There’s increasing recognition that improvement in pain and functional outcomes after total knee arthroplasty (TKA) varies widely between patients. Attention has turned to determining what patient characteristics are associated with outcomes after TKA and identifying individuals whose pain is unlikely to improve.
Robert R. Edwards, PhD, a psychologist in the Center for Pain Management at Brigham and Women’s Hospital, Jennifer A. Haythornthwaite, PhD, of Johns Hopkins University, and colleagues conducted the first multicenter study that comprehensively examined clinical, functional, psychosocial and sensory factors as predictors of pain and functional outcomes after TKA. They present the results in BMC Musculoskeletal Disorders.
The study was prospective and enrolled 248 adults with knee osteoarthritis (60% female, mean age of 65) who underwent unilateral primary TKA at one of two academic medical centers.
About two weeks before surgery, patients completed a wide-ranging assessment:
- Four prespecified primary outcome measures, repeated six months postoperatively—The Brief Pain Inventory (BPI) Pain Severity subscale, BPI Pain Interference subscale, Western Ontario McMaster Universities (WOMAC) Pain scale, and Function scale
- Self-report measures of general health and quality of life—The Widespread Pain Index, the General Health subscale of the SF-36, the EuroQOL, and the Godin exercise questionnaire
- Self-report measures of psychosocial factors—The Pain Catastrophizing Scale (PCS), the Patient-Reported Outcome Measurement Information System (PROMIS) Depression and Anxiety scales, the Brief Symptom Index 18-Somatization Scale, the NEO personality inventory, the positive affect subscale of the Positive and Negative Affect Schedule, the ENRICHD Social Support Instrument, and 0–100 scales assessing expectations for improvement after surgery
- Self-report measures of sleep—The Pittsburgh Sleep Quality Index, the Insomnia Severity Index, and a 0–100 scale assessing the severity of fatigue
- Physical function tests—Stair climbing and six-minute walk, with pain ratings recorded during the tasks and situational pain catastrophizing scales completed afterward
- Quantitative sensory testing—Evaluation of mechanical pain at the trapezius muscle and patella, cold pain sensitivity test, noninvasive assessment of endogenous pain inhibition during the cold pain sensitivity tests, and Pain Sensitivity Questionnaire
- Actigraphy—Patients were asked to wear an actigraphy wristband for one week to measure sleep efficiency, total sleep time, and wake after sleep onset time
The mean scores on all four primary outcome measures improved significantly from baseline to six months post-surgery. However, there was substantial interindividual variability, and approximately 30% of participants reported clinically significant ongoing pain and/or physical limitations.
Factors Influencing Primary Outcomes
In multivariable models, significant (P<0.05) factors that predicted results six months after surgery were:
- BPI Pain Severity—Risk factors for more pain were higher catastrophizing, opioid use, presence of another chronic pain condition, the number of painful body areas, and the degree of reported anxiety. Protective factors were higher sleep efficiency and higher agreeableness on the NEO (suggesting the person is reliably cooperative, warm, kind and friendly)
- BPI Pain Interference—No significant risk factors; greater social support was protective
- WOMAC Pain—Risk factors were insomnia, higher levels of fatigue, and the number of painful body areas. Protective factors were higher educational attainment and higher agreeableness
- WOMAC Function—Risk factors were increased pain during the six-minute walk and higher levels of general fatigue. Higher agreeableness was protective
Some of the psychophysical variables assessed during quantitative sensory testing influenced the results:
- BPI Pain Severity—The beneficial effect of agreeableness was most evident in patients with high temporal summation (greatest increases in pain in response to repeated stimuli)
- WOMAC Pain and WOMAC Function—Similar to the pattern observed for BPI Pain Severity; high agreeableness buffered the adverse impact of high temporal summation, and low agreeableness enhanced the detrimental impact of high temporal summation
- BPI Interference—Not tested because no sensory measures were significant predictors in univariate analyses
Toward Better Risk Stratification
Improving the precision of predictive models could help surgeons make better decisions about who is likely to benefit from TKA. The risk factors and protective factors ascertained in this study should be considered for inclusion in these models, along with a measure of agreeableness or a similar personality trait. Considering synergy or other interactions among risk factors also appears to be important.