Preoperative prognostic nutritional index value as a predictive factor for postoperative delirium in older adult patients with hip fractures: a secondary analysis

This secondary analysis study was approved by the Peking University Third Hospital Medical Science Research Ethics Committee (approval No. M2022578) and registered in the Chinese Clinical Trial Registry (ChiCTR2300070569). The original two prospective, observational, single-center, cohort studies were approved by the Medical Science Research Ethics Committees of Beijing Jishuitan Hospital (JLKS201901-04, JLKS202009-10, respectively) and registered in the Chinese Clinical Trial Registry (ChiCTR1900027393, ChiCTR2000038924, respectively). All patients provided informed consent for the collection and analysis of their clinical data and serum samples.

This study was a secondary analysis performed at Beijing Jishuitan Hospital between 1 June 2020 and 30 March 2022. According to the previous study [13], the inclusion criteria were: age ≥ 65 years, hospital admission for surgical treatment of hip fracture, and American Society of Anesthesiologists physical status classification I–III. The exclusion criteria were: central nervous system disorders already existed preoperatively or with a previous history of such diseases, including preoperative delirium, Parkinson’s disease, dementia (including dementia due to Parkinson’s disease, Alzheimer’s disease, and Lewy body dementia), stroke within the previous 6 months or other central nervous system disorders; other factors that might affect the incidence of POD including multiple traumas and transfer to the intensive care unit postoperatively; communication difficulties and severe hearing or vision impairment which were unable to finish assessment scales, and unwillingness to participate in the study or unexpected discharge.

Nutritional status was assessed via PNI, which was calculated as 10 × serum albumin (g/dL) + 0.005 × total lymphocyte count (/μL). Serum albumin and total lymphocyte count were obtained from preoperative laboratory results.

All participants were interviewed the day before surgery, and the following baseline data were collected: demographic information, laboratory results, American Society of Anesthesiologists physical status, age-adjusted Charlson comorbidity index [14], Mini-Mental State Examination (MMSE) [15], education level, and Activities of Daily Living [16]. Pain intensity was assessed using a numerical rating scale [17]. Other information namely comorbidities, medical history, and fracture classification was collected from the patients’ medical records. The laboratory results were collected from the preoperative examinations processed 2 ~ 5 days before surgery. History collection and physical evaluation were performed by trained investigators.

The Confusion Assessment Method (CAM) was used to exclude patients with preoperative delirium and to diagnose POD, as previously described [13, 18]. Four items (acute onset and fluctuating course, inattention, disorganized thinking, altered level of consciousness) constitute the CAM. When diagnosing delirium, it is important to obtain information from a reliable source about the patient's acute onset and fluctuating course. Inattention may be indicated by difficulty focusing attention, while disorganized thinking may be shown by incoherent thoughts. The patient's level of consciousness may be altered, categorized as consciousness, vigilant, lethargic, stupor, or coma. To diagnose delirium using the CAM method, the first two items must be present along with either of the last two items. All participants were followed for the first 2 postoperative days, a period in which POD is usually diagnosed after hip fracture surgery in older adult patients. Cognitive assessment was performed through twice-daily visits and by a fixed geriatric-care physician.

No sedatives and/or anticholinergic drugs were administered to any patients as adjuvant medications before anesthesia. The anesthesia method (general anesthesia or spinal anesthesia) was determined by anesthesiologists in communication with the patients. After entering the operation room, patients received inhalational mask oxygen at 6 L/min. Heart rate, electrocardiography, and pulse oximetry were monitored. Radial artery catheterization was performed to directly measure mean arterial pressure and for blood gas analysis. All patients received an ultrasound-guided iliac fascia block and were given 30 mL of 0.33% ropivacaine for regional anesthesia and postoperative analgesia. Patients who received spinal anesthesia underwent single-dose subarachnoid spinal anesthesia as an injection of 8–10 mg of 0.3% ropivacaine at the L2–3 or L3–4 levels, and the anesthesia plane reached T10. Patients who received general anesthesia underwent intravenous induction with propofol and fentanyl. Rocuronium or cisatracurium was administered intravenously when the patient lost consciousness. Endotracheal intubation or laryngeal mask placement was performed 2 min later. Sevoflurane, remifentanil, rocuronium, or cisatracurium were used for anesthetic maintenance. The end-tidal partial pressure of carbon dioxide and bispectral index value were monitored intraoperatively in patients who underwent general anesthesia. During the operation, mean arterial pressure and heart rate were maintained within 20% of the baseline values, partial pressure end-tidal carbon dioxide was maintained at 30–35 mmHg, and the bispectral index value was maintained at 40–60 for patients who underwent general anesthesia.

