How would we know organising works? A published research plan
We have not yet measured whether any feature in Rifd prevents a medication error or improves coordination between family members, and we will not claim a measurement without data. This is a published plan: three questions, the measures for each, the consent and anonymisation rules, and what counts as a negative result.
· The Rifd team · Review policy
This article has not been pharmacist-reviewed yet. read the review policy.
What this page says, and what it does not
There is no number on this page, deliberately. Any accuracy or improvement figure published before data collection is invented, and any study published without a stated method cannot be disproved. What we publish today is the method and the commitment, not the result.
Nor do we say Rifd "improves medication adherence" or "reduces errors" — the sentence a great many pages write with no source behind it. What we say is that organisation addresses a real problem caregivers describe every week: information is scattered, responsibility is unnamed, and a mistake goes unnoticed until it happens.
The three questions and what we measure for each
Each study has a question whose answer could be no. That is a precondition for any honest measurement: if a study is designed so that it cannot fail, it is a demonstration rather than research.
| Study | The question that could be answered no | What we would count as a negative result |
|---|---|---|
| Extraction accuracy on Arabic prescriptions | Does the extraction return the correct drug name, dose and frequency from an Arabic prescription image as reliably as from an English one? | A lower exact-match rate for Arabic names, or repeated systematic errors in numbers and units (ml, mg) that do not appear in English. |
| Coordination between family members | Do reported duplicate or missed doses fall when more than one person can see the same record? | No measurable difference, or a difference that appears only in single-caregiver households and does not extend to shared ones. |
| Medication organisation at home | Does the household's list come closer to what a pharmacist verifies after two weeks of use? | A longer but still inaccurate list, or an improvement that disappears after the first update — a tool filled in once and then abandoned. |
How the data would actually be collected
Every study needs real participants, so research ethics is not an appendix at the end of the plan but a precondition at the start. The following rules are non-negotiable, and a study that cannot meet them does not run.
- Explicit prior consent: no data is taken from an account unless its owner agreed to that specific study, in a sentence they had to read at least once.
- Anonymise at collection, not afterwards: no names, national ID numbers, phone numbers or addresses, and free-text fields that could carry a name are stripped.
- Shift the dates: keep the interval between events rather than the exact date, so no participant can be tied to an appointment at a named hospital.
- A right to withdraw: a participant can ask for their data to be removed from a study, and it is removed unless it has already entered a published analysis — said clearly in advance.
- No medical advice during a study: researchers do not answer medication questions, and every such question is sent back to a pharmacist or doctor.
- No sale, no commercial sharing: study data is used for the published question and nothing else, and is never sold to a third party.
How we publish, and how we handle failure
Measures get recorded before collection, not after, because choosing the measure once the data is in is how positive results get manufactured out of any dataset. Then we publish the method, the sample size and the limitations, and a negative result is published as a result.
We also state in advance what a study this size cannot say: a small sample does not generalise to every household, an improvement in use does not establish a health outcome, and the absence of a difference does not mean a tool is useless in every case. Those limits get written with the result, not in a separate footnote.
And if a study concludes that a feature does not work, the choice is to change or remove the feature, not to change the question. What happens at that moment is the real test of whether any of this is binding.
Why put this on a public page?
Because a claim without a method is what makes this category untrustworthy: plenty of pages tell you an app "reduced errors" with no figure, no source and no sample, and nobody asks them for the detail. We cannot ask you to accept our claims on a lower standard than the one you apply to everyone else.
And if you want practical help now rather than to wait for a study: the printed schedule and the handover checklist work today, and neither waits on any proof — because every tool, digital or paper, needs one accurate list first.
Take it with you: printable files
No account and no sign-up: one page you print, or fill in on screen.
- Blank printable medication scheduleOne page, six columns: the medication name as printed on the box, the dose, the times, the course length, the pharmacy and one note. Print it and fill it in by hand.Printable page (A4)
- Family caregiver handover checklistA written handover of everything the person covering for you needs for a week: medications, the week's appointments, emergency numbers, and the small details that get lost.Printable checklist (A4)