Skip to main content
Expert Talk
Amakomaya — Aamako Maya Apps Pvt. Ltd.

Case Study: Measuring Maternal and Neonatal Mortality in Kapilvastu

Case studyBy , Co-Founder & CHIO, Aamako Maya Apps Pvt. Ltd.Published 9 min read

How this project ran​

  1. CARE NEEDDistrict-level maternal mortality is not directly measured, least of all where reporting is weakest.
  2. CURRENT WORKFLOWA death at home may never reach facility records, and a district average conceals the variation.
  3. IMPROVED WORKFLOWCensus-equivalent enumeration, records attached to individual women, geography kept as a variable.
  4. DIGITAL SOLUTIONRegistration and longitudinal follow-up on a central record under role-based access.
  5. DATA EXCHANGEExtraction, formatting and delivery to NPHF on a defined cycle — a data-quality mechanism as much as a reporting one.
  6. IMPLEMENTATIONA division of labour in which NPHF led and financed the rollout and Amakomaya ran the digital component.
  7. RESULTThe mechanism through which the information was collected, maintained, mapped and reported. The study owns its findings.

The problem​

Nepal's maternal mortality ratio is a national estimate. It is not a number that exists at district level in any directly measured form, and the places where it is highest are the places where routine reporting is weakest. A woman who dies at home, undelivered or shortly after birth, may not appear in facility records at all. The event that most needs counting is the one least likely to be recorded.

Kapilvastu, in Lumbini Province on the Terai plain, is a hard case by design. The district is eco-ethnographically diverse — Madhesi, Tharu, Muslim and hill-origin communities, different languages, different care-seeking behaviour, and a settlement pattern that varies from municipal centres to villages that see a health worker infrequently. A district average would conceal exactly the variation the study existed to find.

So the study needed three things at once: enumeration wide enough to be census-equivalent, records attached to individual women rather than to visits, and geography preserved as an analytic variable.

The study​

The Nepal Public Health Foundation is an independent non-profit established in 2010 by a group of public health specialists and activists, based at Upahar Marga, Baluwatar, Kathmandu. Its stated vision is ensuring health as the right and responsibility of the Nepali people, with health policy and systems research, social determinants of health, and maternal, child and reproductive health among its focus areas.

The study — Reducing MMR and NMR in an Eco-Ethnographically Diverse Kapilvastu District of Lumbini Province of Nepal through Community Engaged Digitalized Comprehensive Continuum of Care — was conducted by NPHF in collaboration with the Nepal Health Research Council. Dr. Gehanath Baral served as Co-Principal Investigator, and presented the work at the Health Research and Innovations in Public Health platform convened by the Indian Council of Medical Research.

The published description of the method is specific: a modified Motherhood Method, complemented by GIS mapping and the Amakomaya digital application, generating census-equivalent data on maternal and neonatal mortality across the district's municipalities.

The Motherhood Method estimates maternal mortality by systematic community enumeration rather than by facility reporting — it asks households about deaths among women of reproductive age and establishes, case by case, whether each was pregnancy-related. It is field-heavy by construction. Every household visit produces a record that has to be captured accurately, attributed to a location, and reconciled against the rest of the enumeration. That is where the digital layer earns its place.

What the digital layer had to do​

Pregnant woman and household → mobile application, community mobilizer → DHIS2 eRecord → validation and monitoring → extraction → analysis and GIS

Registration and longitudinal follow-up​

Community mobilizers and other authorized users worked in the mobile application, registering women digitally and updating each record at subsequent follow-up contacts. Each woman held a single continuing record rather than a series of disconnected encounters — which is the difference between a dataset you can analyse and one you can only count.

This matters more in a mortality study than in a service deployment. To classify a death as maternal you need the pregnancy that preceded it. If registration and outcome live in separate tables joined by name and village, the join fails on exactly the cases that matter most.

A central record under role-based access​

Field data was maintained as structured electronic health information in a centralized DHIS2 environment. Credentials were issued by role — community mobilizers and data-entry personnel, implementing organizations working in the district, and NPHF monitoring users — so that field users, partners and the study team each saw what their work required, with individual-level records under controlled access.

Building on DHIS2 rather than a bespoke database was deliberate. DHIS2 is the platform Nepal's health information system already runs on, its tracker model is built for exactly this individual-plus-longitudinal shape, and its organizational-unit hierarchy carries the district, municipality and ward structure that the analysis needed anyway.

Geography as a variable​

Because organizational units and coordinates travel with each record, the enumeration could be resolved to municipality and settlement rather than to the district as a whole, and joined to the study's GIS mapping. In a study whose premise is eco-ethnographic variation, that is not a reporting nicety — it is half the finding.

Extraction on a schedule​

Data was extracted, formatted and delivered to NPHF on a defined cycle: a first report early in implementation, then monthly reporting through the field period. Regular extraction is a data-quality mechanism as much as a reporting one. A study that discovers its completeness problem at analysis has discovered it too late to send anyone back.

