This review included studies that evaluated interventions aimed at improving timely initiation of ANC – defined as attendance within the first trimester – among pregnant adolescents and women in sub-Saharan Africa. Eligible study designs comprised randomised controlled trials, non-randomised studies, controlled before-and-after (CBA) studies, and interrupted time-series studies, in line with established methodological guidance for systematic reviews of interventions.^1,12 Only studies published in English between 1990 and 2022 were considered for inclusion.

Eligibility parameters were defined using the Population, Intervention, Comparison, Outcome (PICOS) framework and presented in the logic model. The review included studies from sub-Saharan Africa focusing on interventions to improve timely ANC initiation (first trimester) among pregnant women, including adolescents and young women (15–49 years). Studies addressing multiple health behaviours were eligible, with no restrictions on baseline maternal health conditions. General population, children, and elderly studies were excluded.

Types of interventions: Eligible interventions were those designed to promote or improve initiation of ANC within the first trimester among women of reproductive age in sub-Saharan Africa. Studies were included if they evaluated any intervention that resulted in early ANC attendance or were delivered during the antenatal period with the intention of influencing early care-seeking behaviour, including interventions addressing multiple health behaviours during pregnancy.

Types of comparators: Studies that compared multiple health behaviour interventions with non-intervention control, standard of care, or another active intervention addressing one health behaviour were included.

Types of outcomes: The primary outcome of interest was early initiation of antenatal care, defined as ANC attendance within the first trimester, as a modifiable health behaviour among pregnant adolescents and women. Outcome data collected using any method – including self-report, direct observation, or objective measures – were considered eligible, including data obtained from service audits and medical or pregnancy records.

Titles and abstracts retrieved through the search were independently screened for eligibility by two reviewers, with disagreements resolved through consultation with a third reviewer (Ronald Fisa, Mutale Sampa and Mercy W. Monde) using the eligibility criteria as prescribed in the study protocol. After screening of titles and abstracts, full texts of potentially eligible studies were retrieved. Two authors (Ronald Fisa and Mutale Sampa) independently screened full texts for eligibility. Study selection and screening were managed in Covidence, a software for managing and streamlining systematic reviews (Covidence systematic review software, 2019). Studies that did not met the criteria were excluded. Disagreements were resolved through discussion with a third reviewer (Choolwe Jacobs). A final decision on the included studies was made, leading to the extraction phase. A PRISMA flow diagram was used to illustrate the screening procedure.

Community-based interventions were delivered through diverse mechanisms, including community outreach,^15,23,24 home visits,^{14,16,18,22,24} community dialogues^14,23,24 and through referral of mothers to the health facilities.^15,18

The distribution points ranged from homes^{14,15,16,19,22,23,24} to community centres,^15,16 village centres,²⁴ waiting homes¹⁶ to health facilities.¹⁴ The most popular, and perhaps relatively systematic, approach to distributing care was home visits. However, home visits were complemented with escorts to health facilities.

Mechanisms for community-based interventions: Community health volunteers and outreach nurses were trained to identify pregnant women through home visits, offering education and counselling on preventive care and maternal health services, including prevention of mother-to-child transmission (PMTCT).^14,15,22,23 Both existing and newly trained CHWs supported women to attend health facilities. Household-based approaches, where CHWs were assigned specific households or participants, proved more effective in Malawi¹⁶ and Uganda.²²

Notably, CHWs had a responsibility of connecting identified women to the health facilities by either escorting them to the facilities¹⁴ or encouraging them to do so.¹⁵ While homes may be an important location for the provision of services, the linkage to health facilities is crucial in dealing with referral and management of those who need more professional care.²⁵

Some studies reported integration of other health services, such as PMTCT programmes^14,16,18 and a broader child survival programme.²⁴ For some studies, training, education, and awareness is presented as part of many other interventions, while for other studies these components are used to optimise other interventions. Several components and pieces are added to the home visits mechanism, depending on the type and purpose of the intervention. The CHWs visited women enrolled in PMTCT care and provided information and counselling.

In Tanzania¹⁴ CHWs visited women enrolled in PMTCT care and provided information and counselling. In their study, Geldsetzer et al.¹⁴ used trained CHWs to revisit pregnant women at home to verify whether they had attended ANC and to explore who missed an ANC or PMTCT and understand why and to encourage them to attend.

The study by Lema et al.¹⁸ is also another example of continuous support for women, in which routine revisits in all identified pregnant women were conducted within 2 weeks after the initial visit to verify whether women had visited ANC. Lema et al.¹⁸ argues that through revisits, CHWs became increasingly effective in identifying pregnant women in the community. This observation could be explained by the fact that CHWs might have learned how to approach women and families in the community best and elicit information on pregnancy status.

