Prevalence of Similar or Worse Symptom and Osteoarthritis Severity of Index and Contralateral Knees Prior to Knee Arthroplasty: A Cross-sectional Multicenter Cohort Study

Abstract

Objectives:

Patients considering knee arthroplasty (KA) and consulting orthopaedic surgeons use different criteria to judge surgical appropriateness with surgeons placing high priority on osteoarthritis (OA) severity and age while patients place greatest emphasis on pain severity. In the face of fluctuating symptoms and high rates of bilateral knee OA, selection of the optimal knee for KA surgery can be challenging. This study examined osteoarthritis and symptom severity profiles of index versus contralateral knees of persons preparing for KA.

Method:

The nine-year Osteoarthritis Initiative data from 427 participants and five-year Multicenter Osteoarthritis Study data from 377 participants undergoing KA were included. The Western Ontario and McMaster Universities Osteoarthritis Index Pain, Function and Stiffness Scales for both knees and radiographic readings were obtained at the visit prior to knee arthroplasty. Radar plots illustrated symptom and OA severity. Prevalence estimates of the proportion of contralateral knees with comparable or worse symptoms and OA severity relative to the index knee were reported.

Results:

The prevalence of comparable or worse knee OA on the contralateral knees as compared to index knees is 26% (95% CI = 22%, 30%). Prevalence of the same or worse WOMAC Pain for the contralateral knees relative to the index knees is 17% (95% CI = 14%, 21%).

Conclusions:

The prevalence of comparable/worse knee OA or symptoms on the contralateral knees relative to index knees of persons undergoing unilateral KA is substantial. Methods for selecting the surgical knee should be reconsidered for this substantial subgroup of patients.

Patients considering knee replacement surgery may have different outcome priorities as compared to their orthopaedic surgeons. For example, Hawker and colleagues found that patients placed greatest priority on pain reduction while surgeons rated overall patient satisfaction as the highest priority [1]. When judging whether patients may be appropriate candidates for surgery, surgeons place greatest emphasis on knee osteoarthritis (OA) severity, age and the number of knee compartments affected by OA with less emphasis placed on patients’ pain severity and functional deficit [2,3].

Differences between patients and their surgeons regarding knee arthroplasty (KA) candidacy and outcome priorities may lead to differences in surgeon recommendations and patients’ hopes and expectations. One potential area of confusion is the decision as to which knee should be operated on first when a patient with bilateral symptomatic knee OA is being considered for KA surgery. Knee OA is often a bilateral disease with fluctuating symptoms and many patients are symptomatic bilaterally. For persons undergoing KA, prevalence estimates of contralateral knee OA range from 56.9% [4] to 87.4% [5]. Because patients place high priority on pain relief while surgeons generally place greater emphasis on OA severity assessment, there may be disagreement as to which knee is highest priority for an upcoming arthroplasty. Even when there is concordance between patient and surgeon on which knee is highest priority, the extent of knee pain, functional loss and OA severity may be very similar between knees. Symptomatic knee OA is on the rise [6] with obesity being key but not the only contributor to the high rate.

Given the increasing rate of symptomatic knee OA, the high prevalence of bilateral disease with a fluctuating symptom trajectory and the sometimes discordant opinions of patients and their surgeons regarding criteria for KA candidacy and outcome expectations, this study examined bilateral knee OA severity and symptom profiles of patients preparing for unilateral KA. A substantial proportion of patients experience variation in knee OA pain severity over time, [7,8] and the resulting uncertainty in selecting the knee for KA can create challenges for both patients and surgeons. Prior reports of the extent to which knee OA severity and symptoms differed in both knees when only one KA was planned were not found. If either knee OA severity or symptoms were either the same or worse in the contralateral knee as compared to the index knee, the data would suggest need for a potential alternative approach to selecting the knee that should be considered the highest priority for surgery.

Objectives of the study were to: 1) determine the symptom (i.e. pain with activity and knee stiffness), function (i.e. difficulty with daily activity) and osteoarthritic severity profiles of the contralateral knee relative to the index knee in persons undergoing unilateral KA; 2) contrast these profiles with those generated from persons undergoing bilateral KA in the same year, and; 3) estimate the prevalence of contralateral knee OA severity or symptom severity that was comparable to or worse than that for the preoperative index knee. Data from the extremely well characterized and rigorously collected Osteoarthritis Initiative (OAI) and Multicenter Osteoarthritis Study (MOST) National Institutes of Health funded longitudinal cohort studies were used to achieve study objectives.

