Costs and cost-effectiveness of community health workers: evidence from a literature review

Objective This study sought to synthesize and critically review evidence on costs and cost-effectiveness of community health worker (CHW) programmes in low- and middle-income countries (LMICs) to inform policy dialogue around their role in health systems. Methods From a larger systematic review on effectiveness and factors influencing performance of close-to-community providers, complemented by a supplementary search in PubMed, we did an exploratory review of a subset of papers (32 published primary studies and 4 reviews from the period January 2003–July 2015) about the costs and cost-effectiveness of CHWs. Studies were assessed using a data extraction matrix including methodological approach and findings. Results Existing evidence suggests that, compared with standard care, using CHWs in health programmes can be a cost-effective intervention in LMICs, particularly for tuberculosis, but also – although evidence is weaker – in other areas such as reproductive, maternal, newborn and child health (RMNCH) and malaria. Conclusion Notwithstanding important caveats about the heterogeneity of the studies and their methodological limitations, findings reinforce the hypothesis that CHWs may represent, in some settings, a cost-effective approach for the delivery of essential health services. The less conclusive evidence about the cost-effectiveness of CHWs in other areas may reflect that these areas have been evaluated less (and less rigorously) than others, rather than an actual difference in cost-effectiveness in the various service delivery areas or interventions. Methodologically, areas for further development include how to properly assess costs from a societal perspective rather than just through the lens of the cost to government and accounting for non-tangible costs and non-health benefits commonly associated with CHWs.


Introduction
In recent years, community health workers (CHWs) have received renewed attention in light of critical shortages in the health workforce and emphasis on strengthening primary healthcare systems for achieving global health goals [1][2][3][4]. CHWs are generally assumed to be a less expensive alternative compared with other cadres of health workers, notably with regard to salary and incentives as well as training costs. In parallel, more and more evidence has accumulated in recent years on the effectiveness of CHWs in delivery of essential health services in low-and middle-income countries (LMICs) [5][6][7]. However, studies assessing the costs and/or costeffectiveness of CHW programmes are limited due both to data and methodological problems [1,8,9]. Therefore, we conducted an exploratory literature review to:

provide an overview of what is globally known about
CHWs' costs and cost-effectiveness 2. identify methodologies and elements of costs, effects and cost-effectiveness included in and excluded from studies to date 3. discuss appropriate methodologies for evaluating the costs and cost-effectiveness of CHWs.
As this was an exploratory review, no estimates in monetary terms (dollar values) are presented. Nonetheless, the overall conclusions about the costs and costeffectiveness of CHWs will serve to inform policy dialogue around the role of CHWs in health systems, and findings about methodologies will encourage researchers to properly assess the costs and costeffectiveness of such programmes.

Definitions
For the purpose of this literature review, the definition of CHW that we used is the following [6]: "Any health worker carrying out functions related to health care delivery; trained in some way in the context of the intervention, and having no formal professional or paraprofessional certificate or degree in tertiary education". Costs are defined as the resources, either expended or foregone, associated with implementing a health programme or treatment. Costeffectiveness as a study type is defined as "one form of economic evaluation where both the costs and consequences of health programmes or treatments are examined" [10]. When comparing two programmes or scenarios, intervention A is said to be more "cost-effective" than intervention B when programme cost per unit effectiveness for A is less than for B. "Cost-effective" may also refer to a comparison with a threshold or benchmark. In the rest of the paper, we refer to consequences as benefits or effectiveness, defined as the change in desired outcome due to the intervention or programme. We refer to final patient outcomes (change in health status and/or well-being) wherever possible; where not available, measurable intermediate patient outcomes (for example, number of patients visited and number of visits conducted) and measurable CHW provider outcomes (for example, improved CHW productivity) are used.

Search strategy
We used the search results from a larger, systematic review on factors influencing performance of close-tocommunity providers, which included searching the EMBASE, PubMED, Cochrane, CINAHL, POPLINE and NHS-EED databases for the period January 2003 to April 2013 [2] as well as a manual search of reference lists of all papers. This broader review included quantitative, qualitative and mixed method studies, all in English, about CHWs working in promotional, preventive or curative primary healthcare in LMICs. From that search, we extracted costing studies, studies that assessed the costs and effects of a single CHW intervention and economic evaluations assessing the costs and benefits of at least two CHW interventions. We conducted an additional search in PubMed for articles published during the same period to verify the existence of any further relevant papers (see Table 1); this search was later updated to include articles published from May 2013 to July 2015. The search strategy is summarized in Figure 1 while the full search strategy is presented elsewhere [2].

