British Journal of Obstetrics und Gynaecology Octobcr 1991, Vol. 98, pp. 1001-1008

Randomized controlled trial of antenatal social support to prevent preterm birth ROBERT L. BRYCE, FIONA J. STANLEY, J. BARRY GARNER Abstract Objective-To test the effect of a programme of additional antenatal social support on the occurrence of preterm birth (a birth from 20 to 36 weeks gestation) in women at risk of preterm birth. Design-A prospective randomized controlled trial. The design was one of randomization before consent for a new treatment. Setting-Three public hospital antenatal clinics in Perth and the offices of 87 obstetricians and general practitioners in Western Australia. Subjects-1970 pregnant women with poor obstetric histories entered the trial; 983 of these were randomly allocated to the programme group and 987 to the control group. Interventions-Normal antenatal care was provided for both groups. In addition, members of the programme group were offered an intervention aimed at providing expressive (emotional) social support, consisting of antenatal home visits and telephone calls by midwives. Of the women allocated to the programme group, 24 refused consent and 69 were not traced before completion of their pregnancies, the remaining 890 women (90-50/,) agreed to enter the programme, and each received at least one intervention. Main outcome measures-Gestational age at completion of the pregnancy. A pregnancy ending before 20 weeks was labelled a miscarriage. Results-There were 1261981 (12.8%) preterm births in the programme group and 1471986 (14.9%) in the control group. The outcome data for two women in the programme group and one in the control group could not be found. The unadjusted odds ratio for preterm birth in the programme was 0.84 (95% CI 0.65-1.09). The observed relative reduction in preterm births associated with the programme was 13.8% (95% CI -8.2% to +31.5%) and the trial had a 60% power to exclude a true relative reduction of 25%. Conclusions-The results of this trial and those of other controlled clinical trials provide little evidence for the effectiveness of social support interventions in the prevention of preterm birth in women with poor obstetric histories.

Programmcs dcsigncd to reduce the occurrence of preterm birth or low birthweight have been Department of Obstetrics & Cynaerology, Flinders Medical Centre, Bedford Park, South Australia 5042 R. L. BRYCE Senior Lecturer Western Australian Research Institute for Child Health, Princess Margaret Hospital for Children, Roberts Road, Subiaco, Western Australia 6008 F. J. STANLEY Director J . B. GARNER Biostatistician

Correspondence: Dr R. L. Brycc

established in several countries in an attempt to reduce the mortality and morbidity associated with preterm birth and to reduce the costs of neonatal intensive care (Institute of Medicine 1985). These programmes have combined several approaches to the prevention of preterm birth including health education, social support and the use of tocolytic drugs. Assessments of the effectiveness of these programmes could have been subject to bias as the incidcnccs of preterm birth in women enrolled in the programmes have been compared either with the 1001

1002

R. L. Bryce et at.

incidence in women attending thc same hospitals before introduction of the programme (Papicrnik et al. 1985; Herron et al. 1982), or with the concurrent incidence in women attending a hospital without the programmc (Hcrron et (11. 1982). There is less opportunity for bias with a randomized controlled trial, but the only randomized controlled trial of a prcterm birth prevention programme published before thc start of this trial failed to demonstrate an effect (Main et al. 1985). Even if these programmes combining several approaches to the prevention of preterm birth wcre effective, it would not be clear which componcnt was the most effective, nor would it help elucidate the causcs of preterm birth. For these reasons, it was considered preferable to study the effectiveness of a single intervcntion in the prevention of preterm birth and in the prescnt trial, social support was chosen as the intervention to be tested. The choice of social support was based on the observation of scvcral associations. Stress in pregnancy (gencrally measured in terms of life events) is associated with preterm birth and low birthweight (Newton & Hunt 1984). Preterm birth is most common in women of the lowest social classes (Fedrick & Anderson 1976; Bcrkowitz & Kasl1983) and these women experience the highest levels of stress and have the lowest levels of social support (Brown & Harris 1978). However, the association betwccn stress in prcgnancy and preterm birth is not entirely an effect of social class since this association was also found in a study which controlled for social class (Ncwton et al. 1979). Thc observation that, among women exposed to high lcvcls of stress in pregnancy, those with the most social support expericnccd thc fcwcst prcgnancy complications (Nuckolls et a/. 1972) led our group to design a randomized controlled trial to cxaminc the hypothesis that the provision of additional social support in pregnancy would, for some women, ameliorate the effects of stress and allcviatc thc cffects of social disadvantage, thereby improving the outcome of prcgnancy by a reduction in preterm births. There have becn several previous randomized controlled trials of interventions in pregnancy which may have provided some social support and which have found some improvements in psychological or behavioural outcomes but. in general. they have not had the statistical power to address physical outcomes (Oakley 1985). Our trial was designed to have sufficient power to demonstratc any clinically important effects on preterm births.

Subjects and methods A randomized controlled trial was designed to test the primary hypothesis that il programme of antenatal social support for women at high risk of preterm birth would significantly reduce the incidcncc of preterm births (Stanley & Bryce 1986). A 25% relative reduction in the incidence of prctcrm births was proposed to be the minimum effect that would justify the introduction of the programmc as a public health measure. Secondary hypothcscs proposed that the effect of the programmc would be greatest in women in the lowest social class and in women with the least existing social support. A preterm birth was defined in the trial as a birth between 20 and 36 weeks gestation as this is a widely uscd definition in Australia. (Other countries, including the United Kingdom, use different definitions, for example 28-36 weeks gestation). The supply of social support involves an exchangc between people of expressive and instrumental aid. For thc purposes of the trial, we characterized cxpressive support as sympathy, empathy, understanding, affection, acceptance and being a confidante; instrumental support was characterized as information, advice and material aid (Thoits 1982). Participants The trial design was one of randomization prior to conscnt for a new treatment (Zelen 1979). Women attending any of the three public hospital antcnatal clinics in Perth, or the offices of any of 87 obstctricians and general practitioners in Wcstern Australia who attended more than 50 births in 1984, were offered enrolment at their first antenatal visit. The enrolment form informed the women that some of those who agreed to enrol would be offered visits at home by a midwifc to talk to them and listen to their problems. Womcn were ineligible for enrolment if they were non-English speaking, previously enrolled in the trial, more than 25 completcd weeks gestation, or if their fetus was dcad. They were eligible for the programme if they had a history of one or more preterm births, one or morc low birthweight births (

Randomized controlled trial of antenatal social support to prevent preterm birth.

To test the effect of a programme of additional antenatal social support on the occurrence of preterm birth (a birth from 20 to 36 weeks gestation) in...
611KB Sizes 0 Downloads 0 Views