Determinants and underlying causes of frequent attendance in midwife-led care: an exploratory cross-sectional study

The care for pregnant women in the Netherlands is organized as follows: low-risk pregnant women receive prenatal, natal and postnatal care from a midwife, high-risk women receive care from an obstetrician while women with intermediate risk received joint care. In case of a change in the medical condition of a low-risk woman, an obstetrician is consulted and in case of a determined medical condition, the woman is referred to an obstetrician, who will provide care during the remaining pregnancy period and birth. Otherwise, the woman is referred back to the midwife.

Women with psychological problems are often referred to the Psychiatrist-Obstetrician-Pediatrician-clinic, an outpatient setting where pregnant women are evaluated by obstetricians, pediatricians, and psychologists and receive advice from this multi-disciplinary team. For the clarity of this paper, the word ‘visit’ is used related to midwifery-led care, the word ‘consultation’ is used related to consultation of an obstetricians and other specialists in hospitals.

This exploratory cross-sectional study was conducted in a large midwifery practice, located in a medium-sized city near Amsterdam, the Netherlands. The practice was organized in two teams, each consisting of four midwives. On the 25th of January 2018, the Medical Ethics Committee of the VU University Medical Center (ref. 2018.019) approved this study to collect and analyse retrospective data of January 2015 to January 2017. Women gave permission to use their anonymous data, their verbal consent was obtained during the intake and noted in their files while they watched; the beforementioned Medical Ethics Committee approved this procedure of obtaining consent.

Eligible participants were pregnant women at the specified midwifery practice during the study period with an ongoing pregnancy. Women with a miscarriage and women who only received postnatal care were excluded. Included participants were classified based on the level of health care utilization according to the Kotelchuck Index-Revised, see Appendix [3, 12].

In brief, the Kotelchuck Index-Revised, a validated tool for measuring the level of care utilization, is based on the number of face-to-face visits with the midwife, adjusted for gestation and onset of care and aligned with the number of recommended face-to-face prenatal visits by the KNOV. Utilization of care was defined as inadequate (less than 50% of the recommended visits), intermediate (50–79%), adequate (80–109%) and adequate plus (110% or more) [3, 12]. In this study, women with adequate plus utilization of care were considered to be frequent attendance (cases) and were compared to women with the adequate utilization of care (controls). Women with the intermediate or inadequate utilization of prenatal care were excluded from the study because these groups were irrelevant in view of the goal of our study.

The digital midwifery database was searched to identify eligible participants. Two researchers (JG and BW) extracted data from the database, which were anonymized and de-identified. In this database, the medical charts of all pregnant women guided by this midwifery practice are included. The chart includes medical care records of pregnancies, demographic characteristics, lifestyle and medical information of the pregnant woman. Possible reasons for frequent attendance were identified based on the additional information in the charts of frequent attenders documented by the midwife.

Based on the Andersen’s model, several variables concerning predisposing, enabling, need and health behavior factors were considered as potential determinants of frequent attendance [3, 15]. In this study, Predisposing variables consisted of sociodemographic factors such as age, ethnicity (Dutch, western non-Dutch, non-western non-Dutch), marital status (relationship yes/no), occupation (yes/no), and level of education. The level of education was subdivided into low (primary school and some vocational training), intermediate (secondary school and completed vocational training), and high (college and/or university). Enabling variables included deprivation (yes/no), year of the start of prenatal care (2015/2016), and team (team 1 or team 2); both teams consisted of four midwives. Needs variables consisted of perceived and evaluated needs. Perceived needs in this study were telephone consultations initiated by clients (numbers), psychosocial problems (no/yes in the past, yes at present), and sexual violence (no/yes). Evaluated needs included parity (nulliparous or multiparous), the start of pregnancy (spontaneous or assisted), duration of getting pregnant (in months), reasons for frequent attendance, attendance to health care professionals other than the midwife, the number of ultrasound examinations, and prescribed drugs. Reasons for frequent attendance were defined based on the information in clients’ dossiers. Consultation of other healthcare professionals was subdivided in visits with a general practitioner, a medical specialist e.g. an obstetrician or an internist, or a visit to the Psychiatrist-Obstetrician-Pediatrician-clinic (POP). Health behavior variables consisted of smoking (no/yes), alcohol consumption (no/yes), intake of folic acid (no/yes), and body mass index before pregnancy (BMI). We classified BMI according to the World Health Organization Classification of adults: underweight and normal weight, overweight, and obesity [16].

Descriptive analysis included contingency tables to summarize the study population. Categorical data were reported as absolute numbers and percentages. Normally distributed continuous variables were reported as means with standard deviations and non-normally distributed continuous variables as medians with interquartile ranges.

