The Journal of Arthroplasty 37 (2022) S110eS120

Contents lists available at ScienceDirect

The Journal of Arthroplasty

journal homepage: www.arthroplastyjournal.org

Proceedings of The Knee Society 2021

Combinations of Preoperative Patient-Reported Outcome Measure

Phenotype (Pain, Function, and Mental Health) Predict Outcome
After Total Knee Arthroplasty
Melissa N. Orr, BS a, Alison K. Klika, MS a, Ahmed K. Emara, MD a,
Nicolas S. Piuzzi, MD a, *, The Cleveland Clinic Arthroplasty Group
a
Department of Orthopaedic Surgery, Cleveland Clinic, Cleveland, OH

a r t i c l e i n f o a b s t r a c t

Article history: Background: Value-driven healthcare models prioritize patient-perceived benefits to quantify the quality
Received 17 November 2021 of care through patient-reported outcome measures (PROMs). The Patient Acceptable Symptom State
Received in revised form (PASS) is the highest level of symptom beyond which a patient considers his/her condition satisfactory.
14 February 2022
We identified preoperative phenotypes of PROMs associated with not achieving PASS at 1 year following
Accepted 21 February 2022
total knee arthroplasty (TKA) and explored the relationships between such phenotypes with hospital
Available online 28 February 2022
utilization parameters.
Methods: A prospective institutional cohort of 5,274 primary TKAs for osteoarthritis from 2016 to 2019
Keywords:
veteran rand-12 mental component
with 1-year follow-up were included. Preoperative scores on Knee Disability and Osteoarthritis Outcome
summary (VR-12 MCS) Score (KOOS) Pain, KOOS-Physical function Short form (PS), and Veterans RAND 12-Item Health Survey
phenotypes (VR-12) Mental Component Summary (MCS) were used to develop patient phenotypes. Associations
patient reported outcomes (PROMs) between preoperative “phenotype” and 1-year PASS, discharge disposition, length of stay, 90-day
pain readmission, and 1-year reoperation were evaluated using multivariate regression.
function Results: In total, 16.3% (n ¼ 862) of patients reported their state as “not acceptable” at 1 year. A com-
healthcare utilization bination of low scores in each of the presently examined PROMs was associated with the highest odds of
1-year dissatisfaction (odds ratio 2.18, 95% confidence interval 1.74-2.74). The PROM phenotypes were
the greatest drivers compared to sociodemographic variables in predicting satisfaction. Combinations of
https://www.kneesociety.org/ low scores in VR-12 MCS and KOOS-PS were significantly associated with both non-home discharge
status and prolonged length of stay.
Conclusion: Patients with combined lower preoperative scores across multiple PROMs (KOOS-Pain <41.7,
KOOS-PS <51.5, and VR-12 MCS <52.8) have increased odds of dissatisfaction after TKA. Measuring pain,
function, and mental health concurrently as phenotypes may help identify TKA patients at risk for not
achieving a satisfactory outcome at 1 year.
© 2022 Elsevier Inc. All rights reserved.

The value-driven healthcare model has prioritized understand-

ing patient-perceived benefits to quantify the quality of care through
Investigation performed at the Department of Orthopedic Surgery, Cleveland patient-reported outcome measures (PROMs) [1e3]. To facilitate the
Clinic, Cleveland, Ohio. interpretation of these scores, there has been increasing focus on the
use of categorical thresholds to define improvement and to use these
Funding Sources: This research did not receive any specific grant from funding cut-offs to contextualize, align expectations, and better predict
agencies in the public, commercial, or not for-profit sectors.
future patients’ postoperative outcomes [4,5]. The Patient Accept-
One or more of the authors of this paper have disclosed potential or pertinent able Symptom State (PASS), one such threshold of clinical relevance,
conflicts of interest, which may include receipt of payment, either direct or indirect, has been defined as the highest level of symptom beyond which a
institutional support, or association with an entity in the biomedical field which patient considers himself/herself well [6,7].
may be perceived to have potential conflict of interest with this work. For full
Although total knee arthroplasty (TKA) has been established as a
disclosure statements refer to https://doi.org/10.1016/j.arth.2022.02.090.
* Address correspondence to: Nicolas S. Piuzzi, MD, Department of Orthopaedic safe and effective procedure for patients with osteoarthritis (OA) to
Surgery, Cleveland Clinic, 9500 Euclid Ave, Cleveland, OH 44195. reduce pain, improve function, and improve quality of life as

https://doi.org/10.1016/j.arth.2022.02.090
0883-5403/© 2022 Elsevier Inc. All rights reserved.
M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120 S111

measured by PROMs [8,9], a subset of patients report dissatisfaction patients were successfully enrolled at baseline into the data
after surgery [10]. Overall, PROMs can be classified into a range of collection system, of which 33 (0.44%) suffered 1-year mortality.
domains including physical health, mental and emotional health, Patients were excluded if they did not complete a 1-year follow-up
daily functioning, symptom burden, and scores in such components (n ¼ 2,169), leaving 5,274 patients (70.8%) who completed 1-year
have been associated with patient satisfaction [7,11e13]. However, follow-up and were subsequently analyzed. The median [Q1, Q3]
the domains collected across unique PROMs are interrelated, and thus age was 67 [62, 73], BMI was 31.5 [27.5, 36.1], 60.8% were female,
a holistic understanding is integral to a patient’s overall health status. and 86.1% were White.
Therefore, the current study aimed to identify combined pre-
operative phenotypes of PROMs and associate such phenotypes Assessment of Satisfaction
with not achieving PASS at 1-year following TKA. A secondary aim
is to test for associations of such phenotypes with healthcare uti- For this study, patients completed the PASS question at 1 year
lization measures of the following: (1) discharge disposition, (2) postoperative to assess satisfaction with their operative joint as
prolonged length of stay (LOS; 3 days), (3) 90-day readmission, part of a 1-year follow-up electronic data capture [14,15]. This
and (4) 1-year reoperation. electronic capture system contacts patient through their follow-up
preference of internet-based portal, email or text message. If pa-
Methods tients do not follow-up within 3 weeks they are contacted by a
phone call.
All patients who underwent primary elective unilateral TKA The PASS, a value beyond which patients can consider them-
with a diagnosis of primary OA between January 2016 and selves well, has been established and utilized for outcomes
December 2019 at one of the 9 hospitals within a North American following treatment for OA [4,6,7]. Dissatisfaction was measured by
integrated healthcare system were included. a “no” answer in response to the PASS question [23e25], worded as
Preoperative and postoperative data were prospectively follows: “Taking into account all the activity you have during your
collected within the Orthopaedic Minimal Data Set Episode of Care daily life, your level of pain and also your activity limitations and
(OME) database, which captures over 97% of elective surgical pro- participation restrictions, do you consider the current state of your
cedures within the orthopedic department [14,15]. Baseline PROMs knee satisfactory?”
included the Knee Injury and Osteoarthritis Outcome Score (KOOS)
Pain subscale [16,17], KOOS-Physical function Short form (PS) [18], Preoperative Patient-Reported Outcome Measures Phenotypes
and the Veterans RAND 12-Item Health Survey Mental Component
Summary (VR-12 MCS) [19] and were recorded in a consecutive Preoperative scores on KOOS-Pain [17,26], KOOS-PS [18], and
series of patients using Research Electronic Data Capture (REDCap) VR-12 MCS [27] were used to develop patient phenotypes. These
software. Demographic characteristics included age, gender, body PROMs were selected as they are commonly utilized in the US
mass index [BMI], race, socioeconomic determinants (including the arthroplasty literature [28]. The cohort was stratified by below ()
national Area Deprivation Index [ADI], a scaled measure with or above (þ) median scores on each of the 3 preoperative PROMs to
higher scores representing residence in a neighborhood of higher create 8 unique phenotypes [29]. Phenotypes were denoted as
socioeconomic deprivation) [20], baseline comorbidities, and sur- follows: (1) PainþPSþMCSþ, (2) PainþPSþMCS, (3) Painþ
gical history. Surgical and procedure details are recorded, as well as PSMCSþ, (4) PainþPSMCS, (5) PainPSþMCSþ, (6) PainPS
healthcare unitization parameters of LOS, discharge destination, MCS, (7) PainPSMCSþ, and (8) PainPSMCS. All PROMs,
readmissions, and emergency department visits within 90 days of including KOOS-PS which could be scored in either directions, were
procedure, and reoperations and mortality up to 1 year. PROMs on a scale 0-100 with 0 being worse, thus () denoting poorer
were collected preoperatively and at 1 year [2,21,22]. scores (Fig. 1, Supplementary Fig. 1). Median values and inter-
quartile range [IQR] for each PROM were 41.7 [30.6-50.0], 51.5
Eligibility Criteria and Population Characteristics [42.1-59.7], and 52.8 [43.0-61.1] for KOOS-Pain, KOOS-PS, and VR-
12 MCS, respectively.
All patients who underwent primary elective unilateral TKA The primary outcome of the study was to associate the 8 pre-
with a diagnosis of OA were screened for inclusion. A total of 7,476 operative PROM phenotypes with dissatisfaction after 1-year

