Introduction

The term cascade of interventions describes the tendency of interventions to accumulate during labor [18, 23]. Epidural analgesia, oxytocin augmentation and amniotomy are three interventions commonly administered during the process of labor. These interventions have been investigated for their association with labor duration and mode of birth, taking into account early versus late administration [1, 10, 11, 22, 24, 25]. A few authors have also considered the optimal timing of interventions as measured by cervical dilation [10, 14]. However, most previous studies of intrapartum interventions focus on their combination and not their timing or sequence.

At present, there is no evidence that oxytocin [3, 24] or amniotomy [22] increase the use of epidural analgesia when administered alone. However, studies suggest that the rate of epidural analgesia increases when amniotomy and oxytocin are combined [25]. The reverse association has also been studied. Epidural analgesia may increase the need for oxytocin augmentation [1, 10, 17]. These findings are important because epidural analgesia alone [1] or in combination with oxytocin has been found to be associated with an increase in adverse birth outcomes [18, 23]. Recent studies also express concern about the risks of administering oxytocin augmentation during labor [2, 4, 5, 1921].

A few studies have addressed not only the combination of intrapartum interventions, but also the relationship of intervention sequences to birth outcomes [7, 18, 23]. Several authors used predefined intervention sequences in their analyses [18, 23]. Dickinson et al. [7] investigated both the sequence and the timing of epidural analgesia and oxytocin augmentation. They found labor duration to be slightly longer when oxytocin was administered before epidural analgesia compared to after, though this was not statistically significant (P > 0.05). The possible differential effects of these common interventions when used in different sequences and at different points throughout labor remain largely unclear.

Our aim was to document the sequences of three intrapartum interventions—oxytocin augmentation, epidural analgesia, and amniotomy—and the cervical dilation preceding each, in comparison between nulliparae and multiparae, as well as to understand how the various sequences were associated with labor duration and mode of delivery.

Materials and methods

Data on birth processes were collected in 47 maternity units in Lower Saxony, Germany [16]. The study was designed as a longitudinal cohort study. Pregnant low-risk women were considered eligible if they had a vital single fetus in vertex presentation and planned a vaginal birth. As a German guideline recommends different forms of care in the event of preterm labor below 34 weeks, the study was limited to pregnancies of at least 34 completed weeks of gestation [9]. Induced labors (by oxytocin, prostaglandins, amniotomy, misoprostol or castor oil) were included.

In Lower Saxony there were 63,719 live births in 2005. From April to October 2005, attending midwives prospectively documented detailed information on 1,169 birth processes for the study. The sample size for studying the cascades of interventions and their multivariate associations was designed to detect a hazard ratio of 1.2 between two groups at the level alpha (two-sided) = 0.05 with a power of 80 %. We therefore calculated the necessary number of events to be 944. To achieve this power for events of a 50 % probability, the target number of labors to be analyzed was n = 1,888 for the nulliparae as well as multiparae group. We supplemented the prospective sample with additional births (n = 2,786). Data on these births were retrieved from the state-wide perinatal auditing database at the Center for Quality and Management in Health Care, Lower Saxony Chamber of Physicians. The timing of interventions during the births was obtained from hospital medical records. The births were to women who met the same eligibility requirements as the prospective sample. The form for documentation was the same for prospective and retrospective data. Institutional approval for the anonymous gathering of information was granted by the ethics committee of Hannover Medical School and by the ethics committee for all public hospitals in Lower Saxony. We excluded eight nulliparae of the complete sample of 2,090 nulliparae as we had no information whether they experienced an amniotomy or a spontaneous rupture of membranes.

Our results are stratified for nulliparae (np: n = 2,082) and multiparae (mp: n = 1,873). Women with a previous cesarean section but no previous vaginal births were classified as nulliparae (n = 209, 10.0 %). In Germany, there are only few differences in the management of care for vaginal births after cesarean section (VBAC), for instance misoprostol is not used for induction in VBAC women if there are no further complicating factors in their history [8]. For each birth, the attending care provider determined the onset of labor, which they defined on the basis of regular or irregular contractions in association with cervical dilation. Timing was documented in minutes. For our analysis, we modeled the sequence of the three intrapartum interventions—amniotomy, oxytocin augmentation and epidural analgesia—according to the first intrapartum application of each. The sequence of the three interventions was modeled by dividing the birth process into intervals. This resulted in intervals that represented the time between onset of labor and the first intervention, between interventions, and from the last intervention until birth (Fig. 1a–c). Interventions which were used before onset of labor, for example oxytocin as induction, were not included. When these women received oxytocin during labor, the first intrapartum application was documented as the intervention of interest.

