Cardiovascular disease has become one of the greatest threats to human health in the 21st century (1). The number of patients suffering from cardiovascular disease has increased dramatically over recent years worldwide (1, 2), and the amounts of cardiovascular operations have also increased rapidly. According to the latest report, 1.5 million cardiac surgeries are performed globally every year approximately (3), and the incidence of complications varies from 2 to 60% following cardiac surgery (4). POD is the most common complication among cardiac surgical patients with an incidence of 25–52% (5). It is defined as an acute disturbance of consciousness characterized by acute and fluctuating changes in attention, awareness, and cognition (6), with a poor prognosis. An analysis published in Lancet reported that delirium costs more than $164 billion in health care expenses in the United States each year (7), bringing a heavy economic burden to society. Thereby, the prognosis of POD is receiving greater public attention (8).

In recent years, a large body of evidence indicates that POD in patients undergoing cardiac surgery is significantly associated with poor prognosis. However, the results remain controversial. Compared with patients without POD, cardiac surgical patients who develop POD have higher mortality (9, 10). While others suggest that POD is not significantly related to mortality (11, 12). According to our literature search, only one review evaluated the association between POD and mortality in patients undergoing TAVR which was published in 2020 (13), seven studies are included and the sample size is relatively small. Individual studies have insufficient power to detect the association and to persuade convey conflicting results. Furthermore, a quantitative evaluation of the influence of POD on hospital days, ICU time, and MV time has not been performed.

It should be noted that we do not draw enough attention to delirium since the insufficient recognition and under-reporting (14, 15). Knowledge of the true magnitude of POD and its associated burdens in cardiac surgical patients would allow healthcare professionals to allocate much-needed resources toward reducing morbidity and mortality associated with delirium after cardiac surgery. Therefore, we conducted a systematic review and meta-analysis to explore the relationship between POD and outcomes in these patients, including short-term and after-discharge mortality, hospitalization, ICU stays, and MV time, to provide scientific data for POD management after cardiac surgery.

The present systematic review and meta-analysis identified 42 studies enrolling a total of 19,785 patients, to summarize the relationship between POD and outcomes in patients undergoing cardiac surgery. The overall incidence of POD was 17.0%, and the results of the meta-analysis showed POD has been associated with increased mortality, longer duration of MV time, ICU stays, and hospitalization among cardiac surgical patients. Given certain heterogeneity among studies, we next conducted subgroup analysis based on study design, sample size, male proportion, surgery type, study region, etc. Despite remaining heterogeneity after subgroup analysis, it was partly reduced in some subgroups. This suggests that POD is a very common and severe neuropsychiatric syndrome, that seriously affects the prognosis of patients.

In this study, POD was significantly linked to mortality. However, due to the significant heterogeneity across studies, this relationship may be limited. Subgroup analyses were conducted based on the timing of mortality, and the results indicated that POD is related to short- and long-term mortality of cardiac surgery patients, which was partially consistent with Crocker et al. (58). The difference was that Crocker et al. indicated that POD was no significant association with short-term mortality. This may be linked to no meta-regression analysis performed to assess the influence of POD on short-term mortality in their study, as only two of the included studies had reported 6-month mortality. However, short-term mortality was reported in 13 studies in our article, where a larger sample size may yield different results.

The mechanism by which POD increases mortality risk is not understood. We propose the following explanations to comprehend the results. First, higher levels of postoperative pro-inflammatory cytokines (e.g., IL-2, IL-6, TNF-α, MCP-1) were associated with a higher risk to develop POD in cardiac surgery patients (59, 60). Elevated levels of pro-inflammatory cytokines reflect an active inflammatory response in the body, which may cause systemic inflammatory response syndrome. Systemic inflammation can alter the brain’s inflammatory status, produce acute cognitive impairments, such as POD, and drive new pathology and accelerated decline (61). Thus, the presence of POD can be considered a marker of hyperinflammatory conditions, which are associated with higher mortality (62, 63). Second, Holmes et al. (64) indicated that POD may represent an extreme non-adaptive presentation of sickness. Patients who develop delirium after cardiac surgery could cause cognitive impairment, decreased consciousness, behavioral abnormalities, etc., which increase cerebrovascular accidents, bleeding, infection, and other complications risk (50, 65, 66), and patients with hypoactive motor-type delirium may present with more severe systemic disease, increased complications of inactivity (e.g., dehydration, pressure ulcers, hypoventilation, and venous thrombosis; 67), these complications are associated with higher mortality (68). Finally, POD is closely associated with the presence of hemodynamic or electrical instability, and disorders of fluids and electrolytes, which may increase the risk of mortality (69). The accidental extubation, difficulty in weaning, or reintubation in patients with MV were also increased (70–73), which required an increased duration of MV, and it has been well documented that prolonged MV time is an independent predictor of increased mortality (74). The data also showed that POD could result in prolonged hospital and ICU stays of the patients, concordant with the results of Salluh’s study (8). The longer the patients stay in the hospital, the more they are at risk of complications and death. All these factors may explain the increased mortality risk among patients who develop POD. Further investigation regarding the pathophysiological mechanism of POD is still warranted to fully understand the reasons why POD led to poor outcomes among cardiac surgical patients.

