- Top of page
Sexual intercourse is commonly believed to hasten labour. Semen contains prostaglandin E, breast stimulation has been shown to hasten the onset of labour, and coitus and orgasm stimulates uterine activity, thereby underpinning the expectation that sexual activity at term may expedite labour.
In a prospective diary-based study from our centre, coitus at term in healthy women is associated with a shortened gestation and fewer labour inductions, and there is a direct correlation between the frequency of coitus and expedited onset of labour, which provided the impetus for this trial.
A Cochrane review on sexual intercourse for cervical ripening and induction of labour identifies only a small trial with 28 women, from which no meaningful conclusions can be drawn. A recent systematic review on the methods of labour induction published in 2011 cites only the same Cochrane review, demonstrating the sparse literature currently available concerning the relationship between sexual intercourse and induction of labour.
The UK's National Institute of Clinical Excellence (NICE) antenatal care guideline states that ‘pregnant woman should be informed that sexual intercourse in pregnancy is not known to be associated with any adverse outcomes’, reflecting the accumulated literature on the safety of coitus in pregnancy. However, 20–80% of pregnant women may have safety concerns about sexual intercourse during pregnancy.
We postulate that couples can be reassured about the safety of coitus in late pregnancy and persuaded to engage in coitus as a natural method to initiate labour, so as to avoid labour induction or to hasten an anticipated birth. We anticipate that promoting coitus as a means of expediting labour will be most effective in women who have been recently abstinent, as this population starts from a zero base. We performed a randomised trial to evaluate the effect of advising vaginal intercourse to expedite labour amongst pregnant women in late pregnancy on their pregnancy duration and need for labour induction.
- Top of page
Ethical approval for the trial was obtained from the University of Malaya Medical Centre Medical Ethics Committee (approval number 547.3, dated 16 August 2006). Written consent was obtained from all participants. The study was conducted in accordance with the Declaration of Helsinki by the World Medical Association on experimentation in human subjects. The trial was registered in a public trial registry (registration number ISRCTN 82333699) before the start of enrolment.
The charts of women attending the antenatal clinic in our hospital were reviewed by a research nurse to ascertain whether they fulfilled the inclusion criteria of gestation of 35–38 weeks, with a straightforward pregnancy (i.e. viable, singleton and cephalic presentation, with no placenta praevia, antepartum haemorrhage, ruptured membranes, previous caesarean section, hypertension, diabetes, significant medical history, fetal growth restriction, fetal anomalies or previous stillbirth). Women at 35–36 weeks of gestation were approached preferentially to give the maximal opportunity for coitus after enrolment. Women who fulfilled the criteria were approached and given an information leaflet about the study.
At the initial contact stage with potential participants, the study was presented as an activity study on coitus in late pregnancy to minimise any influence on control group behaviour. The potential participants were then asked about their coital activity in the last 6 weeks. Women who said they had been coitally active or who declined to answer were excluded. We also excluded women who said they did not have a current male partner or who had medical contraindication to coitus (e.g. human immunodeficiency or hepatitis-B positive – our antenatal population was universally screened).
The methodology and intervention for this trial is adapted from and closely mirrored that of an earlier trial performed at our centre on advising coitus in women scheduled for a non-urgent induction of labour. We decided to proceed with this trial as the proposed study population and circumstances are widely divergent from those of the earlier trial, and a prospective study of coitus from 36 weeks of gestation onwards, also carried out at our centre with a similar study population and circumstances, has shown a strong correlation between coitus and the onset of labour. The groundwork and funding for this trial was in place before the non-significant result of the earlier trial was reported.
As 99% of the parturients at our centre are married, and the use of condoms by married couples for the purpose of preventing sexually transmitted disease in those not already known to be infected is believed to be rare in our setting, we did not advise on or obtain any information on condom use in this trial.
Eligible women who were willing to participate then had a one-to-one session with a practicing medically qualified member of the research team, who obtained their written consent and randomly allocated them to the intervention or control groups. The counsellor was clearly identified to the women as a medical doctor. The counsellors involved in providing the intervention were trained on the intervention protocol, which is as described below.
