Journal Logo

Scientific Articles

Patient Preferences for Surgical Treatment of Knee Osteoarthritis

A Discrete-Choice Experiment Evaluating Total and Unicompartmental Knee Arthroplasty

Hutyra, Carolyn A. MMCi1,a; Gonzalez, Juan Marcos PhD2; Yang, Jui-Chen MEM3; Johnson, F. Reed PhD2; Reed, Shelby D. PhD2; Amendola, Annunziato MD1; Bolognesi, Michael P. MD1; Berend, Keith R. MD4,5,6; Berend, Michael E. MD7; MacDonald, Steven J. MD8; Mather, Richard C. III MD, MBA1

Author Information
The Journal of Bone and Joint Surgery: December 2, 2020 - Volume 102 - Issue 23 - p 2022-2031
doi: 10.2106/JBJS.20.00132
  • Open
  • Supplementary Content
  • Disclosures
  • Infographic
  • Video Summary

Abstract

Total knee arthroplasty (TKA) and unicompartmental knee arthroplasty (UKA) are 2 common and successful treatments for knee osteoarthritis1-4. Revision rates for TKA have fallen to approximately 5% at 10 years, with good functional outcomes5,6, but 17% of patients remain dissatisfied with the results of the procedure7. Furthermore, in as many as 47% of patients, osteoarthritis is restricted to 1 tibiofemoral compartment8. UKA is an alternative to TKA in patients who retain cruciate ligaments and preserved compartments and is more prevalent in a younger population compared with TKA6,9.

Outcomes following UKA and TKA are well documented in multiple randomized controlled trials and international registries3,6,9. UKA is associated with lower complication rates but higher revision rates4-6,10. Functional and patient-reported outcomes appear to favor UKA, particularly in data from the National Joint Registry for England, Wales, Northern Ireland and the Isle of Man4,9,11,12.

Because patients value specific outcomes, complications, and downstream events of surgical interventions differently, those values or preferences may impact the ideal treatment choice13. Stated-preference data are essential to delivering patient-centered care and establishing evidence regarding the value that patients with knee osteoarthritis place on processes and outcomes from arthroplasty14. The purpose of this study was to evaluate patient preferences for UKA and TKA by quantifying utility weights and the relative importance of treatment outcomes. Stated choice of TKA or UKA was examined to determine patient willingness to undergo either procedure.

Materials and Methods

A discrete-choice experiment was developed following good-practice standards for discrete-choice research by stated-preference researchers with >70 years of combined expertise13,15-20. Discrete-choice experiment are one of several experimental options used to elicit patient preferences for features of treatment21. This methodology was selected to quantify benefit-risk tradeoffs between cohorts for given clinical alternatives consisting of attributes and levels.

Respondents reviewed arthroplasty information and answered demographic questions. The Oxford Knee Score was used to evaluate baseline knee function before respondents completed the discrete-choice experiment22. The Oxford Knee Score is a licensed, validated 12-item patient-reported outcome measure scored from 0 to 48, with 0 representing the worst outcome and 48 representing the best. The Oxford Knee Score test was administered electronically, and patients were categorized according to functional status on the basis of the Kalairajah redefined scale: excellent (>41), good (34 to 41), fair (27 to 33), or poor (<27)22-24.

Respondents with excellent functional scores were directed to the end of the survey as unlikely surgical candidates based on surgeon recommendation25,26. Remaining respondents with good Oxford Knee Scores were directed to a discrete-choice experiment that evaluated preferences for improvements in knee function beyond good function (good-function cohort), and those with fair or poor scores were grouped together to receive a discrete-choice experiment evaluating preferences for improvements beyond fair function (fair/poor-function cohort). The only variance between surveys was the inclusion of 1 additional functional attribute level in the fair/poor-function cohort.

The basis of a discrete-choice experiment is selection of attributes that represent general categories under which treatments can be evaluated and of levels that convey how a treatment satisfies an attribute27. Attributes and levels are used to generate experimentally designed profiles that require tradeoffs27. The rate at which patients state they are willing to accept such tradeoffs indicates the relative importance of the attributes.

Attributes and levels (Table I) were selected through an extensive, iterative process involving literature review3-6,8,9,11,12,24,28-75, review from stated-preference experts, and input from fellowship-trained orthopaedic surgeons with varying arthroplasty views—ranging from strong support of TKA to strong support of UKA—to ensure consideration of all factors. To further assure objectivity, clinical information was compiled by an independent reviewer. Expert physicians (n = 5), including an independent orthopaedic surgeon not involved in the treatment of the study population, reviewed all clinical information for appropriateness and objectivity. Face-to-face interviews were conducted with 7 patients with knee osteoarthritis following a think-aloud protocol. Survey length, relevant attribute levels, and understanding of attribute definitions were iteratively evaluated. Additional testing occurred via an online survey marketplace to ensure appropriate attribute graphics and phrasing. Stakeholders evaluated survey length, validated experimental design efficiency, modified information delivery, and improved study design. Attributes most strongly supported by the evidence were selected: (1) chance of complications within 3 months of the surgical procedure, (2) functional ability following recovery, (3) awareness of the knee implant following recovery, and (4) chance of needing another operation within 10 years to replace the implant. Attribute levels were constructed to account for the range of variation within the knee-osteoarthritis population, from healthy individuals to those with comorbidities.

