Background

Globally, an estimated 2.3 million newborns died in 2022 [1], with 99% occuring in low-and middle-income countries (LMICs) [2]. Similarly, 99% of maternal deaths worldwide in 2017 occurred in LMICs, with sub-Saharan Africa (SSA) accounting for 66.3% [3]. The first target of Goal 3 of the Sustainable Development Goals (SDGs) is to reduce the maternal mortality rate to less than 70 per 100,000 live births and the neonatal mortality rate below 12 deaths per 1000 live births [4]. Despite the substantial decline in maternal mortality, from 556 deaths per 100,000 live births in 2015 to 104 deaths per 100,000 live births in 2022, neonatal mortality remains a public health challenge in Tanzania, with a rate of 24 deaths per 1000 live births [5]. Prematurity and perinatal hypoxia are the leading causes of neonatal deaths in LMICs, including Tanzania [6]. Notably, most stillbirths occur at term and during labour [7], highlighting an urgent need for appropriate interventions aimed at the detection of antepartum fetal hypoxia to prevent adverse perinatal outcomes [8].

Previous studies indicate that existing methods employed in clinical practice do not significantly improve perinatal outcomes in LMICs [9, 10]. Therefore, it is postulated that continuous monitoring will enhance the likelihood of detecting signs of fetal compromise [10]. Various reliable devices for screening and continuous fetal monitoring to indicate a normal condition of the fetus and identify fetuses at risk for hypoxia exist within the literature [11,12,13]. However, the technology is often immobile and has high maintenance costs, making it unsuitable and inaccessible in low-resource settings [14]. In Tanzania and many other LMICs, fetal heart monitoring is mainly done by a Pinard fetoscope and occasionally by a fetal doppler, notwithstanding the severe shortage of staff. Innovative technology can improve the efficiency of care in LMICs amidst ongoing limitations and shortages of medical resources, including personnel [15].

Melody International Ltd. recently developed the iCTG, a mobile cardiotocography device designed to improve fetal surveillance in LMICs [16]. This device includes a wireless fetal heart monitor, a uterine contraction monitor, and a tablet, offering advanced technology that outperforms conventional devices while maintaining reliable measurements. The user-friendly iCTG helps healthcare providers in early detection, prompt intervention, and reducing emergency delivery risks, thereby potentially lowering perinatal deaths. Moreover, it enables real-time data transmission to a physician during patient transfers, allowing receiving hospitals to prepare in advance and improving survival chances. The iCTG’s introduction in Thailand has led to increased antenatal checkups for high-risk pregnancies, ultimately improving survival rates for both mothers and babies. For example, in 1,500 cases, 50 women were transferred to higher-level facilities, 10 complications were diagnosed, and five lives were saved [17]. The remarkable results of using iCTG in LMICs have resulted in its recognition as a world health organization (WHO) recommended medical device for low-resource settings [18].

In Tanzania and other LMICs, pregnant women continue to face challenges in achieving the recommended four ANC visits, increasing their risk for pregnancy-related complications, maternal mortality, and stillbirth [5, 8, 19, 20]. These challenges are linked to factors such as financial constraints, limited knowledge, perceived quality of care, service availability, distance to facilities, cultural beliefs and personal preferences [21]. A review indicated that at least one ANC visit by a skilled provider reduces neonatal mortality risk by 39% in SSA countries [22]. In Tanzania, it was found that women with four or more ANC visits had a higher likelihood of facility delivery (76.8%) and postpartum care (43.5%) [23] which contribute to better perinatal outcomes by facilitating timely interventions [8, 24].

Recognizing this, between October 2023 and May 2024, Tanzanian pregnant women living in semi-urban areas were offered antepartum fetal monitoring with iCTG at the primary health facilities. We hypothesized that pregnant women’s access to FHR monitoring with iCTG would influence the change in their health-seeking behaviour [25, 26]. The latter would lead to more ANC visits, timely detection of fetuses at risk of stillbirth and intervention, and hence, improved perinatal outcomes. ANC visits can prevent adverse perinatal outcomes by providing cost-effective interventions [27]. Therefore, the current study aim was to evaluate the effect of iCTG in monitoring fetal heartbeats during late pregnancy on the number of ANC visits, detection of abnormal FHR, mode of delivery, institutional deliveries, Apgar score, and perinatal outcomes.

