Logging side effects

Why free-text notes fail

The natural way to record side effects — write something down when something happens — produces a record with a specific, predictable defect: it contains only the notable days. There is no baseline, no denominator, and no way to tell whether nausea occurred on three days out of ninety or three days out of ten.

It also drifts in vocabulary. "Bit queasy", "nausea", "felt rough" and "stomach off" describe similar experiences and are stored as four unrelated strings, so nothing can be counted even in principle.

A structure that works

1. Choose three to five items at the start and keep them fixed. Adding a sixth mid-cycle is fine; changing the definition of an existing one is not.
2. Use one severity scale. A 0–5 scale where 0 means absent is enough, and recording the 0s is what makes the rest meaningful.
3. Record at a fixed time, so the sampling is consistent rather than triggered by how you feel.
4. Add a free-text field for anything unusual — but as a supplement to the structured items, never a replacement.
5. Timestamp everything. For weekly medications, the relationship to injection day is often the entire finding.

Choosing what to track

Pick items that are specific, observable and relevant to what you are actually taking. Two or three is usually better than eight, because an eight-item daily form gets abandoned in a fortnight.

• Prefer specific over general — "nausea" is more countable than "feeling off".
• Include at least one positive or neutral measure, so the record is not exclusively negative.
• For GLP-1 medications, gastrointestinal effects and appetite are commonly the items people and clinicians care about; ask what your prescriber wants to know before choosing.
• Keep an objective measure alongside, such as weight on a fixed schedule, so subjective and objective data can be read together.

Preparing it for a clinician

The purpose of all this structure is that a clinician can read it in thirty seconds. Before an appointment:

• Export the period since the last visit rather than the whole history.
• Bring the timeline of dose changes alongside the side-effect record, since the question is nearly always about the relationship between them.
• Highlight anything new since last time, and anything that resolved.
• Do not interpret. Bring the record; let the clinical reasoning happen in the room.

PepCycle exports to CSV and JSON, so a period can be handed over as a table rather than read aloud from a phone. See exporting your log.

What a self-log cannot establish

An uncontrolled personal record can show that two things co-occurred. It cannot establish that one caused the other, and expectation shapes self-reported outcomes strongly — particularly for subjective measures like mood, energy and sleep quality. That limitation is not a reason to skip the record; it is a reason to hand it to someone qualified to interpret it, and to be sceptical of your own conclusions.

Anything severe, sudden or worrying is a reason to seek medical attention rather than to add an entry.