During counselling, women assigned to the advise-coitus arm were told that vaginal intercourse at term can be used as a natural method to safely expedite labour, and may hence reduce the need for induction of labour. They were further told that frequent intercourse can be more effective. They were advised to start vaginal intercourse after 36 weeks of gestation. Women who wanted more information after the standard counselling were consistently informed that vaginal intercourse at term is a natural, safe and effective method for initiating labour. For women allocated to the control group, interaction after randomisation was kept as short as possible, with the impression given that the women were participating in a diary-based activity study of coitus. The controls were told that sexual intercourse is safe but its effect on labour was uncertain. To keep our approach to the control arm as neutral as possible, controls that wanted additional information were referred to the common information leaflet given to all participants. The allocation of women to the advise-coitus or control arms was not revealed to providers. There was no further intervention. All participants subsequently received standard obstetric care.
The randomisation sequence was in random blocks of eight or 12, with a 1:1 ratio, generated using a computerised random number generator (www.random.org) by N.S. and P.C.T. Numbered opaque envelopes were prepared containing the allocation to either advise-coitus or control groups, which were opened in sequence to effect randomisation. An additional information leaflet was given to participants allocated to advise-coitus. This leaflet reinforced the information given in the advice or counselling session. All participants were given a simple diary sheet, instructed to regularly update their vaginal intercourse diary by circling days where vaginal intercourse had occurred and to return the diary by hand to postnatal ward staff or to use the stamped addressed envelope after delivery.
Prelabour rupture of membranes (PROM) was defined as confirmed liquor amnii leakage, as evidenced by amniotic fluid trickling through the cervix at speculum examination, or the observation of amniotic fluid discharging from the introitus in the absence of uterine contractions at the time of first occurrence. In the event of PROM, women were given the option of immediate labour induction or to wait up to 24 hours for spontaneous labour as an inpatient. If required for PROM, labour induction was typically performed with oxytocin infusion or vaginal dinoprostone. Labour induction was also offered for prolonged but otherwise uncomplicated pregnancy at 41 weeks of gestation, with either vaginal dinoprostone if the cervix was unfavourable or amniotomy followed by oxytocin infusion when the cervix was favourable (typically for a Bishop score > 6). We defined induction of labour (IOL) as the application of these standard techniques to procure labour. The standard management for labour induction and intrapartum care in our institution has been described previously.
Labour was deemed to be established when contractions were at least once every 4 minutes and the cervical dilation was at least 3 cm. Oxytocin given after established labour was considered to be for labour augmentation. The length of active labour was taken as the interval from the time of diagnosed established labour to the time of delivery.
The charts of participants were retrieved after their delivery for their hospitalisation, labour and delivery details to be extracted onto a standardised case report form. Women who did not submit their coital diaries after delivery or who delivered in other hospitals were contacted by telephone to obtain as much coitus and birth data as possible.
Primary outcomes were: (1) duration of pregnancy (as represented by gestational age at delivery and the interval between intervention and delivery); and (2) rate of induction of labour. Secondary outcomes included any non-birth related hospitalisation since randomisation, PROM, method of labour induction, indication for labour induction, length of labour, epidural analgesia use in labour, oxytocin augmentation, mode of delivery, indications for operative delivery, maternal hospital stay, Apgar score at 5 minutes, arterial cord blood pH, admission to the neonatal care unit and indication for neonatal admission.
Our sample size was calculated based on the following rationale: in an earlier report, 20/84 women (23.8%) who were abstinent from 36 weeks of gestation to birth required IOL for prolonged pregnancy, compared with 5/116 women (4.3%) who reported coitus during that time. In abstinent women allocated to the control group, we assume a background 5% coitus rate before delivery and their IOL rate would thus be [(0.238 × 0.95) + (0.043 × 0.05)] = 22.8%. We postulated that if one in three allocated to advise-coitus, could be persuaded towards coitus, then [0.05 + (0.95 × 0.333)] = 36.7% would be coitally active, and their IOL rate would be [(0.238 × 0.633) + (0.043 × 0.367)] = 16.6%. Applying IOL rates of 22.8 versus 16.6%, with α = 0.05, a power of 0.8 and a 1:1 recruitment ratio, and applying the chi-square test, 649 women in each arm would be required. Factoring in a 20% drop-out rate, a total of 1623 women would be required for the study.
