What makes an exercise “super Responder”?
A Roundtable Discussion with Alan Aragon, Dr. Conrad Earnest and Dr. Jose Antonio
Most people most of the time people will respond to exercise in a fairly predictable manner. When doing research on a group of people put on a specific exercise regimen, the majority will respond in a fairly typical and expected fashion. However, what they will also find, are a small percent of people who fall on both ends of the extreme, what might be termed non-responders on one side and super responders on the other. They are always a very small percent of a study as expected, but have been a consistent finding since, well, forever. Those two groups that fall well outside the typical response of the larger group, were once considered a statistical anomaly, or perhaps non-compliant, or other possibilities, without much attention given to them. However, more recent studies have focused on those non-responders, and studies have found identifiable genetic differences in non-responders to more typical responders. I cover that topic in a prior post HERE for those interested.
What of super responders? What do we know of them? It’s my view if we can identify what makes these genetic elites respond as they do, we may be able to apply some of that information to us mere mortals in the future. There’s surprising little hard data on super responders, but anyone who has been involved in exercise physiology research, and or worked with athletes over the years has seen them.
What’s a super responder? As the name implies, a person who responds to exercise in a manner so unusually and un expectedly, you just marvel at their elite genetics to the point it’s hard to view them as the same species as the rest of us! I will admit my bias here in that I’m mostly interested in resistance training super responders, but they will exist in other endeavors too to be sure.
I have seen what I’d consider genetic elite super responders a handful of times in my thirty plus years in the “biz” and they never fail to amaze me. I worked with one guy who trained sporadically at best following a haphazard program and his nutrition, in a word, sucked. He’d diet for a few weeks, and win every local and regional show he did, then with a little actual effort – via my assistance with keeping him on track – won a national level show and got a pro card. After he won his pro card, he decided to take time off from the gym. Six months later he comes to a cookout I was having and after six months of no exercise, looked only slightly smaller! He still looked better than 99% of the people busting their butts in the gym year in and year out and I wanted to kill him. I kid!
This super responder phenomena is not limited to men in the least. Just recently I met a gal who looked like someone who was in great shape and exercised regularly. Lean, muscular, shapely was she. Looked like she was perhaps four-six weeks out from a figure show. I asked her where she worked out and she said she didn’t. After a lengthy discussion I came to learn she did no exercise other than her work as a physical therapist (working on my torn rotator cuff but that’s another story) ate mostly fast food until recently, and in her late 30s. Yet, she looked better than 90% of the people you’d see in the gyms! As expected, everywhere she goes people ask her what she does for a workout and nutrition plan she follows, etc. Frankly, I was gobsmacked! I can’t even fathom what she’d look like if she regularly hit the gym. She said her entire family was like that. She showed me a picture of her brother who does lift weights, and he was a monster.
I have been in the biz a long time, worked with pro bodybuilders, Olympic track athletes, SWAT teams, and so forth, and it takes a lot to impress me. The super responders I have known blow my mind. I could go on with a few more stories about such people with elite genetics a fraction of the population possess, but I decided to bring in additional help to discuss the topic and see what, if any, consensus may exist on these “genetic freaks” as some will refer to them. Dr. Jose Antonio, Dr. Conrad Earnest, and Alan Aragon are some of the most knowledgeable and experienced researchers I know, who also have extensive “real world” experience in the exercise physiology arena, so I threw some Qs at them in this roundtable discussion.
Readers Note: I’m intentional over simplifying the topic of statistics in terms of where people may fall in data sets and such. Bio-statistics gets real complicated real fast, and for all but science nerds like me and such, deathly boring.
Q #1: Per my intro gents, no doubt you have experienced the same thing I have over the years. I know there are data on non-responders, and some interesting preliminary findings on non-responders specific to some genes identified between them, but what of super responders? Is there any data you’re aware of regarding super responders? Would seem an interesting and potentially useful area of inquiry for us mere mortals.
