ABSTRACT
Introduction:
Health risks associated with short interpregnancy intervals, coupled with women’s desires to avoid pregnancy following childbirth, underscore the need for effective postpartum family planning programs. The antenatal period provides an opportunity to intervene; however, evidence is limited on the effectiveness of interventions aimed at reaching women in the antenatal period to increase voluntary postpartum family planning in low- and middle-income countries (LMICs). This systematic review aimed to identify and describe interventions in LMICs that attempted to increase postpartum contraceptive use via contacts with pregnant women in the antenatal period.
Methods:
Studies published from January 2012 to July 2022 were considered if they were conducted in LMICs, evaluated an intervention delivered during the antenatal period, were designed to affect postpartum contraceptive use, were experimental or quasi-experimental, and were published in French or English. The main outcome of interest was postpartum contraceptive use within 1 year after birth, defined as the use of any method of contraception at the time of data collection. We searched EMBASE, Global Health, and Medline and manually searched the reference lists from studies included in the full-text screening.
Results:
We double-screened 771 records and included 34 reports on 31 unique interventions in the review. Twenty-three studies were published from 2018 on, with 21 studies conducted in sub-Saharan Africa. Approximately half of the study designs (n=16) were randomized controlled trials, and half (n=15) were quasi-experimental. Interventions were heterogeneous. Among the 24 studies that reported on the main outcome of interest, 18 reported a positive intervention effect, with intervention recipients having greater contraceptive use in the first year postpartum.
Conclusion:
While the studies in this systematic review were heterogeneous, the findings suggest that interventions that included a multifaceted package of initiatives appeared to be most likely to have a positive effect.
INTRODUCTION
There is substantial evidence that short interpregnancy intervals (the time from delivery to subsequent conception is less than 18 months) increase risks of adverse maternal and perinatal outcomes such as low birth weight, small for gestational age, maternal mortality, and severe morbidity,1–3 and that postpartum women have strong desires to avoid pregnancy following childbirth. Yet, interest in postpartum family planning (PPFP) programs has waxed and waned since the 1960s.4 For example, a literature review on community-based interventions from the 1970s to 2004 aimed at improving postpartum care found a dearth of programs promoting postpartum birth spacing despite community-based FP distribution programs being widespread.5 However, the last decade has seen a renewed interest in voluntary PPFP, following the World Health Organization’s (WHO) issuance of guidelines in 2013 for PPFP strategies,6 and with strong support from the U.S. Agency for International Development and other donors through tools, such as the High Impact Practices brief for immediate PPFP,7 and processes to identify priority actions to support global PPFP scale-up.8,9
The 2013 WHO guidelines identified a continuum of 4 points of contact for PPFP, beginning with contacts during pregnancy (via facility-based ANC or community-based pregnancy-screening programs), and moving on to facility-based contacts in the intra- and immediate postpartum time periods, contacts via postpartum programs, and contacts via infant-care programs. A 2019 systematic review and meta-analysis of observational studies found that counseling during antenatal or postnatal care was a predictor of postpartum contraception use.10 Because antenatal care (ANC) coverage (percentage attended at least 4 visits) is 66% globally, 45% in low-income countries, 61% in lower-middle-income countries, and 91% in upper-middle- income countries11 and with the WHO now recommending 8 visits,12 there is potential ample opportunity for FP counseling during the antenatal period.
There have been several literature reviews on topics related to PPFP. A 2012 Cochrane review assessed the effectiveness of ante- and postnatal counseling on the uptake of copper intrauterine devices (IUDs).13 A 2010 review examined integration of FP with other health services, followed by a 2014 systematic review that also assessed the use of integration and outreach programs to promote PPFP.14,15 A 2015 systematic review reviewed studies of interventions explicitly intended to have an effect on the contraceptive practices of postpartum women in LMICs.4 A 2016 review assessed intervention strategies to decrease the unmet need for contraception among postpartum women in LMICs.16 Finally, a 2022 systematic review synthesized evidence on the coverage and effect of both routine FP counseling and new FP counseling interventions on postpartum modern contraceptive uptake in sub-Saharan Africa.17 With the exception of the most recent review, these reviews mostly included studies conducted before the 2013 WHO recommendations, and the early reviews were mostly based on studies from high-income countries.6,12,17 The most recent review was restricted to sub-Saharan Africa and only considered counseling interventions.17
This article summarizes the methods and results from a systematic review of interventions in low- and middle-income countries (LMICs) that attempted to increase voluntary postpartum contraceptive use, including the lactational amenorrhea method (LAM), through contacts with pregnant women in the antenatal period. This review aimed to describe the interventions identified and assess their effectiveness on postpartum contraceptive use within the first year postpartum and other related outcomes.
This review aimed to describe interventions that attempted to increase voluntary postpartum contraceptive use and assess their effectiveness on postpartum contraceptive use within the first year postpartum and other related outcomes.
