Plain counts across 265 circulars, computed from the parsed index. No conclusions are drawn beyond what the records state.
Read this first: the archive is incomplete before 2024
Each circular carries a serial number. Comparing the highest serial issued in a year against how many of that year’s circulars this index actually lists shows how much of the record survives.
20193%
20203%
20213%
202210%
202317%
202494%
202590%
2026100%
Coverage = circulars held ÷ highest serial issued that year. Years under 50% cannot support any trend claim.
This changes how the charts below should be read. A low bar in 2019 to 2023 reflects missing archive, not quiet years. Only 2024 onward is close enough to complete to compare.
Quality reporting gave way to enforcement
Splitting each year into lab reports on the one hand, and withdrawals, revocations and recalls on the other.
Quality failure Enforcement2023 shows no enforcement at all, but only 16% of that year's circulars are archived. Treat 2023 as unknown, not as zero.
Jan to Sep 2025: 1 of 77 enforcement (1%)
Same nine-month window as below, so the comparison is not distorted by the current partial year.
Jan to Sep 2026: 32 of 50 enforcement (64%)
Enforcement appears inside the same MSD/Q/P series that previously carried only lab reports, so this is not a new classification scheme. The “failing sample” template fell from 92% of 2024 to 0% in 2026.
Circulars by year
Shape tracks archive completeness above, not activity.
Counts include every circular in the index with that issue year.
Action types
What the MSD actually did, by action recorded on the index.
Counts are circulars, not separate incidents. Where two strengths are issued as separate circulars on one day, that day is one event.
Most repeated manufacturers, with a warning
176 of 265 circulars name no manufacturer at all, and 43 of 63 makers appear exactly once. Ranking companies by count is therefore weak evidence and should not be read as a quality league table.
Only makers appearing more than once are listed. This covers a minority of the corpus.
Injectables dominate
Keyword classification of the description text. The categories partition the corpus and sum to 265, but a product can still be misfiled, so read this as a sense of proportion rather than a clinical classification.
Keyword matching on the description only. A product can be misfiled here, and the original circular always governs.
Where the products in this index come from
Manufacturing country as written in the circular text. These are counts of circulars, not failure rates. Sri Lanka imports the overwhelming majority of its medicines, so countries appear here largely as a reflection of what is imported and in what volume, not of any difference in quality.
89 of 265 circulars state both a manufacturer and a country. A country figure here says nothing about the share of that country's products which passed, because the index records only failures and withdrawals.
Read this as supply origin, not as a verdict on any country. Only products that failed appear in this index at all. A country with a high number here may simply supply more of what Sri Lanka buys. No inference about the quality of medicines made in any country can be drawn from these counts.