This prospective, interventional study was conducted at the Department of Gynecology and Obstetrics of Ziekenhuis Oost-Limburg (ZOL, Genk, Belgium) in collaboration with Hasselt University between March 2021 and May 2024 (ClinicalTrials.gov ID: NCT04031430 – CAPROM sub-study). One of the secondary research objectives was to evaluate beliefs about medication and adherence during pregnancy and in the postpartum period, and to explore the relationship between medication adherence and self-reported aspirin intake during pregnancy with pregnancy and neonatal outcomes.

Pregnant women at risk for GHD who were followed up via remote blood pressure monitoring as part of the Pregnancy Remote Monitoring (PREMOM II) project (Lansse et al., 2020) at Ziekenhuis Oost-Limburg (ZOL, Genk, Belgium) between March 2021 and May 2023 were eligible for participation in this study.

Low-dose aspirin (Asaflow^©) 160 mg once daily was prescribed to all subjects and was initiated before 16 weeks of gestation and continued until 36 weeks of gestation, consistent with the Aspirin for Evidence-based Pre-eclampsia Prevention Trial (Lansse et al., 2020; Rolnik et al., 2017).

All participants were asked to install the application Medisafe^© (MediSafe Inc.) on their smartphones to register their daily medication use during pregnancy. Medisafe^© is a free medication schedule and reminder app compliant with the General Data Protection Regulation and ISO 27001:2013 certified and is available in Dutch. The applicability and feasibility of Medisafe^© for medication management have already been demonstrated in the literature (Morawski et al., 2017; Huang et al., 2019) but were not previously used in the pregnant population.

In the Medisafe^© app, medication intake could be registered as “taken,” “missed,” or “skipped.” If the medication was “taken” or “skipped,” subjects had to actively press the corresponding button in the app. Medication reminders could be snoozed for 10 min with a maximum rappel of 4 reminders per medication, after which the medication intake was automatically registered as “missed.” The research team only accessed the Medisafe data at the end of the study and thus did not interfere in case of non-adherence during the study period.

Women’s overall medication beliefs and adherence were measured using two standardized questionnaires, with approval from the original authors. First, the beliefs about medication use in general and during pregnancy were assessed using the “Beliefs about Medicine Questionnaire” (BMQ) (Nordeng et al., 2010; Ceulem et al., 2019; Horne et al., 1999). Second, women’s medication adherence was assessed using the Probabilistic Medication Adherence Scale (ProMAS) (Kleppe et al., 2015). All study participants were asked to complete the Dutch-language version of the BMQ and ProMAS at three time points: during pregnancy (at inclusion), between day 10 and day 21 postpartum, and between 4 and 6 months postpartum. The study participants received a personal link via email to complete the questionnaires online (via Castor EDC) in their home setting at the corresponding measurement time. Reminders were sent after approximately 2 weeks of non-completion, with a maximum of 4 reminders per time point.

The following maternal demographics and clinical characteristics were recorded at inclusion: age, (pre-gestational) weight and BMI, race, smoking status, family history of cardiovascular and gestational hypertensive disorders, medical history of relevant co-morbidities (such as chronic hypertension, thrombophilia, kidney disease, diabetes mellitus type II, thyroid problem), medication use during pregnancy (at and after inclusion), own birth weight, information regarding the current pregnancy (estimated date of delivery, single or multiple pregnancy, mode of conception, gravidity, parity, spontaneous or induced abortion), and prior pregnancy complications among multipara.

The following pregnancy and neonatal outcomes were extracted after delivery from the electronic patient files: date of delivery, gestational age at delivery, mode of delivery, diagnosis of GHD according to the ISSHP guidelines (Brown et al., 2018), neonatal birth weight and percentile, sex, admission to the neonatal intensive care unit (NICU), and neonatal mortality. Birth percentiles were calculated based on the Belgian curves published by the Studiecentrum voor Perinatale Epidemiologie (SPE, Brussel, 2009).

The statistical analysis was performed with IBM SPSS^® Statistics. The study population, BMQ results, ProMAS results, and self-reported aspirin intake were analyzed using descriptive statistics. Continuous data are presented as mean ± standard deviation (SD) or median (interquartile range), depending on the normality of the data. Normality was checked via the Shapiro-Wilk test. Categorical data are presented as numbers (%). Comparison of the paired BMQ and ProMAS sum scores at the different time points (i.e., during pregnancy and in the postpartum period) were evaluated using the repeated measures ANOVA and the Wilcoxon signed rank Test for continuous and categorical (ordinal) data, respectively. Spearman’s Rho correlation analyses were performed between: the ProMAS adherence sum score assessed during pregnancy and in the early postpartum phase (10–21 days postpartum) and the self-reported aspirin intake; and the pregnancy BMQ sum score of the general statements (harm and overuse) and the self-reported aspirin intake. Pearson correlation analysis was performed between the BMQ sum score of the general statements (harm and overuse score) and the ProMAS adherence sum score assessed during pregnancy. The relationship between the ProMAS results assessed during pregnancy and in the early postpartum phase (10–21 days postpartum), the self-reported aspirin intake, and pregnancy and neonatal outcomes were explored descriptively and tested via one-sided Independent Samples T-test or the Mann-Whitney U test for continuous data, depending on the normality, and via the Chi-squared test for categorical (binary) data. A p-value <0.05 was considered statistically significant.

