Association of Early Postoperative Pain Trajectories With

Original Investigation | Orthopedics

Association of Early Postoperative Pain Trajectories With Longer-termPain Outcome After Primary Total Knee ArthroplastyJasvinder A. Singh, MBBS, MPH; Celeste A. Lemay, RN, MPH; Lisa Nobel, PhD; Wenyun Yang, MS; Norman Weissman, MD, MPH; Kenneth G. Saag, MD, MSc;Jeroan Allison, MD, MSc; Patricia D. Franklin, MD, MPH

Abstract

IMPORTANCE Studies to date have not comprehensively examined pain experience after total kneearthroplasty (TKA). Discrete patterns of pain in this period might be associated with pain outcomesat 6 to 12 months after TKA.

OBJECTIVES To examine patterns of individual post-TKA pain trajectories and to assess theirindependent associations with longer-term pain outcome after TKA.

DESIGN, SETTING, AND PARTICIPANTS This prospective cohort study combined data from anational US TKA cohort with ancillary pain severity data at 2 weeks and 8 weeks after the index TKAusing a numeric rating scale. All participants received primary, unilateral TKA within the Function andOutcomes Research for Comparative Effectiveness in Total Joint Replacement (FORCE-TJR) nationalnetwork of community sites in 22 states or at the lead site (University of Massachusetts MedicalSchool). Participants had a date of surgery between May 1, 2013, and December 1, 2014. The dataanalysis was performed between January 13, 2015, and July 5, 2016.

EXPOSURES Pain trajectories in the postoperative period (8 weeks).

MAIN OUTCOMES AND MEASURES Index knee pain at 6 months after TKA using the Knee Injuryand Osteoarthritis Outcome Score (KOOS) pain scale. Group-based trajectory methods examined thepresence of pain trajectories in the postoperative period (8 weeks) and assessed whethertrajectories were independently associated with longer-term pain (6 months).

RESULTS The cohort included 659 patients who underwent primary TKA with complete data at 4points (preoperative, 2 weeks, 8 weeks, and 26 weeks). Their mean (SD) age was 67.1 (8.0) years,64.5% (425 of 659) were female, the mean (SD) body mass index (calculated as weight in kilogramsdivided by height in meters squared) was 30.77 (5.66), 94.5% (613 of 649) were white, and the mean(SD) preoperative 36-Item Short Form Health Survey physical component summary and mentalcomponent summary scores were 34.1 (8.2) and 53.8 (11.4), respectively. Two pain trajectorysubgroups were identified at 8 weeks after TKA: patients who experienced fast pain relief in the first8 weeks after TKA (fast pain responders, composing 72.4% [477 of 659] of the sample) and patientswho did not (slow pain responders, composing 27.6% [182 of 659] of the sample). After adjusting forpatient factors, the pain trajectory at 8 weeks after TKA was independently associated with the meanKOOS pain score at 6 months, with a between-trajectory difference of −11.3 (95% CI, −13.9 to −8.7).

CONCLUSIONS AND RELEVANCE The trajectory among slow pain responders at 8 weeks aftersurgery was independently associated with improved but greater persistent index knee pain at 6months after TKA compared with that among fast pain responders. Early identification of patients

(continued)

Key PointsQuestion Is knee pain recovery

different between patient subgroups in

the early postoperative period after

total knee arthroplasty (TKA), and is it

associated with intermediate-term

TKA pain?

Findings In this cohort study of 659

patients who underwent primary TKA, 2

pain trajectories (fast pain response and

slow pain response) were evident at 8

weeks after the procedure. These

trajectories were independently

associated with clinically significant

differences in pain outcome at 6 months

between the 2 pain trajectories.

Meaning Early identification of patients

in the slow pain response trajectory at

8 weeks after TKA may offer an

opportunity for interventions in the

perioperative period to potentially

improve the intermediate-term and

long-term pain outcomes after TKA.

+ Supplemental content

Author affiliations and article information arelisted at the end of this article.

Open Access. This is an open access article distributed under the terms of the CC-BY License.

JAMA Network Open. 2019;2(11):e1915105. doi:10.1001/jamanetworkopen.2019.15105 (Reprinted) November 13, 2019 1/17

Downloaded From: https://jamanetwork.com/ on 05/10/2022

https://jama.jamanetwork.com/article.aspx?doi=10.1001/jamanetworkopen.2019.15105&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamanetworkopen.2019.15105

Abstract (continued)

with a trajectory of slow pain response at 8 weeks after TKA may offer an opportunity forinterventions in the perioperative period to potentially improve the long-term pain outcomesafter TKA.

JAMA Network Open. 2019;2(11):e1915105. doi:10.1001/jamanetworkopen.2019.15105

Introduction

Postoperative pain is commonly underestimated and undertreated.1-3 Undertreated postoperativepain not only leads to distress and poor patient satisfaction but also is associated with longer hospitalstay, reluctance to engage in rehabilitation exercises, poorer health-related quality of life, andincreased morbidity related to complications.4-8 Untreated postoperative pain is associated withpersistent chronic pain after cardiac surgery.6,9

Total knee arthroplasty (TKA) is one of the most common elective surgical proceduresperformed in older patients to treat pain and functional limitation owing to refractory knee arthritis10

and is associated with optimal arthritis pain relief in most but not all patients. In the United States,8% to 15% of patients who undergo TKA have residual moderate to severe index joint pain thatpersists 2 to 5 years after the procedure.11,12 A systematic review concluded that unfavorable long-term pain outcomes were seen in 10% to 34% of patients after knee arthroplasty.13 Few studies haveexamined intermediate-term postoperative pain and its association with long-term pain outcomes.In a single-center study14 of 116 patients, pain during the 1-year postoperative period generallyimproved, and 1-year pain status was associated with long-term pain outcomes. In 2 studies,researchers used preoperative and postoperative pain and function scores to assess pain outcome at5 years after TKA15 or used preoperative scores to assess pain outcome at 3 months and 12 monthsafter TKA16; these studies focused on preoperative pain and did not collect postoperative pain data(within 2 to 3 months of TKA).15,16 In addition, research to date is limited by single-center enrollment,small sample sizes with high dropout rates, and the lack of adjustment for common confounders (eg,race/ethnicity, body mass index [BMI; calculated as weight in kilograms divided by height in meterssquared], and comorbidities).

To our knowledge, studies to date have not comprehensively examined pain experience in thepost-TKA period or evaluated for the presence of discrete subgroups of individuals with differentpatterns of pain at 6 to 12 months after TKA. This factor is an important knowledge gap given (1) theco-occurrence of postoperative pain during post-TKA physical rehabilitation,17,18 (2) the treatmentof early postoperative pain with opioids19-21 and the opioid epidemic in the United States,22-25 and (3)the significant prevalence of persistent pain after TKA.11-13 If early patterns of pain can be used toidentify patients likely to experience prolonged pain, clinical treatment options could be developedand tailored to alter this outcome. The emerging tools for trajectory analysis shown to be associatedwith different care patterns and health care expenditures in patients with cancer26 offer the potentialto provide new insights into patients’ pain experience and outcomes after TKA.

Our study objective was to examine whether discrete subgroups of patients undergoing TKAhave different trajectories of pain in the period (8 weeks) after TKA in a nationally representativesample. We evaluated (1) whether the postoperative period pain trajectory is associated with thelevel of 6-month postarthroplasty pain outcome and (2) the patient characteristics associated withpostarthroplasty pain trajectories.

JAMA Network Open | Orthopedics Early Postoperative Pain Trajectories and Long-term Pain Outcome After TKA




Methods

Study Cohort and SurveyThis research was an ancillary study of the Function and Outcomes Research for ComparativeEffectiveness in Total Joint Replacement (FORCE-TJR) study,27 a nationally representativeobservational prospective cohort study that enrolled 28306 patients undergoing knee or hiparthroplasty (total or partial knee arthroplasty, total hip replacement, hip resurfacing, and knee or hiprevision arthroplasty) across the United States, including more than 130 surgeons in academic andprivate centers. The FORCE-TJR study enrolled cases from April 11, 2011, to December 31, 2016. Thestudy was approved by The University of Alabama at Birmingham and University of Massachusettsinstitutional review boards. Patient written informed consent was obtained before study enrollmentand before any study activities were conducted. This study followed the Strengthening the Reportingof Observational Studies in Epidemiology (STROBE) reporting guideline.