Methods

Study design and setting

We employed a pragmatic, non-randomized clinical trial to evaluate the effectiveness of implementing iCTG for antepartum FHR monitoring across four primary health facilities in the Pwani region of Tanzania. The facilities were randomly assigned to either the intervention group (n = 2) or the control group (n = 2). We adopted a hybrid type 2 implementation to assess the effectiveness of fetal monitoring intervention [28]. The intervention was implemented for eight months between October 2023 and May 2024. This study was carried out in two districts: Kisarawe (101,598 people) and Bagamoyo (106,484 people) [29]. We purposefully chose the Kisarawe and Bagamoyo districts due to their rural and semi-urban population similar characteristics. Research shows that pregnant women living in rural settings have limited access to quality ANC [30, 31]. In addition, the 2022 Tanzania DHS indicates that the Pwani region had 60.5% of 4 + ANC visits coverage and did not attain a recommended percentage of at least 70% of 4 + ANC visits across districts [5, 32]. This implementation study was conducted in health centers that fall under primary-level health facilities. In these facilities, pregnant women who attend ANC should receive all available services at no cost. However, a woman who needs to undergo blood tests must contribute a minimum of 3000 Tshs (~ 1.5 USD) as cost sharing for service. Those who must undergo an ultrasound exam are directed to a facility that offers this service and must pay a minimum of 15,000 Tshs (~ 7 USD). These factors provided the opportunity to evaluate the effectiveness of implementing iCTG under standard ANC clinic conditions. The capacity of the selected health centers, including the resources available, staff competence, number of clients and providers, were comparable.

Study participants and sample size

Recruitment was completed between October 2023 and May 2024. We recruited pregnant women who were residents with gestational age ≥ 32 weeks and aged 16 years or above. We excluded pregnant women who did not speak either Kiswahili or English, as the study materials and interviews were conducted in these two languages. In addition, women who did not have access to a phone, whether their own, a partner’s, or a close relative’s, were excluded due to the need for follow-up communication. Finally, we excluded women who planned to travel outside the study site region for childbirth, as this would have limited our ability to conduct the follow-up survey 4 to 6 weeks after delivery. Before recruitment, participants were briefed about the study’s objective and had an opportunity to ask questions. We enrolled participants who provided written informed consent and collected the baseline data within health facilities. At intervention sites, participants received the standard care with an additional of 20–40 min of continuous FHR monitoring using iCTG, while those at control sites received the standard care. We estimated the sample size and power analysis using G-power software for experimental study designs. Based on the 2022 Tanzania Demographic and Health Survey [5], the proportion of pregnant women who attended 4 + visits in rural settings (p1) was 61%. We assumed that the availability of iCTG in intervention sites would increase ANC attendance beyond four visits by 9%; the proportion in the intervention (p2) was estimated to be 70%. A significance level was set at 0.05, with a power of 80% (1-beta = 0.8) and considering an attrition rate of 10%, the estimated sample needed in each group was 403 for a total sample size of 806.

Intervention

The intervention was designed based on the need assessment, informed by interviews with healthcare providers and service users, a literature review, and researchers’ experience with the barriers and facilitators of implementing FHR monitoring with iCTG devices to improve perinatal outcomes [33]. The intervention comprised two main components: 1) training of hand coaching healthcare providers and 2) Antepartum FHR monitoring using iCTG.

Training of healthcare providers

We trained 16 healthcare providers (doctors and midwives) on using iCTG for FHR monitoring—a four days training targeted at enhancing fetal monitoring knowledge, skills, and iCTG usage among trainees. The skills included applying, reading, and interpreting the CTG waveforms and decision-making on management based on the diagnosis. The training program included lectures, video demonstrations, group discussions, and bedside hands-on practice. We used the pre-and-post test to assess the knowledge and skills required to use the iCTG. We set a minimum score of 70% post-test for the trainee to use the iCTG for fetal heart monitoring.