METHODS
Research Questions
This review focused on any intervention (or a component of an intervention) delivered during ANC with the explicit aim of increasing voluntary postpartum contraceptive use after the index birth. Specifically, this review sought to answer the following research questions: (1) Can interventions delivered in the antenatal period increase voluntary postpartum contraceptive use? (2) What intervention components/content are present across all the effective interventions?
Context and Population
We focused on interventions in LMICs that were delivered in the antenatal period (delivered in the community or in primary or secondary health facilities) to increase postpartum contraceptive use (including use of LAM).
Eligibility Criteria
Studies published from January 1, 2012, to July 31, 2022, were included in this review if they were conducted in LMICs; evaluated any intervention that was delivered at some point (but not necessarily exclusively) during the antenatal period that was explicitly intended to have a distinct effect on postpartum contraceptive use (including use of LAM); were experimental (i.e., were randomized or nonrandomized trials) or quasi-experimental (e.g., controlled before-after or interrupted time-series) designs; and were full peer-reviewed articles published in French or English. Descriptive studies, qualitative research, literature reviews, opinion papers, conference proceedings, and unpublished studies were not eligible.
Search and Screening
We used a comprehensive set of search terms around 4 themes: (1) postnatal; (2) contraceptive methods; (3) antenatal interventions, and (4) low- and middle-income countries, combined using the Boolean “AND” (the Supplement contains the full search strategy for each database). Search terms were based on those from a previous review, which included literature published through the end of 2013.4 We limited our search to studies published from 2012 until the end of July 2022. We also searched 2 relevant reviews published after 2012 for additional studies.4,17
We searched EMBASE, Global Health, and Medline and manually searched the reference lists from eligible studies included in the full-text screening for additional relevant studies. Before running the formal search, we consulted the London School of Hygiene & Tropical Medicine Library services librarian for advice. We validated our preliminary search by checking that the search results included articles of relevance suggested by experts and found in previous reviews. The search result files for each database were uploaded into a systematic review software program (Rayyan, www.rayyan.ai). After piloting the study selection process, we formally screened search results against eligibility criteria. All abstracts were double-screened, with screeners (NF, OM, OC, AC) masked to the other screeners’ decisions. Decisions were unmasked when all screening was complete. Discordant decisions were then discussed among the team and consensus was reached on whether to include or exclude these studies.
Outcomes
The main outcome of interest was postpartum contraceptive use within 1 year of childbirth. If studies reported postpartum contraceptive use at multiple time points, these were all recorded. If a study did not report the main outcome, we required that it included a clear description of the intervention and reported at least 1 of the following outcomes: postpartum contraceptive use extending beyond a year; use of specific contraceptive methods; contraceptive continuation; postpartum unmet need; pregnancy; length of birth interval; knowledge/awareness of available methods; or intention to use a method in the postpartum period. Any other data on relevant outcomes were also extracted. All measures of intervention effect were considered (e.g., prevalence ratios or prevalence differences). Measures of intervention effect were considered significant at the 5% level.
Data Extraction
An Excel-based extraction tool was developed by 1 team member (OM) and refined in several iterations until the final tool was agreed upon. Studies were assigned to 4 people (OM, NF, OC, and AC) for the initial data extraction. After the initial extraction was complete, the tool was separated into 4 sections: (1) general study information, (2) further details on the intervention, (3) assessment of outcomes, and (4) quality appraisal. In further discussions among the core team (OM, NF, OC), the extracted data were “harmonized” (i.e., we agreed upon a set of categories for each component of the data). For example, because the initial data extraction was based on the authors’ reports, the same study designs were often described in different ways; in these cases, we standardized the descriptions. Data were then extracted again by another team member (MP), and discrepancies were reviewed and discussed within the team.
