Data sources

We used a recently assembled database containing assessments of 548 studies, projects or programs (referred to collectively as projects) that used CBPHC (defined in the initial paper in this series [24]) to improve neonatal or child health (henceforth referred to as child health) and to document these improvements. In brief, CBPHC was considered to be one or more interventions carried out in the community outside of a health facility. The additional presence of one or more facility–based interventions did not disqualify the project from inclusion.

The database and its assembly have been described elsewhere in this series [24]. In short, peer–reviewed documents, reports and books assessing the impact of one or more CBPHC interventions on child health (coverage of a key child survival indicator, nutritional status, serious morbidity, or mortality) in LMIC settings, among children in a geographically defined population, were selected. Two independent data extraction reviews were carried out and followed by an independent consolidated summative review. Data from the latter review were transferred to electronic database.

From this database, we identified a sub–set of 42 projects that had carried out an equity analysis as part of their assessment using the process described in the following section.

Article review and inclusion process

Using the PRISMA guidelines for systematic reviews on health equity [25,26], we identified a sub–set of 138 articles in which equity was mentioned in one or more of the following fields in the CBPHC project database: 1) the title of the article, 2) the documentation of the process of the intervention, 3) part of the data analysis strategy, or 4) in the notes provided by the reviewers of the assessment for inclusion in the systematic review. We carefully reviewed this sub–set of equity–relevant assessments and excluded assessments in which equity was not actually analyzed across population subgroups. After this focusing phase, we were left with 43 projects to examine further.

Two of the authors (MS and RK) separately reviewed each of these 43 projects and extracted additional data on how equity was defined in each assessment, what data sources were utilized for assessment of equity effects, and what the outcome on equity actually was. The metrics from each project being assessed were stratified into log–frame categories (input, process, output, outcome, impact). One article was excluded from the analysis because it did not provide sufficient information on how equity was analyzed, leaving 42 articles in the final data set ( Figure 1 ). Aside from the availability of adequate information on equity analysis in each article, the quality of the study was not assessed.

Figure 1.  Overview of sequence of article review and inclusion/exclusion criteria.

Criteria for equity analysis

In order to identify the diverse criteria utilized among the studies to analyze equity, we created open–text responses as we reviewed each assessment, and then categorized them into common themes as we identified commonalities among the identified categories. We summarize the categories below and provide examples for some of the less–common categories. In our literature review, we identified a USAID report [27] on incorporating equity into project designs for MNCH that offered guidance on identifying disadvantaged groups that should be considered in implementing equitable MNCH projects. The USAID report referred to these groups by the acronym PROGRESS ( P lace of residence, R ace, O ccupation, G ender, R eligion, E ducation, and S ocioeconomic Status) [28]. This typology provided guidance for the kinds of characteristics to look for and how to organize the findings from the reports we analyzed.

Categorization of equity outcomes

We created three categories of outcomes for the various equity indicators used by the assessments included in our analysis (pro–equitable, equitable, and inequitable, as defined in Box 1 ). We categorized indicators as pro–equitable if findings favored underserved populations and were statistically significant or, if tests of statistical significance were not carried out, the study authors described their results as having practical significance. Indicators with findings that were similar for underserved groups as for the other groups were categorized as equitable. Indicators with findings that showed unfavorable outcomes for underserved populations were categorized as inequitable.

These categories helped us to differentiate between several important equity outcomes –namely when disadvantaged sub–groups were benefitting less, equally, or more than other sub–groups. If disadvantaged groups were benefitting less, this was an inequitable outcome. When disadvantaged groups were benefitting equally, this was noted as a good sign, though not a fully optimal outcome since disadvantaged groups often need to make additional progress in order to overcome inequities.

Organization of identified metrics for health equity into a logical framework

Barros et al. [19] offer a framework for analysis of health equity from the standpoint of an individual person’s experience with an illness, beginning with the socioeconomic context through exposures to disease, vulnerability to succumbing to disease, and the outcomes and consequences of illness. While this approach helped us think through the various ways that equity can influence child health work, we opted to organize the indicators of health equity used by the assessments included in our analysis by utilizing a different framework of analysis from the standpoint of project implementation: beginning with inputs and processes, and then moving to outputs, outcomes, and impacts [29] to track at what point in project implementation equity dimensions were assessed. This made it possible to identify gaps and opportunities from a project planning and implementation perspective. Figure 2 below provides a graphic representation of the conceptual flow of this log–frame matrix from one phase to another.

