TL;DR

This post introduces Notify Health, an AIM-incubated charity that reduces vaccine-preventable child deaths in Nigeria via SMS and voice vaccination reminders.

Nigeria has the highest number of children missing vaccines globally by count. Over 50% of missed vaccines are caused in part by caregivers forgetting or misunderstanding the schedule. SMS and call vaccination reminders are a scalable solution that has strong evidence of effectiveness. A meta-analysis in LMICs shows sizable effects, and there are programs that prove this intervention can work at scale by serving millions of children already. Implementation remains neglected in many contexts, including Nigeria.

Founded one year ago, Notify Health piloted reminders with ~2,200 caregiver-child pairs from 44 clinics in Kogi state, Nigeria. The program works by digitising widespread health facility paper records to enrol caregivers. During the pilot, we sent 42,000+ reminders and saw timely vaccination increase by 12–24 percentage points in our analysis of coverage data for ~10,000 children. Despite limitations in the M&E design, these results build on prior evidence and offer encouraging early signs that the intervention is working.

Our current cost-effectiveness estimate, built on published GiveWell models and using conservative discounts, is ~5x GiveDirectly cash transfers or $11,000 per death averted. Modelled program tweaks (higher-gap states, wider enrollment, lower cost) lift this above 10x and up to 24x cash cost-effectiveness, showing levers we will pursue next.

Next, we’ll expand to a higher-burden state, find ways to enrol harder-to-reach children, and refine the programme to cut costs and raise effect sizes. Our ultimate aim: ensure millions of children across countries in sub-Saharan Africa get the vaccines that protect them from entirely preventable diseases.

You can support us by joining our team as a generalist, contributing to closing our 2025 funding gap or offering feedback. Read to the end to learn more.

The problem - Millions miss vaccination due to simple knowledge barriers

Globally, over 20 million children miss routine vaccinations each year. Progress has stalled since 2011, leading to more than 700,000 entirely preventable child deaths annually, over half of them in sub-Saharan Africa. Nigeria, where we operate, has the most “zero-dose children” in the world, numbering between 1.2 to 2.2 million. If you are an unvaccinated child born in Nigeria today, you have a roughly one in 37 chance of dying from causes that are vaccine-preventable^[1].

There are still major gaps in the immunization system, even though organisations like GAVI and the WHO have made strong progress on the supply side — from lowering procurement costs for LMICs to strengthening cold chains and improving service delivery through health facilities and health workers. The vaccine demand side, however, remains a persistent challenge. One organisation that has found an effective approach is the GiveWell top charity New Incentives, whose conditional cash transfer (CCT) program has enrolled over 5 million children in Nigeria. However, according to GiveWell’s CEA, the New Incentives program would not be cost-effective in many Nigerian states where vaccination rates remain relatively low — in these areas, a gap exists for a lower-cost program focused on driving demand for vaccination.

One persistent barrier to vaccination demand is simple "lack of knowledge or information". There are many Nigerian states where this type of barrier exists for ~50% of caregivers who miss a vaccination appointment.^[2] This includes caregivers who mistakenly believe their child is already fully immunized, are unaware that multiple visits are required, or simply forget to go. Nigeria’s routine immunization schedule requires at least six separate visits at different intervals during a child’s first fifteen months of life — a sequence that can be difficult for caregivers to remember. This schedule will become ever more complicated when considering the imminent introduction of the R21 malaria vaccine, which requires doses at 5, 6, 7, and 15 months (adding three visits on top of the existing schedule). In a system where health workers are incredibly stretched and records are paper-based, many caregivers and their children fall through the cracks once they leave the facility.

