Wednesday, October 22, 2014

Update on the 6-week 20 lb Class Data

This is just a quick update to last week's post.

I took photos this morning, but they were blury. Not very good. I was only able to read 42 lines with confidence. But, for a quick update on the statistics, that's probably a relevant sample. Out of 109 total contestants (13% of whom I called "drops" last week), 42 lines of data is a 38% sample. That's probably good enough for an 80% confidence level.

I had full 5 week data on 27 contestants, and 4 week data on another 15.

In the 5 week data, the median is 1.9 lb/wk, and the average is 2.5 lb/wk. (remember, the magic number is 3.33 lb/wk). The distribution looks like this.
In the combined 4 and 5 week data, the median is 1.8 lb/wk and the average is 1.8 lb/wk. And, the data looks like this:
I'm tempted to say that the higher average of the 5 week only data indicates a "week 5" bump. (Its pretty logical to suspect that week 5 would be a good week.) But, I really couldn't say that without doing a bit more analysis. Its possible.

I could also look the other way. Last week, the averages and means were in the range of 2 lb/wk, now it is down at 1.8 lb/wk. But, again, my numbers this week are pretty cursory.  I'd want to really prove that.

In general, I'm going to say that this data check-in still supports last week's conclusions: Last week, I predicted a 25%-40% success rate for losing 20 lb in 6 weeks. And, all things being equal, I'd still assert that prediction. But, all things are not equal in week 6!

Week 6 Changes

The coaches made some pretty serious changes to the program in the last week. We got a notice last Friday:

  • They advised everyone to cut all carbs and fats for week 6, with the exception of one day (Wednesday) carb consumption of  1g/lb of body weight. 
  • They've also hinted that they are going to play some water weight games at the end, before the final weigh in. 
  • They recommended adding an hour brisk walk as a tactic.

This type of thing will certainly boost the program's success rate. Overall, cutting carbs will mean a 25% reduction in calories.  But, additionally, in most of the contestants, it will induce ketosis. Putting people into ketosis, while sustaining a daily workout plan, could have some pretty big results. Back when Atkins was a craze, I remember hearing reports of people loosing 10+ lb/wk on the induction phase of the diet. I don't think it is unhealthy per se. I doubt that a week of ketosis would do anybody any harm. (Heck, everyone was doing it in the 90s.)

The Wednesday instruction is a carbohydrate cycling thing.  I don't really know much about carb-cycling. So, I can't make any commentary on how well it will work. But, I can only assume it will improve results.  If contestants follow the instructions, they'll put themselves into ketosis on Saturday, out of ketosis on Wednesday, then back into ketosis on Thursday.

The no-carb thing doesn't really work for me.  I'd tried it in the 90s, and I get light-headed. I tried going no-carb for a few days, starting last Friday. And, I made if through the weekend. But, I my mental focus was bad, and I lost all energy to function like a human.

Jenn says all this week 6 stuff is "cheating". And, I kindof agree.  From a numbers basis, if it takes a pound off the scale, it counts.  The ketosis and carb-cycling will certianly do that. So, I don't consdier them "cheating". For me, I'd look at them as short-term strategies. I think she was upset that the rules changed in the last week, giving her the "I didn't sign up for this" syndrom.  I do, though, think  loosing water weight is quite different from loosing fat. Those types of games should be reserved for wrestlers! I signed up to loose fat.

Extra workouts, I'd say, isn't cheating. I've been doing those anyway.  I'm planning to put in a 4-mile run tonight, and probably a 9-mile on Sunday.

For me, personally, I need to loose about 1 - 1.5 lb this week to make the goal.  So, I'm skipping all the carb stuff and sticking to the normal program. I've been loosing over 3 lb/wk. So, I'll probably make it without making changes.