Fig. 1
figure 1figure 1

The sequence of interventions and duration of intrapartum intervals. a Oxytocin as the first intervention, b epidural analgesia as the first intervention, c amniotomy as the first intervention. CS cesarean section, VO vaginal operative, NA not analyzed due to low number of cases, h hours, md median, *duration of the intervals was compared between nulliparae and multiparae by log-rank test, P < 0.05. These figures present how many nulliparae (n = 1,525) and multiparae (n = 1,117) with at least one intervention received which intervention first and in median how many hours after the onset of labor. Following the first intervention the possible pathways in the intervention cascade with frequencies and median duration between the two subsequent interventions are displayed. The interventions were considered as competing risks in a Kaplan–Meier’s estimate. For the duration from one intervention until spontaneous birth operative deliveries were treated as censored

In several cases, more than one intervention was documented as having occurred at the same time (epidural analgesia and amniotomy: np = 2, mp = 1; amniotomy and start of oxytocin augmentation: np = 22, mp = 16; epidural analgesia and start of oxytocin augmentation: np = 16, mp = 2). We held an expert discussion with ten professionals to address this issue, as a woman can only receive one intervention at a time. It was agreed that epidural analgesia can be considered as the first intervention when it was documented at the same time as either oxytocin or amniotomy. Because women need to remain still during epidural analgesia administration, health care providers would be more likely to promote contractions afterward. When amniotomy and oxytocin were documented at the same time, amniotomy was considered as an intervention administered before oxytocin was applied. We added 0.02 h (1 min = 0.0167 h) to the later intervention for the purposes of our analysis. Importantly, this minimal adjustment does not change the value of the median duration of an interval. Therefore, only the sequences of interventions were significantly affected by this adjustment.

An intervention at the same time as the onset of labor was resolved in the same way (oxytocin: np: n = 14, mp: n = 19; epidural: np: n = 2, mp: n = 0; amniotomy: np: n = 4, mp: n = 23). In cases where an intervention and birth were documented as having occurred at the same time (np: n = 46, mp: n = 18)—mostly amniotomy at cesarean section—we adjusted the intervention to 0.02 h prior to birth.

We calculated medians of the intervals using Kaplan–Meier’s estimate. Each subsequent intervention was analyzed as a competing risk [12, 15], excluding women without further interventions. To calculate the median duration from the onset of labor or an intervention until birth, we conducted separate analyses where operative deliveries were treated as censored data (Tables 1, 3). Furthermore, we investigated the last documented cervical dilation prior to an intervention with Kaplan–Meier’s estimate without censoring (Tables 1, 2). We used the log-rank test to determine if there were significant differences in cervical dilations or the duration of intervals between events for nulliparae compared to multiparae. We also used the log-rank test to determine if there were significant differences in cervical dilation or labor duration for the various intervention sequences. P < 0.05 was regarded as significant.

Table 1 Baseline characteristics

For statistical analysis we used transition data analysis (TDA, http://www.stat.ruhr-uni-bochum.de/tda.html), software specialized for longitudinal analyses.

Results

Baseline characteristics are presented in Table 1. Oxytocin augmentation was the most frequent intervention in nulliparae (n = 1,095, 52.6 %), followed by amniotomy (n = 718, 34.5 %) and epidural analgesia (n = 711, 34.2 %; Table 1). The most frequent intervention in multiparae was amniotomy (n = 783, 41.8 %) followed by oxytocin augmentation (n = 505, 27.0 %). Epidural analgesia was administered to 12.1 % of multiparae (n = 226; Table 1).

Of the total 3,955 participants, 1,525 nulliparae (73.2 %) and 1,117 multiparae (59.6 %) experienced at least one intervention. The intervals from onset of labor until the first intervention and from the first until the second intervention were significantly shorter in multiparae than in nulliparae (P < 0.01). The duration from the second until the third intervention did not differ between nulliparae and multiparae (P = 0.48; Table 1).

