Why women do not accept randomisation for place of birth: feasibility of a RCT in the Netherlands


  • Study conducted at the Maastricht UMC+, PO Box 5800, 6202 AZ Maastricht, the Netherlands

Dr M Hendrix, Department of Obstetrics and Gynaecology, Maastricht UMC+, PO Box 5800, 6202 AZ Maastricht, the Netherlands. Email mhendrix@ved2.azm.nl


Objective  The purpose of this study was to investigate why low-risk nulliparae were not willing to participate in a randomised controlled trial (RCT) of place of birth.

Design  Prospective study.

Setting  The Netherlands.

Population  All low-risk nulliparous women starting their pregnancy under midwife.

Methods  A questionnaire for 107 nulliparae who were willing to participate in a cohort study on place of birth, but at an earlier stage in their pregnancy declined to participate in a RCT of place of birth. This questionnaire included 12 items on a 4-point Likert scale but was not subjected to formal validation.

Main outcome measure  Reasons why nulliparae did not accept randomisation of place of birth.

Results  The most important reason why women refused participation in the trial was that they had already chosen their place of birth before they were asked to participate at 12 weeks of pregnancy. From their answers, it became clear that pregnant women strongly value their autonomy of choice. The decision not to participate in the trial was not influenced by the information given by the midwife and the additional written information.

Conclusions  Factors that prevent randomisation for place of birth are difficult to influence. There is a need to explore why there is such certainty of view among women having their first child. Until we have an understanding of why women select information to make these choices and why women are reluctant to participate in trials that challenge choice, it may well be impossible to mount a trial of place of birth.


With increasing interest in home birth, there is a continuing debate about the need for a randomised controlled trial (RCT) comparing home birth versus hospital birth that would assess perinatal outcomes and maternal safety and satisfaction. An attempt at a trial in Britain recommended the Netherlands as the ideal setting for this type of RCT because the Dutch obstetric system has a well-established system that promotes home or home-like births.1 Compared with other western countries, the Dutch obstetric system is unique: women, who have a low-risk pregnancy, are free to choose between a home birth (30% of all deliveries) and a home-like short-stay hospital birth (12% of all deliveries). In both cases, the birth is supervised by a registered independent midwife.

A study of place of birth is of interest as the process of childbirth is influenced by both biological and emotional factors, such as feelings of security and confidence, which are related to the place of birth.2

We designed a RCT to investigate differences in interventions, satisfaction, referrals to the obstetrician and costs between home birth and home-like short-stay hospital birth. Low-risk nulliparous women were to be randomly allocated to giving birth at home or at a hospital, in both cases assisted by an registered independent midwife. They would be studied from the first trimester of the pregnancy until 6 weeks after delivery. Outcome variables of interest included expectations and experiences of pregnant women and their partners, indications for referral to specialist care and clinical interventions.

A crucial question was whether pregnant women would be prepared to participate in such a randomised trial, given the UK experience. Therefore, we conducted a pilot study in 2003. One hundred nulliparous women were given a hypothetical scenario of being randomly allocated to either a home birth or a short-stay hospital birth. Sixty women (60%) answered this question, and 30 women indicated that they would certainly be willing to participate. Based on this result, a power calculation was performed; the trial was registered at Clinical Trials (registration number NCT00237601) and accepted by the medical ethical committee. Funding was raised and recruitment started in March 2006.

The trial was conducted in different parts of the Netherlands. Thirty-five midwives in 14 primary care midwifery practices participated in the trial by recruiting pregnant women. The midwives gave information about the trial during the first prenatal visit, usually between 8 and 10 weeks of pregnancy. Only nulliparous women were eligible to participate. Inclusion was possible up till the 18 week of pregnancy. A week after this prenatal visit, these women were contacted by the researcher to ask if they wanted to participate in this trial. During this contact, more information about the study was given to the women and their questions were answered. When a pregnant woman declined to participate in the trial, the reason for this was asked and noted and the woman was asked to participate in a nonrandomised cohort study. Participants were free to choose their place of birth, and data for both groups would be analysed separately. The reasons for nonparticipation in the trial or in cohort study at all were registered by the researcher.

After an inclusion period of 6 months during which all nulliparous women were asked to participate in the trial, only one woman had given informed consent for randomisation. Another 115 women declined the RCT but were willing to participate in the cohort study.