Alan : I’ll be candid here and admit that the first thing that came to mind was a series of studies by Jacob Wilson’s group on HMB-FA, and subsequently, HMB-FA/ATP. Even though I personally cannot help but be skeptical of the findings, they are nevertheless impressive if taken on face-value. In their first study, Jacob specified in an interview that, “Now, in most studies you just recruit anybody, and you could have non-responders, moderate responders, high-responders. We filtered so we had purely high-responders to training.” The results of their selective recruitment were quite astounding. The HMB group had a LBM gain of 7.4 kg (16.28 lb) while dropping 3.5 kg (7.7 lb) body fat. In contrast, the placebo group gained 2.1 kg (4.62 lb) LBM while losing 1.7 kg (3.74 lb) body fat. These are both favorable scenarios, but the HMB group’s LBM gain –is definitely on the freakish side, especially since these subjects were resistance-trained; they were not novices primed for huge gains. In a subsequent study out of the same lab, this time using HMB-FA/ATP, subjects gained 8.5 kg LBM (18.7 lb) in 12 weeks. This LBM gain surpasses the 7.5 kg gain in the previous HMB-FA study, which already was greater than the gains seen by Bhasin et al (PMID: 8637535), whose untrained subjects were on exogenous testosterone (600 mg/week). Subjects in the training + testosterone group gained 4.2 kg of LBM in 10 weeks. Compare this to Lowery et al’s trained subjects on HMB-FA/ATP, who gained 8.4 kg LBM in 12 weeks. Furthermore, Bhasin et al’s subjects’ quadriceps thickness increased 7 mm, while 7.8 mm was gained in the HMB-FA/ATP study. In sum, the subjects in the HMB-FA studies showed LBM gains that surpassed exogenous testosterone by twofold, and were four times the typical LBM gains seen from creatine supplementation. So, at least on face-value (if you believe the data reported), these subjects were super responders; they were packing on an average of more than a pound of LBM a week for months on end – and these were not newbie gains. There’s also the alternative explanation that HMB-FA is just that magical, and even regular Joes can become supermen on it.
Conrad: To be quite honest, I think we’ve had the data all along but never really talk about it. As you’ve already discussed, we spend a lot of time focusing on non-response or low response because we want to see people improve and are puzzled when they don’t. I think this applies more to health than performance.
So, how we have had the data all along? If you look at any study you will see an average response accompanied by what is known as an error term. When reported correctly these take the form of standard deviations of confidence intervals, which basically show readers the variance in the response to the study. The minus sides of the average are low responders and the positive sides are the more responsive individuals.
Thus, if we took everyone above the average we’d have a responder pool to examine; at least theoretically. One has to appreciate that variance related to testing can factor into this as well. And, of course, there are “outliers” who over or under respond beyond the point of what is reasonable.
A classic example of this are a number of JAMA reports comparing diets like Mediterranean, DASH, Atkins and Ornish, etc. If you look at the error terms you will see that some people clearly improve more than others.(5, 4)
To gain insights into performance one would have to be committed to genetic analysis for all study participants and (it’s a big “and”) be willing to recruit people into studies in the hundreds. It would be an expensive undertaking.
This is actually going on in the endurance world right now. Claude Bouchard, PhD is leading a world-wide effort looking at elite athletes with VO2max’s over 70 ml/kg/min to identify gene patterns of truly elite level athletes. (1) Tuomo Rankinen, just last year, published a paper showing no evidence of a common DNA variant profile specific to world class endurance athletes.(2)
Jose: I published a study on high protein intakes (> 3 g per kg daily) showing that a few subjects responded quite a bit (gaining about 6 kg of LBM in 8 weeks) while others didn’t gain a thing. Some even lost LBM. Whether that classifies as scientific evidence for a ‘super response’ is certainly debatable. If you really want to watch super-responders, watch the World Championships in Track and Field each year. Or if you want to wait every 4 years, the Olympics. Those men and women are by definition super- responders.
Q #2: Some years ago there was a great deal of attention placed on myostatin-related muscle hypertrophy, animals with myostatin deletion animals being incredibly muscular, and many being convinced that was the magic bullet to muscularity. Of course a bunch of worthless “myostatin inhibitor” supplements sprung up following that. I recall some pro bodybuilders being tested for alterations in the myostatin gene, but the results were inconsistent. While myostatin clearly plays an important role, it’s obviously not the magic bullet first thought. It seemed the focus on myostatin dropped off as data on the importance of mTOR and PI3K pathway increased, and people then jumped on that as the new be-all-end-all of pathways and related genes for manipulating muscle mass and such. What area of genetic related research do you feel worthy of focus that might show what essential differences exist between super responders and the rest of us? It’s obviously an immensely complex topic.
