In Reply

I would like to thank Raiten and Gutsche for their thoughtful comments on our article.1 

Despite uncertainty on the clinical course of subsyndromal delirium (SSD), there is evidence that it may herald manifest delirium.2,3Nevertheless, although SSD may be viewed as a penumbra between normal mentation and clinical delirium,4the transition from a normal mental status to SSD may not be distinct temporally and is often missed clinically. Accordingly, we presumed that early identification and treatment of SSD might halt its progression to full-fledged delirium. In other words, targeted delirium prophylaxis was actually the main theme of our trial, which aimed at rationalizing pharmacological prophylaxis by offering it to those at assumedly particular risk for delirium. This approach stands in contradistinction to the conventional pharmacological approach to prophylaxis, which counts on the arbitrary commencement of antipsychotic prophylaxis just before5or shortly after6surgery. In this respect, we believe that a fundamental implication for targeted prophylaxis is that administration of antipsychotics should be commenced as early at the inception of SSD as possible. To achieve this, we started to screen our patients for SSD as soon as they were deemed eligible for assessment using our screening tool, the Intensive Care Delirium Screening Checklist. The time frame of 4 h after extubation, however, may not be regarded as premature in view of the prevailing practice of having the patients extubated within 12 h of surgery unless otherwise indicated,7which is also the regular practice at our institute. To achieve this target, sedatives are tapered off once patients are considered for weaning so that drug effect has vastly waned by the time they are being extubated. In fact, one of the criteria for extubation we adopt is that patients should be awake without stimulation.7The authors have contended that residual drug effect, particularly benzodiazepines and opioids, may have been the cause of SSD presenting shortly after extubation. Although this could be a possibility, delaying our screening in the hope that lingering drug effect would vanish may have been unwarranted because analgesics, including opioids, were to be administered for pain control for as long as needed to make our patients comfortable. Virtually, this would have translated into precluding the screening for SSD for an indefinite period, thus radically undermining the concept of early detection and treatment of SSD, which was the major object of our trial. On the other hand, the administration of pharmacological prophylaxis to those with SSD, even as early as employed in our study, may at worst be regarded as a step forward in rationalizing the management of delirium prophylaxis compared with the conventional practice of nonselectively administering the antipsychotic before surgery5or immediately upon regaining consciousness after operation.6Notably, none of our patients had SSD on the first time they were assessed. To put this in context, the minimum time from extubation to detection of SSD and commencement of antipsychotic medication was 17 h. We did not provide information on this issue in the expectation that the time lapse from extubation to randomization (i.e. , detection of SSD) would not be of much clinical relevance. However, reanalysis of pertinent data showed that the median (interquartile range) time from extubation to detection of SSD was 21 (19–27) h for the whole population studied, 21 (18–26) h for the risperidone group, and 21 (19–29) h for the placebo group.

The authors have also alluded to the fact that administration of risperidone in our trial was not associated with shortening of the length of stay in the intensive care unit or the duration of clinical delirium. This may be accounted for by the fact that we based our sample size calculation on the primary outcome measure, i.e. , the incidence of delirium. Because we conducted no a priori  power analysis for any of our secondary outcome measures, including the length of the intensive care unit stay and duration of clinical delirium, our study could have been underpowered in this respect. For instance, applying post hoc  power analysis to our data showed that our sample size of 51 and 50 patients in either group had barely a power of 0.49 to detect a medium effect size of 0.4, or a power of 0.68 to detect a larger effect size of 0.5 for the length of the intensive care unit stay, assuming a type I error of 0.05 and using the two-tailed Mann–Whitney U test. Using the same assumptions, our observation of seven delirious patients in one group and 17 in the other returned a meager power of 0.13 to detect an effect size of 0.4, or a power of 0.18 to detect an effect size of 0.5 for the duration of clinical delirium. We have already addressed this relevant issue in the Discussion section of our article and ended up with a call for larger studies to confirm our results and to explore other aspects that we failed to uncover in our trial.

The authors have also raised a question regarding the use of risk-factor management techniques8and the impact of this issue on the validity of our results. Although we do not employ formal programs such as the Hospital Elder Life Program,8we do adopt most components of these programs into our regular care, including practices for cognitive stimulation, nonpharmacological sleep-enhancing maneuvers, encouragement of early mobilization, access to personal visual and hearing aids, and attention to hydration status. Such interventions, though not implemented through standardized services, are provided to our patients as basic aspects of proper postoperative care. On the other hand, although these programs are believed to be associated with lower incidence of delirium,8,9a Cochrane review on interventions to prevent delirium10has concluded that evidence in favor of proactive geriatric consultation programs is based merely on a single randomized controlled trial11comparing a proactive geriatric consultation program with usual care in elderly hip fracture patients. Moreover, current guidelines for prevention of delirium recommend that multicomponent delirium prevention strategies may be considered for use in older hospitalized patients based on grade C evidence (i.e. , lacking direct evidence).12Thus, in view of available evidence, multicomponent programs can by no means be considered as a standard of care, and may be regarded at best as bundles of good care that should be widely offered to hospitalized elderly patients. In this respect, the effectiveness of multicomponent programs has been demonstrated to be vulnerable to the degree of adherence on the part of healthcare providers.13–15This issue may cast significant doubt on the feasibility of wide-scale application of these programs. For instance, in the single rigorously conducted randomized controlled trial by Marcantonio et al. ,11the overall adherence to recommendations made by the geriatrics team was in the order of 77%, whereas adherence to specific recommendations (e.g. , to prescribe around-the-clock analgesic) was as low as only 32%. Although there is no sufficient data, there is some reason to expect even lower adherence rates in real-life uncontrolled settings owing to evident logistic and financial constraints. In this regard, although the proportion of facilities actually implementing multicomponent geriatric programs is not known exactly, available data suggest that fewer than half of the facilities in the United States may implement such programs and that only a modest proportion of potential candidates actually utilize the available services.16 

Finally, in our article, we did stress the fact that our results should be regarded as quite preliminary and that larger studies are still required to examine the validity and reproducibility of our findings.

, Ain Shams University, Cairo, Egypt. drsmichel@hotmail.com