Rationale, methods, and progress of the ArthroCaP Study: A prospective cohort study exploring the associations between chronic postsurgical pain and postoperative cognitive dysfunction after elective knee or hip arthroplasty

ABSTRACT Background Chronic postsurgical pain (CPSP) may increase the risk of postoperative cognitive dysfunction (POCD) directly or by negatively impacting mobility. A comprehensive measure of satisfaction with surgery that accounts for ability to perform activities might be even more strongly associated with POCD. There might be complex mechanisms underlying the interplays between postoperative pain and cognition. Aims The primary objective is to explore whether CPSP at 6 months is associated with POCD (≥2-point decline in the Montreal Cognitive Assessment [MoCA] compared to preoperative) at 12 months. Secondary objectives are to explore the association between satisfaction with surgery and POCD, the role of pain medications (opioids) in the association between CPSP and POCD, the role of patient preoccupation/optimism and coping in determining the effect of pain on cognition, and the hypothesis of neurogenesis interference as an underlying mechanism. Methods We will prospectively recruit ≥100 patients ≥50 years old undergoing elective total knee/hip arthroplasty. Sociodemographic characteristics, comorbidities, frailty, pain, and pain medications will be assessed preoperatively. The Somatic Preoccupation and Coping questionnaire will be administered preoperatively and 4 to 6 weeks postoperative. Pain and the Self-Administered Patient Satisfaction Scale will be measured at 3 and 6 months. MoCA and neuropsychological tests will be administered at baseline, 4 to 6 weeks, and 6 and 12 months. Blood will be longitudinally collected for biomarker analysis. Progress Forty-six participants have been enrolled in the study so far. Conclusion ArthroCaP will provide preliminary data on the association of CPSP and patient satisfaction with POCD and underlying mechanisms. It will inform larger confirmatory and interventional studies to enhance the benefits of surgery.


Epidemiology of Cognitive Changes after Surgery and Rationale to Evaluate the Association between Persistent Postsurgical Pain and Postoperative Cognitive Dysfunction in Patients Undergoing Total Knee or Hip Arthroplasty
Surgery is associated with short and long-term changes in cognition, traditionally referred to as postoperative cognitive dysfunction (POCD), more recently defined under the concept of postoperative neurocognitive disorders. 1,2 A delayed neurocognitive recovery or transient POCD during the first 1 to 3 months after surgery is what had been most frequently described. 3 The more recent NeuroVISION study was a large international cohort study including a representative sample of 1114 patients aged 65 years or older undergoing different types of noncardiacsurgery. 4 In NeuroVISION, up to 30% of patients experienced a decline of ≥2 points in their Montreal Cognitive Assessment (MoCA) score at 1 year after surgery compared to prior to surgery. 4 A study in patients poststroke comparing MoCA with formal neuropsychological testing showed that a decline of ≥2 MoCA points corresponded to a diagnosis of cognitive impairment based on formal neuropsychological testing. 5 Moreover, a decline of ≥2 MoCA points over 1 year as observed in the NeuroVISION postsurgical population exceeds what is expected in communitydwelling older people. 5,6 The risk factors and determinants of POCD are poorly understood. 7 Up to 23% of patients undergoing total hip arthroplasty (THA) and 34% of patients undergoing total knee arthroplasty (TKA) experience persistent pain many months after surgery. 8,9 Chronic pain has been associated with structural brain changes. Neuroimaging studies in patients with chronic pain have shown a decrease in gray matter in the anterior cingulate cortex, insula, and dorsolateral prefrontal cortex. 10,11 These are areas known to be involved in the experience and anticipation of pain; however, they have also been linked to cognitive function across all neuropsychological domains and to performance on the MoCA. [12][13][14][15] It has been reported that people with chronic pain often show memory and concentration deficits. 16 A bidirectional interaction between pain and cognitive processing is typically hypothesized. 16 However, in a longitudinal cohort study of 10,065 community-dwelling older people, a history of persistent pain (of any origin) at enrollment was associated with accelerated memory decline and incident dementia over 10 years of follow-up. 17 In the surgical setting, clinical experience and evidence suggest that acute perioperative pain as well as perioperative opioid administration are possible triggers of perioperative delirium. 18,19 Small neuroimaging studies have also showed that procedures that successfully treat chronic pain can reverse brain anatomical and functional abnormalities. 10,20 In the NeuroVISION study, the odds of cognitive decline were lower with elective orthopedic surgery when compared to elective general surgery at 1 year (adjusted odds ratio 0.64, 95% confidence interval 0.44-0.93), 4 which could be explained by the expected beneficial effect of orthopedic surgeries, in most of the cases, in resolving chronic pain and impaired mobility, which may have a positive impact on cognitive performance. 21 However, whether CPSP or prolonged postoperative use of analgesics, particularly opioids, has a role in the cognitive changes observed in the 1 year following surgery remains uncertain. There is paucity of studies on the topic, and all focused on postoperative analgesia and cognitive changes in the very short term. [22][23][24]