We compiled a concise quality appraisal tool to extract data on the elements of quality most relevant to our included studies. This tool drew upon the Cochrane Effective Practice and Organization of Care risk of bias criteria and the UK National Institute for Health and Care Excellence Quality appraisal checklist for quantitative intervention studies.18,19 One person (OM) extracted the quality appraisal data from all studies. We used the PRISMA 2020 statement as a guide for reporting our procedures and results.20
Data Synthesis
This stage aimed to understand how and why the interventions had or did not influence the different outcomes and, in doing so, to identify the factors that could explain the differences between the intervention effects. We anticipated the interventions in this review would be heterogeneous, that is, diverse regarding the features of the intervention, such as the type of intervention, its components, who delivers it, how it is delivered, when, and how often. To address this heterogeneity, we planned a narrative synthesis of the measures of intervention effect, guided by the United Kingdom Economic and Social Research Council Methods Program framework.21 In the preliminary synthesis, we developed an initial description of the data and organized it in terms of geographic distribution, chronological frequency of publication, and intervention type. We stratified studies by intervention type and then alphabetically under these headings. Studies were then categorized by their study design, setting, participants, intervention components (e.g., counseling, home visits, integrated, and multimodal), intervention delivery details, and contraceptive use outcomes. We looked for general patterns in effect direction and size. While we extracted data on a range of contraceptive-related outcomes, the focus of the synthesis was on postpartum contraceptive use within 1 year. Finally, we interrogated the initial patterns we observed by exploring the relationships within and between the studies. The summaries of the interventions were guided by the template for intervention description and replication checklist so that the recommended items for describing an intervention are included.22
RESULTS
Search, Screening, and Extraction
We began the search in July 2022 and completed double-screening in a systematic review software program on August 26, 2022. We double-screened 771 records. After unmasking the screening decisions, there were 16 discordant decisions and 30 records that required further discussion among the team. Thirty-eight records were identified for full-text review. After full-text review and double data extraction (completed March 22, 2023), 34 reports on 31 unique interventions were deemed eligible and included in the review (Figure 1).20,23–56
Preliminary Synthesis
Several reports were based on the same intervention, sometimes in the same population. Although Guo et al., Huber-Krum et al., Pradhan et al., and Puri et al. had different outcomes and analytical methods on the same study in Nepal, we considered these 4 reports together and reported on the multiple outcomes under 1 study.23–26 Karra et al. and Pearson et al. were also based on the same intervention package as those 4 studies but were in Sri Lanka and Tanzania, respectively.23–28 In this case, we kept the data separate as the intervention mechanism could have been modified by the context. Similarly, Tran et al.29 and Tran et al.30 were based on the same intervention (Yam Daabo) but were conducted in Burkina Faso and the Democratic Republic of the Congo, respectively; these reports were also kept separate in the synthesis. Grouping the 4 Nepal reports together reduced the total number of studies included in the synthesis to 31.23–56
Study Characteristics
The main characteristics of the included studies are presented in Table 1. Figure 2 and Figure 3 show where studies were conducted. The majority of studies (68%, n=21) were conducted in sub-Saharan Africa: Kenya (n=5),31–35 Burkina Faso (n=2),29,36 Democratic Republic of the Congo (DRC, n=2),30,37 Nigeria (n=2),38,39 Tanzania (n=2),28,40 Ethiopia (n=1),41 Ghana (n=1),42 Guinea (n=1),43 Malawi (n=1),44 Rwanda (n=1),45 Uganda (n=1),46 Zambia (n=1),47 and Zimbabwe (n=1).48 Asia had 9 reports (29%): Nepal (n=4),23,–26,49–51 Afghanistan (n=1),52 Bangladesh (n=1),53 India (n=1),54 Sri Lanka (n=1),27 and Thailand (n=1).55 One study was conducted in Egypt.56
Reports were published in all years from 2012 to 2022 inclusive, with 68% (23 of 34) published in the previous 5 years (2018–2022) (Figure 4, n=34, the grouped Nepal reports are separated in this figure).
Around half of the study designs (52%, n=16) were randomized controlled trials, and half (48%, n=15) were quasi-experimental. Most studies (77%, n=24) were conducted in health facility settings: hospitals (n=9), lower-level facilities (n=11), mixed-level facilities (n=2), and mixed health facilities and community (n=2). The remaining 7 studies (23%) were conducted in community settings only.
Just over half the studies specified that participants were pregnant women (55%, n=17). The rest of the studies enrolled postpartum women only (23%, n=7), pregnant and postpartum women (10%, n=3), pregnant women and their male partners (6%, n=2), women aged 15-49 (pregnancy status not specified) (3%, n=1), and postpartum women and female community health volunteers (3%, n=1). When women were enrolled in the postpartum period, they were still assessed for the effect of an intervention delivered in the antenatal period.
Interventions
Details of the intervention components are presented in Table 1. Twelve studies described interventions in the antenatal period only and 18 in the antenatal and other periods, including in the intra- and postpartum periods. One was unclear. Despite the interventions being heterogeneous, we were able to categorize them into 6 main types: counseling (23%, n=7), digital (6%, n=2), educational (16%, n=5), financial (6%, n=2), package (45%, n=14), and training of providers (without an explicit follow-up intervention with pregnant women) (3%, n=1). Interventions were classified as package if they contained at least 2 distinct components. Interventions were classified as counseling if the provision of counseling was the dominant component (e.g., 1 intervention provided one-to-one counseling but also offered a pamphlet containing similar information and was classified as counseling rather than package) and if they were delivered one-to-one or to a couple, whereas interventions delivered to a group only were classified as educational. The distinction between counseling and education was not always clear (e.g., Bang et al. described an educational intervention that consisted of using village-level sessions and small-group sessions; we could not be sure what the village-level sessions comprised [i.e., they could have consisted of one-to-one or couple sessions]).41 In unclear cases, we classified the intervention according to the description used by the authors. We further categorized the types into 13 subtypes: counseling: one-to-one (13%, n=4); counseling: one-to-one plus pamphlet (3%, n=1); counseling: one-to-one with spouse involvement (3%, n=1); counseling: mixed couple and group sessions (3%, n=1); digital: SMS (6%, n=2); educational: campaign (3%, n=1); educational: group sessions (13%, n=4); financial: client vouchers (3%, n=1); financial: provider pay-for-performance (3%, n=1); package: multifaceted (35%, n=11); package: digital and one-to-one education and counseling (3%, n=1); package: systems strengthening (6%, n=2); and training: providers (3%, n=1).
