Summary of main results

Ten cluster RCTs and four individually randomised controlled trials met our inclusion criteria. These were conducted in Georgia, Ghana, Honduras, India, Mali, Mexico, Nepal, Nicaragua, Pakistan, and Zimbabwe. The interventions evaluated in the studies included community‐based health education, facility‐based health education, home visits, household monetary incentives, and integration of immunisation services to intermittent preventive treatment of malaria. These were implemented either as single interventions or as multi‐faceted interventions.

We found moderate‐certainty evidence that giving information and discussing vaccination with parents and other community members at village meetings or at home probably improve immunisation coverage. We also found low‐certainty evidence that giving information to parents about the importance of vaccinations during visits to health clinics combined with specially designed 'reminder‐type' immunisation cards may improve immunisation coverage. There was low‐certainty evidence that regular immunisation outreach, home visits, and integration of immunisation with other primary healthcare services (such as intermittent preventive treatment of malaria) may improve immunisation coverage. However, there was currently low‐certainty evidence that household monetary incentives (in the form of conditional or unconditional cash transfers) may have little or no effect on immunisation coverage.

Overall completeness and applicability of evidence

In the context of the GVAP, there is an urgent need for effective interventions that would ensure equitable uptake of existing vaccines by people in all communities around the world (WHO 2012a). However, immunisation coverage remains uneven between and within the world's regions and countries. For example, in 2014, DTP3 coverage was 96% in Europe and the Western Pacific and only 77% in sub‐Saharan Africa (WHO 2015). The GVAP coverage target was to achieve DTP3 coverage of 90% in all countries by 2015; but only 129 (66%) countries have achieved this coverage target. The 10 countries with the largest numbers of unimmunised children are all low‐income or lower‐ to middle‐income countries (SAGE 2015).

Barriers to improving immunisation coverage could be broadly categorised into factors that affect the demand for vaccines, barriers to the supply of vaccines, or both (Lewin 2011). Around 2014, the concept of vaccine hesitancy emerged as a factor hindering the demand of vaccines (Larson 2014). The Strategic Advisory Group of Experts on immunisation identified factors that influence vaccine hesitancy. These were grouped into three major areas, namely contextual influences, vaccine and vaccination specific issues, and individual and social group influences. The influences of these factors are said to be complex and context‐specific, varying in time, place, and vaccine (Larson 2014). It is unclear if the interventions tested in the included studies were derived from identified barriers in the settings, though specific concepts were tested. In general, though, interventions to improve coverage should focus on identified barriers within settings.

The included studies evaluated interventions that varied enormously in content and in the intensity of delivery, raising questions regarding the likely impact of interventions in different settings and regarding how best to implement the interventions. For instance, how effective will a three‐minute health education intervention (Usman 2009; Usman 2011; low‐certainty evidence) be in a typical clinical setting in improving completion of the immunisation schedule? Will the same effect be obtained for more than one vaccine? How feasible is evidence‐based discussion (Andersson 2009; moderate‐certainty evidence) in a community with low literacy? How feasible is a monetary incentive intervention (Morris 2004; low‐certainty evidence) in a resource‐poor setting without donor support? The limited number of studies makes it difficult to explore these issues and restricts the wider applicability of the evidence.

Home visits to promote childhood vaccination uptake or to deliver vaccination are common in many settings. The applicability of the home visit intervention as implemented in Brugha 1996 may be affected by several factors. First, the use of first‐degree university students as lay/community health workers to deliver this type of intervention may not be feasible in many resource‐poor settings. Two reviews reported moderate‐certainty evidence on the effectiveness of lay health workers in promoting the uptake of childhood immunisation services (Glenton 2011; Lewin 2010). In these reviews, the level of education of the lay health workers varied from primary school graduates to high school graduates, with some studies not reporting this information. Furthermore, the settings of the studies were middle‐ and high‐income countries. Second, referring caregivers of children who need immunisation to a health facility requires that there be a facility within reasonable distance of the community. This type of intervention may not be useful in settings in which households do not have easy access to health facilities. Finally, administering injectable vaccines at home has implications for vaccine quality and injection safety; given the need to maintain the cold chain from manufacture to administration of vaccines as well as the need to dispose of injection material safely (preferably by incineration). This approach may not be cost effective or sustainable in a resource‐constrained economy. The high diversity of the interventions and the contextual differences, therefore, make it difficult to draw conclusion on their effectiveness in improving vaccination coverage.

