Child Marriage in Bangladesh: Convert the Health System into the Frontline Detection and Deterrence Network

Diagnosis

Child marriage is not a fringe practice in Bangladesh, it is the median experience for girls. The note records that 51% of girls are married before age 18 per BDHS, and that Bangladesh holds the world's 4th-highest rate. At that prevalence the harm is structural rather than episodic: early marriage drives adolescent pregnancy, obstructed labor, maternal and newborn mortality, school dropout, and a lifetime of lower earnings and bargaining power. Because more than half of girls are affected, this is a population-level health and human-capital loss, not a set of individual misfortunes, which is why it sits in the health and injury domain with a structural horizon.

The practical problem is visibility and incentives. Marriages are arranged inside households and solemnized locally, often without the state seeing them until a girl appears pregnant at a clinic. The system that already touches nearly every adolescent girl, community health, is not currently organized to detect, deter, or document underage union. That is the leverage point.

Recommended actions

1. Embed an age-verification and flagging step at every adolescent health contact. Owner: Directorate General of Health Services (DGHS). Mechanism: a DGHS circular mandating that community clinics, family welfare centers, and EPI/antenatal touchpoints record date of birth, marital status, and pregnancy status for every girl seen, with any married-or-pregnant minor auto-flagged in the DHIS2 health information system. Observable signal: a rising count of flagged adolescent cases entering the system, which means cases that were previously invisible are now being seen.
2. Stand up a community-health referral protocol that routes flagged cases to local authorities. Owner: DGHS, working through its community health workers and health assistants. Mechanism: a standing operating procedure that triggers a confidential referral to the Upazila Nirbahi Officer and Marriage Registrar within a fixed window whenever a minor is flagged, paired with a counseling visit to the household. Observable signal: share of flagged cases with a documented referral and follow-up visit climbing toward full coverage.
3. Make the marriage-registration and birth-registration records talk to each other. Owner: DGHS, with Department of Public Health Engineering supporting on the registration-infrastructure and data side. Mechanism: link the birth-registration record (which DGHS facilities help generate at delivery) to the marriage registrar's check, so a registrar can verify age against an authoritative record before solemnizing. Observable signal: registrars rejecting or pausing underage applications, recorded as a verifiable rejection count.
4. Tie a conditional incentive to staying unmarried and in school through the health channel. Owner: DGHS as the delivery and verification arm. Mechanism: a budget line for an adolescent-girl conditional stipend, with eligibility verified at the same community health contact that does the flagging, so the same visit both detects risk and rewards delay. Observable signal: a measurable drop in the share of newly flagged underage marriages in stipend-covered areas versus comparison areas.
5. Publish a quarterly district scorecard. Owner: DGHS. Mechanism: an open dashboard built from the flagging data, ranking districts by detected underage-marriage and adolescent-pregnancy load. Observable signal: districts moving down the load ranking over successive quarters.

Sequencing (first 12 months)

Start with action 1, the circular and the DHIS2 flag, because nothing else works until cases are visible. Detection is the asset that every later step consumes. Once flagging is live in a set of pilot upazilas, switch on action 2, the referral protocol, so the new data produces an action and not just a number. In parallel, begin action 3, the records linkage, since it is the slowest to build and should be maturing by the time pilots scale. Reserve the conditional stipend (action 4) and the public scorecard (action 5) for the second half of the year, once the flagging data is clean enough to verify eligibility and to rank districts honestly. The unlock is sequential: visibility unlocks referral, referral plus records unlock deterrence, and clean data unlocks both incentives and accountability.

Risks and constraints

The binding constraints are political and administrative, not technical. Underage marriage is socially sanctioned in many communities, so households may avoid flagged clinics or under-report, which means coverage and trust have to be protected or detection collapses. Registrars and local officials face local pressure to look away, so the referral and rejection mechanisms need protection from capture. Fiscally, the conditional stipend is the heaviest item and competes with every other health priority for the same budget line, so it should be piloted and proven before any national commitment. Confidentiality is a hard requirement: a flagging system that exposes girls to retaliation would do more harm than the status quo.

Bottom line

With more than half of girls married before 18 and Bangladesh fourth-highest in the world, the fastest available lever is the health system that already meets nearly every adolescent girl, reorganized by DGHS to detect, refer, and deter. Make underage marriage visible first, then make it consequential, and reserve the expensive incentive for after the data proves where it will work.