When Screening Meets the Meal Route: The Boundary MBG Needs for Mothers, Toddlers, and Developmental Support

MBG Watch · 2026-09-27

The premise

Indonesia's meal route is getting wider than the school tray.

On August 7, 2026, ANTARA reported that the government had decided to intensify MBG implementation for pregnant women, breastfeeding mothers, children under five, and residents in 3T regions. The same report said roughly 11.15 million eligible pregnant women, breastfeeding mothers, and toddlers had not yet received the benefit, and that 352 MBG kitchens in 3T regions were being prepared across high-stunting provinces and remote areas. By August 31, BKKBN said the 3B route had reached 10.8 million recipients: 964,000 pregnant women, 2.4 million breastfeeding mothers, and 7.4 million toddlers not attending early-childhood education programs.

That shift matters because MBG is no longer only meeting schools. It is meeting posyandu, puskesmas, family-assistance teams, early-childhood settings, village records, stunting programs, and household-facing care routes. Those same routes are also where developmental, perinatal, psychosocial, and digital support tools increasingly appear.

The question is not whether MBG should become a developmental-screening or mental-health program. It should not. The narrower question is what public coordination record should exist when nutrition support and screening or support activity touch the same mother, child, kitchen, cadre, service day, or referral route.

MBG Watch has already argued for an outcome loop for mothers and toddlers, a privacy boundary for beneficiary validation, a limit on what metabolic sensing can prove, low-burden support after harm, a boundary between immunization and meal routes, and a special-diet and referral record. This piece extends that line. When screening meets the meal route, the public record should show coordination without turning food access into surveillance.

What the evidence supports

The evidence supports a modest claim: screening and digital support tools can help identify need earlier when they are embedded in care systems that have referral capacity, workforce preparation, and privacy safeguards.

WHO's nurturing-care work starts from a broad child-development frame: young children need adequate nutrition, good health, safety and security, responsive caregiving, and opportunities for early learning. Its handbook is deliberately operational. It calls for governance, family and community engagement, stronger services, monitoring of progress, and tools that help systems learn where support is missing. Nutrition is part of that frame, but it is not the whole frame.

Indonesia already has a maternal and child health route where some of this work is expected to happen. The Ministry of Health's public guidance tells families to bring children under five to posyandu regularly, including for growth monitoring, maternal-child health checks, nutrition counselling, and referral to puskesmas when a child is ill or growth falters. The SSI/ASIK help material describes Satusehat Indonesiaku as a centralized application supporting recording and monitoring of community health services outside buildings by primary-care health workers. In other words, MBG is not entering an empty field. It is entering a field with its own health records, service responsibilities, and referral obligations.

The perinatal mental-health evidence points in the same direction. A 2024 systematic review in Archives of Women's Mental Health found digital screening for mental health in pregnancy and postpartum to be acceptable, feasible, and effective across 34 studies, but its implementation barriers were practical: environmental context and resources, skills, professional roles, beliefs about consequences, training, workload, organizational support, and the choice of screening application and setting. A 2026 systematic review of mHealth apps for maternal mental well-being found potential value but cautioned that apps should be treated as supplementary rather than standalone interventions until larger efficacy trials are available. It also noted common limitations: small sample sizes, high dropout rates, and lack of long-term follow-up.

Developmental screening evidence carries a similar boundary. The Watch Me Grow Integrated protocol in Australia was designed as a web-based developmental surveillance approach linked to primary care and referral pathways. Its study questions included whether the approach could improve identification, follow-up, and cost-effectiveness compared with surveillance as usual. That is the right shape of evidence: screening as part of a care pathway, with referral and follow-up, not screening uptake as proof that children have improved.

The useful lesson for MBG is therefore precise. Screening tools can help a care system notice need earlier. They cannot, by themselves, prove that MBG improved nutrition, development, maternal wellbeing, or child safety. They also cannot become a shortcut for deciding who deserves a meal.

What the evidence does not support

The evidence does not support using developmental flags, perinatal screening scores, mobile-sensing traces, app engagement, or psychosocial screening uptake as MBG proof.

Those measures answer different questions. A screening score may suggest that a mother or child needs care. It does not prove that an MBG meal was safe, arrived on time, matched the dietary need, improved growth, reduced distress, or changed a developmental trajectory. App engagement may show that a person opened or completed a tool. It does not show that support was useful, voluntary, trusted, or followed by a real service. A mobile or wearable signal may describe behavior. It does not become consent to expose a household, label a child, or rank families inside a food program.

This distinction protects both sides. It protects MBG from overclaiming outcomes that belong to health, early-childhood, and family-support systems. It also protects mothers and children from having food access entangled with sensitive status: pregnancy, depression risk, developmental concern, household location, attendance irregularity, or whether a caregiver accepted screening.

The least-harm rule is simple: MBG may coordinate with care; it should not make meals conditional on screening.

The public coordination record MBG needs

The record MBG needs is an operating record, not a person-level health file. It should tell the public whether coordination is safe and functional without revealing who was screened, who was referred, who was diagnosed, or where a vulnerable household lives.

A useful public record would have six parts.