Fig. 1. Patients were categorized into 8 preoperative PROM phenotypes by above (þ) or below () median scores in Knee Disability and Osteoarthritis Outcome Score (KOOS) Pain,
KOOS-Physical function Short form (PS) and Veterans RAND 12-Item Health Survey (VR-12) Mental Component Summary (MCS). PROM, patient-reported outcome measure.
S112 M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120

Table 1
Baseline and Demographic Variables by PROM Phenotype.

Variable PainþPSþMCSþ PainþPSþMCS PainþPSMCSþ PainþPSMCS

24.9% (N ¼ 1,315) 12.3% (N ¼ 650) 8.5% (N ¼ 447) 6.7% (N ¼ 352)

Age
Mean (SD) 68.0 (8.2) 68.0 (8.9) 68.3 (8.9) 68.7 (9.0)
Gender
Female 634 (48.2%) 352 (54.2%) 256 (57.3%) 220 (62.5%)
Male 681 (51.8%) 298 (45.8%) 191 (42.7%) 132 (37.5%)
Race
Black 78 (5.9%) 69 (10.6%) 41 (9.2%) 41 (11.6%)
Other 13 (1.0%) 14 (2.2%) 12 (2.7%) 10 (2.8%)
White 1,224 (93.1%) 567 (87.2%) 394 (88.1%) 301 (85.5%)
BMI category
Underweight, BMI <18.5 2 (0.2%) 1 (0.2%) 0 (0.0%) 0 (0.0%)
Normal weight, BMI 18.5-24.9 164 (12.5%) 81 (12.5%) 39 (8.7%) 30 (8.5%)
Overweight, BMI 25.0-29.9 485 (36.9%) 209 (32.2%) 138 (30.9%) 90 (25.6%)
Obese class I, BMI 30.0-34.9 386 (29.4%) 185 (28.5%) 123 (27.5%) 126 (35.8%)
Obese class II, BMI 35.0-39.9 192 (14.6%) 112 (17.2%) 89 (19.9%) 64 (18.2%)
Obese class III, BMI 40.0 86 (6.5%) 62 (9.5%) 58 (13.0%) 42 (11.9%)
CCI
Mean (SD) 0.9 (1.4) 1.1 (1.7) 1.1 (1.7) 1.2 (1.6)
Insurance
Commercial 658 (50.0%) 288 (44.3%) 218 (48.8%) 139 (39.5%)
Medicaid/Medicare 655 (49.8%) 357 (54.9%) 226 (50.6%) 211 (59.9%)
Self 2 (0.2%) 5 (0.8%) 3 (0.7%) 2 (0.6%)
Education (y)
Mean (SD) 14.9 (2.7) 14.3 (3.1) 14.6 (2.7) 14.1 (3.1)
Smoking status
Current 54 (4.1%) 37 (5.7%) 13 (2.9%) 21 (6.0%)
Never 747 (56.8%) 356 (54.8%) 239 (53.5%) 187 (53.1%)
Quit <6 mo 19 (1.4%) 15 (2.3%) 10 (2.2%) 10 (2.8%)
Quit >6 mo 495 (37.6%) 242 (37.2%) 185 (41.4%) 134 (38.1%)
National ADI
Mean (SD) 42.3 (24.0) 47.0 (24.5) 43.7 (23.9) 46.0 (24.2)

Variable PainPSþMCSþ PainPSþMCS PainPSMCSþ PainPSMCS P Value

4.3% (N ¼ 227) 3.1% (N ¼ 164) 15.7% (N ¼ 829) 24.5% (N ¼ 1,291)

Age <.001
Mean (SD) 65.8 (9.1) 65.6 (8.9) 65.8 (9.0) 65.6 (9.7)
Gender <.001
Female 147 (64.8%) 111 (67.7%) 569 (68.6%) 916 (71.0%)
Male 80 (35.2%) 53 (32.3%) 260 (31.4%) 375 (29.0%)
Race <.001
Black 23 (10.1%) 30 (18.3%) 114 (13.8%) 221 (17.1%)
Other 0 (0.0%) 1 (0.6%) 25 (3.0%) 39 (3.0%)
White 204 (89.9%) 133 (81.1%) 690 (83.2%) 1,031 (79.9%)
BMI category <.001
Underweight, BMI <18.5 1 (0.4%) 1 (0.6%) 2 (0.2%) 2 (0.2%)
Normal weight, BMI 18.5-24.9 29 (12.8%) 16 (9.8%) 72 (8.7%) 98 (7.6%)
Overweight, BMI 25.0-29.9 60 (26.4%) 39 (23.8%) 221 (26.7%) 311 (24.1%)
Obese class I, BMI 30.0-34.9 71 (31.3%) 50 (30.5%) 234 (28.2%) 387 (30.0%)
Obese class II, BMI 35.0-39.9 42 (18.5%) 39 (23.8%) 177 (21.4%) 265 (20.5%)
Obese class III, BMI 40.0 24 (10.6%) 19 (11.6%) 123 (14.8%) 228 (17.7%)
CCI <.001
Mean (SD) 0.9 (1.4) 1.2 (1.5) 1.0 (1.6) 1.1 (1.5)
Insurance .004
Commercial 103 (45.4%) 75 (45.7%) 429 (51.7%) 574 (44.5%)
Medicaid/Medicare 122 (53.7%) 88 (53.7%) 397 (47.9%) 712 (55.2%)
Self 2 (0.9%) 1 (0.6%) 3 (0.4%) 5 (0.4%)
Education (y) <.001
Mean (SD) 14.4 (2.8) 13.5 (2.1) 13.9 (2.7) 13.5 (2.7)
Smoking status <.001
Current 9 (4.0%) 12 (7.3%) 41 (4.9%) 125 (9.7%)
Never 140 (61.7%) 91 (55.5%) 476 (57.4%) 686 (53.1%)
Quit < 6 mo 9 (4.0%) 6 (3.7%) 30 (3.6%) 49 (3.8%)
Quit > 6 mo 69 (30.4%) 55 (33.5%) 282 (34.0%) 431 (33.4%)
National ADI <.001
Mean (SD) 47.4 (24.4) 51.4 (25.9) 47.5 (24.9) 52.4 (25.9)

Bold indicates statistically significant.