Our public hospital delivers about 5000 women per year, of which about 4675 (93.5%) would have attended for prior antenatal care as booked cases. We expect 42% to be abstinent at 35–36 weeks of gestation. Factoring in a 20% exclusion rate because of criteria infringement, we anticipated 1570 suitable women to be available for enrolment each year. Assuming a 70% enrolment pick-up rate, 1099 women per year could be recruited. The trial enrolment target of 1623 participants was achievable in 18 months. The study was funded for 12 months by an internal research grant from the University of Malaya. Because of a far slower rate of enrollment than anticipated, funding was secured for another 12 months. After this funding was exhausted (after 40 months of enrollment), we decided to perform an analysis of the data with 1200 women enrolled to determine if the trial should be continued. The funder played no role in the trial design, conduct, data analysis or drafting of the article.
Data were entered into spss 17 (SPSS Inc., Chicago, IL, USA) for data analysis. We excluded those mistakenly recruited who did not satisfy the inclusion criteria. Analysis was performed by intention to treat. For comparison of means and distribution of continuous data we used the Student t test, for ordinal data or non-normally distributed data we used the Mann–Whitney U test, for 2 × 2 categorical data sets we used Fisher's exact test and for categorical data sets larger than 2 × 2 we used the chi-square test. Tests were two sided and P < 0.05 in any test was considered to be statistically significant.
- Top of page
Enrolment into the trial was from 28 March 2008 to 15 June 2011, and the last participant was discharged from hospital following admission for childbirth on 19 July 2011. The recruitment and in-trial flow chart is depicted in Figure 1. A total of 1200 women were enrolled onto the study, and 25 women were excluded because of study criteria infringements discovered after randomisation. Pregnancy outcomes were extracted from the women's charts, hospital central records or obtained through telephone contact if the woman did not deliver at our hospital. Pregnancy outcomes were unobtainable in another 25 women because they did not deliver at our hospital and could not be contacted by telephone. In a further 13 women, we could trace pregnancy outcomes from their charts but they did not return their coital diaries and they could not be further contacted by telephone to obtain their coitus data. Of the 1200 trial participants, 1086 (90.5%) went on to deliver at our hospital. Only 144/1200 (12.0%) of the participants returned their coitus diaries (four returned diaries contained no useful data); for the remainder, coitus data were obtained via telephone contact. However, post-hoc analysis showed that the proportion of women reporting ‘any coitus’ was similar in the returned diary and telephone contact groups: 80.7 versus 82.8% (P = 0.552).
The characteristics of participants analysed on an intention-to-treat basis (n = 1175) are shown in Table 1. There was no significant difference in any characteristic between the trial arms.
Table 1. Characteristics of trial participants, stratified according to treatment allocation (advise coitus or control)
| ||Advise coitus n = 585||Control n = 590|
| Age (years) ||29.1 ± 3.9||28.9 ± 3.6|
| Ethnicity || || |
|Malay||237 (40.5)||261 (44.2)|
|Chinese||178 (30.4)||187 (31.7)|
|Indian||141 (24.1)||118 (20.0)|
|Others||29 (5.0)||24 (4.1)|
| Gravidity ||1 (1–2)||1 (1–2)|
| Parity ||0 (0–1)||0 (0–1)|
| Nulliparous ||350 (59.8)||364 (61.7)|
| Gestation at recruitment (weeks) ||36.2 ± 0.8||36.2 ± 0.7|
Primary outcomes of pregnancy duration and labour induction rate, as well as the data of main outcomes such as reported coitus, are displayed in Table 2. Pregnancy duration, as represented by the gestational age at delivery of 39.4 ± 1.2 versus 39.5 ± 1.2 weeks (P = 0.112), the intervention to delivery interval of 3.2 ± 1.4 versus 3.3 ± 1.3 weeks (P = 0.417) and the labour induction rates of 126/574 (22.0%) versus 120/576 (20.8%) (RR 1.1, 95% CI 0.8–1.4, P = 0.666) did not differ across the advise-coitus versus control trial arms.