There is paucity of data on the sustainability of the interventions presented in this review, as none of the included studies reported long‐term follow‐up data. All had two data points that were at baseline and post‐intervention, making it impossible to ascertain the long‐term effects of the interventions. However, two studies aimed to build the capacity of the providers (Djibuti 2009; Morris 2004), and to upgrade the physical structure (Morris 2004). These strategies can contribute to sustainability if other supporting resources are available. A sustainability framework for projects aimed at strengthening immunisation systems in LMICs should include maintenance or continuance of health benefits from projects, institutionalisation of projects within the system, and capacity development (Gruen 2008; Shediac‐Rizkallah 1998). It has been observed that for a programme to be sustained, early and active planning is required (Shediac‐Rizkallah 1998). Sustainability of quality improvement interventions has been particularly challenging in LMICs, especially when a programme is supported by external funds (Gruen 2008). Withdrawal of external funds may not only impact negatively on the gains of the programme but may jeopardise support for future programmes (Gruen 2008). This is particularly so when the intervention is cost intensive.

Information on the resource implications of interventions may be helpful in determining their long‐term sustainability and cost effectiveness. Only one study provided data on the cost of the intervention (Andersson 2009). This study reported that community‐based health education in Pakistan costs USD9.00 per child. The cost of interventions would depend on the context of the intervention, as the cost to vaccinate a child fully has been reported to vary between USD1 and USD40 in LMICs (Shea 2009). Therefore, the cost of interventions should be reviewed within the context and settings of the studies. As part of their systematic review on the effects of lay or community health workers in primary health care (Glenton 2011; Lewin 2010), Glenton and colleagues reviewed the costs and cost‐effectiveness of vaccination programme interventions involving lay health workers (Corluka 2009). The authors found that studies did not adequately address affordability and sustainability and were also highly heterogeneous in terms of settings and outcomes, limiting their comparability. In addition, they found insufficient data to allow any conclusions to be drawn regarding the cost‐effectiveness of lay health worker interventions to promote vaccination uptake. Studies focused largely on health outcomes and did illustrate to some extent how the institutional characteristics of communities, such as governance and sources of financial support, influence sustainability (Corluka 2009). Considering that the interventions assessed in this review were set up as parallel programmes, rather than being integrated into routine services, it is unclear how effective they will be if integrated with other services within the system, with the typical levels of human and other resources available. This calls for cost‐effectiveness evaluations of these interventions, particularly as integrated rather than stand‐alone programmes (Dicko 2011; Okwo‐Bele 2012). Such evaluations also have limitations as it can be difficult to translate these findings from one setting to another. Therefore, there is need for study authors to provide the details of the required resources to implement the intervention.

Many immunisation programmes in LMICs are delivered as mass immunisation on set 'immunisation days', following mass immunisation campaigns (Balraj 1986; Bandyopadhyay 1996; Berry 1991; Cutts 1990; Gomber 1996; Kumar 1990; Lin 1971; Linkins 1995; Shaikh 2003). None of the reports of this commonly used strategy met the criteria for inclusion in our review or for the Saeterdal 2014 Cochrane review on interventions aimed at communities to inform and /or educate on early childhood vaccination. Shea 2009 has noted that it may be difficult to randomise mass media interventions. However, ITS designs could be used to assess the effects of these mass immunisation campaigns on immunisation coverage (Nglazi 2014).

Quality of the evidence

This review included 14 studies, three had unclear risk of bias (Andersson 2009; Owais 2011; Pandey 2007), while all the others had high risk of bias. The main study limitations were non‐concealment of allocation, no blinding, lack of protection against contamination, and extraneous sources of bias. The cluster RCTs were adjusted for cluster effects. Overall, the certainty of the evidence for most interventions was low. This implies that the currently available research provides some indication of the likely effect of the interventions. However, the likelihood is high that the true effect of the interventions will be substantially different.

Potential biases in the review process

We minimised bias in the process of conducting and reporting the current review by adhering to standard Cochrane guidelines (Higgins 2011). However, access to studies from LMICs is limited to those studies published in indexed journals. There may be a need to identify non‐indexed local journals and the grey literature in low‐income countries, and to conduct handsearching of these sources. In addition, as noted by Machingaidze and colleagues, due to the broad nature of childhood immunisation (encompassing many different components) identifying a search strategy that includes all aspects of childhood immunisations is challenging (Machingaidze 2014). However, in one 2014 Cochrane review on community‐aimed interventions to inform and educate about childhood vaccination, the authors did not identify additional studies to those included in this review (Saeterdal 2014).

Agreements and disagreements with other studies or reviews

Several previous systematic reviews assessed the effectiveness of interventions for improving childhood immunisation coverage (Batt 2004; Bordley 2000; Giles 2014; Glenton 2011; Jacobson Vann 2005; Johri 2015b; Kaufman 2013; Kendrick 2000; Pegurri 2005; Ryman 2008), although very few of them included studies from LMICs (Batt 2004; Glenton 2011; Pegurri 2005; Ryman 2008), and many were already out‐of‐date as the dates of the most recent searches for the reviews were pre‐2005 (Batt 2004; Bordley 2000; Jacobson Vann 2005; Kendrick 2000; Pegurri 2005).

