Mar 10, 2014
Followup to: Lifestyle interventions to increase longevity.
What does it mean for exercise to be optimal?
There may be even more criteria.
We're all likely going for a mix of outcomes, and optimal exercise is going to change depending on your weighting of different factors. So I'm going to discuss something close to a minimum viable routine based on meta-analyses of exercise studies.
Not knowing which sort of exercise yields the best results gives our brains an excuse to stop thinking about it. The intent of this post is to go over the dose responses to various types of exercise. We’re going to break through vague notions like “exercise is good” and “I should probably exercise more” with a concrete plan where you understand the relevant parameters that will cause dramatic improvements.
Optimality aside, I recommend starting with a very minimal routine for 6-ish weeks to build the habit of exercise in to your life. You'll want a program that causes you little mental stress that you can actually stick with. You've got a few options for achieving this. The gains from weightlifting can be surprisingly quick—you'll see dramatic changes in your appearance in 4 months—and seeing yourself lift more weight every session can be a great motivator. Couch to 5k is a basic running progression designed for sedentary people. A daily bodyweight routine is a good way to achieve habit formation through consistency. I recommend making a firm choice and sticking with it until it becomes easy.
Once you've made exercise a habit, you'll want to gradually nudge yourself towards the level that's optimal. So what is that level? Most of the rest of the claims in this post are supported by this review by Swiss researchers. As far as I know, this is the largest systematic review of exercise studies ever undertaken, reviewing 7000 studies with 80 meeting inclusion criteria covering over 1.3 million subjects. Sheer size, however, is not the only reason to take this study very seriously. As someone who has read hundreds of exercise studies, I can say that the methodology of the meta-analysis done to determine dose-response to exercise is excellent. What is most encouraging is that the study authors repeatedly point out shortcomings, and ways their findings should not be interpreted because the underlying data does not warrant it. They also check for publication bias. One potential caveat is that this a review of cohort studies, not RCTs. But the authors note that RCTs of exercise almost always show greater effect sizes, not smaller. This is likely because people over-report how much exercise they do in observational studies.
In order to compare the intensity of different activities, exercise researchers use a unit called a MET, or metabolic equivalent. The MET is defined so that your weight (in kg) * METs = Calories you're burning per hour. An example MET table can be found here. For the purposes of exercise studies, activities are typically classified as low-intensity, moderate-intensity, and vigorous-intensity. These roughly correspond to 1-3, 4-6, and 7+ METs per hour. For typical individuals, this will translate to approximately 200, 400, and 600+ Calories burned per hour.
On the low end, some studies have found dramatic benefits from just the first 15 minutes of moderate-intensity exercise per week. These studies indicate that you gain about as much going from no exercise to some exercise as you do going from some exercise to optimal exercise.
The Swiss review finds that the first hour per week of vigorous-intensity activity gets you 2/3rds the benefit of 10 hours per week, but the study authors make sure to point out that this is an implausible effect size and that there are almost certainly some confounding and reverse causality issues going on. Which is to say that people who have better health are simply going to be capable of more exercise.
How about on the high end? Studies differ on where the point of diminishing returns is. Some put it at 1000-1500 Calories; others as high as 3500 Calories. (Remember, a typical individual burns ~400 Calories per hour of moderate-intensity exercise.) I'll shoot for 1500 Calories in my recommendations; 3500 Calories is pretty hard to reach without exercising like a pro athlete.
Estimates indicate that each minute of exercise gets you 3-7 minutes of extra life on average, with higher returns for more intense exercise. So every week, you have the opportunity to get a 3-7x ROI on time spent exercising up to the point of diminishing returns. I recommend high-intensity exercise—not only does it save time, it's also been shown to improve health more on a per-Calorie basis.
You may have shied away from weight training in the past because you thought you would turn into some huge gross bodybuilder. But bodybuilders and fitness models take drugs and spend years training intensively to look the way they do. You are not going to gain 20lbs of muscle overnight magically. This goes double if you’re a woman. You do not have testosterone; you are not going to be building huge muscles no matter what you do.
Of the forms of exercise I cover, weight training has the most rigorous evidence separating what works and what doesn’t. This study (pdf warning) examines what sort of resistance training results in the most rapid improvements.
In weight training lingo, AxB means A sets of B repetitions. So 4x10 would mean 40 reps with rest periods every 10 reps. Our study recommends starting with 4x10 3 times a week, and transitioning to 4x4 2 times a week as you become stronger. Aim for a weight you can barely complete all the reps with.
