05-14-2016 20:16
05-14-2016 20:16
05-16-2016 13:13
05-16-2016 13:13
With respect to the 220 - age formula, the following study shows a SEE estimate of 12.4 for the Fox formulate (220 - age) and 11.4 bpm for Tanaka (208 – 0.7 × age):
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3935487/
220 - age is 10 bpm off for me. At 183 bpm, that's a 5% error.
05-16-2016 14:45 - edited 05-16-2016 14:46
05-16-2016 14:45 - edited 05-16-2016 14:46
Regarding the accuracy of BMI, most studies seem to focus on specific subsets of populations -- I guess because this is the simplest way to use existing data. The accuracy is going to vary by demographic ... I think.
This study shows 83% of individuals were accurately identified, 6.5% were falsely classified as overweight and 10.5% were falsely classified as normal weight. These were 448 males between 18 - 20.
http://www.ncbi.nlm.nih.gov/pubmed/18419029
This study is quite large and concludes: A BMI ≥ 30 had a high specificity (95% in men and 99% in women), but a poor sensitivity (36% and 49 %, respectively) to detect BF %-defined obesity.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2877506/
If your complaint is that BMI falsely identifies a lot of people as obese, that doesn't appear to be justified ... it appears to have a high specificity. The real problem with BMI is that it misses a lot of obese people (poor sensitivity).
Interestingly enough, I've never heard someone on the forum complain that BMI failed to identify their obesity.
05-16-2016 17:20 - edited 05-16-2016 17:25
05-16-2016 17:20 - edited 05-16-2016 17:25
@FitBeforeFifty wrote:With respect to the 220 - age formula, the following study shows a SEE estimate of 12.4 for the Fox formulate (220 - age) and 11.4 bpm for Tanaka (208 – 0.7 × age):
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3935487/
220 - age is 10 bpm off for me. At 183 bpm, that's a 5% error.
The study you cited seems to support what I've been saying (even though the study was for a pretty narrow cross section of individuals):
Conclusion:
Our findings show that based on the SEE, the prevailing age-based estimated HRmax equations do not precisely predict an individual’s measured-HRmax.
Did I miss something?
Regarding the 220-Age formula for me, while I've never been formally tested (at least not in recent years), I can get my heart rate up into the low 180s when doing repeat quarters (at least so says my Polar chest strap), and yet my calculated HRmax is 161 which works out to a 13% error.
05-16-2016 17:30 - edited 05-16-2016 17:31
05-16-2016 17:30 - edited 05-16-2016 17:31
@FitBeforeFifty wrote:Regarding the accuracy of BMI, most studies seem to focus on specific subsets of populations -- I guess because this is the simplest way to use existing data. The accuracy is going to vary by demographic ... I think.
This study shows 83% of individuals were accurately identified, 6.5% were falsely classified as overweight and 10.5% were falsely classified as normal weight. These were 448 males between 18 - 20.
http://www.ncbi.nlm.nih.gov/pubmed/18419029
This study is quite large and concludes: A BMI ≥ 30 had a high specificity (95% in men and 99% in women), but a poor sensitivity (36% and 49 %, respectively) to detect BF %-defined obesity.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2877506/
If your complaint is that BMI falsely identifies a lot of people as obese, that doesn't appear to be justified ... it appears to have a high specificity. The real problem with BMI is that it misses a lot of obese people (poor sensitivity).
Interestingly enough, I've never heard someone on the forum complain that BMI failed to identify their obesity.
The issue isn't failure to identify obesity, it is identifying obesity when it doesn't exist. To my way of thinking, using young men as a study group is decidedly biased. Why? In my experience, young men have not yet fully developed from a muscular perspective, heck, I would have been classified as "normal" back when I was that age; I carry quite a bit more muscle now then I did 40 years ago (and a little more fat too), now I am officially "obese" in spite of the fact I am anything but. That applies not only to me, but to pretty much the entire group of seniors I run with.
Even still, I wouldn't exactly call an 83% accuracy rating very good.
05-17-2016 05:11
05-17-2016 05:11
@shipo wrote:The study you cited seems to support what I've been saying (even though the study was for a pretty narrow cross section of individuals):
Conclusion:
Our findings show that based on the SEE, the prevailing age-based estimated HRmax equations do not precisely predict an individual’s measured-HRmax.
Did I miss something?
Yes, this study proves your point and mine. Your point is the formula isn't precise. My point is it is a pretty good estimate.
You want a precise predictor. That's great. One doesn't exist, so we use the best predictor we have. None of us have said the formula is precise/perfect, we've only said it's a reasonable estimate and the best we have.
All models are wrong, some models are useful. Apparently your criteria for useful is very low error rate and works for you. Your opinion is if it isn't precise it isn't useful, so therefor people shouldn't use HR if they haven't measured their max HR. This is definitely throwing the baby out with the bath water.
