How to Track Whether a New Antidepressant Is Helping

Week three. You wake up and run the same audit you ran yesterday: is this better? Some days the answer is obvious. Most days your memory of last Tuesday has already been rewritten by how this morning feels. Here is a way to answer the question with a record instead of a mood.

One thing before the rest. This is about observing and reporting, not about medication decisions, and it names no drug and no doses. Every decision about what you take, how much, and for how long belongs to the person who prescribed it, and nothing here is a reason to change, stop or skip anything without talking to them first.

Why this is so hard to judge from memory

You are being asked to compare this week against a week you never measured, using an instrument that quietly rewrites itself.

Ben-Zeev and Young ran that comparison in the Journal of Nervous and Mental Disease in 2010. They collected momentary symptom reports across a week, then asked the same people to summarize it afterwards. The summaries did not match the week. Recall was biased in both directions: some symptoms came back worse in memory than people had reported at the time, others came back milder.

Two more problems are specific to your situation. You have no record of the four weeks before you started, so there is no baseline. And your life did not hold still while you waited.

So it becomes a question for a written record rather than a Tuesday morning impression.

What the timelines actually look like

The uncertainty has a shape, and it lasts longer than most people expect. The NHS puts it plainly: antidepressants usually take 1 to 2 weeks to start having an effect and can take up to 8 weeks to work fully. So the stretch where you genuinely cannot tell is not evidence that something has gone wrong. It is the normal middle of the process.

One finding about that window is worth knowing. Szegedi and colleagues pooled 41 trials and 6,562 patients in the Journal of Clinical Psychiatry in 2009. Early improvement meant at least a 20% drop on a clinician-rated depression scale within the first two weeks. It predicted later stable response with sensitivity of at least 81%, and stable remission with at least 87%. Negative predictive values ran from 82% to 100%.

The caveat matters more than the numbers. Positive predictive values in the same analysis were much weaker, between 19% and 60%. So the absence of early movement carried more information than its presence, and neither is a forecast for any individual. These are group patterns from trials, measured with clinician-administered scales. The authors were writing about when clinicians should reconsider a plan, not about what you should conclude from a chart on your phone.

What is worth writing down, and what is noise

The goal is a record another person can read in 90 seconds. Four items, not 14, the same four every day.

Sleep, appetite and energy, before mood

Clinicians often describe these as the things that shift before mood does. I am not going to dress that up as a finding, because what gets repeated online is a rule of thumb. Track them anyway, for a duller reason: they are concrete. You can rate last night's sleep the same way on a Monday and on a Friday. "How am I doing" gets a different answer at 8am and at 11pm, which makes a month of it almost unreadable.

The activity test

What are you doing this week that you were not doing a month ago? Answered a message that had been sitting there for two weeks. Cooked instead of ordering, three times. Went to the thing you would normally have canceled. Write the fact and skip the interpretation. Behavior is harder to argue with than a self-assessment, and it is the part hardest to get out of a questionnaire.

Side effects, with dates

Keep this as a separate dated list, not a feeling folded into your daily number. What appeared, roughly when, roughly how strong, and whether it faded. The NHS notes that many side effects ease after a couple of weeks while some carry on. Dates separate a side effect that showed up in the first days and went away from one that arrived in week five and is still here. Those are two different conversations, and both belong to your prescriber.

One line of context

A week with a house move in it is not the same week as the one before it. One short line a day is enough: deadline Thursday, ill Monday to Wednesday, saw family. Without it, four weeks of ratings is a chart with no legend, and you will read your worst week as evidence about the medication when it was evidence about February.

A simple two minute record

Same time, same scale, same items.

1. Sleep last night, on a 1 to 10 scale.
2. Energy today, same scale.
3. The one or two things you and your prescriber actually talked about, same scale.
4. One line of context.
5. Separately, any new side effect, with the date it started.

That is the whole method. It is not a symptom questionnaire, it produces no score, and it diagnoses nothing. Its only job is to survive four weeks and still be readable at the end. If a record like this has collapsed on you before, usually somewhere in week three, that failure mode is well understood and fixable: tracking that survives past week two.

One borrowed rule needs an amendment. When you test a coping technique, the one variable rule says change one thing at a time so you can read the result. That still holds. The difference is that with medication you are not the one who changes the variable. You record. Your prescriber decides what moves.