Table 2. Main outcomes after randomisation to advise-coitus or control groups
| ||Advise coitus n = 574||Control n = 576||RR (95% CI)|| P |
| Labour induction a ||126 (22.0)||120 (20.8)||1.1 (0.8–1.4)||0.666|
| Gestational age at delivery (weeks) ||39.4 ± 1.2||39.5 ± 1.2|| ||0.112|
| Intervention to delivery interval (weeks) ||3.2 ± 1.4||3.3 ± 1.3|| ||0.417|
| Coitus reported b ||481 (85.3)c||458 (79.9)c||1.5 (1.1–2.0)||0.019|
| Frequency of reported coital acts d ||3 (2–5)c||2 (1–4)c|| ||0.006|
Coitus 481/574 (85.3%) versus 458/576 (79.9%) (RR 1.5, 95% CI 1.1–2.0, P = 0.019) was more frequently reported and the coital frequency reported [median 3 (IQR 2–5) versus 2 (IQR 1–4), P = 0.006] was higher in the advise-coitus arm compared with the control arm. Although participants appeared to respond to the suggestion to use frequent coitus, presumably to expedite labour onset, there was no significant impact downstream on pregnancy duration or need for labour induction.
As labour typically followed shortly after PROM in a previous trial report on coitus in pregnancy from our centre, PROM was considered as the onset of labour. Our protocol for uncomplicated PROM allowed patient choice of immediate labour induction or an inpatient wait of up to 24 hours for spontaneous labour before labour induction. Hence, the labour induction rate following PROM was dependent on this choice. If we took PROM to be the onset of labour, the spontaneous labour rates were 503/574 (87.6%) versus 497/576 (86.3%) (RR 1.1, 95% CI 0.8–1.6, P = 0.278) for the advise-coitus compared with the control arms, respectively, again with no statistical difference between the groups.
Table 3 shows the secondary outcome measures of the two trial arms. There was no significant difference in non-birth related hospitalisations after enrolment, interval between enrolment to admission for birth, rate of PROM, indication for labour induction, need for oxytocin augmentation of labour, length of active labour, mode of delivery, indication for caesarean delivery, hospital stay for delivery, postpartum blood loss, umbilical arterial cord pH, Apgar score at 5 minutes, birthweight, perinatal mortality rate or neonatal intensive care admission rate between the trial arms. The three cases that resulted in perinatal mortality all presented with intrauterine deaths and macerated stillbirths, without obvious external malformations. In one of these cases, the fetus was probably growth restricted, with a birthweight of 2.23 kg at 38 weeks of gestation, and herpes simplex IgM was only detected in the maternal serum at presentation for fetal death after routine investigation. In the other two cases, the intrauterine deaths remained unexplained after routine investigations. Autopsies were not performed, in accordance with parental requests. All three women reported coitus, and post-hoc analysis after categorisation into women that reported coitus versus women who reported abstinence showed no difference in perinatal mortality rate (3/939 [0.3%] versus 0/198 [0%], P = 1.00); in contrast, the neonatal unit admission rate of 10/938 (1.1%) versus 7/198 (3.5%) (RR 0.3, 95% CI 0.1–0.8, P = 0.018) was higher in abstinent women. The latter finding may be a ‘false-positive' type-1 error. The similarity in occurrence rates of all the secondary outcomes within the trial and the post hoc analysis based on reported coitus on neonatal outcomes would suggest that coitus in late pregnancy is safe but ineffective in initiating labour.