Conrad: At this point in time, genetics can only point to a few genes that may influence strength performance. They are not (yet) the magic bullet because, as you have stated, genetics are complex and involve hundreds of genes. In my previous example, there are hundreds (probably more) of genes in involved in cardiovascular adaptations and no reason to think the same does not apply to strength.
Thus, hundreds of subjects will eventually need to be analyzed and then run through some fairly elaborate modeling – perhaps machine modeling – to ferret out the top candidate genes. Once we have a clue on those, they would further have to be aligned with muscle fiber content and potentially muscle architecture.
Jose: I’ve actually wrapped up a study on the ACTN3 gene (gene for ‘speed’). A few things you can learn from it. One, you ain’t a slave to your genes. But two, your genes can limit you. If you ain’t got the ACTN3 gene, then forget about performing well at a speed sport. It would be like entering a math contest with just half a brain. Gene research is certainly the next frontier in exercise and sports science. I’ll be looking at genetic markers for ‘fatness’ (FTO gene) and others. So you may have super-responders (i.e., speed gene, ACTN3); but on the flip side, you might be a ‘super-responder’ for gaining fat. It’s really different sides of the same coin.
Q #3: To that end, do you feel it’s likely to be some combination of genes science is already aware of, or some yet to be discovered? Any pathways you’re currently looking at that might be a piece of this puzzle?
Alan: I’ll answer the previous two questions in one swoop here since they’re similar. Funny you should bring up myostatin inhibitors, Wilson et al have an abstract published in JISSN (PMC: 4271642) showing positive results from a follistatin derived from fertile chicken egg yolk isolate. 1.7 kg was gained in 8 weeks in recreationally trained subjects – not nearly as impressive as the gains seen from HMB-FA. Based on the current literature on molecular drivers of muscle hypertrophy, many of the areas are too disjointed and poorly understood to draw and firm conclusions or base future directions on. While it’s true that the PI3K/AKT/mTORc1 (collectively abbreviated to mTOR) pathway has received a ton of study and recognition for its role in muscle anabolism, it was also relatively recently found by Phillips et al (PMID: 23555298) that subjects with the greatest degree of hypertrophy showed an inhibited mTOR activation signature, including the down-regulation of 70 rRNAs. A bit of a head-scratcher there. I think the areas of focus that we might more fruitfully sink our teeth into are the activation and incorporation of satellite cells, and what agents or protocols might be most potent at achieving this. For those unaware, satellite cells are precursors to full-blown skeletal muscle cells; they’ve been referred to as the most abundant myogenic stem cells. They are capable of providing additional myonuclei for augmenting muscle fiber growth. The implications here are potentially profound, since the ceiling of muscle growth is thought to be influenced by the number of myonuclei within the fibers. Creatine is widely considered to be the most effective supplement for muscle hypertrophy, and one of its mechanisms of action is the proliferation of satellite cells & myonuclei (PMID: 16581862). Another area of interest where the findings have been a bit more concrete is muscle protein synthesis (MPS), and which agents or protocols most potently increase it over the short- and longer-term. As for nutritionally mediated gainz, gene expression of cyclin-dependent kinase 2 (cdk2) – a factor that increases cell proliferation and body mass – has been shown to increase both acutely and chronically as a result of resistance training, but only after the ingestion of whey protein near the training bouts (PMID:18661258).
Conrad: Definitely a combination of genes as they relate to fiber type.
Q #4: During your own research, be it exercise training of various types, supplements etc, have you found super responders once the data came in? When you saw it, did you put it down to some statistical anomaly? Voodoo? Or, did you wonder how and why a small portion of your study participants were so genetically gifted? It would seem until very recently, the small fraction of people who fell well outside the statistical norms, the non-responders and super responders where viewed as more a curiosity of unknown origin and essentially ignored. At least in the non-responder side, that appears to be changing. Thoughts?
