2.1. Study design

This investigation is designed as a prospective, randomized controlled trial with a triple-blind protocol. The study will be conducted at an academic hospital in China: Honghui Hospital, Xi’an Jiaotong University. This protocol adheres to the standard protocol items: recommendations for interventional trials (SPIRIT) 2013 guidelines, including the checklist and diagram (Fig 1), and describes the study design, methods, and statistical analysis plan; results will be published separately upon trial completion [45].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study assessment schedule (Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) Figure).

BPI: Brief pain inventory; SF-36: 36-Item Short Form Health Survey; WOMAC: Western Ontario and McMaster Universities Osteoarthritis Index; ICOAP: Intermittent and Constant Osteoarthritis Pain Questionnaire; ROM: Range of motion; CSI: Central sensitization inventory; HMAD: Hamilton Depression Rating Scale; PD-Q: PainDETECT questionnaire; CGI-I: Clinical global impression of improvement; PGI-I: Patient Global Impression of Improvement.

https://doi.org/10.1371/journal.pone.0334400.g001

This protocol was approved by the Independent Ethics Committee of Honghui Hospital, the Affiliated Hospital of Xi’an Jiaotong University (No. 202403056) and registered (registration date: 18 March 2024) at Chinese Clinical Trial Registry (http://www.chictr.org.cn, identifier: ChiCTR2400081990). Details of the study will be explained thoroughly to the potential participants by the study investigator. The informed consent form must be signed and provided by all eligible individuals before enrollment. Participants may withdraw from the study at any time without any reason. The participants can withdraw from the study without any reason at any time. This trial does not involve collecting biological specimens for storage. The privacy of all participants will be strictly protected. Any confidential information, such as name, identification number, hospital administration number, etc. will not be exported, and data anonymity will be executed during the process of data analysis and management.

Eligible candidates for primary unilateral TKA will be enrolled and assessed based on a rigorous set of predefined inclusion and exclusion criteria. The study is planned to commence in August 2024, with a projected recruitment phase exceeding 12 months, to be succeeded by a comprehensive 12-month follow-up period for the enrolled participants. Adhering to the principles outlined in the SPIRIT 2013 guidelines, the protocol for this study has been meticulously formulated to guarantee both clarity and scientific rigor in the design and execution of the trial [45]. The trial flow chart is shown in Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Patient disposition flow chart.

CSI: Central sensitization inventory; DLX: duloxetine; PGB: pregabalin.

https://doi.org/10.1371/journal.pone.0334400.g002

2.2. Study population

Within the scope of this study, a skilled investigator who is proficient in clinical research practices and possesses the necessary medical qualifications will be responsible for identifying potential candidates from a group of consecutively recruited individuals who have agreed to participate. The selection of participants for the trial will be carefully carried out by applying established inclusion and exclusion criteria to ensure their suitability for enrollment. The process of obtaining informed consent will be conducted by this trained researcher, who is knowledgeable about good clinical practice (GCP) standards. This will protect the ethical integrity of the study and ensure compliance with all regulatory requirements.

2.3. Inclusion criteria

To ensure the rigor and appropriateness of participant selection in this investigation, the following inclusion criteria have been meticulously established:

1. Eligible participants must have a diagnosis of knee OA, confirmed through clinical evaluation and imaging studies, and must be scheduled for their primary TKA;
2. Participants must be aged fifty years or older;
3. They should possess clear communication capabilities to facilitate precise conveyance and comprehension of study-related information;
4. The physical status of the participants should be classified between Class I and Class III, according to the American society of anesthesiologists (ASA) physical status classification system;
5. Voluntary participation is mandatory, with participants providing signed informed consent.

2.4. Exclusion criteria

To maintain the integrity and safety of the study, participation will be precluded for individuals meeting any of the following exclusion criteria:

1. History of other surgical interventions (e.g., hip, major thoracic or abdominal surgery) within the past year;
2. Prior open surgery on the same side of the knee joint;
3. Intra-articular injections or arthroscopy of the surgical knee within the last three months;
4. Planned contralateral TKA or any other surgical procedures during the study period;
5. Severe preoperative comorbidities anticipated to result in hospitalization or significant impairment of study participation;
6. Documented history of significant peripheral nerve injury;
7. Have neurological disorders or cognitive impairments affecting questionnaire completion (e.g., Alzheimer’s disease, dementia);
8. Allergy to sulfa drugs;
9. History of alcohol or substance abuse or dependence within the last five years preceding enrollment;
10. History of arrhythmias, heart failure, myocardial infarction, or cardiac dysrhythmias at the time of enrollment;
11. Uncontrolled hypertension with systolic blood pressure >180 mm Hg or diastolic blood pressure >110 mm Hg at screening;
12. History of glaucoma (or elevated intraocular pressure), uncontrolled thyroid disease, or uncontrolled seizure disorders;
13. Hyponatremia (sodium <135 mmol/L) or a history of recurrent hyponatremia;
14. History of peptic ulcer disease or bleeding disorders;
15. Impaired liver function (ALT or AST > 100 IU/L or international normalized ratio (INR) >1.5), known cirrhosis, or liver transplantation;
16. Severe renal impairment (creatinine clearance <30 mL/min), history of renal transplantation, or undergoing dialysis;
17. Allergy or intolerance to duloxetine, pregabalin, or anesthetic agents;
18. Long-term use of gabapentin or pregabalin (regular use for more than three months);
19. Previous use of duloxetine or alternative serotonin-norepinephrine reuptake inhibitors (SNRIs), selective serotonin reuptake inhibitors (SSRIs), monoamine oxidase inhibitors (MAOIs), and tricyclic antidepressants;
20. A score >7 on the Hamilton Depression Rating Scale (HAMD);
21. Inability to complete study questionnaires;
22. Refusal to undergo randomization.

2.5. Randomization and allocation

To ensure the scientific rigor of the study design and minimize bias, the randomization scheme will be designed and supervised by an independent statistical expert who is not involved in patient recruitment or clinical management. The random allocation sequence will be generated using validated statistical software (SAS 9.4). Eligible patients will undergo a preoperative assessment for CS using the central sensitization inventory (CSI) as the evaluative instrument [46–48]. Patients with a CSI score of less than 40 will be assigned to Group 6, where they will receive the same intervention as Group 1. Patients with a CSI score of 40 or above will be randomly allocated to Groups 1–5 in a 1:1:1:1:1 ratio, with block sizes of 5 or 10 randomly chosen to reduce the predictability of allocation.

2.6. Blinding

This investigation is designed as a prospective, randomized, triple-blind clinical trial. Throughout the trial, all individuals involved in data collection, including patients, surgeons, clinical researchers, and statisticians, will strictly adhere to a blinding protocol regarding group allocations. The blinding will only be revealed once the final data analysis is completed. To ensure the blinding is maintained for all participants, including medical professionals and researchers, the subjects will receive capsule formulations of duloxetine and pregabalin, or placebos of equal appearance, in a random manner determined by a computer. The capsules have been carefully created to be identical in color, size, taste, smell, labeling, and packaging. This ensures that subjects are unable to determine the medication they are ingesting.

A designated staff member at the medical institution’s pharmacy will be responsible for dispensing duloxetine, pregabalin, and placebos to participants. They will have no knowledge of the specific group assignments. The medications will be distributed to participants according to a computer-generated randomization schedule, following the study protocol strictly. It is crucial to adhere to this process to ensure that each participant receives the study medication or placebo at the prescribed dosage and timing. The capsule count and administration intervals will be kept consistent across all the groups. The staff member will also meticulously document each participant’s medication compliance, which is essential for analyzing and interpreting the data later on.

The task of collecting, managing, and analyzing the data will be entrusted to a separate, unbiased team. This team will not be informed of the group assignments, ensuring the objectivity and reliability of the study results. The process of revealing the randomization and assignment will only occur once the required number of participants have been enrolled and the data analysis has been thoroughly completed. To validate the effectiveness of the blinding process, we will use bang’s blinding index (BBI) to assess the degree of blinding maintained throughout the trial [49]. After the intervention, patients will be asked to complete a blinded assessor questionnaire. This questionnaire will include a single question asking them to guess which treatment they received (such as “Guess Duloxetine,” “Guess Pregabalin,” “Guess the co-medication,” “Guess Placebo,” or “Do not Know”). This measure will help us evaluate how successful the blinding process is.

Data regarding participant treatment guesses and the calculated BBI will be reported descriptively in the Results section to quantitatively assess the effectiveness of blinding. These findings will be used to aid in the interpretation of the study outcomes and will be discussed in the context of potential study limitations in the Discussion section.

2.7. Treatment

2.7.2. Interventions.

This study will be structured into three distinct phases: an initial treatment phase that lasts six weeks, a high-dose/combination therapy phase that also lasts six weeks, and a two-week tapering-off period. The therapeutic protocol, described in Fig 3, will involve daily administration of 60 mg of duloxetine to patients in Groups 2 and 3, and 300 mg of pregabalin to those in Groups 4 and 5 during the initial treatment phase. Groups 1 and 6 will be given a placebo.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Study design.