Review approach
Three reviewers jointly developed two separate data extraction matrices. The first matrix captured the study or model's overall methodological approach as well as specifics regarding how costs, effects and cost-effectiveness of the CHW programme were assessed. The overall methodological approach included study design, perspective, time horizon, discounting, year of costing and currency, intervention and comparator(s), setting, scenarios, sensitivity analysis and software. In terms of costs, the data extraction form captured programme (training, recurrent, capital and overhead/indirect) and patient costs, costs averted, how costs are reported and data sources. A review of the outcomes included both patient and provider outcomes and was defined as final patient outcomes (change in health status and/or wellbeing), measurable intermediate outcomes (for example, number of patients visited and number of visits conducted) #2 "health economics" OR "economics, medical"[MeSH] OR "economic evaluation" OR "health care costs" OR "health resource allocation" OR "health resource utilization" OR costs OR "costs and cost analysis" [MeSH] OR "cost analysis" OR "cost-benefit analysis"[MeSH] OR "cost effectiveness" OR "cost effective" OR "health care costs" OR "cost benefit analysis" OR "cost-benefit analysis" and measurable CHW provider outcomes (for example, improved CHW productivity). The cost-effectiveness measure was also indicated. Although elements of quality were included in our data extraction and analysis, our review did not exclude studies based on a full assessment of study quality, because of the high diversity in types and focuses of the studies. The second matrix captured the study or model's findings in terms of costs, outcomes and cost-effectiveness. Findings from any sensitivity analyses were also extracted. Systematic reviews were summarized in terms of main CHW-related findings.
We piloted the abstraction process by having the three reviewers jointly analyse and discuss one article and then discuss as a team questions that arose during data extraction. All papers were then read and abstracted by a single reviewer. Each reviewer completed the data extraction matrix separately, and review results were compiled into a single matrix for analysis. Analysis was done by summarizing and discussing the data within the team, following the categories as presented above. For an overview of the review approach, see Figure 2.

Search results
The search strategy resulted in 32 individual articles about 31 studies being fully reviewed using the data extraction matrix and four review/discussion papers summarized.  Table 2 presents an overview of the methodological characteristics of included studies, while Table 3 outlines the included studies with their location, type of CHW, intervention description, perspective, cost components included, and any assessment regarding cost-effectiveness. These aspects of the included studies are discussed below.
The studies reported a variety of geographical areas and settings. Eighteen articles presented results from sub-Saharan Africa, nine from Asia and three from Latin America. Two articles included various countries in Africa (see Table 3). A number of different settings were included: home (n = 10), villages or general community (n = 7), health facility or health centre (n = 3) and workplace (n = 1). One study reviewed the experience of CHWs and mobile health (m-health). Several studies included CHWs operating in various settings (n = 4), while seven studies did not specify the exact setting.
Health priority areas addressed by CHWs included reproductive, maternal, newborn and child health (RMNCH, n = 13, including two reviews), tuberculosis (TB) (n = 6), malaria (n = 7) and a range of other disease areas or multiple areas including (problems regarding) hypertension, diarrhoea, malnutrition, pneumonia, common mental disorders and a range of primary care services (n = 7). CHWs working in RMNCH performed a wide variety of activities including basic curative activities, counselling and health promotion, referrals, prenatal care and support during home deliveries. CHWs involved in TB and malaria mainly administered directly observed therapy (DOT) of TB medicines and dispensed drugs. CHWs working in other disease areas were involved in different types of activities, ranging from health education and promotion, screening, diagnosis and management of some conditions to referrals.

