Better Functional and Similar Pain Outcomes in Osteoarthritis Compared to Rheumatoid Arthritis After Primary Total Knee Arthroplasty: A Cohort Study

Authors

  • Jasvinder A. Singh,

    Corresponding author
    1. Birmingham VA Medical Center and University of Alabama at Birmingham, and Mayo Clinic College of Medicine, Rochester, Minnesota
    • University of Alabama, Faculty Office Tower 805B, 510 20th Street South, Birmingham, AL 35294. E-mail: Jasvinder.md@gmail.com

    Search for more papers by this author
    • Dr. Singh has received consultant fees, speaking fees, and/or honoraria (less than $10,000 each) from Allergan, Regeneron, and Savient and (more than $10,000) from Takeda, and has received research and travel grants from Takeda and Savient.

  • David G. Lewallen

    1. Mayo Clinic College of Medicine, Rochester, Minnesota
    Search for more papers by this author
    • Dr. Lewallen has received consultant fees, speaking fees, and/or honoraria (more than $10,000 each) from Zimmer, Orthosonic, and Osteotech, owns stock or stock options in Pipeline Biomedical, holds a patent or license from Zimmer and receives royalties from Zimmer, Orthosonic, and Osteotech for hip and knee implants, and his institution has received research funds from DePuy, Stryker, Biomet, and Zimmer.


Abstract

Objective

To determine the association of the underlying diagnosis with limitation in activities of daily living (ADL) and pain in patients undergoing primary total knee arthroplasty (TKA).

Methods

Prospectively collected data from the Mayo Clinic Total Joint Registry were used to assess the association of diagnosis with moderate–severe limitation in ADL and moderate–severe pain at 2 and 5 years after primary TKA, using multivariable-adjusted logistic regression analyses. We calculated odds ratios (ORs) and 95% confidence intervals (95% CIs).

Results

There were 7,139 primary TKAs at 2 years and 4,234 at 5 years. In multivariable-adjusted analyses, compared with rheumatoid arthritis (RA)/inflammatory arthritis, osteoarthritis (OA) was associated with significantly lower moderate–severe ADL limitation at 2 years (OR 0.5 [95% CI 0.3–0.8]) (P = 0.004) and at 5 years (OR 0.5 [95% CI 0.3–0.9]) (P = 0.02). There was no significant association of diagnosis of OA with moderate–severe pain at 2 years (OR 1.2 [95% CI 0.5–2.7]) (P = 0.68) or at 5 years (OR 1.0 [95% CI 0.3–3.7]) (P = 1.0).

Conclusion

We found that patients with OA who underwent primary TKA had better ADL outcomes compared to patients with RA/inflammatory arthritis at 2 and 5 years. On the other hand, the pain outcomes after primary TKA did not differ in OA versus RA. This suggests a discordant effect of underlying diagnosis on pain and function outcomes after TKA. These novel findings can be used to better inform both patients and surgeons about expected pain and function outcomes after primary TKA.

INTRODUCTION

Total knee arthroplasty (TKA) is an effective surgery for patients with end-stage refractory knee arthritis. It is associated with significant improvement in pain, function, and quality of life ([1, 2]). The most common underlying diagnoses for primary TKA are osteoarthritis (OA) or rheumatoid arthritis (RA) ([3, 4]). Most literature to date has focused on surgical and implant factors affecting implant outcomes (such as the risk of revision) after TKA and, in a systematic review of 40 studies, revision, infection, and dislocation rates after total joint arthroplasty were higher in RA compared to OA ([5]). There is emerging literature related to factors associated with patient-reported outcomes (PROs; pain, function, quality of life) after TKA ([6-9]). However, fewer studies have assessed the association of underlying diagnosis with PROs after TKA.

The current literature regarding the effect of underlying diagnosis on pain and function outcomes after primary TKA is contradictory. Sledge and Walker found that range of motion and pain were similar between RA and OA in 798 TKAs with a minimum 2-year followup ([10]). On the other hand, Kirwan et al studied 293 patients who underwent total hip or total knee replacement at 2.5 years and found greater improvements in OA patients in pain and function compared to RA patients ([11]). These studies consisted of small sample sizes and often did not adjust for important confounders, making the results potentially biased. A recent study of 3,608 patients found that compared to OA, RA was associated with better pain outcomes at 6 months post–primary TKA, but had no association with function outcomes ([6]). However, pain and functional improvements continue to evolve greatly after 6 months post–primary TKA ([12-14]), and therefore a study of short-term outcomes does not tell us about the intermediate or long-term pain and function outcomes and their predictors. Another study that combined data from 1974–1998 from a UK surgeon group showed that in a sample of 812 patients with 5-year followup, composite pain and function score was worse in RA versus OA ([15]). Thus, it is not clear whether RA is associated with worse pain and function outcomes after TKA compared to OA. The purpose of this study was to assess the association of diagnosis with TKA pain and function outcomes up to 5 years postoperative by using a large sample from a prospective total joint registry and performing analyses adjusted for important covariates and confounders. We hypothesized that patients with OA will have better pain and functional outcomes after primary TKA compared to patients with RA/inflammatory arthritis as the underlying diagnosis. Our hypothesis was based on a lower anticipated complication rate in patients with OA in absence of systemic inflammation and/or immunosuppressive medications.