DLX: duloxetine; PGB: pregabalin.

https://doi.org/10.1371/journal.pone.0334400.g003

The transition to the high-dose/combination therapy phase will depend on evaluating the effectiveness of the treatment, which will be determined by changes in the brief pain inventory (BPI) 24-hour average pain score during the initial treatment period. Patients who experience a reduction in pain of over 30% will be considered “treatment responders” and their treatment will be stopped. Conversely, “non-responders,” who experience less than 30% pain relief, will continue to receive intensified therapy [43]. Group 2 will receive a daily dose of 120 mg of duloxetine, while Groups 3 and 4 will receive a combination of 60 mg of duloxetine and 300 mg of pregabalin daily. Group 5 will be treated with a daily dose of 600 mg of pregabalin. Non-responders in Groups 1 and 6 will remain on placebo. The dosages of the study drugs will be systematically reduced during the subsequent tapering period. Dosages will also be tapered for patients who discontinue treatment earlier.

In cases of significant intolerance to the prescribed dose levels, patients will be advised to decrease their dosage under the guidance of the investigating team for one week. The purpose will be to then gradually increase the dosage back to the originally prescribed level. If patients still experience intolerance after dosage adjustment, the study medication will be stopped to prioritize patient safety.

2.8. Withdrawal criteria

At any point during the study, participants will retain the freedom to withdraw their participation without needing to give a reason. This will not affect their relationship with healthcare providers. Participants can be assured that their decision to withdraw will not impact their access to medical care, their rights, or any follow-up procedures. There will be no negative consequences or penalties for withdrawing. In certain circumstances, such as concerns for patient safety or other significant factors, the chief investigator or regulatory authorities may decide to end a participant’s involvement in the clinical trial. To ensure the well-being of participants, individuals who are considering stopping their medication or withdrawing from the study will be strongly encouraged to consult with the research team for guidance on how to manage the discontinuation process.

2.9. Outcome measurement

Outcome metrics and patient characteristics will be systematically collected from electronic hospital information systems and medical charts. In addition, physical examinations will be conducted, and patients will be asked to complete questionnaires themselves.

2.10. Patient characteristics

In this investigation, the researchers aim to gather baseline descriptive data to assess the participants. The data collected will include age, gender, height (cm), weight (kg), body mass index (BMI), and education level. Additionally, this study will document the duration of OA pain symptoms and evaluate the severity of OA using the Kellgren-Lawrence (K-L) grading system. The consumption of medications related to OA pain will also be recorded. Furthermore, the health status of the participants will be assessed using the ASA Physical Status Classification System. Lifestyle factors, including smoking behaviors and alcohol intake patterns, will be duly considered. The researchers will document the participants’ medical history, existing comorbidities, medication history, and previous health issues to gain an understanding of their health status. All the data will be collected and processed confidentially to ensure participant privacy.

2.11. Safety

The safety evaluation will entail rigorous monitoring, recording, and reporting of all adverse events (AEs). The assessment will employ a combination of systematic and non-systematic methods. During each scheduled study visit in the six months following the procedure, participants will be systematically queried about preset common AEs related to the study medication using a standardized checklist. These AEs include drowsiness, insomnia, fatigue, dizziness, headache, dry mouth, decreased appetite, constipation, nausea, vomiting, peripheral edema, and diplopia [44]. Furthermore, all AEs reported spontaneously by participants or observed by investigators at any time, whether expected or related to the study treatment, will be thoroughly documented.

All recorded AEs will be coded using Medical Dictionary for Regulatory Activities (MedDRA, version 27) terminology, including Preferred Terms and System Organ Class. Serious adverse events (SAEs), defined as events leading to death, life-threatening experiences, prolonged hospitalization, or persistent/significant disability/incapacity, will be promptly reported to the principal investigator, ethics committee, and relevant regulatory authorities. The HAMD will be used at specific time points to monitor mental health status, including depressive symptoms and potential suicidal tendencies, as specific safety outcomes. Any clinically significant safety findings or events that may require discontinuation of treatment will be timely evaluated by the clinical team and managed according to best clinical practices and the study protocol.

In the final trial publication, the safety data report will include: (1) all SAEs, (2) AEs leading to treatment discontinuation, and (3) treatment-emergent adverse events (TEAEs) occurring in at least 5% of patients in any treatment group, presented by MedDRA System Organ Class and Preferred Term. A summary table detailing the overall incidence of AEs across treatment groups will also be provided. If appropriate, statistical tests may be performed to report differences in the frequency of key AEs between groups.