The FORCE-TJR cohort was based on total knee replacement surgical procedures referred by astratified sample of surgeons from across the United States; 75% of surgeons were from community-based practices. Operating room schedulers at each site referred patients scheduled for primary,unilateral TKA to the University of Massachusetts Medical School (UMMS) data coordinating centerfor enrollment each day. A centralized UMMS research coordinator reviewed participation bytelephone with all patients. Almost 95% (12 585 of 13 283) of TKA patients who returned the signedinformed consent form completed the preoperative assessment surveys, including demographics,preoperative pain and function (36-Item Short Form Health Survey [SF-36] and Knee Injury andOsteoarthritis Outcome Score [KOOS] pain scale), and medical and musculoskeletal risk factorsaccording to the FORCE-TJR protocol. Specific enrollment and measures have been describedpreviously.27-29 All FORCE-TJR participants repeated these measures at 6 months after total jointreplacement. In addition to the comprehensive preoperative and outcome data collected by FORCE-TJR, patients who enrolled in the FORCE-TJR substudy during the ancillary study period at a subsetof sites completed a 2-week and 8-week assessment of pain severity and pain treatment strategies.

All patients receiving TKA within the FORCE-TJR national network of community sites in 22states or at the lead site (UMMS) were invited to enroll in the ancillary pain study, as well as theparent study.27 Exposures were pain trajectories in the postoperative period (8 weeks). For theancillary pain study from 24 practice sites, patients who underwent primary, unilateral TKA alsocompleted a pain assessment at 2 weeks and 8 weeks after the index TKA using a numeric ratingscale (range, 0-10; higher scores indicate worse pain). Of the 12 585 FORCE-TJR study participants,2209 (17.6%) of those undergoing primary total knee replacement were enrolled from May 12, 2011,to March 4, 2015. The dates of surgery were between May 1, 2013, and December 1, 2014. The datesof analysis were January 13, 2015, to July 5, 2016. Pain study sites included all community-basedorthopedics offices covered by the University of Massachusetts institutional review board. TheFORCE-TJR patients agreed to complete baseline, 6-month, and 12-month surveys, and theFORCE-TJR study staff supported collection of these data. Ancillary study resources supported adedicated study coordinator to perform pain assessments at 2 weeks and 8 weeks for a subsampleenrolled into the pain study. The consent forms were given to the patients at a preoperative visit, andsurgeons requested each patient to participate in the registry. Next, coordinators called patientswithin 3 days to review the consent form to confirm participation. By reviewing operating roomrecords, we verified that 79.4% (13 283 of 16 721) of people with TKA invited to join this ancillarystudy of the FORCE-TJR registry participated in the study. Patients received a reminder telephone callif the 2-week or 8-week survey was not returned within 2 weeks of the target date. Patients or othermembers of the public were not involved in the development of the study protocol.

Pain Assessment ComponentsThe 2-week and 8-week pain surveys contained the validated Revised American Pain Society PatientOutcome Questionnaire (APS-POQ-R),30 which includes pain severity (0-10 numeric rating scale;




http://www.equator-network.org/reporting-guidelines/strobe/

higher scores indicate worse pain),31 consequences of pain, common adverse effects of narcotics,and nonpharmacological strategies for managing postoperative pain. The pain numeric rating scale31

is valid, reliable, and feasible and has been used in multiple musculoskeletal conditions.32-36 We usedthe preoperative pain assessment along with 2-week and 8-week post-TKA pain severity to derivepain trajectories (detailed in the Statistical Analysis subsection).

Long-term Pain Outcomes and CovariatesEarly post-TKA pain trajectories (3 assessments, as well as preoperative pain severity) were theprimary variables of interest. The main study outcome was index knee pain at 6 months after TKA asassessed with the KOOS pain scale as a continuous variable.37,38 The KOOS39 is a 42-item,participant-administered, validated, knee-specific questionnaire for osteoarthritis commonly used inclinical trials, registries, and patient follow-up and has been used in patients aged 13 to 79 years. TheKOOS is valid and reliable, and the thresholds for minimal clinically important change have beendefined.40,41 The KOOS covers 5 dimensions, including pain (9 items), knee-specific symptoms (7items), activities of daily living (ADLs) function (17 items), sport and recreation function (5 items),and knee-related quality of life (4 items). Each item is rated by the patient on a 5-point Likert-typescale ranging from no problems to extreme problems. Each of the 5 scores is calculated as the sum ofthe items included. Scores are transformed to a scale of 0 to 100, with 0 representing extreme kneeproblems and 100 representing no knee problems. A difference of 8 to 10 points represents aclinically meaningful difference for the KOOS pain scale.39,42

Covariates included factors previously shown to be associated with longer-term pain outcomeup to 6 months after TKA, including sex, age, BMI, race, medical comorbidity assessed using themodified Charlson Comorbidity Index43 (a validated measure), and quality of life assessed using theSF-36.44,45 The SF-36 physical component summary (PCS) and mental component summary (MCS)scores were calculated using standard algorithms; both are norm based, with a mean (SD) of 50 (10).Back pain was measured using the Oswestry Disability Index, with scores ranging from 0 to 50,categorized as no (0-4), mild (5-14), moderate (15-24), severe (25-34), or complete (35-50)disability.46

Statistical AnalysisWe performed all analyses using StataMP statistical software, version 13 (StataCorp LLC). We used a2-sample Wilcoxon rank sum (Mann-Whitney) test (2 sided) for continuous variables and a Pearson χ2

test (1 sided) for categorical variables, with a significance level of .05 to examine whether there wereany differences between the FORCE-TJR source cohort and the pain trajectory cohort. Descriptivestatistics explored demographic characteristics, clinical variables, and pain scores before and afterTKA. We used group-based trajectory models to assess trajectories of pain between pre-TKA and8-week TKA scores.47,48 Specifically, we developed trajectories of the probability of having moderateto severe pain from before TKA to 8 weeks after TKA. We then examined factors associated withthese trajectories and whether the trajectories could predict a pain outcome at 6 months. We usedthe trajectory program developed by Jones and Nagin.49

We assessed multicollinearity by examining the variance inflation factors for each of thevariables in the model. A variable for which variance inflation factor values are greater than 10 maymerit further investigation; the variance inflation factors were 1.08 or lower in our final linearregression model for 6-month outcome and in the logistic regression model for variables associatedwith trajectories. We examined the linearity assumption for variables that were continuous usingthe augmented partial residual plot, and no evidence of nonlinearity was found in our final linearregression model. The shape of the trajectories was forced to be linear owing to overfitting ifquadratic or cubic terms were used with the limited number of time points that were available.

In sensitivity analyses, we evaluated the extent of missing surveys at 2 weeks and 8 weeks andwhether any patterns of missingness altered trajectory assignment or outcome predictions. Missing




pain values at 2 weeks or 8 weeks were imputed based on the predicted values from multilevelmixed-effects linear regression models.

Assessing the Trajectory ModelAfter basic exploratory data analysis has been performed, trajectory analysis requires assessment of(1) the number of trajectories, (2) the shape of each trajectory, and (3) the predictors of membershipto each trajectory. We used several criteria to evaluate the group-based trajectory model to use. Weassessed a priori 2 to 6 trajectory subgroups because models with more than 6 trajectory subgroupsmay be difficult to interpret clinically. We used the Bayesian information criterion to choose theoptimal number of subgroups or trajectories.48 We also used the following criteria to assess modelquality: (1) average posterior probability for membership in each trajectory of at least 0.70 and (2)nonoverlapping 95% CIs for the 2 trajectories.48 To avoid overfitting the data, we only examinedlinear trajectories for 3 points.50 We examined all combinations of linear trajectories and differingnumbers of trajectories. Next, to create trajectory membership, all individuals were assigned to thetrajectory to which they had the highest probability of belonging. This strategy was selected becauseour posterior probabilities of trajectory subgroup membership were high and using probabilisticmembership in trajectories may not be useful in clinical practice. In other words, by using hardassignment to trajectories, we explicitly subgrouped patients according to pain scores.

Assessing the Factors Associated With Trajectory MembershipWe used logistic regression to assess the factors associated with post-TKA pain trajectorymembership, including patients who experienced pain relief in the postoperative period (fast painresponders) and patients who had less pain relief in the postoperative period (slow pain responders).Then all relevant factors associated with trajectory membership were screened for inclusion in themodel. Those that were predictive (P < .10) were considered for inclusion in the final model. Wecalculated the area under the receiver operating characteristic curve to assess the ability of factors inthe final model to explain the variability in the pain trajectory subgroups.51

Assessing the Predictive Ability of Trajectory MembershipWe sought to assess whether the post-TKA pain trajectory was associated with pain severity at 6months after TKA. We adjusted for preoperative covariates that are known factors associated withpain after TKA to assess the independent capability of trajectory subgroup assignment to predictpain. Model fit was examined using the Hosmer-Lemeshow test.52 We used adjusted R2 to assess theability of trajectory subgroups to explain the variability in the KOOS pain score at 6 months (ie, 26weeks) after TKA.51 Based on resources for this ancillary study and the timing of its funding, we aimedto enroll more than 500 FORCE-TJR patients without using formal sample size calculations.

