Importance of self-rated health and mental well-being in predicting health outcomes following total joint replacement surgery for osteoarthritis

Authors


Abstract

Objective

The determinants of outcomes and the scope of outcomes examined in total joint replacement (TJR) typically have been limited to aspects of physical health. We investigated mental well-being, physical and social health, and self-rated health (SRH) as predictors of future health status within a cohort undergoing a TJR for hip or knee osteoarthritis. We also investigated the interrelationships among these health dimensions as they relate to SRH.

Methods

Participants (n = 215 hip, n = 234 knee) completed measures presurgery and 3 and 6 months postsurgery, including pain, physical function, fatigue, anxiety, depression, social participation limitations, passive/active recreation, community mobility, and SRH. Structural equation modeling was used to investigate the interrelationship between 3 latent health dimensions (physical, mental, social) and the predictive significance of SRH for future health status.

Results

The mean age was 63.5 years (range 31–88 years) and 60% were women. Prior dimension status strongly predicted future status. Adjusted for prior dimension scores, comorbidity, and sociodemographic characteristics, SRH predicted future scores for all 3 health dimensions. Worse prior SRH predicted less improvement at all time points. The effects of physical and social health on SRH were fully mediated through mental well-being. Only mental well-being significantly predicted SRH, within and across time.

Conclusion

Mental well-being is critical for understanding the relationship between physical health and SRH. In addition, SRH significantly predicts TJR outcomes, above and beyond prior physical health. The exclusive focus on any one health dimension may lead to missed opportunities for predicting and improving outcomes following surgery, and likely improving overall health generally.

INTRODUCTION

Osteoarthritis (OA) is the most common joint disorder in the world and is the leading cause of disability in Western populations. When conservative treatment fails to alleviate hip and knee joint pain and dysfunction caused by OA, total joint replacement (TJR) is an elective surgical option that can provide significant pain relief and improved function with proven cost-effectiveness (1–3). However, while this is generally the case, a number of studies have documented significant variability in the degree of improvement in outcomes following TJR, with a significant minority of patients receiving TJR reporting little or no improvement in pain and function (4–6).

Different types of outcomes have been used to evaluate TJR procedures. Traditionally, these outcomes centered on surgeon perceptions, were technically oriented, and focused primarily on pain and function. More recently, a greater emphasis has been placed on patient-centered outcomes and health-related quality of life (3).

Improvements in pain and function following hip and knee TJR have been well established (2, 7, 8). There is some work that has reported on improvements in other dimensions of health as well, including mental health and social functioning (2, 9, 10). However, while there has been work that has examined the determinants of these outcomes, the primary focus has been on evaluating demographic and socioeconomic characteristics and preoperative comorbidity and pain and function as determinants of postoperative pain and function (11–13). These factors have modestly, although significantly, explained variations in postoperative pain and function. Prosthetic and procedure-related factors also have been examined as determinants (4, 13). More recently, patient expectations and preoperative psychological factors have been discussed and assessed as determinants of postoperative pain and function and health-related quality of life generally (14–16).

Although individual determinants of hip and knee TJR outcomes have been identified, what has been neglected is the concomitant consideration of patient-reported mental and social health status with pain and physical function status. Moreover, a number of studies have documented the predictive significance of self-rated health (SRH) for a number of health outcomes, including social–psychological well-being, morbidity, health care utilization, and mortality (17–19). These findings have been relatively consistent across varying chronic condition groups, clinical and community cohorts, and varied time periods, even after controlling for a variety of health and health-related factors. However, SRH has not been fully considered in TJR patient-reported outcomes. While changes in SRH in response to surgery have been examined (20–22), the influence of pre- and postoperative SRH on subsequent outcomes has not been fully investigated. Long et al (23) examined the impact of preoperative SRH on postsurgical Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC)–defined health categories (physical aspects of health) and found that patients reporting better preoperative SRH reported greater improvement in this dimension of health. The aim of this study was to identify novel determinants of TJR outcomes and identify potential areas for targeted intervention beyond physical aspects of health that may improve the overall health of patients with arthritis and TJR. We longitudinally examined the predictive significance of SRH for future health status (i.e., physical health, mental well-being, and social health [labeled health dimensions]) in a cohort of individuals within 6 months of undergoing TJR surgery for hip or knee OA. Prior to investigating these longitudinal relationships, we explored the interrelationship between the health dimensions as they relate to SRH.

