The study was a multicentre, randomised controlled trial with prognostic stratification by hospital. Before the trial a group of nurses or midwives at each hospital were trained in the structured approach, with the number trained at each site reflecting the projected rate of enrolment. Additional training sessions were held during the recruitment phase in sites with substantial staff turnover. In the North American hospitals the two day training programme was carried out by nurse experts who worked in an organisation providing continuing education to labour nurses in Ontario, Canada. In the UK hospitals the training programme was carried out by an expert midwife instructor after consultation with the Canadian trainers, using an adaptation of the North American curriculum. At the end of the workshop each participant was given a manual, which reiterated course content and provided opportunities for the techniques to be practised before the onset of the trial. The box details the topics in the training programme, and the accompanying manual covered the components of structured care.
Components of structured care
The following components were taught in the training programme and used in the structured care group*
- Normalise the environment—if the labour assessment unit is a clinical environment or offers little privacy, move to a lounge or empty room away from medical equipment; encourage use of comforting objects and spontaneous self comforting behaviours
- Palpate to assess fetal position
- Encourage maternal positions that promote fetal head rotation or relieve pain—standing and leaning forward; asymmetrical upright (standing, kneeling, sitting); sitting upright; sitting, leaning forward with support; kneeling, leaning forward with support; on hands and knees; side lying positions (pure side lying and exaggerated Sims or semiprone); abdominal lifting
- Assess labour pain, both contraction pain and backache; assess physiological and behavioural indicators of pain; assess pain using a visual analogue scale
- Demonstrate cognitive, behavioural, and sensory interventions to manage labour pain; be present continuously; encourage comfort measures, including breathing and relaxation, application of heat and cold, massage, shower or bath, movement; encourage visualisation techniques, including recalling past experiences when contentment has been felt, picturing the results of the contractions (cervical dilation, fetal rotation, and descent), and visualising each contraction as a mountain and each breath as a step towards the peak; help create a ritual behaviour to use with each contraction; suggest music and rhythmic techniques
- Assess maternal emotional status; be aware of behavioural indicators of emotional distress; know how and when to ask direct questions, such as “What was going through your mind during that last contraction?” to identify coping related responses compared with distress related responses
- Use interventions to reduce emotional distress: offer helpful replies to distress related responses; reframe negative thoughts to positive ones; discuss the importance of relaxation, rhythm, and ritual; scan the body for areas of tension to focus relaxation efforts
- If hospital admission is not planned, discuss the importance of carrying on normal activities of daily living, with specific suggestions appropriate to time of day; offer anticipatory guidance about coping with labour as it becomes more active
Throughout the trial providers of structured care could communicate with the trainers and with each other on an electronic listserv. Before the launch of the trial at each centre the principal investigator (EH) held meetings with the staff at which she emphasised the uncertainty about the value of structured care and the importance of staff continuing to provide usual care to the control group, to maintain distinctions between both groups.
To be eligible to participate the hospitals had to have a pre-existing, geographically distinct labour assessment unit and a spontaneous vaginal birth rate of 75% or less. In all study sites the labour assessment units were the entry points for women who thought they were in labour. In North America the units were staffed by nurses and in the UK by midwives, but the approach to care was the same—that is, the purpose was to determine whether a woman should be admitted to the labour ward or sent home to await active labour. Women were eligible for the study if they were nulliparous, had a live singleton fetus in the cephalic position, had no contraindications to labour, were competent to give informed consent or had a parent or guardian who was competent to give informed consent, and were experiencing contractions but did not meet labour ward criteria for admission. Women were not eligible if the gestational age was less than 34 weeks, an induction of labour or caesarean delivery was planned, they had complications that necessitated hospital admission, they were likely to be transferred to the labour ward within one hour, or they had a doula or midwife providing continuous support.
After basic assessment of labour (duration and frequency of contractions, status of membranes, status of the fetal heart rate, and assessment of cervical dilation as per hospital protocol) the women were assessed for trial eligibility. Baseline data were obtained after eligible women gave written, informed consent to participate and before randomisation. Determination of race or ethnicity was a combination of self assignment and observer assignment. Randomisation was centrally controlled and concealed, using an internet based service. The nurse or midwife accessed the trial website to obtain the participant’s study group allocation.
Immediately after randomisation women assigned to the experimental group received one to one care by a nurse or midwife trained in structured care. The structured care provider palpated the participant’s abdomen to assess fetal position and asked her to describe her thoughts during the last contraction and to rate pain intensity on a visual analogue scale. The provider used positioning techniques, comfort measures, and simple cognitive restructuring techniques such as positive visual imagery and reframing negative thoughts, and offered anticipatory guidance about coping with active labour (see box). If the woman was sent home without having delivered and if circumstances permitted, the provider was to telephone her later to check on her progress. When the woman returned, efforts were made to repeat structured care.