Materials and Methods

Sources of data - The Osteoarthritis Initiative AND the Multicenter Osteoarthritis Study

The Osteoarthritis Initiative (OAI) is a National Institutes of Health and privately funded prospective community-based multicenter nine year longitudinal study of subjects with radiographic knee osteoarthritis (OA) or at high risk for knee OA [9]. Knee OA risk was confirmed if the subjects were overweight or obese, reported prior knee injury or surgery, had knee symptoms or family history of knee replacement surgery. Subjects were recruited from the communities of four study sites and Institutional Review Boards of all sites approved the study. The sites were the following: (1) Memorial Hospital of Rhode Island, in Pawtucket, Rhode Island, (2) University of Maryland in Baltimore, Maryland, (3) Ohio State University in Columbus, Ohio, and (4) University of Pittsburgh in Pittsburgh, Pennsylvania. All data in the current study were obtained from a publically available website (https://ndar.nih.gov/).

A total of 17,457 women and men between the ages of 45 and 79 years were screened and 4,796 were enrolled. Enrollment began in 2004 and publically available data are available with nine years of yearly follow-up occurring via either telephone or a clinic visit. A total of 4.796 persons consented to the study. Missing clinic visit data for the entire OAI sample ranged from 10% at the one-year clinic follow-up visit to 35% at the 8-year clinic follow-up visit.

The MOST study subjects were recruited from two sites, the University of Iowa in Iowa City, Iowa and the University of Alabama, Birmingham, Alabama. Much like OAI, persons either had or were at risk of developing knee OA. Participants ranged in age from 50 to 79 years. All participants (N = 3,026) signed site IRB approved consent forms. Over the 7-year study period, missing data from follow-up visits or phone-based data collection ranged from 1% at the 15 month visit, to 13% at the 7-year follow-up visit.

Subjects

To be eligible for the current study, participants had to have a KA during the study. Participants were excluded if they had a KA prior to baseline data collection because a prior KA precluded comparisons to contralateral anatomically intact knees. Over the nine-year OAI study period, 427 subjects underwent knee arthroplasty (KA) on at least one knee. Of these, 9 participants were excluded because they had a KA prior to baseline data collection. Of the remaining 418 participants, 56 had bilateral KA in the same year.

Over the first 5 years in MOST (only years with publically available KA data), a total of 377 subjects had KA on at least one knee. Of these, 23 participants were excluded because they had a KA prior to baseline. Of the remaining 354 participants, 70 had bilateral KA in the same year. A total of 97% of all knee replacements in both studies had been confirmed by medical record confirmation at the time of this study and the remainder were self-reported and had not yet been adjudicated. Pre-operative descriptive data for both the MOST and OAI datasets are provided in Table 1.

Table 1.

Characteristics of the Samples at the Preoperative visit

	Unilateral OAI Preopa Data (n=362)	Bilateral OAI Preop Data (n = 56)	Unilateral MOST Preop Data (n=284)	Bilateral MOST Preop Data (n = 70)
Age	68.2 (8.4)	67.9 (8.4)	67.6 (7.6)	66.8 (7.6)
Sex (% female)	61.3	55.4	70.1	68.6
Race (% African American)	13.3	8.9	13	4.3
Body Mass Index kg/m2 (mean, sd)	30.1 (5.0)	30.3 (4.3)	32.5 (6.7)	32.1 (5.9)
Comorbidity score ≥1 n (%)	113 (33.1)	15 (28.8)	115 (43.4)	28 (41.2)
Kellgren & Lawrence Grade n (%)
0 Index, contrab	2 (0.8), 26 (12.4)	0, 1(3.0)	4 (1.7), 46 (20.6)	0, 1 (1.6)
1 Index, contra	5 (1.9), 30 (14.4)	2 (5.6), 0	6 (2.5), 27 (12.1)	1 (1.6), 2 (3.2)
2 Index, contra	15 (5.8), 61 (29.2)	2 (5.6), 7 (21.2)	21 (8.9), 37 (16.6)	3 (4.7), 3 (4.7)
3 Index, contra	54 (20.7), 65 (31.1)	8 (22.2), 7 (21.2)	76 (32.2), 66 (29.6)	18 (28.6), 12 (19.1)
4 Index, contra	184 (70.8), 27 (12.9)	24 (66.6), 18 (54.5)	129 (54.7), 47 (21.1)	41 (65.1), 45 (71.4)
WOMAC Pain (Mean (sd))
Index, contra	7.7 (3.8), 3.0 (3.3)	7.0 (3.8), 6.7 (3.7)	8.1 (3.8), 4.4 (4.1)	6.5 (3.6), 6.8 (3.6)
WOMAC Stiffness (Mean (sd))
Index, contra	3.6 (1.7), 1.6 (1.6)	3.2 (1.8),3.0 (1.8)	3.7 (1.8), 2.3 (1.9)	3/5 (1.5), 3.6 (1.6)