The solution - Make it easy with proven, low-cost, and scalable reminders

Intervention: Vaccination Reminders

A strong evidence base shows that SMS and voice reminders can improve both vaccination coverage and timeliness. Case studies like Suvita in India and MomConnect in South Africa show that it’s feasible to scale such programs to millions. Caregivers receive reminders a few days before their child’s vaccination, helping them plan. With 88% of Nigerian households owning a mobile phone, most caregivers are now reachable. Messages can also include information like what vaccines are due and where to go — for example, one of our messages reads:

Fatima should get their next free immunization tomorrow. 
Join other caregivers from your community and visit your nearest clinic on their immunization day. Adavi-Eba Primary Health Centre does immunization on Tuesdays.

A 2021 meta-analysis by Eze et al. found that SMS reminders significantly improved vaccination coverage. GiveWell revised this meta-analysis in a recent intervention report on SMS reminders, estimating that reminders might reduce the share of children who are unvaccinated by 15%. Though they think it could be a highly cost-effective program, GiveWell has some reservations, primarily about the variance in program design and effectiveness of SMS reminder programs across studies and contexts, making it hard to predict their impact in any specific setting.  

Despite their potential, reminder systems are still not integrated into routine immunization services in nearly any Nigerian state, even though SMS and voice calls cost just fractions of a cent, and key data like birth dates and phone numbers are already collected. This leaves a large, evidence-based, and highly tractable opportunity to boost vaccine uptake untapped: simply use technology already in caregivers’ hands.

A major barrier to this happening already has been the lack of a cheap, scalable way to digitize the key data shortly after birth — an obstacle our implementation is designed to overcome.

Our implementation

Our program design is inspired by Suvita, an AIM-incubated charity in India that is supported by GiveWell and Founders Pledge. Children-caregiver pairs are enrolled into the program by taking photos of existing paper registers and then digitizing the key information.

First, we photograph immunization registers containing the child’s name, date of birth, and caregiver phone number — a natural touchpoint reached early in life when children receive their first vaccine^[3]. Photos are taken by our staff or facility-based health workers using smartphones, on a regular schedule, to capture children before their 6-week appointment. During visits, we train health workers on record keeping and how to ask for phone numbers to improve register quality over time.

Second, during the Data Entry Stage, the images are reviewed for clarity and completeness, then manually transcribed into structured digital spreadsheets. Errors are flagged to inform health worker feedback.

Finally, digitized data feeds into our mass messaging system. Automated SMS and voice call reminders are sent in line with Nigeria’s immunization schedule — four days and one day before each due vaccine. Caregivers receive messages via SMS (in English) and voice (in their local language), the latter recorded by native speakers. While most studies focus on SMS-only reminders, our needs assessments show many caregivers are illiterate and prefer voice calls — which have minimal marginal cost.

Progress - Notify Health piloted reminders in Nigeria with ~2,200 caregivers

Notify Health launched 13 months ago after going through the AIM/Charity Entrepreneurship Incubation Program. Since then, we have run a needs assessment in two Nigerian states, completed a pilot in one state and signed an MoU allowing us to scale up to 130,000 caregivers per year. We have also hired our first two Nigerian staff, a Program Manager and a Monitoring, Evaluation & Learning Officer, to further strengthen our capability in these areas and to help us reach our 2025 goals.

We piloted our vaccination reminders with ~2,200 caregiver-child pairs from 44 health facilities^[4] in Kogi State over 3 months from mid-October 2024 to mid-January 2025. Our M&E concept followed the key uncertainties in the theory of change:

• Operational metrics - can we enrol caregivers in large numbers cost-effectively and maintain a mass messaging system that is capable of sending thousands of reminders every month?
• Caregiver feedback - are people receiving, reading and understanding the reminders?
• Changes in immunization rates - do we have initial signs that our intervention works to increase immunization rates?

Operational metrics - Feasibility & Reach

Given existing RCTs and meta-analyses showing SMS reminders generally work to increase vaccination outcomes, the part of our theory of change that we were most uncertain about was our ability to enrol children into the program cost-effectively – particularly in the absence of an already existing electronic database – and then automatically send them reminders.

This step depends heavily on the completeness of paper-based immunization registers, particularly whether each child has a date of birth and a caregiver phone number. To improve enrollment rates, we trained health workers on the importance of these fields and provided them feedback based on register quality metrics.