The division of labour​

The partnership was structured so that each side did what it was equipped to do.

Nepal Public Health Foundation led and financed the rollout across its programme areas in Kapilvastu, organized training, held the data-access credentials, and undertook to observe data-sharing requirements under the Statistical Act.

Amakomaya Apps Pvt. Ltd. configured and customized the eRecord to the study's requirements, issued user accounts and system access, delivered IT support and training to NPHF's technical team, maintained the platform and resolved reported defects within a fixed window, carried out extraction and formatting, and held responsibility for data security and server operations.

Intellectual property in what was created sits with the Government of Nepal and NPHF. That is the correct arrangement for publicly-funded health research and worth stating plainly: the study's data is not a vendor asset.

What a research deployment demands​

Most of our maternal health work is service delivery — get information to a woman, get a record to a health worker. A study is a different discipline, and three differences did most of the work.

The denominator is the deliverable. In service mode, an incomplete register degrades gracefully; women you missed simply do not receive the service. In mortality estimation, the women you missed are the error. Completeness stops being a quality indicator and becomes the measurement itself.

Every field has to survive a protocol. Service systems can add a field when someone asks. A study cannot — data elements are configured against the protocol, the approved instruments and the ethical approval, and changing one mid-field is a change to the study, not to the software.

The record has to age well. Analysis, peer review and reporting to NHRC and the municipalities happen long after the last household visit. What matters at that point is not the interface but whether the audit trail, the versions and the location data are still coherent.

Outcome​

The study generated granular insight into mortality patterns across Kapilvastu's municipalities and has been presented internationally. The digital component's contribution is well-bounded: it provided the mechanism through which maternal health and mortality information was collected, maintained, monitored, mapped and reported.

It did not, and could not, reduce mortality on its own. Enumeration is not intervention. What a census-equivalent district picture makes possible is the next step — municipalities that can see where their own deaths are occurring, and a baseline precise enough that a later change is measurable rather than asserted. Whether and how mortality moves is for the study's own design, analysis and findings to establish.

What I would tell the next team

Agree the endpoint definition before you configure anything. "Pregnancy outcome" is four different fields depending on who is asking, and reconciling them retrospectively across thousands of records is a quarter of work that a half-day conversation at design time avoids.

Instrument the enumerators, not just the enumeration. Data-entry timing, device activity and follow-up status told us more about field quality than the clinical fields did. In a household-enumeration study, knowing which mobilizer covered which ward when is the audit trail.

Extract early, even when there is nothing to see. The first extraction two months in is not for the data — it is to discover that the export you assumed would work does not, while there is still field time left to fix it.

Say plainly where your contribution ends. A technology partner that lets a mortality reduction claim attach to its platform is doing the study a disservice. The platform produced the measurement. The study owns the finding.

Frequently asked questions​

What is the Motherhood Method for measuring maternal mortality?
It is a community-based enumeration method that estimates maternal mortality by systematically asking households about deaths among women of reproductive age and establishing case by case whether each death was pregnancy-related. It is used where facility-based reporting misses most maternal deaths. The Kapilvastu study applied a modified version complemented by GIS mapping and digital data collection.
Why use DHIS2 for a research study rather than a survey tool?
Survey tools capture events; a mortality study needs individual women followed over time, with pregnancy and outcome connected in one record. DHIS2's tracker model is built for that shape, its organizational-unit hierarchy carries the municipality and ward structure the analysis needs, and it is the platform Nepal's health information system already runs on.
Who owns the data collected in a study like this?
In this partnership, intellectual property created under the agreement is owned by the Government of Nepal and the Nepal Public Health Foundation. The technology partner is responsible for data security and server operations but the data is not a vendor asset, and any extraction or sharing follows the study's approved procedures and access controls.
Did the digital platform reduce maternal mortality in Kapilvastu?
No such claim is made here. The platform provided the mechanism to collect, maintain, monitor and report maternal health and mortality information. Assessment of maternal and neonatal health outcomes and of intervention effectiveness rests with the study's design, analysis and findings.

Sources

  1. Nepal Public Health Foundation — organizational vision, focus areas and establishment. nphfoundation.org
  2. Study title, collaboration with the Nepal Health Research Council, Co-PI Dr. Gehanath Baral, the modified Motherhood Method, GIS mapping, use of the Amakomaya digital application and census-equivalent mortality data — NPHF announcement of the ICMR Health Research and Innovations in Public Health platform (March 2026). nphfoundation.org
  3. NPHF field presence in Kapilvastu — record of a visit to Jahadi Health Post, Kapilvastu Municipality, June 2024. nphfoundation.org
  4. Scope of the partnership, division of responsibilities, role-based user groups, defect-resolution and reporting obligations, data governance and intellectual property: memorandum of understanding between NPHF and Amakomaya Apps Pvt. Ltd.
Last updated on