Results

Cohort Characteristics and Assessment of Nonresponse BiasIn total, 2209 patients undergoing TKA returned either 2-week or 8-week surveys (or both) duringthe enrollment period. We identified 659 patients with primary TKA from the FORCE-TJR cohort whohad complete preoperative, 2-week, 8-week, and 26-week pain data and constituted the studycohort. Details are shown in the patient flowchart (eFigure in the Supplement). The mean (SD) agewas 67.1 (8.0) years, 64.5% (425 of 659) were female, the mean (SD) BMI was 30.77 (5.66), 94.5%(613 of 649) were white, 68.8% (440 of 640) had an educational level of high school graduate orgreater, and 66.9% (355 of 531) had a median annual household income exceeding $45 000(Table 1). In addition, 63.8% (415 of 650) had a Charlson Comorbidity Index of 0, and 31.8% (203 of638) had undergone a previous joint replacement surgery; the mean (SD) preoperative SF-36 PCSscore and MCS score were 34.1 (8.2) and 53.8 (11.4), respectively (Table 1).





Table 1. Demographic and Preoperative Characteristics of the Study Cohort

VariableNo./Total No. (%)(N = 659)

Sex

Male 234/659 (35.5)

Female 425/659 (64.5)

Age, mean (SD), ya 67.1 (8.0)

BMIb

<25 90/640 (14.1)

25-29.9 214/640 (33.4)

30-34.9 202/640 (31.6)

35-39.9 97/640 (15.2)

≥40 37/640 (5.8)

Race

White 613/649 (94.5)

Nonwhite 36/649 (5.5)

Marital status

Married or living with someone as married 467/642 (72.7)

Widowed, separated, or divorced 142/642 (22.1)

Never married 33/642 (5.1)

Adults living in household, No.

1 153/595 (25.7)

2 355/595 (59.7)

3 59/595 (9.9)

≥4 28/595 (4.7)

Educational level

Less than high school graduate 183/640 (28.6)

High school graduate or greater 440/640 (68.8)

Missing 17/640 (2.7)

Insurance, plus secondaryc

Medicare 385/642 (60.0)

Private or HMO 230/642 (35.8)

All othersd 27/642 (4.2)

Annual household income, median, $e

≤45 000 176/531 (33.1)

>45 000 355/531 (66.9)

Charlson Comorbidity Index Score

0 415/650 (63.8)

1 143/650 (22.0)

2 56/650 (8.6)

≥3 36/650 (5.5)

Previous joint replacement surgery

No 435/638 (68.2)

Yes 203/638 (31.8)

Oswestry Disability Index low back pain

None 282/647 (43.6)

Mild 191/647 (29.5)

Moderate 126/647 (19.5)

Severe 48/647 (7.4)

Nonsurgical joints with moderate to severe pain, No.

0 475/650 (73.1)

1 142/650 (21.8)

2 22/650 (3.4)

3 11/650 (1.7)

(continued)




In a comparison of patients in the ancillary pain study (patients responding to the 2-week and8-week pain surveys) with all patients in the main FORCE-TJR cohort (the parent study), nomeaningful clinical differences were identified despite some statistically significant differences in thepreoperative variables (including sex, BMI, emotional health, and comorbidities). Within the painsurvey cohort, 1655 of 3614 (45.8%) TJR patients (total knee replacement plus total hip replacement)completed both 2-week and 8-week surveys. No clinically meaningful differences in preoperativefactors were found between pain survey respondents (at both 2 weeks and 8 weeks) and allFORCE-TJR participants with the exception of a difference in percentage of nonwhite patients (5.5%[36 of 659] in the pain cohort vs 10.3% [2145 of 20 783] in the FORCE-TJR cohort). eTable 1 in theSupplement summarizes the KOOS pain scores, and eTable 2 in the Supplement compares otherpatient characteristics.

Factors Associated With Post-TKA Pain TrajectoriesTwo pain trajectory subgroups were identified at 8 weeks: fast pain relief in the first 8 weeks afterTKA (fast pain response; 72.4% [477 of 659] of the sample) and no fast pain relief (slow painresponse; 27.6% [182 of 659] of the sample) (Figure). The mean (SD) KOOS pain score at 6 monthswas 87.3 (13.4) in fast pain responders vs 74.5 (19.4) in slow pain responders.

Significant factors associated with the slow pain responder trajectory in univariate analyseswere lower (worse) preoperative SF-36 MCS scores (51.5 vs 54.6; P = .002), more pain before surgeryin the surgical knee (46.4 vs 49.2; P = .04), and worse preoperative KOOS ADL scores in the surgicalknee (51.2 vs 57.0; P < .001) (eTable 3 in the Supplement). Sex, age, BMI, marital status, educationallevel, insurance status, Charlson Comorbidity Index, and previous joint replacement surgery were notsignificantly associated with the pain trajectory (eTable 3 in the Supplement). In multivariable-adjusted analyses, preoperative SF-36 MCS scores and preoperative KOOS ADL scores in the surgicalknee were associated with the pain trajectory (Table 2). Each unit increase in SF-36 MCS and KOOS

Table 1. Demographic and Preoperative Characteristics of the Study Cohort (continued)

VariableNo./Total No. (%)(N = 659)

Preoperative SF-36,mean (SD)

PCS score 34.1 (8.2)

MCS score 53.8 (11.4)

Preoperative KOOS pain score for surgical knee,mean (SD)f

48.4 (17.7)

Preoperative KOOS pain score for surgical kneef

None 3/659 (0.5)

Mild 65/659 (9.9)

Moderate 258/659 (39.2)

Severe 333/659 (50.5)

Preoperative KOOS pain score for nonsurgical knee,mean (SD)f

76.1 (23.3)

Preoperative KOOS pain score for nonsurgical kneef

None 246/646 (38.1)

Mild 245/646 (37.9)

Moderate 118/646 (18.3)

Severe 37/646 (5.7)

Other preoperative KOOS scores for surgical knee,mean (SD)f

Activities of daily living score 55.4 (18.5)

Symptom score 50.1 (20.1)

Sport score 18.7 (19.2)

Quality of life score 27.9 (17.7)

Abbreviations: BMI, body mass index (calculated asweight in kilograms divided by height in meterssquared); HMO, health maintenance organization;KOOS, Knee Injury and Osteoarthritis Outcome Score;MCS, mental component summary; PCS, physicalcomponent summary; SF-36, 36-Item Short FormHealth Survey.a In total, 221 of 659 (33.5%) were younger than 65

years, and 438 of 659 (66.5%) were 65 yearsor older.

b The mean (SD) BMI was 30.8 (5.7).c Medicaid as secondary insurance was an option for

12 of 385 (3.1%) with Medicare, 2 of 278 (0.7%) withprivate/HMO, and 5 of 27 (18.5%) with all others.

d All others include uninsured and self-pay.e Income exceeding $45 000 was chosen for

categorization based on the national median income.f The KOOS pain scale ranges from 0 to 100 (0

represents extreme knee problems, and 100represents no knee problems). A preoperative scoreof 50 to less than 70 indicates moderate pain, and0 to less than 50 indicates severe pain. The KOOSitems are rated by the patient on a 5-point Likert-type scale (range, 0-10; 0 represents no problems,and 10 represents extreme problems), with each ofthe 5 scores calculated as the sum of the itemsincluded; using a Likert-type 2-week and 8-week painscale, a preoperative score of 5 to 6 indicatesmoderate pain, and 7 to 10 indicates severe pain.








ADL score in surgical knee scores was associated with odds ratios of 0.98 (95% CI, 0.96-1.00;P = .02) and 0.97 (95% CI, 0.95-0.99; P = .007) for the slow pain responder trajectory, respectively.

Unadjusted and Multivariable-Adjusted Association of Pain TrajectoriesWith 6-Month PainIn univariate analyses, pain trajectories at 8 weeks were statistically significant and clinicallymeaningfully associated with the KOOS pain score at 6 months after TKA. Race, CharlsonComorbidity Index, and preoperative SF-36 MCS scores were also statistically significantly associatedwith the KOOS pain score at 6 months after TKA (Table 3). Nonwhite race was associated with a9-point-lower KOOS pain score (75.4 vs 84.3, P = .01), Charlson Comorbidity Index of at least 3 witha 10-point-lower score compared with a score of 0 (74.7 vs 85.4, overall P < .001), and preoperativeMCS score of at least 45 with a 10-point-higher score (85.7 vs 75.9, P < .001).