Figure 2.  Generalized log frame for health projects.

We created a matrix for each phase of the logical framework and, for each of the included indicators, tabulated the equity effects of each project. For each cell of the matrix, we described the content of the project and drew conclusions from the available evidence.

Each assessment was further analyzed to determine the criteria used to define equity, the type of data used to assess equity, and the scope of the assessment as well as the types of indicators measured in the assessment. The definition of equity was not pre–determined, and the definitions of equity used in the assessments were categorized after the list of equity indicators used in the projects had been reviewed. This was done to avoid missing any relevant equity indicators that might not have fit into a pre–determined definition of equity.

The type of data used to assess equity was defined as primary or secondary. The term primary data refers to data collected by the project, while the term secondary data refers to data which were gathered by another entity. Secondary data included those obtained from DHS and MICS data sets. Finally, each indicator was further classified as to whether it was referring to a project input, process, output, outcome or impact.

Location of included projects

The assessments included in our analysis were for projects from various regions of the world ( Table 1 ). One of the studies included data from 28 African countries, and another had data from four African countries. All other studies focused on one country or a smaller sub–population within that country as shown in Table 1 .

Kinds of data used in the assessments

The data utilized in 37 of the 42 projects including equity analyses collected specifically by the project within the project’s geographic area. However, five analyses exclusively utilized data from DHS and MICS surveys, and two utilized both project–level data collected for assessment of the project and also publicly available national data.

Criteria through which equity effects were assessed

Across the 42 projects included in our analysis, 82 equity indicators — for example coverage of prenatal home visits analyzed across household income categories (Callaghan–Koru, 2013; reference [S15] in Online Supplementary Document(Online Supplementary Document) ) — were identified. Equity was measured by comparing changes in health program characteristics or health status over time for more disadvantaged groups with changes in the identical indicators for more advantaged groups. Table 2 summarizes the criteria by which disadvantaged groups were distinguished from more advantaged groups.

We grouped several equity indicators under a category we refer to as socioeconomic status (SES). These included income categories, maternal education, and household characteristics. By far, the most common indicator for assessing equity was a measure of wealth, often based on household income, household assets, household size, or maternal earnings. Other SES equity indicators included in the analysis were agricultural production by heads of household and specific assets present in the household such as a working toilet, running water, or a refuse collection system. Other SES criteria included the ethnic group of the family, religion, marital status of child’s parents, occupation of the parents, and demographic characteristics such as maternal age. These equity indicators aligned well with those identified by the USAID PROGRESS report (shown in the right–hand column of Table 2 ); the only PROGRESS category that was not identified in our analysis was religion [27].

Assessments of equity of inputs

After careful analysis and discussion among co–authors and colleagues, we determined that no projects that we included in our data set explicitly analyzed or reported inputs from an equity perspective. The dearth of input–related efforts in project design, implementation, and evaluation is concerning and is noted as an area where further work is needed.

Assessments of equity using processes

A number of the assessments included in our review measured process indicators through an equity lens, as shown in Table 3 (references in Tables 3–6, are prefixed with an S and appear in Online Supplementary Document(Online Supplementary Document) ). Two–thirds (10/13) of the measurements of equity involving process indicators concerned whether the household had received a home visit from a health worker or had contact with the health system. Eleven out of 13 of the measurements yielded a pro–equitable result, and the remaining two yielded an equitable result. Thus, for the process indicators in the assessments selected for analysis, equity had been achieved in all cases and a pro–equity result is observed in almost all. The findings for this portion of the log frame consistently support the equitable nature of home visiting practices, a central feature of many CBPHC projects, as also discussed in the in this supplement that directly address the effectiveness of CBPHC in improving MNCH [30–32]. Many of these home visits either implicitly or explicitly included promotion and support of breastfeeding, which has also been noted in the literature as an intervention that can be supported equitably through community–based approaches with multiple benefits to MNCH.

Assessments of equity of outputs

The assessments of equity using output indicators are listed in Table 4 . Two–thirds (4/6) of the six equity assessments using output indicators among the projects selected for our analysis concerned the utilization of specific services or the expected immediate output of an intervention. Half (3/6) of these equity assessments used household income as the equity criterion. The number of assessments is too small to make major generalizations from, but the indicators demonstrating a pro–equity effect in the output category focus on access to health services (either in a facility or in the home). Indicators that demonstrated an inequitable effect were both from the same study and related to the hygienic practices across several equity dimensions.