TABLE I - Attributes and Levels for the Discrete-Choice Experiment Trade-Off Questions
Attribute Levels
Chance of complication 1 out of 100 people (1%) will have a serious complication within 3 months of operation
4 out of 100 people (4%) will have a serious complication within 3 months of operation
11 out of 100 people (11%) will have a serious complication within 3 months of operation
Functional ability* You are able to lead an active lifestyle after recovery (Super Excellent)
You are able to do all daily activities after recovery (Excellent)
You are able to do daily activities with some limitations after recovery (Good)
Normality: awareness of knee implant You are never aware of your knee implant after recovery
You are seldom aware of your knee implant after recovery
You are mostly aware of your knee implant after recovery
Chance of needing another operation 3 out of 100 people (3%) will need another operation within 10 years to replace their implant
12 out of 100 people (12%) will need another operation within 10 years to replace their implant
19 out of 100 people (19%) will need another operation within 10 years to replace their implant
*Respondents who scored Fair/Poor function on the Oxford Knee Score were shown the Good level of the functional ability attribute. Respondents who scored Good function on the Oxford Knee Score were filtered to a separate study design with 2 functional attribute levels that did not include Good function.

Respondents in the good-function cohort received a survey with 2 functional levels indicative of “Super Excellent” and “Excellent” function. “Super Excellent” was created to gauge patient interest in the creation of highly durable implants that would allow participation in high-impact sports, an activity currently not recommended with marketplace implants. “Excellent” was created according to the functional abilities of someone scoring excellent on the Oxford Knee Score, as verified through patient interviews and physician panel discussions. Respondents in the fair/poor-function cohort received a survey with 3 functional levels. The third level corresponded to good function according to the Oxford Knee Score. Level selection assumed that surgical intervention would result in some functional improvement.

Experimental Design

Respondents completed a fixed-attention check, 8 random discrete-choice experiment trade-off questions, and 2 fixed final trade-off questions. Each random trade-off was followed by an opt-out in which respondents could choose to remain in their current functional state (Fig. 1). Lighthouse Studio (version 9.5.3; Sawtooth Software) generated 300 questionnaires that were randomly assigned to ensure adequate combinations and variations. The orthogonal discrete-choice experiment featured connectivity and positional balance to reduce order bias and did not include attribute randomization, concept sorting, or attribute prohibitions.

Fig. 1
Fig. 1:
Sample choice question from the survey.

Sample and Recruitment

The instrument was released at a large academic medical center to orthopaedic patients 40 to 80 years old with a knee osteoarthritis diagnosis verified through International Classification of Diseases, 10th Revision (ICD-10) coding (M17.9) and supported by corresponding Oxford Knee Scores76. Enrollment occurred through clinic visits and emailed survey invitations. The study received institutional approval and was registered with ClinicalTrials.gov (NCT03058380).

Statistical Analysis

Response quality was addressed by identifying respondents who straightlined or always selected Operation A or always selected Operation B in the discrete-choice experiment because such responses are considered a strong indicator of a lack of respondent attention77. Statistical comparison of respondent characteristics in the cohorts was conducted. P values were computed with use of the Fisher exact test and Student t test (2-tailed p values). Chi-square 2-way contingency tables tested for independence of variables. Attribute dominance—a classification of respondents unwilling to trade off among attributes, and thereby choosing a specific attribute with the better level every time it is displayed within a choice question—was examined between cohorts. Individuals with extended (i.e., >1 hour) survey completion times were considered outliers and removed only from the time analysis.

Random trade-off question data were analyzed following good-practice standards78. We used a random-parameters logit model in StataSE (version 15; StataCorp) to separate population-level preference estimates for both cohorts. Respondent treatment choices were estimated as a function of attribute levels. All attributes were effect-coded. With effect coding, 0 indicates the mean effect of all attributes as opposed to the effect of all omitted levels in dummy coding, and the coefficient for the omitted level is the negative sum of coefficients for all other levels within an attribute79. Discrete-choice experiment results are preference weights or log-odds indicating average preference for levels in each attribute. These weights are not directly interpretable. Only differences between preference weights in an attribute are meaningful. Attribute importance weights were calculated as the greatest change in preference weights that can be achieved with levels for each attribute.

We normalized preference weight estimates to facilitate comparisons of preferences across cohorts. The difference between the largest and smallest coefficients for the chance of serious complications was calculated. Attribute importance was normalized to 10. Remaining preference weights were rescaled to preserve relative attribute importance within cohorts. Rescaled preference weights represented respondent preferences for each attribute level as an equivalence to the chance of serious complications and can be compared across cohorts.

Confidence intervals (CIs) around rescaled preference and importance weights were estimated with use of the Krinsky and Robb procedure with 10,000 draws and rescaled accordingly80. If CIs between levels of a single attribute do not overlap, preference weights are statistically different from each other at better than the 5% level. Similarly, if CIs between attribute importance measures do not overlap, importance weights are statistically different from each other at better than the 5% level.

Two final-choice questions—reflecting actual attribute rates for UKA and TKA—examined respondent choices between operations (see Appendix). The first question displayed attributes and levels corresponding with the discrete-choice experiment design. The second included a modified functional attribute presented as a probabilistic outcome aligning with values in the literature and using person diagrams with rates for the ability to perform daily activities following recovery. Respondents were blinded to intervention type. Analysis of the final choice was completed with use of stated-choice probabilities and coefficients from random-parameters logit models to calculate preference weights for probabilistic function levels (TKA, 36%; UKA, 46%). We leveraged a probability-weighting function81 proposed by cognitive psychologists and empirically validated as a way to uncover the impact of uncertain outcomes. The function included 2 parameters ([δ],[γ]) that defined risk attitudes (i.e., patients are risk-averse, risk-neutral, or-risk tolerant) and discriminability (i.e., whether patients are able to discriminate differences in risk). Based on utility scores and the 2 final questions, predicted-choice probabilities for UKA and TKA were calculated for both cohorts.

Results

Three hundred and forty-eight respondents met the inclusion criteria and were enrolled, of whom 300 completed the survey. In addition to the 48 participants who did not complete at least 1 discrete-choice experiment question, 42 participants were excluded from analysis for scoring “Excellent” on the Oxford Knee Score (n = 32) or for always choosing Operation A or B on the random trade-off questions (n = 10). The remaining 258 surveys were analyzed (Fig. 2). No exclusions were made on the basis of attribute dominance because no differences were found between cohorts for any attribute.