In summary:

• A total of 2,168 children were enrolled throughout the pilot period
• The rate of enrollment increased from ~100 per week in October to ~250 per week in December as paper registers got better
• Record quality improved significantly: the proportion of children with a phone number associated with their name was 48% in the first month vs 78% in the final month - getting close to a natural ‘cap’ based on phone ownership rates.

Additionally, our automated SMS system was able to cope with a high volume of messages

• Our system sent over 42,000 reminders (SMS + call and call re-tries)
• Roughly 9 out of 10 SMS messages were successfully sent, according to the feedback automatically generated from the telecom companies.  
• About half of all voice calls were picked up, and 35-40% were listened to long enough to convey the key appointment details.

Caregiver Feedback

To test the next step in our theory of change, we opted to conduct phone surveys. The goal was to ensure reminders are received on the mobile phone of the caregivers, that the right caregiver reads the message, and that they are reminded to go to vaccination.

• We attempted phone surveys with 692 caregivers across two phone survey rounds.
• 60% of these calls were picked up (with a maximum of 3 attempts)
• Of the phone numbers that picked up, 31% were a ‘wrong number’
• Of the correct caregivers who answered the survey:
  • 69% actively recalled having received a reminder, after an average time since the last reminder of 20 days
  • 87% of respondents who remembered the reminders labelled the reminders “very helpful”, and 8% found them ‘a little helpful’
  • 49% found the voice reminder additionally helpful on top of the SMS
  • None of the respondents reported any problems with the reminders.
  • We were worried that two SMS and two calls are too many reminders for each vaccine, but 95% reported that we sent just the right number, and 5% reported that we sent them too few.

There were clear limitations to our phone survey — and to phone surveys in general — which make the results difficult to interpret. Still, the findings highlight a critical concern: a meaningful share of respondents were not the intended caregivers, pointing to issues with phone number accuracy. In many cases, we have visibility into the source of the problem. Our data audits reveal frequent discrepancies between independent data entry staff on key fields like phone number and date of birth. However, errors can occur at multiple points — from caregiver reporting, to health worker recording, to data entry. As a result, the rate of ‘wrong numbers’ has emerged as a key monitoring metric for improving record quality, and we have clear ideas on how to improve it.

Changes in immunization rates

Measuring impact on vaccination coverage within a short pilot is challenging. Standard metrics require children to reach at least 6 or often 12 months of age to allow enough time to complete their vaccination schedule, including any catch-up doses. To get an early signal of whether our reminder system was driving on-time vaccinations, we focused on measuring timely completion of the first vaccines after the birth vaccines (Penta-1 and Penta-2) up to 4 weeks after they are due. Whilst multiple vaccines are given at each appointment, we focus on Penta-1 and Penta-2 as standard proxies for the 6- and 10-week vaccination appointments, following common practice in the literature and by WHO and UNICEF.

To assess children’s vaccination status, we digitized the immunization dates of around 10,000 children using the same immunization registers we use to enrol children (these registers also contain the dates vaccines are given). After the enrollment period of the pilot ended, we revisited 27 of the 44 clinics where we had enrolled >90% of the total children in the pilot and digitized the entirety of the previous 12 months' worth of data from the immunization registers.

For our analysis, we compare vaccination outcomes between children enrolled to receive reminders and those not enrolled. The non-enrolled group includes children from the 12 months before the pilot or from the pilot period who were not captured during enrollment (usually because they did not have an associated phone number).

Children were eligible for inclusion if, by the time of data collection, they had either reached the relevant age (10 weeks for Penta-1, 14 weeks for Penta-2) or had already received the respective vaccine. Among included children, we assess timely vaccination, defined as receiving Penta-1 by 10 weeks or Penta-2 by 14 weeks.

