Introduction

Top Abstract Introduction Methods Results Discussion References

In recent years several studies have reported a higher early neonatal mortality, particularly mortality related to asphyxia, in infants born during the night than in those born in the day.^1-3 This has important implications for health care as millions of births take place at night throughout the world. An exploration of the time of birth in relation to such mortality will help us to assess the importance of this problem in preterm infantsthose most at risk. We assessed whether infant mortality and early neonatal mortality is related to the time of birth in both full term and preterm infants born in Sweden.

	Methods

Top Abstract Introduction Methods Results Discussion References

We used data from the Swedish birth register, 1973-95.^4-6 The registry came into effect in 1973 and covers virtually all births in the country. A wide range of data about mothers and newborns are collated from interviews with mothers and from various records. The data include records on the exact time of birth. The inputs include standardised sets of forms used in all antenatal clinics and delivery units and during paediatric examinations of newborn infants. The records are linked to the registry of population that includes the death registry. ^{5 6} Data on mortality during the first year of life are merged with the data from the birth registry. There were 2 392 263 valid births recorded by the registry during 1973-95. To assess the effect of the time of birth in comparison with the recent result by Heller¹ we retained only those data on spontaneous live births (and thus not planned time of birth) of infants without considerable congenital malformation (2 102 324 infants). We also assessed which hours of birth over 24 hours were "high risk" periods for infant and early neonatal mortality and early neonatal mortality related to asphyxia.

During the study period there were no alterations in routines surrounding staff shift changes in Sweden. The obstetric nurses change shifts at 7 am, and the nurses have full responsibilities for each delivery. The obstetricians and neonatalogists change shifts at 8 am. Larger hospitals have a neonatalogist on call 24 hours; medium and small hospitals have a senior paediatrician on call 24 hours. The general principle is that all complicated deliveries will have a neonatalogist present, and premature and sick babies will be cared for in special neonatal wards. Early neonatal mortality was similar in babies born in large cities and those born elsewhere (1.57 v 1.53 per 1000, P=0.51), an indication of homogeneity of neonatal care across the country.

Night time births refers to births that occurred in the hours from 9 pm to 7 am.¹ Early neonatal death refers to death in the first six days of life. ^{1 7} The occurrence of death, as noted from the Swedish death register, was cross validated by the hospital based Swedish birth register. In a few cases of discrepancy between the hospital based records and the death register, we used the hospital records. We recorded the cause of death as defined by ICD-8 (international classification of diseases and related health problems, eighth revision) for 1973-86 and ICD-9 (ninth revision) for 1987-95. ^{4 8} Asphyxia related deaths were represented by code ICD-8 code 776 and ICD-9 codes 767, 768, 769, and 770.

We computed the absolute and relative risks to evaluate the impact of the time of birth on the risk of infant and early neonatal mortality and early neonatal mortality related to asphyxia. We assessed preterm and term births separately in the analysis of infant and early neonatal mortality. The analyses were done for 1973-9, 1980-9, and 1990-5 separately so that we could make comparisons with results from other countries for 1990-5 ^{1 7 8} and assess changes over decades.

View larger version (19K): [in this window] [in a new window]   Fig 1.   Infant mortality for spontaneous live births of infants without congenital malformation, Sweden 1973-95 (Cochran-Armitage test for trend, P<0.0001 for different rates during 1973-95)

	Results

Top Abstract Introduction Methods Results Discussion References

There were around 2.39 million live births during 1973-95. Table 1 shows the average infant mortality per thousand births. Early neonatal deaths (in the first six days) accounted for more than half of all deaths during the first year of life. A steady decline in infant mortality and early neonatal mortality was observed from 1973 to 1995 (fig 1). There was a slight increase in early neonatal mortality related to asphyxia during 1982-3, but otherwise there was a general downward trend from 1973 to 1995.

There was a higher risk of infant mortality and mortality related to asphyxia during the first year of life in infants born during the night than in those born during the day (table 2). The difference was mostly small but was greater for preterm infants. The most striking difference was in 1990-5, with an increase of about 60% in the risk of mortality related to asphyxia in preterm infants born during the night compared with preterm infants born during the day. The results for early neonatal mortalitythat is, in the first six dayswere similar to those for mortality in the first year (table 3). The increase of such mortality in 1990-5 was about 30% for early neonatal mortality (relative risk 1.31, 95% confidence interval 1.10 to 1.57) and about 70% for early neonatal mortality related to asphyxia (1.70, 1.22 to 2.38) in preterm infants born at night.