Fig. 2
Fig. 2:
Map of survey enrollment process. OKS = Oxford Knee Score.

In total, a pivot design was utilized to filter 72 respondents into the good-function cohort and 186 into the fair/poor-function cohort. Respondents in the good-function cohort were significantly more likely to be married (p < 0.001) and have a postgraduate degree (p < 0.001) compared with those in the fair/poor-function cohort (Table II). These respondents were also more likely to report a desire to return to impact sports (p = 0.004), whereas those in the fair/poor-function cohort reported being more focused on activities of daily living (p = 0.004).

TABLE II - Patient Characteristics and Information on Knee Status*
Respondent Characteristics Whole Sample (N = 258) Good-Function Cohort (N = 72) Fair/Poor-Function Cohort (N = 186) P Value
Age(yr) 65.2 ± 8.4 64.3 ± 7.6 65.4 ± 8.7 0.347
Sex
 Male 35.9% (92) 37.5% (27) 35.3% (65) 0.746
 Female 64.1% (164) 62.5% (45) 64.7% (119) 0.746
Race
 Black or African American 9.9% (25) 4.2% (3) 12.2% (22) 0.061
 White or Caucasian 86.9% (219) 91.5% (65) 85.1% (154) 0.117
 Other 1.2% (8) 2.8% (3) 2.8% (5) 0.770
Marital status
 Single, never married 6.0% (15) 4.3% (3) 6.6% (12) 0.484
 Married/cohabitating 70.5% (177) 84.3% (59) 65.2% (118) <0.001
 Divorced/separated 17.9% (45) 10.0% (7) 21.0% (38) 0.021
 Widowed 5.6% (14) 1.4% (1) 7.2% (13) 0.075
Education level
 <4 years of college 19.5% (75) 11.4% (8) 22.7% (67) <0.001
 Bachelor’s degree 30.7% (77) 31.4% (22) 30.4% (55) 0.873
 Postgraduate degree 39.4% (99) 57.1% (40) 32.6% (59) <0.001
Employment status
 Full time (≥30 hours) 29.1% (73) 40.0% (28) 24.9% (45) 0.026
 Part time (<30 hours) 8.8% (22) 10.0% (7) 8.3% (15) 0.669
 Not employed 6.0% (15) 4.3% (3) 6.6% (12) 0.484
 Retired 56.2% (141) 45.7% (32) 60.2% (109) 0.041
Functional goals for treatment
 High-impact sports 15.4% (39) 28.2% (20) 11.0% (19) 0.004
 Low-impact sports 59.3% (150) 70.4% (50) 57.8% (100) 0.066
 Activities of daily living 57.3% (145) 45.1% (32) 65.3% (113) 0.004
Length of time with knee pain
 <1 year 4.1% (10) 4.2% (3) 4.0% (7) 0.949
 1-3 years 12.3% (40) 14.1% (10) 11.6% (20) 0.587
 3-5 years 16.4% (40) 16.9% (12) 16.2% (28) 0.891
 5-10 years 26.6% (65) 31.0% (22) 24.9% (43) 0.327
 >10 years 40.6% (99) 33.8% (24) 43.4% (75) 0.169
*Data are missing for some categories.
Values are given as the mean and standard deviation.
Values are given as the percentage of patients, with the count in parentheses.

Preference Weights

Estimated preference weights aligned with the natural ordering of the levels for each attribute; that is, better clinical outcomes had larger preference weights, excluding the super-excellent functional level, in the fair/poor-function cohort (Fig. 3). Within the good-function cohort, there were significant differences between preference weights for several attributes, including complications and revision. Within complication rates, there was a difference between the 4% and 11% levels (p = 0.05). Significant differences were found between all levels of revision rates. Within the fair/poor-function cohort, differences were found between preference weights for all 4 attributes. All levels of the complication and revision attributes showed significant differences. Differences were present between excellent and good (p < 0.001) and super-excellent and good (p < 0.001) functional levels and between “seldom aware” and “mostly aware” (p < 0.001) and “never aware” and “mostly aware” (p < 0.001) levels of the normality attribute. Comparison of preference weights between cohorts revealed no differences for any attribute.

Fig. 3
Fig. 3:
Mean preference weights for both cohorts, with vertical bars indicating the 95% CI.

Overall Attribute Importance

Overall attribute importance indicated that adverse events were most important for both cohorts (Fig. 4). Respondents with good, poor, or fair function valued serious complications and revision rates equally. Respondents in the good-function cohort valued the functional attribute the least (attribute importance, 6.7; 95% CI, 0.6 to 26.9), whereas those with worse function in the fair/poor-function cohort were least concerned with normality (attribute importance, 18.1; 95% CI, 13.9 to 21.9). Within the fair/poor-function cohort, there were differences in attribute importance for the following comparisons: (1) complications (32.6; 95% CI, 28.1 to 37.2) and function (21.2; 95% CI, 17.3 to 25.4; p = 0.001), (2) complications (32.6; 95% CI, 28.1 to 37.2) and normality (18.1; 95% CI, 13.9 to 21.9; p < 0.001), (3) function (21.2; 95% CI, 17.3 to 25.4) and revision (28.1; 95% CI, 24.4 to 31.7; p = 0.018), and (4) normality (18.1; 95% CI, 13.9 to 21.9) and revision (28.1; 95% CI, 24.4 to 31.7; p = 0.001).

Fig. 4
Fig. 4:
Relative attribute importance weights for both the good-function and fair/poor-function cohorts, with vertical bars indicating the 95% CI. *Indicates significant difference between cohorts.

Within the good-function cohort, there were differences in attribute importance for the following comparisons: (1) complications (34.4; 95% CI, 13.2 to 40.7) and function (6.7; 95% CI, 0.6 to 26.9; p = 0.003), (2) function (6.7; 95% CI, 0.6 to 26.9) and normality (24.6; 95% CI, 8.6 to 29.9; p = 0.032), and (3) function (6.7; 95% CI, 0.6 to 26.9) and revision (34.3; 95% CI, 24.2 to 57.0; p = 0.013).

