Correspondence  |   March 2019
Improving External Validity May Jeopardize Internal Validity
Author Notes
  • Prince Sultan Military Medical City, Riyadh, Kingdom of Saudi Arabia.
  • (Accepted for publication November 19, 2018.)
    (Accepted for publication November 19, 2018.)×
Article Information
Correspondence   |   March 2019
Improving External Validity May Jeopardize Internal Validity
Anesthesiology 3 2019, Vol.130, 508-509. doi:10.1097/ALN.0000000000002575
Anesthesiology 3 2019, Vol.130, 508-509. doi:10.1097/ALN.0000000000002575
I read with interest the recently published article by Ramgolam et al.1  that sought to find the safer way to induce anesthesia in children at risk of developing perioperative respiratory complications. I would like to congratulate the authors for their tenacity to complete the study despite the obstacles in the recruitment of the staff and patients. However, I have few concerns regarding the current study by Ramgolam et al.
To preserve internal validity of controlled trials and eliminate confounders, all participants in each group should receive the same treatment, and all the groups should be treated equally apart from the intervention.2  Unfortunately, this was violated multiple times in the current study by Ramgolam et al. The cause of this violation is not clear. It might be an attempt to make the setting more natural to improve the external validity, or it might be because of the recruitment of new staff to complete the study. For example, a drop of oxygen saturation less than 95% was one of the outcomes; however, the fraction of inspired oxygen was variable in the inhalational group, nitrous oxide was used in half of the patients in the inhalational group, and preoxygenation was not routine in the intravenous group. Another example was that anesthesiologists were free to administer propofol in the inhalational group. Forty-nine percent of patients in the inhalational group received propofol in a dosage that is roughly equal to one third of the dose of propofol in the intravenous group. The rationale for administering propofol was not mentioned. Propofol might be administered because of the fear of, or the actual, light anesthesia in the inhalational group. This light anesthesia in the inhalational group and not the inhalational induction may be the cause of the perioperative respiratory adverse events. Patient who received propofol in the inhalational group (with possible light anesthesia) had more perioperative respiratory adverse events (49%) compared with those who did not receive propofol (39%). Although the post hoc analysis demonstrated that this difference was not statistically significant, this statistical insignificance may be unreliable because the subgroup analysis was underpowered. The current study by Ramgolam et al. was powered to determine the difference in the incidence of perioperative respiratory adverse events between children receiving an inhalational induction and an intravenous induction. For comparison of subgroups of the same size and with the same power as the overall effect, the sample sizes should be inflated fourfold.3  It would be more appropriate if a specific minimum alveolar concentration value was targeted that was equipotent to the dose of propofol given in the intravenous group. Also, targeting a bispectral index value might be used to ensure equal depth of anesthesia in all participants.