Duration of the active first stage of labour and severe perineal lacerations and maternal postpartum complications: a population‐based cohort study

The impact of first stage labour duration on maternal outcomes is sparsely investigated. We aimed to study the association between a longer active first stage and maternal complications in the early postpartum period.


| I N TRODUC TION
Maternal postpartum morbidities can have both short-and long-term effects in a woman's life.One major complication of childbirth is infection, including postpartum endometritis and wound infections, 1,2 potentially leading to life-threatening maternal sepsis. 3Urinary retention with bladder overdistention may damage the detrusor muscle and cause persisting voiding difficulties. 4,5Severe perineal lacerations with anal sphincter injury can lead to anal incontinence, persistent perineal pain and dyspareunia. 6,7 longer second stage labour duration has been associated with increased maternal morbidity in numerous studies.][10][11][12][13] The effect of active first stage duration is less investigated, which is of concern as labour is a continuous process and its first and second stages are closely interrelated. 14,150][21] It is still unclear if the association between labour duration and adverse maternal outcomes primarily relates to the duration of the second stage, the duration of the active first stage, or both, as few studies have included both stages.
In the last decade, large cohort studies in contemporary populations have redefined the patterns of normal active first stage progression.3][24][25] Based on these findings, the World Health Organization (WHO) has changed the definition for the onset of the active first stage of labour to a cervical dilation of 5 cm, 26 and worldwide, clinical guidelines have shifted, allowing more time for the active first stage.
In light of this knowledge, the association between active first stage duration and maternal complications needs to be investigated further in population-based and contemporary cohorts of sufficient size, taking second stage duration into account.The aim of this study was to explore the association between active first stage duration distribution, instead of preset duration thresholds, and maternal complications in the early postpartum period.

| Study design and data sources
For this population-based cohort study, data on pregnancies and deliveries were obtained from the Stockholm-Gotland Perinatal Cohort. 27The cohort consists of prospectively collected data automatically retrieved from the electronic medical record system Obstetrix (Cerner, Kansas City, MO, USA), covering all births in the Stockholm and Gotland regions.About one-quarter (approx.28 000) of annual births in Sweden from 1 January 2008 through to 15 June 2020 are included.The database contains information from antenatal care, including maternal medical and reproductive history, pregnancy-related information, labour details, and maternal and neonatal outcomes in the postpartum period until discharge from the postnatal care unit.For each delivery, partograph data, including granular longitudinal data on cervical dilation, temperature, blood pressure, labour management, interventions and mode of delivery, are also retrieved.Cervical examinations during labour are routinely performed by the attending midwife and are entered into the partograph.The database further includes diagnoses of diseases and conditions according to the Swedish version of the International Classification of Diseases, tenth revision (ICD-10).

| Study population and exposure
Women with a singleton pregnancy and an infant in vertex presentation, with delivery at term (≥37 +0 weeks of gestation), who had a spontaneous onset or induction of labour were included.Exclusion criteria were: missing information for maternal or infant personal identification number; prenatal stillbirth; deliveries without partograph data; prelabour CD or CD before cervical dilation of 10 cm; missing time point of full dilation (10 cm or retracted cervix); and fewer than two records of cervical dilation between 3 and 10 cm, or first record of cervical dilation at ≥8 cm.Outliers with unrealistic values of active first stage duration (negative or >25 h) or second stage duration (negative or >500 min) were excluded (Figure 1).
The exposure was defined as the duration of the active first stage of labour.The duration of the latent phase of labour was not considered, and consequently was not included in the analysis.Active first stage duration was defined according to the WHO definition, as the time interval between the first recording of 5 cm cervical dilation and the first time point of full cervical dilation. 26The recording of 5 cm dilation was missing in the partograph for a substantial section of the source population.To minimise the potential risk of selection bias and improve the precision, the missing time for cervical dilation of 5 cm was estimated and imputed (Figure S1).By using the mean cervical dilation pace from the study by Lundborg et al., 25 12 different imputation patterns with at least two records from 3 to 10 cm were hierarchically used for imputation of the missing time point for dilation of 5 cm (Figure S1). 28he source population included all women who met the inclusion criteria (Figure 1).The study population included women with a recorded time point of dilation of 5 cm together with women with an imputed time point of dilation of 5 cm, and this group was used for the main analyses (Figure 1).These populations were stratified into parity groups: (i) primiparous; (ii) parous with no previous CD; and (iii) parous with previous CD, respectively, owing to clinical disparities in both labour duration and the risk of complications.Active first stage duration was categorised into percentile categories (<50th, 50th to <75th, 75th to <90th and ≥90th percentile).These percentile categories were derived from the distribution of the active first stage duration overall, and from the distribution of the durations of the three stratified parity groups, respectively.