PROM, patient-reported outcome measure; PS, Physical function Short form; MCS, Mental Component Summary; ADI, Area Deprivation Index; SD, standard deviation; BMI,
body mass index; CCI, Charlson Comorbidity Index.
M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120 S113

Fig. 2. Relative OR of preoperative PROM phenotype and baseline factors in predicting 1 year dissatisfaction after TKA. OR, odds ratio; TKA, total knee arthroplasty; CI, confidence
interval; BMI, body mass index; CCI, Charlson Comorbidity Index; ADI, Area Deprivation Index; P, pain.

following TKA. Secondary outcomes included associating the 8 Charlson Comorbidity Index (CCI), ADI, smoking status, and in-
preoperative PROM phenotypes with healthcare utilization pa- surance status, and PainþPSþMCSþ was used as a reference
rameters including the following: (1) discharge disposition, (2) phenotype for the analysis. Performance metrics of area under
prolonged LOS (3 days), (3) 90-day readmission, and (4) 1-year the curve were calculated for each model utilizing the same
reoperation. cohort. Supplemental models were built to include NarxCare
Narcotic scores; however, as this score only became available in
Statistical Analysis 2018 and our cohort included patients from 2016, NarxCare
Narcotic score was available for <50% of the cohort and not
Continuous variables were reported with mean and a stan- included in the full models. The outcomes of 1-year dissatisfac-
dard deviation and categorical variables with a frequency and a tion, non-home discharge, prolonged LOS, 90-day readmission,
percentage. Continuous variables were compared between and 1-year reoperation were chosen due to their availability in
groups with analysis of variance and categorical variables with the OME data collection system and being clinically relevant to
chi-squared tests. Multivariable regression analysis was con- the patient population. All tests were 2-sided, and a P-value .05
ducted to evaluate independent associations between the pre- was considered statistically significant. All statistical analyses
operative PROM phenotypes and each outcome while adjusting were implemented in R v4.1.2 (R Project for Statistical
for potential confounders of age, BMI classification, gender, race, Computing, Vienna, Austria).
S114 M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120

Table 2 combination of low scores in KOOS-Pain and KOOS-PS

Relative Odds Ratio of Preoperative PROM Phenotype and Baseline Factors in Pre- (PainPSMCSþ) similarly increased odds of 1-year dissatisfac-
dicting 1-y Dissatisfaction After TKA.
tion (OR 1.55). A combination of low scores in both KOOS-Pain and
Characteristic Odds Ratio 95% CI P Value VR-12 MCS or KOOS-PS and VR-12 MCS was associated with twice
Phenotype the odds of 1-year dissatisfaction for both PainPSþPS and
PainþPSþMCSþ e e PainþPSMCS. Finally, having a combination of low scores in all 3
PainþPSþMCS 1.42 1.07-1.87 .014 PROMs (PainPSMCS) was associated with the highest odds of
PainþPSMCSþ 1.25 0.90-1.73 .2
1-year dissatisfaction (OR 2.18).
PainþPSMCS 2.00 1.45-2.74 <.001
PainPSþMCSþ 0.82 0.49-1.30 .4 Although the preoperative phenotypes were the greatest drivers
PainPSþMCS 2.08 1.35-3.13 <.001 of 1-year dissatisfaction, significant associations were also
PainPSMCSþ 1.55 1.20-2.01 <.001 observed between race and age with 1-year dissatisfaction. Race
PainPSMCS 2.19 1.74-2.75 <.001
was significantly associated with increased odds of 1-year dissat-
Age grouped
45-64 e e
isfaction (Black: OR 1.41), and age (65þ: OR 0.74) was significantly
18-44 1.18 0.55-2.33 .7 associated with lower odds of 1-year dissatisfaction (Fig. 2, Table 2).
65þ 0.74 0.61-0.89 .002 Associations were observed between PROM phenotypes and
BMI grouped discharge disposition. Combinations of low scores in both PS and
Normal weight
MCS, PainþPSMCS and PainPSMCS both had more than
e e
Overweight 0.81 0.62-1.07 .13
Obese class I 0.79 0.60-1.04 .083 double the likelihood of a non-home discharge (OR 2.7 and 2.3),
Obese class II 0.78 0.58-1.05 .093 respectively. However, sociodemographic factors were greater
Obese class III 0.77 0.56-1.05 .1 drivers of non-home discharge status with the greatest odds being
Gender that of age group 65þ (OR 4.45) and CCI group 5þ (OR 2.87). Higher
Female e e
Male 0.98 0.83-1.15 .8
BMI and ADI were also significantly associated with increased odds
Race of non-home discharge (Fig. 3, Table 3).
White e e Similar to discharge disposition, the PROM phenotypes which
Black 1.41 1.11-1.77 .004 included combined low scores in both PS and MCS significantly
Other 1.07 0.63-1.72 .8
increased the likelihood of prolonged LOS (3 days). Using
Insurance status
Commercial e e PainþPSþMCS þ as the reference group, PainþPSMCS and
Medicaid/Medicare 1.05 0.88-1.26 .6 PainPSMCS were both associated with 1.7 higher odds of a
CCI prolonged LOS. However, again, other sociodemographic variables
0-2 e e also predicted this outcome with higher likelihoods. Age group
3-4 0.96 0.80-1.15 .7
5þ 1.48 0.95-2.24 .07
65þ, obese class III BMI classification, and the highest National ADI
Smoking status group (81-100) were all associated with twice the odds of a pro-
Never e e longed LOS. CCI 5þ was associated with the greatest likelihood of
Current 1.26 0.92-1.70 .14 prolonged LOS (OR 2.3) (Fig. 4, Table 4). The performance metrics of
Quit <6 mo 1.41 0.91-2.11 .11
the full models were area under the curve of 0.63 for 1-year
Quit >6 mo 1.36 1.15-1.59 <.001
National ADI dissatisfaction, 0.76 for non-home discharge, and 0.68 for pro-
1-20 e e longed LOS (Supplementary Table 2).
21-40 1.03 0.81-1.31 .8 Multivariable regression did not demonstrate any significant
41-60 1.02 0.80-1.31 .9 associations between the preoperative PROM phenotypes and 90-
61-80 1.21 0.93-1.57 .15
81-100 1.21 0.90-1.62 .2
day readmission or 1-year reoperation after controlling for poten-
tial confounders (P > .05 for all). The proportion of each adverse
Bold indicates statistically significant.
outcome (1-year dissatisfaction, non-home discharge, prolonged
PROM, patient-reported outcome measure; TKA, total knee arthroplasty; CI, confi-
dence interval; PS, physical function short form; MCS, Mental Component Sum- LOS, 90-day readmission, and 1-year reoperation) was greatest for
mary; BMI, body mass index; CCI, Charlson Comorbidity Index; ADI, Area phenotypes PainPSMCS and PainþPSMCS (Table 5).
Deprivation Index. In recognizing the impact of preoperative narcotics usage on
outcomes following total joint arthroplasty (TJA) [30], each model
was repeated with the addition of the NarxCare Narcotic score as a
Results categorical variable in groups of 100 with prescription naive as the
reference (Supplementary Figs. 2-4). NarxCare Narcotic score >400
Of the 5,274 patients in the analysis, 16.3% (n ¼ 862) of the was significantly associated with increased likelihood of 1-year
cohort reported their state as “not acceptable” (ie, “no” to PASS) at 1 dissatisfaction, non-home discharge, and prolonged LOS.
year. Patients who achieved PASS had significantly higher scores in
all measured PROMs at 1 year, both as raw values and as deltas (P < Discussion
.001) (Supplementary Table 1). There were significant associations
among phenotypes with baseline variables (Table 1). Patients who With the shift toward value-centered healthcare systems,
presented with below median pain (“pain”) were younger (P < patient-reported satisfaction is critical in evaluating the success of
.001) and more often female (P < .001). Significant associations orthopedic surgery, and specifically TKA. The patient dissatisfaction
were observed among phenotypes and National ADI (P < .001) with rate of 16.3% at 1 year after TKA reported in the current study is
the PainPSMCS phenotype having the highest score and rep- comparable to other studies, which have reported dissatisfaction
resenting patients from the most deprived neighborhoods. rates up to 21% [31e33]. Thus the current study aimed to charac-
Multivariable regression demonstrated that compared to the terize patient profiles by preoperative PROMs associated with
reference phenotype (PainþPSþMCSþ), a phenotype which increased likelihood in reporting satisfaction after TKA. Overall, low
included low preoperative scores in VR-12 MCS alone but similar preoperative scores in knee injury and OA-specific PROMs, KOOS-
pain and function (PainþPSþMCS) significantly increased a pa- Pain (<41.7), or KOOS-PS (<51.5) alone do not significantly predict
tient’s odds of 1-year dissatisfaction (odds ratio [OR] 1.42). A dissatisfaction after TKA; however, combined preoperative low
M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120 S115

