Labor patterns vary widely, and both slow and fast labor are associated with complications [1–4]. We have studied unplanned out-of-institution births [5], which might occur because labor progresses faster than anticipated. A majority of the women experiencing such births are multiparous women with singleton, cephalic pregnancies in spontaneous labor at term. Improved prediction of time to completion of dilation in this group could improve planning and management of transportation to birthing institutions.

Many studies have examined the total length of the stages or phases of labor, or the median time of dilation in 1-cm integers (traverse time) [4, 6–8]. However, there is little consensus on the definitions of labor onset or the transition from latent to active phase [9, 10]. The shape of the classic Friedman curve is also contended because the average rate of cervical dilation appears to increase over time without distinct points of acceleration or deceleration [10, 11]. This gradual transition between the latent and active phases of the first stage of labor makes it difficult or impossible to objectively identify the onset of active labor. Average labor curves and tables showing median traverse time in centimeter integers can help diagnose abnormally slow labor progression but are not designed to estimate time to completion of cervical dilation conditional on the cervical opening and other factors known to be associated with the length of labor, including maternal age [12], height [13], body mass index (BMI) [14], parity [8], gestational age [15], and birthweight [6]. As well, timing of admission to the delivery unit [16], epidural analgesia [15, 17], and oxytocin use [18] are known to be associated with length of labor. Clinical protocol with early amniotomy and early oxytocin [19]–but not amniotomy alone [20]–is associated with slightly shorter time from admission to delivery.

The first stage of labor is normally the longest, and the transition to the second stage of pushing and imminent delivery is an important decision point for pre-hospital health workers. Little research, though, has been conducted to improve methods to predict when the transition is likely to happen conditional on cervical opening and other covariates. Therefore, we aimed to use factors associated with the speed of labor to predict time to transition from first to the second stage, using data from multiparous women with singleton, cephalic pregnancy in spontaneous labor at term.

Our main finding is that, in addition to cervical dilation, both birthweight and spontaneous rupture of the membranes markedly affect conditional time to completion of the first stage of labor in multiparous women. The association of fetal macrosomia (>4 kg) with longer labor and more frequent use of oxytocin has been well documented in previous research [25, 26]. In studies of labor 6 decades ago, Friedman found an large increase in the length of active nulliparous labor with increasing birthweight [27], but this association was much weaker among multipara [28]. This association was also found in the work of Nesheim [13] and in a large study which included women who delivered singleton, cephalic babies with gestational age >34 weeks at 19 hospitals in the United States [6]. This recent study analyzed birthweight in 0.5-kg increments and found that birthweight of more than 3 kg apparently became a factor in prolonging labor in multiparous women. However, a high proportion of the sample received epidurals and/or induction with oxytocin, and the researchers made no attempt to adjust for these factors. Incerti et al. found an association between gestational age and duration of active labor in nulliparous women at term but did not analyze birthweight [15]. We found evidence of collinearity between birthweight and gestational age, suggesting that the effect of gestational age is at least partially mediated by birthweight.

Artificial rupture of the membranes is performed to prevent or treat labor that progresses slowly. However, a recent review of the use of amniotomy in spontaneous labor found no evidence that it shortened the first stage of labor [20]. Unfortunately, we could not assess this association, but we found that the spontaneous rupture of the membranes predicted much faster progress. The clinical impression holds that spontaneous rupture often signals the descent of fetal parts and the acceleration of dilation. The apparent effect of spontaneous rupture was quite large among the subjects in the present study and could not be explained by other factors studied (e.g., different cervical measurements at admission). A likely mechanism is the reduction in uterine wall tension when the volume is reduced, as explained by LaPlace’s law [29]. This enables the uterus to maintain or generate higher pressure, which speeds up labor. Further studies on this possible explanation are warranted.

It is well known that higher parity (number of previous births) is associated with faster labor, and our findings in this regard were expected [7, 8, 13, 30, 31]. The relationship between maternal age and labor progress is unclear, although dystocia, including the need for oxytocin augmentation and prolonged labor, might increase with higher maternal age [32]. One study showed maternal age of more than 30 years to be associated with longer first stage labor (4 cm to complete dilation) in nulliparas and multiparas at term who did not receive oxytocin or epidurals [33]. However, other studies have found no association between maternal age and the duration of the first stage [13, 34]. In our previous study of unplanned out-of-institution births in Norway, we found that older women were less likely than younger women with the same parity to have unplanned out-of-institution deliveries [5]. We then speculated that this could be due to less vigorous labor or more precautions with age, and our findings from the present study support the latter.