Provision of tools for implementing interventions: Several reviewed studies highlighted the importance of providing additional support to implementers to improve ANC outcomes. This support included equipment, supplies, transport, and infrastructure.^14,15,19,22 In Tanzania, registers were given to CHWs for tracking household visits to pregnant women.¹⁴ In Zambia, community transport suited to local terrain was provided,¹⁵ while in another study, bicycle ambulances eased access to health facilities during obstetric emergencies.²⁴ In Zanzibar, mobile phones were used to send short messaging service (SMS) reminders for ANC, which increased first trimester attendance.¹⁹

In Uganda, CHWs received essential supplies, supervision, and refresher training to support early diagnosis and management of hypertensive disorders in pregnancy.²² This intervention was found to be cost-effective across the settings.²⁶ In Malawi, maternity waiting homes were constructed and an additional nurse was hired to meet rising demand at intervention facilities.¹⁶ This led to a 200% increase in the proportion of women initiating ANC early. These interventions suggest that targeted logistical and material support to implementers – both at community and facility levels – can significantly improve ANC uptake and service quality, especially when adapted to local needs and supported by health system strengthening measures.

Additional support for care by implementers of interventions is another mechanism being used in some reviewed studies. The support to implementers came in the form of equipment, supplies and materials.^14,15,19,22 In Tanzania, Geldsetzer et al.,¹⁴ provided registers to CHWs for monitoring CHWs’ activities in the intervention wards by logging in each household visit to a pregnant woman. In Zambia, Ensor and colleagues’s¹⁵ intervention was to provide community transport to groups of two or three villages. While in a study by Hembling et al.,²⁴ bicycle ambulances were also provided to ease transport from the community to the facility during obstetric emergencies.

Facility-based interventions form the cornerstone of current health systems. With the growing demand for quality facility-based services, evidence of facility-level interventions to improve initiation of ANC within the first trimester was explored in articles under review. We found two quality improvement studies, one in Ghana²¹ and another in Kenya²⁰ that implemented interventions aimed at improving the timeliness, completeness and accuracy of the data being submitted to and reported by the District Health Information Management System (DHIMS) offices, in efforts to improve maternal and child health outcomes at a scale.

Another facility-based intervention conducted in Tanzania¹⁹ was aimed to evaluate the association between a mobile phone intervention named ‘wired mothers’ and ANC in Zanzibar and hypothesised that the ‘Wired Mothers’ intervention can increase ANC attendance as well as improve content and timing of antenatal care services provided to individual women.

Mechanisms for facility-based interventions: Two of the articles described interventions that employed an approach of quality improvement^20,21 trained District Health Management Team (DHMT) officers as implementers. Both studies were non-randomised studies that routinely collected maternal newborn, child health (MNCH) data in the facilities and involved training of key personnel in quality management of data.

Mwaniki et al.²⁰ implemented a quality improvement intervention to increase utilisation of integrated health services (ANC, PMTCT, and skilled delivery) in an entire rural district that was performing below the national average in most of the health indicators over a 20-month period. Trained DHMT members were assigned to two or three 7–12-member quality improvement teams composed of facility healthcare personnel, community healthcare volunteers, and community representatives to mentor/coach them on quality improvement as part of their regular supportive supervision. Primary indicators in the intervention were ANC.

Similarly, Singh et al.,²¹ in their quality improvement intervention in Ghana, aimed at improving the health outcomes of mothers, infants, and children under-5 years by improving the coverage, quality, reliability and patient-centredness across all public and faith-based facilities in Ghana. In this vein, a quality improvement team in each facility was formed, and each member attended four learning sessions and shared progress with other quality improvement (QI) teams. In addition, supervision and coaching visits were made by project staff in conjunction with district health supervisors.

This approach entailed the identification of process failures by health staff and testing these process changes in the health facilities to address those failures²⁷ seemed to have been effective as demonstrated in Sigh et al.,’s study²¹ where there were some improvements in maternal health outcomes over time, with an increase in early ANC from 37% to 42%.²¹

In Tanzania, Nielsen et al.,¹⁹ implemented mobile phone innovative training initiatives known as the Wired Mothers’ Intervention, consisting of an automated short messaging service (SMS) system providing wired mothers with unidirectional text messaging between wired mothers and their primary healthcare providers. The aim of the SMS component was to provide simple health education and appointment reminders to encourage attendance at routine ANC, skilled delivery attendance, and postnatal care.

This intervention demonstrated the efficacy of mobile phones in improving ANC and also made remembrance of ANC appointments easier, which can be helpful even in different contexts.¹⁹ Furthermore, Nielsen et al.¹⁹ argued that mobile phones are an economic innovation that could influence the success or outcome in similar contexts.

