Arya M. Sharma, MD

One of the questions many readers and colleagues have asked, is whether or not the issue of ‘causality’ should matter in the the context of the Edmonton Obesity Staging System (EOSS).

In other words, should only conditions be counted that are ‘causally’ linked to obesity or is it enough that these conditional are merely more common in people with excess weight.

There are essentially two important but distinct aspects to this question that ultimately relate to how EOSS is to be used in clinical practice.

If the primary purpose of EOSS is to identify patients who would benefit from weight loss, then, yes, it matters whether or not the co-morbidities considered, are ‘causally’ related to obesity and can be reversed or ameliorated by reducing and sustaining a lower body weight.

However, if the primary purpose of EOSS is merely to identify obese patients, who are at high risk and need to be prioritized within the health care system in order to receive the appropriate care for their conditions (irrespective of whether or not this ‘care’ involves weight loss or just better management of their comorbidities), then the question of causality is really irrelevant.

Thus, in the first case, one would only count ‘comorbidities’ that are actually ‘causally’ related to excess weight - an example being sleep apnea. There is ample evidence that weight loss reduces symptoms of sleep apnea (while weight gain makes it worse) and so sleep apnea would count as an important comorbidity that can be addressed by obesity treatment.

In the second case, it does not actually matter if the comorbidity is in any way related to excess weight. All that really matters, is whether this comorbidity is present or not. An e.g. would be depression, which, while not caused by obesity and not likely to improve with weight loss (it may sometimes even get worse), may, when present, help identify obese patients, who do have a higher risk of premature death.

So while in the first example, EOSS would be used to decide who needs to lose weight, in the second example, EOSS simple serves to identify obese people, who are at highest risk of complications and death.

Apart from the second scenario being the real reason that EOSS was developed, it is also a far more practical approach to using EOSS, because for many comorbidities it may be impossible to answer the ‘chicken or egg’ question or even determine if these are simply two different chickens.

In clinical medicine this phenomenon is referred to as ‘phenocopy’, a term used to describe a case where two distinct and unrelated conditions, present clinically with the same symptoms or ‘phenotype’.

In obesity, this is particularly common, because, while many symptoms may be ‘causally’ related to obesity, these same symptoms may just happen to present in an obese individual but have nothing to do with that patient’s excess weight.

For EOSS, this question would not really matter - whether the knee pain is from an accident or from carrying around the excess weight makes no difference - the only thing that matters is that this is a patient with excess weight AND knee pain and therefore, this patient is an obese patient, who is at higher risk for mortality than an obese patient without knee pain.

Remember, bariatric care, as I define it, is not about losing weight but rather about medical care for the bariatric patient. As resources are limited, all I want EOSS to tell me is, who to see first.

Of course there are ‘normal-weight’ people with knee pain, but they are not my problem. There are also ‘normal-weight’ people with high blood pressure, sleep apnea, diabetes and depression - again, they are not my problem.

As someone, who works in an obesity clinic, my job is to assess obese patients and help them the best I can, whether their treatment requires weight loss or not. If EOSS can help me decide, who to see most urgently, then EOSS has done its job.

Of course, further research is needed to determine whether EOSS actually works well in clinical practice (e.g. outside of a speciality centre) and we may perhaps need to simplify and clarify the criteria. But the principle stands: it is simply not enough to look at BMI to decide who needs (urgent) medical attention and who does not.

AMS
Edmonton, Alberta

As anyone watching the news over the past 24 hours will probably have noticed, yesterday saw the release of two large studies looking at whether or not the Edmonton Obesity Staging System is a better predictor of mortality risk than BMI alone - the short answer is “yes”.

The results from these studies was reported by all major media outlets around the globe including CNN, TIME, MSNBC, and virtually all national print, radio, and tv stations. It was also picked up by the blogsphere.

Interestingly, when I first suggested the use of a staging system for obesity in 2008, it was born out of both a practical need and my own medical experience with obese patients.

The practical need was to better determine, who needs to be seen in our obesity program, given the rather limited resources and long waiting times.

My medical experience had long taught me that the commonly used BMI classification of obesity, or even the suggested use of waist circumference were rather blunt instruments in determining which patients needed obesity treatments and which did not.

So, the idea was simply to create a clinical tool that would help us decide, which obese patients required our attention most urgently.

However, as readers will imagine, with all the talk about ‘healthy” weights and ‘benefits’ of weight loss, our proposal was met with considerable scepticism - not about whether or not obese people with obesity related health problems (EOSS 2-4) needed treatment, but rather whether or not obese people who appeared pretty healthy (EOSS 0/1) were indeed at a low risk from their excess weight.

So we looked for large datasets in which we could apply EOSS and compare it to BMI in predicting death.