The pain trajectory at 8 weeks was significantly and clinically meaningfully associated withhigher KOOS pain scores at 6 months after TKA in both the full model and reduced multivariablemodels. These results are summarized in Table 4. After adjusting for patient factors, the paintrajectory at 8 weeks after TKA was independently associated with the mean KOOS pain score at 6months, with a between-trajectory difference of −11.3 (95% CI, −13.9 to −8.7; P < .001).

We performed sensitivity analyses by imputing pain scores for those missing them at 2 weeksor 8 weeks. The same variables that were significant in the main analyses were significant in thesensitivity analyses, with minimal attenuation of coefficients.

Discussion

In this prospective cohort study of a representative US cohort, we combined data from theFORCE-TJR national cohort of primary TJR outcomes with ancillary pain survey data at 2 weeks and 8weeks after TKA. We examined the patterns and characteristics of discrete posttrajectory subgroupsin the first 8 weeks after surgery and how these trajectories predicted longer-term pain outcomeafter primary TKA. The 659 patients with preoperative, 2-week, and 8-week pain trajectory datawere similar in patient profiles and comorbidities to the FORCE-TJR cohort. Two pain trajectorysubgroups were identified at 8 weeks: two-thirds of patients who experienced fast pain relief in thepostoperative period (fast pain responders) and one-third of patients who did not (slow painresponders). After adjusting for patient factors, the pain trajectory subgroup at 8 weeks after TKAwas independently associated with the mean KOOS pain score at 6 months, indicating relevance tointermediate-term and long-term TKA outcomes.

Figure. Pain Trajectory in Fast Pain Responders and Slow Pain Responders at 8 WeeksAfter Total Knee Arthroplasty

Preoperative 4 6 8

1.0

0.8

Prob

abili

ty o

f Pai

n

Time, wk

0.6

0.4

0.2

0

2

Fast pain responder

Slow pain responder

Probability of moderate to severe pain is scaled 0 to 1(lower indicates less probability, while higher indicatesgreater probability). The shaded areas are 95% CIs.Using the Knee Injury and Osteoarthritis OutcomeScore (KOOS) pain scale (range, 0-100; 0 representsextreme knee problems, and 100 represents no kneeproblems), a preoperative score of 50 to less than 70indicates moderate pain, and 0 to less than 50indicates severe pain. The KOOS items are rated by thepatient on a 5-point Likert-type scale (range, 0-10; 0represents no problems, and 10 represents extremeproblems), with each of the 5 scores calculated as thesum of the items included. Using the Likert-type2-week and 8-week pain scale, a preoperative score of5 to 6 indicates moderate pain, and 7 to 10 indicatessevere pain.




Table 2. Multivariable-Adjusted Logistic Regression Model AssessingPre–Total Knee Arthroplasty (TKA) Factors Associated With the SlowPain Responder Trajectorya

Variable

Full Modelb

OR (95% CI) P ValuePreoperative SF-36 MCS score,with 1-unit increase

0.98 (0.96-1.00) .02

Preoperative KOOS activities of dailyliving score for surgical knee,with 1-unit increase

0.97 (0.95-0.99) .007

Sex

Male 1 [Reference] NA

Female 1.13 (0.73-1.73) .58

Age, y

<65 1 [Reference] NA

≥65 0.97 (0.63-1.49) .21

BMI

<25

25-29.9 1.97 (1.00-3.84) .05

30-34.9 1.12 (0.56-2.24) .74

≥35 1.43 (0.69-2.95) .34

Race

White 1 [Reference] NA

Nonwhite 1.56 (0.70-3.49) NA

Charlson Comorbidity Index Score

0 1 [Reference] NA

1 1.25 (0.79-2.00) .34

2 1.20 (0.60-2.40) .42

≥3 0.78 (0.32-1.91) .35

Previous joint replacement surgery 0.82 (0.52-1.27) .37

No 1 [Reference] NA

Oswestry Disability Index low back pain

None 1 [Reference] NA

Mild 0.93 (0.58-1.49) .76

Moderate 0.92 (0.53-1.58) .76

Severe 0.87 (0.40-1.90) .72

Nonsurgical joints with moderateto severe pain, No.

0 1 [Reference] NA

1 0.91 (0.54-1.51) .70

2 1.17 (0.43-3.15) .76

3 0.73 (0.17-3.07) .67

Preoperative SF-36 PCS score,with 1-unit increase

1.00 (0.97-1.03) .95

Preoperative KOOS pain scorefor surgical knee, with 1-unit increase

1.02 (1.00-1.04) .06

Area under the receiver operatingcharacteristic curve

0.64 NA

Abbreviations: BMI, body mass index (calculated as weight in kilograms dividedby height in meters squared); KOOS, Knee Injury and Osteoarthritis OutcomeScore; MCS, mental component summary; NA, not applicable; OR, odds ratio;PCS, physical component summary; SF-36, 36-Item Short Form Health Survey.a No evidence of multicollinearity was found because the variance inflation

factors for each of the variables were 1.08 or lower (as a rule of thumb, avariable whose variance inflation factor values are greater than 10 may meritfurther investigation). Multivariable-adjusted logistic regression included thefollowing factors: sex, age, BMI, race, preoperative Charlson ComorbidityIndex score, previous joint replacement surgery, preoperative OswestryDisability Index low back pain, preoperative number of nonsurgical joints withmoderate to severe pain, preoperative SF-36 PCS score and MCS score,preoperative KOOS pain score for surgical knee, and KOOS activities of dailyliving score.

b Only significant variables are listed in the table (significance level, P � .05).




Total knee arthroplasty is the most common and costly joint surgery, with approximately700 000 TKA procedures being performed annually, making it a major public health burden.53 To ourknowledge, the present work is the first study of a multisite, representative cohort of patientsundergoing primary TKA in the United States that has examined the trajectory of pain after primaryTKA and its association with pain outcomes at 6 months after surgery. Although previousstudies14-16,54 examined preoperative pain (and in rare cases postoperative pain) and longer-termpostoperative pain, to our knowledge, no systematic evaluation was done to assess pain trajectoriesand their association with longer-term outcomes (eg, patient pain experience). These retrospectivestudies also did not control for important confounders and were single-center studies, reducing

Table 3. Factors Associated With 6-Month Knee Injury and OsteoarthritisOutcome Score (KOOS) Pain Scores in Univariate Analysesa

Variable

KOOS PainScore,Mean (SD) P Value

Pain trajectory 8 wk after total knee arthroplasty

Fast pain respondersb 87.3 (13.4)<.001

Slow pain respondersc 74.5 (19.4)

Sex

Male 84.4 (17.4).16

Female 83.4 (15.7)

Age, y

<65 82.2 (18.8).43

≥65 84.5 (14.9)

BMI

<25 85.9 (13.6)

.08

25-29.9 82.9 (17.2)

30-34.9 84.7 (16.4)

35-39.9 80.9 (16.9)

≥40 87.0 (14.3)

Race

White 84.3 (15.8).01

Nonwhite 75.4 (22.2)

Annual household income, $

≤45 000 82.3 (18.5).17

>45 000 85.3 (14.6)

Charlson Comorbidity Index

0 85.4 (15.5)

<.0011 82.5 (17.3)

2 80.1 (17.5)

≥3 74.7 (16.3)

Preoperative SF-36 MCS score

<45 75.9 (19.3)<.001

≥45 85.7 (14.8)

Abbreviations: BMI, body mass index (calculated as weight in kilograms dividedby height in meters squared); MCS, mental component summary; SF-36,36-Item Short Form Health Survey.a Kruskal-Wallis test was used for the categorical variables in the table.b Patients who experienced pain relief in the immediate postoperative period

are referred to as fast pain responders. The mean (SD) preoperative KOOS painscore for this subgroup was 49.2 (17.5), which was not different from the 46.4(18.0) of slow pain responders. The improvement from preoperative to6-month KOOS pain scores was 38.1 (19.2) in fast pain responders vs 28.1 (21.0)in slow pain responders.

c Patients who experienced minimal pain relief in the immediate postoperativeperiod are referred to as slow pain responders.




confidence in the study findings and raising questions about generalizability. In the period afterprimary TKA, we found that 72.4% (477 of 659) of patients were fast pain responders and 27.6% (182of 659) of patients were slow pain responders in the first 8 weeks. Evidence of the 2 pain trajectoriesexisted as early as 8 weeks after the primary TKA and was possibly present even earlier. Thisobservation was surprising and important because pain during this period includes postsurgical acutepain.55,56 Previous attempts at performing an analysis of pain trajectory were limited because theyassessed preoperative factors only and did not investigate pain in the postoperative period.15,16 Thefindings of an association between preoperative variables and post-TKA pain in those single-centerstudies were similar to the results of other observational studies.11,57-61 Our study highlights the needto assess pain in patients during the period after primary TKA.