Understanding the Possible Complexity between the Experience of Surgery and Cognitive Changes after Total Knee or Hip Arthroplasty
Pain is only one aspect of a patient's experience of surgery; persistent pain might not lead to negative cognitive changes per se but rather through its persistent impact on function and well-being. It is plausible to hypothesize that an overall measure of the patient's experience with surgery or of the impact of surgery on patients' lives, more than simply a measure of pain, might be linked with cognitive performance. The Self-Administered Patient Satisfaction Scale (SAPS) is used as a more comprehensive measure of patient recovery after surgery. 25 It consists of the following four items: the overall satisfaction with surgery, self-reported extent of pain relief, the ability to perform home or yard work, and the ability to perform recreational activities. 25 The instrument has excellent psychometric properties and has been used in large studies. [26][27][28] Studying the association of pain and cognition should take into account possible factors affecting the experience of pain but also possibly influencing how pain may affect the performance on certain cognitive domains. Pain and satisfaction are patient-reported outcomes; their reporting is the result of a subjective experience that is affected by individual tolerance and by how reality meets expectations. For instance, the risk of POCD increases with age; in contrast, older people complain less frequently about persistent pain after surgery. 29 This can be due to an altered perception of pain in an aging body system or because older people may not indicate that they are in pain despite perceiving it because this is normal or expected. 4,30 The Somatic Preoccupation and Coping (SPOC) questionnaire was developed by Busse and colleagues to assess and measure patients' somatic complaints, coping, energy, and optimism about their recovery from surgery. 31 It has been shown to predict 1-year pain persistence, interference of pain with normal work, and health-related quality of life 1 year after a traumatic, open extremity fracture repair, independent of age. 32 The SPOC questionnaire was originally validated in a population with a mean age of 39 years. 31

Understanding the Neurobiological Mechanisms Underlying the Association between Persistent Pain or Poor Satisfaction and Impaired Cognition after Total Knee or Hip Arthroplasty: The Neurogenesis Interference Hypothesis
The biological and neurobiological mechanisms by which pain and patient satisfaction with surgery may interact with cognitive abilities are unclear. One interesting hypothesis is that pain and satisfaction might selectively impair memory discrimination (or pattern separation) as measured through high interference memory tests (like the Mnemonic Similarity Test [MST]) and possibly interfere with hippocampal neurogenesis. 16 This is similar to what is observed in major depression and depression-related cognitive alterations. 33 Deficits in the MST have been associated with elevated scores on scales of stress, depression, and binge drinking and with increased age, whereas several weeks of high-intensity exercise improves MST scores. To assess whether the association between CPSP and POCD has similar features as to what is observed in major mood disorders, we can assess the correlation between changes in pain intensity (and incidence of CPSP) with the MST and separately with other cognitive/neuropsychological testing. As shown in patients with depression recovering from their depressive state, we might see improved MST scores in patients who experience a postoperative improvement in their pain while still showing a postoperative decline in other cognitive domains/tests and vice versa. 33 Numerous studies in animals have linked depression with hippocampal-reduced neurogenesis. [33][34][35] In the absence of a noninvasive measure of neurogenesis in humans, correlational evidence in human studies has identified performance on the MST as putatively neurogenesis sensitive because it is similarly impacted by factors that alter neurogenesis levels in rodents. 36,37 Although the contribution of adult neurogenesis to pattern separation remains unclear, 38 the hypothesis of a neurogenesis interference as a mechanism through which pain could affect cognition remains plausible and worth further evaluation. Neurotrophic factors such as brain-derived neurotrophic factor (BDNF) and insulin-like growth factor-1 (IGF-1), which support the survival and function of hippocampal cells, are regulators of neurogenesis. 39,40 BDNF and IGF-1 have been shown to enhance the association between higher aerobic fitness and better memory performance. Conversely, a reduction in neurotrophic factors might mediate the interaction of pain (and the consequent possible reduction in mobility) or patient satisfaction (as measured by SAPS, which includes also satisfaction with their ability to perform physical tasks) with cognition.

