The influence of perioperative interventions targeting psychological distress on clinical outcome after total knee arthroplasty

Our aim was to assess the effect of perioperative interventions targeting psychological distress on clinical outcome after total knee arthroplasty (TKA). We searched studies on the effect of perioperative interventions focused on psychological distress used in conjunction with TKA on pain, function, and quality of life (QoL) on PubMed, Embase.com, PsycINFO/OVID, CENTRAL, the Cochrane Database of Systematic Reviews, Scopus, and Web of Science. We included 40 studies (22 RCTs, ten cohort studies, and eight quasi-experimental studies) with a total of 3846 patients. We graded the quality of evidence as low for pain and function and as moderate for QoL. Patients receiving music, education, cognitive behavioural therapy, guided imagery, pain coping skills training, Reiki, occupational therapy with self-monitoring, and biofeedback-assisted progressive muscles relaxing training had lower pain scores or declined opioid prescriptions after TKA. Pain coping skills training, audio recording-guided imagery scripts, video promoting self-confidence, psychological therapies by video, Reiki, music, occupational therapy with self-monitoring, education, and psychotherapy improved postoperative functional outcome. Education through an app improved QoL after TKA. The studies in our systematic review show that perioperative interventions targeting psychological distress for patients receiving TKA seem to have a positive effect on postoperative pain, function, and QoL. RCTs with strict methodological safeguards are still needed to determine if perioperative interventions focused on psychological distress should be used in conjunction with TKA. These studies should also assess which type of intervention will be most effective in improving patient-reported outcome measures and declining opioid prescriptions. Electronic supplementary material The online version of this article (10.1007/s00296-020-04644-y) contains supplementary material, which is available to authorized users.


Introduction
Total knee arthroplasty (TKA) is the treatment of choice for medically operable patients with end-stage osteoarthritis (OA) of the knee joint if non-surgical therapies fail to obtain adequate pain relief and functional improvement [1]. TKA proved to be a cost-effective procedure with excellent postoperative implant-related outcomes, such as radiographic appearance and implant features [2]. Nevertheless, a significant number of patients report pain (8.0-26.5%) on long-term follow-up after TKA [3] and as many as 11-19% of the patients are not satisfied with their procedure [4,5]. Persistent pain after TKA is commonly treated with opioids after surgery [6]. Currently, increasing misuse and addiction to opioids are a rapidly evolving public health issue [7]. Improving pain scores after surgery by understanding factors influencing postoperative pain may help prevent further expansion of this opioid crisis [7].

3
Unfavourable outcome after TKA is related to age, gender, level of education, pre-operative function and pain [8], comorbidities [9], social support [9], Body Mass Index (BMI) [10], and surgical factors [11][12][13]. Preoperative psychological factors such as mental health status, symptoms of anxiety and depression, and poor coping skills have also been examined [13][14][15]. Systematic reviews [16][17][18] and meta-analyses [19,20] on this subject reported that psychological distress might affect the postoperative outcome (pain and function) after TKA. Perioperative interventions targeting these psychological factors may improve clinical outcome after surgery. Previous studies have examined the effect of interventions influencing psychological factors to improve postoperative clinical outcome after TKA [21][22][23][24]. We found three previous systematic reviews on psychological interventions in conjunction to orthopaedic surgeries [25][26][27]. The systematic review of Bay et al. [25] did not support the effectiveness of psychological interventions in improving patient-reported joint outcomes after TKA as the interventions explored by studies were found to be ineffective at the latest follow-up. The results of Szeverenyi et al. [26] and Tong et al. [27] indicated that psychological interventions might improve postoperative outcome of orthopaedic surgery. These previous reviews included several types of orthopaedic procedures (among which TKA, total hip arthroplasty (THA) and spinal procedures) and did not focus on TKA. Besides, the most up-to-date search was performed in January 2018 [27].
To our knowledge, focused systematic reviews of studies on TKA patients with wide search and inclusion criteria investigating the effect of interventions targeting psychological distress on patient-reported outcome measures pain, function and/or quality of life (QoL) after surgery have not yet been reported. The aim of our systematic review was to assess the effect of perioperative interventions focused on psychological distress on pain, function and QoL after primary TKA for OA of the knee.

