Effective Linkages of Continuum of Care for Improving Neonatal, Perinatal, and Maternal Mortality: A Systematic Review and Meta-Analysis

Background Continuum of care has the potential to improve maternal, newborn, and child health (MNCH) by ensuring care for mothers and children. Continuum of care in MNCH is widely accepted as comprising sequential time (from pre-pregnancy to motherhood and childhood) and space dimensions (from community-family care to clinical care). However, it is unclear which linkages of care could have a greater effect on MNCH outcomes. The objective of the present study is to assess the effectiveness of different continuum of care linkages for reducing neonatal, perinatal, and maternal mortality in low- and middle-income countries. Methods We searched for randomized and quasi-randomized controlled trials that addressed two or more linkages of continuum of care and attempted to increase mothers’ uptake of antenatal care, skilled birth attendance, and postnatal care. The outcome variables were neonatal, perinatal, and maternal mortality. Results Out of the 7,142 retrieved articles, we selected 19 as eligible for the final analysis. Of these studies, 13 used packages of intervention that linked antenatal care, skilled birth attendance, and postnatal care. One study each used packages that linked antenatal care and skilled birth attendance or skilled birth attendance and postnatal care. Four studies used an intervention package that linked antenatal care and postnatal care. Among the packages that linked antenatal care, skilled birth attendance, and postnatal care, a significant reduction was observed in combined neonatal, perinatal, and maternal mortality risks (RR 0.83; 95% CI 0.77 to 0.89, I2 79%). Furthermore, this linkage reduced combined neonatal, perinatal, and maternal mortality when integrating the continuum of care space dimension (RR 0.85; 95% CI 0.77 to 0.93, I2 81%). Conclusions Our review suggests that continuous uptake of antenatal care, skilled birth attendance, and postnatal care is necessary to improve MNCH outcomes in low- and middle-income countries. The review was conclusive for the reduction of neonatal and perinatal deaths. Although maternal deaths were not significantly reduced, composite measures of all mortality were. Thus, the evidence is sufficient to scale up this intervention package for the improvement of MNCH outcomes.


Introduction
The Millennium Development Goals (MDGs) call for reducing global maternal and child mortality by 2015. However, worldwide, approximately 289,000 women still died in 2013 due to complications of pregnancy and childbirth [1,2]. About 99% of such deaths occurred in lowand middle-income countries (LMICs) [1]. Furthermore, the total number of under-five children's deaths was 6.6 million in 2013, of which 80% occurred within 48 hours postpartum [1]. Only one-third of the 68 Countdown Countries are on track to achieve MDG4 (reduce child mortality), and the average annual rate of decline is far below that needed for MDG5 (improve maternal health) [1].
To improve maternal, neonatal, and child health (MNCH), various interventions have been recommended. However, most of the over 190 MNCH studies addressed only a single intervention [3][4][5][6][7][8]-few of them looked at how these interventions could be combined. Moreover, MNCH care programs are often vertical in nature; because newborn health is largely dependent on maternal health, care for both the mother and newborns should be provided in a continuous manner [9]. Indeed, around 20% of diseases among infants can be attributed to the mother's poor health status, malnutrition, and inadequate care during the perinatal period [10]. For instance, maternal treatment of infectious diseases such as HIV and malaria influences infants' health [11,12].
The continuum of care has been well known as a potential means of improving MNCH through integrated service delivery [13][14][15]. Recent MNCH studies have begun to focus more on interventions to strengthen continuum of care [16,17], although researchers have yet to establish a clear definition of continuum of care [3,14]. In MNCH, continuum of care has two core dimensions: sequential time and space [3,14,15,18]. The time dimension includes service delivery from pre-pregnancy to childhood, while the space dimension includes communityfamily care, outreach-outpatient care, and clinical care [19]. Previous studies estimated that up to 67% of neonatal deaths could be prevented by improving the coverage of common MNCH care to 90% of reproductive age women [7,20]. However, as resources are limited, providing all possible MNCH services is unrealistic, particularly in LMICs. For instance, about 0. 68-1.32 billion USD was required for providing a total continuum of care package from antenatal care to postnatal care to avert 38-67% of neonatal deaths in Sub-Saharan Africa [20]. As such, identifying the most effective MNCH package must be prioritized in health policies to ensure more effective use of continuum of care.
Postnatal care frequency and timing are also inconsistent, possibly due to changes in global postnatal care recommendations. The WHO's practical guidelines from 1998 encourage mothers with babies to access postnatal care services at "six hours, six days, six weeks, and six months" [41]. This formula targeted "crucial" moments of need for the mother and baby, though rigorous evidence was absent. Since then, different frequencies and timings have been recommended: in 2006, the timings were within the first week, preferably 2-3 days after delivery, and 4-6 weeks [42], and in 2009, within the first 24 hours after birth, on day three, and on day seven, if necessary [30]. This was based on evidence indicating high maternal deaths on the first and second days after birth [43]. Additionally, almost 40% of under-five deaths occur within the first 28 days of life [30,44], and three-quarters of neonatal deaths occur during the first week of life, with 25-45% in the first 24 hours [37]. Thus, we need to examine whether these standards for frequency and timing of care delivery should be considered in defining the continuum of care.
Although the expectation for continuum of care effectiveness is high and interventions have been planned, we found no prior review article or review protocol in this area. Comprehending the effectiveness of continuum of care linkages will encourage further opportunities for intervening in MNCH care and guide in tailoring interventions. Thus, in this systematic review and meta-analysis, we quantitatively synthesized evidence proposing effective linkages of continuum of care. Our focus was which linkages reduce neonatal, perinatal, or maternal mortality in LMICs in the continuum of care time and space dimension stages.

