Abstract
<title>Abstract</title> <p> <bold>Aim:</bold> To assess how haemoglobin measurement and sampling error affect district anaemia prioritisation in India. <bold>Subject and Methods:</bold> We analysed National Family Health Survey 2019–21 (NFHS-5) haemoglobin data for 1,685,994 people in 707 districts and five beneficiary groups. Four Indian paired-sample studies of capillary point-of-care and venous automated-analyser haemoglobin were combined in a random-effects synthesis. For each draw, observed haemoglobin values were shifted, weighted anaemia prevalence was recounted, and district design-based sampling error was added. We examined priority-quintile stability across a sensitivity range for between-district measurement-bias variation. <bold>Results:</bold> The pooled venous-minus-capillary difference was 0.294 g/dL (standard error 0.087; I² 92.8%). For children aged 6–59 months, national prevalence was 68.1% in NFHS-5 and 63.1% (95% uncertainty interval 52.7–70.4) under the Indian-study predictive distribution. With sampling error alone, about half of districts in the measured highest-prevalence quintile remained there with at least 80% probability; this fell to 33.3% when the district-bias standard deviation was 0.163 g/dL and 16.3% at 0.380 g/dL. The national correction reversed direction under an international, two-sided bias prior. <bold>Conclusion:</bold> District anaemia league tables should not be treated as certain under currently available evidence on measurement error. Uncertainty-aware priority bands and a probability sample of paired capillary and venous measurements embedded in a future survey would provide a more defensible basis for targeting. </p>