Reported infectious-disease case counts under differing case definitions, 18 diseases, 2018–2024 — with the reporting-coverage denominator

Two questions, one dataset. (1) How much does case definition change the apparent size of an epidemic? Confirmed andsuspected series are published paired, per country and year, from named originatingauthorities. In 2024 Brazil reports 6,579,385 laboratory-confirmed dengue cases and10,330,277 suspected — a 57% difference for the same event. (2) How much of an apparent multi-year trend in an aggregated surveillance collection isactually reporting coverage? Every annual row carries the number of countries reporting,so the two can be told apart. WHY THE COVERAGE COLUMN MATTERSCountries reporting to this collection grew from 22 in 2018 to 99 in 2024. Seasonalinfluenza rises from 25,104 cases in 2018 to 917,338 in 2024 — a 37-fold increase thatreads as an epidemic and is almost entirely one country joining in 2022. Dengue rises25-fold over the same period while its reporting countries only double, and its largestreporter is present in every year: that rise is a real epidemic. Same table, oppositestories. Totals without a denominator invite the error. CONTENTS alertx_open_panel_by_country_year.csv 1,622 rows: 18 diseases x 108 countries x 2018-2024, confirmed and suspected columns, originating authority named per row alertx_open_panel_annual_with_coverage.csv annual totals WITH countries_reporting alertx_open_panel_withheld_cells.csv 44 cells excluded, each with its reason dataset_metadata.json method, limitations, SHA-256 checksums CHECKSUMS.sha256 integrity METHODCounts aggregated with period-fit rules; running totals differenced rather than summed;national reporting supersedes subnational for the same population. Calendar-year windowswith full prior history so cumulative counters anchor correctly. Confirmed and suspectedare classified separately and are NEVER summed — the suspected series generally containsthe confirmed one. Every cell passes a published quality gate before inclusion. WITHHELD CELLSCells failing a gate are withheld and listed with the reason rather than dropped silently.Gates: unnamed authority, thin support, single-period bulk, unclassified metric,implausible jump, and situation-report figure. The last exists because two national malariafigures (7,300,000 and 3,400,000) are single running-outbreak totals quoted in situationreports, not surveillance series.reports, not surveillance series. LIMITATIONS* Reported surveillance counts, not modelled incidence or true burden.* Confirmed and suspected are two measurements of one population; adding them double-counts.* Coverage is uneven and changes over time. Do not read any annual series as a trend without the coverage columns.* Absence of a country means it did not report to this collection, not that the disease was absent.* Only complete calendar years are published, released once the following year has also closed. SUPERSEDES version 1.0 (dengue only), 10.5281/zenodo.21855171. METHOD SOFTWARE: https://doi.org/10.5281/zenodo.21855526