Attribute importance comparisons between cohorts revealed that the importance of function was significantly higher for the fair/poor-function cohort compared with the good-function cohort (p = 0.03).

Final Choice

The 2 final-choice questions yielded slightly different results (Table III). Given the assumed functional form for the value of outcomes under uncertainty and known preferences for outcomes elicited through the main discrete-choice experiment, preference weights were 0.23 for 36% and 0.54 for 46% in the good-function cohort and 0.47 for 36% and 0.59 for 46% in the fair/poor-function cohort. Based on these utility scores and combined final 2 choice questions, predicted-choice probabilities for TKA and UKA were 42% and 58%, respectively, in the good-function cohort and 54% and 46%, respectively, in the fair/poor-function cohort.

TABLE III - Results from Final-Choice Questions and Combined Results of the Final Treatment Choice Question Between TKA and UKA*
Final Treatment Choice Good-Function Cohort (N = 70) Fair/Poor-Function Cohort (N = 182)
Final-choice question #1
 TKA 31 (44.3%) 84 (46.2%)
 UKA 39 (55.7%) 98 (53.8%)
Final-choice question #2
 TKA 18 (25.7%) 61 (33.5%)
 UKA 52 (74.3%) 121 (66.5%)
Final predicted-choice probabilities
 TKA 41.6% 53.7%
 UKA 58.4% 46.3%
*Values are given as the number of patients, with the percentage in parentheses.

Discussion

Although various methodologies have been employed to elicit patient preferences regarding TKA, willingness for a surgical procedure, and costs, these studies have not specifically addressed the question of TKA versus UKA and the associated tradeoffs of each82-85. To our knowledge, the present study is the first to directly contrast these 2 options in a clinical population and evaluate preferences on the basis of functional outcome scores.

Overall, the findings of the present study indicate that patients with knee osteoarthritis place the greatest value on adverse events such as complications and revision when considering treatment. Unsurprisingly, patients with better functional status value function following the surgical procedure less than those who are functionally worse off (fair/poor-function cohort). Those with good function are also more concerned with the ability to participate in sports compared with those with poor or fair function, who are more focused on activities of daily living. These findings suggest that there is a population of individuals with good function for whom a surgical procedure is appropriate and aligned with patient preferences. There was also minimal difference in value between achieving excellent or super-excellent function, indicating that an extremely durable implant for participation in high-impact sports is not necessarily a better option compared with the current marketplace implants.

The results of this study indicated that those with good functional status tended to prefer UKA, whereas those with worse function preferred TKA. The choices between procedures, however, were split nearly evenly (good-function cohort: TKA, 42%, and UKA, 58%; fair/poor-function cohort: TKA, 54%, and UKA, 46%). These findings have important implications for policy and suggest that the trend of increased utilization of UKA is consistent with patient preferences and is patient-centered. Some may argue that the percentage of the knee osteoarthritis population who qualify for UKA is far lower than the proportion who desire the procedure86. This disagreement is based on several factors, including data discrepancies between national registries regarding revision rates, design of knee implants, and patient qualification criteria86. Historically, to qualify for UKA, ideal patients have included those <60 years old with symptomatic end-stage osteoarthritis in the medial compartment of the knee, low levels of activity, weight of <180 lb (<82 kg), intact cruciate ligaments, and certain flexion requirements86. These criteria have lessened with the development of new implants and the expanded use of UKAs87-89. A comprehensive review of reported estimates from peer-reviewed publications indicates that 6% to 50% of patients with knee osteoarthritis may qualify for UKA87,89-91. The results of the present study highlight the role of preference elicitation and application through shared decision-making because there are a larger number of patients who, if eligible for UKA, would choose this treatment over TKA. Furthermore, patient preferences may not necessarily align with physician preferences, with physicians tending to be more risk-averse. Physicians should thoroughly discuss treatments and tradeoffs with patients to promote more informed shared decision-making.

Study limitations included lower enrollment numbers in the good-function cohort than in the fair/poor-function cohort; however, the sample size for both cohorts was adequate for analysis. Following de Bekker-Grob et al., the minimum sample sizes were 61 and 24 for the good and fair/poor-function cohorts, respectively92. Also, although patients with good function may undergo arthroplasty, it is more common for patients with worse functional status to undergo a surgical procedure25,26. This discrepancy is reflected in enrollment numbers between cohorts. Additionally, all study respondents were enrolled at a single academic medical center. Although we cannot guarantee that this cohort is representative of the overall knee osteoarthritis population, it was representative of the local population. A final limitation was the respondent evaluation of hypothetical treatments in the discrete-choice experiment; actual treatment choices may vary from stated choices.

In summary, patient preferences for outcomes of knee arthroplasty vary among patients. Unsurprisingly, patients with worse baseline functional status valued functional improvements significantly more than those with good baseline function. Overall, complication and revision rates were most important. Clinicians should focus on these when discussing treatments. There were no differences among preference weights between cohorts. There appeared to be a preference for UKA that, if combined with the appropriate patient criteria, may support increased utilization of the procedure for a segment of the population.

Appendix

Supporting material provided by the authors is posted with the online version of this article as a data supplement at jbjs.org (http://links.lww.com/JBJS/G142).

Note: The authors thank Diane B. Covington, PA-C, MHS, and Justin R. Lynch, PA-C, for their assistance with participant enrollment.