Objectives
We designed the ArthroCaP study as a proof-of-concept study with the main objective of exploring the association between postoperative pain and postoperative cognitive trajectories in people ≥50 years old undergoing elective TKA or THA. Specifically, our primary objective is to evaluate whether CPSP at 6 months is a risk factor for POCD at 12 months defined as ≥2-point decline in MoCA.
Secondary study objectives are to explore (1) whether CPSP at 3 months is a risk factor for POCD (based on MoCA) at 6 and 12 months; (2) whether postoperative long-term opioid use is associated with POCD and whether the association of CPSP and POCD is fully mediated by, or is independent of, the long-term use of opioids; (3) the association between patient satisfaction with surgery (as measured by the SAPS scale; i.e., a measure of the patient's experience of surgery more comprehensive than pain) and POCD; in particular, whether the SAPS score at 3 months predicts POCD at 6 and 12 months and whether the SAPS score at 6 months predicts POCD at 12 months; (4) whether changes in pain and satisfaction are associated with changes in (a) a high-interference memory test as a possible neurogenesis-sensitive memory test and (b) plasma neurotrophic factor levels as possible mediators of postoperative cognitive changes; (5) the association of preoperative level of somatic preoccupation and coping, as measured by the SPOC questionnaire and its change after surgery, with 6-and 12-month POCD; and (6) whether the preoperative level of somatic preoccupation and coping, as measured by the SPOC questionnaire and its change after surgery, affects (confounder) or modifies (effect modifier) the association between CPSP and POCD.
The current article reports the study methods and progress to date following the STrengthening the Reporting of OBservational studies in Epidemiology guidelines for reporting of cohort studies (see supplemental material).

Study Design
The ArthroCaP study is a prospective cohort study conducted at two tertiary hospitals in Hamilton, Ontario, Canada, St. Joseph's Hospital and Juravinski Hospital. It was initially launched as a substudy of a prospective cohort study looking at the incidence of CPSP titled Determinants of long-term outcomes in patients undergoing TKA (also known as SPOC study approved by the Hamilton

Participants
Patients are eligible for the ArthroCaP study if they (1) are 50 years old or older, (2) are scheduled for elective TKA or THA for osteoarthritis, (3) provide informed consent. Patients are not eligible if they present any of the following exclusion criteria: scheduled for revision arthroplasty; known history of dementia; unavailability of tablet or computer with an internet connection for remote assessment; inability to interact with a tablet or computer due to language, visual, or hearing impairment or any severely limited mobility of the upper limb joints; or inability to understand spoken or written English. Participants who have access to a tablet or computer with internet but have no experience in using them can be included as long as a family member or friend is available to provide technical assistance.

