Endocrine Feedback Loop - Episode 58

BMI vs Percent Body Fat in Defining Obesity
Date: February 20, 2025
Host: Dr. Chase Hendrickson (Vanderbilt University)
Panel: Dr. Andrew Craftson (University of Michigan) & Dr. Beverly Chang (Weill Cornell Medicine)

Episode Overview

This episode examines a forthcoming article from the Journal of Clinical Endocrinology & Metabolism that explores the relationship between Body Mass Index (BMI) and percent body fat in defining obesity. The discussion centers on whether percent body fat should replace BMI in clinical practice, analyzes the new study’s methods and findings, and explores the broader implications for obesity medicine.

Key Discussion Points & Insights

1. Historical Context and Limitations of BMI

[02:43–10:19]

• BMI as an Epidemiological Tool:
  BMI (body mass index) has been the primary standard for classifying overweight and obesity for decades but was originally designed for population-level risk stratification, not individual assessments.

  “BMI was originally designed as an epidemiologic screening tool to risk stratify populations. Not necessarily the individual patient you see in front of you in clinic…” – Dr. Beverly Chang [06:50]

• Correlations and Shortcomings:
  Higher BMI correlates strongly with risks for conditions like diabetes, hypertension, and heart disease, but it is an imperfect proxy for actual fat mass.

  • BMI encompasses total body weight (including bone, muscle, water, organs—not just fat).
  • Athletes, for example, may have a high BMI due to muscle, not excess adiposity.
  • Some individuals have normal BMI but high body fat (“normal weight obesity”), potentially increasing risk not flagged by BMI.

• Obesity as a Disease:
  There’s an ongoing shift from viewing obesity as a personal responsibility to recognizing it as a chronic, biologically-based disease.

  • The limitations of BMI highlight the need for better tools that capture fat distribution and quality, not just quantity.

  "BMI does not estimate [adipose quantity, functionality, distribution]... It does estimate it in some surrogate ways that helps lead us to this current discussion." – Dr. Andrew Craftson [10:19]

2. Study Design and Methodological Overview

[13:02–22:15]

• Study Basis:
  The discussed study is a cross-sectional correlation analysis using NHANES data from 1999–2018, with participants aged 18–85.

• Measuring Adiposity:

  • DEXA (Dual-Energy X-ray Absorptiometry):
    Used as the gold standard for total and regional body fat measurement in this study; considered superior to techniques like skin calipers or BIA (bioelectrical impedance analysis) for body composition.

    "DEXA... has been considered our gold standard for a long time for measuring percent body fat because it's easy to use or easy to administer rather, and it's a low amount of radiation..." – Dr. Beverly Chang [16:36]

  • Bioelectrical impedance is more accessible for general clinicians but less precise.

• Defining Metabolic Syndrome (Key Variables):

  • Waist circumference, HDL cholesterol, fasting glucose, blood pressure, triglycerides.

• Anchoring Cutoffs:
  The key critique: the study derives percent body fat thresholds for overweight/obesity by mapping back to existing BMI cutoffs (e.g., what percent body fat corresponds to BMI cutoffs for risk of metabolic syndrome).

  "They are still anchoring their cutoffs to our current BMI thresholds. And I think... it makes their point moot as far as trying to get away from BMI." – Dr. Beverly Chang [21:03]

3. Results & Statistical Findings

[23:41–31:32]

• Study Population:

  • ~17,000 participants (roughly equal men/women); mean age ~42 years.
  • Well-representative demographically of the US.
  • Average body fat: Men ~28%; Women ~40%.

• Key Findings on Risk:

  • Overweight (by BMI): 5% risk of metabolic syndrome.
  • Obese (by BMI): 35% risk.

• Translating to Body Fat %:

  • For a 5% risk (“overweight”): Men = 25% body fat; Women = 36%.
  • For a 35% risk (“obesity”): Men = 30%; Women = 42%.

• Predictive Validity:

  • Receiver Operating Characteristic (AUC) for metabolic syndrome:
    • BMI: 0.83 (men), 0.75 (women)
    • Percent body fat: 0.80 (men), 0.71 (women)
      (Similar performance at the population level.)

• BMI vs. Percent Body Fat - Misclassification:

  • Women: Notably, many women with “normal” BMI can actually have high body fat and therefore higher risk (“BMI misses high-fat women”).
  • Men: Some men with high BMI may actually have low body fat (often muscular), meaning BMI can “over-treat” men.

