An N-of-1 trial gives one participant two or more treatments in several randomized periods, with the same outcome measured over and over. The software job is unusually specific: fix the order, remind the participant, record the daily scores against the right period, and give back a per-person dataset.
Free sandbox · No credit card · 21 CFR Part 11 aligned
Study day
Baseline diary
What to know first
Design in brief
The basic unit is a pair of periods: one on treatment A, one on treatment B (or placebo or usual care), with the order within the pair randomized. Repeating the pair two or three times, as in the timeline above, gives each treatment several independent exposures in the same person. That repetition is the point: one A-B sequence cannot separate a treatment effect from a drift in the condition, but three randomized pairs can start to.
Several design choices follow. Period length must be long enough for the treatment to reach its effect and for the outcome to be measured reliably, which depends on the drug. Washout is the gap or the discarded early part of each period during which the previous treatment wears off; if carryover is plausible, a washout or an analysis that ignores the first days of each period is needed. Blinding, where it can be done, needs a treatment supplier who prepares identical-looking products and a code that the participant and clinician do not see until the end. Outcomes should be patient-reported or simple measures taken often, such as a daily pain score, a sleep rating or a symptom count.
N-of-1 design is also used in rare diseases where a conventional trial is not possible, and in treatment decisions where several options are reasonable. Other pages here cover related ground: edc for rare disease clinical trials discusses small populations, and the crossover trial sample size calculator helps size a series of N-of-1 trials combined. A Bayesian framing is described in the Bayesian clinical trial design guide.
Design parameters
Each parameter has a direct effect on forms, reminders and exports.
| Parameter | Typical options | What it changes in the data system |
|---|---|---|
| Number of pairs | 2 to 4 pairs of periods | Number of period labels and the length of the casebook. More pairs mean more diary days to review. |
| Period length | 1 to 4 weeks, depending on the drug | Number of expected diary entries per period and the reminder schedule. |
| Washout | None, a gap, or early days discarded in analysis | Whether diary entries on washout days are collected and flagged or collected and excluded. |
| Order | Randomized within each pair, or a fixed sequence | The sequence list the study team prepares in advance; shown in the period schedule. |
| Blinding | Open, participant-blinded, or double-blind | Which roles can see the treatment label per period; handling of supply codes. |
| Outcome measure | Daily 0 to 10 score, weekly questionnaire, a count | Which of the 10 question types you use and how often the task repeats. |
| Stopping rules | Participant request, adverse event, clinician decision | A discontinuation form and a clear record of the last completed period. |
| Analysis plan | Paired comparison per pair, or a model over all days | Export shape: one row per entry with participant, day, period and treatment label. |
The diary
In an N-of-1 trial the diary is not a side-measure; it is the primary data stream. Capture's ePRO runs in the participant's browser from a link, with no app to install, and supports the 10 question types: text, numeric, linear scale, slider, single choice, multiple choice, dropdown, date, time and long text. Diary tasks repeat on a schedule and reminders go out to participants, so a missed day is visible rather than silent.
Daily diary · Pair 2, period 2
Participant 001-0007 · Day 61 of 106
Pain right now
PAINVASMove the slider to the number that fits best today.
Sleep last night
SLEEPQTime you took your capsule
DOSETMAnything unusual today?
NOTEOptional
Build a diary in the free sandbox, open the participant link on your phone and run it as a participant would. Every feature, no credit card, no time limit.
Order of periods
With three pairs and the order within each pair randomized, there are 2 x 2 x 2 = 8 possible sequences (for example ABBAAB). Preparing the list is a design task: choose the sequences, decide whether the participant and clinician are blinded to them, and keep the key where only the people who need it can see it. The randomization list generator and the minimization randomization calculator are useful for building or checking lists, and how to blind a clinical trial covers supply and code handling.
In the data system the sequence must be visible at the right level. Staff who need to know which period a participant is in should see it; blinded roles should not see the treatment. Capture supports blinded and open-label randomization and blinded roles that never receive treatment-arm values. How a series of per-participant sequences maps onto its allocation settings is a point to confirm with the team when you scope the build.
Whatever tool you use, the property you protect is simple: every diary entry is attached to a period, and every period to a treatment, in a way that cannot be changed silently. If an entry is edited, the audit trail records who changed it, when, the old and new value, and the reason. If a period is cut short, the discontinuation is recorded rather than the data being overwritten.