Recruitment and Procedures
Eligible patients will be approached 1 week to 1 month before surgery, during the preoperative assessment visit or during a mandatory standardized group educational session prior to surgery (i.e., "knee/hip class"). In accordance with the COVID-related restraints to research activities and their changes over times and as per participant choice, the study baseline and longitudinal assessments will be done in person (e.g., in coincidence with the preoperative knee/hip class or any visit performed for clinical visit), via telephone, or by e-mail (selfcompleted). All cognitive/neuropsychological assessments will be done remotely via videoconference or using ad hoc online test batteries. The biomarkers study component and its timeline have been arranged to facilitate blood collection even during the pandemic. Table 1 summarizes the study timeline, measurements, and data collection. The following sections provide additional details.

Measurements
Baseline Assessment. The baseline (i.e., preoperative) data collection includes the following: (1) demographic characteristics: age, sex, body mass index, family support (lives alone/family/partner or friend) (2) comorbidities (3) preoperative opioid usage; dose (morphine equivalent dose/day) and duration of use (4) pain a. target knee/hip joint pain: pain intensity using a 0 to 10 numeric rating scale b. chronic pain in other parts of the body aside from the target joint (needs attention or medication) (5) anxiety and depression using the Hospital Anxiety and Depression Scale (6) American Society of Anesthesiologists grading status (7) a FRAIL scale that combines components of the physical frailty phenotype with the number of illnesses.

Assessment of Pain, Satisfaction with Surgery, and Somatic Preoccupation and
Coping. Pain intensity (based on a 0-10 numeric pain scale) and pain medications will be self-reported at baseline and then 4 to 6 weeks and 3, 6, and 12 months after surgery. In our study, CPSP is defined as pain in and around the site of surgery of average intensity ≥4/10 over the past week and that started or increased after TKA/THA without any other identifiable cause and has lasted ≥3 months. The presence (incidence) and severity (intensity) of pain will be measured at 4 to 6 weeks and 3, 6, and 12 months after surgery.
Participants will rate their satisfaction with surgery according to the SAPS questionnaire at 3, 6, and 12 months. Each of the four SAPS items is scored on a 4-point Likert scale with response categories including very satisfied (100 points), somewhat satisfied (75 points), somewhat dissatisfied (50 points), and very dissatisfied (25 points). 25 The total score (0-100) is the unweighted mean of the scores from the individual items; the higher the score, the higher the level of satisfaction with surgery.
Participants will complete the SPOC questionnaire at baseline and again at 4 to 6 weeks after surgery. The questionnaire includes 27 items covering themes of somatic complaints (ten items), coping (six items), energy (seven items), and optimism (four items). Each item is rated on a 7-point Likert scale (e.g., from completely agree to completely disagree, with uncertain as the central point). Total scores range from 0 to 162, and higher scores represent worse coping, increased somatic The subset of patients who stay in hospital at least one night after surgery will have blood collected every morning on postoperative days 1-3 or until discharge (whichever comes first). b Whenever an in-person study visit is feasible, depending on restrictions in place due to the pandemic and patient compliance.
complaints, lower energy, and pessimism regarding recovery. 31