Logistic regression models were fitted to estimate the unadjusted and adjusted effects of frequent attendance associated with the variables described above. Variables with a prevalence of less than 5% were excluded from the regression analysis. The team was included in the models as a fixed effect because the two teams provided care to an equal number of participants. Age was re-coded to be centered around the mean to have a meaningful baseline. Backward elimination was performed to obtain the leanest model possible, where each time, the variable with the largest p-value was removed from the model until all variables were significant (p < .05). The model assumptions were checked and the final model was assessed on overall performance (Nagelkerke’s R2) and discrimination (specificity, sensitivity, and overall accuracy).

Referral to obstetrician was considered an effect modifier of prenatal healthcare utilization.3 Therefore, additional analyses were performed comparing referred and non-referred pregnant women with χ2 tests for categorical variables and separate logistic models were fitted on the subsets of non-referred and referred participants. All analyses were executed using SPSS (version 24).

The characteristics of the study population are shown in Table 1, column 2. The prevalence of frequent attendance in 2015 and 2016 were 19 and 24%, respectively. In our cohort, the average age was 29.2 years, 48% of the women were primipara was 48, and 53% had a non-Dutch background. Of all women, 96 (9%) participants reported to use prescribed drugs, mainly Selective Serotonin Reuptake Inhibitors and thyroid medication.

The underlying reasons for more than the recommended visits in the midwifery practice in the 243 frequent attenders, independent of referral, are shown in Table 2. The most reported reasons were being worried/having vague complaints (44%), followed by abdominal pain (15%), medical problems (13%), psychosocial problems, e.g. behavioral, emotional and/or relational problems (12%) and experiencing high pregnancy burden (8%).

Table 3 shows reasons for consultation with an obstetrician and reasons for referral to an obstetrician. More than half of the study population, 572/1053 (55%), had at least one consultation with an obstetrician, ranging from one to seven; the mean of the number of consultations for frequent attenders and adequate users was 2.7 (SD 2.97) and 1.9 (SD 4.83), respectively. Of the whole study population, 493/1053 (47%) were referred to an obstetrician. Half of the women referred, 252/493 (51%) had at least one prior consultation with an obstetrician. Frequent attenders were more likely to be referred compared to adequate users (respectively 54.7 and 44.4%; p = 0.005). Women of non-western non-Dutch origin were more likely to be referred compared to women of Dutch or women of western non-Dutch origin (respectively 51.6, 44.8, and 31.8%; p = 0.002). Women living in a deprived area were more likely to be referred, compared to women not living in a deprived area (respectively 55.8 and 44.6%; p = 0.003).

The final logistic regression model on the entire study population included enabling, need, and health behaviors variables, but no predisposing variables. Women who started their prenatal care in 2016, with an assisted start of pregnancy, a consultation with an obstetrician, a history of sexual violence, more than average ultrasounds, and were overweight had higher odds of frequent attendance, whereas women living in a deprived area had lower odds of frequent attendance (Table 4).

Analyses showed referral to an obstetrician as an effect modifier, therefore, we split the results by referral (Table 4).

The characteristics of the non-referred population (frequent attenders and adequate users) are shown in Table 1, columns 3 and 4.

The final logistic regression model on the subset of non-referred participants included only need variables: consultation with an obstetrician and a history of sexual violence. Women with one or more consultations with an obstetrician and women with a history of sexual violence had higher odds of frequent attendance (Table 4, columns 4 and 5).

The characteristics of the referred population (frequent attenders and adequate users) are shown in Table 1, columns 5 and 6.

The final logistic regression model on the subset of referred participants included enabling, need, and health behavior variables: deprived area, year, consultation with an obstetrician, psychosocial problems, number of ultrasounds, and BMI (Table 4, columns 6 and 7). Women living in a deprived area had lower odds of frequent attendance, whereas women who started prenatal care in 2016, with one or more consultations with an obstetrician, psychosocial problems in the past or present, who had more than average ultrasounds, and were overweight had higher odds of frequent attendance.

To provide the best possible care, a trustful midwife-woman relationship is needed for women to share more details in case of vague complaints or worries to identify their implicit needs. Furthermore, the presence of sexual violence should be explored at least twice during pregnancy, because only a third of women disclose a history of sexual violence during pregnancy. Such information can be used by midwives to attune the prenatal care to the woman’s specific needs. This implies that (student-)midwives have to learn to be flexible with the guidelines about the prenatal visits schedule and have to decide together with every individual pregnant woman when and how prenatal care is provided.