Outcome of planned home and planned hospital births in low risk pregnancies: prospective study in midwifery practices in the Netherlands

STUDY DESIGN

The study was conducted prospectively in two periods between 1990 and 1993 among women with low risk pregnancies receiving midwifery care in the province of Gelderland. A total of 97 midwives in 54 practices enlisted 2301 women, who signed an informed consent form and received a questionnaire about their social background and their preference for birth at home or in hospital. The midwives also received questionnaires about their clients, one to complete before delivery and one afterwards. A copy of the birth notification form (a voluntary registration system used by most midwives and obstetricians) with data on medical and obstetric background, labour, and delivery was added to the completed questionnaires.

For 294 women (13.8%) the birth notifications indicated obstetric referral before the onset of labour, which was an exclusion criterion, and for another 171 (8.0%) information from the midwives could not be checked against birth notifications because they were not available. The study population thus consisted of 1836 women, 840 primiparae and 996 multiparae, of whom 1140 had chosen home birth and 696 hospital birth. For 116 (6.3%) women, information was confined to what had been received from their midwife and the birth notification form; these women were excluded only from the subanalyses relating social background to outcome.

DATA ANALYSIS

Data variables were divided into background and outcome variables according to whether the variable was or could be known before the onset of labour. A value of 1 or 0 was awarded to each, based on the optimality concept originally developed by Prechtl6 and Touwen et al,7 in which optimality indicates “the best possible”; it avoids judgments on what is normal or abnormal when defining, for example, “no episiotomy” as optimal. The items were then summed into separate indexes for perinatal background and perinatal outcome,8 reflecting the number of optimal items in each index.

The perinatal background index,8 consisting of 31 items, considers as “best possible” the absence of any social, medical, or obstetrical problem before and during pregnancy. Because of its expected skewness in a low risk population and the poor internal consistency expected with many unrelated items (verified by Cronbach's (alpha) = 0.29),9 the index was used in a simplified, dichotomous manner. Women at or above the median were considered to have a relatively favourable background, the others as having a less favourable background.

The perinatal outcome index consists of 36 items, of which 22 relate to childbirth, nine to the condition of the newborn, and five to the condition of the mother afterwards.8 Optimal values were based on the principle that a maximally healthy mother and baby with minimal intervention for both of them constitutes the best possible birth.8 10 The perinatal outcome index therefore considers not only the result, but also the means by which it is achieved.

Primiparous women and multiparous women were considered separately because of well known differences in outcome. All analyses were based on the planned rather than the actual place of birth because referral to hospital during labour is usually indicative of anticipated or existing problems. Including these women among hospital births would bias the results of planned hospital births negatively and home births positively.

Power analysis, based on detecting a significant difference in the combined frequency of non-optimal factors during and after childbirth, led us to aim for a sample size of 1600 women, with approximately half being multiparous and preferably half choosing hospital birth. Because women in Gelderland more often choose home than hospital birth, only women choosing hospital birth were recruited in the final four months of the study.

Differences in individual background and outcome items were assessed by the ² test and differences in the composite indexes by Student's t test.

MEASURING PERINATAL OUTCOME

Measuring the quality of maternity care has never been easy. For many years, perinatal mortality rates were used for this purpose, often with little regard for the value and validity of such data.11 Now, with rates well below 10 per 1000 births, they have lost virtually all of their utility for measuring quality of care in the Western world. Other measures have yet to find acceptance, but it is unlikely that a single measure will ever be satisfactory for a process that involves mother and baby and for which the end result is not the only outcome that matters. We therefore opted for a differentiated approach that considers both the mother and the baby and that takes both the results and the way in which they are achieved into account. To this end and with a view to obtaining a single measure for maximal outcome with minimal intervention8 10 we constructed a composite perinatal outcome index based on an optimality concept developed in the 1970s for identifying a cohort of infants with a flawless start in life.6 7

OUTCOME IN RELATION TO BACKGROUND

Using this tool we compared the outcomes of planned home births with those of planned hospital births for primiparous and multiparous women after controlling for the confounding effects of social, medical, and obstetric background. Without control for this background, the perinatal outcome in primiparous women was significantly better for planned home births than for planned hospital births. This is mainly because nulliparous women with a less favourable background tend to prefer hospital, whereas those with a favourable background tend to choose home birth. This may be different in other countries, but it is not unexpected in the Netherlands, where home birth has been an approved option for a long time.1 5 12 After background variables were controlled for, the perinatal outcome for primiparous women with low risk pregnancies was similar for those who planned home births and those who planned hospital births.

For multiparous women with low risk pregnancies, the perinatal outcome of planned home birth was significantly better than that of planned hospital birth, whether or not background was controlled for. A closer look at the background characteristics shows that multiparous women with a complicated previous pregnancy, including instrumental delivery in our study, were more likely to opt for hospital birth than for home birth. Their history may put them at higher risk of encountering problems again, and this may account for some difference in outcome between home and hospital. However, the multiparous women in our study were at low risk and their history would not have prompted referral to an obstetrician. We also analysed our data after excluding women with a less than optimal obstetric history, and the perinatal outcome index remained better for planned home birth than planned hospital birth (t = 4.75, P<0.001). Further research will be necessary to determine how much of the difference in outcome can be attributed to obstetric history and how much to the chosen place of birth. In the meantime and on the basis of our results, the place of birth seems to affect perinatal outcome in women at low risk.

IMPACT OF CHOICE

Ideally—and particularly when offset against virtually 100% hospital births in the rest of Europe—better evidence is needed before generalisations are made on the merits of planned home birth. Such evidence is not easy to gather. It is well known that a variety of psychological factors can influence people's health and interfere with medical treatment. In obstetrics, levels of anxiety have been found to predict obstetric complications.13 Choice itself (allowing women to choose home or hospital birth) may influence levels of anxiety and apprehension and thereby also the outcome of maternity care. Evidently, the elimination of choice—as would be necessary in a randomised trial—could by itself have a major impact on perinatal outcome by inducing insecurity and anxiety in women assigned to give birth in a manner that they do not prefer. In areas where the patient's choice has a profound effect on outcome, random comparisons eliminating choice will give unreliable estimates of true differences.14 Therefore, in the Netherlands, where choosing between home or hospital birth is an integral feature of the system, randomised controlled trials between home birth and hospital birth would not produce generalisable results even if it were possible to mount such trials.

Our research has shown that, for women with low risk pregnancies in the Netherlands, choosing to give birth at home is a safe choice with an outcome that is at least as good as that of planned hospital birth. We also found indications that there is some self selection among women who can decide for themselves where to have their baby, and that this preordains outcome, albeit to a limited extent. It is important, therefore, that the home birth option remains available, but especially that women at low risk are really given a free choice.