An interim analysis will be conducted after 50% of participants have completed the 6-month follow-up, using a group sequential design with an O’Brien-Fleming alpha-spending function to control the type I error rate at 0.05 [58]. The Data and Safety Monitoring Board (DSMB) will review interim results to assess both efficacy and futility. Based on emerging data, the DSMB may recommend trial continuation, modification, or early termination, even in the absence of formal stopping rules. Throughout the trial, the DSMB will monitor adverse events and efficacy data to ensure participant safety and ethical conduct. After the 12-month follow-up, participants in the monotherapy group will be offered access to the combination intervention if it is shown to be superior and no new contraindications arise.

2.12. Data collection and management

Trained personnel will collect demographic and baseline characteristic data during patient recruitment and registration (Table 1). Independent researchers will document clinical outcomes, survey responses, complication rates, and adverse events during hospitalization and follow-up. The researchers will follow strict data collection protocols to ensure accurate and reliable information. Additionally, an independent data monitoring committee will oversee the data collection process to maintain the quality and completeness of the data. No new participants will be enrolled during the re-evaluation phase of the study to avoid data confusion and potential bias. Each participant will receive a unique identification code to ensure patient privacy and data anonymity. Clinical and survey data will be collected and stored using this code. A patient identification checklist will link the data to the respective patients. The principal investigator will securely hold the key to access these codes, ensuring the security and confidentiality of the data.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics.

https://doi.org/10.1371/journal.pone.0334400.t001

2.13. Sample size

The objective of this study is to evaluate the therapeutic effects of different treatment strategies on postoperative pain in patients with CS following TKA. Assuming a standard deviation (SD) value of 2.5, a two-sided test with 90% statistical power was used as the evaluation criterion. The study aims to detect a minimum difference of 1 point in the 24-hour average pain score on the BPI between patients subjected to high-dose monotherapy groups (Combination Groups 2 and 5) and those receiving combination therapy (Combination Groups 3 and 4) [43]. This 1-point threshold was chosen based on the methodological precedent established in the COMBO-DN trial, which employed the same effect size and analytic framework [43]. Furthermore, previous studies have shown that a 1-point change on the BPI scale (ranging from 0 to 10) represents the MCID for chronic pain conditions [59]. From the patient perspective, a 1-point reduction is considered a small but noticeable and meaningful improvement in pain intensity [60–62].

Sample size calculations are being carried out using the statistical software nQuery Advisor 7.0 (Statistical Solutions, Saugus, MA, USA). To ensure the validity of the analysis, each treatment group requires a minimum of 135 participants, as determined by these calculations. The clinical trial design takes into account the expected response rate to the treatment and the possibility of treatment discontinuation. It is anticipated that approximately 60% of participants will respond positively to the initial treatment, while an estimated 15% might discontinue treatment for various reasons during the study [43]. Based on these clinical scenarios, we intend to randomly assign a total of 1200 patients. This strategy aims to ensure that each treatment group has an adequate sample size for the data analysis phase, thereby enhancing the reliability and validity of the study’s findings.

Furthermore, multiple strategies will be implemented to maximize participant retention and minimize loss to follow-up, including: (1) a comprehensive informed consent process that clearly outlines study expectations and commitments; (2) regular follow-up contacts via participants’ preferred methods (clinic visits, phone calls, emails, or text messages); (3) flexible scheduling options for follow-up assessments; (4) appointment reminders; and (5) providing participants with contact information for a dedicated study coordinator to address any questions or concerns promptly. We will closely monitor the loss to follow-up rate throughout the entire trial process.

2.14. Statistical analysis

For the data analysis in this study, we will use IBM SPSS software (version 26.0, IBM Corp., New York, NY, USA). Continuous and descriptive numerical data will be reported as the means ± SDs, while categorical variables will be presented as frequencies and percentages. Statistical significance will be determined by a P value of less than 0.05. All analyses will adhere to the principles of intention-to-treat (ITT). The efficacy population for the analysis will include all patients who received treatment and had at least one post-baseline efficacy assessment during the high-dose monotherapy/combination therapy period. The safety population will consist of all patients who are randomly assigned and received treatment. The baseline assessment prior to the initiation of treatment will be used as the reference point, and the follow-up at six weeks post-surgery will mark the beginning of the high-dose monotherapy/combination therapy period.