Although the interventions were heterogeneous, they could be categorized into counseling, digital, educational, financial, package, and training of providers.
Most counseling interventions consisted of multiple sessions during pregnancy. One36 consisted of mixed sessions with 2 couple counseling sessions and 1 male group session. One intervention was a small subcomponent of a post-HIV test counseling that included sexual negotiation skills and empowerment, information about HIV, prevention of mother-to-child transmission, FP, and communication skills related to sex and condom use.40 Half the counseling interventions (n=4) were delivered by providers, mostly in ANC clinics.
The 2 digital interventions were delivered by SMS. Both contained a 2-way element, and the trial31 also included a 1-way SMS arm and classified participants into tracks (routine, adolescents 14–19 years, first-time mothers, women with a previous cesarean delivery, and those with multiple gestations), with tailored messages for each.
The educational campaign was led by community health workers.54 Of the 4 educational group sessions, 1 was delivered by study trainers48 and 1 by community health volunteers35; Lori et al.42 and Bang et al.41 did not mention who delivered the educational sessions.
One financial intervention targeted clients,34 included pregnant women attending ANC at a private-sector health care facility, and provided vouchers for free contraceptive methods. The vouchers had different validity periods; 1 was valid for 1 year and 1 valid for 8 weeks after the estimated date of delivery. Participants were also randomly assigned to receive an SMS reminder to use the vouchers, which was sent at 5 weeks postpartum. The other financial intervention was provider-targeted and used “pay for performance,” where facilities were given bonus payments based on the number of maternal and child health services that they provided (ANC, skilled birth attendance, postnatal care, pentavalent vaccination, and TB case detection), with additional annual payments based on a balanced scorecard that addressed quality of services, and the contraceptive prevalence rates in the health facility catchment areas.52
One of the 2 package interventions that focused on system-strengthening was based on accountable care principles,49 where “a group of providers were held accountable for achieving pre-specified outcomes for a specific population over a period of time for an agreed cost.” This involved strengthening community health worker active surveillance, integrating digital health information, and increasing monitoring and supervision capabilities. The other system-strengthening package intervention had multiple interacting components, including improvements in maternity waiting home infrastructure (in the form of equipment and supplies), policies, management and financial structures, and linkages to health systems with skilled midwives.47 Three package interventions consisted of one-to-one counseling plus other components. One study53 added community-based meetings led by community mobilizers, flyers, and distributed contraceptives. The second45 added provider training on insertion and removal of postpartum intrauterine device (PPIUD) and a refresher on postpartum implants, financial incentives, and stakeholder engagement. The third study55 included a digital component using computerized media to promote self-efficacy in oral contraceptive use. The remaining multifaceted package interventions contained a wide variety of components, which included counseling, training of providers, and the provision of equipment.
The training intervention consisted of training providers on PPFP, including the advantages and disadvantages of each method, focusing on immediate PPIUD.50
Characterizing the Included Studies
Outcomes
Twenty-four of the 31 studies reported on the main outcome of interest: postpartum contraceptive use within 1 year after birth. Eight studies reported prevalence at 12 months exactly. Others reported use at 48 hours (about 2 days), 1 week, and various other time points and periods within 12 months, with 1 study reporting the primary outcome up to 24 months. The other 7 included studies reported on 1 of the additional outcomes.
We assessed the studies regarding the main outcome of interest (postpartum contraceptive use within 1 year after birth) by considering the intervention components and effect sizes within the studies that demonstrated a positive effect against those that did not find a positive intervention effect. The outcome measures were heterogeneous (Table 2), which supports our decision to conduct a narrative synthesis rather than a meta-analysis.
Contraceptive Use Outcomes: General Patterns in Effect Direction and Size
Of the 24 studies (77%) that reported on voluntary postpartum contraceptive use within 1 year after birth, 18 reported a positive intervention effect (75%). Of these 18, 9 (50%) were package interventions,23–26,29,30,37,47,49,51,53,56 4 (22%) were counseling interventions,36,38,39,43 3 (17%) were educational interventions,35,42,54 1 (6%) was digital,33 and 1 (6%) was financial.34 Nine of the 14 package interventions (64%), 3 of the 5 educational interventions (60%), 4 of the 7 counseling interventions (57%), 1 of the 2 digital interventions (50%), and 1 of the 2 financial interventions (50%) reported a positive effect on the primary outcome. The 1 training intervention did not report on the primary outcome.
