Does a slow answer mean a refusal? Not really - not here
Our piece on response times measured how long health FOI responses actually take. The obvious next question is whether the slow ones are slow because something is being withheld - stalling as a soft form of no. Cross-referencing the same date-extraction against refusal and exemption language on the same releases gives a fairly flat answer here, which is itself worth reporting.
Barely any difference
Releases carrying refusal or exemption language ("not held", a cited section, "vexatious") have a median response time of 24 working days. Ones without any - a clean disclosure - have a median of 21 working days. The share landing within the statutory 20-day limit is almost identical too: 37.4% for refused/exemption releases versus 38.2% for clean ones. Statistically, in this dataset, whether a health body is about to say no tells you almost nothing about how long you'll wait for the answer.
That's a genuinely different picture from our companion Policing & Justice tracker's identical piece on this, where the two are entangled in a stranger way - fast administrative "not held" dismissals and slow substantive exemption fights pull the average in opposite directions there, producing refused responses that are simultaneously slower on average and more likely to land within the deadline. Health doesn't show that split. Whatever is driving response time in health FOI, it doesn't look tightly coupled to whether the answer ends in a refusal.
What might explain the difference
We'd guess this is at least partly about the underlying mix of requests. Policing FOI includes both very routine administrative queries (fast, often ending in a quick "not held") and case-specific ones that can trigger genuine exemption review (slow). Health FOI in our dataset, by comparison to the policing side, skews more toward statistical and service-level requests where the answer is either a table of numbers or a straightforward "we don't hold that broken down that way" - fewer of the substantive individual-case fights that stretch out policing's slow tail. That's a plausible read of the pattern, not something we've separately verified.
Reading this responsibly
"Refusal or exemption language" is a text-presence signal, not a case-by-case legal outcome. The 246 releases behind this use the same extraction method as our response-time piece: pattern-matched dates from free text, sanity-filtered, spot-checked by hand rather than independently audited in full.