Fig. 3. Relative OR of preoperative PROM phenotype and baseline factors in predicting non-home discharge after TKA.

scores in multiple subdomains were associated with increasing dissatisfaction in a univariate analysis (P < .001); however, this
odds of dissatisfaction at 1 year. association was not observed after adjusting for other factors. Low
Dissatisfaction has been previously linked to sociodemographic preoperative scores in Oxford Knee Score function (P ¼ .014), and
patient variables; however, such variables often lose significance in 12-Item Short Form Survey MCS remained significant predictors of
a multivariable analysis, or have not been consistently replicated satisfaction [34]. Low preoperative Medical Outcomes Study Short
[10,34,35]. Younger age has been associated with increased Form-36 MCS demonstrated independent predictive ability for
dissatisfaction, with higher patient expectations and presence of patient dissatisfaction after adjustment with covariates (P < .05) in
residual symptoms in this population as potential explanations a study of 1,720 primary hip or knee patient by Gandhi et al [39].
[36,37]. Although the current study only had 44 patients in the 18- Although a greater degree of improvement in several unique
44 age category, patients in the 65þ age category had a decreased PROMs has been associated with an increased odds of 1-year
likelihood of reporting dissatisfaction at 1 year. satisfaction, other studies have not been able to find significant
Individual preoperative PROMs and the degree of improvement associations and suggested that this is potentially due to ceiling
in PROMs have been associated with dissatisfaction. In a study of effects [7,13,39]. Halawi et al [40] demonstrated a moderate cor-
1,703 primary TKA patients, Bourne et al [38] found that low scores relation between the Western Ontario and McMaster Universities
in preoperative Western Ontario and McMaster Universities Oste- Osteoarthritis Index pain subscale and satisfaction (R ¼ 0.49,
oarthritis Index components of both pain and function increased P < .001), and therefore suggested satisfaction be directly queried
the odds of 1-year dissatisfaction, by 2.4 and 2.5, respectively. In a from the patient rather than rely on postoperative PROMs. Preop-
study of 12,017 consecutive primary knee replacements, Scott et al erative PROMs were especially of interest to the current study to
[34] found preoperative pain to be significantly associated with determine the ability of PROMs to predict which candidates may
S116 M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120

