How Dekona measures whether it works

Status: no results yet, on purpose. Dekona has not completed a measured cohort. This page fixes our definitions, measures, and limits before the first cohort finishes, so that when we publish numbers, you can check them against rules we wrote down in advance. We will not change these definitions after seeing results. Any material change gets its own dated note on this page.

What Dekona claims today

Dekona brings a family's medical records together with permission, flags what changed when new records arrive, and turns what the record shows into work a family can see and finish. That is a coordination claim, not a clinical one. Dekona organizes and coordinates. It does not diagnose, and it is not a substitute for your clinicians.

We make no claim today that Dekona improves adherence, appointment attendance, readmissions, or any health outcome. If we ever make a claim like that, it will meet the standards on this page first.

Primary measures

1. Time to a working care picture. The hours from a family's first record connection to a usable current view: medications, care team, and upcoming appointments present and reviewed. Measured from time-stamped product events for every new family, reported as a median with the spread, never as a best case.

2. Verified coordination follow-through. The share of coordination items Dekona finds in the record (a refill coming due, an unbooked follow-up, a document gap) that reach a verified end state within 14 days. Every item ends in exactly one state:

Done, confirmed. The outcome is confirmed by a source: the record itself, a pharmacy or office confirmation, or a photo of the label or paperwork. We note which source.
Done, unconfirmed. Someone marked it done, but no source confirms it. We count this separately and never fold it into the confirmed number.
Declined. The family understood the item and chose not to act. That is their call, and it is not a failure of theirs.
Blocked. The item cannot proceed, and we name the dependency, such as an authorization that has not come through.
No longer needed. A clinician or plan change ended the need.

Secondary measures

A cohort report may also include: how many caregivers in a family actively share the same picture, whether the ER packet is complete when it is needed, caregiver-reported time spent coordinating each week, and how often the weekly change summary is opened. Caregiver-reported numbers are self-reported and labeled as such.

Cohort rules

A published result will state: the enrollment window, how families were invited, how many were invited, enrolled, and completed, the observation period, and every exclusion. We will not merge unlike groups into one rate, drop inconvenient families from a denominator, or pick a reporting window after seeing the data. Numerator, denominator, and missing data are reported with every rate.

Limits we accept up front

Early cohorts will be small and self-selected. They will not represent every family.
Completing coordination work is not the same as improving health. We will not present one as the other.
Families chose to sign up for Dekona, which selects for motivation. We will say so wherever that matters.
A named reviewer checks the method, the source table, and the math before anything is published.

Why publish this before results

Because every product in this category describes outcomes it has not measured. Writing the rules down first costs us the ability to flatter ourselves later. That is the point.

Questions about the method: support@dekona.care