Of the 24 studies that reported on voluntary postpartum contraceptive use within 1 year after birth, 18 studies reported a positive intervention effect.
While the measures of effect captured and the effect sizes were wide ranging (e.g., relative risk 1.10 to odds ratio [OR] 6.69),36,42 the largest, most consistent effects were seen in the package interventions. Among the package interventions where the main outcome was reported (n=10), 9 (90%) reported a positive intervention effect (this compares to 4 of the 6 counseling interventions where the main outcome was reported (67%); 3 of the 5 educational interventions (60%); 1 of the 2 digital interventions (50%) and 100% of the digital interventions (n=1)). Two package intervention studies reported immediate postpartum contraceptive uptake,25,45 all of which reported a positive effect of the intervention. Among the 6 studies where the main outcome was reported but where there was no significant effect observed,30,31,40,41,46,48 1 was a package intervention, 2 were counseling interventions, 2 were educational group sessions, and 1 was a digital intervention.
Four studies23–26,44,53,56 reported on postpartum contraceptive use beyond 1 year after birth; 3 of these reported a statistically significant intervention effect.
Twenty-two studies reported on the use of specific methods of contraception (Table 2, column title Use of specific methods of contraception), with 1324,25,28–30,33,36,37,45,46,51 reporting the intervention had a positive effect on use of at least 1 of the following methods: PPIUD, implants, permanent methods, or use of highly effective methods. Five studies reported on LAM and 3 on exclusive breastfeeding. Three of the studies increased use of LAM in the intervention group, 1 decreased it, and 1 showed no use of LAM whatsoever. The studies on exclusive breastfeeding all showed increased use. A further 2 studies trained on the correct use of LAM but did not report results because of measurement problems. Four studies reported on contraceptive continuation (continuation for the first 12 months after adoption, discontinuation at 6 months, continuation among women starting contraception at 10 weeks postpartum, and stopping contraception at 6.5 months),31,33,40,53 none of which reported a statistically significant intervention effect.
Randomized Controlled Trials
Of the 24 studies that reported on the main outcome of interest, 11 were randomized controlled trials (RCTs) and 13 were quasi-experimental designs. Of the 11 RCTs, 5 used cluster or stepped-wedge designs23–26,29,30,35,46 and 6 were individually randomized trials.31,33,34,36,38,39 Utilizing our quality appraisal tool (Table 3), we found that all but 131 of these 11 RCTs described the source population well and included an eligible population that appeared representative of the source population (Unger31 was unclear on both of these criteria). All 11 RCTs described the intervention and comparator well. However, only 430,31,35,39 had a low risk of bias regarding the random sequence generation. Two38,46 had a high risk of bias, and the sequence generation was unclear in the remaining 5 trials. This means that in 7 of the 11 RCTs, there was likely a lack of comparability between the trial groups. The trials with the low risk of bias in sequence generation had moderate differences in effect between the groups: increased contraceptive adoption at 12 months (risk difference: 7.2%, 95% confidence interval [CI]=2.6, 12.9, P=.034)35; use of modern contraception at 6 months postpartum: 57.4% vs. 35.4%, P=.00239; prevalence of modern contraceptive methods at 12 months (adjusted prevalence ratio: 1.58, 95% CI=0.74, 3.38)30; and increased contraceptive use in both intervention arms (1-way SMS: 72% and 2-way SMS: 73%; P=.03 and .02, respectively, vs. 57% control; however, this difference was not significant when correcting for multiple comparisons).31
Intervention Descriptions
The 18 studies reporting a positive intervention effect on the main outcome were also heterogeneous regarding intervention type. Nine of these studies23–26,29,37,47,49,51,53,55,56 a “package”-type intervention and 6 of these 9 (66%) were characterized as “multifaceted” with several interacting intervention components. All but 1 (Tran et al.29) of the multifaceted interventions that reported PPFP use within a year reported a positive intervention effect. Although they did not report on the primary outcome of interest, the 7-component package described by Karra et al.27 (FIGO’s PPIUD intervention, described later) to institutionalize immediate PPIUD services as part of routine ANC and delivery-room services reported a doubling of women choosing PPIUD at 24 months after the rollout of the intervention (from 4.1% before to 9.8% after).
The Jarvis et al.37 quasi-experimental study evaluated a 3-armed intervention that provided (1) free FP; (2) a “quality inputs” intervention consisting of clinical training and provision of PPIUD insertion equipment, training staff on WHO’s Medical Eligibility Criteria for contraceptive use, and a systematic FP screening and referral tool; (3) a combination of free FP and quality inputs (1 and 2). The authors reported benefits in the quality and free arms (arms 1 or 2), but the greatest effect in the free plus quality arm (arm 3) long-acting, reversible contraceptive (LARC) use within 12 months postpartum (OR=8.4; 95% CI, 3.4, 20.6). However, this study had a high risk of bias because several baseline group characteristics were dissimilar between the intervention arms compared to the control group, and the outcome data were incomplete.