Significance & Innovations

  • In addition to physical aspects of health, mental well-being and self-rated health also predict a range of health outcomes following total joint replacement surgery.

  • The effects of physical health on self-rated health appear to be mediated through mental well-being.

  • Since self-rated health significantly predicted total joint replacement outcomes, this study suggests that the exclusive focus on any one health dimension may lead to missed opportunities for predicting and improving outcomes following surgery.

MATERIALS AND METHODS

Individuals undergoing primary unilateral TJR for hip or knee OA were consecutively recruited from 4 Toronto, Ontario, Canada academic hospitals. Eligibility criteria included age ≥18 years and ability to read and comprehend English. Individuals having surgery for other than OA, a hemiarthroplasty, or a revision arthroplasty were ineligible.

The study was approved by the research ethics board of each participating institution. Written informed consent was obtained from all of the study participants. Mailed health questionnaires were completed presurgery and 3 and 6 months postsurgery.

SRH.

SRH was serially captured by the question: “In general, would you say your health is …,” with the responses excellent, very good, good, fair, and poor, and scored from 1 to 5. SRH was treated continuously. It is not uncommon for an ordinal variable with ≥5 levels, particularly when it is viewed as a proxy for an underlying continuum, as is SRH (24), to be operationalized as such. Furthermore, since repeated measures were being examined, it was particularly important to account for reliability. We specified SRH to have a reliability of 0.78, the average of reported estimates from population and clinical trial data (which operationalized SRH as continuous) (25, 26).

Health status measures (serially collected).

Pain on activity was assessed with the Hip Disability and Osteoarthritis Outcome Score (HOOS) (27) and the Knee Injury and Osteoarthritis Outcome Score (KOOS) (28) pain subscales. These measures assess the frequency and extent of pain during activities such as “walking on a flat surface” and “going up/down stairs.” The measures have documented reliability, validity, and responsiveness in TJR (27, 28).

Limitations in activities of daily living were captured with the WOMAC physical function subscale. A commonly used measure in TJR (29) with demonstrated validity and responsiveness (29, 30), this measure assesses individuals' ability to move around and look after themselves, eliciting the degree of difficulty experienced due to their hip or knee.

Limitations in more physically demanding activities were assessed with the HOOS/KOOS function in sport and recreation subscale (27, 28). Individuals rate their degree of difficulty when active on a level beyond that captured by the WOMAC. These subscales have documented reliability, validity, and responsiveness (27, 28).

Fatigue was assessed with the Profile of Mood States (POMS) fatigue subscale (31). The POMS is a frequently used measure of fatigue in the literature and has been used to study fatigue in a range of chronic conditions. The POMS has documented validity (32).

Anxiety and depression were captured with the Hospital Anxiety and Depression Scale (HADS) (33). The HADS has been widely used in outpatient populations (34), and the reliability of each subscale is sufficient for group comparisons in multiple populations (34).

Social participation limitations were assessed with the Late Life Function and Disability Instrument limitation subscale, disability component (35). Respondents rated the extent to which they felt limited in their ability to personally perform in 16 socially expected life tasks. The dimension's reliability and validity have been demonstrated (35).

Difficulty traveling within the community/neighborhood and difficulty with recreation/leisure activities were derived from an adaptation of the Calderdale Health and Disablement Survey (36). The questionnaire was designed to assess the extent to which a respondent's condition limited their mobility or ability to travel within their community and the extent of difficulties in participating in social and leisure activities.

For comorbidity, individuals responded yes/no to a list of 14 medical problems identified in the Self-Administered Comorbidity Questionnaire (37).

Time-independent covariates included age at surgery, sex, hip/knee, and level of education (high school or less, postsecondary or more).

For ease of comparison, all measure scores (except SRH and comorbidity count) were transformed to a 0–100 scale, where higher scores indicated worse health/more difficulty.

Statistical analyses.

Structural equation modeling was used throughout with Mplus, version 3.13. There were a number of advantages to this approach. First, there was an account for the measurement error associated with each measured variable, resulting in a more accurate and robust estimate of the relationships among the latent variables (later described). Second, the measured variables for each latent construct were handled as if they were elements of a factor, eliminating the problem of multicolinearity. Third, pertaining to the longitudinal aspect, error terms of repeated measures could be specified to co-vary over time, accounting for “method effects”; the measurement properties of the model (i.e., the relationships between indicators and latent variables) could be examined over time to ensure their stability, and consequently the proper interpretation of change over time; and the relationship between repeated latent constructs and repeated outcomes could be explicitly examined over time.

