What adherence looks like when you actually measure it
Published · 6 min read
Every rehabilitation product reports adherence. Very few say what they counted, and the definition is where the number is won or lost.
The easy definition, and why it flatters
The simplest measure is sessions completed divided by sessions prescribed. It is what most dashboards show and it has an obvious flaw: it treats every absence identically.
A patient who skipped three days because they had flu and a patient who stopped opening the app produce the same percentage. One of those is a clinical fact and the other is a problem, and a single ratio cannot tell them apart.
Suspensions, so the denominator stays honest
A programme in Ormeda can be suspended. Illness, travel, a competing medical event. While it is suspended, days are not counted as missed.
That sounds like a feature for the patient. It is really a feature for the data. Without it, every legitimate interruption degrades the adherence figure, and once clinicians learn that the number punishes normal life they stop trusting it. A metric people have learned to discount is worse than no metric.
The programme states are pending, in process, done and stopped. Suspension is a state, not a gap in the record.
Adherence is not the headline
The programme level recovery score runs 0 to 100, and adherence is reported beside it as one of its components, with its own raw start and end values.
That is deliberate. A patient can complete every session and not improve, and a programme that rewards attendance over outcome will produce exactly that. Showing adherence next to the score, where a clinician can read the two against each other, means a high figure can be checked against what actually changed.
Traffic lights, and one that cannot be scrolled past
Scored questionnaires carry thresholds that produce three kinds of event: continue, repeat the day, or stop and review.
A stop event blocks the next day until a clinician acknowledges it. That is the part that matters. An alert that can be dismissed without a decision is a notification, and notifications get trained out. An alert that halts the programme forces someone to look.
Over a cohort, the distribution of those events is itself the signal. Many repeat events clustered at the same point in a protocol says something about the protocol, not about the patients.
What the clinic sees that the patient does not
Pain scores, completed exercises, questionnaire answers and medication logs are structured at the moment they are captured. That means a report is a query rather than a data entry project.
Because the rehabilitation instance is attached to the encounter, the outcome report is part of the medical record rather than an export from somebody else's app. The audit trail that covers the rest of the record covers it too: every view and edit is traced, and corrections are amendments rather than deletions. That is described on the security page.
The limitation worth stating
None of this measures whether the patient performed the exercise correctly. A completed entry means the patient marked it complete. Photo entries and video review help, and a supervised programme helps more, but self reported completion is self reported.
Any vendor telling you their adherence figure proves correct technique is selling you something. What it proves is engagement, which correlates with outcome and is not the same thing.
See the reporting
The charts described here are the ones clinicians see inside the platform, on the rehabilitation module page. The figures shown publicly are illustrative, because patient data is not published. The structure they are computed from is public, though: the published ACL programme shows the days, the dosage and the check-ins a report is built on.
Programme content is written and assigned by the treating hospital or clinic. To see the reporting against your own cohort, request a demo.