Patterns and outcomes of preterm hospital admissions during pregnancy in NSW, 2001–2008

In Australia, regionalised maternity care aims to ensure provision of the appropriate level of care to all women, through assessment of risk, referral and transfer. An important element of this care is providing high-quality, safe and accessible services to women and babies in rural and remote areas.1,2 For women who experience pregnancy complications, this means their needs may not necessarily be met by their local hospital. They may be admitted and then be discharged or transferred or give birth. Babies born in a tertiary centre to women experiencing pregnancy complications (including those transferred from another hospital before the birth) have lower rates of morbidity and mortality than babies who are transferred after they are born.3,4

Pregnancy and childbirth is one of the most common reasons for hospital admission in Australia, and accounted for about 8% of all overnight acute admissions in 2009–10.5 Decisions to admit or transfer women who present to hospital with complications during pregnancy depend on various factors, including: severity of the complication; stage of pregnancy or labour; bed availability; availability of specific clinical expertise; hospital services; proximity of services to the woman’s home; women’s or clinician’s preferences; and availability of support at home.

The few studies that have examined admissions during pregnancy or the subsequent pathways for these women have been limited by non-representative populations6 or inability to track individuals through multiple admissions and transfers.7-9 Because of a lack of data on single or multiple admissions during pregnancy and transfers between facilities, it is not possible to evaluate how well the policy aims of equitable and efficient use of resources and optimal outcomes for neonates10 are being met.

The aim of this study was to assess the number, factors associated with, and outcomes of preterm admissions to hospital, with a focus on transfers to hospitals with a higher level of maternity care.

The study population comprised women who gave birth to a liveborn singleton infant ≥ 24 weeks’ gestation in 2001–2008 in New South Wales and who were admitted to hospital during weeks 20–36 of their pregnancy (“preterm”). Women were included for each pregnancy that met the criteria.

We obtained data from the NSW Perinatal Data Collection (PDC) and the NSW Admitted Patients Data Collection (APDC). The PDC (“birth” records) is a population-based surveillance system, covering all live births and stillbirths at ≥ 20 weeks’ gestation or ≥ 400 g birthweight, and includes demographic, medical and obstetric information and information on the condition of the infant. The APDC (“hospital” records) is a census of all inpatient admissions (public and private) in NSW. It includes demographic, administrative and clinical information. Diagnoses related to the admission are coded according to the 10th revision of the International Classification of Diseases, Australian Modification (ICD-10-AM).

Probabilistic linkage of birth and hospital records for mothers and babies was conducted by the Centre for Health Record Linkage and was approved by the NSW Population and Health Services Research Ethics Committee. These datasets have low rates of missing data and generally high levels of agreement when compared with medical records.11-13

For each birth, records of the mother’s hospital admissions (including hospital-to-hospital transfers) where the admission occurred at 20–36 weeks’ gestation were retained. Day-only admissions were included. Gestational age (in weeks) at admission was estimated from the baby’s date of birth and gestational age in the birth record, and the date of admission in the hospital record.

The primary outcome was whether a preterm admission ended in discharge, maternal transfer to a hospital with a higher level of maternity care, or birth (Box 1). Outcomes of transfers to higher care were examined as discharge, transfer again to still higher care, transfer back to lower care, or birth. Neonatal outcomes for pregnant women admitted preterm were investigated, including length of stay, transfer, morbidity and mortality.14

Maternal characteristics (age, number of previous pregnancies, smoking during pregnancy, Indigenous status and area of residence) were obtained from the birth records. Quintile of socioeconomic status was determined by the Index of Relative Socio-economic Disadvantage,15 and remoteness category by the mean Accessibility/Remoteness Index of Australia (ARIA+) score for postcode of residence16 (Box 2, full version online only at mja.com.au). Reason for admission was based on the principal admission diagnosis on the hospital record. Hospitals were classified into seven groups based on available neonatal care and number of births, and urban or regional area.17

From 2001 to 2008 there were 688 902 liveborn singleton births at ≥ 24 weeks’ gestation, of which 110 439 (16.0%) involved a mother who was admitted to hospital during weeks 20–36 of the pregnancy (Box 1). These births were to 98 183 women, of whom 68 458 (69.7%) were admitted preterm during one pregnancy, 23 970 (24.4%) were admitted preterm during two pregnancies, and 5755 (5.9%) were admitted preterm during 3–6 pregnancies. The proportion of pregnancies during which the mother was admitted increased from 15.5% in 2001 to 16.2% in 2008 (trend P = 0.01).