  "There are all these individuals who do not quite fit as a good correlation...by BMI they should be low risk, but by body fat they're actually high risk." – Dr. Andrew Craftson [29:20]

• Correlation Analysis:

  • The scatter plot shows a wide, non-linear (curvilinear) relationship between BMI and body fat %, especially at the extremes.
  • Authors highlight large variability (“large scatter”)—BMI is less predictive at individual level.

4. Discussion of Implications and Critique

[31:32–39:48]

• Study Conclusions (Authors):

  • Percent body fat offers alternative thresholds (25/30% for men, 36/42% for women) to define overweight and obesity.
  • Advocates for moving from “anthropometric estimates” (BMI) to “direct measurement” of adiposity (body fat %).

• Panelists’ Critique & Reservations:

  • Anchoring Flaw: Because percent body fat thresholds are constructed by mapping onto BMI/metabolic syndrome risk, they are not truly independent.

    "Given that this report uses BMI to anchor those definitions of obesity and overweight via body fat... I don't think we can define overweight and obesity by percent body fat instead of BMI with this report." – Dr. Beverly Chang [38:56]

  • Measurement Variability: Small differences in %fat (e.g., 5%) may be within the margin of error for body composition methods (affected by hydration, variance by machine, etc.).

    “[Small differences] can see that difference from one measurement to the next, depending on a person's water content... hydration status actually.” – Dr. Beverly Chang [33:17]

  • Visual Statistics Critique: The study relies heavily on visual inspection of scatter plots rather than rigorous quantification.

    “Some of the conclusions drawn being derived from derived primarily from a visual interpretation of a scatter plot.” – Dr. Andrew Craftson [37:52]

  • Clinical Real-World Application: There’s skepticism that DEXA or even percent body fat estimations will replace BMI:

    • Cost, time, and accessibility of DEXA are prohibitive for routine primary care.
    • Simple measures like waist circumference, though imperfect, are easier and potentially as helpful.

    "For some patients, doing waist circumference is uncomfortable. For some people, they're not trained well... And certainly the time that it would take to do a DEXA that in addition to the cost, doesn't necessarily warrant widespread implementation." – Dr. Andrew Craftson [40:18]

5. Broader Debate: Redefining Obesity and What’s Next

[39:48–42:51]

• Current Trends:
  The field debates how best to combine weight, body fat, and other markers (waist, waist:hip, BIA, DEXA) to meaningfully assess risk and track treatment.

• Moving Forward:
  Both panelists highlight the necessity for prospective, longitudinal research that directly links percent body fat to health outcomes—independent of BMI—to inform meaningful clinical thresholds.

  “I want to see real outcome data, real longitudinal outcome data where we enroll that one person, thousands of people, but we follow that same person over the decades, measuring their body fat, assessing their risk and incident of diseases that would be able to inform us risk thresholds with body fat percentage without relying on vmi.” – Dr. Beverly Chang [42:21]

Notable Quotes & Memorable Moments

• Dr. Beverly Chang on the Catch-22 of Moving from BMI:

  “Moving away from BMI is very challenging because... how do we interpret a body fat percentage number? How do we interpret a waist circumference? We don't know yet.” [07:50]

• Dr. Andrew Craftson on Clinical Practice:

  “In a busy clinic... you don't have a lot of time to devote to obesity care, maybe using BMI a lot to make some reflexive decisions. But for those who are having a concentrated visit on obesity medicine, one wonders how much of an advantage the imaging would have over your own clinical assessment of their body habitus.” [35:12]

• Dr. Beverly Chang on Study’s Limitation as Practice-Changer:

  “It's definitely not the study that would change clinical practice.” [39:42]

Timestamps for Important Segments

• [02:43–10:19] — History and critique of BMI as an obesity metric.
• [13:02–22:15] — Methods: NHANES dataset, DEXA, definitions, and statistical approach.
• [23:41–31:32] — Study results: sample, group comparisons, risk thresholds, and correlation findings.
• [31:32–39:48] — Implications, panelist critique, clinical relevance.
• [39:48–42:51] — Broader debate and next steps for obesity definition and research.

Conclusions

• Despite longstanding criticism of BMI, this study does not provide a definitive answer for replacing it with percent body fat in clinical practice.
• Percent body fat, while more directly tied to adiposity, introduces its own challenges—accuracy, accessibility, and lack of clear, risk-based thresholds.
• Both BMI and percent body fat show similar predictive value for metabolic syndrome at the population level, but individual misclassification remains a problem with both methods.
• Substantial, prospective outcome-based research is needed before guidelines can recommend a shift away from BMI toward direct measures of body fat for defining obesity or guiding therapy.