Box 1. Significance & Innovations

  • Patients with osteoarthritis (OA) had better functional outcomes compared to patients with rheumatoid arthritis (RA)/inflammatory arthritis at 2 and 5 years after primary total knee arthroplasty (TKA).
  • Pain outcomes after primary TKA did not differ between OA and RA patients.
  • Underlying diagnosis has a discordant effect on pain versus functional outcomes after primary TKA.

PATIENTS AND METHODS

Methods

We describe the methods and results as recommended in the Strengthening of Reporting in Observational Studies in Epidemiology statement ([16]).

Settings, patients, and data sources

This study included all patients who underwent primary TKA at the Mayo Clinic, Rochester, Minnesota, from 1993 to 2005 and who responded to the Mayo Knee Questionnaire at the 2- or 5-year followup. The Mayo Knee Questionnaire is a validated, standardized knee questionnaire that has construct validity and reproducibility ([17]). A very high correlation (correlation coefficient 0.74) between mailed and in-person, physician-administered Mayo Knee Questionnaires has been demonstrated ([17]). This questionnaire is similar to the Knee Society Scale ([18]), a validated scale ([19, 20]) that is most commonly used in knee arthroplasty studies ([21]). Several studies using data from this validated questionnaire have been published ([8, 9, 22-24]). It assesses knee pain and function and is administered to all patients who undergo TKA at the Mayo Clinic. It is administered by mail, phone call, or during an in-person clinic visit at the 2- and 5-year time points post-TKA by experienced, dedicated joint registry staff. Pain and function data have been captured in the Mayo Clinic Total Joint Registry electronically starting in 1993.

We used data from the Mayo Clinic Total Joint Registry, which prospectively collects survey and other data electronically on all joint replacements performed at the Mayo Clinic. Data included patient demographics, operative diagnosis, surgery and implant details, dates of evaluation, reoperations and complications, current radiographs, and pain and function assessments ([25, 26]). The study was approved by the Institutional Review Board at the Mayo Clinic. Since it was a database study, informed consent was waived.

Predictor of interest

Our main variable of interest was the operative diagnosis. It was categorized as OA, RA (including other inflammatory arthritis), and other, as previously described ([8, 24]).

Covariates

Several covariates known to be associated with pain and function after TKA ([27-32]), as well as potential confounders, were obtained from the Mayo Clinic Total Joint Registry and linked databases and included in our analyses; namely: demographics (age, sex), body mass index (BMI), American Society of Anesthesiologist (ASA) class, implant fixation (uncemented, hybrid/cemented), and distance from the medical center. ASA class is defined as follows: I = a normal healthy patient, II = a patient with mild systemic disease, III = a patient with severe systemic disease, and IV = a patient with severe systemic disease that is a constant threat to life ([33]). We also included anxiety, depression, and medical comorbidity, assessed using the validated Deyo-Charlson Index ([34]) based on the presence of International Classification of Diseases, Ninth Revision codes in the Mayo Clinic electronic databases at the time of surgery. The activities of daily living (ADL) limitation multivariable model was also adjusted for preoperative limitation in 3 ADL and the pain multivariable model was adjusted for preoperative pain severity.

Outcomes of interest

The outcomes of interest were moderate–severe ADL limitation and moderate–severe pain obtained from the self-reported validated Mayo Knee Questionnaires at 2 years or 5 years. We defined these outcomes a priori as undesirable, as in previous studies ([8, 9, 22-24]), since the main purpose of TKA is to relieve pain and improve function. Three key ADL, including walking, climbing stairs, and rising from a chair to a standing position, were assessed via survey, with limitations categorized into “no,” “mild,” “moderate,” or “severe” limitation for each ADL, as previously described ([24]). Presence of ≥2 activities with moderate or severe limitation was classified as overall moderate–severe activity limitation (reference, all other categories), as in previous studies ([9, 24]). Postoperative index knee pain was assessed with the question, “do you have pain in the knee in which the joint was replaced?” The responses were: no pain, mild (occasional), stairs only, walking, and stairs (all combined into reference category); moderate (occasional), moderate (continuous), and severe categories combined into the outcome variable, moderate–severe pain. This pain question is the same as in the validated Knee Society Scale ([17-19]).