For our confirmatory factor analysis and overall structural equation modeling model, 4 indices were used to determine overall model fit: root mean square error of approximation (RMSEA), standardized root mean square residual (SRMR), comparative fit index (CFI), and Tucker-Lewis index (TLI). Good fit was supported by RMSEA ≤0.05 (with a 90% upper confidence limit <0.08) and a nonsignificant P(RMSEA ≤0.05), SRMR ≤0.08, and CFI and TLI ≥0.95 (38). Standardized covariance residual matrices, modification indices, and expected parameter change values also were examined. The final model was assessed using robust maximum likelihood to account for any minor departures in non-normality and obtain reliable statistical results.

Latent health dimensions.

Three latent health variables were developed. For physical health, indicators included pain on activity, limitations in activities of daily living, limitations in more physically demanding activities, and fatigue; for mental well-being, indictors included anxiety, depression, and fatigue; and for social health, indicators included social participation limitations, difficulty traveling within the community/neighborhood, and difficulty with recreation/leisure activities. Confirmatory factor analyses confirmed that the health measures corresponded with 3 health dimensions and confirmed the longitudinal stability of our measurement model (39), ensuring that temporal changes in health dimension scores were due to true changes and not changes in the measurement structure over time.

Since comorbidity count was not a measure of degree of severity or extent of symptoms, it was specified as a predictor of the health dimensions along with the covariates.

Overall model.

To examine the interrelationship between the health dimensions as they related to SRH, a structural regression model was specified whereby SRH was initially regressed on all of the health dimensions; the health dimensions were allowed to co-vary. The covariates and comorbidity count were specified as predictors of SRH and the health dimensions. Subsequently, exploring the potential for mediated effects between the health dimensions and SRH, analyses were undertaken whereby initially each of the health dimensions was considered as the only predictor of SRH (adjusted for covariates). This was followed by analyses examining the significance of pair combinations of the health dimensions as predictors of SRH (adjusted for covariates). The analyses were replicated at each of the 3 time points.

To test the predictive significance of SRH for physical health, mental well-being, and social health, a model was specified whereby the health dimension statuses at the 3- and 6-month time points were regressed on prior SRH. This model controlled for comorbidity count and the covariates. Additionally, to ensure that the relationships of interest were not confounded, this model also controlled for prior health dimension status, prior SRH status, and the effect of changes in the health dimensions over time on future SRH. Repeated measures were specified to have their error variances co-vary over time.

RESULTS

A total of 529 individuals were enrolled in the study. Of these, 449 individuals (85%; 215 hips and 234 knees) completed the study, while 35 individuals were lost to followup and 45 dropped out. In comparative analyses of baseline values between the analytical and “out-of-study” sample, “out-of-study” individuals reported lower education levels (although only 30 of the 80 “out-of-study” individuals reported their level of education), were older (mean age 63.5 versus 67.6 years), reported greater pain on activity (mean 59 versus 66), and reported more difficulty with travel in the community (mean 45 versus 56). Although statistically significant, the magnitude of the differences was not substantial considering the wide range of the scales. While the distribution differences were greatest for level of education, the analytical sample nonetheless covered well the range of education levels, with nearly one-third of the sample reporting the more infrequent category of high school education or less. For all of the health measures considered, the range of values among the analytical sample was greater and substantially overlapped the ranges found among the “out-of-study” sample.

Near equal proportions of hip and knee surgeries were represented (47.9% and 52.1%, respectively). Women comprised 60% of the sample. The mean age of the sample was 63.5 years (median 64.0 years, range 31–88 years). The majority (72%) reported postsecondary education. Presurgery, 13.5% reported “excellent” SRH, 36% and 37% reported “very good” and “good” SRH, respectively, and 13.7% reported “fair” or “poor” SRH.

The mean health measure scores across time are shown in Table 1. As expected, presurgical scores for the physical health dimension measures showed a high level of severity in this TJR sample. The mean presurgical anxiety score was relatively similar to the mean score derived from normative data provided for the HADS. The mean depression score, however, was higher than that from normative data. Similar levels of limitation were observed between the social health dimension measures. A mean score of 41.4 for social participation limitations reflected a moderate to severe level of limitation (35).