Table 3. Secondary outcomes of women randomised to advise-coitus or control groups
|Outcome|| n a ||Intervention||RR (95% CI)|| P |
|Advise coitus n = 574||No advice n = 576|
| Additional hospital admission before birth ||1149||52 (9.1)||57 (9.9)||0.9 (0.6–1.3)||0.687|
| Interval from intervention to admission for birth (weeks) ||1150||3.2 ± 1.3||3.2 ± 1.3|| ||0.522|
| PROM ||1150||94 (16.4)||80 (13.4)||1.2 (0.9–1.7)||0.250|
| Indications for induction of labour ||245|| || || ||0.118|
|PROM|| ||64 (51.2)||50 (41.7)|| || |
|Non-reassuring fetal status|| ||26 (20.8)||28 (23.3)|| || |
|Prolonged pregnancy|| ||23 (18.4)||26 (21.7)|| || |
|Others|| ||12 (9.6)||16 (13.3)|| || |
| Epidural analgesia ||1150||141 (24.6)||130 (22.6)||1.1 (0.9–1.5)||0.455|
| Oxytocin augmentation ||1150||152 (26.5)||148 (25.7)||1.0 (0.8–1.4)||0.788|
| Length of active labour (hours) ||1024||4.9 ± 3.6||4.8 ± 3.6|| ||0.762|
| Mode of delivery ||115|| || || ||0.482|
|Spontaneous vaginal||0||421 (73.3)||414|| || |
|Instrumental vaginal|| ||52 (9.1)||(71.9)|| || |
|Caesarean section|| ||101 (17.6)||46 (8.0)|| || |
| Indications for caesarean || || ||116 (20.1)|| ||0.050|
|Non-reassuring fetal status|| ||36 (35.6)||57 (49.1)|| || |
|Failure to progress|| ||27 (26.7)||34 (29.3)|| || |
|Failed induction|| ||13 (12.9)||9 (7.8)|| || |
|Malpresentation|| ||8 (7.9)||9 (7.8)|| || |
|Miscellaneous|| ||16 (16.8)||7 (6.0)|| || |
|Timing of caesarean |
|After spontaneous labour||57 (56.4)||61 (52.6)|| || ||0.814|
|After induced labour||37 (36.6)||45 (38.8)|| || || |
|Planned electively||7 (6.9)||10 (8.6)|| || || |
| Maternal hospital stay for delivery (days) ||1149||1.9 ± 1.5||1.9 ± 2.5|| ||0.743|
| Peridelivery blood loss ||1065||316 ± 189||323 ± 207|| ||0.532|
| Blood loss ≥1000 ml || ||6 (1.1)||11 (2.1)||0.5 (0.2–1.5)||0.232|
| Umbilical cord blood pH ||102||7.28 ± 0.08||7.29 ± 0.09|| ||0.274|
| pH ≤7.1 ||2||16 (3.1)||20 (4.0)||0.8 (0.4–1.5)||0.500|
| Apgar score at 5 minutes ||1064||10 [10–10]||10[10–10]|| ||0.405|
| Apgar <7 at 5 minutes || ||1 (0.2)||1 (0.2)||1.0 (0.1–16)||1.00|
| Birthweight (kg) ||1149||3.12 ± 0.41||3.12 ± 0.41|| ||0.941|
| Perinatal mortality ||1150||1 (0.2)||2 (0.3)||0.5 (0.05–5.5)||1.00|
| Neonatal admission ||1149||5 (0.9)||12 (2.1)||0.4 (0.1–1.2)||0.094|
| Indications for neonatal admission ||17|| || || || |
|Observation || ||1||5|| || |
| Suspected meconium aspiration || ||0||4|| || |
| Low Apgar score || ||1||1|| || |
| Jaundice || ||2||1|| || |
| Lower limb deformity || ||0||1|| || |
| Congenital pneumonia || ||1||0|| || |
Post hoc, we also compared the pregnancy duration and labour induction rate of the coitally active versus abstinent participants. Intervention to delivery interval was (mean ± SD) 3.3 ± 1.3 versus 3.2 ± 1.3 weeks (P = 0.563) in the 939 participants who reported coitus compared with the 198 who were abstinent. Labour induction rates were 195/939 (20.8%) versus 49/198 (24.7%) (RR 0.8, 95% CI 0.6–1.1, P = 0.217. Women in the trial whose labour was induced reported fewer episodes of coitus, median [IQR] 2 [1–4] versus 3 [1–5] [P = 0.03]). We then performed a multivariable logistic regression analysis to evaluate the risk factors for labour induction, incorporating frequency of reported coitus, recruitment to delivery interval, ethnicity, premature rupture of membranes and nulliparity into the model (these variables had P < 0.1 on bivariate analyses against labour induction). After adjustment, the frequency of reported coitus (as a continuous variable) was not independently associated with labour induction (AOR 0.94, 95% CI 0.87–1.02, P = 0.11). With these findings, and also taking into account the primary outcome results, as shown in Table 2, we concluded that it would be futile to continue with the trial.
- Top of page
We observed a significant difference in reported coital activity in the advise-coitus group compared with the control group of 85.3 versus 79.9% (RR 1.5, 95% CI 1.1–2.0, P = 0.019). This effect is consistent with the results of a previous trial that investigated advising coitus to women scheduled for a non-urgent labour induction, which reported rates of coitus of 60.2 versus 39.6% (RR 1.5, 95% CI 1.1–2.0, P = 0.004) in the advised-coitus versus control arms, respectively. The frequency of reported coital activity was also higher in the advise-coitus group in the current trial. These findings demonstrate that coital activity at term can be influenced by medical advice. However, despite the increase in reported coitus, there was no observed impact on the onset of labour or the labour induction rate in our current trial or in the earlier trial. This lack of correlation between reported coitus and downstream outcome of labour onset is also reported by a prospective study.