Alan: In the research I’ve collaborated on thus far, I unfortunately can’t report anything spectacular. In fact, it’s some of the most anticlimactic/unsexy results in the literature (LOL). However, in the most recent RCT of ours that’s in publication (PMID: 28070459), we did report individual data, and one of the subjects showed approximately an 18 mm gain in biceps thickness (0.71 inches), measured via ultrasound, over an 10-week period. The rest of the subjects – all of whom were resistance-trained – stayed pretty much the same, which was not surprising since they were in hypocaloric conditions. It’s interesting to imagine what type of response this high responder would have if conditions were optimized for muscle growth, and/or supplementation was implemented. On the poor responder side, one of the subjects experienced a huge drop in quadriceps thickness (25 mm; about 1 inch) during this study. Interindividual response variability is just the nature of the beast, and a very poignant example of this is work by Ahtiainen et al (PMID: 26767377), who demonstrated vast heterogeneity in subjects’ responses to resistance training. Their study is openly accessible, have a look at this figure showing the full range of differences in strength & size gains (and losses): https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5005877/figure/Fig4/. High responders gained 4-5 times the muscle mass of average responders. Of course there are the unfortunate subjects (2% of the subjects were classified as low responders) who included those who experienced losses in size and strength as a result of resistance training (and in my speculation, other factors as well).
Conrad: Not really, no. It’s a great question. From a statistical point of view one has to separate out potential outliers who may over or under respond vs. a “super responder.” Outliers are more times than not a matter of study variance (typically unintended) where participants might not have been fully compliant to a protocol.
For example, a few years a go I performed a study looking at the effects of Astaxanthin on cycling time trial performance and found one person who responded amazingly well. Too well and outside the realm of reason. I also found one person to under-respond in a similar fashion. Everyone else fit neatly within the bounds of the average change and confidence intervals of the study. Some a bit better than average, some less so.(3)
Jose: to me, the outliers have always been fascinating (especially the ones who respond well, not those who barely respond). The nature of my research can’t pin down specific mechanisms. For instance, the graph below is from one of my high protein diet studies. You will note that in general, higher protein intakes promote a loss of fat mass. But look at the individual variability (NP is normal protein; HP is high protein). Perhaps it is a mistake in logging food. Perhaps some people gain fat really easy while others lose it easily. I mean who knows? But the GENERAL effect is that eating lots of protein tends to produce a loss of fat. And that’s good enough for 75% of the population. The other 25% will respond in a manner that isn’t ‘average.’
Q #5: OK, going into pure hypothesis time. One thing we can say about super responders is they don’t appear to have say higher levels of anabolic hormones such as testosterone or IGF-1, etc, nor far superior insulin sensitivity, or other readily apparent differences between them and other people. My point being, whatever it is that differentiates these super responders from the rest of humanity is not what we may consider “the usual suspects” it seems. In general terms, if science does finally figure what combination of factors makes a super responder, what do you expect to find once they do? What do you think are the primary differences between super responders and the rest of the world?
Alan: According to work by Thalacker-Mercer et al (PMID: 19767798), it doesn’t appear that self-reported habitual dietary intake influences myofiber hypertrophy extreme responders compared to non-responders to resistance training. However, they found that training-induced changes in IL-6 and acute increases, in p70S6K activity and changes in muscle androgen receptor protein content were positively associated with muscle hypertrophy. These findings collectively indicated that intramuscular factors could be more potent mediators of muscle hypertrophy than systemic factors (i.e., pre- or post-training exercise-induced hormonal flux). Other potentially important players in the game are an abundance of microRNA , and as I previously mentioned, greater myonuclear increases via the proliferation or increased activity of satellite cells.
Conrad: It will be a very complex interaction between genes, gene expression and how they affect everything in body relative to response. Nerves and nerve impulse, hormone production, muscle fiber type, brain activity, pain tolerance, etc. etc. etc.
Jose: I think what science will find are the genes that are key for certain kinds of athletic performance. And then perhaps in utero (sounds Frankenstein-ish) alter those genes? Crazy huh!?
Q#6: Totally anecdotal of course, but do you have a story similar to mine in the intro of a person or people you have worked with or know who just astounded you with how little they had to do to be more muscular, or leaner, or stronger than everyone around them?
Alan: One particular client case stands out (we’ll call him JC), who was near the start of prep for a bodybuilding contest, and weighed 170 lbs (77.3 kg), at what I would estimate was 10% body fat. I witnessed him do 12 reps on the barbell bench press with 315 lbs (3 plates per side; 143.2 kg). His final rep was almost as clean and effortless as the first, indicating that he definitely had a few more reps in reserve before failure. This guy never trained for powerlifting. He was just freakishly strong. Healso had one of the most massive backs I’d seen on a non-pro, and even he admitted it felt like having a whole other person attached to his torso. Another example of giftedness is lifetime natural Kelechi Opara (just Google him to see what I mean), who I helped prep for one of the earliest Men’s Physique competitions. All we needed was 5 weeks to get him ready. The judges told him he was too big AND too lean for the division, and was better suited for bodybuilding. I’m not sure those criteria would still stand in the present day, where the physique competitors are just as massive, or even more massive than many of the bodyduilders were at the time.
