I have a chronic condition that requires daily medication. And until about four months ago, my approach to tracking it was a notes app, a vague memory of whether I'd taken something that morning and the occasional panic at 10pm.

That's not good enough. So I decided to get serious about it.

I wanted to know, with actual data, how consistent I really was and whether my consistency correlated with the health metrics my Apple Watch was quietly collecting. I gave myself 90 days to find out.

What follows is an honest account of what I tracked, what the data showed and what actually changed as a result.

The setup

I'm on two daily medications: one taken in the morning, one in the evening. For privacy I'll leave out the specifics, but they're both time-sensitive, meaning there's a narrow window where taking them is most effective.

For health metrics I have about 18 months of Apple Watch data, including resting heart rate, heart rate variability, sleep duration and respiratory rate. I also have blood pressure data from a Withings cuff I'd been using inconsistently.

The plan was simple. Log every dose, mark it taken or skipped, note the time and let the data accumulate. I'd check in at the four-week and eight-week marks, then do a full review at 90 days.

I started on a Monday because of course I did.

Week one: more chaos than I expected

I assumed I was a reasonably adherent person. The first week corrected that assumption.

Monday through Wednesday went fine. Thursday I was in back-to-back meetings until 7pm and completely forgot my evening dose. Friday was fine. Saturday I took the morning dose at 1pm because I'd slept in and the whole routine was off. Sunday I forgot the evening dose again.

Five out of a possible eight doses in the first four days. Sixty-three percent. For someone who thought they were pretty consistent, that was a wake-up call.

The reminders helped but only when my phone wasn't on silent. The real problem was that taking medication had never been a deliberate habit, it had always been a vague intention that sometimes happened and sometimes didn't. Making it visible, with a log I had to actually update, changed how I thought about it.

Week two: the routine starts forming

Something shifted in week two. I started thinking about the doses proactively rather than reactively waiting for a notification. I put my morning medication on the kitchen bench next to the kettle. I moved my evening reminder 30 minutes earlier so I'd have a buffer.

By the end of week two I was at 86% adherence for the fortnight. Not perfect, but the trend was moving in the right direction.

More interestingly, I started noticing the times I was skipping. It was almost always the evening dose. And it was almost always on Friday or Saturday.

Month one: patterns start to emerge

After 30 days the picture was clearer. My morning adherence was sitting at 91%. My evening adherence was 76%. That's a 15-percentage-point gap between two doses I was supposed to be taking with equal consistency.

The weekend effect was real. Monday through Thursday I averaged 93% adherence. Friday through Sunday I averaged 71%. The disruption to routine was the culprit, not laziness or forgetfulness in any meaningful sense. I just didn't have the same environmental cues on weekends.

I made one change: I set a second reminder for the evening dose on Fridays and Saturdays, 90 minutes after the first. If I confirmed the first one the second wouldn't fire. If I didn't, I'd get a nudge.

That single change lifted my weekend evening adherence from 68% to 84% in the following four weeks.

Month two: the health correlations get interesting

By month two I had enough data for patterns to become meaningful rather than just suggestive. This is where having Apple Health data alongside the adherence log started earning its keep.

I looked at my resting heart rate across weeks with high adherence (90%+) versus weeks with lower adherence (below 75%). The difference was small but consistent: my resting heart rate averaged 2.4 bpm lower in high-adherence weeks. Over a single week that's noise. Across eight weeks of data it starts to look like a signal.

Sleep was more interesting. In weeks where I missed multiple evening doses, my average sleep duration was about 23 minutes shorter and my HRV was noticeably lower. Correlation is not causation, and I'm aware the relationship likely runs both ways: poor sleep makes it easier to forget doses, and missed doses may affect sleep quality. But seeing the numbers together made the relationship feel concrete rather than theoretical.

I also noticed something about my mood entries. On days following a skipped evening dose, my journal check-ins were more likely to note feeling foggy or flat in the morning. Again, not a controlled experiment. But consistent enough to be worth paying attention to.