Cognitive/Neuropsychological Assessments
Cognitive/neuropsychological assessments will be conducted preoperatively and then at 4 to 6 weeks and 6 and 12 months after surgery. All assessments will be performed employing instruments validated for remote administration or online/electronic versions for selfadministration. Through videoconference, trained research personnel will administer or supervise the selfadministration of the tests, as appropriate. The videoconference platforms allow screen sharing to help the participant in case they need technical assistance. Participants will be allowed to have a friend or caregiver attending the help with technicalities, without interfering with the test completion. Overall, the assessments require between 45 and 60 min for completion. Each cognitive/neuropsychological evaluation will include the following: MoCA: this is a sensitive global cognitive screening tool that evaluates multiple cognitive domains, including orientation, memory, language, attention, calculation, constructions/visuospatial skills, executive function, and abstraction. We will use the MoCA version for administration through videoconference, which includes all subtests of a regular in-person MoCA. Time to administer the MoCA is approximately 10 to 15 minutes.
MST: the classical MST consists of a series of 192 color photographs of everyday objects on a white background. 38 Participants are first engaged in an encoding phase (no right or wrong answer) followed by a subsequent recognition test with a three-choice response: exact repetitions of earlier items, lures that are similar but not identical to earlier items, and novel foils. During the test, participants are asked to identify each item as "old," "similar," or "new." The recognition memory score is calculated as the difference between the rate of "old" responses given to repeat items minus "old" responses given to foils. The lure discrimination index (which measures mnemonic discrimination) is calculated as the difference between the rate of "similar" responses given to the lure items minus "similar" responses given to the foils. Typically, "normal" aging is associated with a decline in mnemonic discrimination (i.e., identifying lure objects that are similar to memory set objects as "similar") with preserved recognition memory (identifying repeated memory set objects as "old"). 41 We will use a validated modified online version of MST adapted to increase the efficiency of the test (i.e., to obtain the same information but with a shorter test) and therefore its acceptability to older participants. 42 Additional neuropsychological testing, including the following: • Visual paired associate learning (PAL), a computer task that assesses visual memory and new learning by testing a participant's memory for object location pairs (hippocampaldependent visuospatial memory). Boxes are displayed on the screen and opened in a randomized order. One or more of them will contain a pattern. The patterns are then displayed in the middle of the screen, one at a time, and the participant must touch the box where the pattern was originally located. If the participant makes an error, the patterns are represented by their locations. The difficulty increases throughout the test. The average accuracy across the last two completed levels will be analyzed. 33 • Rapid serial visual presentation, which assesses executive functions such as working memory and sustained attention. A white box is shown in the center of the screen, inside which digits from 2 to 9 appear in a pseudorandom order, at the rate of 100 digits per minute. Participants are requested to detect target sequences of digits (for example, 2-4-6, 3-5-7, 4-6-8). When the participant sees the target sequence, they must respond by selecting the button in the center of the screen as quickly as possible. The level of difficulty varies with either one-or three-target sequences that the participant must watch for at the same time. The average accuracy throughout the task will be analyzed. • Stroop Color-Word Interference Test (CWIT), which measures inhibitory control and cognitive flexibility. The CWIT utilizes the "Stroop effect" to identify deficits in inhibition of overlearned behaviors. In addition to standard color naming, word reading, and color-word interference trials, the CWIT includes a trial requiring the examinee to switch between inhibitory and noninhibitory responses. Process measures enable assessment of general difficulties. ○ Condition 1: Color Naming provides a baseline for parceling out basic naming skills from higher-level functions of verbal inhibition (Condition 3) and switching (Condition 4). ○ Condition 2: Word Reading serves as a baseline for parceling out the contribution of basic reading skills. ○ Condition 3: Inhibition reflects the ability to inhibit the more salient, automatic task of reading words in order to name the color of the ink in which the words are printed. ○ Condition 4: Inhibition/Switching requires adequate naming speed, reading speed, verbal inhibition, and cognitive flexibility. Scores include the time needed to complete each condition, the number of errors, and the number of selfcorrected errors. 43 • N-back task, which assesses visual working memory. Participants are shown a series of numbers on the screen and are asked to enter the number they saw two numbers ago. The outcome measures are correct detection and reaction times. 44 Subjective memory complaints (SMCs) 45 : participants are asked the following four questions: (1) "Do you have any difficulty with your memory?" (2) "Do you forget where you have left things more than you used to?" (3) "Do you forget the names of close friends or relatives?" and (4) "Do other people find you forgetful?" A positive response to any of these questions indicates SMCs. 29,46 Other Data Collections Information regarding surgery and clinical complications will also be collected through participant interview and chart review.