The sequences of interventions are presented in Fig. 1a–c. In nulliparae, the first intervention was most often epidural analgesia (n = 579, 27.8 %); (Fig. 1b). It was administered in median 9.56 h after onset of labor (Fig. 1b). In 59.2 % of these births, epidural analgesia was followed by oxytocin augmentation, after a median of 1.57 h. In 20.6 % of these births, epidural analgesia was followed by amniotomy, after a median of 4.39 h. Other nulliparae received oxytocin (n = 512, 24.6 %) or amniotomy (n = 434, 20.9 %) as the first intervention. Nulliparae who received oxytocin or amniotomy as first interventions most often experienced spontaneous birth and not a second intervention afterward (Fig. 1a, c).

In multiparae, amniotomy (n = 629, 33.6 %) was the most commonly used first intervention, in a median of 4.92 h after the onset of labor (Fig. 1c). Following an amniotomy, 80.0 % of multiparae experienced spontaneous birth in median 0.42 h. Few multiparae experienced an epidural as the first intervention (n = 189, 10.1 %; Fig. 1b). When they did, it was more often followed by another intervention than not. For multiparae, oxytocin followed in median 1.24 h (n = 83, 43.9 %) and amniotomy followed in median 2.01 h (n = 42, 22.2 %).

Measured by time after onset of labor, epidural analgesia was the only first intervention to be administered later in multiparae than in nulliparae (P < 0.01). However, median cervical dilation prior to epidural analgesia did not differ significantly (np: 2.8 cm; mp: 2.6 cm; P = 0.16; Table 2). Median cervical dilation prior to epidural analgesia slightly increased when administered as a second or third intervention, i.e. later in the cascade (Table 2). However, we found reverse tendencies for oxytocin and amniotomy. Oxytocin was administered at the lowest median cervical dilation when it followed epidural analgesia (Table 2). Oxytocin as a first intervention was mainly administered either between 2 and 4 cm cervical dilation or during the second stage of labor (np: n = 176, 35.1 %; mp: n = 48, 16.6 %; data not shown in table). Amniotomy as the first intervention was mostly administered at an advanced cervical dilation (median cervical dilation: np: 7.1 cm, mp: 7.4 cm). In women who received three interventions (np: n = 206, mp: n = 62), all three of them were most often administered during the first stage of labor (np: n = 169, mp: n = 47; data not shown in table). In these cases, epidural analgesia was most often the first intervention in the cascade (Table 2).

Table 2 Median cervical dilation before interventions in sequence

The number of spontaneous births decreased with increasing numbers of interventions. Nulliparae and multiparae without any of the three analyzed interventions experienced a spontaneous birth in 85.3 and 97.4 % of cases, respectively. Of those who received one intervention, 73.1 % of nulliparae and 94.6 % of multiparae delivered spontaneously, compared to 56.3 % of nulliparae and 80.7 % of multiparae who received three interventions (data not shown in table). Among women with one or two intrapartum interventions, those who received solely epidural analgesia and those who received it in combination with oxytocin were least likely to have a spontaneous birth. Second to the women who received no interventions, the highest incidences of spontaneous birth were found in women who received solely an amniotomy or an amniotomy and oxytocin (Fig. 2a, b).