F I G U R E 1
Flow chart of the study population and women with recorded time points for cervical dilation of 5 cm. 1 Source population includes all women meeting the study criteria (used for sensitivity analysis). 2Women with a recorded time point for cervical dilation of 5 cm (used for sensitivity analyses). 3Study population includes women with a recorded time point for cervical dilation of 5 cm and women with an imputed time point for cervical dilation of 5 cm (used for main analyses).

| Outcomes
The outcomes were diagnoses of maternal complications in the early postpartum period, defined as occurring during the delivery hospitalisation, recorded by ICD-10 codes at discharge from the delivery hospital.Outcomes included: (i) severe perineal lacerations, defined as third-or fourth-degree tears; (ii) postpartum infections, including endometritis, fever after delivery, urinary tract infection, wound infection in perineal sutures or caesarean surgical site, or other specified infections in the urinary or genital tract after delivery; (iii) postpartum urinary retention, defined as ≥1000 mL at catheterisation or >300 mL postvoid residual volume; and (iv) haematoma in birth canal or ruptured perineal sutures.The corresponding ICD-10 codes are presented in Table S1.

| Covariates
Baseline covariates of interest were: maternal body mass index (BMI, kg/m 2 ), measured at the first antenatal care visit; smoking in early or late pregnancy; maternal age at delivery (years); diagnosis of pre-eclampsia, pre-gestational or gestational hypertension, pre-gestational diabetes (i.e.diabetes mellitus type 1 or 2) or gestational diabetes; gestational duration at birth (completed weeks and days); epidural analgesia; oxytocin augmentation (initiation before active first stage, during active first or during second stage of labour); mode of delivery (spontaneous vaginal, operative vaginal or CD); birthweight (small for gestational age, SGA, ≤2 SD; appropriate for gestational age, AGA, from −2 to ≤2 SD; and large for gestational age, LGA, >2 SD).

| Statistical analyses
The distribution of maternal, pregnancy and delivery characteristics, and the incidence of each postpartum outcome, was calculated overall and by percentile category of active first stage duration (<50th, 50th to <75th, 75th to <90th and ≥90th percentile).Crude and adjusted relative risk (cRR and aRR, respectively) with 95% confidence interval (95% CIs) were estimated by Poisson regression analysis with a robust error variance, 29 with active first stage duration <50th percentile as the reference category.
Informed by previous knowledge, directed acyclic graphs (DAGs, Figure S3) were created for each outcome separately, seeking to identify potential confounding factors and mediating factors.A mediating factor is a variable in the causal pathway between the exposure and the outcome of interest, as opposed to a confounding factor, which affects both exposure and outcome (visualised in Figure S2).Potential confounding factors were included in the multivariable analyses.For postpartum infections, results were adjusted for BMI, maternal age, induction of labour and diabetes.For urinary retention we adjusted for maternal age, gestational duration and epidural analgesia initiated before the active first stage.An epidural during the active first stage was neither considered a confounding factor (as it occurs simultaneously as the exposure) nor a mediating factor (as it is initiated during the active first stage when its duration is not yet known).For severe perineal lacerations, haematoma and ruptured sutures, the adjusted models included maternal age, BMI and gestational duration.
Mediation analysis was performed to investigate the effect of second stage duration on the association between active first stage duration and maternal outcomes.The second stage was considered a mediating factor in the pathway between active first stage duration and maternal complications, as a longer active first stage duration is associated with a longer second stage duration, 14,15 along with increased risk of maternal postpartum complications. 9Mediation analysis is based on a counterfactual framework and is described elsewhere. 30,31sing this analytical framework, we aimed to separate the total effect into the direct effect (i.e. the effect of active first stage duration on maternal complications) and the indirect effect (i.e. the effect mediated through a longer second stage).
The proportion of the total effect mediated by second stage duration was estimated as the proportion mediated on the risk-difference scale.For the mediation analysis, active first stage duration was modelled as a binary variable with the 75th percentile duration used as the cut-off.Second stage duration was used as a continuous variable on the log scale.The 95% CIs were identified by bootstrap bias-corrected confidence intervals.Confounding factors of the relationships between active first stage duration and each separate outcome, between active first stage duration and second stage duration, and between second stage duration and each outcome were adjusted for accordingly.
A sensitivity analysis was performed to compare the results of the study population (i.e.including women with either recorded or imputed time points for cervical dilation of 5 cm) with the results of women with only recorded time points for cervical dilation of 5 cm.Baseline characteristics and the overall absolute risks of the outcomes were further investigated in the source population (i.e.all women meeting the inclusion criteria, even if they had no recorded or imputed time point for cervical dilation of 5 cm) as a secondary sensitivity analysis.This was done to examine the comparability of the study population, the source population and the population of women with recorded time points for cervical dilation of 5 cm, to validate the imputation technique.