Assessments of equity of outcomes

Table 5 below lists the equity assessments carried out using outcome indicators. Many relate to knowledge and behavior change related to breastfeeding or to the population coverage level of an intervention. Of the 35 measurement carried out, only 14% (5/35) yielded an inequitable result; 11% (4/35) yielded an equitable result, and the rest (74%) yielded a pro–equitable result. Inequitable indicators included several interventions requiring significant equipment or knowledge such as vaccine coverage and antenatal and delivery care. Some indicators — such as ITN coverage, availability, and use — showed mixed results across different studies, with some having equitable results across household income categories or urban and rural settings and others not. Equitable and pro–equitable programs commonly focused on equitable behaviors such as breastfeeding and newborn and child health practices that can be implemented in the home without complex or expensive supplies or knowledge.

Assessments of equity of health impact

Finally, Table 6 lists the assessments of health equity that were carried out for health impact–related indicators (nutritional status, morbidity or mortality). Of the 28 projects that included an equity assessment of health impact, 20 were based on a measure of mortality; four were based on a measure of morbidity and four on a measure of nutritional status. Overall, 23 of the 28 assessments demonstrated pro–equitable results and one yielded an equitable result. Only four of the 28 yielded an inequitable result.

Overall summary of equity effects using household wealth as the equity criterion

We have summarized all the findings reported above in which household income was the equity criterion ( Table 7 ). Overall, 75% (33/44) of these effects were pro–equitable outcome, 9% were equitable outcome, and only 16% (7/44) yielded an inequitable effect.

Overall summary of all equity effects

Finally, we have summarized equity effects in Table 8 . Overall, 78% (64/82) of the equity assessments carried out yielded a pro–equitable outcomes; 9% (7/82) yielded an equitable outcomes, and only 13% (11/82) yielded an inequitable outcome.

While in–depth analysis of the impact of packages of interventions was not the focus of this paper (another paper in this series [33] addresses this strategy in general – not limited to equity), we reviewed which projects constituted a single intervention vs a package of interventions. Of the 42 projects, 11 (26%) included a single intervention while eight (19%) included 2 interventions, and 23 (55%) of projects had a package of three or more services. We could not identify any clear patterns between the number of interventions and how equitable the findings were; the only clear pattern was that, in general, all interventions and equity dimensions within any particularly project tended to be the same in terms of equity outcomes (eg, all of the findings for Ahmed 1993 were inequitable).

Of the 42 projects that conducted an equity analysis, we also reviewed which ones analyzed more one or more dimensions of equity. 27 (64%) included an analysis for only one dimension of equity while nine (21%) included two dimensions of equity, and only six (14%) included three or more dimensions of equity. We also did not identify any obvious patterns among the small groups of projects in each of these categories. Household income as part of SES was by far the most common dimension of equity, and was utilized across all of these categories followed closely by comparing urban vs rural populations. The projects with inequitable findings included a number of SES analyses and also child gender and an urban vs rural comparison.

Explaining the pro–equity effects of community–based primary health care

Most CBPHC projects are designed to reach every household with health education and information about how to access outreach services (if not to actually provide services including curative care), and outreach services are generally distributed more evenly throughout target populations than facility–based services [35]. Meanwhile, some countries, such as Peru, where great investment in health facilities has taken place—including expansion of community health centers—these efforts have resulted in only very small improvements on equitable utilization of health facilities [36].

Research on the equity of facility utilization in low–income settings is limited; more evidence is available for high–income settings in the Americas, Europe, and Asia. In LMIC settings, health facilities tend to be few and far between, often expensive from the perspective of the poor, and lacking high quality of care, including provision of care that is seen by certain sub–groups as disrespectful [36–38]. Factors such as education level, income, and urban and rural residence play key roles in determining whether someone is more or less likely to seek care at a health facility [36,37]. Thus, the effort and resources that patients and their families have to expend to reach a health facility and the uncertain return on that family’s investment contributes to low utilization of facility–based services.

The challenge of providing interventions that are often only available in health facilities – or require infrastructure and skills difficult to deploy in communities outside of facilities—is significant as well. A growing literature, including but also going beyond the database used in this study, points to inequitable usage of health facilities in terms of the SES and urban/rural characteristics of users [23,34,39].