Fig. 2
figure 2

Ratio of nulliparae and multiparae with a spontaneous birth depending on the number and sequence of interventions. EA epidural analgesia. a Spontaneous births (%) in nulliparae (n = 2,082). Proportion of spontaneous deliveries of nulliparae (n = 2,082) with the following sequence of interventions: no intervention, n = 558; oxytocin, n = 360; epidural analgesia, n = 118; amniotomy: n = 254; oxytocin–epidural analgesia: n = 51; oxytocin–amniotomy: n = 70; epidural analgesia–oxytocin; n = 277; epidural analgesia–amniotomy: n = 52; amniotomy–oxytocin: n = 129; amniotomy–epidural analgesia: n = 7; oxytocin–epidural analgesia–amniotomy: n = 20; oxytocin–amniotomy–epidural analgesia: n = 9; epidural analgesia–oxytocin–amniotomy; n = 66; epidural analgesia–amniotomy–oxytocin: n = 67; amniotomy–oxytocin–epidural analgesia: n = 11; amniotomy–epidural analgesia–oxytocin: n = 33. b Spontaneous births (%) in multiparae (n = 1,873). Proportion of spontaneous deliveries of multiparae (n = 1,873) with the following sequence of interventions: no intervention: n = 756; oxytocin: n = 203; epidural analgesia; n = 64; amniotomy: n = 515; oxytocin–epidural analgesia: n = 12; oxytocin–amniotomy: n = 80; epidural analgesia–oxytocin: n = 55; epidural analgesia–amniotomy: n = 25; amniotomy–oxytocin: n = 93; amniotomy–epidural analgesia: n = 8; oxytocin–epidural analgesia–amniotomy: n = 2; oxytocin–amniotomy–epidural analgesia: n = 2; epidural analgesia–oxytocin–amniotomy; n = 28; epidural analgesia–amniotomy–oxytocin: n = 17; amniotomy–oxytocin–epidural analgesia: n = 6; amniotomy–epidural analgesia–oxytocin: n = 7. The relative frequency for a spontaneous birth for women without any intervention was 85.1 % of 558 nulliparae and 97.4 % of 756 multiparae. The higher the number of interventions the lower was the mean frequency of spontaneous birth. Taking all women with three interventions without stratifying the sequence of interventions into account, 56.3 % of 206 nulliparae and 80.7 % of 62 multiparae with three interventions delivered spontaneously

Regarding the durations from the last intervention until spontaneous birth, there were increases in the median durations with the number of interventions. They increased from 1.52 h in nulliparae and 0.61 h in multiparae when only one intervention was administered, to 3.32 h in nulliparae and 1.52 h in multiparae when three interventions were administered (data not shown in table).

Overall, the number of interventions increased with labor duration (Table 3). Second to women with no interventions, the shortest labors were found in women who only experienced an amniotomy (np: 6.78 h, mp: 4.01 h). Few significant differences were found between median labor durations in women with different combinations and sequences of interventions (Table 3). When amniotomy followed epidural analgesia, labor duration was significantly shorter in comparison to oxytocin following epidural analgesia (np: P = 0.04; mp: P < 0.01). Nulliparae who received first oxytocin and then epidural analgesia experienced a longer labor duration (15.35 h) than women with the reverse sequence (11.71 h; P = 0.01). When women received three interventions, labor durations did not differ significantly between the different sequences of interventions (Table 3).

Table 3 Median labor duration stratified by interventions in sequence

Discussion

The temporal sequence of intrapartum interventions varied in association with parity, labor duration and mode of birth. An increase in interventions was observed in relation to longer labor durations and more operative deliveries.

Epidural analgesia was the intervention performed most often and earliest in a nullipara’s labor. In both nulliparae and multiparae, epidural analgesia was most often administered during early labor and was most often followed by another intervention, mainly by oxytocin. Women with epidural analgesia experienced significantly longer labors when oxytocin was the subsequent intervention compared to amniotomy as the second intervention.

In multiparae, amniotomy was most often the first intervention and most often performed in advanced labor. Few nulliparae and multiparae with an amniotomy experienced further interventions or operative deliveries.

Oxytocin as the first intervention was often administered during the second stage of labor. The intervals from epidural analgesia or amniotomy until oxytocin augmentation were some of the smallest intervals between interventions in our analysis. The median cervical dilation before oxytocin augmentation tended to be lower when this intervention was further along in the cascade.

Limitations

We chose a low-risk sample for our study which may not be representative for all women in labor. Using prospective information, data from perinatal records as well as hospitals with different participation rates may have resulted in selection bias. We tried to minimize this bias by using the same documentation tool for all births. One may criticize our classification of women with one previous cesarean section as nulliparae. We assumed that women with no previous vaginal birth would have labors more similar to nulliparae than to multiparae. Induced labors may attract more interventions than spontaneous labor. Midwives may have altered in their definition of the onset of labor in induced labor. The hospitals may have used differing policies for administering epidural analgesia and oxytocin, which we did not consider in this analysis and may have led to differing intervention rates and birth outcomes. Another weakness of the study may be seen in the definition of onset of labor. Despite this, the start of labor seemed the appropriate beginning point in our analysis for this study, as we mainly focused on the sequence of interventions and the intervals between the interventions.

