Explanatory factors for first and second-generation non-western women’s inadequate prenatal care utilisation: a prospective cohort study

Data from the national DELIVER study, a multi-centre prospective dynamic cohort study that aimed to evaluate the quality, organisation and accessibility of primary midwifery care in the Netherlands, was used for this study [18]. Primary midwifery care is maternity care provided by autonomous midwives to women with no complications or no increased risk for complications during pregnancy, childbirth and the puerperium. Of the 2852 practicing midwives in the Netherlands on the first of January 2013, 1621 - one per 10.000 women within the age range 15 to 39 - worked in primary care [19]. Between 1999 and 2008, 83.1% of the pregnant women in the Netherlands had their first prenatal appointment at a primary care midwife. Eventually, 35.8% gave birth under supervision of a primary care midwife [20].

The cohort consisted of primary midwifery care clients who had completed up to three questionnaires between their first prenatal appointment and six weeks postpartum. Clients could enrol at any moment in their pregnancy and were invited to respond to the appropriate questionnaire. The first or early prenatal questionnaire was completed before 35 weeks of gestation, the second or late prenatal questionnaire between 35 weeks of gestation and birth, and the third or postpartum questionnaire around 6 weeks postpartum. By completing a questionnaire, clients implicitly gave consent to participate in the study. Ethical approval was obtained from the medical ethics committee of the VU University Medical Center in the Netherlands (WC 008–100).

The women in this cohort were recruited between September 2009 and February 2011 by the midwives from 20 midwifery practices spread all over the country. These midwifery practices were selected according to three stratification criteria: region, level of urbanisation and practice type. A total of 14148 clients were invited to participate. Of these, 12398 were eligible. The response rate for at least one questionnaire was 62% [18]. For each participating client, questionnaire data was linked to data from the national Netherlands Perinatal Registry and the electronic client record in the midwifery practices by means of unique anonymous identifiers for the client and midwifery practice.

This study included native Dutch and non-western women (see the definitions in the description of the main independent variable) who had completed the early prenatal questionnaire of the DELIVER study. Because of our specific interest in non-western women’s prenatal care utilisation compared to that of native Dutch women, migrant women of western origin were excluded. From the literature it was obvious that non-western women were at higher risk for inadequate use. Furthermore, women lacking additionally linked data and women who did not start prenatal care with a primary care midwife were excluded.

Figure 1 shows a schematic draft of the selection process of the study population. Of the 5590 native Dutch and non-western women who had completed the early prenatal questionnaire of the DELIVER study, 3749 (67.1%) had additionally linked data on prenatal care utilisation. Of these, 3300 women had started prenatal care with a primary care midwife. Table 1 shows the characteristics of these 3300 women according to ethnic origin.

Of the 2998 native Dutch women, 234 (7.8%) made inadequate use of prenatal care – 6.5% entering late and 1.3% with an insufficient number of visits. Of the 203 first-generation non-western women, 45 (22.2%) made inadequate use of prenatal care – 20.7% entering late and 1.5% with an insufficient number of visits. Of the 99 second-generation non-western women, 16 (16.2%) made inadequate use of prenatal care – 14.1% entering late and 2.0% with an insufficient number of visits.

Table 2 shows the ethnic origin of the 302 non-western women. The majority (64.9%) belonged to one of the four major non-western groups in the Netherlands: Turkish, Moroccan, Surinamese, Dutch Antillean/Aruban. Turkey and Morocco supplied the Netherlands with large numbers of guest workers during the economic growth in the 60’s and 70’s. Suriname and the Dutch Antilles/Aruba are former Dutch colonies.

An index was compiled to assess women’s prenatal care utilisation as comprehensive as possible by taking both prenatal care entry and the number of prenatal visits into account (see Additional file 1). This index was derived from the Kotelchuck index [21], and adapted to the Dutch primary midwifery care context.

Prenatal care entry was determined based on the gestational age at the first prenatal visit (derived from the ultrasound scan or the first day of the last menstrual period) and classified into ‘on time’ (gestational age at onset < 12 weeks) and ‘late’ (gestational age at onset ≥ 12 weeks).

The number of prenatal visits was derived from the electronic client record, and compared to the expected number of visits derived from the guidelines of the Royal Dutch Organisation of Midwives (KNOV) [22]. The expected number of prenatal visits was based on the gestational age at which women gave birth. For women who were referred to secondary care, the expected number of prenatal visits was based on the gestational age at referral.