Materials and Methods
We conducted a systematic review according to the guidelines of the Cochrane Collaboration [45], and followed the four phases indicated in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement [46]. The PRISMA statement for our systematic review is available in S1 Table. We registered the protocol for this systematic review at the PROSPERO International prospective register of systematic reviews on October 25, 2013, and updated it on January 28, 2015. The registration number for this review is CRD42013006089. It is available at http://www.crd.york.ac.uk/PROSPERO (S1 Text).

Study Inclusion Criteria
We included only randomized and quasi-randomized controlled trials in this systematic review, including both individual and cluster-randomized studies. Regarding publication types, we included only peer-reviewed journals and reports from international organizations. We considered papers of all languages, so long as they had English abstracts. Furthermore, we included only articles set in LMICs as defined by the World Bank [47], and excluded non-randomized studies and non-intervention studies such as case series, case reports, and qualitative studies, as well as studies conducted in high-income countries.

Participants
We included women and neonates residing in LMICs. Women had to be in the pregnancy, intrapartum, or postpartum period. We excluded studies that included specific subpopulations that would have made it difficult to generalize the results to a wider population, such as HIVpositive women and groups at high risk for pregnancy complications.

Interventions and Controls
Interventions comprised packaged care/services that addressed different continuum of care time and space stages. According to the WHO and the Partnership for Maternal, Newborn and Child Health, the time dimension of continuum of care starts from pre-pregnancy and progresses to childhood and motherhood [14,[48][49][50]. However, given that we focused on neonatal, perinatal, and maternal mortality in the review, the target periods were only the pregnancy, birth, and postnatal periods of mothers and neonates. The space dimension of continuum of care comprises three care stages-community-family care, outpatient-outreach care, and clinic care [3]. The community-family care interventions included home or community-based care practices addressing mothers and neonates' healthy behaviors. The outpatient-outreach care included population-oriented services that health facilities or mobile services delivered routinely or periodically according to defined schedules. The clinical care interventions included individual-oriented care through facility-based care at primary and referral levels.
We defined the control groups as those who received standard care. In this study, standard care was defined as that provided in health facilities according to local or national guidelines.

Outcomes
We included studies assessing any of the outcomes below [51,52]. Since mortality is often the main outcome measure in MNCH studies, we selected it as the primary outcome [52]. We also included studies that had data from which we could calculate these outcomes.
1. Neonatal mortality is defined as deaths in the first 28 days of life among live births.
2. Perinatal mortality includes stillbirths and early neonatal deaths. Stillbirth is defined as the death of a fetus after 28 weeks of gestation but before delivery of the baby's head. Early neonatal death is death of a neonate within seven days postpartum.
3. Maternal mortality is defined as the death of a woman during pregnancy or within 42 days of termination of pregnancy.
We also reviewed relevant internet sources from the WHO and Google Scholar for additional grey literature. Finally, we searched for additional studies using the snowball method of looking through the reference lists of retrieved articles. We limited the publication period to 15 years, from 1999 to 2014, to ensure that we retrieved a sufficient number of studies. We used the following search strings combining the relevant medical subject headings (MeSH) with additional key/text words: (1) pregnant women, mothers, neonate, newborn, infant, or community; (2) maternal health or neonatal health; and (3) mortality or death. The specific electronic search strategy is provided in the S2 Text.