The need for alternative approaches beyond health facilities to achieve equity in and in fact universal coverage for child health are the following: (i) there is an exponential decline in the utilization of health facilities with increasing distance to the health facility (particularly more than 5 km or 1 hour walk away) [35], and (ii) there is a need for available and affordable public transportation in order to reach health facilities, which is often absent [33,39]. What is lacking from the literature are in–depth assessments of equity of health care utilization in terms of distance from a health facility and the effect of distance from health facilities on health status, taking into account also whether community–based care is available to those further away from those facilities.

Strong community–based programs can encourage facility utilization across income strata as can vouchers provided at the community level for specific services, such as antenatal care, to reduce resource barriers to seeking care [40]. The available evidence suggests that CBPHC approaches that reach all households can be more equitable than solely facility–based approaches in terms of coverage of a number of key primary health care services, particularly for vulnerable populations and those who live further away from facilities, who are also usually more disadvantaged in terms of SES [20,41–43].

There are several assessments that directly compare the degree to which CBPHC approaches as opposed to other approaches improve the health of the poorest segment of the project population compared to that of the better off segment. It makes sense that home–centered, low–resource interventions like breastfeeding promotion and distribution of ITNs would be able to achieve high levels of equity through community–based approaches that often include direct contact with all households [7]. In addition, some of the most promising strategies to improve health equity focus on strengthening community outreach, using CHWs and other lay workers, along with market–driven options such as minimizing or removing user fees and engaging the private sector [3,44,45].

Approaches that make it possible for health workers to reach all households – or at least to reach outreach points that are relatively evenly distributed throughout the project population and close to homes – are inherently more likely to achieve favorable equity effects than facility–based approaches. However, a number of other equity–relevant factors including education, child’s sex, ethnicity [46], and urban vs rural contexts [47] cannot be overlooked even within such a strong outreach approach [48]. Health programs in high–mortality, resource–constrained settings lack the capacity to build and operate facilities within easy reach of all who could need to use them – particularly in low–density rural areas. Thus, the decentralization of services and utilization of innovative and proven strategies to support the coverage, quality, and sustainability of those services is essential for achieving health equity.

While the focus of this review is on low–income countries, inequities are also prevalent in higher–income countries as well. Even where more resources are available to address such issues, political will is needed to direct those resources in ways that decrease inequities. An example of progress and success in the arena of health equity is Japan’s national policies to provide equitable educational opportunities as well as access to health services without financial barriers [49]. Globally, but particularly in low–income countries, much work remains to be done to make this kind of progress a reality for all populations. In addition to our public health–specific tools and approaches, more comprehensive community development and empowerment frameworks, such as the CHOICE ( C apacity–building, H uman rights, O rganizational sustainability, I nstitutional accountability, C ontribution, and E nabling environment) framework [50], can help to frame issues of health equity and provide additional entry points for understanding and addressing them. As Victora et al. note [51], just using the data available and recognizing patterns in inequities is not enough; political will and deliberate design and attention to the causes of inequities in programs for child health are necessary to achieve substantial decreases in child mortality among the most disadvantaged sub–populations where the mortality rates are the greatest.

Community–based approaches can reach those furthest from health facilities and can rapidly expand population coverage of key interventions, so these findings are not surprising. These findings stand in stark contrast to the commonly observed finding that utilization of primary health care facilities is inequitable because those in the lower income quintiles are less likely to obtain services there [52,53]. To our knowledge, this is the first comprehensive review in the peer–reviewed literature summarizing the equity effects of CBPHC in improving child health.

Limitations of our study

This study has several limitations that we want to make explicit. First, we have not further disaggregated the articles based on how strong the equity effect is. Second, some of the 42 assessments qualifying for our analysis are efficacy studies conducted within community settings in which ideal conditions were present for project implementation. Therefore, we must be careful about generalizing these findings to everyday practice settings. But, that said, it still remains true that strong pro–equity effects are achievable through CBPHC. An analysis of the quality of the data included in the 42 assessments included in our review was beyond the scope of this article. Finally, although a thorough search has been conducted that covers articles published over the past six decades through the end of 2015, we know that there are likely to be more recent articles published since that time that are relevant to this analysis.

We have worked to be clear in our language, conservative in our claims, and yet optimistic about the role of community–based approaches to continue to help bolster health equity for children in disadvantaged populations around the world.