In one collaboration, on which my colleague Raj Padwal took the lead, assisted by David Allison and Nicholas Pajewski from the University of Birmingham, Alabama, we looked at the impact of EOSS on mortality in two separate sets of the NHANES study - a representative sample of the US population. The results of this analysis were published yesterday in the Canadian Medical Association Journal.

In another collaboration, on which Jennifer Kuk and Chris Ardern (York University, Toronto) took the lead with help from Timothy Church (Pennington Biomedical Research Center, Baton Rouge, LA), and Xuemi Sui and Steven Blair (University of South Carolina), we looked at the impact of EOSS on mortality in the Aerobics Center Longitudinal Study (n = 29 533). The results were released yesterday in the Applied Physiology, Nutrition & Metabolism.

In today’s post, I will not go into details of the studies or begin a lengthy discussion of the findings - suffice it to say, till someone comes up with an even better way to clinically assess the health status of obese patients, to help decide who does and who does not need obesity treatment, the Edmonton Obesity Staging System may be just the tool that clinicians, payers, and patients have been looking for, to help dispel the notion that health can be measure by simply stepping on a scale.

AMS
Edmonton, Alberta

Long-time readers of these pages may recall that in 2009, Robert Kushner and I published a proposal for a new clinical obesity staging system in the International Journal of Obesity.

Rather than BMI (a measure of weight), the Edmonton Obesity Staging System (EOSS) ranks severity of obesity based on clinical assessment of weight-related health problems, mental health and quality of life. We proposed that this system would provide a far better guide to clinical decision making than using BMI class alone.

For reasons that will become apparent later this week, I would like to repost an excerpt from this original proposal for this system, which was first posted on this blog on March 30, 2008:

Current definitions of obesity based on BMI and waist circumference (WC), while widely accepted, are hardly helpful in counseling individual patients. Readers of my blog are probably quite familiar with my views on this.

As most clinicians will readily agree, when dealing with indiviual patients, both measures lack sensitivity and specificity with regard to identifying the presence or risk of obesity-related risk factors, comorbidities, psychopathology, global functioning or quality of life.

In fact recent epidemiological studies emphasize that good health including low morbidity and mortality is possible over a wide range of BMI. Thus, basing the decision on who to treat and who to leave well alone solely on measures of weight or size is neither sensible nor does justice to the complexity of the relationship between excess body fat and its impact on health and well-being. The well-established obesity-chronic disease paradox makes decisions on who to treat and who not to treat even more uncertain.

Telling healthy large people who have no apparent comorbidities, functional limitations or reduced well-being to lose weight may be counterproductive in that it can introduce and reinforce dissatisfaction with body image, foster frustrations and despair (given the poor long-term success of weight loss attempts) and lead to unhealthy behaviours focusing on weight loss (e.g. excessive exercise or dieting) rather than on healthy lifestyles (which are possible at almost any weight).

Thus, for practical purposes, it is important to move beyond defining who needs obesity treatment simply based on BMI and/or WC to a more clinically meaningful system.

Indeed, what we direly need is a classification of obesity that is clinically relevant in that it helps identify patients who have or are at high-risk of obesity-related complications and are most likely to benefit from treatment.

…..

Now I am no expert on disease classification and realise the large amount of work and consensus meetings that go into developing these classification systems. But I am a clinician, who regularly sees patients and would be happy to see even the simplest form of staging that provides a meaningful framework.

The simplest classification I can think of would be to use a staging system similar to the following:

Stage 0: no apparent obesity-related risk factors (blood pressure, lipids, glucose, etc.), physical symptoms, psychopathology, functional limitations, or impairment of well-being

Stage 1: presence of obesity-related sub-clinical risk factors (elevated blood pressure, impaired fasting glucose, fatty liver, etc.), mild physical symptoms (dyspnea on moderate exertion, occasional aches and pains, etc.), mild psychopathology, mild functional limitations or mild impairment of well-being

Stage 2: presence of established obesity-related chronic disease like hypertension, type 2 diabetes, sleep apnea, osteoarthritis, reflux disease, polycystic ovary syndrome, depression, anxiety disorder, moderate limitations in activities of daily living and/or well being.

Stage 3: established end-organ damage like myocardial infarction, diabetic complications, severe osteoarthritis, significant psychopathology, significant functional limitations and impairment of well-being

Stage 4: severe (end-stage?) disabilities from obesity-related chronic disease, severe disabling psychopathology, severe functional limitations and severe impairment of well-being

Thus, for e.g., a 24 year-old physically active female with a BMI of 32 with no measurable risk factors, functional limitations or self-esteem issues would have Class I, Stage 0 Obesity - benefits of treatment will be marginal or non-existent.

A 32 year-old male with BMI of 36 with hypertension and sleep apnea would have Class III, Stage 2 Obesity - definite indication for obesity treatment.