Data Collection
Selection of studies. The study selection process is summarized in the flowchart in Fig 1. Two review authors (KK and SO) independently extracted data, and screened the quality and content of the included studies. We identified articles by analyzing the titles and abstracts for relevance and compliance with the selection criteria based on the research setting, study design, and reported outcomes. We classified articles as included, excluded, uncertain, or duplicate. We confirmed all potential included or uncertain studies and resolved any disagreements by consensus.
Data extraction and management. KK and SO extracted the features of each study (e.g., study design, setting, components of intervention package, and outcomes) and entered them into a standardized form. KK performed the extraction and SO checked for data accuracy and completeness. Again, when they noted discrepancies, the two review authors discussed until they reached an agreement, or consulted a third review author.

Assessment of Risk of Bias on Included Studies
KK and SO assessed the quality of trials using the "risk of bias" tool presented in the Cochrane Handbook for Systematic Reviews of Interventions [45]. Specifically, we assessed the risk of bias in seven domains: sequence generation, allocation concealment, blinding of participants and outcome assessors, incomplete outcome data, selective outcome reporting, and other potential threats to validity. We resolved all discrepancies by consensus.
The summary of the risk of bias is shown in Table 1. Several studies lacked information related to the risk of bias or these risks were unclear. Due to the nature of the study design, allocation concealment was not an issue in the cluster-randomized studies we extracted [45]. However, we counted baseline imbalance in determining selection bias.

Measures of Treatment Effect
We present results as risk ratios (RR) with 95% confidence intervals (CI) for all randomized and quasi-randomized controlled trials.

Unit of Analysis Errors
We checked whether the study contained a unit of analysis error. In cases where we observed this error, we re-analyzed the available data. No analysis error was found.

Analysis
First, we narratively summarized the included study interventions. Second, we stratified the studies by the linkages of continuum of care time and space dimension stages. Then, we created a hierarchically categorized matrix by service delivery mode with the RRs of applied mortality. Third, we meta-analyzed neonatal, perinatal, and maternal mortality separately for each   linkage of continuum of care time stages (antenatal care + skilled birth attendance + postnatal care; antenatal care + skilled birth attendance; skilled birth attendance + postnatal care; and antenatal care + postnatal care). Then, we analyzed all of these mortality types simultaneously to estimate a composite neonatal, perinatal, and maternal mortality risks for each linkage of time dimension stages. Finally, we stratified the studies by space dimension and estimated the composite risk of neonatal, perinatal, and maternal mortality. The software RevMan 5.1 was used for the analysis. We assessed the heterogeneity of the meta-analyses from the P-values at a significance level of <0.10 and the I 2 statistic. We used random-effects models to adjust for possible heterogeneity.

Results
We retrieved a total of 7,142 articles from the sources (Fig 1). Of them, 3,046 studies were identified from the PubMed database, 4,087 through other databases, and nine through hand searching. After an initial screening, we excluded a total of 7,056 studies because they were duplicates or irrelevant to the research question. Of the 86 studies remaining, we excluded a further 53 because they had no outcome/intervention or treatment (n = 12), no relevant design (n = 22), no clear comparison group (n = 12), or no relevant setting (n = 7). We assessed the remaining 33 studies for eligibility through a review of the full-text articles. After this full-text review, we included a total of 19 studies in the meta-analysis [53][54][55][56][57][58][59][60][61][62][63][64][65][66][67][68][69][70][71]. They were all in English and had been published. Of the 19 studies, 17 were cluster randomized controlled trials (RCTs) [53-67, 70, 71], one was a pre-post intervention design [69], and one was a quasi-experimental study [68]. The included studies were mostly conducted in Asia and three were in Africa. In total, 361,306 births/live births were involved ( Table 2).