1. Aggregate route status

For each district or service cluster, MBG should publish whether its 3B and toddler route is connected to the relevant care settings: posyandu, puskesmas, PAUD or TK where applicable, family-assistance teams, and 3T delivery routes. The record should name the coordination route, not the child.

A public line can say: "SPPG cluster A serves 3B beneficiaries through posyandu and TPK distribution on these days; puskesmas referral contact confirmed." It should not say: "This mother screened positive," "this toddler failed a milestone screen," or "this household receives home delivery because of a health status."

2. Referral availability, not referral identities

If a screening or support activity changes meal delivery or follow-up, the public should see whether referral capacity exists: puskesmas available, child-development or maternal-health service reachable, escalation route named, and follow-up window defined. The public does not need names, diagnoses, NIK, precise addresses, or mental-health scores.

This is especially important in 3T regions. A referral listed in a policy document is not the same as a referral reachable by a family. MBG's record should distinguish "route exists on paper" from "route is active for this service area."

3. Denominator flags

Public numbers should say what population is being counted. Toddlers not enrolled in PAUD, pregnant women reached through TPK, breastfeeding mothers reached through posyandu, schoolchildren reached through class rosters, and 3T households reached through dispersed delivery are not the same denominator.

A denominator flag is not a private label. It is a way to stop false comparisons. If attendance drops on a service day because families were diverted to immunization, screening, flood disruption, heat illness, market disruption, or puskesmas follow-up, the record should identify the denominator change in aggregate. It should not convert a family's reason into a public attribute.

4. Service changes caused by care coordination

When screening or support activity changes the meal operation, the change should be visible. Examples include adjusted delivery timing for posyandu day, separate toddler texture or allergen handling after referral, added waiting-space control, extra cadre staffing, corrected attendance interpretation, temporary suspension of a site, or rerouting after an incident.

The point is not to publish private care facts. The point is to show that the meal system noticed a coordination need and changed the operation safely.

5. A correction trail

If coordination fails, the public record should show correction: what failed, when it was found, who was responsible for the operating fix, what changed, and when the route was restored. This is the same accountability logic MBG Watch has applied to kitchens, complaints, special diets, immunization-route overlap, and post-harm support.

The correction trail should stay operational. It should not become a narrative about a child's condition or a mother's mental health.

6. Suppression rules

The record should have explicit small-cell suppression. Where counts are small enough that a village, PAUD, posyandu, diagnosis, pregnancy status, referral reason, or household can be inferred, the public record should aggregate upward or suppress the field.

This is not a bureaucratic nicety. In a small settlement, "one toddler referred after developmental screening" can identify a family. "One pregnant woman flagged for psychosocial support" can identify a household. "Three mothers absent after mental-health screening day" can become gossip before it becomes governance.

The privacy and coercion boundary

The boundary should be written plainly.

MBG can record and publish, in aggregate:

• whether a service cluster has an active coordination route with posyandu, puskesmas, PAUD/TK, TPK, or 3T delivery teams;
• whether referral capacity was available on the relevant service day;
• whether a denominator changed because the care route changed;
• whether meal delivery, staffing, attendance interpretation, or follow-up changed because of coordination;
• whether an error was corrected;
• whether small-cell suppression was applied.

MBG should not publish, transfer into its public accountability record, or use for eligibility control:

• child identity, NIK, household location, pregnancy status, breastfeeding status, mental-health status, developmental flags, diagnosis, disability status, or referral reason;
• app, wearable, sensor, geolocation, call-log, or passive mobile-sensing traces;
• raw screening scores or caregiver answers;
• school, posyandu, PAUD, or village-level small cells that identify families by implication;
• refusal, noncompletion, or absence from screening as a reason to reduce food access;
• screening uptake as evidence that MBG improved nutrition or development.

The coercion risk is not abstract. If a meal is scarce, recurring, and state-administered, families may reasonably experience adjacent screening as a condition even when a form says it is voluntary. The operational design has to make non-coercion visible: meals continue without screening; refusal is not punished; support is offered through the health or child-development route; MBG receives only the minimum operational signal needed to deliver food safely.

What MBG Watch will monitor next

MBG Watch will monitor five public signs.

First, whether MBG's expanding 3B route publishes clear denominators for pregnant women, breastfeeding mothers, toddlers outside PAUD, early-childhood settings, and 3T areas.

Second, whether public MBG materials distinguish nutrition delivery from health screening, developmental screening, mental-health support, and beneficiary validation.

Third, whether coordination with posyandu, puskesmas, PAUD/TK, ASIK/SSI, and family-assistance teams produces aggregate operating records rather than person-level exposure.

Fourth, whether referral capacity is named before screening is celebrated. A tool that finds need without a reachable referral route can increase burden and anxiety while producing a cleaner-looking dashboard.

Fifth, whether MBG avoids outcome inflation. Screening coverage, app completion, or referral counts may be useful management signals. They are not proof that meals improved child development, maternal mental health, or household wellbeing.

The meal route can cooperate with care. It should not absorb the private life of the family into the meal ledger. The boundary MBG needs is not silence. It is a public record that shows the coordination while leaving the person unexposed.