Costs
In terms of programme costs included, all but six studies [12,13,17,22,25,26] clearly included the value of the CHWs' time spent (either compensated or, for        volunteers, opportunity cost or shadow price) and recurrent expenses such as materials, supplies, transport and supervision, although the individual unit quantities and costs were rarely reported. Eighteen studies included the value of capital items such as vehicles and equipment although the specifics were not always mentioned [11-13, 15, 18, 19, 21, 27-37]. Overhead costs were included in 17 studies [11, 13, 15, 18, 20-22, 27, 28, 30-37], for example, for TB, on the basis of the proportion of total health facility visits or inpatient days for which TB accounted. Three of the 13 studies used a flat rate of 15% or 30% [13,30,32]. Patient costs were included in 11 of the studies [12, 15, 18-22, 29-31, 37], including time for visits and hospitalization as well as transport, medicines, food and other expenses.
The studies relied on a wide range of data sources, including budget and expenditure files from health facilities, hospitals, districts, government price lists, patient questionnaires, literature, time sheets, payroll records, ministries of health and finance and project accounts.
Many of the studies did not estimate costs over a future time period; therefore, a discount rate was unnecessary. Four studies discounted costs at 3% [11,19,33,38] and two others at 5% [27,30]. The costing year ranged from 1996 to 2011, and all but one [23] reported in US dollars.

Outcomes
Various outcome measures reported by included studies are presented in Table 4. As for costs, many of the studies did not estimate benefits over a future time period; therefore, a discount rate was unnecessary. Two studies mentioned discounting future benefits at 3% [19,31] and another at 5% [30] in the base case or standard analysis. Data sources for outcomes included randomized trials, monitoring and evaluation systems, organizational and government offices, demographic surveillance systems and patient treatment registers. One study used assumptions about yearly incidence and disease progression [23].

Overall assessment of cost and cost-effectiveness by disease area
For ease of reporting, both costing and cost-effectiveness findings are presented below. For a summary of all results by disease area, see Table 3.

RMNCH
Findings about the costs and cost-effectiveness of CHWs for a number of different RMNCH conditions and for different types of activities are generally positive. For maternal health, misoprostol prevention and treatment provided by village health workers were found to be more cost-effective than standard care (although standard care was not defined). Treatment was also found to be significantly more cost-effective than prevention in a simulation setting (looking at cost per life saved only) [39,40]. With regard to family planning, Chin-Quee et al. found the cost of adding an intervention to deliver injectable contraceptives to an existing community-based distribution package to be "low" [26]. In the area of neonatal health, women groups facilitated by lay health workers (LHWs) and trained traditional birth attendants (TBAs) were found to be cost-effective ways of reducing neonatal deaths compared to current practice [11,38]. The use of local women trained as peer supporters to individually counsel women about exclusive breast feeding was found to be implementable in sub-Saharan Africa at a "sustainable cost" [28]. For child health, CHWs were found to be costeffective for reducing under-five mortality and resulted in cost savings compared to the facility-based alternative for under-five child growth monitoring, counselling, curative care treatment and free-of-charge medicines as well as home visits as needed [16,17,19]. A study from India found implementation of integrated management of neonatal and child illnesses by CHWs imposed additional costs to the health system, but could not draw a conclusion about cost-effectiveness [36], while a study from Bangladesh on community management of severe acute malnutrition found the practice to be more costeffective than facility-based inpatient treatment [37]. Two reviews of literature on the use of lay and community health workers in vaccination programmes by Corluka et al. [43] and Pegurri et al. [44] found these workers to be more cost-effective options than the comparator which did not include LHWs, including in an outreach setting.

TB
Studies from Brazil [21], Ethiopia [15], Malawi [18], South Africa [14,22] and Uganda [20] found that using CHWs during the non-hospitalized phase of TB treatment is a cost-effective alternative to facility-based treatment. CHWs were found to reduce the cost per patient successfully treated and cured anywhere from 40% to 74% compared with facility-based provision. Okello et al. point out the importance of proper training and supervision in achieving success [20].