The 4 reports in Nepal23–26 evaluated the 7-component intervention through another cluster stepped-wedge trial aimed at increasing PPIUD.25 This FIGO-designed intervention consisted of (1) informational workshops for female community health volunteers and general hospital staff; (2) training maternity care providers in PPFP counseling, PPIUD insertion, and management of complications; (3) PPFP leaflets distributed during counseling; (4) informational wall chart and video displayed in the hospital waiting area; (5) provision of Kelley’s forceps for IUD insertion and provision of IUDs; (6) designated service provider in each hospital as facility coordinator for the program; and (7) regular monitoring of counseling and insertion data. The results showed an increase in PPIUD uptake in the immediate postpartum (adjusted risk difference of 4% (95% CI=3, 6).
The Huber-Krum et al.24 analysis of the Nepal FIGO package intervention looked at the effect of the intervention on modern contraceptive prevalence and method mix rather than just on PPIUD insertion immediately after birth. This analysis found an adjusted risk difference of 4% (95% CI=−0.00, 0.10) more uptake of modern methods at 1 year postpartum, demonstrating a small and borderline statistically significant effect. They also found the 4% increase in PPIUD uptake in the intervention group observed by Pradhan et al. was somewhat diminished to 3% at 12 months postpartum but remained statistically significant (95% CI=0.02, 0.04). By 24 months, most differences between the intervention and comparison groups had disappeared, except for PPIUD, which had an adjusted risk difference of 2% (95% CI=0.01, 0.03).
The Karra et al.44 package intervention consisted of 4 services over a 2-year period: up to 6 FP counseling sessions, free transportation to an FP clinic, free FP services, and treatment for contraceptive-related side effects. The authors of this large cluster stepped-edge trial (39,084 women in 6 hospitals) reported that contraceptive use after 2 years of exposure to the intervention increased by 5.9%, mainly through increases in the use of implants.
Tran et al.29 conducted a cluster-randomized trial in Burkina Faso evaluated a package of 3 facility-oriented interventions (refresher training of service providers, regularly scheduled and strengthened supportive supervision of providers, enhanced availability of services 7 days a week) and 3 individual-based interventions delivered during third-trimester ANC visits and postnatal care follow-up visits (a PPFP counseling tool, appointment cards for women, and invitation letters for partners). The authors reported that the intervention improved the prevalence ratio of modern contraceptive use at 12 months (adjusted prevalence ratio 1.7, 95% CI=1.3, 2.47). The intervention had its greatest effect in promoting uptake of FP at 6 weeks, with an adjusted prevalence ratio of 3.88 (95% CI=1.46, 10.35), but the effect reduced somewhat over time, with the 6-month adjusted prevalence ratio at 2.31 (95% CI=1.44, 3.71).
Cooper et al.56 evaluated a package intervention in Egypt called the SMART project, using a controlled before-after design. SMART aimed to decrease child malnutrition through an “integrated, community-based reproductive and maternal and child health intervention package.” SMART was implemented during a political transition in Egypt and, as such, FP was deemphasized in the beginning to respond to the political and social climate. The intervention consisted of 3 overarching components, within which existed various activities: one-to-one and group counseling, training health care workers, and mobile clinics. The counseling covered the benefits of FP, healthy timing and spacing of pregnancies, postpartum return to fertility and pregnancy risk after childbirth, and LAM and transition to other modern contraceptive methods. The project coordinated with local health directors to advocate for the availability of FP commodities at government health facilities and mobile clinics. Women in the intervention group were more likely to use a modern method of contraception (Upper Egypt: OR=1.45, P<.001; Lower Egypt: OR=1.29, P<.05). However, the difference was not statistically significant for mothers with children aged 11 months or younger (Upper Egypt: OR=1.13, P>.05; Lower Egypt: OR=1.20, P>.05), which authors said could be due to differences in the duration of exposure to the program during the antenatal and postpartum periods among mothers with children aged 11 months or younger compared to mothers with older children.
The package intervention evaluated by Wu et al.51 included active pregnancy identification, home-based ANC and postnatal care counseling and care coordination, patient-centered contraceptive counseling, group ANC, and home-based childcare and counseling. The intervention was longitudinal, starting in the antenatal period and continuing through the first year postpartum. Counseling topics covered recommendations and reasons for birth spacing, contraceptive efficacy, contraindications, and timing for postpartum method initiation. Training materials emphasized best practices for contraceptive counseling (e.g., shared decision-making, patient autonomy, and guidance on potential side effects). This uncontrolled before-after study reported that use of any modern contraceptive method increased from 29% pre-intervention to 46% post-intervention (P<.0001). Participants spoke about how much they valued the continuity of care and relationships with the providers, which the authors said likely helped address women’s changing contraceptive needs during the first year postpartum.