Build steps
A realistic route for a single-site N-of-1 series.
Create a run-in and one visit per period and washout, with the expected length. This gives a casebook that mirrors the timeline and a clean place to hang period-level forms.
Choose one primary outcome and one or two supporting items. Keep it under a minute. Use the slider or linear scale for the primary measure and a choice question for adherence.
At the end of each period, a short staff-completed form records the dose actually taken, adverse events, concomitant medicines and the clinician's global impression. Use the adverse event and concomitant medication templates as a base.
Match reminders to the participant's routine. See the notes on ePRO compliance reminders and completion rates in diary completion rates.
Load or document the sequence, decide who is blinded, and test with a dummy participant that each role sees what it should and no more.
Have the clinician and a volunteer complete a week of diary entries before the first real participant. Fix wording and timing problems then.
Agree the shape with the statistician: one row per diary entry with participant, date, day number, period and treatment label.
After the data
The analysis of an N-of-1 trial is done on one person's series: compare the outcome in A periods with B periods, often using paired differences between the two periods of each pair or a model that allows for autocorrelation and trend in daily scores. Results are usually shown to the participant and clinician as a chart plus a short statement of which treatment looked better, with an interval. Series from several participants can be combined to estimate an average effect and the variation between people. Those methods belong to your statistician, in line with the principles of ICH E9 on statistical design and analysis.
What the platform must give your statistician is clean, labelled, time-ordered data. Capture exports CSV and Excel in wide format with an automatic data dictionary, can filter by date range and site, and produces SDTM datasets as SAS XPT with Define-XML. Ask for the period and treatment labels as explicit columns, and keep the dates and times of every entry. See export EDC data to R, SAS and SPSS for analysis-ready export patterns.
Plan for missing days. An N-of-1 series with 20 percent missing days in one period is harder to read than a complete one. Record missingness reasons where you can (illness, travel, forgot) with a simple choice field on the period check-in, rather than guessing later.
Diary entries
47 of 56
Missed days
9
Periods complete
3 of 6
Open queries
1
travel
in progress
Pitfalls
Choose washout from the treatment's pharmacology, or analyse only the later days of each period, and say which in the protocol.
Use the run-in diary to confirm baseline variability before randomizing periods.
Keep the daily form short and review completion weekly so a failing diary is caught early.
Record the dose and product at each period change on a staff form, and reconcile against the sequence.
Do not show treatment labels to blinded roles; check what each role sees during testing.
Many N-of-1 trials still need ethics approval and informed consent. See eConsent software and check with your ethics committee.
It is a trial in which a single participant receives two or more treatments in several randomized or alternating periods, with the same outcome measured throughout. The result is a comparison of the treatments within that person.
Chronic, stable conditions with symptoms that can be measured often, where the treatment starts and stops working quickly. Acute or progressive conditions and treatments with long-lasting effects are a poor fit.
Yes. A series of N-of-1 trials of the same design can be analysed together to estimate an average effect and between-person differences. The analysis is a statistical choice made in the plan.
The sequence is a design decision. You can prepare a list with a generator and use Capture for blinded or open-label allocation and period tracking. Confirm on a demo how your sequence design maps to the allocation settings.
For a daily entry, aim for under a minute: one primary measure and one or two supporting items. Long questionnaires are better placed at the end of a period.
Yes. ePRO runs in the browser from a link, so there is no app to install. Reminders are sent to participants on the schedule you set.
The CENT extension to CONSORT covers reporting of N-of-1 trials. Use it when writing the protocol and the report, along with your ethics committee's requirements.
It works for one participant or many. Check the pricing page for how study fees and participant fees apply to your design, and test the build free in the sandbox first.
Keep exploring
Crossover trial sample size calculator
Sizing a series of N-of-1 trials.
Randomization list generator
Prepare and check period sequences.
ePRO software for clinical trials
Diaries and questionnaires in the browser.
Patient diary software
Daily diaries and completion tracking.
EDC for rare disease trials
Small populations and single-participant designs.
How to blind a clinical trial
Supply, codes and roles.
Open the free sandbox, build a daily diary and a six-period casebook, and test the reminders and export before you enrol anyone. No credit card, and you pay only once you go live.