Table 3 plus low MCS (<52.8) made up a smaller percentages of the cohort,
Relative Odds Ratio of Preoperative PROM Phenotype and Baseline Factors in Pre- both groups were also significantly associated with double the
dicting Non-Home Discharge After TKA.
increased odds of dissatisfaction. Finally, the combination of low
Characteristic Odds Ratio 95% CI P Value scores across all measured PROMs (PainPSMCS) did demon-
Phenotype strate the greatest likelihood of patient dissatisfaction (Fig. 2).
PainþPSþMCSþ e e The preoperative PROM phenotypes were the primary drivers of
PainþPSþMCS 1.2 0.82-1.75 .3 1-year dissatisfaction compared to the sociodemographic variables
PainþPSMCSþ 1.45 0.96-2.17 .073
(Fig. 2) and also demonstrated significant associations with pro-
PainþPSMCS 2.66 1.81-3.88 <.001
PainPSþMCSþ 0.75 0.37-1.40 .4 longed LOS and non-home discharge, although to a lesser extent
PainPSþMCS 1.81 1.02-3.09 .036 than the sociodemographic variables for those models (Figs. 3 and 4).
PainPSMCSþ 1.13 0.79-1.62 .5 Prior studies have confirmed associations between preoperative
PainPSMCS 2.34 1.75-3.15 <.001
variables and these hospital utilization parameters [42e44]. The
Age group
45-64 e e
current study highlights that patients with combinations of low
65þ 4.45 3.35-5.96 <.001 scores in pain (<41.7) or function (<51.5) with below median MCS
BMI grouped (<52.8) are of increased risk of prolonged LOS and non-home
Normal weight e e discharge. This information may be useful in preoperative planning.
Overweight 0.8 0.56-1.14 .2
The present study should be viewed in the context of its limi-
Obese class I 1.04 0.74-1.47 .8
Obese class II 0.91 0.63-1.33 .6 tations. The non-randomized nature of the present investigation
Obese class III 1.61 1.10-2.39 .016 introduces the potential for selection bias [45]. However, a large
Gender prospective cohort was utilized. Large datasets with robust follow-
Female e e up, representative of the total study population is critical [28,46].
Male 0.62 0.50-0.77 <.001
Race
The follow-up rate of the present study of 70.8% is comparable to
White e e the 1-year follow-up rate reported across the TJA population
Black 1.79 1.35-2.37 <.001 [47,48]. Although loss to follow-up can potentially skew reports of
Other 1.18 0.60-2.14 .6 satisfaction if less satisfied patients are more likely to not follow-
Smoking status
up, literature has also reported a lack of meaningful difference in
Never e e
Current 1.04 0.65-1.59 .9 the PROMs of patients who follow-up automatically vs those who
Quit <6 mo 1.44 0.81-2.42 .2 fail to respond and are contacted manually [2,14]. We also did not
Quit >6 mo 1.14 0.93-1.40 .2 utilize radiographs. Progression of OA has been associated with
Insurance status postoperative satisfaction with Leppa €nen et al [10] finding that
Commercial e e
Medicaid/Medicare 0.96 0.77-1.20 .7
patients with mild radiographic knee OA were significantly more
CCI dissatisfied (28.6%) than patients with more severe OA (8.7%) (P ¼
0-2 e e .003). The current studied relied only on the subjective measures of
3-4 1.65 1.33-2.04 <.001 current health status. Although mental status was included in the
5þ 2.87 1.82-4.43 <.001
preoperative PROM phenotypes through VR-12 MCS, actual diag-
National ADI
1-20 e e nosis of mental illness was not accounted for, and may be the
21-40 1.38 0.97-1.99 .079 subject of future investigations. The present study also found sig-
41-60 1.77 1.25-2.55 .002 nificant associations between Black race and patient satisfaction;
61-80 2.23 1.56-3.23 <.001 however, Pearson’s chi-squared test demonstrated a significant
81-100 2.5 1.69-3.74 <.001
association between race and National ADI category (c2 ¼ 1,026.1, P
Bold indicates statistically significant. < .001). Therefore, the correlation between National ADI and race
PROM, patient-reported outcome measure; TKA, total knee arthroplasty; CI, confi-
makes it difficult to determine which variable explains the change
dence interval; PS, physical function short form; MCS, Mental Component Sum-
mary; BMI, body mass index; CCI, Charlson Comorbidity Index; ADI, Area in 1-year satisfaction. Such information may then be used to further
Deprivation Index. guide patient preoperative counseling, shared decision-making,
and patient’s risk factor optimization. Through a preoperative pa-
tient phenotyping approach based on PROMs, more targeted care
not achieve benefit from TKA for early decision-making. Further- pathways may be developed in an attempt to decrease patient
more, the current study aimed to extend beyond individual PROMs dissatisfaction. Finally, in assessing dissatisfaction, only preopera-
and develop patient phenotypes of preoperative PROMs which tive variables were included in the analysis while the effect of
could indicate postoperative dissatisfaction. Although past studies postoperative major or minor complications on outcomes was not
have validated the PASS threshold with postoperative PROMs directly assessed. Indirectly, major complications which would
[4,13,41], the current study relied on a direct answer from the pa- have warranted readmission or reoperation up to 1 year were
tient at 1 year postoperative as to whether they were satisfied with assessed in an attempt to capture these metrics.
the state of their knee. Although the present study demonstrates the cumulative effect
In the present study, below median scores in pain alone or of below median PROM scores on a patient’s outcome, this study
function alone did not significantly increase the likelihood of pa- does not advocate for denying such patient’s access to surgery.
tient dissatisfaction while low scores (<52.8) in MCS alone did in- Rather, this approach may further aid in understanding which
crease the likelihood of dissatisfaction by 40%. Furthermore, factors contribute to patient dissatisfaction and poor outcomes
although the majority of the patients in this cohort reported either after TKA. Such information may then be used to further guide
all low or all high scores, patients with low scores (<52.8) only in patient preoperative counseling, shared decision-making, and pa-
MCS (PainþPSþMCS) made up 12.3% of the cohort (Table 1). tient’s risk factor optimization. Through a preoperative patient
Providers may wish to attend to this population of patients who phenotyping approach based on PROMs, more targeted care path-
report discrepancies of above average pain and function alongside ways may be developed in an attempt to decrease patient dissat-
below average mental health to consider preoperative counseling. isfaction. Although the phenotypes of this study were developed
Although combinations of low pain (<41.7) or low function (<51.5) based on the median values of preoperative PROMs for this cohort,
M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120 S117

Fig. 4. Relative OR of preoperative PROM phenotype and baseline factors in predicting prolonged LOS after TKA.

the Function and Outcomes Research in Comparative Effectiveness scores below the medians (KOOS-Pain < 41.7, KOOS-PS < 51.5, and
Research in Total Joint Arthroplasty (FORCE-TJR) [49], a national VR-12 MCS < 52.8) significantly drive dissatisfaction over de-
sample of 19,200 TJA patients from 2011 to 2012 reports the mographic values. Furthermore, this effort is novel in examining the
following median national values: 44.4 (range 36.1-50) for KOOS- dynamic between knee injury and OA-specific PROMs (KOOS-PS and
Pain and 51.5 (range 39.7-58.5) for KOOS activities of daily living KOOS-Pain) and a non-knee-specific global health measure (VR-12
or function (PS). Our designated thresholds for KOOS-Pain and MCS). We hope that this proposed development of clinically relevant
KOOS-PS fall within these national ranges and thus represent preoperative PROM phenotypes can be used as an example of ho-
normative data. The VR-12 MCS scores are normalized prior to listic summary of preoperative patient status across institutions.
reporting, with 50 representing the mean score of a non-patient Identifying patients at risk of dissatisfaction following TKA is
control group and every 10 units represent 1 standard deviation important to help set patient satisfactions and stratify risk. Fulfill-
from the mean. Therefore, we believe that the results of this study ment of preoperative expectations has been established as an
may be generalizable to institutions outside of our own. Further important factor associated with favorable patient outcomes after
external validation of the current findings will be required to surgery [50,51]. Patients identified as at-risk for dissatisfaction
determine the validity and usability of our findings. We believe the using the preoperative combined PROM phenotypes can be coun-
present findings to be the first step in explaining the inter- seled on expectations. Furthermore, some patients may benefit for
connected relationship between unique PROMs and patient a concurrent intervention, such as mental health counseling. Full
dissatisfaction. Although the cut-off values for phenotype devel- transparency and shared decision-making with patients about
opment relied on purely statistical methods, the results are clini- postoperative outcomes is of the utmost importance in the patient-
cally relevant in demonstrating the combination of baseline PROM centered, value-based healthcare model [43].
S118 M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120

Table 4

P Value
Relative Odds Ratio of Preoperative PROM Phenotype and Baseline Factors in Pre-

<.001
<.001
<.001
.165
.122
dicting Prolonged Length of Stay (3 d) After TKA.

Characteristic Odds Ratio 95% CI P Value

Phenotype

24.5%(N ¼ 1,291)
PainPSMCS
PainþPSþMCSþ e e
PainþPSþMCS 1.10 0.83-1.46 .5

293 (22.7%)
183 (14.2%)
279 (21.6%)
128 (9.9%)
57 (4.4%)
PainþPSMCSþ 1.30 0.94-1.76 .10
PainþPSMCS 1.71 1.24-2.34 <.001
PainPSþMCSþ 0.86 0.53-1.34 .5
PainPSþMCS 1.17 0.73-1.83 .5
PainPSMCSþ 1.01 0.77-1.32 >.9
PainPSMCS 1.66 1.33-2.09 <.001

(N ¼ 829)
PainPSMCSþ
Age group

(16.9%)

(14.0%)
45-64

(7.5%)

(8.1%)
(4.5%)
e e
65þ 1.99 1.62-2.44 <.001
BMI grouped

140
62
116
67
37
15.7%
Normal weight e e
Overweight 1.21 0.90-1.65 .2
Obese class I 1.35 1.01-1.84 .049
Obese class II 1.37 1.00-1.90 .052
<.001

PainPSþMCS
Obese class III 1.97 1.42-2.77

(N ¼ 164)
Gender

(11.0%)
(10.6%)
(9.3%)
(4.8%)