Sequence of interventions

Thus far, a few studies have described cascades of interventions, and these did so using predefined sequences [18, 23]. Tracy et al. [23] and Roberts et al. [18] analyzed the cascade of oxytocin induction and augmentation (predefined at onset of labor), epidural analgesia (during the course of labor), episiotomy (just before birth) and operative births (at time of birth). The administration of oxytocin alone, epidural analgesia alone, and their combination were found to be associated with higher rates of episiotomies and operative births. Lacking data on timing of interventions in their population, they could not differentiate the sequences of the intrapartum interventions: oxytocin was always considered the first intervention in comparison to epidural analgesia [18, 23].

One study [7] analyzed the sequence of epidural analgesia and oxytocin augmentation. Dickinson et al. [7] found that oxytocin was most often administered before epidural analgesia (63.3 %) and that labor duration was slightly longer in women who received oxytocin before epidural analgesia (P > 0.05). According to our results, labor duration was significantly longer in women with oxytocin as first and epidural analgesia as second intervention, compared to the reverse sequence. One may hypothesize that oxytocin caused a request for epidural analgesia because of increased pain. We can also view oxytocin as an indicator for abnormal labor, particularly in combination with epidural analgesia [7].

Cascade of interventions

We found few results from randomized trials that tested the influence of interventions on other interventions. Studies have shown that epidural analgesia can increase the need for oxytocin augmentation [1, 10]. In our analysis, the interval from epidural analgesia to oxytocin augmentation was one of the shortest inter-intervention intervals. Epidural analgesia was most often the first intervention in the cascade when all three interventions were applied. This may be due to the fact that obstetricians consider epidural analgesia to prolong labor and reduce plasma oxytocin levels [1, 10, 17], and therefore administered an amniotomy and/or oxytocin augmentation. However, we found no evidence that oxytocin administration after epidural analgesia may reduce operative deliveries or improve fetal outcome [6].

We found no current evidence in the literature that oxytocin or amniotomy alone increase the request for epidural analgesia [3, 22, 24]. However, randomized trials of labors with early oxytocin and amniotomy application showed an increased and earlier application of epidural analgesia [25]. We found no randomized trials which demonstrated a reduction of oxytocin use in patients who had received an amniotomy [22]. According to our results, amniotomy as the sole intervention does not seem to provoke further interventions. It is also likely that amniotomies were mostly administered to women with uncomplicated labors.

Overall, the incidence of oxytocin use in our study was very high and oxytocin was often used as the first intervention or subsequent to epidural analgesia. Oxytocin has been shown to be effective in reducing mean labor duration when an antecedent application of epidural analgesia was not considered [3]. One may infer that oxytocin was not always applied appropriately in our study population, which has been also concluded in previous observational studies [20]. Compared to a late application, the early use of oxytocin has been found to cause higher rates of uterine hyperstimulation [3, 24]. This is of concern because hyperstimulation (≥5 contractions in 10 min) causes decreases in fetal oxygen saturation and umbilical artery pH at birth [2, 21]. When contemplating the relative safety of using oxytocin augmentation, obstetricians and midwives should be attentive to the suggestions from current research, such as using checklist protocols [4, 5].

Overall, we found interesting dynamics of labor related to the sequence of interventions. Still, we could not differentiate between interventions that caused poorer labor outcomes and interventions that were administered due to complicated labors. Observational studies are not commonly considered able to clarify cause–effect relationships [11]. This is due to the effect of selection bias, confounding by indication and time-dependent confounding [13]. However, the complex methods of time-to-event analysis may offer the researcher the ability to address the problem of unmeasured confounding [13]. Practitioners can use descriptive information about the sequence of interventions to understand the possible relationships between interventions and the underlying process of labor. As many interventions are administered in varying sequences and combinations, information from randomized trials about the optimal use of interventions in relation to each other is vital to best practice.

This study’s main value is the detailed description of the cascade of three intrapartum interventions, their association with cervical dilation, and labor outcome. While epidural analgesia was most often associated with further interventions during the first stage, women with an amniotomy had a favorable outcome in terms of labor duration and delivery mode. The high incidence of oxytocin augmentation and the short inter-intervention intervals preceding oxytocin found in our study both support the increased attention in literature questioning oxytocin’s risks, as well as continued investigation into this drug. Midwives and obstetricians should think twice whether an intervention is justified in order to reduce interventions during normal labour. Our study results may provide a platform for further research on the best timing and sequence of indicated intrapartum interventions for mothers and their babies.