Table 1. Sample health measure scores at presurgery and 3 and 6 months postsurgery (n = 449)*
 Presurgery3 months postsurgery6 months postsurgery
  • *

    Values are the mean ± SD. Higher scores represent greater symptoms/difficulties or worse health. ADL = activities of daily living.

Pain on activity58.9 ± 17.0924.1 ± 18.1420.5 ± 17.90
Limitation in ADL52.5 ± 17.6922.5 ± 16.4219.1 ± 15.68
Limitation in physically more demanding activities84.3 ± 15.7663.7 ± 27.9457.0 ± 29.14
Fatigue42.0 ± 27.4924.8 ± 22.0824.5 ± 22.88
Anxiety30.3 ± 18.4618.4 ± 17.1418.0 ± 17.78
Depression25.5 ± 16.6616.6 ± 16.8014.9 ± 15.54
Social participation limitations41.4 ± 11.2227.7 ± 17.424.8 ± 16.86
Difficulty traveling in the community45.1 ± 24.7322.8 ± 24.7318.1 ± 22.07
Difficulty with recreation/leisure activities52.2 ± 23.2524.1 ± 23.6618.5 ± 19.69

With the cross-sectional analyses of SRH regressed on all of the health dimensions, adjusted for comorbidity count and covariates, the models displayed very good overall fit at each time point (RMSEA range 0.043–0.059, P[RMSEA ≤0.05] range 0.1830–0.7420, CFI and TLI range 0.959–0.982, and SRMR range 0.028–0.030). Age, sex, and surgical joint were not significantly associated with SRH. Lower education was significantly associated with poorer SRH, as was a greater number of comorbid conditions. Consistently, mental well-being was the only health dimension to independently predict SRH (in all tables, standardized estimates represent the SD change in the dependent variable for a unit SD change in the predictor variable) (Table 2). However, while the effects of physical and social health on SRH were insignificant, significant intercorrelations were observed between each of the dimensions.

Table 2. Path coefficients: structural equation model predicting self-rated health (SRH) in cross-sectional analyses
 Presurgery3 months postsurgery6 months postsurgery
Standardized estimatePStandardized estimatePStandardized estimateP
Health dimensions simultaneously considered as predictors of SRH      
 Physical health0.0420.7482−0.1740.24360.0740.6206
 Mental well-being0.323< 0.00010.449< 0.00010.3480.0005
 Social health0.0670.62410.2240.17380.2040.2176
Health dimensions as independent predictors of SRH      
 Physical health0.240< 0.00010.401< 0.00010.485< 0.0001
 Mental well-being0.406< 0.00010.491< 0.00010.576< 0.0001
 Social health0.266< 0.00010.435< 0.00010.498< 0.0001
Pairwise combinations of health dimensions as predictors of SRH      
 Physical health0.0130.86500.0700.55050.1770.2497
 Mental well-being0.420< 0.00010.554< 0.00010.759< 0.0001
 Social health0.0850.23400.1110.33350.1840.0875
 Mental well-being0.334< 0.00010.4190.00030.430< 0.0001
 Physical health0.0700.64000.0270.86800.2280.1610
 Social health0.2140.04400.4450.00500.3160.0490

When considered independently, each of the health dimensions significantly predicted SRH across time points (Table 2). However, when physical health and social health were each considered in combination with mental well-being, only mental well-being remained significant (Table 2). The intercorrelation between these health dimensions remained significant at all time points, however. This suggested that the effects of both physical and social health on SRH were being mediated through mental well-being. Furthermore, when physical health and social health were considered in combination, only social health remained a significant predictor of SRH (Table 2). Again, the intercorrelation between these two dimensions remained significant, further suggesting that the effects of physical health on SRH were mediated through social health.

Fit statistics for the longitudinal model that investigated whether past SRH predicted changes in the health dimensions over time indicated a very well-fitting model (CFI = 0.964, TLI = 0.952, RMSEA = 0.049 [90% confidence interval 0.041–0.057], P[RMSEA ≤0.05] = 0.590, and SRMR = 0.058). Figure 1 is a diagrammatic representation of this model. This model longitudinally adjusted for past health status, the effects of past SRH on future SRH, and the relationship between the health dimensions and SRH within time points. Although not shown in the figure, this model also adjusted for the covariates and comorbidity count.

Figure 1.

Depiction of final model using longitudinal analysis. Note that only statistically significant paths and covariations are shown. Covariate effects are not shown. SRH = self-rated health.