For postoperative analgesia, all patients received intravenous patient-controlled analgesia with 1.5 μg/kg fentanyl, 200 mg flurbiprofen axetil, and 10 mg tropisetron hydrochloride in 100 ml saline in addition to the preoperative iliac fascia block.

An observational study conducted in China showed that patients with POD had a PNI of 39.4 ± 10.2, whereas those without POD had a PNI of 47.4 ± 11.7 [19]. Based on the calculated PNI mean, the study estimated that both the POD and non-POD groups needed to recruit 29 effective cases to achieve an 80% power at an α value of 0.05. To control for confounding factors such as age, hip fracture type, ASA classification, duration of surgery, intraoperative bleeding, etc., it was proposed to use logistic regression analysis. With an Events Per Variable equal to 5, 25 patients would be required in the POD group. As the incidence of POD in our previous study was 18.9%, the total sample size needed for the study was calculated to be 132 participants. To account for a potential dropout rate of 10%, the study aimed to recruit 147 participants.

The Kolmogorov–Smirnov test was used to analyze the distribution of quantitative variables. Measurement data that followed a normal distribution were presented as mean ± standard deviation, and the independent sample’s t-test was used for analysis. Measurement data that did not follow a normal distribution were expressed as median (minimum, maximum), and the rank sum test was used for analysis. The chi-square test was used to analyze the enumeration data. The patients were divided into a POD group and a non-POD group based on the CAM results. Univariate analysis was used to screen for differences in POD-associated factors between the groups, and binary logistic regression was used to identify the risk factors for POD. The areas under the receiver operating characteristic curve (AUCs) of the risk factors were then calculated. All data were statistically analyzed using SPSS 25.0 (IBM Corp., Armonk, NY, USA), and P < 0.05 was considered statistically significant.

A total of 369 older adult patients with hip fractures were included in this study. Of these, 67 patients were diagnosed with POD on the basis of the CAM results (Fig. 1). Overall baseline and clinical characteristics and the majority of the intraoperative characteristics were well-balanced between the POD and non-POD groups except for the comorbidity of hypertension, MMSE score, and anesthesia method (Table 1). Specifically, hypertension, low MMSE, and general anesthesia were risk factors for POD (p < 0.05).

Patients with low PNI, low serum albumin, and low thyroid stimulating hormone concentrations, were more likely to be diagnosed with POD compared with those without POD. Since the distribution of the PNI data is non-normal, we pre-suppose the median PNI of 45.6 as the cut-off value. We then divided PNI into low and high PNI by the cut-off value of 45.6, and we found that PNI < 45.6 could be used as a predictive factor for the incidence of POD. In contrast, lymphocyte count and other laboratory results showed no significant differences between POD and non-POD patients (Table 2).

After adjusting for confounders, binary logistic regression showed that the significant preoperative risk factors for POD after hip fracture surgery were PNI [adjusted odds ratio (OR) = 0.928, 95% confidence interval (CI): (0.864, 0.997)], general anesthesia [(adjusted OR = 2.307, 95% CI: (1.279, 4.162)), and MMSE score [adjusted OR = 0.956, 95% CI: (0.920, 0.994)] (Table 3).

The risk factors associated with POD shown in Table 3 were examined by receiver operating characteristic curve (ROC) analyses, and the AUC value for PNI was 0.592 [95% CI: (0.519, 0.664), sensitivity = 52.2%, specificity = 64.5%] (Fig. 2A). According to Youden index, the optimal cut-off values were verified as 45.675 for PNI. After combining general anesthesia and MMSE score, the AUC was 0.712 [95% CI: (0.646, 0.778), sensitivity = 75.0%, specificity = 61.2%] with a median predictive accuracy (Fig. 2B), suggesting that PNI combined with the anesthetic method and MMSE score rather than PNI alone may be used as a potential predictive indicator of POD after hip fracture surgery.

The study was approved by the Peking University Third Hospital Medical Science Research Ethics Committee (approval No. M2022578) and registered in the Chinese Clinical Trial Registry (ChiCTR2300070569). All participants and their legal guardians will be informed about the study goals. Written informed consent will be obtained from all participants or their legal guardians prior to the study enrollment. The study follows according to the Declaration of Helsinki. Participants as well as their relatives or legal guardians can withdraw consent at any time. All participant information and data will be stored securely and identified by a coded ID number only to maintain participants’ confidentiality.

Not applicable.

The authors declare no competing interests.

Data can be obtained from the corresponding author upon reasonable request.