Conrad: Too many to recount!
Jose: I’ve always been fascinated by speed whether it’s Bruce Lee’s fists or Usain Bolt’s sprint speed. Or even the speed of a person who can run a 2 and few minutes over marathon. That’s fast no matter how you cut it. I’ve worked with elite athletes like that. And the one thing they have in common is a DRIVE to train and excel. I think the gene for ‘drive’ (or motivation) is inherent as well. Coaches always say they’ll take the average talent that works hard instead of the talent that hardly works.
Q#7: One unfortunate aspect of the super responders in my view, is others tend to look at them as role models as people to emulate, when the super responder often does nothing unique or special for their training, nutrition and so forth. People always look at me sideways when I explain to them the elite bodybuilders for example don’t do anything special nor unique for their training or nutrition, and while PEDs obviously play a part, most can use the exact same PEDs as pro bodybuilders and other athletes, and look nothing like them nor perform as they do. People were shocked when they learned Usain Bolt ate pounds of Chicken McNuggets, obviously expecting him to eat some super strict diet for top athletes they were unaware of. It’s tough for people to understand their responses to exercise and nutrition are genetically determined. How do you deal with that issue when people ask about the value of hard work and consistency when genes are the ‘ghost in the machine’ they have no control over?
I tend to tell them that no matter where on the spectrum of genetics, consistency of training, nutrition, etc. will optimize their response but we all have to work with the chips we are dealt. Most people have “average” genetics which means with consistent exercise and decent nutrition will see results for their efforts, but unlikely to be the next Arnold or Bolt, but as we have no genetic test for super responder, the only way to find out is to try! Thoughts?
Alan: This reminds me of Doug Miller’s latest round of bodybuilding contest wins (here’s how he looked, feel free to ‘mire). Doug is widely regarded as one of the greatest natural bodybuilders of all time. Many consider him The GOAT. As you can imagine, there are hordes of people who vehemently doubt that he’s natural. Prior to his latest round of wins, there’s wasn’t such a huge buzz & debate over whether or not he’s natty. After he showed up and blew the competition away, there was tons of debate over the idea. People who doubted Doug’s natty status forget to realize that he’s in the low to mid 190’s fully carbed up onstage at a height of 5’9”. That plus during his 4-year hiatus from competition, he only gained 7 lbs. He’s pretty much the perfect storm of genetics & ironclad work ethic. Of course, there are folks who will insist on drugs being part of the picture, but folks who are quickest to blow the drug whistle are typically those who have not worked full-time with athletes and have had the chance to see first-hand the full range of responders, including outliers on both ends of the spectrum. I think it’s an epic waste of time & energy worrying about genetic potential or letting the idea of genetic ceilings demotivate trainees in their quest for optimizing muscular size or strength. I’ve made it a policy of mine to avoid these debates over what boils down to pure speculation. There will always be a continuum of gifted to challenged trainees, but every single one of them can make significant improvements over time with smart programming, and/or good coaching. Also, if you place limits on yourself, then those perceived limits can only work against your ultimate achievements.
Conrad: Yeah, pretty much the same strategy. Moving outside of sport, I and into the general realm I am a bit more sardonic and tell clients that those posing with shirtless avatars or bikini photos are advertising their genetics and not their knowledge. Yeah, I know it’s snarky, but there you go.
Jose: Like I’ve said before, you aren’t a slave to your genes. Everyone can make themselves smarter, faster, stronger, etc. You may not be world class, but compared to yourself, you’ll be damn good. And being consistent is the #1 factor in making improvements.
Will: Thanx for your time gents! We didn’t have the answers to the topic of super responders per se, but asking the right questions is what leads to finding the answers, and I’d say this round table discussion very much in the right direction to that end.