The primary efficacy analysis will use a Mixed-Effects Model for Repeated Measures (MMRM) to compare changes in the 24-hour average pain score on the BPI between high-dose monotherapy and combination therapy. The MMRM will include fixed effects for treatment group (high-dose monotherapy vs. combination therapy), visit, treatment-by-visit interaction, initial treatment (duloxetine vs. pregabalin), baseline pain score, and baseline-by-visit interaction, with an unstructured covariance matrix. The MMRM does not include subject-level random effects. Within-subject correlations across visits are modeled using an unstructured covariance matrix to accommodate complex correlation patterns in longitudinal pain score data. Model fit will be assessed using likelihood ratio tests or information criteria, with alternative covariance structures will be considered if convergence issues arise. Results will report estimated means and 95% confidence intervals (CIs) for each treatment group, as well as the between-group mean difference, 95% CI, and corresponding P-value.

Secondary efficacy analyses will be conducted as exploratory measures. The efficacy of different treatments within each treatment period will be compared using the MMRM model. Safety outcomes will be summarized descriptively. The frequencies of patients experiencing TEAEs across different treatment approaches will be compared using Fisher’s exact test.

The MMRM, based on Maximum Likelihood Estimation (MLE), effectively utilizes all available observations under the Missing at Random (MAR) assumption [63,64]. To support this assumption, exploratory analyses will compare baseline characteristics such as age, sex, and baseline pain score between patients with and without missing data to evaluate missingness patterns. If the proportion of missing data for the primary efficacy endpoint (BPI 24-hour average pain score) exceeds 5%, multiple imputation (MI) will be conducted as a sensitivity analysis to assess the robustness of the findings [65]. The MI model will include all fixed-effect variables from the primary MMRM analysis and will generate 50 imputed datasets, selected based on Monte Carlo error convergence [66]. Each dataset will be analyzed using MMRM, and results will be pooled using Rubin’s rules to evaluate the impact of missing data on the primary conclusions [65,67]. To account for potential Missing Not at Random (MNAR) mechanisms, additional sensitivity analyses will be performed, including: (1) pattern-mixture models (PMM) stratified by treatment group or visit timepoint; (2) worst-case scenario analysis, where missing BPI scores are imputed as maximum pain scores to reflect treatment failure; and (3) adjusted MI models incorporating dropout reasons, such as adverse events or loss to follow-up, as covariates. These sensitivity analyses will be compared with the primary MMRM results to evaluate the robustness of conclusions under alternative missing data assumptions [64].

2.15. Cost-effectiveness analysis

To support clinical relevance and guide decision-making, a cost-effectiveness analysis (CEA) will be conducted from the perspective of the Chinese healthcare system, comparing high-dose duloxetine monotherapy, high-dose pregabalin monotherapy, and combination therapy for the treatment of postoperative chronic pain associated with central sensitization. The analysis will evaluate direct medical costs, such as medication use, adverse event management, follow-up consultations, and rehabilitation services, alongside treatment effectiveness measured in quality-adjusted life years (QALYs), derived from EQ-5D-5L utility scores [68]. A trial-based evaluation will be conducted using data collected over a 24-month period, during which costs and QALYs will be estimated for each intervention, stratified by central sensitization levels based on the CSI. Pain outcomes will be assessed using the 24-hour average pain score from the BPI, and generalized linear models will adjust for baseline covariates such as age, sex, and baseline pain score. Incremental cost-effectiveness ratios (ICERs) will be calculated to express the cost per QALY gained. To project long-term cost-effectiveness over a 20-year horizon, a Markov model will simulate patient transitions between defined health states, including pain relief, persistent pain, functional disability, re-treatment, and death, using annual cycles. Transition probabilities will be derived from trial data and supplemented by published literature on chronic pain progression. All future costs and QALYs will be discounted at an annual rate of 3%, and model calibration will be based on Chinese epidemiological and cost data [69]. To assess uncertainty, probabilistic sensitivity analysis (PSA) with 1,000 Monte Carlo simulations will be performed, with results presented as cost-effectiveness acceptability curves (CEACs). Additionally, one-way sensitivity analyses will be conducted to identify key cost-effectiveness drivers, and findings will be illustrated using Tornado diagrams. This comprehensive economic evaluation will provide essential evidence on the value for money of each treatment strategy, thereby informing resource allocation and clinical adoption in the context of China’s healthcare system.

2.16. Consent for publication

Written informed consent for publication of their clinical details and/or clinical images will be obtained from the participants. A copy of the consent form is available for review at any stage if required. Participation in this study is strictly confidential. Any information that is published will not reveal the identity of the participants.