References

1. Maradit Kremers H, Larson DR, Crowson CS, Kremers WK, Washington RE, Steiner CA, Jiranek WA, Berry DJ. Prevalence of total hip and knee replacement in the United States. J Bone Joint Surg Am. 2015 Sep 2;97(17):1386-97.
2. Hansen EN, Ong KL, Lau E, Kurtz SM, Lonner JH. Unicondylar knee arthroplasty in the U.S. patient population: prevalence and epidemiology. Am J Orthop (Belle Mead NJ). 2018 Dec;47(12).
3. Pearse AJ, Hooper GJ, Rothwell A, Frampton C. Survival and functional outcome after revision of a unicompartmental to a total knee replacement: the New Zealand National Joint Registry. J Bone Joint Surg Br. 2010 Apr;92(4):508-12.
4. Liddle AD, Pandit H, Judge A, Murray DW. Patient-reported outcomes after total and unicompartmental knee arthroplasty: a study of 14,076 matched patients from the National Joint Registry for England and Wales. Bone Joint J. 2015 Jun;97-B(6):793-801.
5. Lyons MC, MacDonald SJ, Somerville LE, Naudie DD, McCalden RW. Unicompartmental versus total knee arthroplasty database analysis: is there a winner? Clin Orthop Relat Res. 2012 Jan;470(1):84-90.
6. Niinimäki T, Eskelinen A, Mäkelä K, Ohtonen P, Puhto AP, Remes V. Unicompartmental knee arthroplasty survivorship is lower than TKA survivorship: a 27-year Finnish registry study. Clin Orthop Relat Res. 2014 May;472(5):1496-501. Epub 2013 Nov 19.
7. Dunbar MJ, Richardson G, Robertsson O. I can’t get no satisfaction after my total knee replacement: rhymes and reasons. Bone Joint J. 2013 Nov;95-B(11)(Suppl A):148-52.
8. Willis-Owen CA, Brust K, Alsop H, Miraldo M, Cobb JP. Unicondylar knee arthroplasty in the UK National Health Service: an analysis of candidacy, outcome and cost efficacy. Knee. 2009 Dec;16(6):473-8. Epub 2009 May 22.
9. Liddle AD, Judge A, Pandit H, Murray DW. Adverse outcomes after total and unicompartmental knee replacement in 101,330 matched patients: a study of data from the National Joint Registry for England and Wales. Lancet. 2014 Oct 18;384(9952):1437-45.
10. Cobb JP. Patient safety after partial and total knee replacement. Lancet. 2014 Oct 18;384(9952):1405-7.
11. Lim JW, Cousins GR, Clift BA, Ridley D, Johnston LR. Oxford unicompartmental knee arthroplasty versus age and gender matched total knee arthroplasty - functional outcome and survivorship analysis. J Arthroplasty. 2014 Sep;29(9):1779-83. Epub 2014 Apr 5.
12. Newman J, Pydisetty RV, Ackroyd C. Unicompartmental or total knee replacement: the 15-year results of a prospective randomised controlled trial. J Bone Joint Surg Br. 2009 Jan;91(1):52-7.
13. Lancsar E, Louviere J. Conducting discrete choice experiments to inform healthcare decision making: a user’s guide. Pharmacoeconomics. 2008;26(8):661-77.
14. Institute of Medicine (US) Committee on Quality of Health Care in America. Crossing the quality chasm. A new health system for the 21st century. National Academies Press; 2001.
15. Bridges JF, Hauber AB, Marshall D, Lloyd A, Prosser LA, Regier DA, Johnson FR, Mauskopf J. Conjoint analysis applications in health—a checklist: a report of the ISPOR Good Research Practices for Conjoint Analysis Task Force. Value Health. 2011 Jun;14(4):403-13. Epub 2011 Apr 22.
16. Bridges JF. Stated preference methods in health care evaluation: an emerging methodological paradigm in health economics. Appl Health Econ Health Policy. 2003;2(4):213-24.
17. Marshall D, Bridges JF, Hauber B, Cameron R, Donnalley L, Fyie K, Johnson FR. Conjoint analysis applications in health - how are studies being designed and reported?: an update on current practice in the published literature between 2005 and 2008. Patient. 2010 Dec 1;3(4):249-56.
18. Mühlbacher A, Johnson FR. Choice experiments to quantify preferences for health and healthcare: state of the practice. Appl Health Econ Health Policy. 2016 Jun;14(3):253-66.
19. Coast J, Al-Janabi H, Sutton EJ, Horrocks SA, Vosper AJ, Swancutt DR, Flynn TN. Using qualitative methods for attribute development for discrete choice experiments: issues and recommendations. Health Econ. 2012 Jun;21(6):730-41. Epub 2011 May 6.
20. Telser H, Zweifel P. Validity of discrete-choice experiments evidence for health risk reduction. Appl Econ. 2007;39(1):69-78.
21. Clark MD, Determann D, Petrou S, Moro D, de Bekker-Grob EW. Discrete choice experiments in health economics: a review of the literature. Pharmacoeconomics. 2014 Sep;32(9):883-902.
22. Murray DW, Fitzpatrick R, Rogers K, Pandit H, Beard DJ, Carr AJ, Dawson J. The use of the Oxford hip and knee scores. J Bone Joint Surg Br. 2007 Aug;89(8):1010-4.
23. Kalairajah Y, Azurza K, Hulme C, Molloy S, Drabu KJ. Health outcome measures in the evaluation of total hip arthroplasties—a comparison between the Harris hip score and the Oxford hip score. J Arthroplasty. 2005 Dec;20(8):1037-41.