Wellbeing check-ins
Rate sleep, energy and the two things you are watching in under a minute, at the same time every day. Try it free in MindSync →

How to turn four weeks into one page for your appointment

Follow-up appointments are short. Bring the record and read it in this order.

1. The headline, in one sentence. Sleep is better, mood is about the same, the nausea stopped in week two.
2. The shape of the numbers. Not every reading, just the direction: what a bad day looked like at the start, and what it looks like now.
3. The behavior evidence. Two or three concrete things you are doing, or still not doing.
4. The dated side effect list, read out as dates.
5. Your questions. Write them down beforehand, because they evaporate in the room.

Decisions made against measurements land differently from decisions made against impressions. Guo and colleagues ran a randomized controlled trial with blind raters in the American Journal of Psychiatry in 2015: 120 outpatients with moderate to severe depression, 24 weeks, the same limited set of medication options in both arms. What differed was that in one arm the treatment decisions were guided by scheduled symptom and side effect ratings. Response rates came out at 86.9% against 62.7%, remission at 73.8% against 28.8%.

Now the caveat. That trial measured clinicians measuring, in one setting, under a protocol. It is not evidence that your own notebook improves your outcome, and I would rather say so than let the number do work it cannot do. What it supports is narrower and still worth something: a record changes the conversation the decision gets made in. The page is material for that conversation. It is not a basis for changing a dose.

MindSync has no medication module and does not want one. What it has is the boring half of this: a one minute rating, the same items every day, and weeks lined up next to each other. If that is the part you need, you can see exactly what the app does.

Weekly & monthly summaries
A weekly summary turns scattered entries into something you can read out at a follow up appointment. Try it free in MindSync →

The honest caveats

A record cannot separate the medication from the month. Symptoms fluctuate on their own, expectation does real work, and some of what you write down would have happened anyway. Nobody untangles that from one person's four weeks. The record does not solve it. It means the judgment gets made against something rather than nothing.

There is also a cost to the measuring itself. Rating your mood every morning means thinking about your mood every morning. For some people that is clarifying. For others it becomes one more loop. If daily makes things worse, weekly is a legitimate choice, and a weekly record still beats memory.

Sometimes the first thing tried is not the thing that works. In the STAR*D trial, reported by Rush and colleagues in the American Journal of Psychiatry in 2006, 36.8% of participants reached remission at the first treatment step, and 67% did so cumulatively across up to four successive steps. Read that as the ordinary shape of the process rather than a verdict on you. It is also why the record matters: the second conversation goes better when someone can see the first four weeks.

People ask the same question about the other half of their treatment, and there is a longer piece on the same question, asked about therapy.

Whatever your four weeks say, the decision that follows is not yours to make alone. Do not change, stop or skip anything without talking to the person who prescribed it.

Four weeks you can actually look back on

• Mood & focus-area trends - Weeks charted next to each other, instead of a memory of yesterday.
• Therapy timeline - Every note and check-in in one place, with the dates still attached.
• Wellbeing check-ins - The two minute record that makes the rest of it possible.

Frequently asked questions

How long do antidepressants take to work?

The NHS says antidepressants usually take 1 to 2 weeks to start having an effect and can take up to 8 weeks to work fully. That is a general range, not a schedule, and it varies between people. The practical consequence is that the first few weeks are not a fair test. A written record beats a daily verdict from memory.

How do I know if my antidepressant is working?

Compare a record, not a memory. Rate the same two or three things at the same time each day, note what you are doing now that you were not doing before, and keep a dated list of side effects. Then take it to the person who prescribed it, because interpreting it is their job.

What should I track when starting a new medication?

Four things, kept small enough to survive a bad week: sleep, energy, the one or two things you and your prescriber agreed to watch, and one line of daily context. Keep side effects on a separate list, with the date each one started and whether it faded. A short record is the one you are still keeping in week four.

What if I do not notice anything after a few weeks?

Raise it rather than deciding on your own. No change in the early weeks is information, and the person who prescribed it needs that information early, along with your dated notes on side effects. It is not a reason to change, stop or skip anything by yourself. If you are unsure, asking for the appointment sooner is reasonable.

What should I tell my doctor at the follow up appointment?

Lead with one sentence summarizing the four weeks, then the direction your ratings moved, then two or three concrete things you are or are not doing now, then the dated side effect list, then your questions. That order gets the useful part out before the appointment runs out.