Measures of effect for participant reminders in this review tend to agree with a now out‐of‐date systematic review of interventions aimed at reminding people of their immunisation schedules (Jacobson Vann 2005). Home visits, participant reminders through a redesigned immunisation card, and health education improved the uptake of immunisation in this review. Similarly, telephone calls, sending of letters and postcards, and speaking to clients in person improved the coverage of childhood vaccines in the participant‐reminder review (Jacobson Vann 2005).

We found low‐certainty evidence that monetary incentives (in the form of vouchers, conditional, and unconditional cash transfers) may have little or no effect on uptake of vaccines. This differs from the findings of several related systematic reviews: one systematic review on the effect of conditional cash transfers on health outcomes and the use of health services reported an improvement in the use of health services but, similar to this review, reported mixed results for uptake of immunisation in children (Lagarde 2009a). Two older (and now out of date) reviews also reported on the effects of this intervention (Giuffrida 1997; Kane 2004). One more recent review on the topic included 16 studies from high‐income countries on smoking cessation (10 studies), attendance for vaccination or screening (five studies), and physical activity (one study) (Giles 2014). It reported an increase in vaccination and screening attendance with monetary incentives. However, subgroup analysis showed that cash plus other motivational components was more effective than cash or vouchers alone. The differences between our review findings and those of this review may reflect differences across settings (high‐income compared to low‐ and middle‐income countries) or limitations of the studies included in our review. In addition, the Morris 2004 study findings were of low certainty because of high risk of bias.

Ryman and colleagues conducted a comprehensive search in 2005 to identify peer‐reviewed and grey literature on strategies for improving childhood immunisation coverage in LMICs (Ryman 2008). They identified 25 studies that included an appropriate control group, and grouped the papers into four strategic approaches: bringing immunisation closer to communities (11 studies), using information dissemination to increase demand for vaccination (three studies), changing practices in fixed sites (four studies), and using innovative management practices (seven studies). The studies included RCTs, nRCTs, CBAs, and observational studies, and reported improvements in immunisation coverage of varying degrees. Unlike Ryman and colleagues, we excluded observational studies. We included CBAs if they had more than two units in both the intervention and control groups, in accordance with EPOC guidance (EPOC 2015a). Though the Ryman review identified studies that reported improvements in immunisation coverage, they noted that the indicators of success varied widely making it impossible for the data to be merged in a meta‐analysis (Ryman 2008). We also found that studies reported immunisation outcomes in a variety of ways, for example, proportion of children aged 12 to 23 months who had received measles, proportion of children aged 12 to 23 month who had received full course of DTP (Andersson 2009); probability of receiving at least one immunisation (excluding OPV), the presence of the BCG scar, the number of immunisations received, the probability of being fully immunised (Banerjee 2010); immunisation full coverage of children aged 12 to 23 months with three doses of DTP, BCG, and measles vaccines (Barham 2005); DTP3 coverage at the end of day 90 post‐enrolment (Usman 2011), etc. However, our foreknowledge of childhood immunisation programmes guided our decisions regarding which outcomes were synonymous (and thus can be combined in a meta‐analysis) and which are not.

In a related systematic review, Glenton and colleagues assessed the effects of lay or community health worker interventions on childhood immunisation coverage (Glenton 2011). They conducted the last search in 2009, and identified 12 studies; including 10 RCTs. Five of the studies were carried out in LMICs. In 10 studies, community health workers promoted childhood immunisation and in the remaining two studies, community health workers vaccinated children themselves. Most of the studies showed that the use of lay or community health workers to promote immunisation uptake probably increased the number of children who were fully immunised. Our findings on the effect of community‐based health education and home visits were consistent with these findings.

Johri and colleagues reported a systematic review of "strategies to increase demand for vaccination are effective in increasing child vaccine coverage in low‐ and middle‐income countries". The authors concluded that, "demand‐side interventions are effective in improving the uptake of childhood vaccines delivered through routine immunization services in low‐ and middle‐income countries" (Johri 2015b).

Finally, our review is related to two other Cochrane reviews (Kaufman 2013; Saeterdal 2014); conducted under the auspices of the 'Communicate to Vaccinate' project (Lewin 2011). Kaufman 2013 assessed the effects of face‐to‐face interventions for informing or educating parents about early childhood vaccination on immunisation uptake and parental knowledge and Saeterdal 2014 reviewed interventions aimed at communities to inform or educate (or both) about early childhood vaccination. The two reviews included studies from any setting while this review focused on low LMICs. We included three of the studies (Bolam 1998; Usman 2009; Usman 2011) included in the Kaufman 2013 review in our review and two studies (Andersson 2009; Pandey 2007) from our review were included in the Saeterdal 2014 review. While the findings of this review were similar to the findings of the Saeterdal 2014 review (i.e. that these interventions probably increase immunisation coverage), they differed from the findings of Kaufman 2013 that reported little or no improvement in immunisation coverage. This may be because Kaufman included studies from high‐, middle‐, and low‐income countries.