Factors associated with persistent postsurgical pain after total knee or hip joint replacement: a systematic review and meta-analysis

1. Introduction

Persistent postsurgical pain (PPSP) is a common sequel of surgical procedures. The prevalence varies for different surgical procedures and in different studies, ranging from a low of 3% to a high of 81%,90 with a reported incidence of about 20% after bone and joint surgeries.39 PPSP substantially impairs quality of life, has negative economic consequences,35 and is a potentially preventable condition.106 Patients who undergo elective procedures such as total hip or knee replacement, typically expect alleviation of the pain associated with the damaged joint. Although these procedures have a high likelihood of success in relieving pain and restoring function, PPSP is a major factor that negatively affects patient satisfaction with joint replacement surgery outcomes.72 Therefore, understanding factors predisposing patients to an increased risk of persistent pain is important in attempts to prevent PPSP or minimize its occurrence. The identification of patients at high risk for PPSP will allow allocating perioperative resources for PPSP risk mitigation and may signal a need for enhanced postoperative follow-up and early referral to multidisciplinary evaluation. Several studies been have published on perioperative risk factors associated with PPSP,120 but it is yet unclear which factors show a consistent association with PPSP.47,67

Several systematic reviews have been published on psychosocial factors in general43 or anxiety and catastrophizing specifically107 as risk factors for PPSP, and individual studies have reported on factors associated with orthopedic surgeries such as total hip replacement (THR)27 and total knee replacement (TKR).109 An up-to-date synthesis of the existing literature can help improve perioperative planning and care, as well as guide future study designs aimed at better understanding and possible prevention of PPSP.

The objective of the current study was to perform a systematic review and meta-analysis on the reported factors associated with PPSP after total hip joint and knee joint replacement.

2. Methods

The protocol for this systematic review has been preregistered on PROSPERO (CRD42020152146).

2.1. Inclusion and exclusion criteria for studies

The studies were included if they reported PPSP occurrence after a hip joint or knee joint replacement. We included both prospective and retrospective studies that addressed PPSP as pain at the surgical or related site, present at least 2 months postoperatively. The minimum 2-month cutoff was selected as several definitions of PPSP in the past decade have been proposed,60 with most suggesting pain duration for more than 2 or 3 months after surgery. Exclusion criteria included (1) fewer than 10 participants per arm, (2) follow-up shorter than 2 months after surgery, (3) participants younger than 18 years, (4) studies on patients with established PPSP that did not report the occurrence of persistent pain after surgery, (5) studies not reporting incidence or prevalence rate of PPSP in the cohort, and (6) abstracts.

A systematic literature search was performed in the following databases: Ovid MEDLINE, Embase, Scopus, Cochrane Central Register of Control Trials, Cumulative Index to Nursing and Allied Health Literature and PsychINFO. The full keyword search strategy can be found in Appendix 1 (available at https://links.lww.com/PR9/A184). The search was initially performed on September 17, 2019, with an update on October 14, 2020 (Appendix 1A, https://links.lww.com/PR9/A184), and subsequently expanded on July 9, 2021 (Appendix 1B, https://links.lww.com/PR9/A184), and updated on February 6, 2022 (Appendix 1C, https://links.lww.com/PR9/A184). The duplicates in the resulting abstracts were removed using EndNote.12

The abstracts of the resulting list of articles were screened by 2 authors for eligibility. If the studies were deemed eligible based on abstracts, then the full texts of those articles were retrieved and reviewed independently. The data from eligible articles were extracted into REDCap (Research Electronic Data Capture) and Excel spreadsheets. The data were extracted by 2 authors (A.G. and S.B./S.S.) and compared for inconsistencies. Wherever disagreements arose, a third author (S.H.) was consulted to reach the final decision. Risk of bias (ROB) assessments and data extraction were performed by the 2 reviewers (A.G. and S.B./S.S.) and compared for accuracy.