We also distinguish between children enrolled at any time up to 14 weeks of age and those enrolled on time, defined as being enrolled before their first 6-week vaccine was due.^[5] This allows us to assess whether earlier enrollment — and therefore fuller exposure to reminders — leads to stronger outcomes. It also helps mitigate potential selection effects, since children enrolled later may already have fallen behind schedule, making it less likely for them to receive vaccines within the recommended time frame.

Children enrolled in the program were significantly more likely to receive vaccines on time. For Penta-1, 79% of enrolled children were vaccinated by 10 weeks, compared to 67% of non-enrolled children — an absolute difference of 12 percentage points. This means for every 8 children enrolled, 1 additional child received Penta-1 on time (NNT = 8). For Penta-2, 75% of enrolled children were vaccinated by 14 weeks, versus 51% of non-enrolled children — a 24 percentage point difference (NNT = 4).

When including all enrolled children (up to 14 weeks of age), the effect size for timely Penta-2 drops. This likely reflects a mix of reduced exposure to reminders and selection effects — for instance, children already off-schedule at enrollment or those who had already received the vaccine before reminders began. This underscores the value of early enrollment for maximizing program impact.

Discussion and limitations

These are promising early observational results and must be interpreted as such. Some factors increase our confidence in a causal relationship, however. Firstly, these results are consistent with what we would expect based on causal evidence from randomized controlled trials. Additionally, the intervention has a plausible and direct mechanism of action: reminders address a clear gap in knowledge, as demonstrated by previous surveys. Finally, the results are in line with widespread anecdotal reports from both health workers and caregivers that these reminders help get people to the facility on time.

Timeliness vs coverage

Our results show meaningful improvements in timely vaccination, but they do not directly reflect overall coverage, which is typically assessed after 6 or 12 months of age. In the literature, reminders are shown to improve both timeliness and coverage, though the effect on timeliness is generally larger. We therefore expect our impact on eventual coverage to be smaller than the 12 and 24-percentage-point increases observed for on-time Penta-1 and Penta-2, as relatively more children in the comparison group will catch up.

The key uncertainty is whether our intervention merely shifts the timing for caregivers who would have come anyway, or also causes counterfactual vaccinations that wouldn’t have happened otherwise. We don’t have a clear picture of the relative importance of improving timeliness vs coverage. Our prior is that coverage matters more and that reminders ultimately help improve it. Additionally, timeliness and coverage are positively correlated, suggesting that large gains in timely vaccination are likely to translate into gains in overall coverage. With short follow-up, we can’t say for sure—but based on prior evidence, we strongly expect that our intervention will improve coverage, and the main question is by how much.

Limitations of our analysis

Our analysis has many limitations. In general, this study design lacks a counterfactual control group, making it difficult to attribute observed changes to the intervention rather than to external factors such as seasonality, broader policy shifts, or secular trends. Additionally, differences in the characteristics of participants or the implementation context before and after the intervention can introduce bias. In our case, we highlight some potential key issues:

• The immunization system may have improved independently of our reminder intervention, as it is continually evolving and subject to broader efforts. Notably, a government-led campaign called the “Big Catch-Up” took place in early 2025 in Kogi State. While one of the areas included in the Big Catch Up — Dekina — was also in our pilot, children from there represent less than 3% of those in our analysis, making it unlikely that this meaningfully influenced our results.
• The difference could be due to improved data recording rather than actual increases in vaccination coverage. Our intervention may have led to more complete immunization register entries (indeed, a core component of our program was enhancing register quality).
• Seasonality. It’s possible that our pilot period overlaps with a naturally higher turnout time for immunization. Unfortunately, we lack sufficient historical data to fully assess seasonal trends. The rainy season, which typically runs from April through October, may reduce access to immunization — either due to travel difficulties or health facility disruptions. None of the LGAs in our pilot were flood-prone, but it remains plausible that the tail end of the rainy season affected caregiver behavior, potentially biasing our results upward if vaccine-seeking increased as the rains subsided.