^apreop = preoperative,

^bcontra = contralateral knee

Radiographic tibiofemoral and patellofemoral osteoarthritis assessment

Standardized standing flexed knee posterior-anterior knee radiographs were obtained using identical methods in both the OAI and MOST. Radiographs were obtained yearly on all subjects, except for years 5 and 7 in OAI. In MOST, they were obtained at each clinic visit except for the 72 month visit. Radiographs at the 15 month visit were obtained on a small subset of the sample with no baseline OA had radiographs. The radiographic method is a reliable and valid approach [10]. Tibiofemoral OA grading was conducted by the OAI and MOST investigators using an extensive adjudication process and experienced radiologists and rheumatologists to obtain Kellgren and Lawrence (K&L) grades for all knees over all time periods. All K&L grades were available on the OAI website and can be requested from MOST investigators (http://most.ucsf.edu/studyoverview.asp). Reliability of K&L grades by experienced readers is extremely high with weighted kappa values of 0.79 to 0.83 [11].

Radiographic images of standing knee flexed lateral views of the patellofemoral joints were not obtained in OAI but were obtained during the MOST study. Grading of OA for the patellofemoral joints was coded as either present or absent. Patellofemoral OA was coded as present when either a patellar osteophyte was clearly present or joint space narrowing of the patellofemoral joint was clearly present along with osteophyte(s), sclerosis or cysts.

Self-reported Pain, Function and Stiffness

Both the OAI and MOST collected WOMAC Pain and Stiffness scores for both knees. Only the OAI collected WOMAC Physical Function scores on both knees. The MOST collected person-level WOMAC Physical Function scores and therefore only OAI WOMAC Physical Function data are reported. Both studies collected WOMAC scores at each study visit over the entire study period (yearly for OAI, and at baseline, 15 months, 30, months and 60 months for MOST). WOMAC scores have demonstrated high levels of reliability and validity for persons with knee OA [12–14]. The WOMAC Likert version 3.1 Physical Function score ranges from 0–68, with higher scores indicating greater difficulty with function. WOMAC pain scores range from 0–20, with higher scores indicating greater pain with daily activities. WOMAC Stiffness scores range from 0 to 8 with higher scores equating to greater daily knee stiffness.

Other OAI and MOST data used in the current study

Age during the preoperative visit, race (African American or other), sex and preoperative body mass index, measured in Kg/m² were collected to characterize the sample. A validated modified Charlson comorbidity scale with higher scores indicating higher comorbidity [15], was used to characterize the extent of comorbidity.

Both the OAI and MOST also reported the number of days (in the case of OAI) or the number of months (in the case of MOST) between the KA and the prior study visit. Because the KA data were time varying on both OAI and MOST, that is, the time between KA surgery and the study visits varied for each subject, these data were used to determine the length of time between the pre-operative visit and the date of surgery.