View larger version (31K): [in this window] [in a new window]   Fig 2.   Changes in mortality over 24 hours according to time of birth by one or two hour intervals for spontaneous live births of infants without congenital malformation, Sweden 1973-95. Dashed lines represent reference mean values over 24 hours (2 test for difference over 24 hours, P<0.0001 for infant mortality and early neonatal mortality by both one and two hour intervals; P=0.005 and 0.007 for early neonatal mortality related to asphyxia by one and two hour intervals, respectively)

Figure 2 shows the changes in mortality over 24 hours in the time of birth in one and two hour intervals. A high risk period occurred between 5 pm and 1 am, with another short high risk period around 9 am. There was a low risk period between 2 am and 4 pm, except for the short peak around 9 am. The largest difference was in early neonatal mortality related to asphyxia: infants born during the hours of highest risk (around 9 am and 9 pm) had almost double the risk of such mortality compared with infants born during the hours of lowest risk (around 4 am and 3 pm).

The changes in mortality followed similar trends over 24 hours during weekdays (Monday-Friday) and at the weekend (Saturday and Sunday), as shown for early neonatal mortality in figure 3. Early neonatal mortality was higher from 5 pm to 1 am and around 9 am for both weekdays and weekend days. The absolute early neonatal mortality decreased dramatically over the three periods, though the fluctuations over the 24 hours followed a similar trend with "high risk" periods around 9 am and 9 pm to 1 am (fig 4).

	Discussion

Top Abstract Introduction Methods Results Discussion References

Our observations confirm the previous reports from Germany and the United Kingdom that infants born at night have a higher risk of early neonatal mortality and mortality related to asphyxia than infants born during the day.^1-3 The new message from our observation is that this problem is much more serious for preterm infants and that shift changes and the hours immediately after are high risk periods. Furthermore, there was no improvement in the quality of neonatal health care in the night compared with that in the day over the decades we studied. There was an even greater relative risk of early neonatal mortality and mortality related to asphyxia for infants born at night during the 1990s.

View larger version (17K): [in this window] [in a new window]   Fig 3.   Changes in early neonatal mortality over 24 hours according to time of birth by two hour intervals and day of week for spontaneous live births of infants without congenital malformation in Sweden (2 test for difference over 24 hours: P<0.0001 for weekdays, P=0.036 for weekend)

View larger version (17K): [in this window] [in a new window]   Fig 4.   Changes in early neonatal mortality over 24 hours according to time of birth by two hour intervals for spontaneous live births of infants without congenital malformation in Sweden, 1973-9, 1980-9, and 1990-5 (2 test for difference over 24 hours: P<0.0001 for 1973-9, P=0.08 for 1980-9, P=0.005 for 1990-5)

Strengths and weaknesses in relation to other studies
Quality of data is an important consideration for studies on neonatal mortality. There is a large difference in early neonatal mortality reported within developed countries. ^{1 7 8} This may be due to true differences among populations but could also be due to a lack of consistency in collecting, recording, and coding data on births and deaths in different countries. For 1991-5 the observed early neonatal mortality for live full term infants without congenital malformation was 0.15 per thousand live births in Germany¹ and 0.32 in Sweden. In Canada the rate was 0.39 for 1992-4,⁵ and in the United States it was 0.27 for 1995.⁷ Our results are similar to those from studies in Canada and the United States, while the German study gave an exceptional low value and may be less accurate. The observed difference in early neonatal mortality between infants born during the day and night was much greater in the German study than that observed in our study.

Possible mechanism and implications
Even with the improvements in health care the greatest difference between mortality in infants born day or night was during 1990-5. There was about 30% higher early neonatal mortality and 70% higher neonatal mortality related to asphyxia in preterm infants born at night compared with preterm infants born during the day. There was also a large increase in absolute and relative risk of early neonatal mortality for preterm infants born at night in the 1990s compared with the 1980s. To some extent, there was no improvement or even a worsening in the quality of night time neonatal care compared with the daytime neonatal care in the 1990s. This certainly raises our concern, and efforts should be made to eliminate this difference. The benefit would be enormous considering the millions of births worldwide. The underlying causes are not clear and may be due to excess workloads, inadequate or less experienced staff on night shifts, or out of date systems for managing shift changes within hospitals.