| R E SU LTS
A total of 159 459 deliveries were included in the study population.Maternal, pregnancy and delivery characteristics by active first stage duration percentile categories are described in Table 1.Primiparous women and women with advanced gestational duration were over-represented among deliveries with active first stage duration >75th percentile.With increasing duration, women were less likely to experience a spontaneous vaginal birth.Oxytocin augmentation initiated during the active first stage was more common in women  with a longer active first stage, as was the use of epidural analgesia during labour (Table 1).The overall median active first stage duration was 3.9 h, and the 90th percentile duration was 9.7 h.The active first stage median durations were 5.2, 2.4 and 4.2 h for primiparous women, parous women with no previous CD and parous women with previous CD, respectively, with 90th percentile durations of 11.0, 6.5 and 9.7 h, respectively.The duration for each percentile category in the study population overall is described in Table 1, and for the different parity groups in Figure 2.
Severe perineal lacerations affected 4.7% of women in the study population, postpartum infection affected 2.4%, urinary retention affected 2.5% and haematoma in the birth canal or ruptured sutures affected 0.7%.The risks of severe perineal laceration, postpartum infection and urinary retention increased with a longer active first stage duration, but not for haematoma in the birth canal or ruptured sutures (Table S2).
Compared with the <50th percentile, the aRR of adverse maternal outcomes was augmented with increasing active first stage duration in all parity groups (Figure 2).The aRR of postpartum infection gradually increased with a longer active first stage in all parity groups.With an active first stage of >90th percentile, the aRR for postpartum infection was 1.64 (95% CI 1.46-1.84) in primiparous women, 2.43 (95% CI 1.98-2.98) in parous women with no previous CD, and 2.33 (95% CI 1.65-3.28) in parous women with previous CD (Figure 2).
For perineal laceration and urinary retention, the aRR increased in women with an active first stage duration of >50th percentile in all parity groups.The aRR for urinary retention was 1.77 (95% CI 1.58-1.98) in primiparous women, 4.09 (95% CI 3.29-5.09)in parous women with no previous CD and 1.51 (95% CI 1.07-2.12)for parous women with previous CD, with an active first stage duration of >90th percentile (Figure 2).
For the association between active first stage duration and severe perineal laceration, 34.9% (95% CI 25.5-51.7%)was mediated by second stage duration in primiparous women (Table 2).The corresponding results for postpartum infection and urinary retention were 33.4% (95% CI 25.7-45.0%)and 36.9%(95% CI 29.0-48.5%),respectively.The proportion mediated by second stage duration was generally higher for parous women with no previous CD than for primiparous women.The proportion mediated by second stage duration for severe perineal laceration was 61.7% (95% CI 50.1-79.1%),for postpartum infection was 47.7% (95% CI 36.9-63.9%)and for urinary retention was 48.1% (95% CI 39.1-62.4%) in parous women with no previous CD (Table S3).Mediation analysis for parous women with previous CD was not possible to calculate because of a lack of power.