A combination of care providers and strategies within the community and health facilities has been used in the implementation of interventions. According to Geldsetzer et al.,¹⁴ the intervention engaged both CHW and healthcare providers, known as community outreach nurses, who were tasked with additional maternal healthcare tasks. Community outreach nurses to whom women were referred by CHW for ANC facilitated and monitored the referral system and verified documentation for referred mothers at healthcare facilities, hence adding to supervision and support to CHWs.¹⁴

In a study by Hembling et al.²⁴ village-level councils comprised of faith-based leaders (Protestant ministers, traditional African religious leaders, and Islamic imams), village chiefs, traditional medical practitioners, and female leaders known as ‘queen mothers’ or ‘magazia’. Using this approach of engaging community leaders who are highly influential and respected leaders as custodians of traditional practices, attitudes, and beliefs related to MNCH was assumed as the best approach in addressing deeply held norms and traditional practices and promoting behavioural changes throughout the community.²⁴

Tudor²⁸ argues that integration of services can reduce programme costs and time through more synergistic use of resources. Integrating ANC services with other services can be realised with involvement of all stakeholders and consideration of the health system and socio-political contexts.²⁹ However, if well thought out integration can have undesirable consequences, particularly in low-resourced contexts with already constrained and overloaded health systems.³⁰ Therefore, integrating and linking services need to be carefully considered as they introduce complexity (and uncertainty) to studies and related outcomes.

Table 2 provides a summary of the interventions and outcomes from the reviewed articles. All the interventions demonstrated a significant increase in attendance of ANC in the first trimester, except for two; by Wafula et al.,²³ in Ghana (Difference in Difference = –1.3%) and by Geldsetzer et al.,¹⁴ (59.1% versus 60.7% between intervention and standard-of-care care arm; risk ratio [RR] 0.97; 95% CI: 0.82–1.15; p = 0.754) in Tanzania. The rest of the articles demonstrated a significant increase in attendance of ANC in the first trimester.

All the facility-based interventions improved the early initiation of ANC. For instance, a significant increase in early ANC attendance (odds ratio [OR]: 2.39; 95% CI: 1.03–5.55) was observed in a facility-based intervention called the wired mothers, using mobile phones to send reminders,²⁰ while in a quality improvement intervention by Mwaniki et al.,²⁰ in Kenya, a triple effect (8% to 24% p = 0.002) was observed between the baseline and end line, and another quality improvement intervention by Singh et al.,²⁰ revealed an increase in early ANC from 37% at pre-intervention to 42% at post intervention.

Notably, community-based interventions that were behavioural related increased attendance of ANC in the first trimester, except for a study by Wafula et al.,²³ and Geldsetzer et al.,¹⁴ For instance, a Malawi study,¹⁷ using a household approach integrating other services such as TB and HIV in addition to maternal health services had a significant triple effect (from 10% to 29% – 200%: 95% CI: 163% – 234%) on improving ANC attendance within the first trimester. This intervention managed to triple the proportion of women starting ANC in the first trimester. Most of these interventions also increased the proportion of women completing at least four ANC visits as well as attending skilled birth attendance.

Using the ROB2, all three RCT studies^14,18,19 were assessed and their risk of bias was found to be of high risk. Some of the observations made from these studies were that these RCTs did not blind the participants as well as health workers because of the nature of the interventions. In these studies, the allocation sequence was not concealed from either the health workers or the participants until clusters were enrolled and assigned to interventions. Furthermore, one study¹⁹ had problems in the randomisation of participants to the intervention and control. In this article, it was not clear whether the allocation sequence was concealed until clusters were enrolled and assigned to interventions.

The articles, therefore, scored poorly in these domains of the ROB assessment tool, affecting the overall ROB result. All three articles^14,18,19 had a high risk of bias in the measurement of the outcome.

A summary of the risk assessment for all the domains is given in Table 2.

In assessing the quality of studies, we used the Risk of Bias in Non-randomised Studies (ROBINS-I) of Interventions. For the overall risk of bias for these studies, three of the articles^20,21,24 were found to have low risk bias and five^{15,16,17,22,23} were found to be with moderate risk of bias. Table 3 gives a summary of the risk of bias assessments for all the 8 studies in all the domains.

A summary of the risk of bias is given in Figure 2. Overall, almost all the articles had low risk of bias. All the studies in the review did not deviate from the intended interventions of the study. Only one study¹⁷ representing 12.5% of the studies had bias in the selection of participants, the reported result, and a serious risk of bias because of confounding,²⁰ moderate risk of bias in the selection of participants into the study¹⁷ and a serious risk of bias because of missing data,¹⁷ only a small fraction of the articles had a critical risk of bias in missing data.¹⁶

The majority (87.5%) of the studies^{15,16,20,21,22,23,24} had a low risk of bias in the selection of participants. Low risk of bias in measurement was observed in only 25% of the studies.^20,21

Notably, 75% of the studies had a low risk of bias in the classification of the intervention^{15,16,20,21,24} while 87.5% had a low risk of bias in the selection of participants^{15,16,20,21,24} About 12.5%²⁰ of the studies with a serious risk of bias.

Bias because of missing data was observed in 25% of the studies. Overall, majority of articles had low risk of bias for most of the articles.