Fasting Duration and Blood Pressure in Children: Comment.

Abstract

We read with interest the article by Simpao et al.1 regarding the association between clear fluid fasting duration and postinduction low blood pressure in anesthetized children. In this retrospective analysis, the authors reported that, among children (0 to 18 yr) who underwent inhalational induction of anesthesia for elective surgical procedures, longer duration of clear fluid fasting was associated with increased risk of postinduction low blood pressure during the surgical preparation. Despite the clinical relevance of the topic and the elegant study design, certain methodologic issues require clarification.First, although the exposure (fasting duration) was collected as a continuous variable, the authors transformed this continuous variable into categorical groups using some clinically useful, though arbitrary, cutpoints. The statistical limitations of this approach must be highlighted. As elegantly assessed by the authors, the association between clear fasting time and low blood pressure was not linear. In this setting, percentile categorizations can misrepresent the dose–response relationship between the exposure and outcome because instead of accounting for the nonlinearity, the cutpoints are merely identified according to the distribution of the primary predictor.2,3 This may result in the lumping together of subjects with different risks of low blood pressure, thus violating the assumption of no differences in risk of the outcome between groups. For instance, children in the 6- to 8-h group had a 22% (1.55/1.27) relative higher odds of low blood pressure, relative to children in the 4- to 6-h group, yet both groups have been lumped into the 4- to 8-h category. Similarly, children in the 10- to 12-h group had a 17% (1.16/0.99) relative higher odds of low blood pressure, relative to children in the 8- to 10-h group, yet both groups have been lumped into the 8- to 12-h category. These departures of 22% and 17% appear to be meaningful, given the context of the study, because the highest relative excess in the odds ratio was 33%. In addition, the authors adopted an open-ended categorization of patients with clear fasting time greater than 12 h (alternatively greater than 14 h in the sensitivity analysis). This cutpoint of 12 h (14 h in the sensitivity analysis) may be too far from the most extreme value and may hide important effects. For example, the risk of low blood pressure among children with clear fasting time greater than 18 h (corresponding to about 2.5% of the study population) is unknown because it was averaged with those of the other children in the greater than 12-h group.Second, it appears that about 60% of the study cohort received coinduction of anesthesia with propofol or an intravenous opioid, and 12% of subjects received a neuraxial block (type not stated). Given that propofol coinduction and neuraxial anesthesia are known causes of hypotension under anesthesia, the observed associations do not disambiguate the effect of fasting from the expected hypotensive effects of propofol and/or neuraxial anesthesia. Therefore, the interpretation of the authors’ findings would benefit from a sensitivity analysis by separating patients who received propofol from their counterparts who did not receive propofol.In conclusion, we applaud Simpao et al. for adding to the literature on the association of preoperative fasting with intraoperative hypotension. However, the limitations of their exposure and outcome variables must be considered when interpreting the findings of their report.The authors declare no competing interests.