Biomarkers Study
Participants who consent to the biomarker study will have their blood drawn preoperatively when they are admitted to the hospital on the morning of surgery. For the postoperative time points, blood for the biomarker study will be drawn 2 to 4 h after surgery for each study participant. Participants who stay at least one night in the hospital after surgery will also have their blood drawn daily from postoperative day 1 to postoperative day 3 or the day of discharge (whichever comes first), with the blood being preferentially drawn in the morning and with fasting. Blood will be also collected at 4 to 6 weeks, 6 months, and 12 months after surgery. The blood collection will be performed during clinic visits whenever possible (for example, the 1-month surgical follow-up after the procedure) or at the hospital outpatient labs according to procedures established and implemented during the COVID-19 pandemic for several local studies. Blood collected at each time point will be used in part for the prespecified study biomarkers and, in part, upon specific patient consent, for biobanking.
We will longitudinally measure the levels of the following biomarkers: Neurotrophic factors that are possible mediators of the effect of pain or satisfaction on neuroplasticity and possibly cognitive performance, including the BDNF and IGF-1. We will also include the measurement of nerve growth factor (NGF), another neurotrophic factor that is associated with neuronal survival and cognitive performance. 47 We will measure the plasma levels of neurotrophic factors but also their RNA expression in blood leukocytes 48 ; the N-terminal pro-B-type natriuretic peptide (NT-proBNP); preoperative NT-proBNP is a known independent predictor of postoperative outcomes after noncardiac surgery, including mortality and cardiovascular events. 49 In the nonoperative setting, NT-proBNP has been also shown to correlate with measures of physical performance and tolerance (like peak oxygen consumption, VO 2 peak) that could be affected by persistent pain and could be related with satisfaction. 50,51 We will explore how changes in NT-proBNP correlate with changes in pain and satisfaction, as well as with changes in neurotrophic factor levels.
We will collect venous blood into EDTA tubes (to obtain both plasma and whole blood for DNA measurements) and RNA PAXGene tubes for blood RNA analysis. At each time point, we will collect 10 mL of blood in EDTA and 10 mL of blood in RNA PAXGene tubes. Samples will be collected at the participating hospitals into preprepared study kits and processed and aliquoted on-site based on standard procedures as defined in a specific study manual. In particular, blood samples will be incubated in the RNA PAXgene tubes for a minimum of 2 h at room temperature to completely lyse blood cells; they will then be frozen first at −20°C for 24 h and then at −80°C, stored upright in a wire rack. Samples will be batched and transported to the Clinical Research Laboratory and Biobank (CRLB)-Genetic and Molecular Epidemiology Laboratory (GMEL) in Hamilton, Ontario, Canada. Part of the samples will remain at the CRLB-GMEL and used for NT-proBNP testing (using the Roche Elecsys ProBNP II (V2) assay, run on the Roche Cobas e 602 instrument) and for biobanking. From the CRLB-GMEL, part of the aliquots will be transported to the McMaster University laboratory in Hamilton, led by M.F., for neurotrophic biomarkers analyses (using enzyme-linked immunosorbent assay [ELISA]). Serum BDNF, proBDNF, and IGF-1 will be using the Human BDNF DuoSet ELISA kit DY248, Human proBDNF DuoSet ELISA kit DY3175, and Human IGF-1 Quantikine ELISA kit DG100. The ancillary kit will be used for BDNF and proBDNF. For BDNF measurements, serum samples will be diluted 75×; four readings will be carried out for each sample, each reading representing an independent dilution. For proBDNF measurements, serum samples will be diluted 20×. For IGF-1 quantification, serum samples will be pretreated according to the manufacturer's instructions; final dilution of the samples will be 100×. Samples and standards for IGF-1 ELISA will be analyzed in duplicate, and final values will be the average of two readings, each reading representing a separate dilution. Absorbances for BDNF, proBDNF, and IGF-1 will be read at 450 nm, with a reference reading at 540 nm, using a Multiskan GO plate reader and SKANIT 3.2 software. NGF will be measured using an in-house, two-site sandwich ELISA. 48 The primary antibody is an affinitypurified rabbit polyclonal antibody, following protocols from Boehringer Mannheim, with the fluorescent substrate 4-methylumbelliferyl β-D-galactoside. The polyclonal antibody recognizes proNGF (R-1G) and NGF with roughly equal affinity as estimated by Western blotting. Samples will be measured at three different dilutions, each in triplicate.
For participants who agree to have their blood samples stored and their genetic material tested, we will biobank the samples for use in future studies to explore other possible biomarkers (potentially including protein, metabolites, RNA expression, and DNA variants) as suggested by new data and hypotheses on possible mechanisms underlying the association of CPSP and POCD.