Given this striking discrepancy with our own expectations based on the results of the pilot study, several questions arose: why do pregnant women decline to be randomly allocate to either home birth or home-like short-stay hospital birth, which factors influence their decision and which is most decisive? We decided to perform a questionnaire study explore these questions.


A newly constructed, self-administered postal questionnaire was developed. The study population was all nulliparous women who had declined participation in the trial but had agreed to participate in the cohort study (n = 115). Women who had given birth before the questionnaire was sent out were excluded (n = 8). The gestational age of women who were sent this questionnaire was between 9 and 34 weeks, 1 and 20 weeks after they declined participation in the trial.

The questionnaire was based on the reasons the women had given by telephone and reasons found in previous studies of Verheggen et al.3,4 Twelve items were included on a 4-point Likert scale containing statements respondents could agree or disagree with (Table 2). The statements referred to motivations for not participating in medical research in general and to possible motivations for not participating in this trial in particular. Underlying ideas that had emerged were feelings of autonomy in decision-making, fear of over-medicalisation or under-medicalisation in relation to the participation in the trial or indifference with regard to medical research in general.

Table 2.  Women’s motivations for not participating in this trial in particular or motivations for not participating in medical research in general
PropositionThis proposition does agree to meMean (SD)
Not at all = 0Hardly = 1Moderately = 2Very much = 3
  1. Figures are n (%) unless stated otherwise.

1. I had already decided where to give birth before I was asked to participate in this trial5 (6.0)2 (2.4)23 (27.4)54 (64.3)2.50 (0.814)
2. I hold on to my own choice on where to give birth, researchers should not meddle in this1 (1.2)9 (10.7)23 (27.4)51 (60.7)2.48 (0.736)
3. With the birth of my first child I do not want to run the risk of having to deliver in the wrong place10 (11.9)13 (15.5)26 (31.0)35 (41.7)2.02 (1.029)
4. I wanted to be randomly allocated but my partner was against it75 (89.3)3 (3.6)4 (4.8)2 (2.4)0.20 (0.636)
5. I do not wish to run the risk of receiving a treatment I do not want10 (11.9)5 (6.0)30 (35.7)39 (46.4)2.17 (0.992)
6. The written information about the trial was not clear69 (82.1)11 (13.1)3 (3.6)1 (1.2)0.24 (0.573)
7. The information the midwife gave me about the random allocation was not clear70 (87.5)6 (7.5)3 (3.6)1 (1.2)0.19 (0.553)
8. Basically I do not participate in medical research66 (78.6)11 (13.1)5 (6.0)2 (2.4)0.32 (0.697)
9. I do not like medical research66 (78.6)10 (11.9)7 (8.3)1 (1.2)0.32 (0.679)
10. I do not understand the purpose of medical research72 (85.7)9 (10.7)3 (3.6)0 (0.0)0.18 (0.470)
11. I never participate in medical research63 (75.0)13 (15.5)6 (7.1)2 (2.4)0.37 (0.724)
12. I am afraid that if I consent to randomisation I cannot get out of it39 (46.4)10 (11.9)14 (16.7)21 (25.0)1.20 (1.269)

We asked to indicate to what extent they agreed or disagreed with the statements. To try to determine the content validity of the questionnaire, the participants were also asked to write down their motivation in their own words, if this had not already been worded in the precoded, closed items in the questionnaire.

The questionnaire was sent to 107 participants. The results were analysed using SPSS 12.0 (SPSS inc., Chicago, IL, USA).


Of the 107 participants, 84 women (79%) returned the questionnaire. Twenty-one (19%) did not return the questionnaire. Two women (2%) had a spontaneous miscarriage (Figure 1).

Figure 1.

Follow up of the study population.

The demographic characteristics are summarised in Table 1.