Tuesday, October 14, 2014

And, Here’s what happens when you have a serious dork in your Weight-Loss Class…

Jenn and I are in a 6-week, “lose 20 pound challenge” class at a local gym.  We just did our week-4 weigh in.  So, we’re 2/3 of the way through.  I’m on track to “win”, having lost about 15 pounds.  But, Jenn is a bit behind, and feeling somewhat discouraged.  I've told her she’s probably doing better than most in the class. So, here, I prove it.
Being the dork I am, I also just had to know how well the program works.  So, I “bootlegged” the week 4 weigh-in data with my camera phone. (Don’t worry, I don’t include any names. I may be a dork, but I’m an ethical one! I know to blind my data!).  
Here’s my assessment. The overall data looks like this. Lovely, huh?
Notice the overall trend downward. The biggest numbers, by far, are in the first week.  About 30% of people had their best week in week 1. The average loss in week 1 was 2.7 lb.  In week 2, it dropped to 1.6 lb.  In week 3, it dropped again to 1.3 lb. And, in week 4, so far, the overall has dropped again. 
You can actually see a “split” in the week 4 data.  About 10% of the contestants gained weight in week 4. If you drop them from the data, the weight loss in the rest is 2.2 lb. 
If the theory of muscle benefit works, weeks 5 and 6 should have the biggest numbers. We'll see!
In every week, the statistical deviation is pretty broad.  For the first three weeks, standard deviation is over 2 lb.  So, let’s look at the overall pounds per week data across the four weeks.
Overall, there were 112 lines on the weigh-in charts.  I had to drop the last 7 because my spy photos were bad!  So, I had 105 data points to work with, which is 93% of the contestants. There were 15 people that I had to count as "drops": They missed more than one weigh in. So, let’s assume they quit. That's about 13% of the pool. I’d guess that a 10%-15% drop out rate would be fairly normal in a program like this.
So, there were 61 contestants for which I had four full weeks of data. If someone missed a weigh-in in week 2 or 3, but came back, I left them in the data pool, and averaged their losses across the missing week.
Of those 61, the average pounds lost per week was 2.01.  The median was 2.03. (So, 50% of people lost less than 2.03 pounds per week.). In order to be successful, contestants need to be losing about 3.33 pounds per week. Only 7 of the 61 candidates (11.5%) did. I'd be hopeful for anyone who lost over 3 pounds per week. But, that only adds 5 people, for 19.6%.  In order to succeed, the average contestant will need to loose 5.8 pounds per week in week 5 and week 6.  
I had three weeks of data on an additional 29 contestants. Lumping them in with the four week data gave me a pool of 90 people. For that pool, the average pounds lost per week was 2.11.  But, the median was slightly less, at 2.01. (So, 50% of people lost less than 2.01 pounds per week.) So, it didn’t change the outlook much. I'd consider those two data sets to be aligned. In both cases, the average and medians are within 0.1 pound per week.
So, overall, based on the 61, and the trend of pounds per week, the likely success rate will be 20% or less. (80% or more failure of the non drop-outs.)
A more optimistic prediction could be had if each contestant duplicated their best week (often week 1) in both week 5 and week 6. In that case, the success rate would still be only 20%. But, there would be another 20% that are within 2.5 pounds of the goal. Those people could push themselves over the edge for the final weigh-in, giving the program a success rate of 40%.  (Hey, there’s always hope!)

I was also interested in the relationship of weight loss to starting weight. You'd expect those who have "more to loose" to have a higher weight loss per week.  And, the data looks like it "leans" to the right. But, this really isn't true, to any statistical relevance. A regression test of starting weight versus pounds per week has way too much error (over 80% error).  So, really, your odds of losing 3 lb/week are very nearly the same at 170 lbs as they are if you start at 270 lbs.
Analysis
The heart of the program are (as you’d expect), workouts and diet. First, let’s put some numbers on the workouts.
The workouts are 50 minutes of circuit training, 5 days a week.  Based on livestrong or myfitnesspal, that type of workout burns 3.8 calories/hr/lb of body weight.  There may be a 20% deviation if you “really work hard”. Let’s assume the workouts are a fairly constant daily 550-750 calorie credit. So, everyone gets between 2,750 and 3,750 calories per week of exercise. Across the six weeks, that’s 16,500-22,500 calories. To lose 25 pounds (at 3,500 calories per pound), participants need to burn 87,500 calories. The workouts are giving them somewhere between 19% and 26% of that. There are certainly people – me, for example - who are putting in extra workouts to burn some more. (Jenn calls it “cheating” when I do my evening runs.) But, as is so often and so sadly the case, success is 80% diet.
So, what the diet? Based on the deviation in the data, we have to conclude people are doing some different things.  
The program calls for a very controlled diet with a list of approved foods, and specific quantities per person.  From what we could tell, it seemed to be set up for about 50% calories from protein, 25% from carbs, and 25% from fat.  Both Jenn and I were assigned protiens, carbs, and fats at roughly the following: 1.0 gram of protein per pound of body weight, 0.5 gram of carbs per pound of body weight, and 0.2 grams of fat per pound of body weight. So, it was a pretty traditional body-builder diet (40%p/40%c/20%f), with the carbs cut in half.
But, regardless of the intent, clearly, there are big differences happening in the group.  People who are losing 3 pounds per week have 10,500 calorie deficit. People who are losing 0.5 pounds per week are only at a 1,750 calorie deficit. That means their diets vary (over and above their standard BMR calculation) by about 8,750 calories per week, or 1,250 calories per day.  So, within the program, you could have two 36 year-old female contestants, at starting weights of 170; one is eating 1,250 calories per day more than the other.
Jenn and I suspect some of this is in the meal planning the gym does. Overall, our meal plans were 1,000-1,500 calories per day more than what myfitnesspal or caloriecounter told us to do for a 2 pound a week loss.  So, we adjusted our portions down - which I think was correct (we’re both beating the class averages). I suspect other people did this, too. But, I also know some contestants complained there was “too much food”, and were told to “get it all down”..
Conclusions
For consumers, the program itself is not likely to be successful. If my predictions hold, the success rate will be 25%-40% for non drop-out participants. If the success rate were a bit higher, 40-60%, you could blame the participants for cheating or not giving it their all. But, if the success rate is that lower than 30%, its a programmatic failure.
The best odds for increasing success are likely in the diet. Our experience indicates the meal planning may not be well calibrated to the individual participants. And, the results point to a wide variation in implementation. If you are planning to do this program, my advice would be to double-check your overall diet planning and make sure you maintain an appropriate calorie deficit throughout the period.

EDIT: See the following week for a follow up.