Fetal heart rate monitoring using mobile cardiotocography (iCTG, Melody I Ltd)

We used the iCTG for antepartum FHR monitoring to assess the fetal condition of pregnant women at 32 gestational age or later. The iCTG device consists of a fetal heart monitor, a uterine contraction monitor, and a tablet for displaying CTG (Fig. 1). The iCTG can be used to check maternal and fetal conditions in various medical situations. The obtained data on fetal heartbeat patterns can be sent and assessed remotely by the midwife or doctor using a tablet via the Internet [16]. In this study, the iCTG was introduced at two health centers without an obstetrician to enable pregnant women who had to travel to a secondary or tertiary hospital for a non-stress test examination to receive it at the facility utilizing iCTG. A provider at these facilities examined a pregnant woman and interpreted the results. In case of detection of FHR abnormalities, the midwife recorded and reported to a doctor, followed by intrauterine resuscitation that included a change of maternal position, provision of fluids, and oxygen administration. The device also produced an alarm to alert the provider whenever there were persistent abnormal FHRs to ensure the abnormality was corrected promptly. When there was a need for referral, the client was transferred by ambulance while the iCTG continued to monitor her condition during transit. Additionally, the referring provider could call the receiving hospital in advance to prepare a management/intervention plan to reduce or prevent adverse perinatal outcomes.

Fig. 1
figure 1

Melody International Ltd. iCTG device

Full size image

Data collection procedures

The data collection was conducted between October 2023 and September 2024. We developed a structured questionnaire in English (Supplementary File S1). The translations to Kiswahili were carried out using a forward-backwards method to ensure clarity and accuracy before they were administered to participants. Then, we deployed the questionnaire survey into KoboToolbox Software for data collection. The questionnaire was pre-tested on a small sample to ensure clarity and validity before data collection. We trained four research assistants (RAs) with a background in nursing/midwifery who were not formally employed at the time of study to collect data. The training objective was to familiarise them with the study objectives, data collection tools, research ethics and procedures. We emphasized obtaining informed consent, ensuring participant privacy and confidentiality, and adhering to data collection protocols. During the training, RAs also participated in mock interviews to practice proper data recording techniques and effective participant engagement. Following the training, data collection using an electronically administered questionnaire began across the four health facilities. The socio-demographic characteristics and previous childbirth experience were captured during baseline interviews. ANC checkup factors were extracted from the ANC cards. In contrast, the outcome variables and newborn characteristics were collected during follow-up interviews (4–6 weeks after childbirth) at a health facility or extracted from health records (Supplementary File S2). Gestational age at the time of delivery was considered pre-term (< 37 weeks) and term (≥ 37 weeks and < 42 weeks). Perinatal outcomes were categorized as healthy, stillbirth (defined as an Apgar score of zero at 1 and 5 min), admission or referral after birth, and early neonatal death (END) (death of the newborn occurred within 24 h of life). Birth weight was categorized as low (< 2500 g), normal (2500–3800 g), and high (> 3800 g). Quality control measures were implemented to ensure data completeness and consistency. Data were securely stored and safeguarded to maintain confidentiality throughout the study.

Outcomes

The primary outcome of this study was the proportion of women who complete four or more ANC visits, which was assessed during the postpartum survey. We categorized it into four groups: < 4 visits, 4 visits, 5–6, and ≥ 7. The secondary outcome measures included abnormal FHR detection (FHR < 110 considered as bradycardia and FHR > 160 bpm considered as tachycardia), mode of delivery (vaginal vs caesarean section), institutional births, Apgar score at 5 min (< 7 considered low), and perinatal outcomes (healthy baby, admissions/referral, perinatal mortality (FSB and END)). A newborn having spontaneous breaths at birth, an Apgar score of 7 or above, with no immediate admission or referral requirement, was considered a healthy baby.