Twelve studies used a counseling (n=7)32,36,38–40,43,46 or an education (n=5) intervention,35,41,42,48,54 and 231,33 used an SMS intervention. Thirteen of these 14 measured the outcome of interest. Eight of these 13 (62%) showed a statistically significant impact on uptake of contraception, although most assessed use well before 12 months postpartum. In general, the magnitude of the intervention was strongest the shorter the time period examined, although in 1 case,43 the effect was greater at 9 months than at 6 months postpartum. The magnitudes of the risk differences for modern/effective methods of contraception ranged from -0.1% at 3 months,48 17.5%38 at 16 weeks, 22%39 at 6 months, 7.1%35 and 6.4%36 at 8 months, 4.6%43 and 24.2%54 at 9 months, 40.1%42 at 1 year, no significant differences at 6–15 months,40 and 26.3% in a survey 2 years (midpoint 12 months) after baseline.41
Quality Appraisal Overview
Most studies described their source population well (n=27, 87%), defined an eligible population that was representative of the source population (n=26, 84%), described the intervention well (n=28, 90%), and described the comparator well (27/27, 100%) (Table 3). Among the RCTs (n=16), only 5 (31%) had a low risk of bias regarding the random sequence generation, 2 had a high risk of bias (13%), and 9 (61%) had an unclear risk of bias. Among the 26 studies where it was applicable, 18 (69%) had a low risk of bias regarding the comparability of the baseline (or group) characteristics.
DISCUSSION
Summary of Findings
We double-screened 771 records and included 34 reports on 31 studies in the review. This is a substantial increase from the 16 studies identified by Cleland et al. for the period before 2012.4 Our study is the first review to capitalize on the increasing recent research and the growing interest in PPFP among donors and policymakers. The majority of included studies (n=21) were published since 2018. Most (n=21) were conducted in sub-Saharan Africa, with half of the studies enrolling pregnant women (n=17). Twelve studies had interventions in the antenatal period only.
Approximately half of the study designs (n=16) were RCTs, and half (n=15) were quasi-experimental. The quality appraisal showed generally positive results, with most studies describing their source population, representativeness, intervention, and comparator well. In the trials, random sequence generation could have been improved in many studies. Most evaluations (n=24) were conducted in health facility settings (with 2 also in community settings); 7 were in community settings only (these findings are not very dissimilar to those reported by Cleland et al.). Interventions were heterogeneous, with the distinction between counseling and education interventions not always clear.
Twenty-four of the 31 studies reported on the main outcome of interest in this review (postpartum contraceptive use within 1 year of birth), with 18 of these 24 studies reporting a positive intervention effect—compared to 9 of 16 of the older studies reviewed by Cleland et al. In general, the magnitude of the effect of the intervention was stronger the shorter the time period examined; studies that looked over several time points saw a decrease in the effect over time, although in 1 case, the effect was greater at 9 months compared to 6 months postpartum. This could suggest the effect of the interventions encouraged women to start earlier, but that over time, even women in the nonintervention groups take up contraception. It could also suggest that while women who receive the intervention start a method earlier, they may also discontinue. Interventions may also encourage women to adopt specific methods but not usage overall. For example, interventions that target uptake of LARC uptake, which have a higher continuation rate at later time points may not be comparable to interventions that do not specifically promote a long-acting method. Given that effect sizes tended to decrease over time, the different times at assessment of contraceptive usage could be expected to greatly affect the comparability of the studies.
Although the studies in this systematic review were heterogeneous, the findings suggest that the interventions that included a multifaceted package of initiatives appeared to be more likely to have a positive effect. By contrast, interventions with minimal counseling did not appear to be as effective. Only 2 of the 10 multifaceted packages were delivered during the antenatal period only, and given the relatively small number of studies for any given type of intervention, it is challenging to draw additional conclusions beyond the broad ones made earlier. For the multifaceted package intervention, assessment of the effect of the individual components of the intervention is not possible as they were evaluated overall. Future studies could assess the influence of the different intervention components on each other to understand how they work together to affect postpartum contraceptive uptake.
The findings suggest that the interventions that included a multifaceted package of initiatives appeared to be more likely to have a positive effect.
To aid in the comparability of findings across studies, we recommend methodological improvements. Future evaluations of ANC interventions to increase postpartum contraceptive use could measure use at standardized key time points (e.g., at 3, 6, and 12 months). Moreover, future implementation research needs to more specifically address the perceived susceptibility (i.e., how protected against pregnancy someone thinks they are when using LAM) and actual susceptibility (i.e., how protected against pregnancy they actually are) when using LAM, perhaps by looking at waiting time to conception, and on how reliance on LAM and transitioning from LAM can affect timing of uptake of other methods. Future research could also focus on documenting more definitive results on counseling interventions in the context of group ANC versus individual ANC and on the most promising packages that consider specific contexts and health service gaps.
