17.0% patients were diagnosed as POD by the most widely used CAM criteria from prospective database in this study. Incidence of POD had been inconsistent in previous studies. Dezube et al.[37] found that the overall POD rate was 16.9%, which was very similar to this study. Markar et al.[38] revealed that POD occurred in 46 (9.2%) patients of the 500 included patients undergoing esophagectomy for cancer. But another study found that the incidence of POD was as high as 32% among 84 esophagectomy patients[39]. Takeuchi et al.[40] showed that POD developed in 153 (50.0%) of 306 patients with esophageal cancer. The variability of reported incidence can be possibly explained by heterogeneity of the patients, differences in assessment tools and lack the awareness to identify delirium properly by training and education.
We constructed a nomogram to identify the risk of developing POD based on the results of multivariate analysis. In the training cohort, the nomogram had good discrimination ability with AUC of 0.788 and the calibration plot demonstrated good consistency between the observed and predicted probabilities. More importantly, the nomogram was well validated in an external cohort by independent dataset (data from another hospital), which proved the satisfactory accuracy, high stability and good reproducibility of our model. Moreover, the DCA analysis was performed to confirm that the nomogram had excellent clinical application value in the prediction of POD with superior net benefit.
In this study, age was strongly independently associated with the development of POD. An advanced age has been widely accepted as a major demography-related preoperative risk factor affecting the onset of POD in several studies[41–44]. As the increase of age, degeneration of compensatory physiological mechanisms and organ function, remarkably reduced body adaptability to drastic alterations during the surgical period leaded to a significantly increased risk of POD[29]. In addition, the older patients may be associated with brain atrophy, gradual neurodegeneration, decreased cerebral neurotransmitters production, the narrowing of cerebral vessels due to endothelial dysfunction and atherosclerosis, making older patients more susceptible to POD[45, 46].
This study found that diabetes mellitus was regarded as an important independent predictor for POD. Cognitive decline is documented as a major co-morbidity in people with diabetes by following causes: insulin resistance, oxidative stress, neuronal damage and inflammation[47, 48]. Furthermore, diabetes mellitus has been confirmed to cause macrovascular and microvascular damage: Atherosclerosis of the large-caliber arteries, reduction in the number of capillaries, an increase in the arteriovenous short circuit, the thickening of the basement membrane and endothelial dysfunction[49, 50]. All these effects lead to make the brain tissue decrease of senile cerebral blood flow and more vulnerable to hypoxic damage when the perfusion pressure drops due to surgery or anesthesia[51].
The ASA classification system is quick and easy to assess preoperative functional status. Differing significantly from patients with scores of 1 or 2, Patients with ASA classification ≥ 3 level often suffered more serious systemic diseases, impaired general physical status, higher incidence of postoperative complications and post-operative mortality[52–55]. Consistent with previous studys[56, 57], we divided ASA scores into these two categories. We found that the ASA classification (≥ 3 level) was an independent predictive factor for POD in our study. Surgery and anesthesia increased psychological and physical stress and the patient's preoperative physical condition plays an important role in postoperative recovery[58]. Combined effect of multiple comorbidities and decline in physiologic reserve in patients with higher ASA that increases susceptibility to POD[53, 59].
In this predictive model, low preoperative albumin levels were significantly associated with increased risk of POD. Although several previous studies[60–63]. had also been shown similar to the findings, the cut-off values of serum albumin were not uniform At present, the pathophysiological evidence of interaction between them is still unknown. The possible speculations was that albumin serves as an important biomarker of nutritional statuses, and it is well known that malnutrition had negative impact on cerebral function. Patients with protein malnutrition and vitamin deficiency have difficulty in tolerating surgery and are more prone to lower cognitive performance and POD[64–66]. Additionally, Albumin functions as an anti-inflammatory effects, antioxidant activity and neuroprotective effects by partly attributed to modulation of intracellular signaling of neuronal or glial cells. This defensive mechanism was reduced in hypoalbuminemia and caused cognitive impairment[67, 68].
In our study, open esophagectomy was a risk factor for the incidence of POD. Open approache is associated with significant morbidity and mortality. More recently, Minimally invasive esophagectomy has become increasingly popular due to limit surgical incision, reduce postoperative pain, decrease overall complications, promote early functional recovery, shorten hospital stay, increase quality of life, improve long-term outcomes[69–71]. Ito, et al. [72]also reported that laparoscopic approach was associated with a lower incidence of POD after major abdominal surgery. The possible explanation could be that minimally invasive esophagectomy had advantage to reduce the surgical trauma, attenuate the systemic stress, better preservation of immune function, and lesser postoperative inflammatory response[73]. Some studies[74, 75] have confirmed that a significant association between heightened surgery-evoked inflammatory responses and the development of POD.
The present study still had several limitations. First, Patients in training cohort and validation cohort were from single center rather than multicenter databases, and the resulting selection biases should be acknowledged. Secondly, POD is characterized by fluctuating disturbance of attention and cognition. The incidence of POD in this study may be low due to all participants was routinely evaluated only once on daily basis. Finally, Other variables such as frailty and physical activity were reported to increase odds of POD[76, 77]. However, The variables were not routinely incorporated in preoperative evaluation and some data in this part were missing.