Request for Feedback: Experimental Design for Blood Pressure and Breathing Experiments

Post with nicer formatting here

Summary

I'm trying to identify causes and ways of reducing my elevated blood pressure and am looking for feedback on my experimental design & protocols.

Studies/Experiments:

  • Phase 1: Identify Potential Causes of Elevated Blood Pressure from Existing Self-Tracking Data
    • Approach (details below)
      • Use a mixed-effect model to look for look for significant correlations in data I've already collected.
      • If I find anything promising, design additional studies to confirm the relationship and test interventions.
    • Metrics to look at
      • blood glucose, sleep, exercise, weight/body, pulse, HRV
  • Phase 2: Testing Deep Breathing to Lower Blood Pressure
    • Approach (details below)
      • Measure blood pressure and pulse before & after the most well studied protocols as well as normal breathing.
      • If any protocols show significant reduction in blood pressure, optimize the protocol and design/execute an experiment to test the long term effect.
    • Analysis
      • Student's t-test will be used to test if the blood pressure change for any of the protocols is different from that of normal breathing.

Questions:

  • Phase 1
    • Any other metrics I should be looking at?
    • Does this analytical approach seem reasonable? Are there different statistical approaches I should be taking (details below)?
  • Phase 2
    • Has anyone tried this? If so, what breathing protocols have worked for you?
    • Any suggestions for other interventions to try?
    • Any comments or critiques of the experimental design or analysis?
    • Anything else I should be measuring while doing this?

It would significantly improve these studies to have a larger number of participants. If you’re interested in collaborating on this or other scientifically rigorous self-experiments with blood pressure, low-carb foods, supplements, or other health interventions, please let me know in the comments or via the contact form on the right.


Details

Purpose

  • To identify environmental or controllable factors that have a significant impact on my blood pressure.
  • To quantify the effect of known interventions for reducing blood pressure.
  • To find a set of interventions that enable me to reduce my blood pressure below 120/80 mmHg.

Background

Figure 1. Weekly average of blood pressure as measured by Omron home blood pressure monitors.

I've been measuring my blood pressure over the past 4 months and it's consistently over the American Heart Association target of 120/80 mmHg for "Normal" blood pressure. Of more concern, I frequently measure Systolic blood pressure of >130 mmHg, which is considered Stage 1 Hypertension.

Elevated blood pressure is associated with an increased risk of cardiovascular disease (41.5/100k person years, hazard ratio 1.14 vs. normal BP, see Figure 2 and Table 1).

Figure 2. Cumulative incidence of cardiovascular disease vs. time for different blood pressure groups from a study of the South Korean nationwide health screening database (6.4M participants).
Table 1. Rate of cardiovascular disease for different blood pressure groups from a study of the South Korean nationwide health screening database (6.4M participants).

Given this, I'd like to see if I can reduce my blood pressure and reduce the strain on my heart and circulatory system.

There are numerous medications that lower blood pressure, but all risk of side effects. Before I pursue that route, I'd like to better understand the cause of my elevated blood pressure and see if any diet or lifestyle interventions can ameliorate it.

As mentioned above, I've been measuring my blood pressure for the past 4 months, along with blood glucose, sleep, weight, and exercise. This provides a (hopefully) rich dataset for identifying environmental or lifestyle factors that influence my blood pressure. Notably, I've noticed that my blood pressure is elevated on days after I've had low blood sugar the night before, indicating a possible effect (no statistical analysis done).

From an American Hearth Association evaluation of methods non-medication approaches to reduce blood pressure, with the exception of aerobic exercise (which I already do), the most well evidenced methods of reducing blood pressure are meditation and deep breathing.


Proposed Experiments

Phase 1: Identify Potential Causes of Elevated Blood Pressure from Existing Self-Tracking Data

  • Data
    • Blood pressure:
      • systolic and diastolic blood pressure
      • Measured by Omron Evolve
    • Glucose:
      • Same day: fasting BG
      • Previous day: average BG, time low (70, 60, & 50), time high (120, 140, 160), & coefficient of variation
      • Previous evening (after 7p): same as previous day
      • Measured by Dexcom G6
    • Sleep:
      • Time asleep, number of wake-ups, early rising (time woke before alarm)
      • Measured manually and by Apple Watch (less reliable but more data)
    • Other heart markers:
      • pulse (sleeping, morning, and awake), heart rate variability
      • Measured by Apple Watch and Omron Evolve
    • Body:
    • Exercise:
      • Type of exercise the previous day (aerobic vs. strength training) and frequency of aerobic exercise
      • Manually recorded
  • Analysis
    • A mixed effect model will be used to calculate the effect size, standard error, and p-value for the correlation between each metric and systolic and diastolic blood pressure
    • Effects will be of significant magnitude if a reduction of 5 mmHg can be achieved via a practical variation in the correlating metric.
    • Given the large number of metrics being looked at, I will use p-value thresholds of:
      • 0.02 for planning testing interventions
      • 0.05 for follow up experiments to confirm the correlation
      • 0.1 for further monitoring/assessment as I get more data
  • Questions
    • Any other metrics I should be looking at?
    • Does this analysis seem reasonable? Are there different statistical approaches I should be taking?