2.2. Risk of bias assessment

Most validated tools for assessing the ROB in systematic reviews42 refer to assessing the methodology of randomized controlled trials. Because this study aimed to analyze factors associated with PPSP (from both prospective and retrospective studies), an alternative method for the ROB systematic assessment was applied based on the methodology used in other systematic reviews focusing on the analysis of predicting factors.53,56,116 Potential sources of bias were evaluated and addressed whether (1) the sample adequately represented the population of interest, (2) the data represented the sample, and (3) the outcome of interest was adequately measured to limit potential bias (Appendix 2, available at https://links.lww.com/PR9/A184). Other potential serious sources of bias were documented—eg, selective reporting of outcomes, use of nonvalidated scales, or lack or inappropriateness of sample size calculation.

2.3. Protocol for data extraction

To address the different study designs, we developed guiding principles for data extraction across all studies (Appendix 3, available at https://links.lww.com/PR9/A184). In brief, all studies which addressed PPSP at or more than 2 months were included, and the follow-up timeframe at or closest to 6 months was obtained in case of multiple follow-ups. Although the cutoff for postsurgical duration of pain has most commonly been set at 2 or 3 months, some patients have a longer rehabilitation period after total knee or hip joint replacement surgeries.38,55 Thus, to remove the confounder, we decided to set the primary outcome timeframe at around 6 months after surgery in studies with multiple follow-ups. If PPSP was categorized based on a severity scale (numerical rating scale [NRS] or Likert scale), the data were dichotomized to a “no pain to mild pain” group vs a “moderate-severe pain” group, to allow 2 × 2 table formation. In addition, to allow systematic analysis of data, some factors were combined, eg, epidural and spinal anesthesia were merged under neuraxial anesthesia.

2.4. Data synthesis and analysis

The data analyses were performed using RStudio version 1.4.1103.93 For binary factors, 2 × 2 tables were created, and risk ratios were calculated to be used in generating forest plots. For continuous variables (eg, age and body mass index [BMI]), weighted mean differences were used. Only raw univariate data were used for generating forest plots. Values for weighted mean differences were presented as means with standard deviation (SD). Risk Ratio (RR) or odds ratio was presented as risk or odds, with 95% confidence interval (CI). Meta-regression of multivariable data was not possible because each study accounted for different risk factors. Data on risk factors from multivariable analyses were aggregated in a tabulated form to help make qualitative inferences about common contributing factors.

3. Results

The systematic literature search resulted in a total of 10,851 potential articles. After removing 5,403 duplicate records, a total of 5,448 unique citations were added to the project library. All abstracts of identified studies were reviewed for inclusion criteria, resulting in 231 full-text articles that were considered for full review (see Appendix 1 for full details about literature search strategy and results, available at https://links.lww.com/PR9/A184). These articles were independently evaluated for inclusion by 2 reviewers (AG and SB/SS), and 81 were included in the final analysis (Fig. 1).

F1Figure 1.:

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow chart.

The studies analyzed included total hip and/or knee joint replacement surgeries. The overall study characteristics are outlined in Table 1. The minimum cutoff for determining PPSP in the included studies was 3 months after surgery.98 The total number of patients across all total hip and/or knee joint replacement surgery studies (n = 81) was 171,354 with a mean (SD) sample size of 2,115 (14,226), 28 (34.6%) studies involving less than 100 participants. The mean PPSP rate was 16.4%. Studies investigating PPSP after both total hip and knee joint replacement surgery (n = 11) included 10,399 participants and reported a mean PPSP incidence of 26.7%. Studies after total hip joint replacement surgery alone (n = 16) included 9,086 participants and reported a mean PPSP incidence of 16.5%, and those after total knee joint replacement surgery alone (n = 54) included 151,869 participants and reported a mean PPSP incidence of 15.6%. Overall, 69 of the studies were prospective and 12 were retrospective. A total of 36 studies have provided univariate data to be included in meta-analysis, and 64 studies have reported risk factors in multivariable analyses, from which data were used for qualitative summary (Fig. 1). The studies overall had medium to low ROB (Table 2). The main category to have significant ROB was a “follow-up rate >75% at the first follow-up” (with 14 studies (17%) with high ROB), incomplete characterization of study dropouts (with 65 studies [80%] having unclear or high ROB).Thirty five of the 81 included studies (43%) had high or unclear bias in one of the categories of pain outcome measurement—either in PPSP definition, its method of assessment or classification, or reporting—and 62 of 81 (76.5%) studies reported sources of funding. Although there is not a good way to assess publication bias related to risk prognostic studies,105 funnel plots to assess asymmetry among most commonly reported risk factors did not suggest important publication bias (Appendix 4, available at https://links.lww.com/PR9/A184).