Women assigned to the control group received care by a nurse or midwife who had not been trained in structured care. One nurse or midwife often provided care to more than one woman. Usual care depended on many factors, including the provider’s workload, the provider’s knowledge of and beliefs about assessments of fetal position and cognitive-emotional state, and familiarity with appropriate interventions. Women who were sent home were advised to telephone the unit if they had any questions or concerns.
The length of time women received structured care or usual care was designed to reflect the usual time spent by women in labour assessment units (1-4 hours). In both groups the decision on whether to admit women to the labour ward or to send them home was made as per usual hospital policy. Labour assessment units offer varying degrees of privacy and in some settings it was possible that a patient randomised to the control group could overhear or witness parts of the care to a woman in the experimental group. Therefore no woman was invited to participate while a trial participant was in the labour assessment unit unless complete privacy could be assured. Only the nature of the nursing or midwifery care in the labour assessment unit varied between groups; all other care in the labour assessment unit and labour ward was in accordance with usual hospital practices and policies.
Centres were instructed not to randomise women unless providers of both usual care and structured care were available. Structured care providers could care for non-study women but not for women in the control group. Usual care providers could not care for women in the structured care group. Compliance was assessed in two ways. We expected that over 90% in each group would receive their assigned method of care (care from a provider trained or not trained in structured care) immediately after randomisation. In addition, providers’ reports of their care for women in the structured care group provided evidence of adherence to the main components of the intervention.
The primary outcome was spontaneous vaginal birth. Secondary outcomes were the number of women who had no intrapartum analgesia or anaesthesia, had perineal trauma requiring suturing, and reported negative views of their care (a subsequent paper will report on costs). Other study outcomes included the number of women with more than two visits for assessment of labour; use of intrapartum oxytocics, regional analgesia, and electronic fetal heart rate monitoring; length of hospital stay; and indicators of short term and longer term maternal and neonatal morbidity that have been linked in previous research to method of delivery, including postnatal emotional distress, readmission to hospital of mother or baby for delivery related complications in the first 6-8 weeks after birth, neonatal transfer to a special care baby unit, and intrapartum fetal death or neonatal death.
Trained research nurses or midwives at each hospital abstracted relevant data from the medical records and entered the data into study forms on the trial website. Because the data collection included a question about compliance (whether a provider with training in structured care had cared for the participant), the research staff may not have been fully blinded during retrieval of data from the medical records. The primary and other medical outcomes were, however, objective data that were recorded in the medical records as part of routine practice. Audits of medical records at selected sites during the enrolment period showed neither systematic errors nor important random errors. Attending doctors were not explicitly informed of their patients’ group assignments and were rarely present in labour assessment units.
Participants were asked to complete a questionnaire 6-8 weeks after the birth, which focused on their health, their baby’s health, and their satisfaction with care. The questionnaire included the Edinburgh postnatal depression scale; a score greater than 12 is a reliable indicator of postnatal depressive symptomatology.25
Measurement of satisfaction is problematic because of the “what is, must be best” phenomenon.26 27
A systematic review identified key factors influencing satisfaction with childbirth,27
and the questionnaire items were adapted from one of the most reliable and well validated population based surveys of satisfaction with childbirth.28
Participants could complete their questionnaires directly on the secure trial website,29
by post, or by telephone with the centre research nurse or midwife.
Sample size calculation
During development of the trial protocol 10 hospitals provided the rates of spontaneous vaginal birth for women who would meet trial eligibility criteria. Rates ranged from 42% to 76% (mean rate 58%). In selecting the clinically important difference for the current trial we wanted to detect a 4% absolute difference in spontaneous vaginal birth rate—for example, eliminating one operative delivery for every 25 women treated. With a two sided α<0.05 and α=0.20 in hospitals at the extremes of the range, and when using the mean rate and taking the most conservative approach, we needed a sample size of 4932. The target sample size was 5000.
We analysed the results according to intention to treat. For the primary outcome we used a significance level of 0.05 (two tailed). We set the significance level for secondary outcomes at 0.0125 and for other study outcomes at 0.005. Because we expected variation owing to the effects of unknown characteristics of the hospitals, the analytical approach allowed the proportion of women experiencing spontaneous vaginal birth and treatment effects to vary between hospitals. For binary outcome variables we compared the groups using a logistic regression model with a random hospital effect for the intercept and slope. We present the odds ratios and accompanying confidence intervals (corresponding to the preset P values for primary, secondary, and “other” outcomes). We used a similar logistic regression model to explore the interaction effects between baseline variables and treatment group on the primary outcome. For length of hospital stay we analysed data using a linear regression model with a random hospital effect for the intercept and slope, using the log of length of stay as the dependent variable. Statistical procedures were done using SAS version 9.1. For ratings of women’s views of their care we followed the standard practice of comparing the frequencies with which less than very positive views were reported.26 27 28