The study population involved 73 pregnant women at risk for GHD followed-up via prenatal remote blood pressure monitoring who completed the BMQ and/or ProMAS during pregnancy, and/or used the Medisafe app at least once for registration of their aspirin intake (Figure 1).

Detailed demographics and clinical characteristics of the total study population at inclusion are listed in Table 1. The median gestational age at inclusion was 14.1 weeks (IQR 13.2–15.6). A total of 84.9% of the women were nulliparous. All multipara (15.1%) had previous pregnancy complications. The mean maternal age was 30.1 ± 5.2 years, and the mean pre-gestational BMI was 29.1 ± 5.7 kg/m², with the largest group (45.2%) being categorized with a BMI ≥30. At inclusion, 9.6% were still smokers in the ongoing pregnancy and 17 subjects (23.3%) reported pre-existing co-morbidities. A total of 21.9% used antihypertensive medication, and 9.6% reported no other medication intake apart from aspirin during the index pregnancy (Table 1).

Four women had a termination of their pregnancy because of intrauterine fetal death (n = 2), premature preterm rupture of membranes before 24 weeks of gestation (n = 1), or extreme fetal growth restriction (n = 1). Hence, pregnancy and neonatal outcomes are reported for 69 subjects (Table 2). The mean gestational age at delivery was 39.0 ± 1.5 weeks, with a mean birth weight of 3,212.1 ± 504.3 g. Four pregnancies (5.8%) ended prematurely (<37 weeks), and six (8.7%) small-for gestational age (SGA) neonates were born. The majority (65.2%) of the pregnancies were uncomplicated, although 27.5% and 4.3% developed GH or LPE, respectively. Almost 54% of the deliveries were induced and 11.6% of the neonates were admitted to the neonatal intensive care unit immediately after birth.

A total of 70 subjects completed the BMQ (general and/or pregnancy-specific statements) and/or the ProMAS during pregnancy at a mean gestational age of 17.0 (±4.6) weeks (min 12.4 - max 32.7 weeks). Of these, 50/70 (71.4%) completed both questionnaires during pregnancy. At the time of completion of the BMQ and/or ProMAS, 82.9% (58/70) of the women were already taking aspirin. The remaining group of women received the prescription for aspirin on the same day of completion of the questionnaire(s). Furthermore, 77.1% (54/70) of the women were taking other concomitant medication and/or supplements apart from aspirin at the time of completion of the questionnaire(s) (Supplementary Table S1).

After delivery, 48/70 (68.6%) and 39/70 (55.7%) subjects completed the BMQ and/or ProMAS at an average of 24.2 (±16.8) days and 5.89 (±1.1) months postpartum, respectively. Women completed the questionnaires at either one or both times postpartum. Thirteen women (18.6%) completed both the BMQ and ProMAS at all three time points.

No statistically significant correlation was found between the BMQ general statements (harm and overuse, p = 0.499 and p = 0.649, respectively) and the ProMAS adherence scores assessed during pregnancy (N = 56).

A total of 67 (91.8%) subjects used the Medisafe app at least once to report their aspirin intake during pregnancy. The Medisafe app was used from 15.8 ± 4.3 weeks until 34.4 ± 3.8 weeks of gestation, with a mean duration of 18.6 ± 5.6 weeks. Thirteen (19.4%) subjects did not use the Medisafe app until the prescribed end date of aspirin intake (i.e., 36 weeks of gestation), of which four women had a pregnancy termination and one subject delivered preterm. The mean “minimal” and “maximal” self-reported aspirin intake was 82.5% (min 4.2%, max 100.0%) and 98.6% (min 79.4%, max 100%), respectively. A “minimal” aspirin intake level of ≥80% and ≥90% was reported in 74.6% (50/67) and 53.7% (36/67) of the women, respectively. With respect to the “maximal aspirin intake”, only one woman reported an intake level of <80% (i.e. 79.4%) and 97.0% (65/67) of the women had an intake level of ≥90%.