Strengths and Limitations
Our search incorporated all LMICs in scope, was built on similar preexisting reviews, and reviewed 3 databases. All records were double-screened, with reviewers masked to other reviewers’ decisions. Our synthesis approach was robust, and there appeared to be enough information to assess eligibility of the studies. The quality of the included studies was generally good. The measures taken to minimize bias in the review included double-screening and double extraction; however, the quality appraisal was conducted by 1 person. Compared to the older reviews, which primarily involved counseling, the studies covered a wider range of intervention approaches that focused on, for example, training providers, client voucher provision, systems strengthening, and pay for performance. While this review did not produce clear evidence for the effectiveness of these “new” approaches, it indicates that novel approaches are being considered and highlights the need for further evaluation of them.
The heterogeneity in interventions, population, outcomes, and time period when outcomes were assessed made it difficult to identify themes and precluded conducting a meta-analysis. This means the body of evidence may be too small to make definitive recommendations for any given approach. Moreover, we did not fully consider all the relationships between the intervention characteristics and all the outcomes or how contextual variables could affect the outcomes. For example, a systematic review by Gahungu et al.57 on unmet need for PPFP among women in 5 sub-Saharan African countries identified individual-level factors associated with unmet need, such as low perceived risk of pregnancy, resumption of menses and sexual activity, and factors on the household level, such as husband’s approval and spousal discussion about FP. Few studies in our review reported on these variables, making it difficult to assess potential confounders or contextual differences explaining the magnitude of effects in the different studies. Studies promoting multiple methods (e.g., LARCs and LAM) might see declines in some methods relative to others, whereas studies promoting only 1 method were more likely to see use of this method increase.
There was an additional limitation related to the timing of the intervention delivery. We included interventions that were not exclusively delivered during the antenatal period. The benefit of this is that the review results present a more complete picture of the range of antenatal interventions to increase contraceptive use following birth rather than being restricted to the 12 studies that described antenatal-only interventions. Indeed, it can be argued that an intervention integrated into a continuum of pregnancy care is desirable. The limitation of this, though, is that any effect of an intervention delivered across the different periods cannot be attributed to its delivery during the antenatal period and that comparisons between the periods cannot be made. However, attribution to the antenatal period or other period was not the aim of this review.
CONCLUSIONS
The studies in this systematic review were heterogeneous, and the number of studies with any given approach was small. Nevertheless, the findings suggest that interventions that included a multifaceted package of initiatives appeared to be more likely to have a positive effect and that those with minimal counseling did not appear to be effective. To help avert short interpregnancy intervals and improve health outcomes for mothers and babies, funders, policymakers, and providers may consider strengthening their support for multifaceted ANC interventions delivered over multiple time points that promote voluntary informed choice regarding postpartum contraception.
Acknowledgments
We thank Jane Falconer from the London School of Hygiene & Tropical Medicine Library Services for her review of our search strategy and for sharing her search terms for identifying low- and middle-income countries. We acknowledge the support of Abinaya Chandrasekar in extracting data from some articles. We thank the U.S. Agency for International Development family planning/reproductive health team for their support and review of this report.
Funding
This work was supported by the U.S. Agency for International Development under the MOMENTUM Safe Surgery in Family Planning and Obstetrics award (cooperative agreement # 7200AA20CA00011).
Disclaimer
The contents of this article do not necessarily reflect the views of the U.S. Agency for International Development or the United States Government.
Author contributions
OM, NF, and OMRC conceptualized the review, developed the search strategy, conducted the search and screened the records. OM completed the formal analysis, wrote the original draft and reviewed and edited subsequent versions of the manuscript. VT, RS, KL, FK provided intellectual input regarding key decisions in the review. MP conducted the double extraction and updated tables. All authors critically reviewed the document and approved the final version for publication.
Competing interests
None declared.
Notes
Peer Reviewed
First Published Online: September 16, 2024.
Cite this article as: McCarthy OL, Fardousi N, Tripathi V, et al. Antenatal care interventions to increase contraceptive use following birth in low- and middle-income countries: systematic review and narrative synthesis. Glob Health Sci Pract. 2024;12(5):e2400059. https://doi.org/10.9745/GHSP-D-24-00059
- Received: January 18, 2024.
- Accepted: August 1, 2024.
- Published: October 29, 2024.
- © McCarthy et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly cited. To view a copy of the license, visit https://creativecommons.org/licenses/by/4.0/. When linking to this article, please use the following permanent link: https://doi.org/10.9745/GHSP-D-24-00059