| Sensitivity analysis
In total, 72 701 (45.6%) deliveries in the study population had a record for cervical dilation of 5 cm, and 86 758 (54.4%) deliveries had an imputed time point for the start of the active first stage.When comparing women with recorded time points for cervical dilation of 5 cm with the study population in a sensitivity analysis, baseline characteristics were comparable.The difference in absolute risks between the populations were 0.2% for severe perineal laceration and  postpartum infections, 0.3% for urinary retention, and no difference for haematoma or ruptured sutures (Tables S4  and S5).Moreover, the RRs and overall trends were similar between the two populations.The source population, also containing women with no known or imputed time point for cervical dilation of 5 cm, included a larger proportion of parous women with no previous CD.The absolute risks of the outcomes differed by ≤0.4% between the study population and the source population (Table S4).Importantly, when comparing the characteristics and overall risks, the study population more closely resembled the source population than the population including only women with a recorded time point for cervical dilation of 5 cm did.Additional stratified sensitivity analysis confirmed that a longer active first stage duration was associated with an increased risk of severe perineal lacerations, postpartum infection and urinary retention in both women with spontaneous onset and women with induction of labour (data not shown).

| Main findings
In this large cohort study, a longer active first stage labour duration was found to be associated with an increased risk of severe perineal laceration, infection and urinary retention in the early postpartum period.Both the absolute and the relative risks increased with a longer active first stage duration, already from the 50th to the 75th percentile.In primiparous women, about one-third of the effect was mediated by second stage duration, and half of the association was mediated by second stage duration in parous women with no previous CD.Generally, parous women with previous CD had the highest absolute risk of all outcomes, followed by primiparous women.Even so, the highest relative risks for all outcomes were seen in parous women with no previous CD.There was no association between a longer active first stage duration and haematoma in the birth canal or ruptured sutures in any parity group.

| Strengths and limitations
The strengths of this study include the population-based design of the study and the data collected for deliveries with contemporary obstetrical management.The granular prospectively collected and automatically retrieved data on pregnancy, delivery and postpartum outcomes is a great advantage.The large study population enabled the analyses of higher percentiles of active first stage duration and the investigation of each outcome separately.Imputation of the onset of the active first stage allowed a larger proportion of pregnancies to be included in the study population, reducing the risk of selection bias and misclassification.The method was only used for women who already had information recorded for cervical dilations, to warrant reliable results for an estimated time point of cervical dilation of 5 cm.Parous women, women with previous CD and women with an induction of labour were all included to increase external validity.Mediation analysis, offering insights on the proportion of effects mediated by second stage duration, further improved the quality of the results.
Still, the study has some limitations.For example, even though imputation techniques were used, a number of births were not possible to include because of missing cervical records.However, sensitivity analysis demonstrated that the imputation methods used were valid, and that the study population reflected the source population better than the population including only women with a recorded time point for cervical dilation of 5 cm.As women with CD in the first stage were not included, the results are only generalisable for women reaching the second stage.Further, as parous women are generally discharged earlier than nulliparous women, the absolute risk of infection in parous women could potentially be underestimated in our data, but this would not affect the relative risks or introduce information bias.
There may be residual confounding and remaining mediating factors not taken into consideration.Mediation analysis was performed for the most prominent mediating factor, second stage duration, but the study design did not allow us to disentangle the role of mode of delivery.For example, among deliveries with the longest active first stage duration, some women go on to a CD in the second stage, and therefore no longer be at risk of a severe perineal laceration.This might explain why primiparous and parous women with previous CD were not found to have a higher risk for lacerations in the 90th percentile, compared with the 75th to <90th percentile category.We chose not to restrict by mode of delivery to avoid the introduction of bias.

