Identification of risk factors for postoperative delirium in elderly patients with hip fractures by a risk stratification index model: A retrospective study

Abstract Introduction Postoperative delirium is one of the most common and dangerous psychiatric complications after hip surgery. The aim of this study was to investigate the incidence of postoperative delirium in elderly patients after hip fracture surgery and to identify risk factors for such, as part of developing a risk stratification index (RSI) system to predict a patient's risk of postoperative delirium. Methods Elderly patients (aged 65 years or older) with hip fractures who had received surgical treatment in our hospital between March 2018 and December 2019 were retrospectively included. Clinical data were collected, and multivariate logistic regression analysis was performed to investigate the relevant risk factors of postoperative delirium. An RSI system was developed based on factors identified in the regression analysis. Results Of 272 patients included, 52 (19.12%) experienced postoperative delirium. Drinking history (> 3/ week), the perioperative lactic acid level (Lac > 2 mmol/L), postoperative visual analog score (VAS) > 3, American Society of Anesthesiologists (ASA) physical status > II, application of the bispectral index, and preoperative diabetes were independent risk factors of postoperative delirium. When RSI ≥ 5, the rate of postoperative delirium significantly increased (p < .05). Conclusion The RSI system developed here can safely guide postoperative outcomes of elderly patients with hip fractures, and RSI ≥ 5 may be able to predict the onset of postoperative delirium.


Study population and design
This was a retrospective study, which included elderly patients with hip fractures (femoral neck, intertrochanteric, and subtrochanteric fractures) who were no less than 65 years old and had been scheduled This study received institutional review board approval. The chart review of the cohort of 272 patients was covered by a hospital quality and improvement project, and patient consent was waived.

Data collection
General anesthesia (including that combined with nerve block) and combined spinal and epidural anesthesia (CSEA) were performed.
For intravenous-inhalation combined anesthesia, venous access was opened after entering the room. Routine electrocardiogram, noninvasive blood pressure, pulse oxygen saturation, and body temperature monitoring were performed. Sufentanil, rocuronium bromide, etomidate, and propofol were used for the sequential induction of general

Diagnostic criteria
With reference to the CAM standards developed by the American Psychiatric Association, the diagnostic criteria were the following (Wei et al., 2010): (1) acute onset and fluctuating conditions; (2) inattention; (3) disordered thinking; and (4) changes in consciousness level, delirium could be diagnosed by the presence of (1) and (2), plus either of (3) or (4).

Sample size
We first conducted a chart review from January 2018 to March 2018, which included 16 hip fracture patients with 3 postoperative delirium (POD) (18.75%). Therefore, according to the incidence of POD reported in available studies and a pretrial study (van Meenen et al., 2014), it was assumed that the expected positive rate π is 20.0%, the relative error ε is 20%, the allowable error E is 5%, and the confidence level 1 -α is 95%, so the sample size was calculated as 246 cases.

Statistical methods
Stata 15.0 (StataCorp LP, College Station, TX, USA) statistical software was used for the analysis. Measurement data were denoted asx± standard deviation (SD), and comparisons between groups were performed using t-test. Count data were denoted as sample rate or composition ratio, and comparisons between groups were performed using the χ 2 test. For an analysis of risk factors, multifactor backward stepwise logistic regression analysis was used in order to screen out the independent risk factors (excluding variables with p > .05). And to avoid multicollinearity between variables, those with VIF > 2 were excluded.
All included variables were dichotomous. The weighted RSI model was established according to the odds ratio (OR) of the corresponding independent risk factors. p < .05 was considered statistically significant.

Patient demographic information and risk factors related to delirium
This was an open study with no restrictions on various treatment measures throughout the perioperative period. Patients were excluded due to incomplete information and compliance with exclusion criteria (n = 11). Finally, all of the 272 patients who remained in the study underwent hip fracture surgery.
Among the 272 elderly patients undergoing hip fracture surgery, a total of 52 patients had POD (19.12%). The patients were divided into a POD group and a non-POD group based on whether or not POD had occurred. A univariate analysis was performed on all patients.

Risk factors independent of delirium
Using delirium occurrence as the dependent variable, a multifactor stepwise logistic regression analysis was performed (excluding variables with p > .05 from univariate analysis). The results showed that drinking history (>3 times/week), intraoperative lactic acid measurement (>2 mmol/L), postoperative VAS scores (>3), ASA > II, and preoperative diabetes were independent risk factors of POD in elderly patients undergoing hip fracture surgery, while the intraoperative application of BIS was a protective factor (Table 2).

Establishment of a risk stratification index
According to the odds ratio (OR) of the corresponding independent risk factors shown in Table 2, we established a weighted RSI model.
In order to simplify the model for clinical use, we assigned a weight of 1, 2, and 3, corresponding to OR of 1-10, >10-20, and >20, respectively. If OR < 1, then it was converted to 1/OR (1/applied BIS = 6.25) and weight OR = −1 (protection factor). So, the RSI of POD in elderly patients undergoing hip fracture surgery resulted in a maximum score of 8. When the RSI was −1 to 8, the incidences of POD in elderly patients with hip fracture were 0%, 0%, 1.72%, 9.80%, 14.29%, 26.47%, 61.54%, 100%, 100%, and 100%, respectively. Figure 1 shows the incidences of POD in elderly patients undergoing hip fracture surgery according to different RSI values. When the RSI ≥ 5, the rate of postoperative delirium significantly increased (p < .05, Figure 1).