We found that oxytocin use during the first stage of labor was associated with much longer conditional time, which is not surprising because oxytocin was exclusively used to treat slow progress. Studies have shown that early oxytocin for slow progress is associated with increased speed of labor [18, 35], and we find it likely that this was also the case in our sample. Although we have no way to offer proof, it is reasonable to assume that this relationship has more to do with the reasons or characteristics that led to oxytocin use, rather than its effects. Epidurals were also associated with longer conditional time, which can be related both to the indications for its use and to a negative effect on speed of labor [15, 17].

Maternal height has been found to be inversely related to length of labor, but we did not find this association in our sample [13]. Several studies have shown that higher maternal BMI is linked to longer labor and higher risk for caesarean delivery [36–39]. The underlying cause for this association remains poorly understood, but the effect of BMI on labor might be rather small in multiparous women [37]. We found a modest inverse relationship of a 1% reduction in the time to full dilation for an increment of 1 BMI unit.

The centralization of obstetric care which creates longer distances between homes and birthing institutions demands new solutions. The driving force behind this study was our interest in predicting time to completion of active labor and beginning of the second stage with pushing and imminent birth. The aim is not new, but the application of statistical models with complex functions might have the potential to improve prediction accuracy [15, 40]. However, some consider population-based average labor curves to be poor predictors for the individual progress of cervical dilation [30]. We tested our model in the sample and measured the mean absolute percentage error, which is a score used to measure how close predictions are to actual outcomes. Mean absolute error equally weighs all the individual differences and returns a value from 0 to infinity, with lower values indicating more accurate predictions. The predicted model outcome was not accurate enough to make clinical recommendations, but justified the generation of Fig. 4 and Additional file 2: Table S2 for exploratory purposes, clearly demonstrating the great importance of cervical measurements, gestational age, and spontaneous rupture of the membranes. For example, the estimated conditional time for a woman with 1 previous birth and 6-cm cervical dilation is 53 min with a gestational age of 259 days and spontaneously ruptured membranes but 97 min with a gestational age of 293 days and intact membranes.

Our study has many limitations. History of previous rapid labor might increase the likelihood of subsequent rapid labor, although we are not aware of any research supporting this speculation. Information about rapid or precipitous labor for our sample was not systematically recorded. We also did not include information about significant obstetric risk factors, which are thought to be uncommon in our practice. Other limitations include missing information about the artificial rupture of membranes and other factors that may be associated with the speed of labor (e.g., frequency and intensity of contractions, timing of oxytocin infusion and epidurals).

Including duration of the second stage of labor could also add value to our prediction table. However, the second stage of multiparous labor often lasts only minutes, and managing this stage en route to hospital should be avoided whenever possible. Transportation, particularly by helicopter, should also be avoided when birth is fast-approaching. In these cases, road transport by ambulance, expecting a stop en route for delivery of the baby, can be a better option.

As well, some women delivered multiple times during the study period, but we did not have permission to identify these women. This is unlikely to be a significant source of bias but is important to consider in future research. Finally, obstetric practice likely changed somewhat during the study period (e.g., increased use of epidurals), and different results may well be expected for other populations.

The strengths of this study include complete data from an entire population at a single institution meticulously collected by a small group of midwives. The population was homogeneous and consisted mostly of white women with a low prevalence of health- and social problems and strict obstetrical follow-up during pregnancy. We limited the analysis to a single Robson group of women who are too often exposed to unplanned out-of-institution births, and we excluded important confounders (e.g., previous caesarean, induction of labor and prolonged premature rupture of membranes) to avoid a “mixed bag” that clouds the issue. Labor patterns may have changed over the years, and consequently, groundbreaking studies on this important subject have lost some of their relevance [7, 28]. Thoughtful statistical modelling, considering repeated measures, adjusting for important confounders, and appropriately dealing with the complex, non-linear relationships between the predictors and outcomes enabled us to generate predictions for time to completion of cervical dilation conditional on cervical opening, parity, gestational age, birthweight, and rupture of membranes. However, due to many limitations, the study results can only be used for exploratory purposes and not for clinical recommendations. A larger prospective study could evaluate whether prediction of conditional time could be of a clinical value for parturition women in various Robson groups.

Higher parity, lower birthweight (gestational age), and spontaneous rupture of the membranes are associated with shorter conditional time to completion of the first stage of labor at term among multiparous women. Multiparous women in spontaneous labor should be expected to have fast labor. Increased accuracy in the prediction of labor speed could aid clinical management, including decisions on transfers to birthing institutions.