Table 1.  Characteristics of women in study (n = 84)
Basic characteristicsn (%)Mean (SD)
Mean gestational age (weeks)84 (100)15 (4.1)
Age (years) 
<2514 (16.8)28 (3.8)
25–3566 (78.4)
>351 (1.2)
Unknown3 (3.6)
Height (cm) 
<1.6516 (19.0)1.70 (0.06)
1.651.7545 (53.6)
>1.7515 (17.9)
Unknown8 (9.5)
Weight (kg) 
<6016 (19.0)68.9 (12.9)
607541 (48.8)
>7518 (21.5)
Unknown9 (10.7)
Dutch78 (92.9) 
Other3 (3.6)
Unknown3 (3.6)
Employment65 (77.4) 
Unemployment11 (13.1)
Unknown8 (9.5)
Household income (monthly) 
<1000 euro3 (3.6) 
1001500 euro3 (3.6)
15002000 euro5 (6.0)
20002500 euro18 (21.4)
25003000 euro21 (25.0)
>3000 euro17 (20.2)
No information14 (16.7)
Unknown3 (3.6)
Marital State 
Married30 (36.1) 
Living together48 (57.1)
Single1 (1.2)
Unknown5 (6.0)
Chosen place of birth 
Home birth47 (56) 
Short-stay hospital birth30 (35.7)
Hospital birth2 (2.4)
No idea2 (2.4)
Unknown3 (3.6)

Declining participation in the trial appeared to be based on four propositions (Table 2, proposition 1, 2, 3 and 5). The most important was that they had already chosen the place of birth before they were asked to participate in the trial. In more than 64.3% of the cases, respondents agreed very much with that proposition. 60.7% of respondents agreed very much with the proposition: ‘I hold on to my own choice on where I wish to give birth, researchers should not meddle in this’. Another important proposition was the fear of over-medicalisation: ‘I do not wish to run the risk of receiving a treatment I do not want’ (46.4% agreed with this proposition strongly). Also the fact that they were pregnant with their first child was an important reason for the respondents (Table 2, proposition 3).

The unwillingness of women to participate in the trial did not appear to be influenced by the oral and written information given by the midwives and a member of the research team. Eighty percent of the respondents indicated that the written information was clear enough. The purpose of medical research was clear for the respondents, and women stated that they were not against participation in medical research in general.

No woman gave additional reason(s) for refusing participation that had not already been noted in the closed items of the questionnaire.


The most important reason for pregnant women to decline randomisation was that they highly value their autonomy. Women want to decide themselves about the place of birth. It is their opinion that researchers should not be meddling in this decision. The fact that women highly value their autonomy in these matters is a factor that cannot easily be manipulated or influenced and has to be respected.

The questionnaire used in this study was an ad hoc questionnaire, newly constructed for this occasion. The researcher had registered the reasons the women gave for their trial refusal. These reasons were used to construct items in the questionnaire in combination with some general questions from literature, and the respondents had the opportunity to give their own opinion about the refusal.

We acknowledge that no formal analysis of the overall validity of the questionnaire has been undertaken, although we have examined content validity to some extent. It was felt that the timing of seeking women’s views was important, and this precluded the necessary pilot phase of such a questionnaire that would have provided data for validation. We are aware that this may have influenced the conclusions of our study. Disadvantages of not validating the questionnaire can be found in the external and internal validity. It is important to know if the questionnaire measures what it should measure (internal validity) and if the results are valid for the total population (external validity). Usually, the validating process includes results from psychometric techniques like item analysis, correlational analysis of items, factor analysis, reliability coefficients (Cronbach’s alpha) and scale construction(s).5,6

Especially, the external validity of this questionnaire can be discussed. It is difficult to say whether the results of this study are applicable to all nulliparous women.

The time that women were asked to participate in this trial might play a role in the results. The women were informed during the first visit to their midwife. At that moment, most of the women were 8–10 weeks pregnant. Previous research shows that an early choice in these matters is highly correlated with the final choice.7 The exact time when low-risk women decide where to give birth is unknown and difficult to determine. Women may choose their place of birth before they get pregnant.

It is unclear whether women stay with their first decision for the place of birth. It is possible that this is a dynamic process where women may move from one decision to another. It is interesting to investigate this decision-making process of pregnant women for the place of birth during pregnancy.

There is an increasing awareness that women should participate actively in decision-making, and there is evidence that properly informed women are more likely to adhere to treatment. Clear communication with women and their relatives is decisive. Random allocation of a treatment is a key feature that women have to understand prior to giving informed consent for a trial.8 From our results, we can conclude that the provision of the information seemed to have no effect on the decision to not participate in the randomisation. In this case, more information or other methods of communication would probably not have changed their mind.