(5.3%)
Female e e
Male 0.66 0.55-0.78 <.001
Race

3.1%
21
11
25
24
12
White e e
Black 1.69 1.34-2.13 <.001
Other 0.79 0.42-1.37 .4

PainPSþMCSþ
Smoking status

(N ¼ 227)
Never

(21.3%)
(11.6%)
(17.7%)
e e

PROM, patient-reported outcome measure; PS, Physical function Short form; MCS, Mental Component Summary; LOS, length of stay.
(6.7%)
(3.0%)
Current 1.04 0.74-1.45 .8
Quit <6 mo 1.57 1.01-2.38 .037
Quit >6 mo 1.09 0.92-1.28 .3

35
19
29
11
5
4.3%
Insurance status
Commercial e e
Medicaid/Medicare 0.86 0.72-1.03 .10
CCI PainþPSMCS

(N ¼ 352)
0-2

(20.2%)
(15.9%)
(21.0%)
(10.2%)
e e

(6.0%)
3-4 1.48 1.24-1.76 <.001
5þ 2.32 1.55-3.43 <.001
National ADI
6.7%
71
56
74
36
21
1-20 e e
21-40 1.33 1.02-1.73 .037
41-60 1.26 0.96-1.66 .10
<.001
PainþPSMCSþ

61-80 1.79 1.36-2.37

(N ¼ 447)

81-100 2.03 1.49-2.76 <.001

(13.4%)

(16.1%)
(8.9%)

(8.3%)
(2.5%)

Bold indicates statistically significant.

PROM, patient-reported outcome measure; TKA, total knee arthroplasty; CI, confi-
60
40
72
37
11
8.5%

dence interval; PS, physical function short form; MCS, Mental Component Sum-
mary; BMI, body mass index; CCI, Charlson Comorbidity Index; ADI, Area
Deprivation Index.
PainþPSþMCS

(N ¼ 650)
(15.2%)

(14.5%)
(7.7%)

(9.2%)
(3.1%)

Conclusion
99
50
94
60
20
12.3%

Overall, low preoperative scores in KOOS-Pain or KOOS-PS alone

Rate of Each Outcome by Preoperative PROM Phenotype.

do not significantly predict dissatisfaction after TKA; however,

combined preoperative low scores in multiple subdomains (KOOS-
Pain < 41.7, KOOS-PS < 51.5, and VR-12 MCS < 52) were associated
(N ¼ 1,315)
PainþPSþMCSþ

with increasing odds of dissatisfaction at 1 year. When counseling

(10.9%)

(11.8%)
(5.9%)

(7.1%)
(3.4%)

patients for TKA, measuring pain, function, and mental health

Bold indicates statistically significant.

concurrently as phenotypes could supplement clinical decision-

24.9%
143
77
155
93
45

making in identifying patients at-risk for not achieving a satisfac-

tory outcome.
Non-home discharge

Acknowledgments
1-y Dissatisfaction

90-d Readmission
1-y Reoperation
Prolonged LOS

The authors would like to acknowledge the following for their

effort in data collection: Peter Brooks, MD (Department of Ortho-
Outcome

paedic Surgery, Cleveland Clinic, Cleveland, OH), Joseph W. George,

Table 5

MD (Department of Orthopaedic Surgery, Cleveland Clinic, Cleve-

land, OH), Juan C. Suarez, MD (Department of Orthopaedic Surgery,
M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120 S119

