# Pratirodh evidence map: protocol Written and committed on 27 September 2026, before any record was screened. Changes after this commit are dated below with their reason, as with the seasonal analysis plan (`../PREREG.md`). ## Question Where in India has anyone measured antimicrobial resistance in the environment, in what kind of sample, by what method, and when? The map shows where India has looked, and where it hasn't. It does not pool results or rank places by resistance: studies use different methods and units, and their numbers are never put on one axis. ## Search - **Database:** Europe PMC REST API, PubMed-indexed records (`SRC:MED`). Free, open, and re-runnable by anyone with the exact string below. - **Run:** 27 September 2026. Every record returned is stored as fetched (in the spine's raw store, with a source-fetch record). - **String:** ``` (TITLE_ABS:("antibiotic resistance" OR "antimicrobial resistance" OR "resistance genes" OR "resistome" OR "ESBL" OR "carbapenemase" OR "blaNDM" OR "ARGs") AND TITLE_ABS:(wastewater OR sewage OR "river" OR "surface water" OR "effluent" OR "drain" OR "lake" OR "groundwater" OR "sediment" OR "soil" OR "hospital effluent" OR "treatment plant") AND (TITLE_ABS:India OR AFF:India) AND PUB_YEAR:[2000 TO 2026] AND SRC:MED) ``` The affiliation clause catches Indian studies that don't say "India" in the abstract; it also returns Indian authors' work abroad, which screening removes. Known gaps, stated now: journals not indexed in PubMed, theses, government reports, and preprints are not searched. Reference lists of included reviews are not followed in this round. ## Include A record is included when all of these hold: 1. **Primary data**: the authors collected and analysed samples themselves (not a review, model or re-analysis of someone else's published data, except where it adds sites). 2. **Environment**: water, wastewater, sewage, drains, treatment plants, rivers, lakes, groundwater, sediment, soil, or effluent from hospitals, farms, factories or treatment plants. 3. **Resistance measured**: resistance genes (PCR, qPCR, metagenome, isolate genome) or resistant organisms (culture with susceptibility testing). 4. **In India**: at least one sampling site in India. Excluded, with the reason recorded: clinical samples only; food, animals or people only (gut, milk, meat, faeces from animals); outside India; reviews and commentaries; no resistance measure; laboratory or treatment-technology experiments without field sampling sites. ## Extract For each included study: citation, DOI, year published, sampling years, places as the paper names them (the finest level given), matrix, targets (gene families, organisms, drug classes), method class, number of samples, and whether any site is a hospital or its effluent. Method classes: `metagenome`, `qpcr`, `pcr` (presence), `isolate_genome`, `culture_ast` (culture with susceptibility tests). A study with several gets all of them. Places are matched to the Maplore place registry (LGD) at the finest level the paper supports: city, district or state. Coordinates in a paper are not copied to the map; the site is shown at its city or district. **Hospitals.** A study sampling a hospital's effluent is mapped at its city or district, never at the hospital. The hospital's name is not stored in any field that is published. ## Screening Titles and abstracts are screened by Claude (Anthropic's model), working in Maplore's session, one record at a time, with a decision (`include`, `exclude`, `unclear`) and a one-line reason stored for every record. `unclear` records are resolved from the full text where it is open, and stay `unclear` otherwise. PS reviews a digest (counts, reasons, a sample of each decision), not each record. There is no second independent screener in this round; the map says so. ## Output - Staging: `study_staging` in the spine, one row per included study, with the extraction above. - Layer: places India has studied, by year and method class; places with none are shown as none. - Every study on the map carries its citation and a link to the paper. ## Changes after registration None yet.