Data Analysis

Data were analyzed separately for those with unilateral KA and those with bilateral KA in the same year. I wanted to compare and contrast the extent of knee OA and knee symptoms in the contralateral knees of those who had planned to have only one KA versus those who planned to have KA in both knees. Persons who planned to have KA for both knees were selected based on the following: either they had bilateral KA within a month (i.e., 26 of 56 bilateral KA participants in OAI and 30 of 70 in MOST) or they had bilateral KA within a year. I reasoned that if bilateral KAs were conducted within a year, it was likely that the participant was considering bilateral KA at the time the first KA was done. If KA was conducted bilaterally with more than one year between surgeries (n = 51 in OAI and n = 52 in MOST) it was likely that bilateral KA was not planned at the time of the first surgery.

Because K&L grades of 4 represent non-reversible end-stage disease, if participants had missing K&L data at the preoperative visit but had a K&L grade of 4 for the knee of interest at a prior visit, the K&L grade of 4 was carried forward to the pre-operative visit. No other forms of data carry-forward or imputation were used. Descriptive data for OAI and MOST data are provided in Table 1. Radar plots are used to describe the extent of K&L-based knee OA severity for both the index knees and the contralateral knees of those classified as having unilateral KA and separately for those with bilateral KA in the same year. Only MOST provided radiographic assessments of patellofemoral joint OA and these were graded as either present or absent. Radar plots also were used to compare WOMAC Pain and Stiffness scores for index versus contralateral knees for those in OAI and MOST who underwent unilateral KA and for those classified as having bilateral KA in OAI and MOST. Only the OAI collected WOMAC Physical Function data for both knees and these also are reported.

Prevalence with 95% confidence intervals (CI) was calculated for the proportion of participants in both OAI and MOST in whom radiographic severity of the contralateral knees (only in those with unilateral KA) was equal to or worse than the K&L grade for the index knee. Prevalence also was calculated for the proportion of participants in both OAI and MOST who had WOMAC Pain scores for the contralateral knees (only in those with unilateral KA) that were equal to or worse than the index knees. Prevalence of both WOMAC Pain and WOMAC Stiffness scores of the contralateral knees that were equal to or worse than the index also was determined. Finally, prevalence also was calculated for the proportion of participants with unilateral KA who had both K&L grades and Symptoms (i.e., WOMAC Pain and Stiffness) that was either the same as or worse for the contralateral knees as compared to the index knees.

RESULTS

Characteristics of the samples are reported in Table 1 along with the number of participants with missing data for each variable. The average number of days from the preoperative visit to the surgery date for the unilateral KA participants was 176 (SD = 103) days in OAI and 239 (SD = 135) days in MOST. Radar plots are reported separately for both the OAI and MOST datasets and for persons with unilateral versus bilateral KAs. For the unilateral KA sample in MOST, a total of 55% of the index knees had patellofemoral OA while 38% of contralateral knees had patellofemoral OA. For the bilateral cases in MOST, patellofemoral OA was found in 58% and 68% of knees.

The radar plots in Figure 1 indicate that there are differences in the distributions of K&L grades for OAI and MOST data but that the extent of overlap between index and contralateral knees is somewhat similar in the two datasets with approximately 89% of index knees with K&L grades of 3 or 4 and approximately 45% of contralateral knees with K&L grades of 3 or 4. Figure 2 illustrates the WOMAC Pain and Stiffness scores for OAI and MOST. For WOMAC Pain scores, approximately 25% of contralateral knees scored a 0 indicating no pain with activity. WOMAC Pain scores for the index knees demonstrated a large range for both OAI and MOST. WOMAC Stiffness scores demonstrated a similar pattern as that seen for WOMAC Pain with approximately 30% of contralateral knees with no stiffness and a broad range of scores for the index knees. Figure 3 illustrates the WOMAC Physical Function scores for OAI participants with unilateral KA. Approximately 30% of contralateral knees had no functional difficulty while scores for the index knee demonstrated a broad range of scores. For participants with bilateral KA, substantial overlap of scores with a broad range was found for all measures (see table 1).

Figure 1:

Preoperative Kellgren and Lawrence (K&L) grades for persons with unilateral knee arthroplasty. Panel a shows the distribution of K&L grades from 0 to 4 for the OAI data, and panel b shows K&L grades from 0 to 4 for MOST data. Both panels illustrate K&L grades for the index knees (PreInvKL) and the contralateral knees (PreUnKL). To determine the prevalence, identify the grade of interest (e.g. K&L grade of 4) and then identify the point on the graph where the K&L Grade of 4 intersects the radar graph. In the case of the index knee in panel a for K&L grade of 4, the prevalence was approximately 60%.