24. Rothwell AG, Hooper GJ, Hobbs A, Frampton CM. An analysis of the Oxford hip and knee scores and their relationship to early joint revision in the New Zealand Joint Registry. J Bone Joint Surg Br. 2010 Mar;92(3):413-8.
25. Petersen CL, Kjærsgaard JB, Kjærgaard N, Jensen MU, Laursen MB. Thresholds for Oxford Knee Score after total knee replacement surgery: a novel approach to post-operative evaluation. J Orthop Surg Res. 2017 Jun 12;12(1):89.
26. Yeoman TFM, Clement ND, Macdonald D, Moran M. Recall of preoperative Oxford Hip and Knee Scores one year after arthroplasty is an alternative and reliable technique when used for a cohort of patients. Bone Joint Res. 2018 Jun 5;7(5):351-6.
27. de Bekker-Grob EW, Ryan M, Gerard K. Discrete choice experiments in health economics: a review of the literature. Health Econ. 2012 Feb;21(2):145-72. Epub 2010 Dec 19.
28. Laurencin CT, Zelicof SB, Scott RD, Ewald FC. Unicompartmental versus total knee arthroplasty in the same patient. A comparative study. Clin Orthop Relat Res. 1991 Dec;(273):151-6.
29. Dalury DF, Fisher DA, Adams MJ, Gonzales RA. Unicompartmental knee arthroplasty compares favorably to total knee arthroplasty in the same patient. Orthopedics. 2009 Apr;32(4):orthosupersite.com/view.asp?rID=38057/view.asp?rID=38057.
30. Arirachakaran A, Choowit P, Putananon C, Muangsiri S, Kongtharvonskul J. Is unicompartmental knee arthroplasty (UKA) superior to total knee arthroplasty (TKA)? A systematic review and meta-analysis of randomized controlled trial. Eur J Orthop Surg Traumatol. 2015 Jul;25(5):799-806. Epub 2015 Feb 13.
31. Lombardi AV Jr, Berend KR, Walter CA, Aziz-Jacobo J, Cheney NA. Is recovery faster for mobile-bearing unicompartmental than total knee arthroplasty? Clin Orthop Relat Res. 2009 Jun;467(6):1450-7. Epub 2009 Feb 19.
32. Peersman G, Jak W, Vandenlangenbergh T, Jans C, Cartier P, Fennema P. Cost-effectiveness of unicondylar versus total knee arthroplasty: a Markov model analysis. Knee. 2014;21(Suppl 1):S37-42.
33. Koskinen E, Eskelinen A, Paavolainen P, Pulkkinen P, Remes V. Comparison of survival and cost-effectiveness between unicondylar arthroplasty and total knee arthroplasty in patients with primary osteoarthritis: a follow-up study of 50,493 knee replacements from the Finnish Arthroplasty Register. Acta Orthop. 2008 Aug;79(4):499-507.
34. Smith WB 2nd, Steinberg J, Scholtes S, Mcnamara IR. Medial compartment knee osteoarthritis: age-stratified cost-effectiveness of total knee arthroplasty, unicompartmental knee arthroplasty, and high tibial osteotomy. Knee Surg Sports Traumatol Arthrosc. 2017 Mar;25(3):924-33. Epub 2015 Oct 31.
35. Ghomrawi HM, Eggman AA, Pearle AD. Effect of age on cost-effectiveness of unicompartmental knee arthroplasty compared with total knee arthroplasty in the U.S. J Bone Joint Surg Am. 2015 Mar 4;97(5):396-402.
36. Borus T, Roberts D, Fairchild P, Christopher J, Conditt M, Branch S, Matthews J, Pirtle K, Baer M. UKA patients return to function earlier than TKA patients. Bone & Joint J. 2016;98-B(SUPP_1):50.
37. Witjes S, Gouttebarge V, Kuijer PP, van Geenen RC, Poolman RW, Kerkhoffs GM. Return to sports and physical activity after total and unicondylar knee arthroplasty: a systematic review and meta-analysis. Sports Med. 2016 Feb;46(2):269-92.
38. Hunt LP, Ben-Shlomo Y, Clark EM, Dieppe P, Judge A, MacGregor AJ, Tobias JH, Vernon K, Blom AW; National Joint Registry for England and Wales. 45-day mortality after 467,779 knee replacements for osteoarthritis from the National Joint Registry for England and Wales: an observational study. Lancet. 2014 Oct 18;384(9952):1429-36.
39. Bolognesi MP, Greiner MA, Attarian DE, Watters TS, Wellman SS, Curtis LH, Berend KR, Setoguchi S. Unicompartmental knee arthroplasty and total knee arthroplasty among Medicare beneficiaries, 2000 to 2009. J Bone Joint Surg Am. 2013 Nov 20;95(22):e174.
40. Berend KR, Morris MJ, Lombardi AV. Unicompartmental knee arthroplasty: incidence of transfusion and symptomatic thromboembolic disease. Orthopedics. 2010 Sep;33(9)(Suppl):8-10.
41. Stukenborg-Colsman C, Wirth CJ, Lazovic D, Wefer A. High tibial osteotomy versus unicompartmental joint replacement in unicompartmental knee joint osteoarthritis: 7-10-year follow-up prospective randomised study. Knee. 2001 Oct;8(3):187-94.
42. Patil S, Colwell CW Jr, Ezzet KA, D’Lima DD. Can normal knee kinematics be restored with unicompartmental knee replacement? J Bone Joint Surg Am. 2005 Feb;87(2):332-8.
43. Newman JH, Ackroyd CE, Shah NA. Unicompartmental or total knee replacement? Five-year results of a prospective, randomised trial of 102 osteoarthritic knees with unicompartmental arthritis. J Bone Joint Surg Br. 1998 Sep;80(5):862-5.
44. Soohoo NF, Sharifi H, Kominski G, Lieberman JR. Cost-effectiveness analysis of unicompartmental knee arthroplasty as an alternative to total knee arthroplasty for unicompartmental osteoarthritis. J Bone Joint Surg Am. 2006 Sep;88(9):1975-82.