We also found that the postoperative pain trajectory was independently associated with longer-term post-TKA pain outcome. Specifically, although patients in the slow pain responder subgroup at2 weeks and 8 weeks after surgery had significantly improved pain, they reported greater persistentindex knee pain at 6 months after TKA compared with the fast pain responder subgroup. This findinghas potential implications for patient care after primary TKA. Early identification of patients in theslow pain responder trajectory at 8 weeks after TKA (ie, for the subgroup continuing to experiencemoderate to severe pain) may offer an opportunity to target potentially modifiable factors forinterventions in the perioperative period to possibly improve the long-term pain outcomes inpatients who have undergone primary TKA. Persistent pain after TKA can contribute to long-termopioid use,62,63 which is a major public health crisis.64,65 Comprehensive pain managementprograms (eg, opioid prescribing guidelines66) and other behavioral interventions should be testedfor this subgroup of patients in the postoperative period after TKA, with a goal to improve outcomes.

Table 4. Assessing the Association of 8-Week Pain TrajectoryWith 6-Month Knee Injury and Osteoarthritis Outcome Score (KOOS)Pain Scoresa

VariableFinal Model 6-mo KOOS PainScore (95% CI)b P Value

Pain trajectory 8 wk after totalknee arthroplasty

Fast pain responders 1 [Reference] NA

Slow pain responders –11.3 (–13.9 to –8.7) <.001

Race

White 1 [Reference] NA

Nonwhite –6.6 (–11.6 to –1.5) .01

Charlson Comorbidity IndexScore

0 1 [Reference] NA

1 –1.3 (–4.1 to 1.5) .36

2 –1.7 (–5.9 to 2.5) .42

≥3 –6.8 (–11.8 to –1.8) .008

Preoperative SF-36,with 1-unit increase

PCS score 0.3 (0.1 to 0.4) <.001

MCS score 0.3 (0.2 to 0.4) <.001

Abbreviations: MCS, mental component summary; NA, not applicable; PCS,physical component summary; SF-36, 36-Item Short Form Health Survey.a The reduced 6-month model included pain trajectory, race, Charlson

Comorbidity Index, and preoperative SF-36 PCS score and MCS score.b Because the R2 value for the reduced model was the same as for the full model

that included all the factors based on the principle of parsimony, reducedmodels were more desirable. No evidence of multicollinearity was foundbecause the variance inflation factors for each of the variables were 1.08 orlower (a variable for which variance inflation factor values are greater than 10may merit further investigation). The adjusted R2 was 0.2




Although the mean pain scores in the slow pain responder subgroup improved by 6 months, thesepatients did not reach the same status of the fast pain responder subgroup.

In the present study, preoperative physical health, general emotional and mental health, ADLs,and quality of life were associated with pain trajectory subgroups, whereas the other demographicor clinical characteristics were not. However, we did not assess immediate postoperative events andfactors (ie, during the hospital stay and the first week after discharge to home) and thus are unableto provide additional insights regarding other postoperative variables amenable to intervention toimprove outcomes. Future studies should examine whether the in-hospital, post-TKA period includesany modifiable factors associated with the 2-week to 8-week pain trajectory so that interventionstargeting these factors may be developed and implemented.

The recent institution of bundled payments by the US Centers for Medicare & Medicaid Services(CMS) represents a major shift in the reimbursement of inpatient procedures, including TKA,67 butdoes not explicitly address patient-reported outcomes. According to this newly enacted policy, theCMS will pay for 90 days of care (ie, inpatient stay for TKA plus the first 90 days after discharge)rather than for each of the individual services to avoid fragmented care. This payment change by theCMS, which pays for most TKA procedures in the United States (totaling $7 billion in 2014 for kneeand hip replacements68), will have major consequences on TKA reimbursement. A key change inadapting to the new payment system will be more explicit shared decision-making and a focus onimproving patient-reported outcomes, as well as prevention of medical and surgical complicationsthat lead to rehospitalization and increased cost of care.

Our study findings suggest that tailored pain treatment that targets the slow pain responders inthe postoperative period should be tested for association with final pain outcomes. Taken together,the present study results and the new CMS payment initiative indicate the need for more research inthis area and may ultimately result in improved patient pain outcomes and satisfaction. In addition,this study presents a new application of an emerging analytic method (trajectory analysis technique)to predict postoperative outcomes (using pain as an example). More research is needed in the arenaof perioperative interventions, which have the potential to improve long-term post-TKAoutcomes.69,70

Strengths and LimitationsThe study has notable strengths, including a national, pragmatic sample of contemporary patientsundergoing TKA; use of a replicable, simple pain numeric rating scale at 2 weeks and 8 weeks;application of emerging trajectory tools; robustness of our study results; and a comprehensiveassessment that included potential confounders and covariates in the analyses.

This study has limitations. First, although the patient profiles were comparable between theFORCE-TJR national cohort and our ancillary pain trajectory study sample, fewer nonwhite patientsand patients with less comorbidity were included in the latter because of the nature of patientstreated at the community sites that participated in the trajectory study. Second, this study did notcollect pain data during the inpatient stay and the first week after TKA. This information mayaugment future post-TKA trajectory analyses. We included patients with complete data at 4 timepoints (preoperative, 2 weeks, 8 weeks, and 26 weeks) in these analyses. To ensure that patientswith complete data were similar to the total FORCE-TJR cohort, we also performed sensitivityanalyses, and the analytic subgroup did not meaningfully differ from the cohort.

Conclusions

We identified 2 postoperative pain trajectories after primary TKA. The 2 trajectories were evident asearly as 2 weeks after TKA and suggest that patients at risk of poor pain outcome may be identifiedafter TKA. A simple pain numeric rating scale measure can be used after TKA in orthopedic practicesto identify these pain trajectories. The wide availability of cell phones with texting ability and the useof electronic health records in orthopedic surgeon offices in the United States make this possible




and feasible. We also found that the postoperative pain trajectory was independently associated withlonger-term pain outcome at 6 months after TKA. Preoperative emotional and mental health statusand ability to perform ADLs were associated with the pain trajectory and may guide tailored painmanagement. More research is needed to elucidate modifiable contributors to slow pain responderstatus in the postoperative period. Interventions targeting the postoperative pain trajectory and itscorrelates have the possibility to improve primary TKA outcomes.

ARTICLE INFORMATIONAccepted for Publication: September 21, 2019.

Published: November 13, 2019. doi:10.1001/jamanetworkopen.2019.15105

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2019 Singh JA et al.JAMA Network Open.

Corresponding Author: Jasvinder A. Singh, MBBS, MPH, Department of Medicine, The University of Alabama atBirmingham, 510 Twentieth St South, Faculty Office Tower Room 805B, Birmingham, AL 35294 ([email protected]).

Author Affiliations: Medicine Service, Birmingham Veterans Affairs Medical Center, Birmingham, Alabama(Singh); Department of Medicine, The University of Alabama at Birmingham (Singh, Saag); Division ofEpidemiology, School of Public Health, The University of Alabama at Birmingham (Singh, Saag); Department ofOrthopedics and Physical Rehabilitation, University of Massachusetts Medical School, Worcester (Lemay, Yang,Franklin); Department of Population and Quantitative Health Sciences, University of Massachusetts MedicalSchool, Worcester (Nobel, Weissman, Allison).

Author Contributions: Ms Yang and Dr Franklin had full access to all of the data in the study and take responsibilityfor the integrity of the data and the accuracy of the data analysis.

Concept and design: Singh, Weissman, Saag, Allison, Franklin.

Acquisition, analysis, or interpretation of data: Singh, Lemay, Nobel, Yang, Saag, Allison, Franklin.

Drafting of the manuscript: Singh, Lemay, Nobel, Yang, Saag, Allison, Franklin.

Critical revision of the manuscript for important intellectual content: Singh, Weissman, Saag, Allison, Franklin.

Statistical analysis: Singh, Nobel, Yang, Allison.

Obtained funding: Singh, Weissman, Saag, Allison, Franklin.

Administrative, technical, or material support: Singh, Lemay, Saag, Franklin.

Supervision: Singh.