The provision of clear and accurate patient information is important, but this alone will not ensure consistent interpretation of concepts, such as randomisation. Women may need to discuss the purpose of randomisation to understand them fully enough to give informed consent.9 Other studies, in oncology care, showed that one of the main reasons for accepting trial entry is ‘trust in the doctor’, implying that the act of communication has a greater influence on the decision than the written word.10 Randomised clinical trials pose particular problems, and the concept of randomisation raises many issues for both healthcare professionals and women.11 It is possible that midwives participating in the trial did not fully support randomisation, although we do not know this. This might have influenced their communication with eligible women and may have had an impact on the women’s decision to participate. However, eligible women who received information about the trial were also contacted by a member of the research team. They further explained the study protocol and its purpose. We therefore assume that the aim of the trial and the randomisation were sufficiently explained.

Randomisation in itself is not always understood by both women and the general public.12 A series of studies showed that participants in a randomised trial did not find randomisation itself acceptable nor did they find it acceptable that a clinician would not know which treatment was best.13,14 Clear explanation of the concept and the purpose of the randomisation are fundamental to ensure properly educated consent to clinical trial participation. The purpose of our randomised trial was to compare two common ‘standard’ settings. In the Netherlands, both home births and home-like short-stay hospital births are common choices for a delivery supervised by a registered midwife. We had therefore expected that a RCT for home birth or home-like short-stay hospital birth would be feasible.

This assumption was tested in a pilot study in 2003, before the trial actual started. One of the advantages of conducting a pilot study is that it might give advance warning about where the main research project could fail.15 The result of our pilot study was promising, 50% indicated willingness to participate, and our power analysis was based on these results. However, when the trial actually started, only one woman of the first 116 candidates accepted randomisation. The reason for this unexpected and almost unanimous refusal might be that when it comes to a hypothetical scenario, people prefer to give the socially most acceptable answer, while when dealing with real-life decisions, they answer truthfully. Finally, nulliparous women are preparing for the unknown and might therefore be more reluctant to let others decide on the place of delivery.


The decision not to participate in a trial that randomly allocates nulliparous women to either a home birth or a home-like short-stay hospital birth may be explicable by the fact that women have already decided on their place of birth before their first visit with their midwife. These women highly value their autonomy of choice and do not want others to make these decisions for them, and these factors are difficult to influence.

Healthcare professionals must acknowledge that women value and deserve autonomy of choice, but such choice is only valid if it is based on reliable information about the risk–benefit ratios of their choice. Without such information, choice is not really meaningful because it is not properly informed. Given the repeated publication on the uncertainties of safety of home and home-like hospital births,16–19 there is a need to explore why there is such certainty of view among women having their first child. Until we have an understanding of why women select information to make these choices and why women are reluctant to participate in trials that challenge choice, it may well be impossible to mount a trial of place of birth.

Disclosure of interest

All authors declare that they disclose any financial and personal relationships with other people or organisations that could inappropriately influence (bias) their work.

Contribution to authorship

J.N., J.S., M.N. and M.H. conceived the idea for the study. F.N., J.N., J.S., M.N. and M.H. developed the questionnaire. D.M., M.V.H. and M.H. were involved in data collection and they wrote the first version of the manuscript. Each author edited and approved the final version of the manuscript.

Details of ethical approval

Medical Ethical Committee Maastricht project number MEC 04-234.


This study was funded by Maastricht UMC+ (project number PF197).

Commentary on ‘Why women do not accept randomisation for place of birth: feasibility of a RCT in the Netherlands’

Hendrix et al.1 have addressed an important question related to maternal acceptance or declination of participation in a trial of birthplace. This question is timely, given current professional debates, and the authors should be congratulated on addressing this issue.

Within this study, the authors used questionnaires as their chosen method of data collection. The questionnaire used was developed by them and described as ‘ad hoc.’ The authors rightly acknowledge the limitations of the tool and allude to several important considerations required for questionnaire validation.

Increasingly, questionnaires are used within obstetrics, either as a primary or supplementary data collection tool. However, the reliability and validity of questionnaires are areas that continually challenge researchers, reviewers and readers. As a consequence, all too often there appears to have been less effort put into the questionnaire development than other aspects of study design. A social survey is a complex and often arduous process, but its development is pivotal to the success of a study and the meaningfulness of the results.

Within this commentary, it is impossible to discuss the do’s and don’t of questionnaire design. However, I would like to alert the readers to some important considerations, highlighted by this paper. In the first instance, one needs to determine the difference between descriptive surveys and analytical designs. This study reports the former; the latter are designed specifically to explore the associations between particular variables and would require formal analysis of validity. Validity indicates the degree to which an instrument measures what it is supposed to measure.