Figure 2:

Preoperative WOMAC Pain and WOMAC Stiffness scores for persons with unilateral knee arthroplasty. Panel a shows the distribution of WOMAC Pain scores grades from 0 to 16 for the OAI data, and panel b shows WOMAC Pain scores from 0 to 19 for MOST data. Panel c shows the distribution of WOMAC Stiffness scores grades from 0 to 8 for the OAI data, and panel d shows WOMAC Stiffness scores from 0 to 8 for MOST data. Panels illustrate WOMAC Pain and Stiffnes scores for the index knees (PreInvWP or PreInvWST) and the contralateral knees (PreUnWP or PreUnWST). To determine the prevalence, identify the score of interest (e.g. WOMAC Pain Score of 0) and then identify the point on the graph where the WOMAC Pain score intersects the radar graph. In the case of the contralateral knee in panel a, the prevalence of a WOMAC Pain score of 0 for the contralateral knee (PreUnWP) was approximately 30%.

Figure 3:

Preoperative WOMAC Function scores for persons in the OAI with unilateral knee arthroplasty. Preoperative WOMAC Function scores in MOST were not collected for the index and contralateral knees separately. The figure shows the distribution of WOMAC Function scores from 0 to 55 (in 5 point increments) for the OAI data Both panels illustrate scores for the index knees (PreWF or PreInvWF) and the contralateral knees (PreWFUn and PreContraWF). To determine the prevalence, identify the score of interest (e.g. score of O for the PreWFUn) and then identify the point on the graph where this score intersects the radar graph. In the case of the contralateral knee in persons with unilateral KA (panel a), the prevalence was approximately 28%.

Prevalence of K&L scores for those with unilateral KA in which the contralateral knee was either the same or worse as compared to the index knee was 20.5% (40/195) in OAI and 31.0% (67/216) in MOST. The overall prevalence of K&L scores that were the same or worse for the contralateral knees as compared to index knees was 26% (95% CI = 22%, 30%). Prevalence of WOMAC Pain scores for those with unilateral KA when the contralateral knee was either the same or worse as compared to the index knee was 16.8% (58/346) in OAI and 18.3% in (30/164) in MOST for an overall prevalence of 17% (95% CI = 14%, 21%). Prevalence of WOMAC Stiffness scores when the contralateral knee was either the same or worse as compared to the index knee was 25.3% (70/344) in OAI and 41.6% in (91/219) in MOST. The overall prevalence was 29% (95% CI = 25%, 29%). Prevalence for the same or worse WOMAC Pain and WOMAC Stiffness score of the contralateral knees as compared to index knees was 11.3% (39/344) in OAI and 14.7% (23/156) in MOST for a combined prevalence of 12% (95% CI = 10%, 16%). Prevalence when both K&L and WOMAC Pain scores for the contralateral knees were the same as or worse than the index knees was 9% (16/186) in OAI and 11% (16/147) in MOST for a combined prevalence estimate of 10% (95% CI = 7%, 13%).

DISCUSSION

The current study appears to be the first to generate detailed descriptions of the extent of knee OA and pain severity, stiffness and daily activity difficulty for both the preoperative index and contralateral knees of multicenter samples of persons who underwent KA. The radar plots indicated that there is a moderate degree of overlap between the index knees and contralateral knees leading up to unilateral KA. The overall prevalence of knee OA severity of the contralateral knees that was either equivalent to or worse than the index knees was 26% (95% CI = 22%, 30%) for K&L grades. Approximately 10% (95% CI = 7%, 13%) of contralateral knees had the same or worse combined K&L and WOMAC pain scores. Given that approximately 1 million KA surgeries are done in the US each year [16], this equates to approximately 250,000 persons with comparable or worse bilateral knee OA and 100,000 persons with comparable or worse knee OA and pain in the contralateral knees as compared to index knees.

