Objective: The optimal amount of calories required by critically ill patients continues to be controversial. The objective of this study is to examine the relationship between the amount of calories administered and mortality.
Design: Prospective, multi-institutional audit.
Setting: Three hundred fifty-two intensive care units from 33 countries.
Patients: A total of 7,872 mechanically ventilated, critically ill patients who remained in the intensive care unit for at least 96 hrs.
Interventions: None.
Measurements and main results: We evaluated the association between the amount of calories received and 60-day hospital mortality using various sample restriction and statistical adjustment techniques and demonstrate the influence of the analytic approach on the results. In the initial unadjusted analysis, we observe a significant association between increased caloric intake and increased mortality (odds ratio 1.28; 95% confidence interval 1.12-1.48 for patients receiving more than two-thirds of their caloric prescription vs. those receiving less than one-third of their prescription). Excluding days after permanent progression to oral intake attenuated the estimates of harm (unadjusted analysis: odds ratio 1.04; 95% confidence interval 0.90-1.20). Restricting the analysis to patients with at least 4 days in the intensive care unit before progression to oral intake and excluding days of observation after progression to oral intake resulted in a significant benefit to increased caloric intake (unadjusted odds ratio 0.73; 95% confidence interval 0.63-0.85). When further adjusting for both evaluable days and other important covariates, patients who received more than two-thirds of their caloric prescription are much less likely to die than those receiving less than one-third of their prescription (odds ratio 0.67; 95% confidence interval 0.56-0.79; p < .0001). When treated as a continuous variable, the overall association between the percent of the caloric prescription received and mortality is highly statistically significant with increasing calories associated with decreasing mortality (p < .0001).
Conclusions: The estimated association between the amount of calories and mortality is significantly influenced by the statistical methodology used. The most appropriate available analyses suggest that attempting to meet caloric targets may be associated with improved clinical outcomes in critically ill patients.