Timing and Frequency of Antenatal Care/Postnatal Care
The timing and frequency of antenatal care or postnatal care services delivery differed among the included studies. Most of the studies involved four antenatal care visits, with the minimum numbers of antenatal care visits for the interventions ranging from one to four times during the pregnancy period. The timings of antenatal care, however, were not defined in most of the studies-indeed, only three studies reported the timings of antenatal care: twice at 12-16 weeks and 32-34 weeks in a study in Bangladesh [54], twice during the first and second trimesters in a study in Pakistan [64], and once in the third trimester in a study in Malawi [62].
In the case of postnatal care, the timings varied, ranging from zero days to five months. However, in most studies, postnatal care was provided for seven days at the very least. The frequency was also rather diverse, ranging from one to five times. Six studies did not make clear  Effectiveness of Care Linkages between Antenatal Care, Skilled Birth Attendance, and Postnatal Care Table 3 shows the results of the 13 studies that linked antenatal care, skilled birth attendance, and postnatal care. Except for one study conducted in Malawi, all studies were conducted in Asia (India, Bangladesh, Vietnam, Nepal, and Pakistan). Of the 13 studies, six included interventions that linked the three stages of the space dimension (i.e., community-family care, outpatient-outreach care, and clinical care). Although they linked the three time stages, the frequency of antenatal care visits ranged from two to four times between the studies. The frequencies and timings of postnatal care were not the same across studies. As for the components of the intervention, we noted some particular characteristics. For instance, primary-care-level health officers (e.g. lady health workers, community health workers) provided antenatal care or postnatal care within the community for women and children via home visits [54,56,61,68,69] and outreach services [59]. To improve care delivery, traditional birth attendants (TBAs) received training in the form of basic neonatal care, cleanliness during delivery, and referral of women for emergency obstetric care [54,56,59,68]. Interventions often included community meetings for women to promote MNCH. The facilitators of such interventions were TBAs, community health workers, or other recruited facilitators [54,56,61,65,67].
The intervention packages in these 13 studies resulted in significant reductions in mortality. Twelve studies assessed neonatal deaths. Among them, seven reported significant reductions in neonatal mortality as a result of the interventions. Furthermore, 10 of the 13 studies measured perinatal deaths. Of them, five reported significant reductions in perinatal mortality as a result of the interventions. Six studies measured maternal deaths. Of them, one study showed a significant reduction as a result of the intervention and one study showed a significant increase.
According to the meta-analysis, the interventions that linked antenatal care, skilled birth attendance, and postnatal care significantly reduced neonatal mortality (RR 0.84; 95% CI 0.75 to 0.94, random effects [12 studies, n = 184,260], I 2 82%, P < 0.01) and perinatal mortality (RR   (Fig 3). However, the interventions did not lead to a significant decrease in maternal mortality (RR 0.75; 95% CI 0.46 to 1.22, random effects [six studies, n = 96,691], I 2 68%, P < 0.01). The composite measure of all three mortality types was significantly reduced (RR 0.83; 95% CI 0.77 to 0.89, random effects, I 2 79%, P < 0.01). To explore the considerable heterogeneities observed in this meta-analysis, we explored the publication bias using a funnel plot (Fig 4). The included studies were symmetrically distributed on both sides of the combined effect size line, indicating no obvious publication bias. Table 4 shows the result of the one reviewed study that linked antenatal care and skilled birth attendance [64]. This study was a cluster RCT conducted in Pakistan. The study also linked the space dimension stages (i.e., community-family care, outpatient-outreach care, and clinical care). The intervention included sensitization for mothers only or for couples, TBA training for clean delivery and complication recognition, and emergency referral through vehicles or very high frequency communication systems. The intervention in this study did not lead to any significant reductions in neonatal mortality (RR 0.68; 95% CI 0.41 to 1.10); however, we identified significant reductions in perinatal mortality (RR 0.51; 95% CI 0.38 to 0.68) ( Table 4). No data were available for maternal mortality. We did not perform a meta-analysis for this intervention because of the lack of other studies providing impact estimates on mortality. Table 5 shows the result of the one reviewed study that linked skilled birth attendance and postnatal care [58]. The study was a cluster RCT conducted in India. The study included the three space dimension stages, and included interventions involving newborn care training for physicians, nurses, midwives, and TBAs.

Effectiveness of Care Linkage between Skilled Birth Attendance and Postnatal Care
This study measured only perinatal mortality, which the intervention did not significantly reduce (RR 0.91; 95% CI 0.75 to 1.11) ( Table 5). We did not perform a meta-analysis because of the lack of studies providing impact estimates on mortality. Table 6 shows the result of the four studies that linked antenatal care and postnatal care, but not skilled birth attendance. Two studies were conducted in Asia (Bangladesh and Pakistan) and the other two in Africa (Malawi and Ghana). All studies were cluster RCTs. Of them, three studies had interventions that linked all three continuum of care space stages, while the remaining study was concerned only with community-family care. The frequency of antenatal care visits ranged from one to three times among the studies. The frequencies and timings of postnatal care also differed across the studies.