| Interpretation
0][21] In this study, the outcomes were analysed separately, the results were stratified by parity group and the most central mediating factor (second stage) was accounted for.
A longer first stage duration has been associated with postpartum infection as part of a composite outcome, [16][17][18] but not independently. 16Cheng et al. found a higher risk of endometritis in primiparous women with a prolonged first stage, but after multivariable adjustment including second stage duration, the association was no longer significant. 20horioamnionitis was not included in our study as it occurs before or during labour, and is therefore not diagnosed after the exposure.Previous studies have investigated a mixed outcome of chorioamnionitis and endometritis, 17 or have not specified the timing of maternal fever. 18,21A study by Kawakita et al. showed no increased risk of endometritis for primiparous women with an arrest of labour of ≥4 h diagnosed at a cervical dilation of 6-7 cm, but increased unadjusted risk of endometritis with arrest at a cervical dilation of 8-9 cm. 32][18][19] When studied as an individual outcome, three studies found no increased risk with prolonged first stage, 17,19,20 but Blankenship et al. found a higher risk of severe lacerations in labours with first stage duration of ≥90th percentile. 16In the current study, increased risk was seen already from the 50th to the 75th percentile.
To our knowledge, none of the previous studies have investigated the relationship with urinary retention.Stephansson et al. found an association between longer second stage duration and urinary retention. 9The current study also supports an association with first stage duration, and mediation analysis suggests that this relationship was not fully explained by the effect of second stage duration.
The greater sample size in this study may be one reason for diverging results compared with previous studies.Different definitions for the onset of the first stage and arbitrary thresholds for defining longer durations may also explain disparities.For example, Wang et al. studied the total first stage, including the latent phase, 19 and Cheng et al. used a definition based on the time point reported for regular painful contractions combined with documented cervical dilation. 20The WHO definition for the onset of the active first stage, starting at a cervical dilation of 5 cm, and the distribution of the active first stage duration was used in the current study.
There is a need for future studies focusing on understanding why the duration of labour is associated with maternal complications.Physiological mechanisms behind increased risks may differ between the outcomes included in this study.The duration of labour might be a proxy for actions that occur during that time, for example cervical examinations, emptying of the bladder, and medical interventions, including the use of oxytocin.The effects on associated risks need to be disentangled further to appreciate the effects of the actual duration of labour versus any actions performed within that time.Risk factors or predictors identifying women at risk when labour is prolonged also need further investigation.In our study, adjustment for BMI, maternal age and gestational duration did not significantly alter the results.
Based on the current and other studies, it is not possible to agree on a clear threshold for when the active first stage is too long with regards to maternal complications.Labour is a continuous process where both the active first and second stages have implications for risks of adverse outcomes.The mediation analysis showed that some, but not all, of the risks associated with increased active first stage duration were mediated by second stage duration.Clinicians should keep these population-level risks in mind when either the active first stage or the second stage is prolonged, and prioritise preventive strategies and pay attention to the symptoms of these outcomes.

| CONCLUSION
A longer active first stage duration is associated with increased risks of severe perineal lacerations and maternal complications in the early postpartum period.The second stage duration mediates some but not all of these risks.Active first stage duration should be considered during the labour process, along with other risk factors in obstetric clinical management.

AU T HOR C ON T R I BU T ION S
AS and LK planned and designed the study.LK coded the data analyses under the supervision of AS.XL contributed to coding and examined the data analyses.LL and LVL contributed with methodological expertise.LK and AS interpreted the results, with input from all co-authors.LK wrote the first draft of the article, with input from AS.All authors contributed to the revision of the article.AS obtained funding and ensured administrative support and supervision.All authors accept responsibility for the article as published.

T A B L E 1
Maternal, pregnancy and delivery characteristics, by percentile category of active first stage duration, for all women in the study population with a delivery in the Stockholm-Gotland Regions, 2008-2020, N = 159 495.

F I U R E 2
Absolute and adjusted relative risks of severe perineal laceration and postpartum complications in the early postpartum period in the study population, stratified by parity group and by distribution of active first stage duration in percentile categories.

women, N = 87 313 Mediation analysis second stage duration Estimate (95% CI)
Mediation analysis for primiparous women in the study population.Active first stage labour duration exposure was modelled as a binary variable with the 75th percentile (8.0 h) used as a cut-off.The causal mediating factor was second stage duration as a continuous variable.The 95% CIs were bootstrap biascorrected confidence intervals.Abbreviations: BMI, body mass index; NDE, natural direct effect; NIE, natural indirect effect; OR, odds ratio; PM, proportion mediated; TE, total effect.a Adjusted for maternal age, BMI, gestational duration and fetal occiput posterior.Adjusted for maternal age, gestational duration and epidural analgesia before or during first stage.
T A B L E 2Note: b Adjusted for BMI, maternal age, induction of labour and diabetes.c