Having adjusted for past health dimension scores, past SRH significantly predicted future physical health, mental well-being, and social health status at each of the followup time points, with the exception of social health, by 6 months (Table 3). In all cases, worse past SRH predicted worse future health dimension scores. The effects were greatest at 3 months postsurgery.

Table 3. Path coefficients: final model using longitudinal analyses
Path coefficientsPresurgery3 months postsurgery6 months postsurgery
EstimateStandardized estimatePEstimateStandardized estimatePEstimateStandardized estimateP
 Past self-rated health → current health status        
Presurgery → 3 months postsurgery         
 Physical health   5.5650.281< 0.0001   
 Mental well-being   5.7150.3260.0007   
 Social health   5.8250.448< 0.0001   
3 months postsurgery → 6 months postsurgery         
 Physical health      1.9870.1130.0166
 Mental well-being      2.1460.1310.0483
 Social health      −0.607−0.0450.4827
 Past health status → current self-rated health        
Presurgery → 3 months postsurgery         
 Physical health   −0.016−0.2810.0983   
 Mental well-being   −0.014−0.2510.0143   
 Social health   0.0380.2610.1802   
3 months postsurgery → 6 months postsurgery         
 Physical health      0.0160.3100.2572
 Mental well-being      −0.024−0.4060.0090
 Social health      −0.023−0.2960.3926
 Repeated measures (e.g., self-rated health [time 1] → self-rated health [time 2])        
Presurgery → 3 months postsurgery         
 Self-rated health   0.8190.835< 0.0001   
 Physical health   0.3260.286< 0.0001   
 Mental well-being   0.3890.395< 0.0001   
 Social health   0.5640.294< 0.0001   
 Comorbidity   1.0800.914< 0.0001   
3 months postsurgery → 6 months postsurgery         
 Self-rated health      1.0420.994< 0.0001
 Physical health      0.7010.807< 0.0001
 Mental well-being      0.6740.738< 0.0001
 Social health      0.9130.897< 0.0001
 Comorbidity      0.9910.966< 0.0001
 Current health status → current self-rated health)        
Physical health0.0080.1440.4758−0.003−0.0550.7665−0.015−0.2580.3002
Mental well-being0.0180.3120.00260.0230.4190.00230.0210.3310.0244
Social health0.0030.0210.9283−0.005−0.0700.74140.0290.3740.2016

Longitudinally, SRH was responsive only to changes in mental well-being (Table 3); the greater the improvement in mental well-being, the greater the improvement in SRH during the recovery period. Within time points, mental well-being remained the only health dimension that predicted SRH (Table 3). However, as was the case in the cross-sectional analyses, the intercorrelations between the health dimensions remained significant at all time points in the longitudinal model. This continued to support the view of a mediated effect of physical and social health on SRH through mental well-being. Finally, as expected, past health dimension status strongly predicted future health dimension scores (Table 3).

Covariate effects were predominantly significant at baseline. Results are summarized by health dimension and are followed by effects on SRH; these are shown in Supplementary Appendix A with an accompanying table (available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658).

DISCUSSION

This study considered SRH as a potential determinant of health outcomes following total hip and knee TJR for OA. Having taken into account past health status, changes in health status, and the association between health dimensions, sociodemographic characteristics, type of surgery, and comorbidity, past SRH significantly predicted future health status within the 6 months following TJR. In fact, SRH predicted changes in physical health, mental well-being, and social health. Not surprisingly, poorer past SRH predicted poorer future health status. This is consistent with earlier work that has shown fair/poor SRH to predict a number of negative health outcomes (17, 18, 40). In the present context, poorer SRH, both preoperatively and during the recovery process, predicted less improvement over time. These findings provide a strong case for the predictive significance of SRH for health status in that the analyses concurrently accounted for past SRH and past health status. These findings suggest that a cyclical link between the health dimensions and SRH is operational.

Finding that SRH is capable of representing current health status, changes in health status, and further predicting future health status, the findings in this study suggest that epidemiologic research and clinical practice should consider an individual's SRH when evaluating overall health, when attempting to forecast future health, and when developing intervention and management strategies to improve overall health. Clinically, these findings suggest that to consider SRH, particularly in the event of an intervention and subsequent recovery, is practical and important.