Effectiveness of Care Linkage between Antenatal Care and Postnatal Care
As for the components of the interventions, all four studies included home visits for antenatal care or postnatal care. The practitioners who performed the home visits were primary-level health workers [55,57], community volunteers [60], or trained women's group members [62].
This intervention linkage resulted in significant reductions in mortality. All four studies measured neonatal deaths, but only two reported a reduction in neonatal mortality. Two of the four studies measured perinatal mortality, both reporting significant reductions as a result of the intervention. Maternal death was measured in one of the four studies, but was not significantly reduced [62]. Fig 5 shows the meta-analysis. Neonatal mortality (RR 0.81; 95% CI 0.67 to 1.00, random effects [four studies, n = 69,671], I 2 75%, P < 0.01) and perinatal mortality (RR 0.74; 95% CI 0.65 to 0.84, random effects [two studies, n = 15,591], I 2 0%, P < 0.01) were both significantly decreased in this linkage. We did not perform a meta-analysis for maternal mortality due to a lack of studies. The interventions employing a linkage of antenatal care and postnatal care significantly reduced a composite of the three mortality types (RR 0.79; 95% CI 0.69 to 0.91, random effects, I 2 68%, P = 0.005).

Effectiveness of Care Linkage between Time and Space Dimensions
Six randomized studies linked the three stages of the time dimension (antenatal care, skilled birth attendance, and postnatal care) and the three stages of the space dimension (communityfamily care, outpatient-outreach care, and clinical care). According to the meta-analysis results   (Fig 6). We did not find any obvious asymmetry in the funnel plot (Fig 7).

Discussion
This is the first systematic review to examine the effectiveness of interventions combining the continuum of care time and space dimensions for maternal, neonatal, and perinatal mortality. We included a total of 19 studies in this review; several results of some importance were obtained. First, none of the studies followed the standards of the continuum of care time dimension stages: four antenatal care visits; use of skilled birth attendance; and postnatal care within 48 hours, at seven days, and at six weeks [31]. However, two-thirds of the included studies did contain interventions that linked antenatal care skilled birth attendance, and postnatal care, albeit without considering frequency and timing of care; another two-thirds of the studies linked community-family care, outpatient-outreach care, and clinical care. Second, the interventions linking antenatal care, skilled birth attendance, and postnatal care led to significant decreases in the composite measure of neonatal, perinatal, and maternal mortality. However, the maternal mortality results were inconsistent. The composite measure of mortality also showed reductions in a meta-analysis of the interventions involving linkages between the time and space dimension stages.

Linkage Patterns of Time and Space Dimensions
As mentioned above, none of the studies followed the standards of continuum of care. However, without taking frequency and timing into consideration, linking the three time dimension stages reduced the risk of neonatal and perinatal deaths. In a recent systematic review, the perinatal mortality increased with reduced antenatal care visits, but the difference was of borderline statistical significance (three studies: RR 1.15; 95% CI 1.01 to 1.32) [24]. This could be because taking antenatal care would positively affect skilled birth attendance [72]. This may also support strategies that promote "goal-oriented" antenatal care visits [24,49], which directs attention to the quality of antenatal care, including its components and timing, more than the quantity of such care.
In our review, the frequencies and timing of postnatal care differed across studies. This may be because of the postnatal care changes in global recommendations by the WHO [30,41,42]. In our study, most of the postnatal care interventions included the first postnatal care visit within 1-2 days. However, the timing of additional postnatal care visits was not consistent. The risk of death is significantly elevated even after 42 days postpartum [73,74]. Regarding the frequency of postnatal care, there was no compelling evidence that more postnatal care visits are associated with fewer neonatal or maternal deaths [75,76]. Decisions on the frequency and timing of postnatal care could be better made if they were defined according to local needs [76].
All of the included studies contained community-family care in the space dimension, and two-thirds included links with outpatient-outreach care and clinical care. Community involvement is a positive factor, especially for reducing neonatal mortality, perinatal mortality, maternal morbidity, and the number of stillbirths, as demonstrated in a previous review [77]. Linkage between antenatal care, skilled birth attendance, and postnatal care. According to our meta-analysis, interventions that linked all three stages of the time dimension were most effective. Interventions utilizing a linkage of antenatal care, skilled birth attendance, and postnatal care appeared to reduce the risk of neonatal and perinatal deaths. Only the risk of maternal deaths showed inconsistent results. The study by Azad, which showed an increase in maternal deaths due to the intervention, did not provide a specific reason for this result. However, no deaths among mothers who attended women's groups, making it difficult to consider that the intervention would have influenced the negative results in maternal mortality. Notably, the significant increase was observed only when marginal populations were included; thus, omission error could have affected the intervention results. However, the intervention linkage of antenatal care, skilled birth attendance, and postnatal care was found to be effective for a composite measure of neonatal, perinatal, and maternal mortality. This suggests the importance of considering these three time stages in designing MNCH interventions in LMICs.
Linkage between antenatal care and postnatal care. The intervention packages that combined only antenatal care and postnatal care significantly reduced the risk of neonatal and perinatal mortality. This linkage was also effective in reducing the composite measure of mortality. Antenatal care is an entry point to MNCH care, suggesting that antenatal care may improve overall MNCH care uptake, which leads to better MNCH outcomes. However, this result also suggests that postnatal care might complement the effectiveness of antenatal care. Thus, linkages of antenatal care with postnatal care could be a key to designing effective MNCH interventions.

