Results from my first experiment with a Breathalyzer

Sep 3, 2017 · 5 min read
Image for post
Image for post
Figure 1: Changes in blood alcohol content (BAC) estimated by a BACTrack Breathalyzer after consumption of Fremont Brewery Lush IPA. The beer was consumed at 8:40–9:10PM. 2 measurements were taken at each time point and the average values are plotted above. The graphs were plotted by Benjamin Mako Hill using ggplot2 (thank you!)

The details of the experiment can be found in my previous story but here’s the summary:

  1. I had 1 can of Fremont Brewery Lush IPA ABV 7% over 30 minutes at 8:40–9:10PM every day for 5 days this week.
  2. I used BACTrack Breathalyzer S80 to measure BAC at 20, 50, 80, and 110 or 120 minutes after I finished my drink. Two measurements at each time point. No food or drinks were consumed for 20 minutes before each measurement (following the BACTrack manual).
  3. My housemates wrote down the values and I did not see them until the last measurement on Friday.

Basically, I wanted to measure my “normal” variation in alcohol metabolism. I tried my best to not change my behaviors but I feel like I did experience Hawthorne effect as I was intentionally drinking a little more water throughout the week this week (I was hypothesizing hydration was one of the key risk factor, which was wrong, it seems).

Main observations

Anyways, I did see some variation! A couple things I noticed after plotting these points were:

  • There was some variation in the initial BAC readings: lowest on Thursday at 0.028% and highest on Wednesday at 0.052%.

I was making guesses about my BAC during the week but I was pretty wrong! I guessed that Wednesday (08/30) was lower than Monday or Tuesday (08/28 and 08/29). This is perhaps one good reason to not go by one’s feeling when deciding whether to drive or not…

I think I have a pretty good idea about why my BAC was so low on Thursday. I came home early and ate a lot of snacks including a seitan tamale, vegetarian fish cake, and peanuts for about an hour before 8:40PM. I think having food in my GI system when I started drinking slowed down/cut the absorption of the alcohol. I did eat food and had water throughout the evening every night, except for 20 minutes prior to each measurement (so on Monday, when I measured BAC every 10 minutes, I didn’t eat or drink for 80 minutes). But the food and water intake after I had the beer don’t seem to have influenced the metabolism very much.

I was not feeling so peachy during the day on Wednesday and Friday but it’s not clear if that contributed to the high value on Wednesday.

  • The slopes seem to be consistent across days and linear, meaning the metabolism rates are similar no matter how high or low my initial readings were. Benjamin Mako Hill used z-test to test for a difference in slopes between days and found no evidence:
Image for post
Image for post
Table 1: Pairwise comparison of the slopes between two dates (by Benjamin Mako Hill). Δβ is the difference between slopes reflecting βs in a linear regression model measured in (BAC x 1000)/hour.

When my initial BAC was low on Thursday, my metabolism rate was the same as other days. One potential explanation is that not as much alcohol got absorbed because the alcohol got soaked up by some of the food and since food absorption is not 100%, I may have just pooped out some of the alcohol. My theory before seeing this data was that food may slow down the alcohol absorption but I may be absorbing alcohol for longer at lower levels, which would result in a different slope. But that may not be correct.

From the linear regression model with time since finishing the drink, dates, and interaction terms as independent variables, I got 0.0003% BAC/minute as my rate of decrease in BAC, which is the same as 0.018%/hour. Also, the minutes to reaching 0.00 ranged from 110 minutes (actual time) to 177 minutes (estimated from the model), so that is more than an hour difference.

Measurement errors

Even though this was a very short experiment with few moving parts but there were lots of (potential) measurement errors! You may have already noticed some missing points and varying time points for measurement on the graph. To list a few:

  • Time between measurements was not that accurate, which could influence the slopes. I used a timer set for 20, 30, or 40 minutes and depending on when I reset the timer, I introduced a few seconds to minutes to the time between measurements. I should use the alarm clock instead next time (i.e. 9:30PM, 10PM etc). At one point, I measured a couple minutes early intentionally as it was weird to not eat while everyone else at the table was eating…
  • Not all time points had 2 measurements, although when I successfully measured 2, the values were fairly consistent (typically the first measurement was 0.001 higher than the second), so this may have little impact on the results. I was also in different postures (standing, walking, or sitting), which may have affected the measurements.
  • Of course, the machine accuracy is not high (+/-0.01 @ 0.05 according to the manufacturer), so it’s possible that 0.028 on Thursday might have been lower than the actual BAC (i.e. maybe the machine tends to measure lower values on cooler days). But the consistency between repeated measurements and among slopes on different days make me feel that the trends I saw were not just artifacts but I wouldn’t know for sure. Getting better tools would be the obvious answer but for now, I’m happy with the current setup.

Future plans

All in all, this was fun! I look forward to doing it again. I have a couple things that I want to test next:

  • Maybe eating a bowl of rice before starting to drink is good enough to cut the absorption? I’ll eat a bowl of rice before starting to drink and see if my BAC levels will be low (~0.03 at initial measurement) and consistent throughout the week. Repeat with 100g of seitan, 1 bowl of lettuce salad, or a couple slices of cheese and see if food groups matter.
  • I often feel less tipsy if I bike for 10–20 minutes after drinking. I want to test that.

I’ve uploaded the dataset and R code for those who are curious. Comments and suggestions for future experiments are welcome.

The hardest thing about this experiment was that I sometimes wanted to have other foamy drinks (i.e. midori sour) or have more than one beer. Pat on my shoulder, kisses on my cheeks for making it through the week.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch

Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore

Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store