Ever since the first report of postoperative delirium (POD) following cardiac surgery in 1964, it remains an ardently researched domain^[1-4]. This is principally attributed to the fact that the incidence of POD continues to range between 13 and 50% despite major advancements in cardiac surgical conduct^[2,3]. The POD-associated morbidity-mortality, caregiver suffering, and the healthcare burden buttress the importance of a sound prediction and prevention strategy^[1-4].

Talking of the pharmacological prevention of POD, dexmedetomidine (DEX) has been subjected to an extensive evaluation with mixed results in the available literature^[4,5-8]. While a number of existing meta-analyses support POD-reduction with DEX in a cardiac surgical setting^[5,6], few large scale randomized controlled trials (RCTs) conducted off-late suggest otherwise^[7,8]. In this context, “Dexmedetomidine for reduction of atrial fibrillation and delirium after cardiac surgery (DECADE)” - the placebo-controlled RCT by Turan et al.^[8] contemplated across six academic hospitals in the United State of America and later published in The Lancet in the year 2020 - captivates attention. The DECADE trial studying atrial fibrillation and POD co-primarily randomized 798 patients to receive either DEX or saline infusion at 0.1 µg/kg/hour prior to incision, escalated to 0.2 µg/kg/hour at the end of bypass and to 0.4 µg/kg/hour postoperatively till 24 hours. Following the analysis of data of 794 patients, the trial failed to outline an amelioration of either of the unfavorable outcomes, rather demonstrated a non-significant worsening of POD (17% and 12% in the DEX and placebo groups, respectively)^[8]. This portrays a striking contradiction to the findings of the previous meta-analysis of the decade, quite literally^[5,6,8].

Premised on the evolving controversy, an updated meta-analysis was contemplated by Li et al.^[9] comprising of a total of 2,813 adult patients across 15 RCTs (each featuring comparative DEX and placebo-control/other anesthetic-control groups and POD as a primary/secondary outcome) which also included patients of the DECADE trial^[8]. The pooled results emanating from this very recent meta-analysis published in 2021 revealed that DEX could significantly attenuate the risk of POD in comparison to the controls, with an odds ratio (OR) of 0.56, 95% confidence interval (CI) of 0.36-0.89, P-value = 0.0004, and 64% I2 coefficient^[9]. Ahead of the demonstration of 44% POD risk reduction with DEX, the total POD-incidence funnel plot did not connote a significant publication bias in the meta-analysis. The additional findings of the subgroup analysis are outlined in Table 1^[9].

Although the primary conclusions of the DECADE trial and Li et al.^[9] meta-analysis appear incongruous at first sight, a closer look resolves the dilemma to an extent^[8]. As described previously, the DECADE trial initiated DEX infusion intraoperatively wherein the absence of an eventual POD reduction is actually consistent with the Li et al.^[9] subgroup analysis, which also limits the description of POD-reduction potential to a postoperative drug administration (Table 1)^[8,9]. Li et al.^[9] speculate intraoperative hypotension (IOH) as a likely mechanism of lack of POD protection of intraoperative DEX and elaborate that the former was not recorded in many of the included RCTs^[9]. Turan et al.^[8] also implicate IOH for the non-significant rise in POD in the DECADE trial given the incidence of the same was 57% and 36% in the DEX and placebo groups, respectively. However, no significant association between IOH and POD materialized in the Wang et al.^[10] post-hoc analysis of the DECADE trial including data from three Cleveland Clinic hospitals, outlining an adjusted OR of 0.94 (95% CI: 0.81-1.09; P-value = 0.419) for a doubling in the area under the curve of mean arterial pressure < 60 mmHg, after applying thorough logistic regression models and adjusting for 11 confounders.

As far as POD reduction with postoperative DEX is concerned, an enlightening discussion transpired in a leading cardiothoracic anesthesia journal, the Journal of Cardiothoracic and Vascular Anesthesia^[11,12]. An eleven-RCT meta-analysis by Abowali et al.^[11] concluded an absence of a statistically meaningful POD reduction with postoperative DEX in comparison to propofol sedation. Nonetheless, Heybati et al.^[12] highlighted that Abowali et al.^[11] managed to erroneously include two RCTs, one administering DEX intraoperatively with another employing DEX as an adjunct to propofol. Subsequently, Heybati et al.^[12] reanalyzed the data of the remaining nine RCTs only to discover a significant POD reduction with postoperative DEX (random-effects OR of 0.18, 95% CI: 0.05-0.65, with an I2 = 52% denoting a moderate heterogeneity level). Needless to say, the readjusted results were in agreement with the recent Li et al.^[9] meta-analysis.

Looking beyond the categorical details of the meta-analysis^[13], there are subtle intricacies of the individual DEX-delirium RCTs which necessitate close consideration^[9,14-17]. Table 2 enlists some of the pertinent points to assess while interpreting the results of these RCTs^[9,14-17]. For instance, the POD protective role evaluation of DEX in the Turan et al. DECADE trial may have been compounded by the lack of a protocolized anesthetic regime, titration of the study drug, and other sedatives administration at the discretion of the anesthesiologist and disregard to the group-specific cumulative sedative drug dosages being utilized^[8,18,19]. At the same time, exempting a formal preoperative cognitive status assessment can pose important concerns as adequate literature exists to suggest preoperative anxiety-depression and low mini-mental state examination score as independent risk-factors for POD following cardiac surgery^[14-17]. Notably, the Li et al.^[9] meta-analysis excluded the RCTs which did not rule out a preoperative cognitive impairment.

Even from a holistic research standpoint, the traditionally cited evidence hierarchy of meta-analysis over and above multiple and single RCTs is an oversimplistic rendition, mandating a cautious interpretation^[13]. Amidst an ongoing surge in the number of meta-analysis, it becomes imperative to realize that meta-analysis of small RCTs can be conducive to an overestimation of treatment effects, often reformed by larger RCTs which follow^[13]. In conclusion, large scale multicentric RCTs and meta-analysis of varying RCT counts coexist in the present heterogeneous research ecosystem wherein a superficial perusal of the conclusions can be misleading, particularly with the methodological issues, potential biases, and rhetorical presentation techniques being an inexorable part of the vignette^[13,20].