Sample Size
The ArthroCaP study has been designed as a proof-ofconcept, hypothesis-generating study, with the aim of informing future larger studies. This is consistent with the intent of the catalyst grant that supports the study. The study objectives are therefore exploratory, and the number of participants recruited is primarily dictated by convenience. The study was initially conceived as a substudy of the ongoing "SPOC Study in Knee Arthroplasty," and we initially expected to be able to recruit 150 to 200 of the 388 participants who would be enrolled in the parent study. We expect that 20% of the study participants will experience a ≥2-point decline in MoCA at 1 year postsurgery (considering the possibly younger age and different representation of surgery types in our study compared to NeuroVISION) and that 30% will present with moderate to severe CPSP at 6 months. Given these assumptions, enrolling ≥150 participants would give a power of ≥80% to detect an association between moderate to severe CPSP at 6 months and a ≥2-point decline in MoCA at 1 year, with an odds ratio of 3.0 when we consider some residual correlation between independent variables (i.e., R 2 ≤ 0.2) even after removing collinear covariates. However, with ≥150 participants and a 20% outcome rate, we will be also limited in the number of covariates we can include in the multivariable analysis by the relatively small number of expected events (i.e., 30 patients experiencing ≥2-point decline in MoCA).
Due to the disruptions to recruitment secondary to the pandemic, we identified the recruitment of ≥100 patients eligible for our analyses (i.e., ≥100 patients who completed at least the 6-month follow-up) as a more realistic target, being aware that the study will provide preliminary nondefinitive data to inform future research.

Statistical Analysis
Baseline characteristics will be summarized through descriptive statistics; that is, mean (standard deviation), median (interquartile range), or number (%) as appropriate. We will first evaluate the primary research question; thaht is, the association between CPSP at 6 months and POCD, defined as at least a ≥2-point decline in MoCA score at 1 year by conducting a multivariable logistic regression with 6-month CPSP as the independent variable and 1-year POCD as the dependent variable. We will then perform similar logistic regressions with CPSP at 3 months as the independent variable and 6-month and 1-year POCD (≥2-point MoCA decline) separately as dependent variables.
Subsequently, we will also model longitudinal changes in pain scores (independent variable) and changes in cognitive/neuropsychological scores (dependent variable) over time (4 weeks and 3, 6, and 12 months after surgery) using multilevel (mixed effects) regression models, considering time points nested within patients. 52 We will first use MoCA scores as the dependent variable. Secondarily, we will perform multivariate mixed effects regression models to model longitudinal changes in the neuropsychological test scores (paired associate learning, rapid serial visual presentation, Stroop CWIT, N-back task, and MST score) as multiple dependent variables. We will examine whether changes in pain at 3 or 6 months after surgery predict subsequent changes in neuropsychological performance at 6 and 12 months or at 12 months, respectively. 17 Through post hoc analyses of the same models, we will look at the changes in single test performance, with Bonferroni correction of alpha values to adjust for multiple comparisons. These post hoc analyses will enable us to test the neurogenesis hypothesis that changes in pain (and pain persistence) are associated with changes in MST but not changes in other cognitive domains. We will perform similar univariate and multivariate mixed effects regression models to assess the association between pain and biomarkers (neurotrophic factors and NT-proBNP) and biomarkers and changes in cognitive/neuropsychological performance. We will analyze whether biomarker levels at baseline (before surgery) and right after surgery (or at 4-6 weeks after surgery) predict cognitive changes at 6 and 12 months.
To examine the association between patient satisfaction with surgery and cognitive/neuropsychological performance, we will use the same statistical approaches but using SAPS scores instead of pain as independent variable.
All models will be adjusted for relevant covariates, including baseline characteristics (e.g., patient age, sex, and anxiety/depression level), comorbidities, and pain medications (as time-varying variables when relevant). We will adopt a parsimonious approach in covariate selection, based on clinical relevance and removing collinear covariates (i.e., variance inflation factor >2.5) according to prespecified priorities.
We will conduct subgroup analyses to explore age as a possible effect modifier in our analyses. Including participants with a relatively broad range of ages will allow us to test the interaction with age. To evaluate the role of somatic preoccupation and coping as measured by the SPOC questionnaire, we will first analyze the association between baseline SPOC and baseline cognitive performance in a traditional multivariable linear regression analysis. Similarly, we will analyze the association between early postoperative (4-6 weeks) SPOC and cognitive performance, after adjusting for baseline measurements. Then we will analyze whether preoperative SPOC, postoperative SPOC (4-6 weeks), or change in SPOC at 4 to 6 weeks compared to preoperatively predicts cognitive decline at 6 and 12 months using linear and logistic (i.e., with ≥2-point decline in MoCA as binary outcome) regression models. Secondarily, we will run separate regression analyses to explore the association with cognitive changes of each subset of items covering each of the four themes included in the SPOC questionnaire (somatic complaints, coping, energy, and optimism) separately. For all analyses we will perform a complete case analysis.

