Caste and Ethnicity in Nepal Health Data: Ethical Classification and Equity Analysis

Disaggregated data can show whether access, service quality or health outcomes differ between populations. This helps planners identify exclusion that a national average may hide and direct resources toward avoidable gaps.

The purpose must be explicit and legitimate. Collecting identity data without a plan for analysis, protection and action creates risk without public-health value.

The file groups communities under labels such as Dalit, disadvantaged Janajati, disadvantaged non-Dalit Terai caste, religious minority, relatively advantaged Janajati and upper caste. Such groupings have been used in earlier surveys and equity analyses, but they combine identity with an analyst’s judgement about social position.

Nepal’s population categories, spellings, administrative standards and public terminology have evolved. The list should not be copied into registration systems or research instruments without checking the current National Statistics Office classification, the study purpose and consultation with affected communities.

Ask people to identify themselves rather than assigning a group from surname, appearance, language, address or a staff member’s assumption. State why the question is asked, whether answering is voluntary, who can see the response and how it will be used.

Use current names that communities recognize. Provide an appropriate other or not-stated option when the data standard allows it, and design for people whose identities do not fit a simplified hierarchy.

Choose categories before analysis and document any aggregation. Compare access, experience, quality or outcomes with suitable denominators and uncertainty. Adjusting for geography or income may answer one question but can also hide pathways through which structural discrimination affects health.

Avoid ranking communities as inherently advantaged or disadvantaged. Describe observed conditions, historical exclusion and system barriers, and distinguish association from causation.

Caste or ethnicity can be sensitive personal information. Collect the minimum necessary, restrict access, define retention, remove direct identifiers and assess whether combinations of place, age and group could identify someone.

Suppress, combine or otherwise protect small cells before release. Never publish line lists, use identity data to deny services or expose a person to discrimination, retaliation or social harm.

Disaggregation has value only when institutions respond to inequity. Pair statistics with qualitative evidence and community experience, identify modifiable barriers and assign responsibility for action.

Monitor whether interventions improve access and outcomes without creating new harms. Be transparent when categories or methods change, because trends may otherwise reflect classification changes rather than real progress.