Search strategy and study selection
We registered our review protocol at PROSPERO international prospective register of systematic reviews (https :// www.crd.york.ac.uk/PROSP ERO/) with reference number CRD42016052466 (https ://www.crd.york.ac.uk/prosp ero/ displ ay_recor d.php?ID=CRD42 01605 2466). We performed this systematic review according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) statement criteria [28].
We performed the literature search according to the guidance of Gasparyan et al. [29]. A professional medical librarian (CdH) identified therapeutic studies (published articles and abstracts of major conferences) exploring the influence of any type of perioperative (before TKA, during surgery, or during postoperative rehabilitation) interventions targeting psychological distress on postoperative outcome (pain, function, and/or QoL) after TKA by searching PubMed, Embase. com, PsycINFO/OVID, CENTRAL, the Cochrane Database of Systematic Reviews, Scopus and Web of Science from inception up to May 26, 2020.
The following terms, including synonyms and closely related words, were used as index terms or free-text words: 'total knee arthroplasty' and 'psychological intervention'. Full search strategies for all the databases are available in Supplementary Appendix 1. Duplicate articles were excluded.
Selection of articles was limited to adults > 18 years who had undergone a primary total knee replacement for osteoarthritis of the knee. We included different study designs (RCTs, cohorts, quasi-experimental studies) investigating the effect of any intervention targeting psychological distress on postoperative pain, function and/or QoL. Minimum duration of follow-up was not an inclusion criterion with the aim to create a complete overview of all studies that have investigated the effect of perioperative interventions focused on psychological distress on pain, function and/or QoL. Perioperative interventions influencing psychological factors of patients had to be clearly defined. Full-text availability was required. There were no restrictions with respect to language, age, or publication source of the paper.
Exclusion criteria were studies not meeting domain, determinant, or outcome, case reports, descriptive studies (in which there was no control group), non-primary literature studies (letter to the editor, reviews, thesis, expert opinions) and articles with no separated results of patients after TKA and total hip arthroplasty (THA) or other types of surgery if various surgical procedures were analysed.

Main outcome variables
Two authors (JS & GO) independently screened articles for title and abstract and thereafter full text if the abstract potentially met the inclusion criteria. Subsequently, the authors (JS & GO) individually extracted information regarding study design, baseline patient characteristics, baseline clinical findings, follow-up, number of patients initially included in the study, the number of patients available for follow-up and data regarding the primary outcomes of the systematic review. When there was disagreement with respect to data extraction, a third author (AH or RP) could make the final decision.

Quality assessment
We assessed the risk of bias of the included studies using Cochrane Collaboration's tool for assessing the risk of bias [30]. Using this tool, two authors (JS & GO) independently scored six types of bias (selection bias, performance bias, detection bias, attrition bias, reporting bias, and other types of bias) as low, high, or unclear on potential risk of bias [30].
We used the Grading of Recommendation, Assessment, Development, and Evaluation (GRADE) approach to qualify the overall level of evidence of outcome measures pain, function and/or QoL (https ://www.grade worki nggro up.org/). Using the GRADEpro software (McMaster University, 2015, available from www.grade pro.org), we graded the quality of evidence as high, moderate, low, or very low [31].

Data analysis
We arranged the studies according to the type of perioperative intervention (music, education, psychotherapy, and remaining) and collected data of the effect of perioperative interventions targeting psychological distress on postoperative clinical outcome measures pain, function, and QoL. Initially, our intention was to pool data to perform a meta-analysis.

Results
The search strategy and article selection of articles published from 1964 to 26 May 2020 are shown in the flowchart (Fig. 1). Out of 7835 articles remaining after deduplication, we included 40 studies of which 22 RCTs (one randomised controlled pilot study), 10 cohort studies, and 8 quasi-experimental studies with a total number of 3846 patients.

Interventions
A description of the interventions in the experimental and the control groups and the time at which the interventions were applied are presented in Table 1.

Music
Nine studies examined the effect of perioperative listening to music on postoperative outcome. Eight of these studies [32][33][34][36][37][38][39][40] assessed the effect of music on pain and three [35,36,40] on function. Music was offered at different time points and different types of music were provided.

Education
The effect of education on postoperative outcome was investigated in fifteen studies in which the time of education varied from 12 weeks before surgery to 3 months after surgery (Table 1).

Psychotherapy
Psychological therapies provided with direct support from a professional were examined by eight studies. The patients in the RCTs of Jacobson et al. [61] and Russo et al. [64], who also received psychological therapy, received their psychological intervention by audio recordings, or watching a video instead of direct contact with a health care professional.

Other/remaining interventions
Four remaining interventions (Reiki, biofeedback relaxing training and enhanced reality analgesia, self-monitoring using a diary), applied to six studies, could not be allocated to the music, educational, or psychological therapy intervention groups and were, therefore, classified as remaining interventions (Table 1).