Phase 2: Testing Deep Breathing to Lower Blood Pressure

  • Background
    • Numerous studies, reviews, and meta-analyses have shown deep breathing to lower blood pressure in both the short and long-term (example 1, example 2).
    • Effect sizes are moderate (3-5 mmHg) and statistically significant for large patient populations (>10,000 patients in some studies).
    • Numerous breathing protocols have been tested, with varying results.
  • Approach
    • Measure blood pressure and pulse before & after the most well studied protocols as well as normal breathing.
    • For each protocol, measure at least three times. If the protocol shows a reduction in blood pressure, measure an additional 5 times to confirm.
    • Conduct measurements 1/day in the mornings.
    • If any protocols show significant reduction in blood pressure, optimize the protocol and design/execute an experiment to test the long term effect.
  • Measurement
    • Blood pressure and pulse will be measured with an Omron Evolve.
  • Analysis
    • Student's t-test will be used to test if the blood pressure change for any of the protocols is different from that of normal breathing.
  • Questions
    • Has anyone tried this? If so, what breathing protocols have worked for you?
    • Any suggestions for other interventions to try?
    • Any comments or critiques of the experimental design or analysis?
    • Anything else I should be measuring while doing this?

Thanks in advance for your comments & feedback!


- QD

Great post, I’ll think more, but a couple of details that might be useful.

I’ll try to find a written description, but I remember seeing a Seth Roberts talk in which he showed comparisons of 3 different blood pressure monitors. He made daily measurements with all three and did some kind of averaging. There was quite a bit of noise in the data, and he wanted high confidence in small changes over time. One of the things he found, if I remember correctly, is that there is some drift, so he needed three to tell him if one was drifting. I think there are a number of ways to deal with this problem, but worth considering.

I’ve done quite a bit of blood pressure monitoring and found that I could affect the measure by sitting still for longer and being calmer. I typically took 3 measures with a Homedics arm band and averaged the result. At first my blood pressure mostly went down over the three measures. Then I started to sit for longer and this effect went away.

Interesting. Adding multiple measurements is fairly straightforward. Multiple monitors would be difficult as I need to do these measurements after I get to work and don’t want to carry them all around. I could do a multi-monitor test on weekends though to check for drift.

Dietary sodium, potassium, or the ratio of sodium/potassium are all well-known to influence BP, and some people are much more sensitive than others. This might be a high yield input to track; though not easy to do with high accuracy, there’s such a wide range in various diets (many grams of sodium in a restaurant/processed diet, vs 1-2 grams per day easily achievable with a more careful diet.)

Another thing to think about, making it more difficult to do your n=1 study but ultimately allowing more confidence in the results, would be blinding. That is, measure the BP but you don’t get to see the results. Either your device records it or somebody else puts it in your database, and you don’t get to look at the results until the end of your pre-planned experimental protocol.

Yeah, I’d really like to blind it, but I have to look at the blood pressure cuff to record the data and don’t have anyone else around that can do it for me. What I can/will do is not analyze the data until the end. Not perfect as I get a sense of the result, but better than nothing.

You might want to include the amount of time you spend reading/watching politics and/or news generally, whether on TV or other screens.

Social media is also something that some people track, although I find that most of the time that I spend on Facebook raises my spirits (surprisingly, but it’s because I mostly interact with actual friends there), and time on Twitter makes me less happy (because I almost exclusively see strangers hating on each other there).

I would also suggest you to track weather, temperature and air pressure.

https://www.ahajournals.org/doi/pdf/10.1161/HYPERTENSIONAHA.111.00686

That’s wild. I would not have expected that effect at all. Easy enough to pull the data and see if it has an effect. I’ve probably got enough historical data to check it now…

Please post your findings. I’m very interested in this, my personal observations show that my blood pressure is strongly correlated with changes in air pressure and duration of sleep (and physical activity but that is pretty much expected).

Will do. I’m running a clinical trial for my job this weekend, but will have time next weekend to do the analysis. Will post once I have it.

Stress increases blood pressure. You could look at other measures of stress such as cortisol levels or skin conductivity.

Mindful meditation before BP measurement may clear some external stressors from your mind and give a better result on the effect of other factors.

Salt content of diet is certainly a factor. Are there other food categories that your body is sensitive to that could increase blood pressure.

Thanks for the suggestions. All those seem worth testing. Will need to think about how to modulate salt, as I don’t have an easy way to measure it. Should be able to make qualitative changes that would significantly reduce it though. I’ve also read that potassium supplementing can reduce BP, so I’ll try potassium/sodium substitution as well.

It’s a fantastic design. Personally I have been measuring my blood pressure for 6 months and comparing results regarding whether I take the full doses of medication or only a part of these, mainly because I forget to take them. My diagnosis is very clear, I must lose at least 20 kilos to regulate my blood pressure, in the 6 months that I have been running this activity I have not dropped even 1 gr.
Regarding the design, I do not know if the speed exercises are of any use, my intuition tells me that breathing exercises can regulate the pressure at the time it is high, but I doubt it is a long-term solution. I’m not a doctor, but as far as I know blood pressure is strongly influenced by the levels of fat that exist in the veins. Glucose is a good parameter to measure because normally overweight in addition to having complications with blood pressure can also affect with diabetes, and exercise could also be a good indicator to be able to reduce blood pressure, something in which I have also failed because something exercises one day but I am not constant mainly because I have very little time available. If you are taking medications it would be good to record the amount of medications you are taking for blood pressure, you could also record the number of glasses of water you are consuming daily, I do not know if oxygen levels have any effect on this but if the Apple watch measures this parameter it would be good to consider it within the study.