Antenatal care use in urban areas in two European countries: Predisposing, enabling and pregnancy-related determinants in Belgium and the Netherlands

A secondary data analysis is performed using pooled data from two studies. For Belgium, data were obtained from a prospective observational study conducted in the Brussels Metropolitan Region (the CTP study) [10]. Recruitment occurred between April and July 2008 in nine out of 12 hospital centres for ultrasound to which every woman is referred. All low risk women, at the beginning of their care trajectory (attending a first or second visit or having a gestational age less than 16 weeks) were elegible for inclusion. Data collection comprised a questionnaire about personal characteristics and pregnancy history at the moment of recruitment, a diary recording all antenatal visits in a structured manner (for each visits, 6 questions needed to be filled out, for each question closed answers were provided, women needed to copy the code related to their answer) and bimonthly (once in two months) telephone follow-up interviews to record ANC use (n = 333) [10]. This study was approved by all participating centers and from the Ethics Committee of the University Hospital UZ Brussel.

For the Netherlands, data were obtained from the DELIVER (Dutch acronym for ‘data primary care delivery’) study. Data were gathered in a 12 month study period in 2009-2010. The Deliver study is a descriptive study that aimed to provide information about midwifery care organization, accessibility of midwifery care, and the quality of primary midwifery care in the Netherlands [14]. Midwifery practices were recruited by using purposive sampling. Three stratification criteria were used: region (north, east, south, west), level of urbanisation (urban or rural area), and practice type (dual or group practice) to ensure that different types of practices in different regions were represented. Subsequently, all clients receiving care in the participating primary midwifery practices at any moment in a 12 month study period in 2009–2010 were eligible to participate if they were able to understand Dutch, English, Turkish or Arabic. The participating practices (20 of the 519 midwifery practices in the Netherlands) comprised 110 midwives and a caseload of 8200 clients per year, with all regions of the Netherlands being represented [14]. Data collection with regard to pregnant women recruited in primary care midwifery practices included up to two questionnaires about socio-demographic characteristics and ultrasound scans. One questionnaire was administered before 34 weeks of gestation and the other between 34 weeks of gestation and birth. In addition, information about antenatal care utilization was gathered by extracting data from electronic client records of participating clients. This study was approved by the Medical Ethics Committee of the VU University Medical Center Amsterdam.More study details can be found in the specific papers [10, 14].

To have comparable inclusion criteria for the secondary data analysis, only adult women (>18 years) residing in an urban region (2500 or more households per km²) with a low-risk onset of pregnancy (without pre-existing medical complications) were eligible for inclusion. Application of these criteria meant a reduction of the Dutch study sample to 632 women (Fig. 1). Because of the unbalance in the numbers between both samples, a pooled data set was constructed by combining the entire sample from the Belgian study (n = 333) and a random matched sample from the 632 women remaining in the Dutch study.

To reduce possible pre-existing differences in distribution between both populations, predictors for antenatal care use were used to define a comparable dataset. Our first step was to reduce missings in the Dutch data. Multiple imputation was performed for missing values with regard to household income (97/632), using the hot deck method [15]. Seen Chi-square analyses indicated that non-response concerning household income depends on a woman’s occupational status and educational level (p < 0.05). Missing values of non-respondents were replaced by observed values from a respondent similar to the non-respondent [16] for these variables. Five imputed data sets were generated to calculate the mean household income for each non-respondent. There were no missings in the Belgian study.

After completing the imputation in the Dutch sample we observed that the individual characteristics of both subsamples were distributed differently. Women in the Belgian subsample were significantly (p < 0.05) younger (aged ≤ 20), were more often single, more often less educated, less likely to be active on the labour market and were more often of a foreign nationality compared to the Dutch subsample. In the Belgian sample, women had more often a less educated partner (p < 0.05) and more often a partner with a foreign nationality. Furthermore, these women more often had a low and high equivalent income (p < 0.05) and lack of health insurance and additional health insurance cover. Finally, these women were more often multiparae (p < 0.05), had more unwanted pregnancies, more unplanned pregnancies and attended fewer antenatal information classes. These observed differences might potentially influence differences in health care utilisation, therefore exact matching without replacement [17] was conducted in order to balance the distribution of individual characteristics between the subsamples of the pooled data set. The units of the Dutch subsample were ordered at random and were matched 1:1 to the units of the Belgian subsample for two variables: educational level [6, 7, 9] and maternal age [5‐7]. These variables were chosen because in literature they were observed to be predisposing determinants of ANC use. For 321 women in the Belgian sample we were able to match with someone in the Dutch study. The final pooled dataset (n = 642) therefore consisted of 321 women from Belgium and 321 from the Netherlands.

The CTP tool (Fig. 2) considers three dimensions: the timing of initiation of care, and the number and timing of three specific interventions during pregnancy (blood screening, ultrasound and blood pressure measurement) [10]. Four categories of ANC use are defined by the CTP: inadequate, intermediate, sufficient or appropriate care. This classification reflects the degree to which a minimum amount of care recommended by national obstetric guidelines for every pregnancy was received, regardless of parity or risk status [10]. As the CTP was developed based on evidence about the importance of interventions in pregnancy and the congruence of ANC guidelines, the tool is applicable in the Netherlands [18‐21].

The original data collection instruments used in both studies were explored to determine the variables that had been equivalently operationalised. The common variables to form the predisposing component were age, marital status, educational level, occupational status and current nationality. In addition, educational level and current nationality of the partner were examined. A variable for region referred to the two original study samples: the Metropolitan Region of Brussels, Belgium and urban regions in the Netherlands. The educational level of all women was classified into three categories according to the International Standard Classification of Education (ISCED) [22].

The variables reflecting the enabling component were equivalent income, health insurance cover and additional health insurance cover. Equivalent income was calculated by using the modified Organization for Economic Co-operation and Development (OECD) scale and classified into three categories. This scale involves adjusting monthly household income based on its size and the age of its members [23]. The lowest income group was defined at < 60 % of the respective median national income [24], the at-risk-of-poverty threshold [25]. The moderate and high income groups were delineated at 60–120 % and > 120 % of the national median equivalent net income respectively.

The variables describing the pregnancy-related component were parity, wish for pregnancy, planned pregnancy, continuity of care and attendance of antenatal information classes. Continuity of care was measured by the Continuity of Carer (COC) index, based on the number of visits to each different health care provider and the total number of visits [26]. The index, expressed in percentage, was divided into two categories, with the cut-off point < 50 % and ≥ 50 %.

For each region, the individual characteristics of the study sample and ANC utilisation were summarised. Individual characteristics and ANC utilisation were compared between regions using Chi-squared tests, the association between each of the individual characteristics and ANC utilisation for the whole sample was determined (Chi-squared tests). Subsequently, logistic ordinal regression analysis was used to examine the significance of each individual characteristic in terms of its likelihood of being given a higher CTP classification, while controlling for the remaining significant characteristics. Since this was an exploratory study, backward elimination was used (stay level: p < 0.05) [27]. Our model was constructed in three steps in accordance with the health behaviour model [3, 4]. The first step considered predisposing variables, the second step considered enabling factors, with the selected predisposing variables fixed in the model, and in the final step the pregnancy-related variables were examined while controlling for the selected predisposing and enabling variables. In order to include other variations between the subsamples, the variable region was fixed in this model from the first step onwards. A Score test for the proportional odds assumption and absence of multicollinearity was undertaken for each step. In addition, the final model assessed the percentage of concordant pairs of predicted probabilities and observed responses (>60 %). Multivariate analyses were conducted in SAS 9.1, and all other analyses were performed using SPSS Statistics 20.