Content validity was used in this study. In its simplest form, this means that it seeks to establish that the items or questions are a well-balanced sample of the content area to be measured. Hendrix et al. used their previous research to inform this process. Content validity can be demonstrated through a variety of sources, such as previous research reports, government recommendations, local audit reports, discussions at professional forums and views of members of the target population. In essence, the researcher is asking ‘what information do I have to inform this questionnaire?’ If the information is insufficient, then one has to gather more information, for example by conducting focus group interviews. Triangulation of a number of sources will usually strengthen the validity.

The second consideration is reliability. In questionnaire development, this includes the characteristics of the instrument and the conditions under which it is administered; both of these must be consistent. Pilot work can assist in confirming the reliability, something that was acknowledged as a limitation of this study. Reliability may be measured in many ways, for example by repeat administration of the same questionnaire to the same sample within a short period of time (test–retest reliability).

I would urge researchers, when designing studies, to give questionnaire development adequate care and attention. It is not something to be taken on lightly and for those without the experience, I would suggest seeking help from those who do. As stated by Oppenheim,2 questionnaires do not emerge fully fledged; they have to be created or adapted, fashioned and developed to maturity after many abortive test flights’ (p. 47).

Disclosure of interest

I have no know conflict of interest.

Tina Lavender
University of Manchester, School of Nursing, Midwifery and Social Work, Manchester, UK


  • 1Hendrix M, Van Horck M, Moreta F, et al. Why women do not accept randomisation for place of birth: feasibility of a RCT in the Netherlands. BJOG 2009;116: 549–66.
  • 2Oppenheim AN. Questionnaire Design, Interviews and Attitude Measurement. London: Continuum International Publishing Group Ltd, 2000.

Commentary on ‘Why women do not accept randomisation for place of birth: feasibility of a RCT in the Netherlands’

There is a natural reluctance by clinical trialists to disapprove of refusal by patients to enter their trials and a desire to change such responses. My first reaction as an ethicist is to advise caution: there is a real risk that the fraternity of clinical researchers will make things worse not better if they are perceived as trying to prevent patients from exercising their own free choice in deciding whether to participate in clinical trials. This paper reports that 88% of the women responding thought that ‘researchers should not meddle’ if women had already decided where they wanted to give birth.

That said, there are features of the particular study, applicable more generally to recruitment for clinical trials, that are worth considering based on thorough and honest disclosure and a correlated respect for patients’ autonomy, which nonetheless might yield a higher level of recruitment.

The difference between likely recruitment shown in the pilot study and the actual lack of recruitment in the abandoned randomised controlled trial (RCT) is startling. The authors speculate that answers about hypothetical scenarios might differ from answers given in ‘real life’. Another possible explanation is that recruitment to the abandoned trial might have been more off-putting than ethically necessary or desirable. Apparently, information about the trial was given to potential recruits first by their midwives and then a week later by a researcher, but we are not told whether the information and the way it was given was neutral, negative or positive towards participation in the trial. It is entirely ethically justified to encourage potential recruits to sign up for clinical trials, provided of course that such encouragement is not threatening, misleading or coercive. In future similar trials it might be worth establishing a positively encouraging recruitment process and then checking on the views of the potentially participating midwives and researchers. Do they approve of the trial, do they approve of encouraging patients to volunteer for the trial and do they have strong views about where women should have their babies? It would seem to me to be ethically justified to exclude any such bias, if present, by excluding any midwives and researchers from the recruitment process who disapproved of the trial or of encouraging women to volunteer or who had strong preferences for one or other type of birthplace. In addition, it would be worth emphasising to potential recruits that the reason for the trial is that we really do not know which is safer, nor by how much, home or hospital birthing and that by participating in an RCT they would be helping to provide pregnant women in the future with reliable information upon which to base their choices.

But if, despite such ethically desirable encouragement, women refuse to participate in RCTs their refusal should be accepted with good grace, their involvement in nonrandomised cohort studies welcomed with gratitude, and the Declaration of Helsinki recalled, with its emphasis on the Hippocratic moral norm of medicine, which puts the interests of the individual patient ahead of the interests of science and society.

Disclosure of interest

Nothing to declare.

Raanan Gillon
Emeritus Professor of Medical Ethics, Department of Primary Care, Imperial College, London, UK