For this study, the modified index (see Additional file 1) was dichotomised into:

The main independent variable was women’s ethnicity, which was categorised into native Dutch, first-generation non-western and second-generation non-western. Ethnicity was established by asking women to fill in their country of birth and their parents’ in the questionnaire. The classification into native Dutch and non-western was made according to the definition used by Statistics Netherlands [23]. Women are considered native Dutch when both of their parents were born in the Netherlands, and non-western when at least one of their parents was born in Asia (excluding Indonesia and Japan), Africa, Latin America or Turkey. Non-western women were subdivided into first generation (born outside the Netherlands) and second generation (born in the Netherlands).

Based on Andersen’s healthcare utilisation model [24] and the conceptual framework of Foets et al. [12], several factors were considered as potential explanatory variables for the association between ethnicity and prenatal care utilisation. These variables were derived from the early prenatal questionnaire of the DELIVER study and assigned to the following blocks of conceptually linked variables:

Data analysis consisted of several stages. Firstly, descriptive analyses were carried out on the independent and dependent variables. Secondly, univariable logistic regression analyses were carried out to determine the association between the potential explanatory independent variables and the dependent variable. Only potential explanatory independent variables associated with the dependent variable (p < 0.25) were retained in the corresponding category for further analyses. Thirdly, blockwise multivariable logistic regression analyses were conducted, because of the dichotomous outcome measure and the focus on groups of explanatory factors [25]. The first block consisted only of the main independent variable, namely ethnicity. After this, six separate logistic regression analyses were conducted, adjusting each time for one block of explanatory variables. In addition to these separate logistic regression analyses, a blockwise structured model was constructed by adding the blocks of explanatory independent variables one by one. The blocks of explanatory variables most proximate to the individual were added first and those most distant last, until a final model consisting of the main independent variable and all six blocks was constructed. Besides these ordinary univariable and blockwise mutltivariable logistic regression analyses we also conducted univariable and blockwise multivariable multilevel logistic regression analyses. Latter were conducted to take the hierarchical nature of the data, i.e. clients (level 1) nested within midwifery practices (level 2) into account. If the multilevel model resulted in a significantly better fit than the ordinary univariable or blockwise multivariable logistic regression model, the former was preferred and presented. Lastly, the percentage change in odds ratio (OR) was calculated for each model, using the following formula: (((OR _{(model 2–13)} - OR _{(model 1)})/(OR _{(model 1)} - 1)) ×100) [26].

Using the percentage changes in OR, the overall and direct contributions of each block of explanatory variables to inadequate prenatal care utilisation were calculated with the following formulas:

All analyses were conducted in IBM SPSS Statistics (version 20.0), except for the multilevel analyses which were conducted in Stata (version 12).

Table 3 shows the results of the univariable analysis. Fifteen of the 21 potential explanatory variables were significantly associated with prenatal care utilisation and retained in the corresponding category for use in the blockwise multivariable logistic regression analyses. The potential explanatory variables excluded from further analysis were alcohol use, fear of bearing a handicapped child, pregnancy-related concerns about appearance, perceived health status and accessibility of the midwifery practice (calling the practice and visiting the practice).

Table 4 shows the results of the multivariable logistic regression analyses. The unadjusted odds of first and second-generation non-western women making inadequate use of prenatal care are 3.26 (95% CI 2.13 - 5.00) and 1.96 (95% CI 1.08 - 3.57) times greater than for native Dutch women.

A major strength of this study is the national dataset containing a large number of variables. Another strength is the adjustment of the Kotelchuck index in line with the guidelines of the Royal Dutch Organisation of Midwives (KNOV). A Kotelchuck index adjusted to the guidelines of the Dutch Society of Obstetrics and Gynaecology (NVOG) already existed [12], but this seemed less appropriate as we were focusing on primary midwifery care. The selection and classification of the explanatory variables based on theoretical frameworks may also be considered as another strength. Despite of the strengths, some limitations need to be taken into account. Firstly, we were not able to distinguish different ethnic groups within the groups of first and second-generation non-western women because this would have resulted in small subgroups. Secondly, we were not able to compare first and second-generation non-western women’s prenatal care utilisation against each other, because the numbers were insufficient. Thirdly, the adjusted response rate of 62% with highly educated women being overrepresented in this sample (47.5%), may have led to an underestimation of the number of inadequate users and the contribution of socioeconomic factors.