Outcomes
Outcome measures pain, function, and/or QoL were assessed in 22 RCTs (one randomised controlled pilot study), 10 cohort studies, and 8 quasi-experimental studies. Mean age of the patients ranged from 61.7 to 74.1 years and duration of follow-up ranged between 60 min and 2 years. Due to the heterogeneity of the type of studies, interventions, outcome measures and follow-up there was no possibility to pool data to perform a meta-analysis.

Pain
34 studies examined the influence of a perioperative intervention targeting psychological distress on clinical outcome pain after the TKA. Many different scoring systems were used to score postoperative pain and eight studies assessed pain medication use as an outcome measure for pain ( Table 2).
As shown in Table 2, patients in the intervention groups had significant better postoperative pain scores or declined prescriptions of opioids in 20 studies. Therapies applied in these studies were music during surgery [40] or after surgery [33,36,38,39], education [41,49,[51][52][53]55], cognitive behavioural therapy [57,58], guided imagery [61], pain coping skills training [62], Reiki therapy [66,70], occupational therapy in combination with self-monitoring using a diary [68], weight-bearing biofeedback training [67] and biofeedback-assisted progressive muscle relaxing training [71]. The remaining 14 studies did not show a significant effect on any of the pain-related outcome measures or pain medication use at the latest follow-up when using a perioperative intervention focused on psychological distress in conjunction to TKA.

Function
A total of 29 studies examined the effect of an intervention targeting psychological distress on function after the TKA ( Table 3).
As shown in Table 3, function was significantly improved by perioperative interventions in 18 studies. Pain coping skills training [62], audiorecording guided imagery scripts [61], video promoting self-confidence and psychological support [64], music [35,36], occupational therapy in combination with self-monitoring using a diary [68], various types of education [41,43,45,47,[51][52][53], weight-bearing biofeedback training [67], and psychological therapies (behavioural change intervention [60] and cognitive behavioural therapy [57][58][59]) positively affected any, but not all, of the functional outcome measures after TKA. In the most recent study by Riddle et al. [63], patients receiving pain coping skills training did not have significantly better scores on WOMAC function and the short physical performance battery. Other types of education [42,44,[48][49][50]55], music during physiotherapy [38], enhanced reality analgesia [69], cognitive behavioural therapy delivered by physiotherapists [56], and psychological support from a professional psychologist [23] did also not affect any of the functional outcome measures after TKA.   Preoperative education about care pathway, knee surgery, pain management, expected discharge goals and in-patient and out-patient arthroplasty rehabilitation by an educational nurse and a booklet ** Preadmission preoperative teaching with an instruction booklet during a preoperative outpatient clinic visit. Upon admission to the hospital, they were presented with an educational videotape *** A booklet containing symptom management after TKA, an individual teaching session, and a follow-up support call by the principal investigator **** 25-Min sessions of nurse-patient interaction and discussion regarding specific exercises and physical activity, self-monitoring, goal setting, family support and encouragement, and information prompting $ Guided imagery is a widely used mind-body intervention by the generation of self-or practitioner-guided positive sensory and affective mental images to promote health changes in the body, reducing anxiety and stress, and evoking psychological and physiologic relaxation [61] # Intervention addressed to the recovery of physical function, the concerns during the recovery period and strategies for coping with pain after the operation delivered by trained therapistŝ The video was established to produce positive and therapeutic insight, according to the Videoinsight Methods principles [65] 1 3       Instead of mean and SD:*median (range), **median and mean rank, ***mean and standard error, ^mean rank only, ^^median only, ^^^mean estimate with the 95% CI in parentheses, *^median (interquartile range) instead of mean and SD 1 3     (27) 66 (10) Sit to stand 12 (11,14)

QoL
Two recent studies [49,53] examined the effect a perioperative intervention on QoL (Table 4). Patients receiving postoperative day-to-day education through an app seemed to report significantly better QoL compared to patients who received usual care [53]. Additional psychoeducation did not significantly improve QoL [49]. Figure 2 shows our risk of bias assessment of the included studies. Figure 3 represents our judgement about each risk of bias item presented as percentages across all studies. The most prevalent shortcomings regarding the risk of bias were inadequate blinding participants and/or personnel during the study (performance bias) and "other types of bias". Bias due to inadequate generation of a randomisation sequence or inadequate allocation concealment prior to assignment (selection bias) also caused high scores on the risk of bias (Fig. 3).

Quality assessment
The overall level of evidence of the studies using the GRADE approach was qualified as low for pain and for function and as moderate for QoL. Serious uncertainty in the assessment of the risk of bias, inconsistency, and indirectness were the main reasons for downgrading the overall level of evidence (Table 5).