Cleveland Clinic, Cleveland, OH), Aldo M. Riesgo, MD, John P. initiative. Osteoarthritis Cartilage 2008;16:542e50. https://doi.org/10.1016/
J.JOCA.2007.12.014.
McLaughlin, DO (Department of Orthopaedic Surgery, Cleveland
[19] Kazis L, Miller D, Skinner K, Lee A, Ren X, Clark J, et al. Patient-reported
Clinic, Cleveland, OH), George F. Muschler, MD, Lukas M. Nystrom, measures of health: the veterans health study. J Ambul Care Manage 2004;27:
MD, Brendan M. Patterson, MD, Atul F. Kamath, MD, Michael A. 70e83. https://doi.org/10.1097/00004479-200401000-00012.
Mont, MD, Michael W. Kolczun, MD, David H. Krahe, MD, Frank M. [20] Kind AJH, Buckingham WR. Making neighborhood-disadvantage metrics
accessible d the neighborhood Atlas. N Engl J Med 2018;378:2456e8. https://
Sabo, MD, Richard R. Masin, MD, Steven A. Lietman, MD, Patrick E. doi.org/10.1056/NEJMp1802313.
Sziraky, MD, Michael J. Joyce, MD, Joseph B. Scarcella, MD, James T. [21] Collins NJ, Roos EM. Patient-reported outcomes for total hip and knee
Rosneck, MD, Victor A. Nemeth, MD, Andrew J. Matko, MD, Alfred arthroplasty: commonly used instruments and attributes of a “good” measure.
Clin Geriatr Med 2012;28:367e94. https://doi.org/10.1016/j.cger.2012.05.007.
Serna, MD, Brett W. McCoy, MD, Richard D. Parker, MD, and Bradley [22] Lyman S, Lee Y-Y, McLawhorn AS, Islam W, MacLean CH. What are the min-
A. Pierce, MD. imal and substantial improvements in the HOOS and KOOS and JR versions
after total joint replacement? Clin Orthop Relat Res 2018;476:2432e41.
https://doi.org/10.1097/CORR.0000000000000456.
[23] Tubach F. Evaluation of clinically relevant states in patient reported outcomes
References in knee and hip osteoarthritis: the patient acceptable symptom state. Ann
Rheum Dis 2005;64:34e7. https://doi.org/10.1136/ard.2004.023028.
[1] Squitieri L, Bozic KJ, Pusic AL. The role of patient-reported outcome measures [24] Muller B, Yabroudi MA, Lynch A, Lai C-L, van Dijk CN, Fu FH, et al. Defining
in value-based payment reform. Value Health 2017;20:834e6. https://doi.org/ thresholds for the patient acceptable symptom state for the IKDC subjective
10.1016/j.jval.2017.02.003. knee form and KOOS for patients who underwent ACL reconstruction. Am J
[2] Orr MN, Klika AK, Gagnier JJ, Bhandari M, Piuzzi NS. A call for a Standardized Sports Med 2016;44:2820e6. https://doi.org/10.1177/0363546516652888.
approach to reporting patient-reported outcome measures: clinical relevance [25] Maksymowych WP, Richardson R, Mallon C, van der Heijde D, Boonen A.
ratio. J Bone Joint Surg Am 2021;103:e91. https://doi.org/10.2106/ Evaluation and validation of the patient acceptable symptom state (PASS) in
JBJS.21.00030. patients with ankylosing spondylitis. Arthritis Rheum 2007;57:133e9.
[3] Emara AK, Klika AK, Piuzzi NS. Evidence-based orthopedic surgerydfrom https://doi.org/10.1002/art.22469.
synthesis to practice. JAMA Surg 2020;155:1009. https://doi.org/10.1001/ [26] Collins NJ, Misra D, Felson DT, Crossley KM, Roos EM. Measures of knee
jamasurg.2020.1521. function: international knee documentation committee (IKDC) subjective
[4] Florissi I, Galea V, Shin D, Sauder N, Colon Iban YE, Ingelsrud LH, et al. External knee evaluation form, knee injury and osteoarthritis outcome score (KOOS),
validation of achieving the patient Acceptable symptom state for the EuroQol- knee injury and osteoarthritis outcome score physical function short form
5 dimension 1 year after total hip arthroplasty. J Bone Joint Surg 2021;103:e5. (KOOS-PS), Knee Ou. Arthritis Care Res (Hoboken) 2011;63:S208e28. https://
https://doi.org/10.2106/JBJS.20.00184. doi.org/10.1002/acr.20632.
[5] Khow YZ, Liow MHL, Goh GS, Chen JY, Lo NN, Yeo SJ. The oxford knee score [27] Schalet BD, Rothrock NE, Hays RD, Kazis LE, Cook KF, Rutsohn JP, et al. Linking
minimal clinically important difference for revision total knee arthroplasty. physical and mental health summary scores from the veterans RAND 12-item
Knee 2021;32:211e7. https://doi.org/10.1016/j.knee.2021.08.020. health survey (VR-12) to the PROMIS® global health scale. J Gen Intern Med
[6] Connelly JW, Galea VP, Rojanasopondist P, Matuszak SJ, Ingelsrud LH, 2015;30:1524e30. https://doi.org/10.1007/s11606-015-3453-9.
Nielsen CS, et al. Patient acceptable symptom state at 1 and 3 years after total [28] Wilson I, Bohm E, Lübbeke A, Lyman S, Overgaard S, Rolfson O, et al. Ortho-
knee arthroplasty: thresholds for the knee injury and osteoarthritis outcome paedic registries with patient-reported outcome measures. EFORT Open Rev
score (KOOS). J Bone Joint Surg Am 2019. https://doi.org/10.2106/ 2019;4:357e67. https://doi.org/10.1302/2058-5241.4.180080.
JBJS.18.00233. [29] Goh GS, Liow MHL, Chen JY, Tay DK-J, Lo N-N, Yeo S-J. Do patients with
[7] Galea VP, Ingelsrud LH, Florissi I, Shin D, Bragdon CR, Malchau H, et al. Patient- psychological distress have poorer patient-reported outcomes after total hip
acceptable symptom state for the Oxford Hip Score and Forgotten Joint Score arthroplasty? J Arthroplasty 2020;35:2465e71. https://doi.org/10.1016/
at 3 months, 1 year, and 2 years following total hip arthroplasty: a registry- j.arth.2020.04.077.
based study of 597 cases. Acta Orthop 2020;91:372e7. https://doi.org/ [30] Emara AK, Grits D, Klika AK, Molloy RM, Krebs VE, Barsoum WK, et al.
10.1080/17453674.2020.1750877. NarxCare scores greater than 300 are associated with adverse outcomes after
[8] Murray DW, MacLennan GS, Breeman S, Dakin HA, Johnston L, Campbell MK, primary THA. Clin Orthop Relat Res 2021;479:1957e67. https://doi.org/
et al. A randomised controlled trial of the clinical effectiveness and cost- 10.1097/CORR.0000000000001745.
effectiveness of different knee prostheses: the Knee Arthroplasty Trial (KAT). [31] Robertsson O, Dunbar M, Pehrsson T, Knutson K, Lidgren L. Patient satisfaction
Health Technol Assess 2014;18:1e235. https://doi.org/10.3310/hta18190. after knee arthroplasty: a report on 27,372 knees operated on between 1981
[9] Konopka JF, Lee Y-Y, Su EP, McLawhorn AS. Quality-adjusted life years after and 1995 in Sweden. Acta Orthop Scand 2000;71:262e7. https://doi.org/
hip and knee arthroplasty: health-related quality of life after 12,782 joint 10.1080/000164700317411852.
replacements. JB & JS Open Access 2018;3:e0007. https://doi.org/10.2106/ [32] Noble PC, Conditt MA, Cook KF, Mathis KB. The John Insall Award:
JBJS.OA.18.00007. patient expectations affect satisfaction with total knee arthroplasty. Clin
[10] Leppa €nen S, Niemela €inen M, Huhtala H, Eskelinen A. Mild knee osteoarthritis Orthop Relat Res 2006;452:35e43. https://doi.org/10.1097/01.blo.000023
predicts dissatisfaction after total knee arthroplasty: a prospective study of 8825.63648.1e.
186 patients aged 65 years or less with 2-year follow-up. BMC Musculoskelet [33] Baker PN, van der Meulen JH, Lewsey J, Gregg PJ. The role of pain and function
Disord 2021;22:657. https://doi.org/10.1186/s12891-021-04543-8. in determining patient satisfaction after total knee replacement: data from
[11] Cella D, Hahn E, Jensen S, Butt Z, Nowinski C, Rothrock N, et al. Patient-re- the national joint registry for England and Wales. J Bone Joint Surg Br
ported outcomes in performance measurement. RTI Press; 2015. https:// 2007;89-B:893e900. https://doi.org/10.1302/0301-620X.89B7.19091.
doi.org/10.3768/rtipress.2015.bk.0014.1509. [34] Scott CEH, Howie CR, MacDonald D, Biant LC. Predicting dissatisfaction
[12] Longo UG, De Salvatore S, Candela V, Berton A, Casciaro C, Sciotti G, et al. following total knee replacement: a prospective study OF 1217 patients.
Unicompartmental knee arthroplasty: minimal important difference and pa- J Bone Joint Surg Br 2010;92-B:1253e8. https://doi.org/10.1302/0301-
tient acceptable symptom state for the forgotten joint score. Medicina 620X.92B9.24394.
2021;57:324. https://doi.org/10.3390/medicina57040324. [35] Ethgen O, Bruye re O, Richy F, Dardennes C, Reginster J-Y. Health-related
[13] Kvien TK, Heiberg T, Hagen KB. Minimal clinically important improvement/ quality of life in total hip and total knee arthroplasty: a qualitative and sys-
difference (MCII/MCID) and patient acceptable symptom state (PASS): what tematic review of the literature. J Bone Joint Surg 2004;86:963e74. https://
do these concepts mean? Ann Rheum Dis 2007;66:iii40e1. https://doi.org/ doi.org/10.2106/00004623-200405000-00012.
10.1136/ard.2007.079798. [36] Parvizi J, Nunley RM, Berend KR, Lombardi AV, Ruh EL, Clohisy JC, et al. High
[14] Orthopaedics OCC. Value in research: achieving validated outcome mea- level of residual symptoms in young patients after total knee arthroplasty.
surements while mitigating follow-up cost. J Bone Joint Surg Am 2020;102: Clin Orthop Relat Res 2014;472:133e7. https://doi.org/10.1007/s11999-013-
419e27. https://doi.org/10.2106/JBJS.19.00531. 3229-7.
[15] Piuzzi N, Strnad G, Brooks P, Hettrich C, Higuera-Rueda C, Iannotti J, et al. [37] Scott CEH, Bugler KE, Clement ND, MacDonald D, Howie CR, Biant LC. Patient
Implementing a scientifically valid, cost-effective, and scalable data collection expectations of arthroplasty of the hip and knee. J Bone Joint Surg Br 2012;94-
system at point of care: the Cleveland Clinic OME cohort. J Bone Joint Surg Am B:974e81. https://doi.org/10.1302/0301-620X.94B7.28219.
2019;101:458e64. https://doi.org/10.2106/JBJS.18.00767. [38] Bourne RB, Chesworth BM, Davis AM, Mahomed NN, Charron KDJ. Patient
[16] Roos E, Roos H, Lohmander L, Ekdahl C, Beynnon B. Knee injury and osteo- satisfaction after total knee arthroplasty: who is satisfied and who is not? Clin
arthritis outcome score (KOOS)–development of a self-administered outcome Orthop Relat Res 2010;468:57e63. https://doi.org/10.1007/s11999-009-
measure. J Orthop Sports Phys Ther 1998;28:88e96. https://doi.org/10.2519/ 1119-9.
JOSPT.1998.28.2.88. [39] Gandhi R, Davey JR, Mahomed NN. Predicting patient dissatisfaction following
[17] Roos EM, Lohmander LS. The knee injury and osteoarthritis outcome score joint replacement surgery. J Rheumatol 2008;35:2415e8. https://doi.org/
(KOOS): from joint injury to osteoarthritis. Health Qual Life Outcomes 2003;1: 10.3899/jrheum.080295.
64. https://doi.org/10.1186/1477-7525-1-64. [40] Halawi MJ, Jongbloed W, Baron S, Savoy L, Cote MP, Lieberman JR. Patient-
[18] Perruccio A, Stefan Lohmander L, Canizares M, Tennant A, Hawker G, reported outcome measures are not a valid proxy for patient satisfaction in
Conaghan P, et al. The development of a short measure of physical function for total joint arthroplasty. J Arthroplasty 2020;35:335e9. https://doi.org/
knee OA KOOS-Physical Function Shortform (KOOS-PS) - an OARSI/OMERACT 10.1016/j.arth.2019.09.033.
S120 M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120