For an efficient full-body workout, select one exercise from each movement pattern:
Upper push: bench press, incline press, overhead press, dips.
Upper pull: cable rows, barbell rows, dumbbell rows, chin-ups, face pulls.
Lower push: squats, lunges, leg press.
Lower pull: deadlifts, power cleans, hyperextensions, romanian deadlifts, reverse hyperextensions, glute-ham raises.
So a good starting routine would be
A: 4x10 each of squat, bench press, lat pulldown, hyperextension
B: 4x10 each of squat, overhead press, cable row, hyperextension
alternating A and B workouts on different days of the week e.g. AxBxAxx, BxAxBxx.
You'll try to increase the weight by 5lbs each session. As you improve, you want to decrease the reps and increase the intensity so you can keep advancing. For example, if you stall a couple times doing 4x10 at 125lbs on your squat, switch to 4x8 and keep increasing the weight, then 4x6, etc. until you get to something close to an optimal trained routine:
A: 4x4 each of squats, bench, weighted chins, deadlifts
B: 4x4 each of squats, overhead press, barbell row, power cleans
At this point, you're going to the gym only twice per week to give yourself more recovery time.
For learning exercises, there are many tutorials available online and I recommend checking some out if you are confused about form. You can always search for "<name of exercise> tutorial" and get articles and Youtube videos. Many people feel silly practicing their form with extremely light weights (often just the empty bar). But many world record holders start EVERY session this way to warm up and cement muscle memory. Others are silly NOT to do this. Also keep an eye on your ego. It's easy when setting goals for yourself to try to lift a weight that you can't really lift with proper form, because you want to set that personal best. But you'll feel pretty stupid when you are forced to miss the gym for a month because you hurt yourself.
On exercise selection: I'm not a big fan of deadlifts for absolute newbies, unlike say Mark Rippetoe in Starting Strength. Maybe add deadlifts in after you've gained some muscle and you have better awareness of form. I also differ from Rippetoe in recommending that newbies high bar squat (the distinction being that low bar squats place the bar across the shoulders and high bar squats place the bar on the trapezius). I have taught newbies both forms and most find high bar squatting easier to figure out how to do properly. I spent months learning to low bar squat and still injured myself; high bar squatting can be taught in a couple sessions in my experience. Pay attention to whether a tutorial video is trying to teach you low bar squatting; the cues for each exercise are different.
Free weights are generally better than machine exercises, but I recommend cable rows and lat pulldowns to newbies. The goal is to move from cable rows to barbell rows, and from lat pulldowns to actual chin-ups. The issue here is that a beginner won't be able to do the requisite sets and reps of chin-ups and rows with good form.
A note about equipment: Weightlifting shoes have an incredibly high return on investment. They make back injuries less likely, and drastically improve subjective experience of squatting. You can get Rogue weightlifting shoes (use your size in men's dress shoes to size them regardless of gender) for around $120; there are cheaper options available but good shoes will last years so the amortized cost is low. Here's a full list of options; note that even the cheapest weightlifting shoes are miles better than lifting in tennis shoes. I don't have any personal experience with the Reebok CrossFit lifter, but they seem like a good option under $100 for a shoe with the desired .75-inch rigid heel. I recommend a cheap belt (expensive ones aren't any better) in order to improve your execution of the valsalva maneuver during squats and deadlifts which further protects the spine from flexion under load.
For a beginner, something like this is reasonable. Of course such a program will max out in fitness gains fairly quickly, even if you start doing several cycles of it. But this isn’t our worry as a beginner. For someone serious about progressing with body weight exercises past this stage, I recommend a program like Overcoming Gravity or Building the Gymnastic Body. There is not really formal support for the efficacy of these programs, but they are endorsed by coaches who train many people successfully, and are consistent with the general principles of weight lifting (progressive overload, training frequency, etc.).
For cardio, I recommend against high-intensity intervals when starting out. High-intensity intervals carry a greater risk of injury, especially if you're not used to them. They're also unpleasant and not conducive to building habits. For starting out, I recommend something that is based more on psychological results rather than performance optimality, like Couch to 5k. As you progress, start adding in short bursts of more intense effort. The idea is to tire yourself out quickly. If your cardio routine lasts more than 30 minutes you’re probably going too easy.
