Statistical Sleuthing for Marathon Fueling

Inconsistencies and problems found in Ravikanti S et al. J Appl Physiol. 2025;139:1581–95. 

Final Version of Paper 

Article in Press (AIP)

Data extracted from the figures using graph2table.

Table 1. Running Economy Descriptive Statistics: Inconsistencies across versions 

Table 2. Running Economy Differences in Means: Inconsistencies across versions (AIP vs final) and between text and figures

Table 3. Running Economy Omnibus p-values: Inconsistencies across versions and within text

*The “alternate omnibus p-value” corresponds to the p-value for running economy reported within the parenthetical list of p-values for multiple outcomes in the Results text: "Across the whole 2 h period, exercise significantly increased heart rate, blood lactate concentration, RPE, and impaired running economy. No differences were observed between CHO trials for these variables (P = 0.73, 0.11, 0.79, 0.15).”

Table 4. Running Economy Post-hoc p-value for 60 vs 120: Inconsistencies across versions and between text and abstract

Table 5. Running Economy Post-hoc Results: Inconsistencies in the p-values and confidence intervals, as well as mismatches between corresponding p-values and confidence intervals 

Table 6. Total CHO and Total Fat Oxidation: Inconsistencies between text and figures

Total CHO Oxidation (grams)

Total Fat Oxidation (grams)

Table 7. CHO Oxidation Rate (g/min) Standard Deviations: Inconsistencies across versions and between text, abstract, and figure

Table 8. Peak Exogenous CHO Oxidation (g/min): The reported difference doesn’t match the difference of the reported means

Table 9. Peak Exogenous CHO Oxidation (g/min): Slight inconsistencies in text and figure

Table 10. Mean 2-hour Exogenous CHO Oxidation (g/min): Values reported in text/figure 5C do not match the value shown in Figure 5B. We confirmed that these discrepancies are not due to rounding or extraction error.

Inconsistencies with the constant used for calculating substrate-specific energy expenditure

Fat energy expenditure (kJ/min) appears to be calculated by multiplying fat oxidation rate (g/min) by a constant energy equivalent (~39.4 kJ/g). This relationship holds closely across most time points in all three conditions (60 g, 90 g, 120 g). However, several time points deviate materially from this constant:

• 60 g condition:
  
  

• 90 min: implied constant ≈ 43.6 kJ/g
  
  

• 105 min: implied constant ≈ 43.1 kJ/g
  
  

• 120 min: implied constant ≈ 42.9 kJ/g
  
  

• 120 g condition:
  
  

• 105 min: implied constant ≈ 41.1 kJ/g
  
  

• 120 min: implied constant ≈ 44.2 kJ/g
  
  

These deviations are larger than would be expected from rounding or digitization error, suggesting inconsistent application of the conversion factor at these time points.

Odds ratios reported as infinity

Several odds ratios are reported as infinity. This implies complete separation or zero cells in the underlying contingency tables. In this situation, the odds ratio is not estimable and should not be reported.

Misleading Title

The title of the paper: “Carbon-13 Labeled Glucose Fructose Shows Greater Exogenous and Whole Body Carbohydrate Oxidation and Lower Oxygen Cost of Running at 120 Versus 60 and 90 Grams per Hour in Elite Male Runners,” implies that the 120 condition beat the 60 and 90 gram conditions on oxygen cost of running, which is running economy. But that is inaccurate. For running economy (oxygen cost of running), the 120 condition was significantly better than the 60 condition, but not the 90 condition.