Malaria
Results are limited but generally positive from studies in favour of the cost-effective use of CHWs for malaria programmes compared with regular care. Studies found the delivery of intermittent preventive treatment (IPT) of malaria for children by village health workers was less costly then delivery by nurses in outpatient departments or immunization outreach [29,33]; community-based delivery of sulfadoxine-pyrimethamine (SP) during pregnancy increased access, improved adherence to IPT and was cost-effective according to World Bank criteria [31]; and home management of uncomplicated malaria by CHWs was 36% more cost-effective than the standard care in health facilities [13]. The use of pre-referral artesunate for the treatment of childhood malaria by CHWs was found to be a cost-effective (according to WHO guidelines comparing cost per DALY averted with gross domestic product (GDP) per capita), life-saving intervention with potential application in rural African settings where CHW programmes are already in place (comparing cost per DALY averted with GDP per capita, according to WHO guidelines) [23]. A study from Zambia looking at active and passive case detection by CHWs including testing and treatment concluded that the programme may be cost-effective when community participation in regular testing reached certain levels [24]. Additionally, based on results from two villages, Onwujekwe et al. concluded that starting up a CHW programme for malaria control nationwide in Nigeria is potentially "cost-effective", although no comparator or benchmark was given [41].

Other health priority areas
For wider primary care, studies have found that CHWs increased the coverage and equity of service delivery at low cost compared with alternatives, that using CHWs can be less costly than other community-based programmes and that comprehensive CHW subsystems can be deployed across sub-Saharan Africa at a modest cost compared with the project costs of a primary healthcare system [9,32,34]. A study from India further found that adding an additional primary care community-based health worker to the lowest level of the health facility was cost-effective, though results were only be measured in terms of cost of increasing ANC coverage [35].
Findings related to the cost-effectiveness of CHWs for other disease areas were limited but generally favourable to the use of CHWs to control hypertension (although in one study best results were achieved when combined with general practitioner training as well) [30,42] and for interpersonal therapy and case management of patients with mental disorders [12]. Additionally, an mhealth intervention implemented by CHWs on a variety of healthcare topics including treatment adherence monitoring, appointment reminders and emergency care resulted in monetary as well as time savings [25].
Overall, the results of this analysis are in line with findings from the review/discussion papers included in this study. Perry et al. found that where the costeffectiveness of CHW-provided interventions is compared with that of facility-based interventions, the CHW-provided interventions are generally found to be more cost-effective [45], and Walker et al. found CHWs working in primary healthcare, vaccination and TB control programmes increase the coverage and equity of service delivery at low cost compared with alternative modes of service organization [9].

Discussion
This review has found promising evidence in favour of the cost-effectiveness of CHWs as compared with standard practice or alternative delivery models or when comparing cost-effectiveness findings with a benchmark such as GDP per capita, although results should be interpreted with the understanding of both minor and major methodological challenges.

Methodological issues and limitations
This analysis of 36 articles and reviews from the period January 2003 to July 2015 has revealed the variety of methodological approaches used to assess the costs and cost-effectiveness of CHWs, which limits both comparability and generalizability. Additionally, many articles did not provide sufficient details about study design or methodological assumptions, such as time horizon and study perspective, data quality and sources, limiting their usefulness. However, these were often earlier articles and may reflect the newness of economic evaluation methods. Many studies also failed to recognize the limitations of their data or question the quality. A mixed methods approach to costing and cost-effectiveness studies could enhance insight on the functioning and communityperceived value of CHWs and therefore add much-needed depth to a costing or cost-effectiveness study.
The issue of perspective has emerged as an important methodological challenge in this review: approximately half the studies in this review took a provider or health service perspective. Because of the nature of CHWs, taken from and embedded in the community, as well as health economics methodologies developed to date, current ways of assessing costs and benefits of these programmes (including use of the provider or government perspective) fail to capture many of the important societal costs and benefits associated with CHWs, such as social capital and trust as identified by Walker et al. [9] and improved relationships between patients and care providers. These aspects may have fallen outside the purview of economic evaluations to date because they are not monetizable, but leaving them out means we are failing to capture the true costs and benefits of CHWs in costing studies and economic evaluations. On the issue of perspective, in some cases, CHWs might reduce patient costs (for example, for TB, where patients receiving treatment in the community no longer have to travel to health facilities), making it important to take a wider perspective. And while our review did not touch on financial versus economic costs (how much the project or programme actually pays compared with the overall cost of the project or programme), this is a closely related issue. Additionally, as CHWs often operate as part of larger healthcare teams, it would be desirable to assess their cost-effectiveness as part of the broader health system in which they operate rather than as stand-alone programmes.
Finally, given the large number of CHW programmes, many operating already for decades, this review also reveals that the cost and cost-effectiveness of many CHW programmes have not been extensively and systematically assessed. As CHWs grow in popularity and are incorporated in human resources for health policies and plans in different countries, the need for well-designed and conducted costing and cost-effectiveness studies becomes particularly important.