During these 110 439 pregnancies, the women had a total of 165 174 preterm admissions if transfers are counted as a separate admission (24.0 per 100 pregnancies), or 153 936 preterm admissions if transfers are counted as part of the preterm admission (22.3 per 100 pregnancies). The women had one preterm hospitalisation in 75.2% of pregnancies involving admission, two preterm hospitalisations in 16.8%, three in 4.9%, and 4–33 hospitalisations in 3.2% of these pregnancies. Counting only admissions that did not result in birth gives 18.4 non-birth admissions per 100 pregnancies (126 573/688 902).

Compared with pregnant women who were not admitted preterm, women admitted preterm were significantly more likely (P < 0.01) to be < 25 years and either in their first pregnancy or to have had more than one previous pregnancy resulting in birth (data not shown). They were also significantly more likely (P < 0.01) to be Aboriginal, report smoking during pregnancy and live in regional or remote areas, and less likely (P < 0.01) to be in the highest socioeconomic group.

To our knowledge, this is the first study to use linked population data to examine preterm hospital admissions among pregnant women in Australia. From 2001–2008 in NSW, about one in six pregnant women were admitted to hospital at least once during weeks 20–36 of their pregnancy. Australia-wide, this would represent a significant impact on the health system, translating to about 63 500 preterm admissions each year, almost three-quarters of which will end in the woman being discharged without giving birth. This includes about 3700 women per year transferred to higher care, almost half of whom will be subsequently discharged.

Maternal risk factors for transfer and birth among women admitted preterm in our study (advanced maternal age, extremes of parity, Aboriginality, smoking during pregnancy and social disadvantage), are recognised risk factors for adverse pregnancy outcomes.18,19 Suspected preterm labour was the most common reason for admission and was strongly associated with transfer and birth. However, for women admitted for preterm labour, the high rates of discharge following the initial admission (59%) and transfer (62%) highlight the difficulty in determining which pregnancies will be ongoing. Future improvements in assessment of threatened preterm labour may have considerable potential both to reduce length of stay for the initial admission and to reduce unnecessary transfers.

Preterm prelabour rupture of the membranes accounted for 5% of preterm admissions and was the strongest risk factor for transfer and birth. However, management of this condition remains a dilemma with respect to immediate birth or expectant management, and hospital or outpatient care.20 Hypertension and haemorrhage were the other major contributors to preterm admission, each accounting for 11% of admissions, but were weaker risk factors for transfer and birth. For most pregnancy conditions there is a paucity of evidence for keeping women in hospital for bed rest once their condition has stabilised.21-23

When a maternal or fetal risk is identified that may lead to preterm birth, optimal care is provided by ensuring care at a centre that has the expertise and facilities to provide appropriate management,24 although this may remove a woman from her family and other social supports. In our study, infants born after maternal transfer had lower gestational ages and were less likely to be discharged directly home, suggesting that clinicians do, in general, identify pregnancies that are at increased risk. However, almost half the women transferred were discharged without giving birth, about twice the rate observed in a population-based study in the United Kingdom.25 Differences in health care services, especially the distances involved with access to referral centres, make direct comparisons difficult. There may be more scope for reduction of unnecessary transfers in urban and outlying metropolitan areas in Australia, where distance is a less critical factor. Neonatal transfers, on the other hand, can represent “missed” opportunities for antenatal transfer; our findings suggest that “missed” transfers were mostly due to rapid progression of labour. Clinicians may have identified these women as candidates for transfer but were not confident of sufficient time to effect an antenatal transfer.

The availability of linked hospital admission data enabled us to calculate population rates of admission and transfer, unlike previous studies which have been limited to calculating ratios of admissions to pregnancies.7-9 Our finding of 18.4 non-birth admissions per 100 pregnancies is comparable with the findings of studies from other countries (12–30 per 100).7-9 The main overall reasons for admission in our study (preterm labour and hypertension) and for admission resulting in birth (premature rupture of membranes, preterm labour and hypertension) were also consistent with earlier studies.7-9,26 A US study reporting linked outcome information found that 66% of antenatal admissions ended in the woman being discharged without giving birth, compared with our findings of 72% in NSW; that study only counted antenatal admissions that were at least 4 days before the actual birth date.26

The limitations of using population data include the lack of detailed clinical information on the severity and timing of medical conditions, and on social circumstances of the mother, which would enable a fuller assessment of the appropriateness of antenatal admission and transfer. The availability of maternity services staff at smaller hospitals at any given time is another unknown, and may play an important part in transfer decisions.

Although they are generally short, preterm admissions consume resources that are already stretched — the current workforce is insufficient, and facilities are running at unacceptably high occupancy rates. Strategies aimed at reducing antenatal admissions appear justified and could have important benefits for mothers, their families and health care facilities.

1 Pathways for pregnant women admitted to hospital preterm* in New South Wales, 2001–2008

* During weeks 20–36 of the pregnancy.