Covariates

We included the data collection of covariates in the baseline and follow-up survey. We considered socio-demographic characteristics, previous pregnancy experience, and newborn factors as potential covariates that may affect the outcome variables. The measurements were as follows maternal age (in years), gestational age (GA) (weeks), marital status (married, others), education (no formal education/completed primary education, completed secondary education and above), occupation (housewife, business entrepreneur, and farmer), and personal financial status (spending < 5000Tshs/day or ≥ 5000Tshs/day), the decision maker (pregnant woman, others), household property (no property, ≥ 1 property that included house, car, agricultural land, and cattle), a problem with distance to a health facility (yes or no), parity (nullipara, primipara, and multipara), GA at first ANC (< 13 weeks, 13–20 weeks, and > 20 weeks), health education session an individual received (< 2 times, 2–5 times, and > 5 times), maternal condition; anaemia, hypertensive disorders of pregnancy, infectious diseases (yes or no), complications during labour defined as occurrence of bleeding and/or seizures) (yes or no). Previous childbirth experience variables for multiparous women included an experience of danger signs included antepartum hemorrhage and/or severe features of pre-eclampsia/eclampsia observed during pregnancy (yes or no), place of birth (home, dispensary, health center, and hospital), mode of delivery (vaginal/caesarean section), and experience of childbirth complication included postpartum hemorrhage and pre-eclampsia/eclampsia diagnosed immediately after birth (yes or no). Characteristics of index babies included GA at the time of birth (pre-term: < 37 weeks, term: ≥ 37 weeks), birth weight (low: < 2500 g, normal: 2500–3800 g, and high: > 3800 g), and complication at birth referred to birth asphyxia or respiratory distress identified at or shortly after birth (yes or no).

Statistical analysis

All statistical analyses were performed using Statistical Package for Social Sciences (SPSS) version 28 (IBM Corporation, Armonk, NY, USA). Characteristics of participants were summarised according to the study group. Mean (SD) and proportion were used for descriptive statistics of participants’ characteristics and outcomes. Chi-square and Fisher’s exact tests were used to compare the proportion differences. An independent sample t-test was used to compare group mean for continuous variables. In addition, we treated the primary outcome (ANC visits) as an ordinal variable and analysed using the Mann–Whitney U test. We presented descriptive statistics as median and interquartile range (IQR). We built Poisson regression, binary, and multinomial logistic regression models to compare the rate ratios, odds ratio (OR), and 95% confidence interval (CI) of outcome variables according to study groups. We performed the unadjusted analyses and adjusted the models for relevant covariates. We present unadjusted and adjusted rate ratios and odds ratio (OR) with 95% confidence intervals (95% CI) from the Poisson regression and logistic regressions. A p-value < 0.05 was considered statistically significant.

Results

Description of study participants

Figure 2 presents the participant recruitment flowchart. Of the 806 participants who completed the baseline survey, 763 (94.7%) completed the follow-up interview 4–6 weeks post-delivery and were included in the final analysis. Among these, 492 were from the intervention group and 271 from the control group. Forty-three participants (5.3%) were lost to follow-up due to communication difficulties (unable to reach them by phone) and travel-related issues after childbirth.

Fig. 2
figure 2

Study Participants Flow Chart

Full size image

Table 1 shows the socio-demographic characteristics, pregnancy factors, previous experience, and index baby characteristics by groups. There were no significant differences in age, GA at enrolment, education, financial status, parity, history of hypertensive disorders of pregnancy, infectious diseases, GA at the time of delivery, and complications during labour between pregnant women who received iCTG service and those who received regular care. Most participants in the iCTG group were more likely to be business entrepreneurs, have family assets, decide by themselves on birth choices and less likely to have a problem with distance to the health facility (p < 0.001). Most participants in regular care were married (p = 0.018), received 2 to 5 health education sessions (p < 0.001) and started their ANC before 20 weeks’ gestation (p = 0.002). More women in the iCTG group had anaemia during pregnancy (p < 0.001), previously experienced danger signs (p = 0.036), childbirth complications (p = 0.002), and gave birth through C/S (p = 0.040). Most babies from women who received iCTG service had higher birth weight and were less likely to experience complications: 19% vs 10% p = 0.003 and 8.1% vs 12.5% p = 0.049, respectively.

