Nearly a third (32.4%) of women who had a live, singleton birth at ≥37 weeks gestation with spontaneous onset of labour and without a caesarean section scheduled in advance of labour onset, were admitted in early labour (< 4 cm cervical dilatation). Applying the criteria of at least 5 cm of cervical dilatation at time of admission increased the proportion of women admitted in early labour to 52.9%. Primiparous women were more likely to be admitted in early labour than multiparous women. After adjusting for parity, hospital admission during early labour was significantly associated with several clinical interventions (oxytocic augmentation, artificial rupture of membranes, epidural, pharmacological pain relief, caesarean birth and caesarean for slow progress, assisted vaginal birth, vaginal examinations, neonatal resuscitation), neonatal admission to special care and maternal length of hospital stay.
In this first Australian assessment of the prevalence of hospital admission in early labour, the identified rate of 32.4% falls at approximately the midpoint of the 10.2 to 58.5% range found in previous retrospective cohort studies in Sweden , Iran , Canada , the United States , Bangladesh  and Italy . Our estimates of the 52.9% prevalence of hospital admission in early labour when applying the criteria of at least 5 cm of cervical dilatation at time of admission compare with 60.3% of women admitted in early labour in a sample of Australian women in a randomised controlled trial which also employed criterion of < 5 cm cervical dilatation .
The odds of oxytocic augmentation was more than three times higher for women admitted in early labour than those admitted in active labour, consistent with associations found in all [3, 4, 6, 9,10,11,12, 14] but one  other study comparing oxytocic augmentation rates. Our findings indicate an absolute 24–27% difference (depending on the cut-point for defining early labour) in rate of oxytocic augmentation between women admitted in early and active labour. Others have reported differences in the rate of augmentation between women admitted in early versus active labour from - 0.50%  to 47.0% [10, 14]. Our findings of the absolute differences in rates and the odds of oxytocic augmentation were persistent across the alternative definitions of early labour applied here.
Admission in early labour was associated with more than twice the odds of epidural use over and above the impact of parity and augmentation with oxytocin, which increased to almost four times the odds when applying new WHO criteria of < 5 cm cervical dilatation, with an absolute 25–28% difference in rate of epidural use between women admitted in early and active labour. Other research [1, 3, 6, 12] has consistently found that women admitted in early labour are more likely to have epidural analgesia, with differences in absolute rates of epidural analgesia use between women admitted in early and active labour from 2.4%  to 25.5% .
The association between timing of admission and length of maternal hospital stay was attenuated after accounting for mode of birth in this study, but retained significance even after adjustment when early labour was defined as < 5 cm cervical dilatation. Our concurrent finding that women admitted in early labour had 2.8–3.5 times the odds of caesarean than those admitted in active labour (after accounting for parity), and their infants had 1.5–1.6 times the odds of special care nursery admission, indicates a significant combined burden of early admission on both women and the healthcare system. Our finding that admission before 5 cm cervical dilatation was associated with 2.8 times the odds (95%CI 1.60, 5.05) of caesarean birth was consistent with findings from another study in Australia that employed the same criteria for early labour admission and found it to be associated with 2.4 times the odds of caesarean birth .
In general, extending the definition of early labour to include women with up to 5 cm cervical dilatation revealed associations between stage of labour at admission and a wider scope of outcomes than applying the previously recommended definition of up to 4 cm. Although associations were no longer apparent with caesarean for fetal distress, early labour at admission was significantly associated with artificial rupture of membranes, assisted vaginal birth, neonatal resuscitation, use of pharmacological pain relief and maternal length of stay after all additional adjustments, when these outcomes were not significantly associated with stage of labour at admission using a less inclusive definition of early labour.
Associations between early labour admission, increased obstetric intervention (and subsequent use of other hospital resources) and increased risk of adverse clinical outcomes has not been well understood . It is possible that women who present and are admitted to hospital in early labour have an inherently higher risk of labour warranting medical intervention. It may also be that increased exposure to the medical system results in the amplification of risk for healthy labouring women .
Early labour admission is likely to be driven by interrelated factors. First, clinicians may have difficulty diagnosing active labour. There is little consensus in definitions of onset of any identified stages of labour in the research literature . Lack of evidence-based approaches to clinical decision-making for diagnosing active labour has been identified in several countries . Clinical judgements about the presence of active labour (and appropriateness of hospital admission) lack consistency between clinicians  and are usually made in busy clinical units with limited resources and emotional pressures that add complexity to the judgement . Second, women may prefer hospital admission prior to active labour. Women experience uncertainty about recognising labour, determining progress, and hospital admission decision-making [29,30,31,32] and seek early hospital admission because of pain, anxiety and feeling unsupported during early labour [31, 33, 34]. Although maternity services routinely advise women to return home before active labour, women often prefer admission on initial presentation over repeated discharge . In this study, we did not have the opportunity to assess women’s reasons for admission in early labour. Women may not be aware of the implications of early admission and/or lack confidence for managing pain or uncertainty at home.
A number of small trials have demonstrated that systematised clinician and consumer decision-making support tools can reduce rates of admission to hospital in early labour [16, 29, 35,36,37,38,39]. However, these studies are old, were conducted in non-representative samples and/or involve cost- and time-intensive interventions that are unsuitable for wide-scale implementation in a public hospital system. Currently in Australia, there are no standardised tools to guide clinicians’ or women’s decision-making about timing of admission. However, the combined evidence from this and other research provides sufficient basis for the development of decision tools that may improve the quality of decisions about timing of admission and prudent use of healthcare resources.
This is the first study to estimate the prevalence of early admission to hospital in labour in Australia. We employed rigorous quality assurance processes for the data extraction and found high consistency between data retrievers, and consulted with local clinicians to ensure the practical relevance of outcomes assessed. Nevertheless, there were some noteworthy limitations to our methods. First, we adopted a non-randomised, single-centre study design and limited our sample to women with live, singleton births equal to or greater than 37 weeks. While this was done purposefully to further control for anticipated confounding and optimise the comparability of our findings from Australia with previous research conducted internationally, we recommend caution in generalising our findings to all women. Second, we were limited by the data available in paper records in the study setting. While we assessed and statistically accounted for a several potential confounders of the association between early admission and outcomes, there may be others that remain unmeasured and unaccounted for (e.g., maternal anxiety). Third, it is possible that some women were categorised incorrectly for stage of labour at time of admission, given possible delays between timing of admission and first vaginal examination at which cervical dilation was recorded. If women presenting in early labour had moved to active labour at the time of first vaginal examination, they would have been misclassified as admitted in active labour, resulting in a bias towards no effect. We adjusted for parity in our analyses to reflect the odds of outcomes associated with stage of labour at hospital admission for all women, regardless of parity. Alternatively, we could have stratified our analyses consistent with clinical literature and conducted separate analyses by parity, but this would have resulted in complex probabilities to apply for specific subgroups that may threaten their usefulness for women’s decision making. Further formative work is needed to inform which scientific approaches provide findings that are most useful for all stakeholder groups, so that the type of evidence produced can be assured to meet the decision making needs of women in their timing of presentation to hospital in labour.