Because of potential for mathematical coupling of measurement errors in shared variables used to calculate oxygen consumption (FickVO2) and oxygen delivery (DO2), we asked whether determination of the FickVO2-DO2 relationship in individual patients with ARDS was statistically valid. We studied 17 clinically resuscitated patients with severe ARDS, measuring FickVO2, CalorimetricVO2 (using analysis of respiratory gases), and DO2 at regular intervals while DO2 was increased using an infusion of dobutamine. Overall, we found that DO2 (pre 482 +/- 143, post 616 +/- 170 ml O2/min.m2, p < 0.01) and FickVO2 (pre 130 +/- 23, post 147 +/- 24 ml O2/min.m2, p < 0.02) increased significantly with dobutamine infusion, but CalorimetricVO2 measured simultaneously did not change (pre 128 +/- 22, post 128 +/- 22 ml O2/min x m2, p = NS). In addition, unpooled weighted slope for FickVO2 versus DO2 (0.06) was significantly different from zero, but unpooled weighted slope for CalorimetricVO2 versus DO2 (0.01) was not significantly different from zero. Slopes of the FickVO2-DO2 relationship were significant for only three individual patients. Using methods by Stratton and colleagues to analyze the effect of mathematical coupling in the FickVO2-DO2 relationship, we found that in all patients the slope of measurement errors was greater than observed slope and that observed slope was greater than estimated true slope. Estimated true slope of the FickVO2-DO2 relationship in all individual patients was not significant. Therefore, we suggest that determination of the FickVO2-DO2 relationship in individual patients who are resuscitated and hemodynamically stable is most often not statistically significant.(ABSTRACT TRUNCATED AT 250 WORDS)