Alan Aragon is a nutrition researcher and educator with over 20 years of success in the field. He is known as one of the most influential figures in the fitness industry’s movement towards evidence-based information. His notable clients include Stone Cold Steve Austin, Derek Fisher, and Pete Sampras. Alan writes a monthly research review (AARR) providing cutting-edge theoretical and practical information. Alan’s work has been published in popular magazines as well as the peer-reviewed scientific literature. He co-authored Nutrient Timing Revisited, the most-viewed article in the history of the Journal of the International Society of Sports Nutrition. Alan maintains a private practice designing programs for recreational, Olympic, and professional athletes. alanaragon.com
Conrad Earnest, PhD is currently the Director of Clinical Science at Nutrabolt and a research scientist at Texas A&M University. He formerly held positions at the University of Bath (Professor), Pennington Biomedical (Associate Professor) and The Cooper Institute (Vice President of Research). His CV encompasses approximately 175 publications in peer reviewed scientific journals, including JAMA, Archives of Internal Medicine and Lancet Oncology. His work has been cited over 12,000 times in scientific articles encompassing exercise, health, nutrition and sport performance. www.conradearnest.com
Jose Antonio, Ph.D. is the CEO and co-founder of the International Society of Sports Nutrition (www.theissn.org) and is a Fellow of the National Strength and Conditioning Association. He was the 2005 recipient of the NSCA Research Achievement Award and the 2009 NSCA Educator of the Year; moreover, has published 14 books and 60 peer-reviewed scientific publications. He has extensive media experience vis a vis television, print media, and radio. He is the Editor-in-Chief of the ISSN Scoop, the ISSN’s consumer online e-magazine and the Editor-in-Chief of the Journal of the International Society of Sports Nutrition (www.jissn.com), the ISSN’s official peer-reviewed science journal. He has contributed to Ironman, Muscular Development, Muscle and Fitness, Fitness Rx Men/Women and a host of other consumer publications. Dr. Antonio completed a Ph.D. and post-doctoral research fellowship at the University of Texas Southwestern Medical Center in Dallas Texas. He is an Associate Professor at Nova Southeastern University in Exercise and Sports Science in Ft Lauderdale FL and an avid paddler. You can find his website at www.issn.net
(1) Genomic and Transcriptomic Predictors of Response Levels to Endurance Exercise Training. Sarzynski MA, Ghosh S, Bouchard C. J Physiol. 2017;595(9):2931-39. https://www.ncbi.nlm.nih.gov/pubmed/27234805
(2) No Evidence of a Common DNA Variant Profile Specific to World Class Endurance Athletes. Rankinen T, Fuku N, Wolfarth B, Wang G, Sarzynski MA, Alexeev DG, Ahmetov, II, Boulay MR, Cieszczyk P, Eynon N, Filipenko ML, Garton FC, Generozov EV, Govorun VM, Houweling PJ, Kawahara T, Kostryukova ES, Kulemin NA, Larin AK, Maciejewska-Karlowska A, Miyachi M, Muniesa CA, Murakami H, Ospanova EA, Padmanabhan S, Pavlenko AV, Pyankova ON, Santiago C, Sawczuk M, Scott RA, Uyba VV, Yvert T, Perusse L, Ghosh S, Rauramaa R, North KN, Lucia A, Pitsiladis Y, Bouchard C. PLoS One. 2016;11(1):e0147330. https://www.ncbi.nlm.nih.gov/pubmed/26824906
(3) Effect of Astaxanthin on Cycling Time Trial Performance. Earnest CP, Lupo M, White KM, Church TS. Int J Sports Med. 2011;32(11):882-8. https://www.ncbi.nlm.nih.gov/pubmed/21984399
(4) Comparison of the Atkins, Zone, Ornish, and Learn Diets for Change in Weight and Related Risk Factors among Overweight Premenopausal Women: The a to Z Weight Loss Study: A Randomized Trial. Gardner CD, Kiazand A, Alhassan S, Kim S, Stafford RS, Balise RR, Kraemer HC, King AC. JAMA. 2007;297(9):969-77. http://www.ncbi.nlm.nih.gov/pubmed/17341711
(5) Comparison of the Atkins, Ornish, Weight Watchers, and Zone Diets for Weight Loss and Heart Disease Risk Reduction: A Randomized Trial. Dansinger ML, Gleason JA, Griffith JL, Selker HP, Schaefer EJ. JAMA. 2005;293(1):43-53. http://www.ncbi.nlm.nih.gov/pubmed/15632335