Why Blanket TB Screening Isn't Enough: The Case for Geographic Targeting

Active case finding is central to India's National TB Elimination Program, and it works. When screening is spread evenly across a population, though, it is inefficient. Teams screen a large number of people to find a comparatively small number of confirmed cases. In one recent Pune campaign, 177,322 people were screened across four tuberculosis units to confirm nine cases. That is about 19,702 people screened for each confirmed case. The constraint on the programme is not effort. It is targeting.

Tuberculosis clusters in places

Tuberculosis does not spread evenly. It concentrates in identifiable neighbourhoods, often those that are hardest to reach and easiest to miss in conventional planning. The block or ward, the usual unit of campaign planning, is much larger than the settlement clusters where transmission actually happens. In a dense urban ward there is often no finer official signal, and covering the whole ward is often not practical.

This pattern is an opportunity. Directing a fixed amount of screening toward higher-risk clusters, rather than spreading it evenly, should allow the same effort to find a similar or greater number of cases.

The cost of untargeted search

A look back at three Pune screening campaigns suggests that AI-NETRA's targeting plan would have captured about 89 percent of confirmed cases while directing screening to under one-third of the population. Applied across a district, this concentration could reduce the cost of a full screening cycle from about Rs 7.3 crore to Rs 3.3 crore, a saving of about Rs 4 crore per cycle, if the predicted level of detection is achieved in the field.

The gap inside cities

Ranking tuberculosis burden by block or ward does not show risk below the ward boundary. In dense urban settings, that is exactly where transmission is most concentrated. In Pune's urban units, a person flagged by AI-NETRA as needing a test was 3.6 times more likely to be a confirmed case than a person identified without that guidance. The gain came from more precise referrals, not from testing more people. In rural units, targeting increased the number of people identified as needing a test. Both effects are reported separately in the study, because they arise in different ways.

Targeting and thorough screening belong together

Precision targeting and community-wide screening are not competing ideas. The evidence suggests they should be used in sequence. AI-NETRA helps decide where thorough screening is warranted. Saturation of that area, complete diagnosis, and a fresh map for the next cycle then follow. The Pune pilot did not find a difference in total case detection when every form of confirmed tuberculosis was counted. That result is consistent with a programme that ranked areas without then covering those areas with the same diagnostic capacity. It is not, on its own, a finding against targeting.

Conclusion

Geographic targeting should be shown in the field, not asserted as a principle. The AI-NETRA Pune Pilot Study provides that evidence, together with a clear account of its limits. The complete study is available on the Pune Pilot Study page.