Initially, where all 3 of the health dimensions were simultaneously considered as potential predictors of SRH, only mental well-being (cross-sectionally) and changes in mental well-being (longitudinally) significantly predicted SRH and future SRH, respectively. This was surprising for two reasons. One, it was inconsistent with previous work that pointed primarily toward aspects of physical health as strong predictors of SRH. Two, the largest changes in health status were observed for measures attributed to physical health, primarily pain and activity limitations, in the present sample. Having considered only particular dimensions of health and making use of simple linear or logistic regression analyses, previous research may have missed important mediated effects. Our structural equation modeling analyses made it possible to examine direct and mediated effects. Also, our findings support the view that mediated effects were operational, i.e., the effects of physical health on SRH were mediated through social health and mental well-being and the effects of social health on SRH through mental well-being during recovery.

While previous work has not considered mediated pathways between health dimensions with respect to SRH in this population, there is literature in support of a mediated model. A recent review (41) examined the association of depression or anxiety with medical symptom burden in patients with chronic medical conditions. Findings were consistent across studies with regard to somatic symptoms, including pain, being at least as consistently and strongly associated with depression and anxiety as were the physiologic measures that were thought to be more objective measures of disease severity. Findings suggested that the burden of physical symptoms and resulting functional impairment caused by complications of medical illness were likely to provoke or worsen episodes of anxiety and/or depression. Some evidence suggesting that the effects between anxiety/depression and physical symptoms may be bidirectional also was put forward.

Several studies have assessed the link between integration into social networks and aspects of psychological health, particularly depressive symptoms, in general population samples, elderly samples, and chronic condition groups (42–44). Commonly, the existence of social networks and perceptions of greater engagement in social networks have been linked to fewer depressive symptoms and overall improved mental health. A few studies have further put forward that in groups with chronic conditions, the negative effects of physical symptoms, primarily pain and functional disability, on psychological well-being are partially mediated through social network effects/supports (44, 45).

Irrespective of the mechanism by which the health dimensions interact with each other, however, the predictive significance of mental well-being for SRH in this cohort appears critically important. These findings, however, do not suggest that physical and social health are inconsequential. Rather, the findings suggest that physical and social health are steps in the pathway linking health status to SRH in individuals recovering from TJR. Additionally, these findings suggest that interventions limited to the mitigation of pain only, for example, may limit efforts to improve overall health.

It may appear counterintuitive for the current sample that a majority of patients reported good or better SRH preoperatively. However, this apparent incongruence between levels of physical health and self-assessed health is not unique, and has been reported in a number of studies among individuals with a chronic condition(s) and associated symptoms (46–49). Cumulative findings in extant literature have led many to characterize SRH as a multidimensional construct, involving complex judgments across multiple domains of health. Our results reflect this characterization of SRH.

A recent study (50) compared patient-reported functional outcomes following primary hip and knee TJR between surgeries performed in academic and community hospitals in the Greater Toronto Area. The study included patients recruited from 5 teaching and 5 community hospitals. Significant differences in functional status and patient quality of life were not detected between the academic and community hospital groups at any of the study time points. Therefore, while the current investigation is based solely on academic hospital patients, this evidence suggests that postsurgical trajectories of functional status and quality of life in the present sample are likely to be typical of primary hip and knee TJR outcomes more generally.

While the “lost to followup” and “dropped out” samples fared worse at baseline on 2 of the 9 health measures considered, it is difficult to speculate on their trajectory of recovery should they have remained in the study. Nonetheless, the analytical sample fully covered the possible range of values for each of the health measures. With 85% of the original sample constituting the analytical sample and given the range of trajectories observed within this sample, it is unlikely that the exclusion of the “out-of-study” sample from the analyses would have introduced any sizeable bias as to alter the relationships as reported.

While physical health has always been perceived as one of the major determinants of SRH, our results point to the importance of mental well-being in understanding the relationship between physical health and SRH. These results are particularly intriguing, given that this is a sample of individuals undergoing TJR for which pain and limited physical function are significant indicators. Additionally, since SRH significantly predicted TJR outcomes, this study suggests that the exclusive focus on any one health dimension may lead to missed opportunities for predicting and improving outcomes following surgery.

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 published. Dr. Perruccio 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. Perruccio, Davis, Hogg-Johnson, Badley.

Acquisition of data. Perruccio, Davis, Badley.

Analysis and interpretation of data. Perruccio, Davis, Hogg-Johnson, Badley.

Ancillary