LPG connections but chulha smoke: why the kitchen remains polluted
In a Sundarban fringe village of 640 households, 88% have an LPG connection, but the average refill is only 1.2 cylinders a year. About 72% of households still cook on a mud chulha for part of the day. Winter kitchen PM2.5 measured over 24 hours is 480 µg/m³, while the outdoor concentration is 55 µg/m³. Personal carbon monoxide exposure for the primary cook is 8.1 ppm. The local health sub-centre records acute respiratory infection in 34% of children under five. A trial improved biomass stove reduced kitchen PM2.5 by 35%, but 61% of households discontinued it within four months, citing fuel-size requirements, taste of food and the need for two stoves. The district health officer wants a costed plan before the next winter. A local NGO proposes distributing free LPG refills; another proposes banning chulhas.
1. Interpret the exposure data and explain why high LPG connection coverage has not reduced kitchen air pollution or childhood respiratory infection.
Read the worked answer
Connection coverage measures access, not use. An average of 1.2 refills per year is far below the roughly 8–10 cylinders needed for exclusive LPG cooking. With 72% of households stacking fuels, the chulha remains the everyday stove, so the kitchen concentration is dominated by biomass smoke. A kitchen PM2.5 of 480 µg/m³ is about 8.7 times the outdoor concentration and is the relevant exposure for the cook and young children who stay near the kitchen. Carbon monoxide exposure of 8.1 ppm adds to the risk.
The 34% under-five acute respiratory infection rate is consistent with high household air pollution, but it is not proof of causation. Other factors include malnutrition, crowding, tobacco smoke, poor ventilation, and access to care. Still, the exposure gap is large enough to justify intervention.
The 35% reduction from the improved stove sounds useful, but discontinuation of 61% within four months means real-world exposure barely changes. Fuel preparation, food taste and the need for two stoves are design and social constraints, not irrational behaviour.
2. Design a costed intervention package and an evaluation design. Address affordability, gender and ethics.
Read the worked answer
Package: a local LPG refill depot to cut travel; targeted refill support for the poorest households, especially during winter; a stove-and-cylinder package with repair service; kitchen ventilation improvements such as windows and chimneys where LPG is not yet exclusive; and a behaviour-change campaign led by women's self-help groups. Improved biomass stoves should be offered only as a transitional option, with fuel processing and a trial period. A ban on chulhas without affordable LPG would penalise poor households and is not recommended.
Evaluation: select 12 villages and randomise 6 to the package and 6 to delayed introduction. Measure 24-hour kitchen PM2.5 with gravimetric samplers, personal CO, stove use monitors, refill records and under-five ARI reported by trained health workers at baseline, after one winter and after two winters. Use difference-in-differences, with outdoor PM2.5 and season as covariates.
Ethics: the delayed group must receive the package after one year; do not withhold care; obtain consent; provide referral for children with severe respiratory symptoms. Report results to the panchayat.
3. What measurement and confounding problems could make the trial misleading, and how would you reduce them?
Read the worked answer
Measurement problems: optical PM2.5 sensors are sensitive to humidity and smoke composition; without gravimetric correction they can over- or under-read. A single 24-hour sample may miss cooking peaks, so use real-time monitors plus filter samples. Stove use monitors detect stove use but not fuel quantity. Self-reported refills may be inaccurate; verify with dealer records.
Confounders: season and outdoor PM2.5, tobacco smoking, crowding, child nutrition, health-care access, and wealth. Households that accept the intervention may already be more health-conscious. Randomisation reduces selection bias, but contamination can occur if control villages buy LPG from the same depot.
Reduction: randomise at village level, measure outdoor PM2.5, record smoking and crowding, stratify by wealth, use objective refill records, and analyse by intention to treat. Pre-register the primary outcome and the analysis plan. If refill support is withdrawn after the trial, report whether use persists.