What type of cardio should you do? Cardio that is amenable to high intensity is probably one of: running (especially up hills), swimming, rowing, biking, burpees, or jump rope. But you might be able to adapt others. I'm a huge fan of rowing for a few reasons. One, it works more than just the legs. Two, you can have a rowing machine in your house, which drastically lowers activation cost. Three, I just find it less aversive subjectively. You can keep stationary bicycle mounts in your house as well, and they have the advantages of being compact and very cheap if you already own a bike. Burpees require no equipment, but they bothered my knees. They work great for some people though. Jumping rope is also very space/time efficient but the skill required acts as something of a barrier. If you find learning the skill enjoyable, it's a great option. It is worth noting that runners, bikers, and rowers have among the best VO2 max scores of any athletes.
What does the optimal high-intensity cardio routine look like? Data on this comes from this Meta-analysis of VO2 max trainability. VO2 max has been shown to be a robust predictor of mortality. This relation has held across elite athletes, to average individuals, to the overweight (see Figure 2 from this meta-analysis of vo2 max trainability). Unfortunately, it appears that the protocols eliciting the greatest increases in VO2 max are so arduous as to have high attrition rates. 6 days/week is not a schedule of training I expect anyone but professional athletes to maintain. What sort of realistic routine can still achieve most of these gains? The meta-analysis supports a mixture of 3-5 minute intervals, and longer duration but still intense intervals (30-40 minutes of continuous training). The authors also note that they did not include analysis of the evidence that very high intensity exercise (1 minute or less of max effort) shows unique health benefits. We could simply conclude that a mixture of interval times is good, and that every increment of cardio up to very high levels is likely good for us, but that doesn't feel very motivating. What we want is a clear goal. I'm going to combine data from the VO2 max study and the Swiss review to get a rough estimate. If we want to do resistance training twice per week and cardio at least twice per week can we realistically burn the 1500 calories we want? Let's see. Interval training sessions can vary widely in number of calories burned, but a sprinting session, a 4x4 protocol (one of the more popular protocols in the VO2 max meta analysis, consisting of four 4-minute intervals), and a 30 minute run can burn between 200-450 calories as a first approximation. Twice weekly and we have 400-900 calories. This leaves 600-1100 calories for 2 weightlifting sessions. Estimates for calories burned weightlifting vary extremely widely, most likely due to the huge number of exercises considered "weightlifting", but even the lower estimates put it over 300 calories per hour. This puts 2-4 hours of weightlifting per week at 600-1200 calories expended. How convenient!
All that remains is to suggest specific cardio routines. I don't have the evidence to say with any confidence what a truly optimal routine would look like here, but I can at least give well studied examples of each.
Very high intensity routines follow a pattern of a short warmup (5 minutes at a slow pace) followed by several bursts of 10-60 seconds all out intensity. (30 on 30 off for 10 intervals is popular and close to maximizing vV02max)
VO2 max interval training consists of four 3-5 minute intervals at 85%-95% your max heart rate interspersed with slower jogging for the same interval.
Longer interval training consists of 20-40 minute runs at a consistent pace such that you are exhausted by the end.
I wouldn't worry about the optimal frequency for each one. Don't forget that even training populations just consistently doing one type shows very dramatic improvements in health. I'd suggest freely mixing them up and trying to have fun with it.
Summary of my recommended routine
This is what I recommend gradually working towards once you've made exercise a habit:
Don't do vigorous cardio on the same day as lifting weights! It's a good way to injure yourself, especially your lower back. Exercise doesn't make you stronger; it makes you weaker. It's the recovery from exercise that makes you stronger; give your body time to recover.
Don't try to implement a new diet and a new exercise plan at the same time. If you're trying to choose, do an exercise plan first—effects on health are much larger.
If you are underweight or normal weight, you'll need to eat more when you start exercising. Celebrate after your workouts by eating to reinforce the exercise habit. You may think eating pizza is bad for you, but not exercising is worse, so reward yourself however you want. Or drink my nutrient dense shake, designed to be consumed after workouts. (John_Maxwell_IV and I are planning to commercialize it after we roll out our first nutritionally complete food.)
If you're overweight: I agree with Gary Taubes that exercise is NOT a good way to lose weight. But exercise has bigger effects on health than weight loss, so I actually recommend prioritizing exercise over changing your diet. (Like I said, don't try to do both at once.)
Note that you don't need to stuff yourself with massive amounts of protein to build muscle. Studies have never shown a measured benefit to consumption above .64g/lb of bodyweight, which translates to around 100g for a 150-160lb person. A single serving (3oz) of chicken, for example, contains about 21g of protein.
If you've made it this far, congratulations; you are now as knowledgeable as any personal trainer I've spoken with.