Table 1 - Main characteristics of included studies (n = 81). SN Author-date Surgery Type of study Total number of participants Occurrence of PPSP percentage (number) Follow-up duration* 1 Albayrak 2016 TKR Prospective 274 66.7% (183) 22 mo 2 Aso 2021 TKR Prospective 194 10.3% (20) 6 mo 3 Attal 2014 TKR Prospective 81 39.5% (32) 6 mo 4 Aveline 2014 TKR Prospective 75 17.4% (13) 12 mo 5 Baker 2007 TKR Cross-sectional mail/phone survey 8010 19.8% (1583) 24 mo 6 Bischoff-Ferrari 2004 THR Retrospective 897 42% (377) 36 mo 7 Bossmann 2017 TKR Prospective 47 20% (9) 6 mo 8 Bourne 1994 THR Prospective 94 27.6% (26) 60 mo 9 Brander 2003 TKR Prospective 116 18.4% (21) 12 mo 10 Briggs 1995 TKR Prospective 65 6% (4) 24 mo 11 Bugada 2017 TKR Prospective 614 21.6% (133) 6 mo 12 Buvanendran 2010 TKR Randomized controlled study 228 5.2% (placebo arm: 6 of 115) 6 mo 13 Buvanendran 2019 TKR Prospective 245 14% (34) 6 mo 14 Buvanendran 2020 TKR Prospective 50 34% (17) 3 mo 15 Chen 2021 TKR Prospective 220 13.6% (30) 6 mo 16 Clarke 2009 THR Randomized controlled study 126 (G1: placebo/placebo 26.3% (10); G2: GPN/placebo 31.6% (12); G3: placebo/GPN 23.7% (9)). 6 mo 17 Clarke 2010 THR Prospective 82 37.5% (31) 6 mo 18 Dong 2019 TKR Randomized controlled study 122 36.06% (44) 3 mo 19 Dowsey 2015 TKR Retrospective 689 22% (151) 12 mo–60 mo 20 Dumenci 2019 TKR Randomized clinical trial 384 18% (69) 12 mo 21 Dürsteler 2021 TKR Prospective 146 31.5% (46) 3 mo 22 Erlenwein 2017 THR Prospective 104 17.3% (18) 6 mo 23 Fletcher 2015 THR and TKR Prospective 145 27.6% (40) 12 mo 24 Forsythe 2008 TKR Prospective 48 77.1% (37) 24 mo 25 George 2021 THR and TKR Retrospective 2411 (1146 THR, 1265 TKR) THR 33.6% (385) 6 mo TKR 38.9% (492) 26 Giordano 2020 TKR Prospective 136 16.2% (22) 12 mo 27 Grosu 2016 TKR Prospective 114 10% (11) 6 mo 28 Guimaraes-Pereira 2016 THR and TKR Prospective 19 50% (9) 3 mo 29 Gungor 2019 TKR Retrospective 578 31.1% (180) 3 mo 30 Kornilov 2017 TKR Prospective 79 22.8% (18) 12 mo 31 Kuchálik 2017 THR Retrospective 72 Group ITM 6% (2) 6 mo Group LIA 0% (0) 32 Kuchálik 2017 THR Randomized controlled study 56 (group LIA (n = 29), group FNB (n = 27)) Group LIA 0% (0) 6 mo Group FNB 0% (0) 33 Kurien 2018 TKR Prospective 50 30.4% (14) 6 mo 34 Lavand'homme 2014 TKR Prospective 112 11% (12) 3 mo 35 Lee 2019 TKR Randomized controlled study 24 25% (6) 6 mo 36 Lindberg 2021 TKR Prospective 202 30.2% (61) 3 mo 37 Liu 2012 THR and TKR Cross-sectional mail/phone survey 1030 45.9% (473) (THR:38% (178) 12 mo TKR:62% (295)) 38 Lu 2021 THR Prospective 736 27.2% (200) 6 mo 39 Martinez 2007 TKR Prospective 20 20% (4) 4 mo 40 Masselin-Dubois 2013 TKR Prospective 89 50.6% (45) 3 mo 41 Nazal 2019 THR Retrospective 10 2% (2) 24 mo 42 Neuprez 2020 THR and TKR Prospective 626 (346 THR, 280 TKR) THR 8.99% (31) 60 mo TKR 3.23% (9) 43 Nikolajsen 2006 THR Cross-sectional mail/phone survey 1048 28.1% (294) 