As a rough and imperfect attempt to account for the time-based confounders mentioned above, we compared vaccination rates between enrolled and unenrolled children born in the pilot period. Some children were not enrolled in the program mainly because they didn’t have a phone number on record. This comparison has clear limitations — children whose caregivers don’t have phones may be poorer or live in more remote areas with lower access to healthcare — but it still offers some indicative insights.

We can see that non-enrolled children born during our pilot period have significantly lower timely receipt of Penta-1 and Penta-2. This might offer an additional signal that we are seeing an effect of reminders rather than a general improvement in the vaccination system over the last 12 months.

Taken together, the pilot results provide promising early evidence that reminders are contributing to improved vaccination outcomes in this setting. We cannot yet definitively attribute these effects to our program, but the direction and magnitude of change are consistent with our expectations and align with a strong existing evidence base. The combination of improvements in observational data, supportive caregiver and health worker feedback, and operational success gives us confidence that our approach is on the right track. We recognise there are clear areas for improvement — both in implementation and measurement — which we will address as we move toward stronger, causal evidence.

Cost-Effectiveness – Not Yet 10x, but a clear path there

To calibrate how much to focus on growing the program’s reach in our pilot state vs. larger design changes or working in other locations, we updated our cost-effectiveness analysis (CEA). We now use a more robust model, combining two published GiveWell CEAs: the 2024 analysis of a hypothetical vaccination reminder program in Nigeria, as well as the CEA of GiveWell Top Charity New Incentives by Nigerian state. We further apply a significant cautionary adjustment to account for our enrollment method. Since caregivers enter our program through paper vaccination registers, meaning they have come in for at least one vaccine, we think they are more health-seeking than others. Therefore, the vaccination gap we can close with our reminders is likely smaller than in the general population for which high-quality coverage data exists. The adjustment results in a 58% discount for Nigeria on average.^[6] Considering the source of our model and this adjustment, we think the cost-effectiveness estimates are very conservative. The key model inputs for our pilot State Kogi are:

• Cost per reached child: $2.39: A near-term estimate including all charity costs. The main costs are in regularly obtaining register photos (staff + transport or health worker reimbursement), core staff and overhead (co-founders + non-program, tech, etc.), transcription of photos, and M&E (phone surveys, other data gathering). Message costs are negligible.
• Vaccination gap: 9%: Based on state-level DHS data from 2024. The actual gap is 61%, but our health-seeking population adjustment affects Kogi particularly strongly. The data suggests caregivers either get no vaccine or follow through with most vaccines.

• Reduction in vaccination gap: 16%: Based on GiveWell’s adjusted meta-analysis of Eze et al 2021. We exclude one study and adjust the internal validity discount, which slightly increases the expected effect size.^[7] Note that together with the vaccination gap of our recruited population in Kogi this suggests an estimated effect size of 1.5pp increase in vaccination coverage, a magnitude much lower than what we saw in the timeliness metrics of our pilot and coverage in many studies.

• Probability of death from vaccine-preventable diseases for unvaccinated children: 1.3%: Based on IHME GBD data. Note that this does not yet include malaria mortality addressable by the R21 vaccine currently being introduced in Nigeria. We include an upward adjustment of 42% on the intervention effect to account for this, in line with the addressable child mortality from malaria in Nigeria and expected vaccine efficacy.

Our current program design, in our pilot state, could be ~5x as cost-effective as GiveDirectly unconditional cash transfers, avert a child death for around $11,000, and ensure an additional child is fully vaccinated for $155. Considering the level of evidence behind this intervention, the robustness and conservative nature of the model, and the remaining uncertainty in our cost figures, we think these numbers are a solid foundation to build on.

To inform our strategy and program design decisions, we ran a sensitivity analyses on the key levers to improve our cost-effectiveness.

