31,871 individual submissions have reached the committee that decides which medicines Australia pays for.

How this was built, and what it cannot tell you

The measurement is read directly from each document, with no model involved. The PBAC states all three signals plainly: whether comments were received, how many individuals, health professionals and organisations wrote, and which organisations they were. A model would add nothing there except a way to be wrong.

The grouping is a different matter and worth separating. Therapy area and the condition each submission was for are classified by a language model from the indication text, because the documents do not carry either as a field. So the counts on this page are the document's own; the buckets they are sorted into are an interpretation, and an individual condition label can be wrong. That is why the conditions are browsable one by one rather than only in aggregate.

Three limits worth stating. 33% of decisions have no recoverable consumer-comments section, overwhelmingly before 2014, which is why both the therapy-area and condition views are cut at that year. Read across the whole corpus instead, the quietest conditions come back as hypertension, postmenopausal osteoporosis and major depression, none of which are silent: their decisions simply pre-date the facility. And silence in this register means no comment reached the committee through the consumer facility — not that no one cared, and not that nobody was affected.

People are counted here, never quoted. Organisations are public actors making public submissions and are named; individual accounts of illness stay in the documents they were written for.