Table 1 Participants characteristics by study group
Full size table

Table 2 describes primary and secondary outcomes between pregnant women in iCTG facilities and non-iCTG facilities. Contrary to our initial hypothesis, a significantly higher proportion of pregnant women in iCTG facilities attended fewer than four ANC visits (16.7%) compared to those in non-iCTG facilities (5.2%), with a p-value of < 0.001. In addition, the Median (IQR) for ANC visits was 5(2) for intervention and 5(1) for control with p < 0.001. Overall, the use of iCTG reported a higher total count of abnormal FHR detections, 44(8.9%), compared to facilities not using iCTG 3(1.1%), with a significant difference (p < 0.001). Additionally, FHR documentation was notably higher in facilities using iCTG (84.6%) than in facilities not using iCTG (5.5%), with a p-value of < 0.001. A significantly higher proportion of women delivered through C/S in iCTG facilities (27.6%) compared to non-iCTG facilities (10.3%), with a p-value of < 0.001. A significantly lower proportion of babies with an Apgar score of < 7 (6.1% vs 10.7%, p = 0.020). Regarding perinatal outcomes, iCTG facilities had fewer admitted newborns (8.3% vs 11.4%) and perinatal deaths (stillbirth + END) (2.6% vs 6.6%) with a p-value of 0.008. However, there were no significant differences in institutional births between the two groups.

Table 2 Proportions of outcome variables across study groups
Full size table

Table 3 shows the unadjusted and adjusted associations of intervention and outcome variables. There were statistically significantly lower odds of attending 5–6 ANC visits among iCTG users compared to pregnant women who received regular care in both unadjusted and adjusted analysis (OR = 0.55, 95%IC:0.37–0.806, p = 0.002) and (AOR = 0.6, 95%CI: 0.35–0.91, p = 0.019), respectively. Additionally, using iCTG was significantly associated with higher odds of having < 4 ANC visits (OR = 2.5, 95%CI:1.28–4.75, p = 0.007) than non-iCTG users. After adjusting all covariates, there was no statistically significant difference in the ANC visits of < 4 between the groups (AOR = 1.2, 95%CI: 0.52–2.72, p = 0.674). Moreover, there was no significant effect on the odds of having ≥ 7 visits for both unadjusted and adjusted analysis (AOR = 1.1, 95%CI: 0.56–2.03, p = 0.845).

Table 3 Unadjusted and adjusted odds ratios comparing outcome for those in the intervention and those in usual care
Full size table

Unadjusted analysis showed that the iCTG use was significantly associated with caesarean section among pregnant women in iCTG facilities compared to those who received regular care in non-iCTG facilities (OR = 3.3, 95% CI: 2.14–5.15, p < 0.001). After adjusting for all relevant covariates, caesarean section was significantly higher in iCTG facilities compared to women in non-iCTG facilities (AOR = 3.8, 95% CI: 2.28–6.57, p < 0.001). Institutional births did not significantly differ between the two groups in unadjusted and adjusted analyses (AOR = 0.4, 95%CI:0.46–1.76, p = 0.257). The iCTG use was significantly associated with a decrease in Apgar score of < 7 among newborns born from women who received iCTG care compared to those who received regular care in both unadjusted and adjusted models (OR = 0.5, 95% CI: 0.32–0.92, p = 0.024) and (AOR = 0.4, 95% CI:0.23–0.87, p = 0.018), respectively. No significant association was found between the groups with regard to newborn admission in both unadjusted and adjusted analysis (AOR = 0.7, 95% CI: 0.29–1.65, p = 0.405). However, iCTG use was significantly associated with lower odds of perinatal deaths of babies born from women who received iCTG care compared to those in non-iCTG users for both unadjusted (OR = 0.4, 95%CI:0.18–0.76, p = 0.007) and adjusted (OR = 0.2, 95%CI: 0.10–0.55, p = 0.003) models (Table 3).

Table 4 presents the results of the Poisson regression analysis on the cases of abnormal FHR detection between the iCTG users vs non-iCTG users. Using iCTG for FHR monitoring was significantly associated with a higher rate of abnormal FHR detection than intermittent FHR monitoring, with a rate ratio of 8.08 (95% CI: 2.51–26.02; p < 0.001). After multivariable adjustment, the rate for abnormal FHR detection by iCTG was 10 times higher than intermittent FHR monitoring (10.54, 95% CI: 3.18–34.92; p < 0.001).

