Study design
A single-centre, prospective, observational cohort study was conducted in Renji Hospital, School of Medicine, Shanghai Jiaotong University (Shanghai, China), from January 2018 to March 2020. The study was conducted in accordance with the Declaration of Helsinki and approved by the local ethics committee (RJ198K) (registered at Clinicaltrials.gov; NCT03393676). Written informed consent was obtained from all patients. All investigators were well trained in neuropsychological testing and pain evaluation.
Patient selection
All consecutive patients who underwent elective hemiarthroplasty or total hip arthroplasty with general anaesthesia performed by a single surgeon were enrolled. The inclusion criteria were as follows: (1) >65 years old, (2) speaks Chinese Mandarin, (3) planned to undergo major low limb surgery like hemiarthroplasty or total hip arthroplasty with general anesthesia, (4) signed the informed consent form and (5) assessed as American Society of Anesthesiologists classification(ASA) I to II.
The exclusion criteria included the following: (1) existing cerebral disease or a history of neurological and psychiatric disease, including Alzheimer’s disease, stroke, epilepsy and psychosis; (2) existing cognitive impairment as evidenced by the Mini-Mental State Examination (MMSE) scores below 24; (3) severe hearing or visual impairment; (4) unwillingness to comply with the protocol or procedures; (5) inability to communicate in Chinese Mandarin; (6) presence of serious pulmonary, heart, liver or renal insufficiency; and (7) had undergone anaesthesia or surgery within the past 30 days.
To determine POCD and cognitive decline, it is necessary to use a nonsurgical control group.19 The selection criteria were the same for the subjects controls, where subjects were matched to the elective hemiarthroplasty or total hip arthroplasty replacement surgery sample by age, sex and education, but they had no chronic pain. The controls were recruited from the general community. Both groups were recruited over the same time frame and tested at the same time intervals.
All patients underwent hemiarthroplasty or total hip arthroplasty. All clinical care followed routine clinical practice. All surgical plans were decided and performed in a standard manner by the same orthopaedic surgeon. All patients received general anaesthesia according to routine clinical practice. All patients received standardised perioperative care, including preoperative and intraoperative care and postoperative pain control.
Neuropsychological testing, including a battery of six neuropsychological tests, was performed by a well-trained personnel at baseline (the day before surgery) and at 7 days and 2 months postoperatively. The neuropsychological tests consisted of the MMSE, Visual Reproduction Test, Digit Span Test, Digit Symbol Test, Colour Trail Tests 1 and 2 and Stroop Colour and Word Test. All tests were conducted in the same order at each time point. These measures are not only highly sensitive to the types of cognitive impairments but also have no cultural bias.
Since these tests are prone to the test–retest practice effect, a age-matched control group was tested at the same intervals, providing an indication of practice effect with this test battery for the given time intervals between sessions.
Statistical analysis
Continuous data are reported in the tables as means and standard deviation (SD) or as median and interquartile range, and categorical data were reported as frequency and percentage. We examined the demographic and health characteristics per dichotomous pain variable (‘chronic pain’ versus ‘non-chronic pain’). Between-group differences were analysed by using analysis of variance or the Mann–Whitney rank sum test for continuous measures and the chi-squared test or Fisher’s exact test for categorical measures. Cognitive test scores were examined per dichotomous pain variable. The association between continuously valued variables and POCD was assessed by the difference in medians of patients with or without POCD. The association between categorical variables and POCD was assessed by the difference in risk for POCD between the category and baseline category in percentage points, including the association between POCD identified at 7 days and 2 months after surgery. Univariate logistic regression analysis was performed. All clinically relevant and statistically significant preoperative variables were then entered into a multivariate logistic regression analysis using a forward entry method to identify independent preoperative risk factors for POCD. Data are presented as the odds ratio and 95% confidence intervals. Values of p < 0.05 were considered to be indicative of statistical significance.
POCD identification
The Z-score was calculated according to the methods of Rasmussen.20 Briefly, the preoperative score was first subtracted from the postoperative score (7 days or 2 months) to obtain the difference, △X. Similarly, we calculated the difference in the scores for the healthy controls, △Xcontrol. The calculated difference, △X–△Xcontrol, was then divided by the standard deviation for the changes in test results in the control group, SD(△X)control. The formula was as follows:
where △X is the change in the neuropsychological test performance between different time points.
A combined Z-score was calculated as the sum of Z-scores divided by the standard deviation for this sum of Z-scores in the control group. POCD was defined as two individual Z-scores > 1.96 or a composite Z-score > 1.96. Data from the healthy control group were used to gain information on the practice effect and normal distribution in the test results for this age group, and the Z-scores were calculated.20
Sample size calculation
Pass (Version 15.0, NCSS, LLC, Kaysville, Utah, USA) software was used for the sample size calculation. Logistic regression tests for odds ratios with one binary X procedure were performed. With α = 0.05, a power of 80% and an odds ratio = 3.0, the POCD incidence of major joint replacement was approximately 20%, and chronic pain prevalence in such patients was at nearly 50%. 18,21 We then estimated that a total of 135 patients would be required for the study. To set the attrition rate at 10%, a total of 148 patients were required.