Progress to Date
The ArthroCaP study is currently ongoing at two sites in Hamilton, Ontario. Recruitment began in November 2021. However, due to the intermittent stoppage of elective surgeries due to provincial COVID-19 restrictions (December 2021 and January 2022), recruitment was interrupted. To date, a total of 166 patients have been screened, out of whom 54 participants were enrolled. Currently 46 participants are actively enrolled in the study and 10 participants discontinued participation. The main reason for ineligibility was no access to a computer or inability to use Zoom (27) and having bilateral arthroplasty (12). We were unable to include 12 participants who were eligible due to lack of time to schedule baseline assessments before surgery. Reasons for declining participation included lack of interest (26), participation in other research studies (28), and no reason provided (7). Reasons for discontinuation of study participation included deterioration of health (1), loss of interest (1), no response (3), difficulty concentrating (4), and other (1). Currently, 46 participants are actively enrolled in the study, of whom 17 participants consented to participating in the biomarker study. All participants consenting to the biomarker portion of the study completed their preoperative and postoperative follow-up blood collection according to their scheduled times, with 53% having completed the 6-month follow-up to date. We have observed no loss to follow-up or dropouts from the participants in the biomarker portion of the study. Characteristics of the enrolled participants are provided in Table 2.

Importance of the Study
When asked about their fears related to undergoing major surgery with general anesthesia, people across different ages indicate "brain damage" and "memory loss" as main concerns much more frequently than perioperative death. 53 Moreover, the literature shows that even with only a partial understanding of the concept of likelihood, most older people would not choose a treatment when the expected outcome is survival associated with cognitive and functional impairment. 54,55 Persistent postoperative cognitive dysfunction is highly prevalent in older people undergoing surgery; CPSP can also occur in older surgical patients. Both of these conditions are associated with reduced autonomy and increased burden to patients, caregivers, and health care systems.
In contrast, the existing literature shows a decrease in quality of life in people with chronic pain waiting >6 months for assessment. 56 Individuals waiting for major joint surgery report that their waiting time contributes significantly to the deterioration of their health. 57 Likewise, the current experience with COVID-19 and the  13 (29.5) deferral of elective surgeries highlight what delaying such treatment might mean to older people with chronic conditions. Our study will provide preliminary data on the role of chronic pain and chronic pain medications in the impact of surgery on older people's cognitive abilities and on the underlying mechanisms. It will inform larger confirmatory studies and the design of interventions that aim to enhance the benefits of surgery in older patients.