Systematic review of the costs and effectiveness of interventions to increase infant vaccination coverage in low- and middle-income countries

A large body of evidence has demonstrated the effectiveness and cost-effectiveness of infant immunization for reducing the burden of vaccine-preventable diseases [1, 2]. Routine immunization programs, supplemented by periodic campaigns, cover the majority of the target population in most countries, yet in many settings coverage remains below program goals [3]. Expanding coverage is a major objective of immunization programs, both to increase the magnitude of health benefits from vaccination and to reduce disparities in outcomes among underserved populations [4]. The remaining gaps in target coverage are especially prominent in low- and middle-income countries (LMICs), and so these settings have been a focus for ongoing efforts to improve immunization efforts with both vertical and horizontal programs.

Several recent studies have been undertaken to describe the costs of providing national immunization services [5‐12]. This research provides precise estimates of the cost of providing services at current coverage levels in a range of countries, and describes variation in the costs and operating practices of individual immunization sites. However, these studies do not provide direct evidence on the costs or cost-effectiveness of strategies and interventions used to scale up immunization coverage. Without this information, countries and the international stakeholders have little quantitative evidence on the best ways to use scarce resources to achieve immunization objectives. One possible source of information is studies that report the costs and effects of specific coverage-enhancing interventions.

We conducted a systematic review on the costs and effects of interventions to improve immunization coverage in LMICs, including research reported in the grey literature. This review updates past initiatives to survey the evidence on interventions to improve immunization coverage in low-income, high-burden settings [1, 2]. Several recent and historical studies have reviewed evidence related to immunization coverage improvements, but the majority of these have limited their scope to reporting effects on coverage or related programmatic outcomes, and excluded costs [13‐19]. Three reviews have considered both costs and effectiveness. In 2004, Pegurri, et al. summarized evidence in the peer-reviewed literature [1] and Batt, et al. reviewed the relevant grey literature [2]. Both of these reviews identified a limited number of LMIC studies meeting inclusion criteria, and concluded that heterogeneity in methods adopted by these studies prevented quantitative synthesis of results. Both reviews found that few studies reported costs, with 10 out of 60 and 15 out of 34 identified studies including costs, respectively. A recent review by Ozawa, et al. including articles through 2016, adopted a broader scope, including peer-reviewed studies from both low- and high-income country settings, and covering all age groups, but excluding grey literature [20]. While this study was able to undertake some quantitative synthesis of results, the authors acknowledged difficulties due to the heterogeneity of methods and reporting adopted by included studies, and the majority of estimates in their final sample came from high-income settings.

Our review returns to the approach of the earlier reviews with a focus on LMIC settings. In our review, we extracted data for studies conducted after the period covered by the Pegurri, et al. and Batt, et al. reviews (prior to 2003), as earlier studies were already included in these prior reviews, and following the reasoning that older research would be less relevant to contemporary planning and budgeting decisions. These earlier reviews stressed the importance of the grey literature in documenting findings in this field, and consequently we included both the peer-reviewed and grey literature in our review [1, 2]. The objective of this review was to describe the incremental cost and effectiveness of interventions to increase coverage of infant immunization in LMICs, as defined by World Bank income group. The primary outcomes of interest were the incremental costs and incremental changes in target population coverage associated with a coverage-improvement intervention, as compared to routine program performance.

This systematic review was registered with PROSPERO, an international prospective register of systematic reviews (record number 69586). We included interventions directed solely at increasing infant immunization coverage, as well as interventions designed to improve multiple aspects of immunization performance including coverage. The target age group for infants was defined as age 1 year and below; therefore, measles, mumps, rubella, and varicella vaccination interventions were included. We only included studies that reported empirical data, and excluded modeled analyses to minimize the introduction of additional bias into any summary results. We excluded interventions designed to improve health service delivery generally (e.g., improvements in access to primary health care). We also excluded studies that reported changes in immunization coverage but did not describe specific interventions used to impact infant immunization coverage. Studies with no data on intervention costs were excluded, as were those targeting adult immunization and animal studies. We included studies published between January 2003 (the end date of the period covered by older reviews [1, 2]) and May 2019.

A total of 2325 records were identified from 15 databases (Fig. 1), with an additional 4 records obtained from reference lists of other articles. After removing 114 duplicates, 2215 titles were screened for eligibility, of which 1629 were removed. Of the 586 records remaining, 545 were removed after abstract review, leaving 41 articles which received full-text review. Of these, 27 were excluded for not meeting study inclusion criteria. We examined the LMIC studies included in the Ozawa review and added two relevant articles that we had previously excluded [40, 41]. Fourteen studies were included in the final analysis [38‐51].