3
This is a methodologically well-conducted systematic review for which a professional medical librarian (CdH) has developed the search strategy to conduct a comprehensive search in several databases to identify eligible studies. Two authors (JS & GO) performed the screening, data extraction, risk of bias assessment, and overall level of evidence grading independently. We have created a complete overview of all studies by minimizing our exclusion criteria regarding study design, minimum follow-up, and language. Studies without significant results on the effect of an intervention are often refused for publication. Due to the heterogeneity of the outcome measures of the included studies, it was not possible to conduct a funnel plot to assess this type of bias (publication bias) in our systematic review. However, we included multiple studies [32-34, 38, 39, 42, 46, 55, 56, 68] with small sample sizes (smaller than 30 patients) with no significant results on both outcome measures pain and function. Therefore we assume the risk of publication bias to be low. Unfortunately, drawing meaningful conclusions from the included studies was hampered. First of all, there was a substantial heterogeneity with respect to study design, analysis, domain, interventions, and outcome measures, which precluded pooling for a meta-analysis. Second, according to the GRADE approach, we have graded the quality of evidence as low for outcome measures pain and function. Therefore, the true effect of the interventions targeting psychological distress on postoperative pain and function may be different from our estimate of the effect.
The previous systematic reviews of Szeverenyi et al. [26] and Tong et al. [27] concluded that psychological interventions seem to reduce postoperative side effects and anxiety and to improve recovery and mental components of quality of life after orthopaedic surgeries. However, Szeverenyi et al. [Sweverenyi] did not clarify the type of orthopaedic  procedures (only joint replacement or no joint replacement) and Tong et al. [27] included several orthopaedic procedures (THA, TKA, and spinal procedures) of which only two studies [61,63] represented separated data of patients undergoing TKA. The findings of our review do not support the earlier systematic review of Bay et al. [25], in which most interventions explored by the included studies were found to be ineffective on patient-reported outcome after THA and TKA. Only three studies with patients receiving TKA were included by Bay et al. [25]. Compared to that review, we included fifteen additional RCTs [33, 34, 37, 38, 41, 44, 45, 49, 53, 54, 56-58, 58, 63]. Second, due to the current lack of RCTs on one specific type of intervention focused on psychological distress (for example only pain coping skills training) applied to patients undergoing TKA, we have decided to also include a wider range of study designs to create a complete overview of the perioperative interventions focused on psychological distress that have been used to decrease pain and improve function and/or QoL after surgery. Besides, ten studies [32,34,37,39,48,54,55,66,70,71] in our systematic review evaluated the degree of postoperative pain not only by measuring pain scores, but also by assessing postoperative prescription of opioids or other types of pain medication. Investigating alternative nonpharmacologic methods to reduce postoperative pain and opioid use may help prevent further expansion of opioid misuse and addiction, which is currently a rapidly evolving public health crisis [7].
To the best of our knowledge, except for the mentioned systematic reviews [25,26], no other systematic reviews or meta-analysis with comparable objectives have been published. Therefore, this is the first systematic review with wide search and inclusion criteria focused on TKA patients investigating the effect of interventions focused on psychological distress on patient-reported outcome measures pain, function, and QoL after surgery. Unfortunately, our review also highlighted the limitations of current literature on this subject. To avoid heterogeneity of outcome measures between studies, we would discourage the use of different questionnaires to assess patient-reported outcome measures (PROMs) in orthopaedic research. The reliability and reproducibility of the EuroQOL Five-Dimensional Questionnaire (EQ-5D) and the responsiveness of the Patient-Reported Outcomes Measurement Information System (PROMIS) Global Health survey have been well validated for patients undergoing TKA [72]. We would, therefore, recommend the use of the EQ-5D and PROMIS to allow tracking and evaluation of the effectiveness of perioperative interventions for psychological distress in conjunction with TKA in the following studies [72].

Conclusions
The studies included in our systematic review show the positive effect of multiple perioperative interventions targeting psychological distress for patients receiving TKA to improve postoperative pain (or to decline prescriptions of opioids), function, and QoL. RCTs with strict methodological safeguards (such as long-term follow-up, large number of patients participating in the study, low risk of bias) prospectively comparing outcome for patients with and without perioperative support are still needed to determine if perioperative interventions targeting psychological distress should be used in conjunction with primary TKA for OA of the knee. These studies should also assess which type of intervention will be most effective in improving patientreported outcome measures and declining opioid prescriptions in the future.