Bias, sample size, and statistical analyses

We accounted for the correlation of observations (due to bilateral TKA in a patient, simultaneously or sequential) using a generalized estimating equation approach to adjust the SEs for the correlation between observations on the same subject due to both knees having been replaced ([35]), which is an appropriate statistical method. We selected a large enough sample to study these outcomes without performing formal sample-size calculations by choosing all eligible patients from 1993–2005. With >7,000 patients and >4,000 patients at 2 and 5 years, respectively, and >300 patients with moderate pain and >1,000 patients with moderate–severe ADL limitation, our post hoc calculations confirmed, using a rule of 10 cases per variable, that we had enough power for multivariable regression analyses that included up to 13 variables.

In this study we used univariate and multivariable-adjusted logistic regression models to assess the association between operative diagnosis and moderate–severe pain and overall moderate–severe ADL limitation 2 and 5 years after primary TKA. All multivariable models included age, sex, BMI, ASA class, distance from the medical center, implant fixation, Deyo-Charlson Index score, anxiety, and depression to decrease confounding bias. In addition, ADL limitation multivariable models included preoperative limitation in 3 key ADL and the pain multivariable models included preoperative pain severity. Odds ratios (ORs), 95% confidence intervals, and P values were reported. We specified a priori and examined each ADL limitation as exploratory analyses, in order to avoid multiple comparisons. Responder and nonresponder characteristics were compared using logistic regression analyses. A P value less than 0.05 was considered statistically significant.

RESULTS

For the 2-year cohort, the mean age was 68 years, 56% were women, and 18% were ages ≤60 years (Table 1). BMI was ≥30 kg/m2 in 52% of patients and ASA score was class III/IV in 42%. The mean Deyo-Charlson score was 1.8 and depression and anxiety were present in 11% and 6% of patients, respectively. The 5-year cohort had similar characteristics (Table 1). The survey response rates were 65% (7,139 of 10,957 TKAs) at 2 years and 57% (4,234 of 7,404 TKAs) at the 5-year followup.

Table 1. Clinical and demographic characteristics*
 Primary TKA
2-year cohort (n = 7,139)5-year cohort (n = 4,234)
  1. Values are the percentage unless indicated otherwise. Numbers are rounded to the nearest digit for proportions, so the total may not add up to exactly 100%. TKA = total knee arthroplasty; ASA = American Society of Anesthesiologists; RA = rheumatoid arthritis; ADL = activities of daily living.
  2. aIncludes avascular necrosis, septic arthritis, bone tumor of malignancy, dislocation/instability/bone fracture, failure of previous surgery, etc.
Age, mean ± SD years68 ± 1068 ± 10
Men/women44/5645/55
Age groups  
≤60 years1818
>60–70 years3537
>70–80 years3838
>80 years87
Body mass index (kg/m2)  
<251313
25–29.93536
30–34.92943
35–39.9147
≥4097
ASA class  
Class I/II5858
Class III/IV4241
Deyo-Charlson Index score, mean ± SD1.2 ± 1.91.1 ± 1.9
Psychological comorbidity  
Anxiety65
Depression118
Underlying diagnosis  
Osteoarthritis9493
RA/inflammatory arthritis44
Othera23
Preoperative pain severity  
None/mild3638
Moderate–severe6462
Preoperative ADL limitation  
None/mild3231
Moderate–severe6869

For primary TKA 2- and 5-year followup, men and those with OA as the underlying diagnosis were slightly more likely to respond and older age was associated with significantly greater odds of response (see Supplementary Table 1, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.22090/abstract). Higher ASA class (III or IV) and higher Deyo-Charlson comorbidity score was associated with slightly lower odds of response, while a distance of >500 miles from the Mayo Clinic was associated with much lower odds of response. The distribution of preoperative pain and ADL limitation is shown in Table 2.