This analysis sheds light on some very high-leverage program design choices to increase cost-effectiveness:

• State selection: Of non–New Incentives states, four are between 12x and 20x cash, with another five at ~6.5x (which can reach 10x in combination with other changes). Taken together, around 1.4 million children are born annually in these 9 states.
• Enroll broader population: Enrolling from the general population rather than immunization registers alone, we can achieve up to 24x cash cost-effectiveness, or avert a death for as little as $1,800 — even within our pilot state.
• Refine program for lower cost and larger effect size: Reasonable cost reductions to between $1.5–$2 per child enrolled, or designing a more impactful program than the average program from the meta-analysis, holds potential to increase cost-effectiveness by ~50–100%.

We integrate this thinking into our overall decision-making for our next steps, laid out in the following section.

Up Next – Iterating on the model for greater impact and cost-effectiveness

Taking these CEA considerations into account, and with our pilot demonstrating encouraging early signs of impact, we have a clear path ahead: pilot in another state, enrol children we currently miss, strengthen program quality, and continue our M&E efforts.

Pilot in another Nigerian State

The paper registers used in Nigeria are standard nationwide, and the primary health system is fairly standardized as well. We therefore strongly suspect our program design should be operationally feasible in many states. We want to test this hypothesis and have shortlisted three potentially highly cost-effective candidate states. We are preparing government engagement and local partner selection for the most promising one.

Find a way to enrol a wider range of children from the general population

We will test several approaches to reach children missed in vaccination registers, with door-to-door and community-based enrolment emerging as leading candidates. While these may increase costs, Nigeria’s low baseline coverage and high child mortality suggest that even higher-cost enrolment can remain highly cost-effective. We’ll trial working with CHWs, outreach teams, and vaccination campaigns, and explore partnerships to share costs with organizations already reaching wide populations — such as those distributing ORSZ, Vitamin A, or bednets.

Iterate on the program design to reduce cost and improve impact

We will work through a long list of opportunities in quarterly iterations and monitor our program metric scorecard along the way.

To reduce one of our largest cost drivers — travel to health facilities — we’re shifting to a ‘health worker–led’ model where health workers send us register photos. During the pilot, most owned Android phones, and about 50% reliably submitted photos. With ongoing improvements, this approach can significantly cut costs. For visits we still make, we’ll aim to increase enrolment per visit, improve phone number quality through scorecards, and extend the interval between visits to as close to 36 days as possible — ensuring children are enrolled just before reminders start.

Beyond basic tweaks to message timing and content, we’re also exploring potential add-ons to our program as well as changes to our reminder design. WhatsApp could offer a more salient reminder channel, especially with media elements. Two-way messaging could support deeper behaviour change by encouraging planning or addressing mild hesitancy.

Continue to build our monitoring and evaluation muscle

As we expand, we will further strengthen our monitoring and evaluation system to guide us. This means getting faster, more comprehensive, and reliable feedback from implementation, including through metrics suitable for A/B testing. It also means getting to the next level of evaluation with evidence that our reminder program works beyond the observational analysis from our Pilot. The option space includes a quasi-experimental design, a phased rollout with randomized facilities, or a full RCT. We will work with our advisors to find the best timing and design the most suitable approach as we progress.

Looking further ahead

On a longer time horizon, the impact potential is massive. In a few years, we aim to cover millions of children annually across several countries in sub-Saharan Africa. Reminders can be a scalable core driver of vaccination demand, whilst providing data and natural touchpoints that we can build upon with adjacent interventions, unlocking additional impact for a low marginal cost. Over time, we think technology will enable more impactful messaging — for example, local language text-to-speech or two-way communication — and it will drive down costs further as record digitisation and existing databases become more reliable. In Nigeria alone, more than one million children are born annually in neglected states with large cost-effectiveness potential. Several million more children are born annually in other sub-Saharan African countries, where we might work in the most impactful sub-national areas to make sure every child is vaccinated. Ultimately, our program is a prime candidate for government adoption, integrated into a digital health system.

How you can get involved

There are many ways your support can be impactful:

Join our team - A strong generalist tackling projects across Operations, Tech, Fundraising, Programs, and M&E could greatly contribute to realizing Notify Health’s impact potential. Sign up to our newsletter and get notified when applications are open!