Table 1 describes general characteristics of the 14 studies. One study describes interventions from two countries (Mexico and Nicaragua); these were considered separate observations when extracting cost and effectiveness data [44]. More than half of the observations (n = 10) were conducted in Asia, with the remaining in Africa (Ethiopia, Guinea-Bissau, and Madagascar) and Central America (Mexico and Nicaragua). Observations were nearly evenly split with regard to setting: 6 were conducted in rural areas; 3 in urban; 3 in mixed settings and 3 were not specified. About half of the interventions (n = 7) were delivered in a mobile outreach format; 4 were delivered in a fixed (i.e., health facility) format, and 4 interventions utilized both a mobile and fixed format. Additionally, 10 interventions were administered within routine vaccination systems, while 5 were administered using a campaign format. One study utilized both routine and campaign formats. Intervention duration ranged from 2 weeks to 5 years.

More than half of the studies (n = 9) reported randomized controlled trials, with the remaining studies reporting non-randomized pre-post evaluations (1 with control group, 3 without control groups) or cross-sectional designs. The majority of interventions were aimed at approaches for vaccine delivery (n = 5) or approaches to encourage additional vaccine uptake (i.e., demand generation; n = 5). Additional studies were aimed at health systems strengthening or the introduction of novel vaccine technologies (e.g., syringes). The interventions targeted coverage improvements for a range of vaccines, with DPT3 (diphtheria-pertussis-tetanus vaccine third dose) and measles vaccination most commonly addressed (Fig. 2).

We reviewed the recent literature describing the incremental cost and impact of efforts to improve immunization program coverage in LMIC, and identified 14 studies that containing sufficient cost and coverage data to be included in this study. The interventions reviewed in these studies covered a wide range of geographic settings, vaccines, intervention types, delivery mechanisms and scales. About half of the studies were randomized controlled trials, and the average study score on the CHEC list was 14 out of 17. The majority of included studies reported increases in vaccination coverage following the examined interventions, similar to the results of the Pegurri, et al. 2004 and Batt, et al. 2004 reviews [1, 2], which found coverage improvements of 27 percentage points and 20 percentage points on average, respectively.

Although all reviewed studies provided costs of the intervention itself, only 2 studies also considered changes in the costs of providing routine immunization services, despite the fact that such changes are a likely consequence of efforts to increase coverage. Eleven studies (79%) provided adequate data to calculate ICERs. From these, 14 ICERs of different interventions were calculated, ranging from $0.66 [43] to $161.95 [51] per child vaccinated in 2017 USD. In several cases, calculations of ICERs relied on substantial assumptions. Given the considerable differences in settings and perspectives, as well as different methods and cost categories included, an observed difference between ICERs may well reflect an artifact of the different study designs rather than a true feature of the interventions under study. Therefore, estimates should be compared with caution, and with full knowledge of the methodological and contextual differences between two studies. Funding devoted to coverage improvements may not be utilized efficiently in the absence of better evidence on optimal approaches.

Several factors prevented ICERs from being calculated in all studies. The interventions described in some studies [38, 44] contained a package of health services (immunization and other health activities). However, intervention costs were given only at an aggregate level and therefore immunization ICERs could not be calculated. Additionally, in one study the size of the population exposed to the intervention was not stated [38]. Similarly, in one study the costs were calculated at the level of health workers, so cost per exposed child could not be determined [49]. Due to the small total number and methodological heterogeneity of the extracted studies, we were unable to conduct a quantitative synthesis with cost estimates stratified by intervention type and country category.

Despite the 15-year gap since previous systematic reviews focusing on LMICs, there is still a scarcity of evidence on the cost-effectiveness of options for improving immunization coverage [1, 2, 20]. Studies rarely report estimates of the incremental change in both cost and coverage. Several of the studies that do provide such information failed to report key features of the study methods, which limits the utility of their results. In particular, future studies should include detailed discussions of the intervention type (i.e., aimed at demand generation, delivery mechanisms, etc.), target population, baseline and endline coverage, specific intervention costs, changes in costs of routine service provision as well as information on effect modifiers that could affect incremental costs and coverage. Such information should be reported at a level of granularity that other programs would be able to interpret and modify those costs to fit their individual setting. These details are necessary to navigate the heterogeneity of the studies and to directly compare and synthesize the results to produce generalizable conclusions. The biggest challenge is that efforts to scale up immunization coverage often lack provision for a costing study, which represents a huge missed opportunity to understand the best use of scarce resources for improving coverage. We challenge the immunization community at country, regional, and global levels to incorporate costing studies into scaling up efforts.

This research underscores the need for broad adoption of standardized reporting methods in immunization costing studies [52, 53]. This will be essential to increase our knowledge of improving immunization coverage as well as child immunization intervention costs.