If you find a better formula let us know. Estimates based on static measurements like age and height are always going to be non-precise. The newer formula 208 - 0.7 * age is only slightly more accurate and for you still predicts 166 vs. 161. Better, but still not precise for you.
05-17-2016 05:21
05-17-2016 05:21
@shipo wrote:
The issue isn't failure to identify obesity, it is identifying obesity when it doesn't exist. To my way of thinking, using young men as a study group is decidedly biased. ...
Even still, I wouldn't exactly call an 83% accuracy rating very good.
I agree the first study isn't very diverse (young males), that's why I provided 2 studies. The second one is much more diverse: A cross-sectional design of 13,601 subjects (age 20–79.9 years; 48% men) from the Third National Health and Nutrition Examination Survey
In both studies, the number of false positives are relatively low. The second study had a specificity of 95% in men and 99% in women. In other words, BMI doesn't falsely identify a lot of people as obese. Some, yes. A lot, no.
05-17-2016 06:09
05-17-2016 06:09
@FitBeforeFifty wrote:You want a precise predictor. That's great. One doesn't exist, so we use the best predictor we have. None of us have said the formula is precise/perfect, we've only said it's a reasonable estimate and the best we have.
All models are wrong, some models are useful. Apparently your criteria for useful is very low error rate and works for you. Your opinion is if it isn't precise it isn't useful, so therefor people shouldn't use HR if they haven't measured their max HR. This is definitely throwing the baby out with the bath water.
While you may not present the 220-Age formula as a hard and fast rule, many do, the Fitbit folks included. Judging by the number of threads started by folks concerned about their heart rates, the use of these formulas are, if anything, over done and we should endeavor to encourage individual contributors as well as companies like Fitbit to use alternate language and/or alternate methodologies for establishing workout routines and such.
05-17-2016 06:16
05-17-2016 06:16
@FitBeforeFifty wrote:
@shipo wrote:
The issue isn't failure to identify obesity, it is identifying obesity when it doesn't exist. To my way of thinking, using young men as a study group is decidedly biased. ...
Even still, I wouldn't exactly call an 83% accuracy rating very good.
I agree the first study isn't very diverse (young males), that's why I provided 2 studies. The second one is much more diverse: A cross-sectional design of 13,601 subjects (age 20–79.9 years; 48% men) from the Third National Health and Nutrition Examination Survey
In both studies, the number of false positives are relatively low. The second study had a specificity of 95% in men and 99% in women. In other words, BMI doesn't falsely identify a lot of people as obese. Some, yes. A lot, no.
Here again, false positives are the issue, and while the studies don't show a high number, I strongly suspect if a study was done of healthy/fit middle-age and older individuals, the incidence of false positives would be alarmingly high. As an additional rub, my company hosts a "health screening" forum each year, and if you are within certain health metrics which are established by an outside company, you get a discount on your health care costs. Because I am officially obese per the BMI charts, I am automatically disqualified from the discount, and in order to be granted said discount, I need to file for an exception. Fortunately I'm quite fit and am able to do so, but it is really annoying to even have to file the appeal.
05-17-2016 07:39
05-17-2016 07:39
BMI is correlated with body fat percentage and not fitness. You can be fat and fit.
You can argue that body fat percentages shouldn't be used as a health indicator but that's a separate issue. Statistically speaking, there isn't a high incidince of false positives when BMI is applied as a measure of body fat percentage.
05-17-2016 08:15 - edited 05-18-2016 07:39
05-17-2016 08:15 - edited 05-18-2016 07:39
I don't understand why some folks are wedded to highly flawed formulas, BMI and 220-Age being chief among them.
While some folks may agree with the above statement, they will just as quickly say, "Well, they're better than anything else."
Over the years numerous alternatives have been promoted, some are even being considered for adoption. In an effort to keep thing simple, how about the following:
05-18-2016 07:36
05-18-2016 07:36
@shipo wrote:
- Instead of BMI, use a simple calculation of Height/waist size; if the result is greater than 2, then the subject is overweight.
I think you mean "less than," as the waist size is inversely proportional to the resultant number.
A guy 72" tall with a 36" waist would be exactly 2.
Same guy with a 32" waist would be thinner, but your formula yields a 2.25 result.
05-18-2016 07:38 - edited 05-18-2016 07:41
05-18-2016 07:38 - edited 05-18-2016 07:41
Yup, good catch; I'll edit my post (if I can).
Edit: Yesterday's post edited successfully.
05-18-2016 08:00
05-18-2016 08:00
There is a good Waist to Height Ratio Calculator at ShapeFit.com, which gives definitions of your ratio.
You can check their calculator here: Waist To Height Ratio Calculator – Assess Your Lifestyle Risk