Given the poor outcomes among cardiac surgery patients who developed POD, there exists a great opportunity to improve the outcomes among these patients. A review (75) in Lancet reported that 30 to 40% of delirium may be prevented by early detection, and takes pharmacologic or non-pharmacologic interventions. Current guidelines (76) recommend using a multicomponent, non-pharmacologic intervention to reduce delirium. The strategies include cognitive training, improving sleep quality, improving wakefulness, early rehabilitation, etc. (conditional recommendation, low quality of evidence). However, it is still uncertain as to which interventions result in the effect. In the future, emphasis should be put on improving the awareness of medical staff on delirium, and undertaking studies to validate the intervention effects, to provide widely applicable evidence for healthcare policymakers.

There was a high degree of heterogeneity observed in our meta-analysis. The reasons might be as follows: first, the methods for diagnosing delirium were different. The incidence of delirium may be dependent on the different diagnostic criteria applied, different tools used, and different evaluators. Second, the study periods of the included studies were different. With the progressive developments in delirium research, the attention toward POD has gradually increased. In previous years, POD has not yet attracted enough attention from medical staff, and there is also wide variation across hospitals in the treatment of POD. In addition, sample sources were different. The structure of the population in different studies was different such as age, gender ratio, race, disease severity, and surgery types, which could contribute to the different clinical outcomes. Most of the included studies did not provide adjusted data due to the high risk of confounding bias, for example, age, sex, and disease severity, which is the reason why we did not use the adjusted data for further analysis. Regarding the mortality outcome, we conducted a sensitivity analysis. The heterogeneity was reduced from 75 to 24% after the removal of five studies (12, 26, 32, 50, 57), which indicates that the five studies were the source of heterogeneity.

The present meta-analysis exhibited several strengths, compared to the previously published meta-analysis (13, 58). In the first instance, we used a robust methodology following PRISMA guidelines and a comprehensive search strategy, to ensure the inclusion of all relevant literature. Second, we included 42 studies with a larger sample size. It could provide high statistical power to quantitatively evaluate the association between POD and clinical outcomes. Hence, the validity of the results is more reliable. In addition, the included studies in this meta-analysis had high NOS scores which were strictly following the inclusion and exclusion criteria, thus, reducing the potential selection bias.

However, there remain limitations in this study as well. First, a meta-analysis of MV time, and ICU time showed heterogeneity, but sensitivity analysis and subgroup analysis failed to eliminate it. The random-effects model is used for data processing, which may have a slight impact on the reliability of the results. We speculate that heterogeneity might be partially explained by the differences in factors such as patient characteristics (age, sex, type of surgery, etc.), different diagnostic criteria for delirium, unequal levels of regional medical care, and frequency of delirium assessment. Second, variations in the assessment tools and the assessment time-points of delirium might affect the results. Delirium is a fluctuation in mental status that can change over time and may have occurred before or after assessments. Thus, the true incidence of POD and its effects on clinical outcomes might be underestimated. All included studies screened patients utilizing validated delirium assessment tools, but the latest research states that POD needs to meet DSM-5 diagnostic criteria (13), and not all studies achieve this. Future prospective studies with standardized delirium assessment methods are still needed to detect delirium accurately and reliably. Third, due to insufficient data, we could not further evaluate other potential factors that may affect the heterogeneity between studies, such as Euroscore score, complications, and the use of anesthetic drugs. Finally, publication bias remains a major concern for all kinds of meta-analyses because non-significant or negative results are less likely to be published than studies with positive and significant results. To comprehensively identify negative or insignificant outcomes, we used delirium and cardiac surgery as keywords which meant the kinds of literature published on this topic were eligible, to ensure we identified as many relevant studies as possible; and also incorporated all reported outcome measures from each study. In addition, funnel plots were constructed to assess potential publication bias, and it is worth mentioning that there was no observable publication bias.

In this meta-analysis, we found that POD was involved in poor prognosis among cardiac surgical patients. Patients who develop POD exhibit longer MV time, ICU stay, hospital stay, and greater risk of mortality than patients without POD. Future research should focus on developing and testing interventions for delirium, to reduce its incidence and thereby lower the risk of adverse outcomes in these patients.

The original contributions presented in this study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author/s.

LL: project administration, conceptualization, investigation, date curation, methodology, formal analysis, and writing—original draft, review and editing. XZ: conceptualization, investigation, date curation, formal analysis, and writing—original draft, review and editing. SX: methodology, date curation, and writing—original draft, review and editing. YP: conceptualization and writing—original draft, review and editing. SL: software, date curation, and writing—review and editing. XH: date curation and writing—review and editing. LC: supervision and funding acquisition. YL: conceptualization, supervision, and funding acquisition.