Table 4 Compares abnormal FHR detection among iCTG users vs non-iCTG users
Full size table

Discussion

The current study assessed the effectiveness of antepartum FHR monitoring using iCTG in improving pregnancy and childbirth outcomes at primary health facilities in Tanzania. We found a significantly higher odds of having < 4 ANC visits in the iCTG users group. Additionally, statistically significantly lower odds were found on 5–6 ANC visits among women who received iCTG service. However, we observed no statistical difference in attending seven or more ANC visits. Using iCTG for FHR monitoring was significantly associated with a higher rate of abnormal FHR detection, increased odds of caesarean section, and reduced odds of Apgar scores of < 7 at 5-min. Moreover, perinatal deaths were significantly lower in women who received iCTG care. No significant difference was found in institutional births and newborn admissions between the two groups. Antepartum FHR monitoring with iCTG significantly increased the proportion of FHR documentation in the maternal ANC cards. This study is among the few studies that evaluated the effectiveness of antepartum fetal heart monitor devices in Tanzania, highlighting the importance of implementing digital innovations for FHR monitoring to improve perinatal outcomes.

Improving the quality of ANC and childbirth services has demonstrated an increase in its utilization, especially in low-resource settings [34]. Previous studies have also indicated an increase in ANC visits as a result of using iCTG for antenatal FHR monitoring [17]. On the contrary, our study found a significantly higher proportion of women who attended fewer ANC visits in the health facilities implementing iCTG. Introducing health education during the earliest ANC visits may help encourage women to attend the recommended number of visits throughout pregnancy. In our study, the iCTG intervention was introduced in the third trimester, which limits its potential to influence ANC attendance patterns established earlier in pregnancy. It is also possible that structural or contextual factors unrelated to the intervention influenced ANC attendance in these facilities [35]. Moreover, the attitude of a healthcare provider is vital in motivating women to utilize maternal health services [36, 37]. To enhance the effectiveness of such digital interventions, future implementation of iCTG should be integrated with early and ongoing health education efforts, aimed at reinforcing the importance of receiving the optimum number of ANC contacts.

Our results indicate that iCTG has a significantly higher rate of detecting abnormal FHR, demonstrating its benefits in clinical settings in identifying compromised fetuses. In this study, the iCTG device recorded and displayed the FHR graph via a tablet, which could also indicate the deviations from the normal pattern. The device demonstrates high sensitivity in detecting abnormal FHR and alerts a provider if an abnormal reading persists for over three minutes. The latter assisted the provider in identifying and documenting the data. This result agrees with the previous studies in Tanzania by Kamala et al. (2018) and Mdoe et al. (2018). These studies indicate that continuous FHR monitoring using Doppler (Moyo) in low-risk pregnancies enhances the identification of abnormal FHR, resulting in improved documentation practices and timely obstetric interventions [38, 39]. The comparable effects of interventions on improved FHR monitoring can be related to the functional similarities of the devices, including continuous monitoring ability, portability, user-friendliness, wireless operation, and effectiveness in low-resource settings. Our findings suggest that adopting continuous FHR monitoring technologies into clinical practice could improve FHR monitoring practices and ultimately reduce or potentially replace the current methods.

According to the literature, continuous FHR monitoring has been associated with an increased risk for caesarean section following abnormal FHR detection [40, 41]. For instance, a randomized clinical trial comparing continuous Doppler to intermittent fetoscope FHR monitoring during intrapartum care indicated an increased caesarean section rate due to abnormal FHR detection [39]. This outcome is anticipated given Moyo Doppler’s ability to detect abnormal FHR, prompting healthcare providers’ decision for timely intervention, such as a caesarean section. On the other hand, our results indicate significantly higher odds of caesarean section in health facilities that utilized iCTG. The observed similarity can be related to various factors, such as the timing of interventions, duration of monitoring, characteristics of the study population, and the ability of both devices to detect abnormal FHR, which may necessitate expedited delivery. Despite the benefits of caesarean section in the reduction of maternal and infant mortality, the recent surge increase in the rate of caesarean section births globally has been unacceptably higher than that indicated by WHO [42]. Unnecessary caesarean sections pose significant risks, including increased maternal morbidity, higher costs, strain on limited surgical resources, and complications in future pregnancies [43, 44]. The latter highlights the need for evaluation, clear clinical guidelines, and adequate training to ensure iCTG integration into the ANC system supports evidence-based decisions based on actual fetal hypoxia, rather than routine escalation to caesarean delivery.

