report System open government data.