12–18 mo 44 Noiseux 2014 TKR Prospective 215 4.6% (10) 6 mo 45 Oh 2019 TKR Retrospective 924 16% (148) 12 mo 46 Pagé 2015 TKR Prospective 108 8.3% (9) 12 mo 47 Pagé 2016 THR Prospective 111 21.6% (24) 6 mo 48 Peng 2014 TKR Randomized controlled study 212 (group CFNB [n = 109] group PCIA [n = 103]) Group CFNB 33% (36) 6 mo Group PCIA 52.4% (54) 49 Pereira 2016 THR and TKR Prospective 43 (TKR 22, THR 21) TKR 68.2% (15) 6 mo THR 23.8% (5) 50 Petersen 2015 TKR Cross-sectional mail/phone survey 305 (215 primary surgery patients, 90 revision surgery patients) 19% (41) primary surgery patients 36 mo 47% (42) revision surgery patients 51 Petersen 2015 TKR Prospective 78 22% (17) 12 mo 52 Petersen 2017 TKR Prospective 130 19% (25) 12 mo 53 Petersen 2020 TKR Prospective 26 35% (9) 12 mo 54 Pinedo-Villanueva 2018 TKR Retrospective 128145 15% (19222) 6 mo 55 Pinto 2013 THR and TKR Prospective 92 (THR 48, TKR 44) 41.3% (38) 4–6 mo THR (13) TKR (25) 56 Puolakka 2010 TKR Cross-sectional mail/phone survey 562 35% (197) 4–22 mo 57 Rao 2020 TKR Randomized controlled study 40 47.5% (19) 3–6 mo 58 Remérand 2009 THR Randomized controlled study 142 (70 placebo, 72 ketamine)  14.8% (21) 6 mo 15 placebo, 6 ketamine 59 Rice 2018 TKR Prospective 286 21% (60) 6 mo 60 Sakellariou 2015 TKR Cross-sectional mail/phone survey 272 39.34% (107) 12–16 mo 61 Sanders 2009 TKR Randomized controlled study 56 (baclofen treatment group n = 27, control group n = 29) 27 (baclofen treatment group n = 8, control group n = 19) 3 mo 62 Sayers 2016 THR and TKR Randomized controlled study 560 (283 THR, 277 TKR) 5% (14) THR 12 mo 12% (33) TKR 63 Sen 2020 TKR Retrospective 182 48.4% (88) 6.84 ± 4.10 mo 64 Sideris 2021 TKR Prospective 162 9.3% (15) 6 mo 65 Singh 2010 THR Cross-sectional mail/phone survey 5390 8.1% (435) 24 mo 66 Skrejborg 2019 TKR Cross-sectional mail/phone survey 604 18% (107) 60 mo 67 Sugiyama 2018 TKR Retrospective 298 33% (97) 6 mo 68 Thomazeau 2016 TKR Prospective 104 28.8% (74) 6 mo 69 Vaegter 2017 TKR Prospective 14 28.6% (4) 6 mo 70 Valdes 2012 THR and TKR Prospective 1788 (THR 928, TKR 860) THR 17.0% (158) 38 mo TKR 25.3% (217) 71 Vila 2020 TKR Prospective 112 41.96% (47) 6 mo 72 Von Dincklage 2017 THR Prospective 105 13.3% (14) 6 mo 73 Wang 2014 THR Randomized controlled study 51 (spinal saline 28, spinal ketorolac 23) 15% (8) 6 mo 74 W-Dahl 2014 TKR Retrospective 2123 7.8% (165) 12 mo 75 Wylde 2009 THR and TKR Cross-sectional mail/phone survey 2391 (1112 THR, 613 TKR) THR 13% (144) 60–96 mo TKR 26% (159) 76 Wylde 2011 THR and TKR Cross-sectional mail/phone survey 1294 (662 THR, 632 TKR) THR 27% (179) 24–48 mo TKR 44% (278) 77 Wylde 2013 TKR Prospective 51 29% (15) 13 mo 78 Wylde 2020 TKR Retrospective 3058 11.87% (363) 3 mo 79 Yang 2020 TKR Randomized controlled study 96 27.08% (26) 3 mo 80 Yao 2019 TKR Retrospective 694 9.94% (69) 6 mo 81 Zachodnik 2021 THR Prospective 62 24.19% (15) 12 mo