Month three: taking real data to a real appointment

At the 60-day mark I had a scheduled check-in with my GP. I went in with a printed summary of my adherence data and a few screenshots of the correlation graphs.

The conversation was different to any I'd had before. Instead of "I think I've been pretty good" I could say "My adherence was 84% for the past two months, with morning doses significantly more consistent than evening. I had two weeks in December where I dropped below 75% and here's what was happening." My doctor found it useful. We talked about whether the timing of the evening dose could be adjusted. We discussed the resting heart rate correlation. It was a more productive 15 minutes than most of the appointments I'd had in the previous few years.

The streak psychology is real and I didn't expect that

I want to be honest about something slightly embarrassing. The streak counter worked on me.

Around day 45 I hit a 12-day perfect adherence streak. I knew, intellectually, that it was a simple number in a database. But I did not want to break it. When I was at a dinner party on day 47 and realised I'd forgotten my evening dose, I excused myself to take it rather than just writing off the day. The streak made the abstract cost of a missed dose feel concrete.

At 90 days I'd had three perfect weeks, which would have seemed improbable to me at the start. The longest streak was 19 days. I did not achieve perfect adherence overall, and I'm not sure that's a realistic goal for anyone living a normal life. But the gamification, for want of a better word, shifted my default from "I'll try to remember" to "I actively don't want to break this run." If you want to understand the science behind why this works, our guide on building a medication habit covers the psychology of streaks, habit loops and loss aversion.

Key lessons from 90 days of medication tracking

A few things I now believe that I didn't before I started:

Weekends are harder than weekdays, and it's not about motivation. Routine is the mechanism. When the routine changes the habits built on top of it become fragile. The fix is building weekend-specific habits or cues, not just trying harder. (If your routine varies daily rather than just at weekends, interval dosing might be a better fit.)

Morning doses are structurally easier than evening doses. Morning routines tend to be more consistent. The kitchen, the kettle, the same sequence of actions. Evenings are more variable. If you're on a medication where timing flexibility exists, earlier in the day is generally more forgiving.

The data changes the conversation with your doctor. This one I underestimated completely. Having a log transforms "how have you been going with your medication?" from a question that produces vague reassurance into a question with a real answer.

Imperfect adherence is the norm, not a personal failure. My 90-day average was 87%. I had bad weeks. I had a run of four days in late January where I was travelling and barely held it together. The point of tracking isn't to feel guilty about the gaps; it's to understand them well enough to reduce them.

Health data correlations are motivating in ways that health advice isn't. Knowing I should take my medication consistently is abstract. Seeing that my resting heart rate was 2.4 bpm lower in high-adherence weeks is concrete. The data provides a feedback loop that lectures and reminders alone don't. This is exactly what the Insights dashboard is built to surface.

Where I ended up

At 90 days I'm still using Cadence every day. My overall adherence for the period was 87%, with a clear upward trend as I refined my routines. Month three was my best month at 91%.

More importantly, I understand my patterns now. I know where I'm vulnerable (Friday evenings, travel weeks, late nights when the routine collapses). I know what interventions actually help (the second weekend reminder, the physical cue of medication next to the kettle). And I have a record I can bring to medical appointments that adds something to the conversation.

If you're managing a chronic condition, or even just trying to be more consistent with supplements or preventive care, I'd genuinely encourage starting a log. It doesn't have to be Cadence. But the act of making adherence visible changes how you relate to it.

The data is more honest than your memory. And usually more useful.

If you want to see how Apple Health data connects to your medication habits, read our guide to tracking medications with Apple Health. And if cost is a concern, here's an honest comparison of medication apps that don't require a subscription. For the science behind why consistency matters so much, see our guide on medication adherence. And for practical strategies I wish I'd known earlier, here's how to remember to take your medication every day.

Cadence is free to download on iPhone. Cadence Pro, which includes Apple Health integration, detailed insights and correlation tracking, is a one-time purchase of $9.99. No account required and no subscription.