Current literature demonstrates that the experimental use of antepartum fetal surveillance, including FHR monitoring, has prevented adverse outcomes in multiple cases [10, 45]. Recently, a larger maternity facility in South Asia introduced a new clinical guideline mandating biweekly CTG and amniotic fluid measurement from 39 weeks of gestation to reduce the risk of stillbirth at term. This change in clinical care was associated with a significant decrease in the rates of both stillbirth and neonatal death at term without increasing rates of perinatal morbidity and early-term birth [46]. In our study, the use of iCTG from 32 weeks gestation significantly reduced the likelihood of perinatal mortality, specifically stillbirth and END. Moreover, our results indicate that the intervention decreased the likelihood of newborns getting an Apgar score of less than seven. However, no significant difference was noted in newborn admission or referral due to complications. A previous study in Tanzania reported a higher proportion of fresh stillbirths in the intermittent monitoring group compared to the continuously monitored group when an abnormal FHR was detected during labour [39]. The improved fetal monitoring practice may have enhanced the interaction between pregnant women and midwives, enabling the identification of potential fetal compromise. However, several studies show no differences in perinatal outcomes following the implementation of continuous FHR monitoring, but the effectiveness in detecting abnormal FHR that require timely interventions [24, 47, 48]. The delay in the birth of identified fetuses with abnormal FHR is the most commonly reported factor contributing to unfavourable perinatal outcomes. Adverse perinatal outcomes are more likely to be prevented when the detection of abnormal FHR is coupled with recommended interventions such as intrauterine resuscitation or emergency delivery [49]. Therefore, we urge the combination of innovations in FHR monitoring with other aspects of access to high-quality care such as the Community Maternal Danger Score (CMDS) [50], designed to identify pregnant women who need skilled provider at time of birth to achieve the best outcomes for both mother and baby.

Strengths and limitations

This study involved not only pregnant women at known increased risk of adverse perinatal outcomes but also low-risk pregnancies, intending to identify the fetuses at high risk of stillbirth in late pregnancies that occur without apparent complicating factors. Several limitations should be considered when interpreting the findings. First, the sample sizes between the intervention and control groups were unequal, potentially affecting statistical power and introducing bias. Second, the primary outcome was achieving more than four ANC visits; however, as the iCTG was introduced during late pregnancy, it might not have provided meaningful insight, and other confounding factors could have affected this outcome. Third, although the iCTG device can monitor FHR and contractions, we only used iCTG for continuous antepartum FHR monitoring in this study. We also aimed to find the pre-term onset of labour but did not find a case. Therefore, we could only record the abnormal FHR detection as bradycardia (FHR < 120 bpm) or tachycardia (FHR > 160 bpm), while early, late or prolonged decelerations could not be assessed without contractions. The absence of baseline (pre-intervention) data limits our ability to fully attribute observed effects to the intervention, though this was mitigated by selecting comparable sites and adjusting for relevant covariates. More research is needed on cost-effectiveness, earlier integration, and strategies to reduce unnecessary caesarean sections.

Conclusion

Implementing iCTG for antepartum FHR monitoring improved clinical detection of abnormal FHR and was associated with increased odds of caesarean section and reduced the likelihood of Apgar scores < 7 at 5-min and perinatal deaths. However, iCTG did not increase ANC attendance, possibly due to its introduction during late pregnancy. Moreover, no effect was found on the institutional births and newborn admissions. We implemented a digital medical device that is smart, wireless, user-friendly, and designed to notify providers of fetal distress, aiding clinical decision-making and ultimately preventing adverse outcomes. Although our findings were promising, studies are needed to assess the effectiveness of iCTG across other populations and contexts to ascertain its generalizability and to identify crucial success factors to optimize implementation.