Conflict of Interest Disclosures: Dr Singh reported receiving consultant fees from Crealta/Horizon, Fidia, UBMLLC, Medscape, WebMD, the National Institutes of Health (NIH), and the American College of Rheumatology(ACR); receiving personal fees from Clinical Care Options, Clearview Healthcare Partners, Putnam Associates, andSpherix; receiving other support from OMERACT (an organization that develops outcome measures inrheumatology and receives arms-length funding from 36 companies), Food and Drug Administration arthritisadvisory committee, Veterans Affairs rheumatology field advisory committee, and UAB Cochrane MusculoskeletalGroup Satellite Center on Network Meta-analysis; owning stock options in Amarin Pharmaceuticals and VikingTherapeutics; being a member of the executive committee of OMERACT and the Veterans Affairs rheumatologyfield advisory committee; being editor and director of the UAB Cochrane Musculoskeletal Group Satellite Centeron Network Meta-analysis; and previously serving as a member or chair of the following committees: ACR annualmeeting planning committee and quality of care committee; ACR meet the professor, workshop, and study groupsubcommittees; and ACR criteria and response criteria subcommittee. Dr Nobel reported receiving grants from theNIH and Canadian Institutes of Health Research during the conduct of the study. Ms Yang reported receiving grantsfrom the Agency for Healthcare Research and Quality (AHRQ) during the conduct of the study. Dr Saag reportedreceiving research grants from Amgen, Ironwood/AstraZeneca, Horizon, Merck, Sobi, and Takeda Pharmaceuticalsand receiving consultant fees from Abbott, Amgen, Ironwood/AstraZeneca, Bayer, Bristol-Myers Squibb, Horizon,Lilly, Merck, Pfizer, Radius, Roche/Genentech, Sobi, and Takeda Pharmaceuticals. Dr Franklin reported receivinggrants from the National Institute of Arthritis and Musculoskeletal and Skin Diseases, Patient-Centered OutcomesResearch Institute, and AHRQ outside the submitted work. No other disclosures were reported.

Funding/Support: This study was supported by the AHRQ grant U19 HS021110 (Dr Saag) and grant P50 HS018910(Dr Franklin).





https://jamanetwork.com/journals/jamanetworkopen/pages/instructions-for-authors#SecOpenAccess/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamanetworkopen.2019.15105

mailto:[email protected]

mailto:[email protected]

Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection,management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; anddecision to submit the manuscript for publication.

Additional Contributions: We thank the patients for participating in the study.

REFERENCES1. Chou R, Gordon DB, de Leon-Casasola OA, et al. Management of postoperative pain: a clinical practice guidelinefrom the American Pain Society, the American Society of Regional Anesthesia and Pain Medicine, and the AmericanSociety of Anesthesiologists’ Committee on Regional Anesthesia, Executive Committee, and AdministrativeCouncil [published dosage correction appears in J Pain. 2016;17(4):508-510]. J Pain. 2016;17(2):131-157. doi:10.1016/j.jpain.2015.12.008

2. Aubrun F, Nouette-Gaulain K, Fletcher D, et al. Revision of expert panel’s guidelines on postoperative painmanagement. Anaesth Crit Care Pain Med. 2019;38(4):405-411. doi:10.1016/j.accpm.2019.02.011

3. Cooney MF. Postoperative pain management: clinical practice guidelines. J Perianesth Nurs. 2016;31(5):445-451. doi:10.1016/j.jopan.2016.08.001

4. Rosenberg J, Kehlet H. Does effective postoperative pain management influence surgical morbidity? Eur SurgRes. 1999;31(2):133-137. doi:10.1159/000008631

5. Peters CL, Shirley B, Erickson J. The effect of a new multimodal perioperative anesthetic regimen onpostoperative pain, side effects, rehabilitation, and length of hospital stay after total joint arthroplasty.J Arthroplasty. 2006;21(6)(suppl 2):132-138. doi:10.1016/j.arth.2006.04.017

6. Guimarães-Pereira L, Farinha F, Azevedo L, Abelha F, Castro-Lopes J. Persistent postoperative pain after cardiacsurgery: incidence, characterization, associated factors and its impact in quality of life. Eur J Pain. 2016;20(9):1433-1442. doi:10.1002/ejp.866

7. Shyu YI, Chen ML, Chen MC, Wu CC, Su JY. Postoperative pain and its impact on quality of life for hip-fracturedolder people over 12 months after hospital discharge. J Clin Nurs. 2009;18(5):755-764. doi:10.1111/j.1365-2702.2008.02611.x

8. Wu CL, Naqibuddin M, Rowlingson AJ, Lietman SA, Jermyn RM, Fleisher LA. The effect of pain on health-relatedquality of life in the immediate postoperative period. Anesth Analg. 2003;97(4):1078-1085, table of contents.doi:10.1213/01.ANE.0000081722.09164.D5

9. Choinière M, Watt-Watson J, Victor JC, et al. Prevalence of and risk factors for persistent postoperativenonanginal pain after cardiac surgery: a 2-year prospective multicentre study. CMAJ. 2014;186(7):E213-E223. doi:10.1503/cmaj.131012

10. Ethgen O, Bruyère O, Richy F, Dardennes C, Reginster JY. Health-related quality of life in total hip and totalknee arthroplasty: a qualitative and systematic review of the literature. J Bone Joint Surg Am. 2004;86(5):963-974. doi:10.2106/00004623-200405000-00012

11. Singh JA, Gabriel S, Lewallen D. The impact of gender, age, and preoperative pain severity on pain after TKA.Clin Orthop Relat Res. 2008;466(11):2717-2723. doi:10.1007/s11999-008-0399-9

12. Wylde V, Hewlett S, Learmonth ID, Dieppe P. Persistent pain after joint replacement: prevalence, sensoryqualities, and postoperative determinants. Pain. 2011;152(3):566-572. doi:10.1016/j.pain.2010.11.023

13. Beswick AD, Wylde V, Gooberman-Hill R, Blom A, Dieppe P. What proportion of patients report long-term painafter total hip or knee replacement for osteoarthritis? a systematic review of prospective studies in unselectedpatients. BMJ Open. 2012;2(1):e000435. doi:10.1136/bmjopen-2011-000435

14. Brander VA, Stulberg SD, Adams AD, et al. Predicting total knee replacement pain: a prospective, observationalstudy. Clin Orthop Relat Res. 2003;(416):27-36. doi:10.1097/01.blo.0000092983.12414.e9

15. Dowsey MM, Smith AJ, Choong PFM. Latent class growth analysis predicts long term pain and functiontrajectories in total knee arthroplasty: a study of 689 patients. Osteoarthritis Cartilage. 2015;23(12):2141-2149. doi:10.1016/j.joca.2015.07.005

16. Lenguerrand E, Wylde V, Gooberman-Hill R, et al. Trajectories of pain and function after primary hip and kneearthroplasty: the ADAPT cohort study. PLoS One. 2016;11(2):e0149306. doi:10.1371/journal.pone.0149306

17. Singelyn FJ, Deyaert M, Joris D, Pendeville E, Gouverneur JM. Effects of intravenous patient-controlledanalgesia with morphine, continuous epidural analgesia, and continuous three-in-one block on postoperative painand knee rehabilitation after unilateral total knee arthroplasty. Anesth Analg. 1998;87(1):88-92. doi:10.1097/00000539-199807000-00019

18. Munin MC, Rudy TE, Glynn NW, Crossett LS, Rubash HE. Early inpatient rehabilitation after elective hip andknee arthroplasty. JAMA. 1998;279(11):847-852. doi:10.1001/jama.279.11.847




https://dx.doi.org/10.1016/j.jpain.2015.12.008


https://dx.doi.org/10.1016/j.accpm.2019.02.011

https://dx.doi.org/10.1016/j.jopan.2016.08.001

https://dx.doi.org/10.1159/000008631

https://dx.doi.org/10.1016/j.arth.2006.04.017

https://dx.doi.org/10.1002/ejp.866

https://dx.doi.org/10.1111/j.1365-2702.2008.02611.x


https://dx.doi.org/10.1213/01.ANE.0000081722.09164.D5

https://dx.doi.org/10.1503/cmaj.131012

https://dx.doi.org/10.2106/00004623-200405000-00012

https://dx.doi.org/10.1007/s11999-008-0399-9

https://dx.doi.org/10.1016/j.pain.2010.11.023

https://dx.doi.org/10.1136/bmjopen-2011-000435

https://dx.doi.org/10.1097/01.blo.0000092983.12414.e9

https://dx.doi.org/10.1016/j.joca.2015.07.005

https://dx.doi.org/10.1371/journal.pone.0149306

https://dx.doi.org/10.1097/00000539-199807000-00019

https://dx.doi.org/10.1097/00000539-199807000-00019

https://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.279.11.847&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamanetworkopen.2019.15105

19. Goesling J, Moser SE, Zaidi B, et al. Trends and predictors of opioid use after total knee and total hiparthroplasty. Pain. 2016;157(6):1259-1265. doi:10.1097/j.pain.0000000000000516

20. Patanwala AE, Jarzyna DL, Miller MD, Erstad BL. Comparison of opioid requirements and analgesic responsein opioid-tolerant versus opioid-naïve patients after total knee arthroplasty. Pharmacotherapy. 2008;28(12):1453-1460. doi:10.1592/phco.28.12.1453

21. Singh JA, Lewallen DG. Predictors of use of pain medications for persistent knee pain after primary total kneearthroplasty: a cohort study using an institutional joint registry. Arthritis Res Ther. 2012;14(6):R248. doi:10.1186/ar4091

22. Manchikanti L, Helm S II, Fellows B, et al. Opioid epidemic in the United States. Pain Physician. 2012;15(3)(suppl):ES9-ES38.