Effective Linkages of the Time and Space Dimensions for Reducing Mortality Risks
In the meta-analysis, the linkage of all time and space stages was effective for reducing neonatal, perinatal, and maternal mortality. This was among the primary evidence found in this Effective Linkages of Continuum of Care review study. The importance of both dimensions has been argued in the past, however, until now, evidence justifying the effectiveness of linking the continuum of care time and space dimensions has been scarce. Indeed, the continuum of care has been promoted without solid evidence of its effectiveness. Understandably, further trial evidence will be necessary to demonstrate the effectiveness of the continuum of care with high reliability. Nevertheless, our systematic review will serve as evidence in defending the effectiveness of the continuum of care.

Suggested Definition and Intervention of Continuum of Care
The findings of this review demonstrate the effective continuum of care linkages: thus we can use the results to better define the components of continuum of care and thereby exploit it for the design of future intervention. For the time dimension, the basic components of continuum of care could be linkages between antenatal care, skilled birth attendance, and postnatal care. However, interventions combining antenatal care with postnatal care would be also an effective option when resources are limited. The frequency of antenatal care should range between two and four visits. We could not clearly identify the timings of antenatal care in this systematic review. For postnatal care, timing appears more important than does frequency; since most of the included studies first implemented postnatal care at 0-2 days postpartum, this postnatal care timing should be at least considered. As for delivery, ensuring the presence of skilled birth attendants is not always possible. However, if TBAs receive training in appropriate skills, said TBAs could be included as assistants for post-delivery care.
For the space dimension of the continuum of care, a basic component of interventions is linkages between community-family care, outpatient-outreach care, and clinical care. We particularly recommend interventions based on community involvement, as per previous evidence.
Additionally, our review and meta-analysis results indicated that interventions that include home visit care would be effective for improving neonatal and perinatal health. Particularly, early postnatal care visits between 0 and 2 days would be important. These results suggest that a community-focused intervention would be a favorable design.

Limitations
This systematic review has two limitations. First, the lack of randomized studies may have obscured the real results of the meta-analysis. In particular, we found only one study each including interventions that linked antenatal care and skilled birth attendance or skilled birth attendance and postnatal care. Thus, for this linkage, we could not conduct a meta-analysis. Further evidence of these linkages is required to confirm their effectiveness.
Second, a number of groups showed significant statistical heterogeneity. We could have excluded studies that conflicted with the other studies from the meta-analysis. However, we could not identify the sources of heterogeneity; indeed, even after excluding the outlying studies, we identified heterogeneity. Furthermore, eliminating studies would also be unwise because it can introduce bias [45]. Thus, the findings need to be interpreted with caution.

Conclusions
Our review encourages continuous uptake of antenatal care, skilled birth attendance, and postnatal care to improve MNCH outcomes in LMICs. The review was conclusive regarding the reduction of neonatal and perinatal deaths. Although maternal deaths were not significantly reduced, composite measures of all mortality were. As such, the evidence is sufficient to scale up this intervention package for the improvement of MNCH outcomes.
Supporting Information S1