Fund our 2025 goals - GiveWell contributed $160,000 to cover roughly half of our budget, but we still have a funding gap of around $106,000. You can help us close that so that our vaccination reminder program reaches its full impact potential. Learn how to donate here.

Improve our thinking - Our Strategy and decisions now are crucial for the trajectory of Notify Health. If you have ideas or critiques, we want to hear them. We are also looking for further advisors with experience scaling behaviour change programs in LMICs, particularly through governments. Reach out to us here.

Meet us at EAG London - Both Sam and Daniel will be at the conference from 6th to 8th of June 2025. We would love to meet you there.

Acknowledgements - Who we’re grateful for

We are deeply grateful to the many individuals and organizations whose invaluable contributions have made our work possible. Particular mention goes to Dylan Collins and Madeline Copp for their expertise, and insights. Additionally our thanks go to our dedicated volunteers—Petra Mossop, Brian Foerester, Rasool Somji, Vijay Kotecha, Anand Jeevanandham, Shrilaxmi Patil, and Magdalena Kolczyńska—for generously sharing their time, skills, and commitment.

We are fortunate to be guided by insightful advisors, including Dr. Obinna Ebirim, Patrick Stadler, and Fiona Conlon, whose expertise continues to shape our approach. We also wish to thank Samantha Kagel and Steve Thompson for their guidance in the early stages and ongoing support.

In Kogi State, our partnership with Renaissance Care and Empowerment Foundation, along with their outstanding leadership and staff, has been central to our work on the ground. We also acknowledge the Kogi State Primary Healthcare Development Agency and Ministry of Health for its essential role and ongoing support. We also extend our gratitude to the various health workers in Kogi State who were involved with the pilot, without whom this work would not be possible.

We also extend our heartfelt thanks to Ambitious Impact, GiveWell and our individual donors, especially our seed funders, whose generous funding supports the core of our work.

1. ^{^}

   Based on Givewell’s ‘probability of death’ from a CEA of SMS reminders for vaccination of 2.7% for Nigeria as a whole. Data based on IHME Global Burden of Disease (GBD) 

2. ^{^}

    As defined by the 2021 Nigerian Multi-indicator Cluster Survey and National Immunization coverage survey (MICS & NICS)  page 200 and following

3. ^{^}

    Antenatal and birth registers might reach a slightly wider population and could provide additional opportunities to remind for ANC or birth vaccines, however, due to their current layout and information captured, they are more difficult to use for our program

^{^}

 These were chosen to be a mix of facilities representative of the context in the entire State. We chose facilities across the three main socio-cultural zones in the state, each with different primary languages, as well as a mix of urban and rural facilities.

^{^}

 Sometimes children appear for the first time in vaccination registers when they are already older than 6 weeks, in which case we still enroll them to send caregivers reminders for the remaining vaccines.

^{^}

 We use the Demographic Health Survey (DHS) or Multiple Indicator Cluster Survey (MICS) for vaccination coverage by Nigerian State. These are collected through door-to-door surveys in randomly chosen locations. This approach samples the “general population” including many children that have zero vaccines and therefore do not show up in our registers. We adjust the coverage of our enrolled caregivers by using the strict assumption that only those in the population that have a birth vaccine (BCG) have any other vaccine, meaning everyone else has no vaccine at all, and that we do not enrol anyone that has no vaccine at all. For example, if the general population has a BCG coverage of 80% and a Penta-1 (due at 6 weeks) coverage of 60%, our recruited population is assumed to have 100% BCG coverage and 75% Penta-1 coverage (60%/80%).

^{^}

 The reduction in vaccination gap increases from 15% to 16%. One study showing a high effect size was suspected of baseline imbalance. We take out the study from the meta-analysis and reduce the internal validity discount in return. We further reduce the discount because one main driver mentioned is what we believe to be an imperfect comparison of the effect size found in the meta-analysis to the New Incentives RCT (skeptical prior). More details can be found in the CEA.

Show all footnotes