Editor, We thank Wan et al.1 for their interest in our cohort study on postoperative delirium in patients admitted to the ICU after intracranial surgery2 and appreciate the opportunity to respond. For the diagnosis of postoperative delirium, the Diagnostic and Statistical Manual of Mental Disorders (DSM), fifth edition nomenclature is recommended.3 However, to assist nonpsychiatrically trained clinicians to identify delirium quickly and accurately, several DSM-based instruments have been developed, including the Confusion Assessment Method (CAM)4 and CAM for the ICU (CAM-ICU).5 We used the CAM-ICU for several reasons. All enrolled patients were transferred to the ICU after craniotomy, and all assessments of postoperative delirium were performed in the ICU on postoperative day 1. In our cohort, 2.2% of the patients (18/800, Table 4) were mechanically ventilated when assessing delirium on postoperative day 1, but 27.5% of the patients (220/800, unpublished data used in immobilising factor score) were intubated. As the most widely recognised screening scale of postoperative delirium in the ICU, CAM-ICU is recommended by the American Society of Critical Care Medicine as a delirium assessment tool in adult patients in the ICU6 and is suggested by the European Society of Anaesthesiology as a postoperative delirium screening tool in postoperative patients.7 Therefore, in many cohort and controlled studies, CAM-ICU has been employed as an effective tool for the assessment of postoperative delirium in the ICU.8–11 We agree with Wan et al. that in our study, the incidence of postoperative delirium may have been confounded by the assessment tool and the time window for the monitoring. As we discussed in our article, although previous studies reported that the incidence of postoperative delirium decreased in a time-dependent manner, delirium may occur from postoperative day 4 to day 7 (but less than 5%).11,12 Furthermore, it has been reported that, compared with the Diagnostic and Statistical Manual of the American Psychiatric Association, the sensitivity of CAM-ICU to detect delirium is relatively low in the recovery room setting.7 All these factors may have caused an underestimation of the incidence of postoperative delirium in our cohort. Therefore, our results still indicate the importance of postoperative delirium in the population we studied. Regarding the analysis of risk factors for postoperative delirium, we did follow the overall strategy univariate analyses first (Tables 1 to 4), followed by multivariate analyses (Table 5). Wan et al. may have some misunderstanding of univariate analysis. The main purpose of the univariate analysis is to identify candidate risk factors or covariates, rather than to estimate the size of the effect [the odds ratio (OR) and its 95% confidence interval (CI) were calculated in our study]. Therefore, we can either examine the distribution of variables between the groups or use one independent variable at a time to fit a regression model to identify potential risk factors. We prefer the former because this form of data is more informative with the summary and report of descriptive statistics and the P value in statistical tests. As for the estimation of the effect size, we reported the original and adjusted ORs and 95% CIs of the risk factors in the multivariate logistic regression model in Table 5. These factors are more valuable for preventing postoperative delirium. For more detailed information about our statistical strategy, please refer to other reports.13,14 We used backward stepwise variable selection in multivariate logistic regression because this method can obtain a performance similar to the Bootstrap method, but it is easy to perform.15 We agree with Wan et al. that a number of demographic and peri-operative factors may lead to adverse postoperative outcomes. That is why we did not take the causal relationship between postoperative delirium and adverse consequences as the end point of the study. As we pointed out in the article,2 our data show the potential association between postoperative delirium and adverse outcomes. Considering the relatively high incidence of postoperative delirium and the identified risk factors, our study indicates a comprehensive strategy to prevent postoperative delirium after intracranial surgery. In summary, one study cannot solve all problems. We thank Wan et al. for their suggestions for identifying and investigating safe and effective interventions to reduce the incidence, duration and severity of postoperative delirium, which might provide the readers with more information. With the help of all the readers, we will keep working on postoperative delirium issues to figure it out.
Read full abstract
Jun 4, 2021
Zhe Chu + 4
Open Access