Cost and cost-effectiveness issues
This review has found evidence supporting the costeffective use of CHWs, particularly in the area of TB; there are also studies supporting the cost-effective use of CHWs in the areas of RMNCH, malaria and other disease areas, although their methodology and quality of evidence are less strong. However, even where there is evidence suggesting a better cost-effectiveness of CHWs compared to other service delivery models, results should be interpreted with caution. The reviewed studies used very different methodologies; they compared CHWs to different cadres of health workers, and sometimes, there was no comparator. Furthermore, the studies in this review included and excluded different costs: for example, often they did not include the important and sizeable training and supervision or recruitment and retention costs related to CHWs. Additionally, volunteer time was valued differently in different studies and sometimes excluded altogether. Effectiveness of CHWs was also measured differently in different studies.
On the issue of comparability and generalizability, one fundamental challenge with comparing or generalizing CHW costing and cost-effectiveness findings is the varying nature of CHWs themselves. Although often lumped together, there is a wide typology of CHW models worldwide, with training and competencies varying enormously. Studies should include more details about the type of CHW being assessed and their context, and these differences should be taken into consideration when attempting to compare results [46].
The majority of articles reviewed documented CHW involvement in short-term or limited duration TB, malaria and RMNCH programmes, reflecting the use of CHWs in some countries for specific health areas or conditions. However, evidence is more limited about the costs or cost-effectiveness of CHWs who take on responsibilities across a wider range of disease areas or conditions and on the long-term cost-effectiveness and systemic implications of these programmes. A study by Alam et al. on maternal health from Bangladesh found retention of CHWs to be a problem, and the cost associated with dropout was significant, leading the programme to be less sustainable [27]. Cost-effectiveness analyses are often presented as snapshots of a certain short period of time, while longer term issues of retention and sustainability should be considered as well.
Further mixed method research is needed to better understand why CHWs are sometimes cost-effective and sometimes not and if there are fundamental aspects of different health areas that lend themselves to a costeffective use of CHWs. For example, it could by hypothesized that CHWs are cost-effective in the area of TB because the activities performed are limited and easy to standardize, whereas CHW activities in the area of RMNCH may be more varied. It would be interesting to examine CHW cost-effectiveness on the different components of RMNCH separately (for example, antenatal care and deliveries separately) and analyse cost and costeffectiveness differences between these activities. Additionally, research is needed to understand the impact of the task sharing on efficiency, costs and cost-effectiveness of both the programme from which the tasks were split, the CHW and the system as a whole [47].
Besides the limitations of the reviewed studies, limitations of this review itself should be taken into consideration. Publication bias is a potential issue; some relevant studies may have been missed if they were not identified by the larger search from which these results were taken or the supplementary PubMed search, and we did not consider grey literature. We have also not specifically evaluated the quality of the reviewed studies, though the review points out methodological shortcomings of the reviewed studies as a whole.

Conclusions and policy implications
This literature review suggests that using CHWs in health programmes can be a cost-effective intervention in some settings, particularly for TB, with less strong evidence but promising indications of cost-effectiveness in RMNCH and malaria. These findings may relate to the fact that some areas have been evaluated less (and less rigorously) than others, rather than reflecting an actual difference in cost-effectiveness in the various service delivery areas or interventions.
Notwithstanding the caveats mentioned above about the heterogeneity of the studies and methodological weaknesses, this review shows that CHWs programmes have potential to represent good value for money for governments and donors for delivery of essential health services in LMIC. In developing or scaling up CHW programmes, however, more attention needs to be given to understanding costs and cost-effectiveness from both a government and societal perspective and to integrating community health workers in national healthcare systems in terms of employment, supervision, support and career development [48,49].