23. Sehgal N, Manchikanti L, Smith HS. Prescription opioid abuse in chronic pain: a review of opioid abusepredictors and strategies to curb opioid abuse. Pain Physician. 2012;15(3)(suppl):ES67-ES92.

24. Manchikanti L, Abdi S, Atluri S, et al; American Society of Interventional Pain Physicians. American Society ofInterventional Pain Physicians (ASIPP) guidelines for responsible opioid prescribing in chronic non-cancer pain:part 2–guidance. Pain Physician. 2012;15(3)(suppl):S67-S116.

25. Manchikanti L, Abdi S, Atluri S, et al; American Society of Interventional Pain Physicians. American Society ofInterventional Pain Physicians (ASIPP) guidelines for responsible opioid prescribing in chronic non-cancer pain,part I: evidence assessment. Pain Physician. 2012;15(3)(suppl):S1-S65.

26. Downey L, Engelberg RA. Quality-of-life trajectories at the end of life: assessments over time by patients withand without cancer. J Am Geriatr Soc. 2010;58(3):472-479. doi:10.1111/j.1532-5415.2010.02734.x

27. Franklin PD, Allison JJ, Ayers DC. Beyond joint implant registries: a patient-centered research consortium forcomparative effectiveness in total joint replacement. JAMA. 2012;308(12):1217-1218. doi:10.1001/jama.2012.12568

28. Ayers DC, Fehring TK, Odum SM, Franklin PD. Using joint registry data from FORCE-TJR to improve theaccuracy of risk-adjustment prediction models for thirty-day readmission after total hip replacement and totalknee replacement. J Bone Joint Surg Am. 2015;97(8):668-671. doi:10.2106/JBJS.N.00889

29. Franklin PD, Lewallen D, Bozic K, Hallstrom B, Jiranek W, Ayers DC. Implementation of patient-reportedoutcome measures in U.S. total joint replacement registries: rationale, status, and plans. J Bone Joint Surg Am.2014;96(suppl 1):104-109. doi:10.2106/JBJS.N.00328

30. Gordon DB, Polomano RC, Pellino TA, et al. Revised American Pain Society Patient Outcome Questionnaire(APS-POQ-R) for quality improvement of pain management in hospitalized adults: preliminary psychometricevaluation. J Pain. 2010;11(11):1172-1186. doi:10.1016/j.jpain.2010.02.012

31. McCaffery M, Beebe A. Pain: Clinical Manual for Nursing Practice. St Louis, MO: CV Mosby Co; 1989.

32. Eriksson K, Wikström L, Årestedt K, Fridlund B, Broström A. Numeric rating scale: patients’ perceptions of itsuse in postoperative pain assessments. Appl Nurs Res. 2014;27(1):41-46. doi:10.1016/j.apnr.2013.10.006

33. Michener LA, Snyder AR, Leggin BG. Responsiveness of the numeric pain rating scale in patients with shoulderpain and the effect of surgical status. J Sport Rehabil. 2011;20(1):115-128. doi:10.1123/jsr.20.1.115

34. Childs JD, Piva SR, Fritz JM. Responsiveness of the numeric pain rating scale in patients with low back pain.Spine (Phila Pa 1976). 2005;30(11):1331-1334. doi:10.1097/01.brs.0000164099.92112.29

35. Strout TD, Burton JH. Clinically significant change in physician-assigned numeric pain rating scale scores. Am JEmerg Med. 2004;22(3):243-245. doi:10.1016/j.ajem.2004.02.039

36. Joos E, Peretz A, Beguin S, Famaey JP. Reliability and reproducibility of visual analogue scale and numericrating scale for therapeutic evaluation of pain in rheumatic patients. J Rheumatol. 1991;18(8):1269-1270.

37. Roos EM, Roos HP, Lohmander LS, Ekdahl C, Beynnon BD. Knee Injury and Osteoarthritis Outcome Score(KOOS): development of a self-administered outcome measure. J Orthop Sports Phys Ther. 1998;28(2):88-96.doi:10.2519/jospt.1998.28.2.88

38. Roos EM, Toksvig-Larsen S. Knee Injury and Osteoarthritis Outcome Score (KOOS): validation and comparisonto the WOMAC in total knee replacement. Health Qual Life Outcomes. 2003;1:17. doi:10.1186/1477-7525-1-17

39. Roos EM, Lohmander LS. The Knee Injury and Osteoarthritis Outcome Score (KOOS): from joint injury toosteoarthritis. Health Qual Life Outcomes. 2003;1:64. doi:10.1186/1477-7525-1-64




https://dx.doi.org/10.1097/j.pain.0000000000000516

https://dx.doi.org/10.1592/phco.28.12.1453

https://dx.doi.org/10.1186/ar4091

https://dx.doi.org/10.1186/ar4091

https://www.ncbi.nlm.nih.gov/pubmed/22786464






https://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2012.12568&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamanetworkopen.2019.15105

https://dx.doi.org/10.2106/JBJS.N.00889

https://dx.doi.org/10.2106/JBJS.N.00328


https://dx.doi.org/10.1016/j.apnr.2013.10.006

https://dx.doi.org/10.1123/jsr.20.1.115

https://dx.doi.org/10.1097/01.brs.0000164099.92112.29

https://dx.doi.org/10.1016/j.ajem.2004.02.039


https://dx.doi.org/10.2519/jospt.1998.28.2.88

https://dx.doi.org/10.1186/1477-7525-1-17

https://dx.doi.org/10.1186/1477-7525-1-64

40. Collins NJ, Misra D, Felson DT, Crossley KM, Roos EM. Measures of knee function: International KneeDocumentation Committee (IKDC) Subjective Knee Evaluation Form, Knee Injury and Osteoarthritis OutcomeScore (KOOS), Knee Injury and Osteoarthritis Outcome Score Physical Function Short Form (KOOS-PS), KneeOutcome Survey Activities of Daily Living Scale (KOS-ADL), Lysholm Knee Scoring Scale, Oxford Knee Score (OKS),Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), Activity Rating Scale (ARS), andTegner Activity Score (TAS). Arthritis Care Res (Hoboken). 2011;63(suppl 11):S208-S228. doi:10.1002/acr.20632

41. Alviar MJ, Olver J, Brand C, Hale T, Khan F. Do patient-reported outcome measures used in assessing outcomesin rehabilitation after hip and knee arthroplasty capture issues relevant to patients? results of a systematic reviewand ICF linking process. J Rehabil Med. 2011;43(5):374-381. doi:10.2340/16501977-0801

42. Ingelsrud LH, Terwee CB, Terluin B, et al. Meaningful change scores in the Knee Injury and OsteoarthritisOutcome Score in patients undergoing anterior cruciate ligament reconstruction. Am J Sports Med. 2018;46(5):1120-1128. doi:10.1177/0363546518759543

43. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity inlongitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383. doi:10.1016/0021-9681(87)90171-8

44. Gandek B, Ware JE, Aaronson NK, et al. Cross-validation of item selection and scoring for the SF-12 HealthSurvey in nine countries: results from the IQOLA Project: International Quality of Life Assessment. J ClinEpidemiol. 1998;51(11):1171-1178. doi:10.1016/S0895-4356(98)00109-7

45. Ware J Jr, Kosinski M, Keller SDA. A 12-Item Short-Form Health Survey: construction of scales and preliminarytests of reliability and validity. Med Care. 1996;34(3):220-233. doi:10.1097/00005650-199603000-00003

46. Fairbank JC, Pynsent PB. The Oswestry Disability Index. Spine (Phila Pa 1976). 2000;25(22):2940-2952. doi:10.1097/00007632-200011150-00017

47. Nagin DS. Group-based trajectory modeling: an overview. In: Piquero AR, Weisburd D, eds. Handbook ofQuantitative Criminology. New York, NY: Springer New York; 2010:53-67. doi:10.1007/978-0-387-77650-7_4

48. Nagin DS, Odgers CL. Group-based trajectory modeling in clinical research. Annu Rev Clin Psychol. 2010;6:109-138. doi:10.1146/annurev.clinpsy.121208.131413

49. Jones BL, Nagin DS. A Stata plugin for estimating group-based trajectory models. https://ssrc.indiana.edu/doc/wimdocs/2013-03-29_nagin_trajectory_stata-plugin-info.pdf. Published May 21, 2012. Accessed October 9, 2019.