This study showed that pregnant women at risk for GHD have positive beliefs about medication use in general, which is in line with other studies reporting positive attitudes toward medicines in general in the pregnant population (Nordeng et al., 2010; Ceulem et al., 2019; Zaki and Albarraq, 2014; Tefera et al., 2020). The positive perceptions of medication use observed in our at-risk study population also remained in the postpartum period. The mean scores for overuse and harm showed no statistically significant difference between pregnancy, at 10–21 days postpartum, and at 4–6 months postpartum. Based on these results, at the population level, the pregnancy or postpartum status may not change women’s overall perception of medication use. However, individual changes throughout the perinatal period cannot be ruled out and require clinical attention.

While 95% of the women at risk for GHD agreed to have a higher threshold for taking medicines during pregnancy, only 37% agreed to refrain from using medicines during pregnancy. Higher thresholds for taking medicines in pregnancy have previously been reported multiple times (Nordeng et al., 2010; Ceulem et al., 2019; Zaki and Albarraq, 2014). However, in the “general” pregnant population, including both low- and high-risk pregnancies, more women (58%–78.9%) agreed that it is better for the fetus to refrain from using medicines during pregnancy (Nordeng et al., 2010; Ceulem et al., 2019; Zaki and Albarraq, 2014; Tefera et al., 2020). Despite a limited sample size and the high percentage of nulliparous women, our study showed that an at-risk pregnant population enrolled in a clinical trial context may be more aware of the potentially beneficial effects of medicines used during pregnancy compared to the general pregnant population. This could be the result of our study population having been well-informed about their risk of developing GHD and the importance and effectiveness of prophylactic aspirin intake in pregnancy. The importance of being well-informed by a healthcare professional on medication adherence during pregnancy has been reported in the literature (Vinogradov et al., 2024; Nordeng et al., 2010; Ceulem et al., 2019; Mortelmans et al., 2024). However, similar perceptions regarding the use of natural remedies during pregnancy were found between our at-risk and the general pregnancy population and require careful consideration from clinicians as risks of herbal products in pregnancy cannot be ruled out (Ceulem et al., 2019). In the postpartum period, intra- and interindividual differences were noted for certain statements compared to during pregnancy. This highlights again that clinicians, when providing counseling, should be well aware of the likelihood of differences and potential changes in women’s perceptions of medication during and after pregnancy (Bjertrup et al., 2025).

In our study cohort, 55.8% reported (medium)-high medication adherence rates during pregnancy, with a mean score of 10.3 ± 3.9 out of 18. Similar adherent rates were observed at 10–21 days postpartum but were lower at 4–6 months postpartum compared to pregnancy levels. This decline may be explained by the importance that these at-risk women attribute to medication use in pregnancy, with perceived benefits not only for themselves but also for their unborn child, in contrast to the (late) postpartum period when GHD and associated risks for the infant are no longer applicable. Another possibility is that over time, pregnancy experiences gradually fade away while women are regaining pre-pregnancy feelings and perceptions (Bjertrup et al., 2025). Furthermore, most women continue to take “pregnancy-related” medication early postpartum (e.g., antihypertensive medication, anticoagulants, etc.), which is no longer needed at 4–6 months postpartum (Brown et al., 2018). The lack of a significant result for the decrease in ProMAS sum scores during and after pregnancy could be due to the relatively small sample size in our study.

At the individual level, some women reported higher medication adherence sum scores late compared to early postpartum. Despite the wide variation in PROMAS scores among individual participants, a very high to maximum level of self-reported aspirin adherence in pregnancy was observed in this cohort. The mean “minimal” self-reported aspirin intake was 82.5%, while three-fourths of the pregnant women reported an intake level of ≥80%. Importantly, the calculation of the minimum self-reported aspirin intake clearly entails limitations (i.e., “missed” registrations could mean that aspirin was taken but not recorded in the app), potentially underestimating the actual adherence. In contrast, the maximum self-reported intake may provide a more reliable estimation of the actual adherence, approaching 100% for nearly all women. The high adherence rate observed in our cohort is likely an overestimation of the situation in the real world, caused by the inclusion of women willing to participate in a trial and to communicate about their medication adherence while being aware that their adherence was monitored. The use of the app, with its automatic notifications timed for aspirin and medication intake, may also have improved the adherence or at least minimized the likelihood of forgetting aspirin intake. Other cohorts in a clinical trial setting on aspirin compliance also reported high adherence levels (Rolnik et al., 2017; van Montfort et al., 2020).

Finally, a positive correlation was found between ProMAS scores, both during pregnancy and in early postpartum, and the self-reported aspirin intake during pregnancy. Although further research is needed, this finding paves the way for the potential clinical use of the ProMAS instrument in early pregnancy as a predictive tool for medication adherence in the (at-risk) pregnant population. Equally important, a trend toward higher ProMAS scores was noted among women with uncomplicated pregnancies compared to women experiencing complications such as GHD, preterm birth, or SGA. This intriguing observation also points to the potential value of assessing women’s medication adherence early in pregnancy with the ProMAS instrument, as it may help identify pregnant women at risk for adverse pregnancy and neonatal outcomes. This will enable timely clinical and behavioral interventions to reverse the course.