Clinical prognosis of patients
Compared with the non-POD group, the POD group showed a significant increase in the average length of hospital stay (p < .05) and a moderate increase in medical costs and mortality (Table 3).

DISCUSSION
The aim of this study was to investigate the incidence of POD in elderly patients who underwent hip surgery and to ascertain factors that increase the risk of POD. This information was then used to develop Abbreviation: CI = confidence interval, OR = odds ratio, ASA = American Society of Anesthesiologists, BIS = bispectral index, VAS = visual analogue score.

F I G U R E 1
Graph showing the incidence of postoperative delirium (POD) in patients graded according to their risk stratification index (RSI) score maximum score of 8. Patients with an RSI score ≥5 had a significantly increased risk of POD.
To date, there have been few studies that have concentrated on risk factors for POD after hip fracture. One previous study found older age, lower albumin, a history of stroke, higher blood glucose, higher total bilirubin, higher C-reactive protein, longer duration of surgery, and a higher volume of red blood cell transfusions were independent risk factors of POD in elderly patients following total hip arthroplasty for hip fracture (Guo et al., 2016). While another study found that age over 75 years old, diabetes, and ASA classification > II were independent risk factors for POD (Wang et al., 2018). Our results show some differences with both these studies. Diabetes or higher blood glucose were identified in both the studies in agreement with our results, and ASA > 2 was identified in one of the studies. However, age was not identified as an independent risk factor. While preoperative drinking, perioperative lactic acid level > 2 mmol/L, postoperative VAS, and application of BIS that were identified in this study were not identified in the previous studies. Differences between the studies highlight the need for larger multicenter studies to evaluate risk factors for POD after hip surgery.
The risk factors identified by this and prior studies suggest three major contributors for the occurrence of POD: underlying brain health (age, drinking, and diabetes), the effects of specific medications and anesthetics (propofol and general anesthesia), and clinical distress during and after surgery (lactic acid and VAS scores). Each of these contributors will be discussed in turn.
The age limit of susceptibility to delirium remains controversial. The statistical results of the elderly patients in this study found that the age of >85 years old is not an independent risk factor, and the number of patients aged >85 years old in the delirium group is less than that in the non-POD group (51.92% < 71.36%). However, it is generally believed that elderly patients are the high-risk group for POD (Elie et al., 1998;Winkler et al., 2011). In addition to age, long-term drinking can also cause toxic reactions to multiple regions in the cerebral nervous system, especially the prefrontal lobe, resulting in amnestic cognitive impairment. This study showed that patients with a longterm drinking habit (≥3 times/week) are more likely to have delirium after surgery (Dickov et al., 2012). The elderly patients undergoing hip fracture surgery can be subjected to complicated perioperative conditions, greater surgical trauma, and stronger stress response.
Preoperative ASA classification is a preliminary assessment of the patient's tolerance to anesthesia, and it can also basically reflect the physical status of the patient. In this study, preoperative pulmonary, cerebrovascular, and cardiovascular diseases were excluded from the independent risk factors. However, preoperative diabetes was an independent risk factor for POD in elderly patients undergoing hip fracture surgery may be caused by the impaired cerebral blood flow due to the involvement of the cerebral and carotid arteries, as well as the autoregulation that affects cerebral blood flow (Gummert et al., 2002).
The impact of type of type of anesthesia on POD is still being debated (Mason et al., 2010). In this study, two types of type of anesthesia, intravenous-inhalation combined anesthesia versus CSEA, were not independent risk factors of POD. Nevertheless, compared with intravenous inhalation combined anesthesia, CSEA could reduce the incidence of POD in elderly patients with hip fractures by 11.4% (Table 1). As for the underlying reason, the role of anesthetics is inconclusive (Chandler et al., 2013). Nishikawa holds that propofol promotes the occurrence of POD through drug redistribution in the body leading to peripheral drug reflux that affects the mental function of patients (Nishikawa et al., 2004), while Gerretsen believes that it works by blocking muscarinic acetylcholine receptors (mAChRs) (Gerretsen & Pollock, 2011). For inhaled anesthetics, Meyer found that both isoflurane and sevoflurane could promote the occurrence of POD (Meyer et al., 2007). The occurrence of POD is generally believed to be associated with intraoperative hypotension, postoperative hypoxemia, and postoperative pain (Cole, 2014;Leung et al., 2009;Wang et al., 2015).
Although all patients in this study were given sufficient analgesia during the surgery, given the varying conditions and individual differences, the patients with VAS > 3 were more likely to develop POD. Due to the particularity of elderly patients undergoing hip fracture surgery, the intraoperative measurement of lactic acid can be used to directly