We specifically targeted women who were recently abstinent from coitus in the anticipation that any advice promoting coitus would have the greatest impact within this group of women as they started from a zero base in coital activity, and that abstinent controls would largely continue to be abstinent by inclination. Our sample size calculation was modelled on anticipated coitus rates of 36.7 and 5.0% for advise-coitus and control arms, respectively. The observed reported coitus rates were far higher in both groups (85.3 and 79.9%), and the difference between the groups was far smaller than postulated – this difference of just 5.4% would have diminished the impact of the trial intervention on downstream primary outcomes of pregnancy duration and labour induction rate. However, post hoc analysis comparing women who reported coitus with those who reported abstinence did not reveal any major differences in pregnancy duration or labour induction rate between these groups.
Sexual activity including vaginal sex towards the end of pregnancy is reported by 62% in the third trimester, 40% in the 2 weeks before labour and 17% in the 2 days before labour. In a recent survey, over half of the postpartum women reported self-application of non-prescribed methods of inducing labour, which included sexual intercourse and nipple stimulation, revealing a huge underlying demand to shorten pregnancy. The relatively high (85.3%) reported coitus rate in our advise-coitus arm could result from our underestimation of the inherent demand to shorten pregnancy and the willingness to apply a medically sanctioned method, which was presented as safe, effective and natural. We postulate that the also high (79.9%) reported coitus rate in the control arm might be because of an increased awareness of coitus from trial participation, allayed safety concerns, as we stressed that coitus was safe, even in the control arm, and hence reinforcement of the folk belief that coitus can initiate labour. As the interval between trial recruitment and admission for birth was more than 3 weeks on average, it was also possible that the positive reinforcement for coitus given to the advise-coitus arm might also have percolated through to the control arm via social interaction across the trial arms, as we did not forbid such discussions.
Although human ejaculate weakens chorioamniotic membranes, the association of sexual intercourse in late pregnancy and PROM is not consistently reported.[16-18] Our study showed no significant difference in PROM (16.4 versus 13.4%; P = 0.25) between the groups, similar to a previous trial report, and consistent with another study that concluded that there were no sexual positioning or sexual activities that were significantly related to PROM.
In our present study, there was no significant difference in length of labour, need of oxytocin, mode of delivery, postpartum haemorrhage, Apgar score at 5 minutes, cord arterial blood pH, admission to neonatal unit and perinatal mortality across the trial arms. These findings suggest that coitus at term is safe and is entirely in keeping with prior reports that coitus is not associated with adverse pregnancy outcome.[20, 21]
There are strengths and limitations in our trial. Our enrolment was in excess of a thousand participants, and the results of analyses, including post-hoc analyses, were consistent across the full range of outcomes. Hence our finding that suggesting coitus to initiate labour is ineffective is likely to be robust. The promotion and control sessions were conducted by a few medically qualified members of the investigating team. Although these individuals were instructed and trained, an identical delivery of the intervention in an interactive environment was improbable. However, participants were randomised and using several doctor-counsellors should provide a robust test of the intervention's effectiveness. Although we did not document the presence of the male partner during the counselling sessions, the male partner was usually absent, which is not ideal, as coitus is an activity for the couple together. However, the high reported coitus rates from our data would imply that in our population, the absence of the male partner in the counselling session is not a major impediment to the uptake of advice in favour of coitus. We had a major problem with the return of the coitus diaries. Only 12% of the responses that we eventually obtained on coital activity were from diary return, the remainder were oral reports taken over the telephone, after the event, and were restricted to participants' recollection of the number of coital acts in the period between trial enrolment and delivery. This delayed recall might have affected results, but women who recalled their sexual activity during pregnancy would typically report fewer episodes of intercourse. Moreover, there was not a significant difference in reported coitus between the diary return and telephone contact groups.
We believe our findings are generalisable as coitus is widely believed to expedite labour, and many women apply this method on their own accord.