*Follow-up time points used for data analysis.

CFNB, continuous femoral nerve block, PCIA, patient-controlled intravenous analgesia; GPN, gabapentin; TKR, total knee replacement; THR, total hip replacement; ITM, intrathecal morphine; LIA, local infiltration analgesia; FNB, femoral nerve block.


Table 2 - Risk of bias assessment of the included studies.
3.1. Results of meta-analysis 3.1.1. Univariate comparisons

The demographic factors with data from multiple studies that allowed testing their association with an increased risk of PPSP were age, the proportion of women in the total study population, and BMI, whereas clinical factors included duration of surgery (in minutes), preoperative factors such as presence of anxiety and depression, knee or hip joint pain, or pain elsewhere in the body.

Age (Fig. 2) and sex (Fig. 3) were not associated with an increased risk of PPSP after TKR or THR (Appendix 5 and Appendix 6 respectively, available at https://links.lww.com/PR9/A184) studies when analyzed separately. However, younger age (SMD −0.18 years [95% CI −0.30 to −0.06]) (Fig. 4) and female sex (RR 1.13, 95% CI 1.02–1.24) (Appendix 7, available at https://links.lww.com/PR9/A184) were associated with an increased risk in a small subset of studies reporting on the combined knee and hip replacement surgeries (n = 2).

F2Figure 2.:

Forest plot for the association between age and PPSP in total knee joint replacement surgeries analyzed together. *Higher/positive SMD represents the difference in age (in years) between patients who developed PPSP vs those who did not, the horizontal box and whisker plots are the group SMDs ±95% confidence intervals, and the blue diamond is the combined SMD of groups.

F3Figure 3.:

Forest plot for the association between the proportion of women in the study population and PPSP in total knee joint replacement surgeries. *Risk ratio (RR) >1 represents higher, RR <1 represents lesser, while RR = 1 represents a similar occurrence of PPSP in women compared with the study population, the horizontal box and whisker plots are the group RRs ±95% confidence intervals, and the blue diamond is the combined RR of groups.

F4Figure 4.:

Forest plot for the association between age and PPSP in total knee and hip joint replacement surgeries analyzed together. *Higher/positive SMD represents the difference in age (in years) between patients who developed PPSP vs those who did not, the horizontal box and whisker plots are the group SMDs ±95% confidence intervals, and the blue diamond is the combined SMD of groups.

Body mass index was not associated with a risk of PPSP in TKR studies (Fig. 5), but higher BMI was associated with higher PPSP occurrence in the subset of studies reporting both hip and knee replacement (SMD 0.15 kg/m2 [95% CI 0.03–0.26]) (Appendix 8, available at https://links.lww.com/PR9/A184).

F5Figure 5.:

Forest plot for the association between body mass index (BMI) in kg/m

留言 (0)

沒有登入
gif