To the best of our knowledge, this was the first study assessing beliefs about medication and adherence among women at risk for developing GHD. Medication adherence was measured using both a standardized, generic instrument (ProMAS) as well as self-reported registrations of aspirin intake in the Medisafe^© app. Moreover, the BMQ and ProMAS instruments were used longitudinally at three time points during pregnancy and postpartum. This longitudinal follow-up provided the first insights into individual medication beliefs and adherence trajectories throughout the perinatal period. To evaluate the evolution in women’s beliefs about medicines during pregnancy, the pregnancy-specific statements of the BMQ were also administered in the postpartum phase, instead of the lactation-specific statements. The self-reported adherence rates of aspirin intake were not checked by the researchers during the study period, and the lack of intermediate intervention or interference can neither have indirectly nor unintentionally increased women’s medication adherence, further enhancing the objectivity of the results. Finally, the Medisafe app has a high ranking in terms of engagement, functionality, desirable characteristics, and overall quality (Carmody et al., 2019; González de León et al., 2021). The app is readily available to the public and can be used outside of study contexts. Study participants indicated that using the app was easy, straightforward, and user-friendly.

This study also has some limitations. A first limitation concerns selection bias and the lack of external validity of the findings. The study cohort consisted of a rather limited sample of women at high-risk of developing GHD who were part of a clinical trial and, hence, received additional care and follow-up, which may also be reflected in the favorable pregnancy and neonatal outcomes observed in this cohort. Participants may have been more motivated to use their medication correctly, already before, as well as during their study participation (Hawthorne effect or information bias). They were also well-informed about their risk and consequences of GHD and the beneficial effect of aspirin as a prophylaxis for PE, potentially leading to high (er) adherence rates compared to the real-world setting. This effect may have been further enhanced by the use of the Medisafe app and its notifications, which likely minimized instances of forgetting to take aspirin. Furthermore, the study design did not allow for a control group without the reminder function of the app, which might have increased overall adherence. Second, with regard to the socio-economic background of the participants, no details on education level or occupation (e.g., occupation in healthcare) were available, although both variables have been described as main determinants of medication beliefs (Nordeng et al., 2010; Ceulem et al., 2019; Zaki and Albarraq, 2014). Third, the ProMAS instrument was used for the first time in the pregnant population, and its statements were not adapted for ‘aspirin’ as the medication of interest. Hence, our results are exploratory, should be cautiously interpreted, and require further investigation. Furthermore, a small number of women (17%) completed the ProMAS on the same day aspirin was prescribed. Fourth, only a minority of the participants completed both instruments at all time points, limiting the internal validity of the findings related to the evolution of beliefs and adherence in the perinatal period. Finally, medication intake was not questioned postpartum.

The study could be replicated with a large(r) sample size of women at risk for developing GHD. The study should preferably be performed in a ‘real-world’ context outside the setting of a clinical trial. Future studies should also focus on the clinical value of the ProMAS instrument by assessing the relationship of the ProMAS scores early in pregnancy with self-reported intake of medication/aspirin in pregnancy, and pregnancy and neonatal outcomes. Since this study only used the MediSafe^© app for the first time in the pregnant population, further research is warranted, e.g., to investigate if the use of the app improves medication adherence among (at-risk) pregnant women by providing reminders, offering education, and facilitating communication between patients and healthcare professionals (Morawski et al., 2018; Santo et al., 2017). This is also relevant from a public health and economic perspective as medication reminder apps may have the potential to reduce healthcare costs as a result of improved clinical outcomes (Santo et al., 2017).

Women at risk for GHD have positive beliefs about medication use in general and during pregnancy throughout the perinatal period. While the women’s overall medication beliefs did not seem to be influenced by their pregnancy, a higher awareness of the beneficial effects of medicine use in pregnancy was observed in this specific population. Medication adherence varied across individuals, however, almost maximum adherence rates for aspirin intake during pregnancy were observed, alongside a trend towards decreased adherence in the postpartum period. ProMAS adherence scores during pregnancy and early postpartum positively correlated with self-reported aspirin intake during pregnancy, while higher ProMAS scores were found among women with uncomplicated pregnancies compared to those with complications. These findings are exploratory and underscore the need for further research to evaluate the potential, clinical utility of the ProMAS instrument as a screening tool in early pregnancy to assess medication adherence and identify risks for adverse pregnancy and neonatal outcomes.