50. Franklin JM, Shrank WH, Pakes J, et al. Group-based trajectory models: a new approach to classifying andpredicting long-term medication adherence. Med Care. 2013;51(9):789-796. doi:10.1097/MLR.0b013e3182984c1f

51. Pepe M, Longton G, Janes H. Estimation and comparison of receiver operating characteristic curves. Stata J.2009;9(1):1-16. doi:10.1177/1536867X0900900101

52. Hosmer DW, Lemeshow S. Goodness of fit tests for the multiple logistic regression model. Commun StatTheory Methods. 1980;9(10):1043-1069. doi:10.1080/03610928008827941

53. Kurtz S, Ong K, Lau E, Mowat F, Halpern M. Projections of primary and revision hip and knee arthroplasty in theUnited States from 2005 to 2030. J Bone Joint Surg Am. 2007;89(4):780-785. doi:10.2106/JBJS.F.00222

54. Brander V, Gondek S, Martin E, Stulberg SD. Pain and depression influence outcome 5 years after kneereplacement surgery. Clin Orthop Relat Res. 2007;464(464):21-26. doi:10.1097/BLO.0b013e318126c032

55. Chan EY, Blyth FM, Cheow SL, Fransen M. Postoperative pain following hospital discharge after kneereplacement surgery: a patient survey. Pain Manag. 2013;3(3):177-188. doi:10.2217/pmt.13.14

56. Chan EY, Blyth FM, Nairn L, Fransen M. Acute postoperative pain following hospital discharge after total kneearthroplasty. Osteoarthritis Cartilage. 2013;21(9):1257-1263. doi:10.1016/j.joca.2013.06.011

57. Burns LC, Ritvo SE, Ferguson MK, Clarke H, Seltzer Z, Katz J. Pain catastrophizing as a risk factor for chronicpain after total knee arthroplasty: a systematic review. J Pain Res. 2015;8:21-32. doi:10.2147/JPR.S64730

58. Hilton ME, Gioe T, Noorbaloochi S, Singh JA. Increasing comorbidity is associated with worsening physicalfunction and pain after primary total knee arthroplasty. BMC Musculoskelet Disord. 2016;17(1):421. doi:10.1186/s12891-016-1261-y

59. Pinto PR, McIntyre T, Ferrero R, Almeida A, Araújo-Soares V. Risk factors for moderate and severe persistentpain in patients undergoing total knee and hip arthroplasty: a prospective predictive study. PLoS One. 2013;8(9):e73917. doi:10.1371/journal.pone.0073917

60. Singh JA, Lewallen DG. Better functional and similar pain outcomes in osteoarthritis compared to rheumatoidarthritis after primary total knee arthroplasty: a cohort study. Arthritis Care Res (Hoboken). 2013;65(12):1936-1941. doi:10.1002/acr.22090




https://dx.doi.org/10.1002/acr.20632

https://dx.doi.org/10.2340/16501977-0801

https://dx.doi.org/10.1177/0363546518759543

https://dx.doi.org/10.1016/0021-9681(87)90171-8

https://dx.doi.org/10.1016/0021-9681(87)90171-8

https://dx.doi.org/10.1016/S0895-4356(98)00109-7

https://dx.doi.org/10.1097/00005650-199603000-00003

https://dx.doi.org/10.1097/00007632-200011150-00017

https://dx.doi.org/10.1007/978-0-387-77650-7_4

https://dx.doi.org/10.1146/annurev.clinpsy.121208.131413

https://ssrc.indiana.edu/doc/wimdocs/2013-03-29_nagin_trajectory_stata-plugin-info.pdf

https://ssrc.indiana.edu/doc/wimdocs/2013-03-29_nagin_trajectory_stata-plugin-info.pdf

https://dx.doi.org/10.1097/MLR.0b013e3182984c1f

https://dx.doi.org/10.1177/1536867X0900900101

https://dx.doi.org/10.1080/03610928008827941

https://dx.doi.org/10.2106/JBJS.F.00222

https://dx.doi.org/10.1097/BLO.0b013e318126c032

https://dx.doi.org/10.2217/pmt.13.14

https://dx.doi.org/10.1016/j.joca.2013.06.011

https://dx.doi.org/10.2147/JPR.S64730

https://dx.doi.org/10.1186/s12891-016-1261-y

https://dx.doi.org/10.1186/s12891-016-1261-y

https://dx.doi.org/10.1371/journal.pone.0073917

https://dx.doi.org/10.1002/acr.22090

61. Singh JA, Lewallen DG. Ipsilateral lower extremity joint involvement increases the risk of poor pain andfunction outcomes after hip or knee arthroplasty. BMC Med. 2013;11:144. doi:10.1186/1741-7015-11-144

62. Bedard NA, DeMik DE, Dowdle SB, Callaghan JJ. Trends and risk factors for prolonged opioid use afterunicompartmental knee arthroplasty. Bone Joint J. 2018;100-B(1)(suppl A):62-67. doi:10.1302/0301-620X.100B1.BJJ-2017-0547.R1

63. Trasolini NA, McKnight BM, Dorr LD. The opioid crisis and the orthopedic surgeon. J Arthroplasty. 2018;33(11):3379-3382.e1. doi:10.1016/j.arth.2018.07.002

64. Volkow ND, Collins FS. The role of science in addressing the opioid crisis. N Engl J Med. 2017;377(4):391-394.doi:10.1056/NEJMsr1706626

65. Kolodny A, Courtwright DT, Hwang CS, et al. The prescription opioid and heroin crisis: a public health approachto an epidemic of addiction. Annu Rev Public Health. 2015;36:559-574. doi:10.1146/annurev-publhealth-031914-122957

66. Holte AJ, Carender CN, Noiseux NO, Otero JE, Brown TS. Restrictive opioid prescribing protocols followingtotal hip arthroplasty and total knee arthroplasty are safe and effective. J Arthroplasty. 2019;34(7S):S135-S139.doi:10.1016/j.arth.2019.02.022

67. Centers for Medicare & Medicaid Services. Bundled Payments for Care Improvement (BPCI) initiative: generalinformation. https://innovation.cms.gov/initiatives/bundled-payments/. Updated July 1, 2016. Accessed October 6,2019.

68. Centers for Medicare & Medicaid Services. Comprehensive Care for Joint Replacement model. https://innovation.cms.gov/initiatives/CJR. Updated September 18, 2019. Accessed October 6, 2019.

69. Niraj G, Rowbotham DJ. Persistent postoperative pain: where are we now? Br J Anaesth. 2011;107(1):25-29.doi:10.1093/bja/aer116

70. Mishriky BM, Waldron NH, Habib AS. Impact of pregabalin on acute and persistent postoperative pain:a systematic review and meta-analysis. Br J Anaesth. 2015;114(1):10-31. doi:10.1093/bja/aeu293

SUPPLEMENT.eFigure. Patient Flow CharteTable 1. Preoperative and Postoperative KOOS Pain Characteristics of FORCE-TJR Cohort, Patients at Eligible SiteDuring Our Pain Study Enrollment, and Those Participating in Our Pain StudyeTable 2. Comparison of Patient and Clinical Characteristics Between the FORCE-TJR Cohort and the Ancillary PainStudy CohorteTable 3. Unadjusted Correlates of Pain Trajectories




https://dx.doi.org/10.1186/1741-7015-11-144

https://dx.doi.org/10.1302/0301-620X.100B1.BJJ-2017-0547.R1

https://dx.doi.org/10.1302/0301-620X.100B1.BJJ-2017-0547.R1


https://dx.doi.org/10.1056/NEJMsr1706626

https://dx.doi.org/10.1146/annurev-publhealth-031914-122957

https://dx.doi.org/10.1146/annurev-publhealth-031914-122957


https://innovation.cms.gov/initiatives/bundled-payments/

https://innovation.cms.gov/initiatives/CJR

https://innovation.cms.gov/initiatives/CJR

https://dx.doi.org/10.1093/bja/aer116

https://dx.doi.org/10.1093/bja/aeu293

Documents

Association of Early Postoperative Pain Trajectories With