45. Berger RA, Meneghini RM, Jacobs JJ, Sheinkop MB, Della Valle CJ, Rosenberg AG, Galante JO. Results of unicompartmental knee arthroplasty at a minimum of ten years of follow-up. J Bone Joint Surg Am. 2005 May;87(5):999-1006.
46. Brown NM, Sheth NP, Davis K, Berend ME, Lombardi AV, Berend KR, Della Valle CJ. Total knee arthroplasty has higher postoperative morbidity than unicompartmental knee arthroplasty: a multicenter analysis. J Arthroplasty. 2012 Sep;27(8)(Suppl):86-90. Epub 2012 May 4.
47. Bourne RB, Chesworth BM, Davis AM, Mahomed NN, Charron KD. Patient satisfaction after total knee arthroplasty: who is satisfied and who is not? Clin Orthop Relat Res. 2010 Jan;468(1):57-63.
48. Parvizi J, Nunley RM, Berend KR, Lombardi AV Jr, Ruh EL, Clohisy JC, Hamilton WG, Della Valle CJ, Barrack RL. High level of residual symptoms in young patients after total knee arthroplasty. Clin Orthop Relat Res. 2014 Jan;472(1):133-7.
49. Fabre-Aubrespy M, Ollivier M, Pesenti S, Parratte S, Argenson JN. Unicompartmental knee arthroplasty in patients older than 75 results in better clinical outcomes and similar survivorship compared to total knee arthroplasty. A matched controlled study. J Arthroplasty. 2016 Dec;31(12):2668-71. Epub 2016 Jun 29.
50. Eymard F, Charles-Nelson A, Katsahian S, Chevalier X, Bercovy M. Predictive factors of “forgotten knee” acquisition after total knee arthroplasty: long-term follow-up of a large prospective cohort. J Arthroplasty. 2017 Feb;32(2):413-418.e1. Epub 2016 Jun 23.
51. Haughom BD, Schairer WW, Hellman MD, Nwachukwu BU, Levine BR. An analysis of risk factors for short-term complication rates and increased length of stay following unicompartmental knee arthroplasty. HSS J. 2015 Jul;11(2):112-6. Epub 2015 Jan 27.
52. March L, Cross M, Tribe K, Lapsley H, Courtenay B, Brooks P. Cost of joint replacement surgery for osteoarthritis: the patients’ perspective. J Rheumatol. 2002 May;29(5):1006-14.
53. Drager J, Hart A, Khalil JA, Zukor DJ, Bergeron SG, Antoniou J. Shorter hospital stay and lower 30-day readmission after unicondylar knee arthroplasty compared to total knee arthroplasty. J Arthroplasty. 2016 Feb;31(2):356-61. Epub 2015 Sep 18.
54. Georgoulis AD, Moraiti C, Ristanis S, Stergiou N. A novel approach to measure variability in the anterior cruciate ligament deficient knee during walking: the use of the approximate entropy in orthopaedics. J Clin Monit Comput. 2006 Feb;20(1):11-8. Epub 2006 Mar 6.
55. Callahan CM, Drake BG, Heck DA, Dittus RS. Patient outcomes following unicompartmental or bicompartmental knee arthroplasty. A meta-analysis. J Arthroplasty. 1995 Apr;10(2):141-50.
56. Sun PF, Jia YH. Mobile bearing UKA compared to fixed bearing TKA: a randomized prospective study. Knee. 2012 Mar;19(2):103-6. Epub 2011 Feb 22.
57. Costa CR, Johnson AJ, Mont MA, Bonutti PM. Unicompartmental and total knee arthroplasty in the same patient. J Knee Surg. 2011 Dec;24(4):273-8.
58. Amin AK, Patton JT, Cook RE, Gaston M, Brenkel IJ. Unicompartmental or total knee arthroplasty?: Results from a matched study. Clin Orthop Relat Res. 2006 Oct;451(451):101-6.
59. Griffin T, Rowden N, Morgan D, Atkinson R, Woodruff P, Maddern G. Unicompartmental knee arthroplasty for the treatment of unicompartmental osteoarthritis: a systematic study. ANZ J Surg. 2007 Apr;77(4):214-21.
60. Weale AE, Murray DW, Newman JH, Ackroyd CE. The length of the patellar tendon after unicompartmental and total knee replacement. J Bone Joint Surg Br. 1999 Sep;81(5):790-5.
61. Hassaballa MA, Porteous AJ, Newman JH, Rogers CA. Can knees kneel? Kneeling ability after total, unicompartmental and patellofemoral knee arthroplasty. Knee. 2003 Jun;10(2):155-60.
62. Hiyama Y, Wada O, Nakakita S, Mizuno K. Joint awareness after total knee arthroplasty is affected by pain and quadriceps strength. Orthop Traumatol Surg Res. 2016 Jun;102(4):435-9. Epub 2016 Apr 1.
63. Behrend H, Zdravkovic V, Giesinger J, Giesinger K. Factors predicting the Forgotten Joint Score after total knee arthroplasty. J Arthroplasty. 2016 Sep;31(9):1927-32. Epub 2016 Feb 27.
64. Zuiderbaan HA, van der List JP, Khamaisy S, Nawabi DH, Thein R, Ishmael C, Paul S, Pearle AD. Unicompartmental knee arthroplasty versus total knee arthroplasty: which type of artificial joint do patients forget? Knee Surg Sports Traumatol Arthrosc. 2017 Mar;25(3):681-6. Epub 2015 Nov 21.
65. Eymard F, Charles-Nelson A, Katsahian S, Chevalier X, Bercovy M. “Forgotten knee” after total knee replacement: a pragmatic study from a single-centre cohort. Joint Bone Spine. 2015 May;82(3):177-81. Epub 2015 Jan 23.
66. Thienpont E, Opsomer G, Koninckx A, Houssiau F. Joint awareness in different types of knee arthroplasty evaluated with the Forgotten Joint score. J Arthroplasty. 2014 Jan;29(1):48-51. Epub 2013 May 18.
67. Heck DA, Marmor L, Gibson A, Rougraff BT. Unicompartmental knee arthroplasty. A multicenter investigation with long-term follow-up evaluation. Clin Orthop Relat Res. 1993 Jan;(286):154-9.