A 45 year-old female with BMI of 54 who is in a wheel chair because of severe gonarthritis with severe hypoventilaltion would have Class III, Stage 4 Obesity - will require aggressive obesity treatment unless deemed palliative.

Stay tuned for more on the Edmonton Obesity Staging System this week.

AMS
Edmonton, Alberta

Sharma AM, & Kushner RF (2009). A proposed clinical staging system for obesity. International journal of obesity (2005), 33 (3), 289-95 PMID: 19188927

While I am on a brief holiday in Berlin, I thought I’d rerun a few earlier posts that discuss the issue of measuring obesity and how such measures may or may not be helpful in obesity management - as many readers may not have seen these posts before, comments are very much appreciated.

The following was first posted on November 18, 2008

This week’s New England Journal of Medicine, features an article by Tobias Pischon on behalf of the EPIC (European Prospective Investigation into Cancer and Nutrition) investigators on the relationship between BMI, waist circumference, waist-to-hip ratio and mortality.

To me, this paper is of considerable interest - not least, because Tobias was one of my students back in Germany, who did his MD thesis on the effect of salt intake and obesity on chronic kidney transplant rejection under my supervision.

Of course, this paper also deals with a topic that I have often blogged about - i.e. the relationship between anthropometric measures and morbidity and mortality.

Pischon and colleagues studied 359,387 participants from nine countries during a mean follow-up of 9.7 years. After adjustments for age, educational level, smoking status, alcohol consumption, and physical activity, the lowest risks of death related to BMI were observed at a BMI of 25.3 for men and 24.3 for women.

After adjustment for BMI, relative risks among men and women in the highest quintile of waist circumference were 2.05 and 1.78, respectively, and in the highest quintile of waist-to-hip ratio, the relative risks were 1.68 and 1.51, respectively.

BMI remained significantly associated with the risk of death in models that included waist circumference or waist-to-hip ratio (P<0.001).

This study, essentially confirms what was already known, namely that the impact of excess body fat on mortality depends not only on the amount of excess fat (BMI) but also on its distribution (waist circumerence, waist-to-hip ratio).

Importantly, the measures of fat distribution are predictive of risk even in normal weight individuals with lower BMI’s, which challenges the use of cutoff points to define abdominal obesity.

On the other hand, as BMI increases, measuring fat distribution adds little to determining obesity-related risk. (which is why obesity guidelines do not recommend measuring waist cirumference in individuals with a BMI>40).

AMS
Edmonton, Alberta

While I am on a brief holiday in Berlin, I thought I’d rerun a few earlier posts that discuss the issue of measuring obesity and how such measures may or may not be helpful in obesity management - as many readers may not have seen these posts before, comments are very much appreciated.

The following was first posted on October 31, 2008

As most readers of these pages probably know, the current definition of obesity is based on the body-mass-index, a number, which, in populations, nicely correlates with body fat.

However, as body fat alone is not the entire picture, other indices that include measurements of fat distribution such as waist circumference or wait-to-hip ration have been suggested, along with cut-offs that would help identify, who is “obese” and who is not.

But how reliably can these indices be measured in clinical practice (as recommended in obesity guidelines)?

This question was addressed by Paul Sebo and colleagues from the Geneva University Hospitals, Switzerland, in a paper just published in Preventive Medicine.

In this study, repeated anthropometric measurements were performed by 12 primary care physicians on 24 adult volunteers, men and women, with an average BMI of 28. While inter-observer reliability for weight, height, and derived BMI were excellent (R>0.99), they were unsatisfactory for waist circumference (R=0.92), hip circumference (R=0.76) and waist-to-hip-ration (R=0.51).

With BMI, only 1% of the volunteers were misclassified as overweight or obese, whereas the use of WC and WHR lead to misclassification in 6% and 23% respectively.

Reliability for the measurements improved after a one-hour training in anthropometric measurements, but the proportion who were misclassified remained high for WC (5%) and WHR (9%).

So, apparently, even with “Swiss precision”, anything that goes beyond height and weight is too complicated to reliably classify obesity in primary care.

But the real question here is whether or not ANY anthropometric measurement can reliably detect who is threatened of affected by excess weight (my definition of obesity). I have argued before that BMI, although fine for population studies, is not useful when making individual decisions about patients.

Not only is there a wide range in individual variability in the actual body fat present in individuals of the same BMI, but, more importantly, there is a huge variability on how that excess fat actually affects that individual’s health.

We have recently proposed the Edmonton Obesity Staging System, which we now use to supplement BMI measurements with stages that reflect the degree of comorbidity and/or reduction in functional status attributable to the excess weight.

Blindly basing decisions to treat or not-to-treat on BMI alone will result in treating a lot of people who have little to gain, while missing out on many who are clearly threatened or affected by excess body fat.

AMS
Edmonton, Alberta