Table 2. Association of underlying diagnosis with preoperative pain severity and ADL limitation*
 OARA/inflammatory arthritisOther
  1. ADL = activity of daily living; OA = osteoarthritis; RA = rheumatoid arthritis.
  2. aP = 0.21 by chi-square test.
  3. bP < 0.001 by chi-square test.
  4. cP = 0.27 by chi-square test.
  5. dP < 0.001 by chi-square test.
2-year cohort   
Preoperative pain severitya   
None/mild36.030.340.9
Moderate–severe64.069.759.1
Preoperative ADL limitationb   
None/mild32.320.114.9
Moderate–severe67.779.985.1
5-year cohort   
Preoperative pain severityc   
None/mild38.432.232.9
Moderate–severe61.667.867.1
Preoperative ADL limitationd   
None/mild32.611.517.1
Moderate–severe67.488.582.9

Univariate association of diagnosis with outcomes at 2 and 5 years

We noted that compared to those with RA/inflammatory arthritis, OA was associated significantly with lower odds of overall moderate–severe ADL limitation post-TKA (Table 3). On the other hand, moderate–severe pain at 2 years or 5 years was not significantly different in OA versus RA patients (Table 3). Patients with OA reported significantly lower limitations in walking, climbing stairs, and rising from a chair, with ORs 0.4–0.6 at 2 years post–primary TKA and ORs 0.6–0.7 at 5 years post–primary TKA, as compared to those with RA (see Supplementary Table 2, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.22090/abstract).

Table 3. Univariate association of diagnosis with ADL limitation and pain outcomes at 2 and 5 years*
 No./total no. (%)2 yearsNo./total no. (%)5 years
OR95% CIPOR95% CIP
  1. Reference (ref.) category is inflammatory arthritis, including rheumatoid arthritis (RA). ADL = activities of daily living; OR = odds ratio; 95% CI = 95% confidence interval.
  2. aIncludes avascular necrosis, septic arthritis, bone tumor of malignancy, dislocation/instability/bone fracture, failure of previous surgery, etc.
Moderate–severe overall ADL limitation        
RA/inflammatory arthritis82/239 (34.3)1.0 (ref.)  75/181 (41.4)1.0 (ref.)  
Osteoarthritis1,377/6,354 (21.7)0.50.4–0.7< 0.011,064/3,739 (28.5)0.60.4–0.8< 0.01
Othera50/158 (31.6)0.90.6–1.40.6034/118 (28.8)0.60.3–1.00.04
Moderate–severe knee pain        
RA/inflammatory arthritis22/243 (9.1)1.0 (ref.)  17/185 (9.2)1.0 (ref.)  
Osteoarthritis462/6,473 (7.1)0.80.5–1.30.31306/3,792 (8.1)0.90.5–1.50.62
Othera20/164 (12.2)1.40.7–2.80.348/117 (6.8)0.70.3–1.80.49

Multivariable-adjusted association of diagnosis with outcomes

In adjusted analyses, we found that compared to patients with RA/inflammatory arthritis, those with OA had significantly lower odds of overall moderate–severe ADL limitation at 2 years and 5 years (Table 4). On the other hand, the odds of moderate–severe pain did not differ by diagnosis (Table 4). Univariate associations noted for each ADL limitation were attenuated in multivariable-adjusted analyses (see Supplementary Table 3, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.22090/abstract).

Table 4. Multivariable-adjusted association of diagnosis with ADL limitation and pain outcomes at 2 and 5 years*
 2 years5 years
OR95% CIPOR95% CIP
  1. Reference (ref.) category is inflammatory arthritis, including rheumatoid arthritis (RA). ADL = activity of daily living; OR = odds ratio; 95% CI = 95% confidence interval.
  2. aAdjusted for 12 additional covariates/confounders: age, sex, body mass index (BMI), Deyo-Charlson comorbidity score, American Society of Anesthesiologists (ASA) class, distance from the medical center, cement fixation, preoperative limitation in 3 activities, anxiety, and depression.
  3. bIncludes avascular necrosis, septic arthritis, bone tumor of malignancy, dislocation/instability/bone fracture, failure of previous surgery, etc.
  4. cAdjusted for 10 additional covariates/confounders: age, sex, BMI, Deyo-Charlson comorbidity score, ASA class, distance from the medical center, cement fixation, preoperative pain, anxiety, and depression.
Overall moderate–severe ADL limitationa      
RA/inflammatory arthritis1.0 (ref.)  1.0 (ref.)  
Osteoarthritis0.60.4–0.90.020.60.3–1.00.05
Otherb1.10.5–2.20.810.60.3–1.40.26
Moderate–severe knee painc      
RA/inflammatory arthritis1.0 (ref.)  1.0 (ref.)  
Osteoarthritis0.90.5–1.80.781.20.5–2.70.68
Otherb1.20.4–3.30.771.00.2–3.70.96