[41] Paulsen A, Roos EM, Pedersen AB, Overgaard S. Minimal clinically important registries: multinational survey and recommendations. Clin Orthop Relat Res
improvement (MCII) and patient-acceptable symptom state (PASS) in total hip 2021;479:2151e66. https://doi.org/10.1097/CORR.0000000000001852.
arthroplasty (THA) patients 1 year postoperatively. Acta Orthop 2014;85: [47] Canfield M, Savoy L, Cote MP, Halawi MJ. Patient-reported outcome mea-
39e48. https://doi.org/10.3109/17453674.2013.867782. sures in total joint arthroplasty: defining the optimal collection window.
[42] Cleveland Clinic OME Arthroplasty Group, Arnold N, Anis H, Barsoum WK, Arthroplasty Today 2020;6:62e7. https://doi.org/10.1016/j.artd.2019.
Bloomfield MR, Brooks PJ, et al. Preoperative cut-off values for body mass 10.003.
index deny patients clinically significant improvements in patient-reported [48] Zheng H, Li W, Harrold L, Ayers DC, Franklin PD. Web-based comparative
outcomes after total hip arthroplasty. Bone Joint J 2020;102-B:683e92. patient-reported outcome feedback to support quality improvement and
https://doi.org/10.1302/0301-620X.102B6.BJJ-2019-1644.R1. comparative effectiveness research in total joint replacement. EGEMs 2014;2:
[43] Anis HK, Strnad GJ, Klika AK, Zajichek A, Spindler KP, Barsoum WK, et al. 21. https://doi.org/10.13063/2327-9214.1130.
Developing a personalized outcome prediction tool for knee arthroplasty. [49] Ayers DC, Li W, Harrold L, Allison J, Franklin PD. Preoperative pain and
Bone Joint J 2020;102-B:1183e93. https://doi.org/10.1302/0301- function profiles reflect consistent TKA patient selection among US surgeons.
620X.102B9.BJJ-2019-1642.R1. Clin Orthop Relat Res 2015;473:76e81. https://doi.org/10.1007/s11999-014-
[44] Campbell A, Emara AK, Klika A, Piuzzi NS, The Cleveland Clinic OME Arthro- 3716-5.
plasty Group. Does implant selection affect patient-reported outcome mea- [50] Lingard EA, Sledge CB, Learmonth ID. Patient expectations regarding total
sures after primary total hip arthroplasty? J Bone Joint Surg 2021;103: knee arthroplasty: differences among the United States, United Kingdom, and
2306e17. https://doi.org/10.2106/JBJS.20.01931. Australia. J Bone Joint Surg 2006;88:1201e7. https://doi.org/10.2106/
[45] Nohr EA, Liew Z. How to investigate and adjust for selection bias in cohort JBJS.E.00147.
studies. Acta Obstet Gynecol Scand 2018;97:407e16. https://doi.org/10.1111/ [51] Kim MS, Koh IJ, Choi KY, Ju GI, In Y. Centrally sensitized patients undergoing
aogs.13319. total knee arthroplasty have higher expectations than do non-centrally
[46] Bohm ER, Kirby S, Trepman E, Hallstrom BR, Rolfson O, Wilkinson JM, et al. sensitized patients. Knee Surg Sports Traumatol Arthrosc 2021. https://
Collection and reporting of patient-reported outcome measures in arthroplasty doi.org/10.1007/s00167-021-06561-4 [Epub ahead of print].
M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120 S120.e1

Supplementary Fig. 1. Visual guide of example patients of preoperative PROM phenotypes.

S120.e2 M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120

Supplementary Fig. 2. Relative OR of preoperative PROM phenotype and baseline factors þ NarxCare Narcotic score in predicting 1 year dissatisfaction after TKA.
M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120 S120.e3

Supplementary Fig. 3. Relative OR of preoperative PROM phenotype and baseline factors þ NarxCare Narcotic score in predicting non-home discharge after TKA.
S120.e4 M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120

Supplementary Fig. 4. Relative OR of preoperative PROM phenotype and baseline factors þ NarxCare Narcotic score in predicting prolonged LOS after TKA. LOS, length of stay.
M.N. Orr et al. / The Journal of Arthroplasty 37 (2022) S110eS120 S120.e5

Supplementary Table 1
The 1 y PROMs Stratified by Achievement of PASS at 1 y.

Did Not Achieve PASS (N ¼ 845) Achieved PASS (N ¼ 4,335) P Value

Year 1 KOOS-Pain <.001

Mean (SD) 58.51 (19.21) 86.99 (13.17)
Delta KOOS-Pain <.001
Mean (SD) 20.46 (20.57) 44.92 (18.11)
Year 1 KOOS-PS <.001
Mean (SD) 64.84 (33.77) 84.86 (29.36)
Delta KOOS-PS <.001
Mean (SD) 19.75 (36.14) 35.57 (32.40)
Year 1 VR-12 MCS <.001
Mean (SD) 48.24 (12.66) 54.69 (8.98)
Delta VR-12 MCS <.001
Mean (SD) 0.12 (11.87) 2.67 (10.95)

Delta: 1 y score minus score at baseline.

PROM, patient-reported outcome measure; PASS, Patient Acceptable Symptom State; KOOS, Knee Disability and Osteoarthritis Outcome Score; SD, standard deviation; PS,
physical function short form; VR-12, Veterans RAND 12-Item Health Survey; MCS, Mental Component Summary.

Supplementary Table 2
Performance Metrics Corresponding to Models (Figs. 2-4).

AUC

1 y dissatisfaction 0.63
Non-home discharge 0.76
Prolonged LOS 0.68

AUC, area under the curve; LOS, length of stay.