68. Bert JM. Unicompartmental knee replacement. Orthop Clin North Am. 2005 Oct;36(4):513-22.
69. Ji JH, Park SE, Song IS, Kang H, Ha JY, Jeong JJ. Complications of medial unicompartmental knee arthroplasty. Clin Orthop Surg. 2014 Dec;6(4):365-72. Epub 2014 Nov 10.
70. Delanois RE, Patel NK, Mistry JB, Mont MA. Economic considerations for obese patients undergoing total knee arthroplasty: commentary on an article by Eric R. Wagner, MD, et al.: “Effect of body mass index on reoperation and complications after total knee arthroplasty. J Bone Joint Surg Am. 2016 Dec 21;98(24):e113.
71. Wagner ER, Kamath AF, Fruth K, Harmsen WS, Berry DJ. Effect of body mass index on reoperation and complications after total knee arthroplasty. J Bone Joint Surg Am. 2016 Dec 21;98(24):2052-60.
72. Meller MM, Toossi N, Johanson NA, Gonzalez MH, Son MS, Lau EC. Risk and cost of 90-day complications in morbidly and superobese patients after total knee arthroplasty. J Arthroplasty. 2016 Oct;31(10):2091-8. Epub 2016 Mar 10.
73. Murray DW, Liddle AD, Dodd CA, Pandit H. Unicompartmental knee arthroplasty: is the glass half full or half empty? Bone Joint J. 2015 Oct;97-B(10)(Suppl A):3-8.
74. Goodfellow JW, O’Connor JJ, Murray DW. A critique of revision rate as an outcome measure: re-interpretation of knee joint registry data. J Bone Joint Surg Br. 2010 Dec;92(12):1628-31.
75. Mohammad HR, Strickland L, Hamilton TW, Murray DW. Long-term outcomes of over 8,000 medial Oxford Phase 3 Unicompartmental Knees-a systematic review. Acta Orthop. 2018 Feb;89(1):101-7. Epub 2017 Aug 23.
76. National Clinical Guideline Centre (UK). Osteoarthritis: care and management in adults. National Institute for Health and Care Excellence (UK); 2014.
77. Johnson FR, Yang JC, Reed SD. The internal validity of discrete choice experiment data: a testing tool for quantitative assessments. Value Health. 2019 Feb;22(2):157-60. Epub 2018 Sep 27.
78. Hauber AB, González JM, Groothuis-Oudshoorn CG, Prior T, Marshall DA, Cunningham C, IJzerman MJ, Bridges JF. Statistical methods for the analysis of discrete choice experiments: a report of the ISPOR Conjoint Analysis Good Research Practices Task Force. Value Health. 2016 Jun;19(4):300-15. Epub 2016 May 12.
79. Bech M, Gyrd-Hansen D. Effects coding in discrete choice experiments. Health Econ. 2005 Oct;14(10):1079-83.
80. Krinsky I, Robb AL. On approximating the statistical properties of elasticities: a correction. Rev Econ Stat. 1990 Feb;72(1):189-90.
81. Gonzalez R, Wu G. On the shape of the probability weighting function. Cogn Psychol. 1999 Feb;38(1):129-66.
82. Vina ER, Ran D, Ashbeck EL, Kaur M, Kwoh CK. Relationship between knee pain and patient preferences for joint replacement: health care access matters. Arthritis Care Res (Hoboken). 2017 Jan;69(1):95-103.
83. Vina ER, Ran D, Ashbeck EL, Ibrahim SA, Hannon MJ, Zhou JJ, Kwoh CK. Patient preferences for total knee replacement surgery: relationship to clinical outcomes and stability of patient preferences over 2 years. Semin Arthritis Rheum. 2016 Aug;46(1):27-33. Epub 2016 Mar 30.
84. Kwoh CK, Vina ER, Cloonan YK, Hannon MJ, Boudreau RM, Ibrahim SA. Determinants of patient preferences for total knee replacement: African-Americans and whites. Arthritis Res Ther. 2015 Dec 3;17:348.
85. Posnett J, Dixit S, Oppenheimer B, Kili S, Mehin N. Patient preference and willingness to pay for knee osteoarthritis treatments. Patient Prefer Adherence. 2015 Jun 11;9:733-44.
86. Campi S, Tibrewal S, Cuthbert R, Tibrewal SB. Unicompartmental knee replacement - current perspectives. J Clin Orthop Trauma. 2018 Jan-Mar;9(1):17-23. Epub 2017 Nov 28.
87. Pandit H, Jenkins C, Gill HS, Smith G, Price AJ, Dodd CA, Murray DW. Unnecessary contraindications for mobile-bearing unicompartmental knee replacement. J Bone Joint Surg Br. 2011 May;93(5):622-8.
88. Kozinn SC, Scott R. Unicondylar knee arthroplasty. J Bone Joint Surg Am. 1989 Jan;71(1):145-50.
89. Ritter MA, Faris PM, Thong AE, Davis KE, Meding JB, Berend ME. Intra-operative findings in varus osteoarthritis of the knee. An analysis of pre-operative alignment in potential candidates for unicompartmental arthroplasty. J Bone Joint Surg Br. 2004 Jan;86(1):43-7.
90. Laskin RS. Unicompartmental knee replacement: some unanswered questions. Clin Orthop Relat Res. 2001 Nov;(392):267-71.
91. Stern SH, Becker MW, Insall JN. Unicondylar knee arthroplasty. An evaluation of selection criteria. Clin Orthop Relat Res. 1993 Jan;(286):143-8.
92. de Bekker-Grob EW, Donkers B, Jonker MF, Stolk EA. Sample size requirements for discrete-choice experiments in healthcare: a practical guide. Patient. 2015 Oct;8(5):373-84.

Supplemental Digital Content

Copyright © 2020 The Authors. Published by The Journal of Bone and Joint Surgery, Incorporated. All rights reserved.