DISCUSSION

Several findings in our study are of interest and deserve further discussion. One of the main findings of our study was that patients with OA had significantly better functional outcomes compared to those with RA at the 2- and 5-year followups. Overall moderate–severe ADL limitation was less frequent in OA versus RA patients at both 2 years and 5 years. On the other hand, we found no significant differences in risk of moderate–severe pain by underlying diagnosis in multivariable-adjusted models. These findings add to the current literature, which consists of mostly small studies and somewhat contradictory findings. Kirwan et al studied 293 patients at 2.5 years and reported greater improvements in pain and function in OA patients compared to RA patients ([11]). However, the group consisted of both TKA and total hip replacement patients. Sledge and Walker studied 794 TKAs with a minimal followup of 2 years and reported the following in OA versus RA patients: 95% versus 92% for pain relief and 90% versus 80% for stair climbing without support ([10]). The analyses were unadjusted and pain relief was not defined. In a study of short-term outcomes, Judge et al studied 3,608 patients with primary TKA and found that compared to OA, RA was associated with better pain but similar functional outcomes at 6 months post–primary TKA ([6]). In a study of 2,032 patients over 3 decades from 1974 that had a 40% response rate, better composite pain and function outcome in OA versus RA at 5 years post-TKA was shown ([15]). Treatment paradigms for OA and RA have changed since 1974 and pain and functional status improvements do not peak as early as 6 months after TKA ([12-14]); therefore, the assessments in these studies represent early outcomes that are still evolving ([6]) and/or results that may not be generalizable 15–30 years later ([15]). Thus, our findings from robust multivariable-adjusted analyses of a large sample from the Mayo Clinic Total Joint Registry at 2 time points add to the current literature.

The greater ADL limitation noted in RA patients compared to OA patients may be due to several potential reasons. Compared to OA, RA has a higher rate of complications such as infection and revision dislocation after primary TKA ([5]). It is also possible that systemic inflammation, a hallmark of RA ([36]), and the polyarticular nature of RA interfere with optimal rehabilitation after primary TKA, which is key to optimal ADL recovery. Since our analyses were adjusted for preoperative ADL limitation, preoperative differences in ADL do not explain the differences in outcomes postoperatively.

These findings have several practical implications. First, this new information can help surgeons inform their patients at the time of informed consent with regard to expected pain and function outcomes after primary TKA, based on their underlying diagnosis. Second, these findings should trigger more studies that investigate the underlying mechanisms explaining these observations, which will then lead to new approaches to improve outcomes in patient groups at risk for worse outcomes (RA patients). This can lead to even better outcomes after primary TKA, a procedure that improves pain, function, and quality of life of patients ([37]).

Our study has several strengths and limitations. Study strengths include prospective data collection by dedicated joint registry staff, multivariable-adjusted analyses, and a large sample size to allow adequate power to test hypotheses. Our findings must be interpreted considering study limitations. Nonresponse may have introduced selection bias. Nonresponders were younger, women, and had higher comorbidity, ASA class, and lived further from the medical center, which implies that had nonresponders to the survey had overall worse pain and functional outcomes and had everyone responded, we would have noted slightly worse pain and function outcomes. However, to our knowledge, there are no published data guiding whether any of these factors also impact the association of OA with pain and functional outcomes (interaction); therefore, the direction of this bias related to nonresponse is unclear. Generalizability of findings to other settings may not be feasible; however, the cohort's clinical and demographic characteristics are similar to those reported in other TKA studies ([38-40]). Our 2-year response rate of 65% was similar to the average 60% response rate reported for large surveys of this size (patient, physician, and others) and the 68% response rate reported for patient surveys ([41]), and was slightly lower than that reported for funded national systems such as the National Health System at 70%. Nonresponse bias should still be taken into account while interpreting study findings. The 5-year estimates should be interpreted with caution due to a lower response rate of 57%. Due to a cohort study design, residual confounding is possible despite inclusion of multiple clinical and demographic variables. Missing data for covariates or confounders were small (<5%), except preoperative variables, which were missing due to nonresponse bias. We had decided a priori not to impute and treat missing data as such, as a conservative approach and given the large sample and enough outcome events.

In conclusion, our study shows that compared to RA, OA is associated with better functional outcomes up to 5 years post-TKA. We also found that pain outcomes after primary TKA were similar in OA and RA patients. Future studies should investigate the reasons for poorer functional outcome in RA patients undergoing primary TKA. More insights into this finding can help us better understand and even improve further pain and function outcomes after primary TKA.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Singh had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Singh, Lewallen.

Acquisition of data. Singh, Lewallen.

Analysis and interpretation of data. Singh, Lewallen.

Ancillary