Patients considering KA surgery most commonly endorse knee pain severity as the most important reason for seeking the procedure [1]. Orthopaedic surgeons, on the hand, rely primarily on OA severity assessment and patient’s age as the most important indicators of KA surgical appropriateness [2,17]. These differing priorities combined with fluctuating OA symptom intensity, the norm for the OA disease process, likely lead to confusion on the part of both patients, surgeons and other clinicians advising patients on likely treatment choices when considering which knee is highest priority for KA. Because most persons undergo unilateral KA (84% of our samples were classified as having unilateral KA while 14% had bilateral KA within one year) most patients and their surgeons must decide which of two knees, when both are symptomatic, is highest priority for KA. A substantial number of these patients and their surgeons likely find this decision challenging given that between 10% and 25% of these patients have very similar OA severity and/or knee symptom severity bilaterally.

Variation over time in pain intensity within a person with knee OA has been clearly established [7,8,18]. The extent varies depending on the study and measures used. Hutchings and colleagues obtained weekly pain measures over a three month period on 287 persons with hip or knee OA and found that 49% of the sample reported one or more changes of at least 2 WOMAC Pain points (baseline WOMAC Pain mean score = 4.2) over the three-month period. Parry and colleagues found that approximately a third of the knee OA sample reported at least a 2 point change (measured on a 0 to 10 numerical pain rating scale) over an 18-month period. For persons with bilateral symptomatic knee OA and undergoing unilateral KA, intra-individual pain variation in both knees likely leads to confusion regarding which knee should undergo surgical treatment. For this select subgroup of patients, it may be that an alternative approach to the standard one-time consultation visit to an orthopaedic surgeon may be most appropriate to increase the likelihood that the most problematic knee is being treated. For example, weekly or monthly internet- or phone-based self-reported pain and disability measures could be completed to increase the likelihood that the most problematic knee is identified. For patients with similarly symptomatic knees, when a consistent pattern of one knee being most problematic is identified, both the patient and surgeon will likely have greater confidence that the most symptomatic knee is the one undergoing KA.

The current study had several strengths including the multicenter data and the sample size but there were some notable weaknesses. The time between the preoperative visit and the surgery varied and this was a result of the study design. It is possible that the varying amount of time between the preoperative data collection and the surgery impacted our findings. If this were the case, one would expect, for example, that WOMAC scores would decrease (i.e., indicate less pain or better function) as the number of days from assessment to surgery increased. In a post hoc analysis, Pearson product moment correlations between WOMAC Pain and Function scores and number of days between preoperative assessment and surgery in OAI data were generated. All associations were positive, indicating that as the days from surgery increased, WOMAC scores were increased (e.g., Pearson r for WOMAC Pain scores for the index knee and days from assessment to surgery was r = 0.20). These data suggest that time from surgery may not have impacted the findings.

Second, only the OAI collected WOMAC Physical Function data for each knee and only MOST conducted lateral knee radiographs to determine if patellofemoral OA was present. Precision of the estimates for these measurements were therefore reduced. Finally, the timeframe for classifying KA surgeries as unilateral versus bilateral was arbitrary and based only on clinical experience. It may be that some persons undergoing bilateral KA with more than one year between surgeries had actually planned on bilateral KA and some patients undergoing bilateral KA within one year had not planned on bilateral KA when the first surgery was done.

In conclusion, this appears to be the first report detailing a comprehensive profile of the knee OA and symptom severity measures for the contralateral knees in persons scheduled for KA. The analyses placed particular focus on patients scheduled for unilateral KA who had the same or worse symptom severity and knee OA in the contralateral knee. Moderate overlap was found between scores for a variety of symptom measures including pain, stiffness and functional difficulty for the contralateral versus the index knees. The estimated prevalence of the same or worse knee OA severity in the contralateral knee as compared to the index knee was 26% (95% CI = 22%, 30%). The prevalence of contralateral knees with the same or worse knee OA and knee pain as compared to the index knee was 10% (95% CI = 7%, 13%). These data support an alternative approach for this subset of patients to identify the most problematic knee for KA. One potential approach would be to obtain weekly or monthly repeated self-report measures of symptom status to determine which knee consistently is most problematic for the patient. This approach would likely reduce uncertainty for both the patient and surgeon when selecting a knee for KA when both knees have similar levels of OA and symptom severity.