The results of this study show for the first time a distinct gender difference for pain during movement but not for pain at rest. We also show for the first time that psychosocial variables (depression, anxiety, pain catastrophizing, and social provisions) are similar between men and women with late-stage osteoarthritis. Similar to prior studies, pain sensitivity, perceived function and function tests are reduced in women compared to men (see ). Surprisingly, while women had significantly worse pain and more impaired function than men, their actual physical activity levels (accelerometry) did not significantly differ and their OA grade was significantly lower, that is, less severe. This study developed predictive models to explain physical activity, function, and pain in people with OA using a comprehensive biopsychosocial approach. When both men and women were considered, physical activity levels were predicted by BMI, age, OA grade, depression, SF-36 PF, and KOOS Pain; pain during movement was predicted by pain at rest, knee extension, state anxiety and pressure pain thresholds; Function was knee flexion and extension, age, sex, opioid medication usage, pain duration, BMI, and heat pain threshold. Different predictive factors were found when the analysis was run with the men and women separately. We therefore, for the first time, were able to model physical activity levels, pain and function with multiple biopsychosocial variables, and to determine if there were differences between men and women in these variables.
Prior studies have modeled a number of different outcomes in people with OA to determine relevant factors that can predict outcomes [49–54]. Of direct relevance, in a sample of 168 OA subjects, sex predicted pain related outcomes (pain, disability and pain behaviors) and catastrophizing mediated the relationship between sex and OA pain-related outcomes . Further, in a study with 106 OA subjects, pain catastrophizing was a significant predictor of pain severity, disability, and function measured by gait . We extended these studies and showed for the first time that quantitative sensory testing (PPTs) predicted pain with movement, both evoked pain measures. We also show that for pain with movement that knee range of motion was an additional predictor for women but not for men. We also extend these findings and show that for function women had more predictors than men, which included opioid medications and pain catastrophizing as predictors of function only for women. However, our studies do not completely agree with prior studies in that pain catastrophizing did not predict pain during movement or physical activity. Differences in sample size (106 vs 268), OA severity (early vs Pre-total knee arthroplasty), and outcomes measures for pain (AIMS and observed pain behaviors vs pain with movement) and function (self-report vs gait speed or accelerometry) could underlie the lack of agreement between prior and the current study.
Pain during rest and movement
The current study found no significant gender difference in resting pain but significant gender differences for pain during movement and self-reported pain as measured by surveys (BPI, SF-36 Pain subscale, and KOOS Pain subscale). This is consistent with larger studies that show worse pain in women compared to men using the Knee Society Score survey instrument and the AIMS [5, 55]. Perceived pain measured by surveys reflects both pain at rest and pain during function. The results of this study suggest that pain during function has the largest impact on the sex differences found when measuring pain using self-report survey instruments.
The current study also found that women had lower Kellgren-Lawrence grades when compared to men, despite higher pain. These results are in agreement with prior studies that show women have more severe symptoms at the same Kellgren-Lawrence grades when compared to men . This difference in pain in relation to OA grade is not manifested in early knee OA . It has been hypothesized that women may have more severe osteoarthritis than men at the presurgical stage and wait longer to have surgery [8, 57, 58]. In fact, women lose articular cartilage from the proximal tibia at four times the annual rate of men and from the patella at a threefold greater rate . In contrast, the current study showed women have less severe Kellgren-Lawrence scores with a similar duration of pain just prior to surgery. These data suggest that women have higher pain despite lower radiographic evidence of OA and wait a similar length of time to have surgery.
Functional differences in OA
In the current study, women had more deficits on self-reported function on the SF-36 PF subscale and KOOS ADL subscale when compared to men, which is in agreement with prior literature . Similar differences have also been reported on the Knee Society and AIMS [3, 5, 6, 55, 58]. The current study showed reduced ability to perform the gait speed test and reduced knee active range of motion. These data are in agreement and extend prior studies that show reduced function on the 6-minute walk test, the timed up and go test, and stair climbing test times [7, 8]. The functional differences in knee OA subjects may in part be due to known differences in quadriceps muscle strength between women and men [8, 9].
The current study showed that physical activity levels measured by accelerometry were similar between women and men immediately prior to surgery, despite differences in perceived function and functional tasks. This is in contrast to prior studies that show lower physical activity levels in women with early OA compared to men . The gender differences in physical activity levels are present in healthy populations, where men spend more time in activities of higher intensity than women [61, 62]. Just prior to surgery, these differences in physical activity levels seem to disappear. However, men with OA have better performance on timed walk tests and stairs than women , which agrees with the results of the current study. Physical function tests are also similar to perceived function in patients with OA. This would suggest that physical function tests and perceived function are similar constructs, but that physical activity, measured by accelerometry, is a different construct. These results also suggest that women with late stage knee OA continue to move as much as men despite more pain during movement, greater pain sensitivity, and less functional ability.
Gender differences in pain sensitivity
The current study, in concurrence with prior literature, shows clear gender differences in pain sensitivity with women having greater sensitivity to heat, cold, and mechanical pressure [10–12]. Across the lifespan, women are more sensitive to heat pain with a nociceptive threshold 1.6°C lower in women than in men . These differences between women and men also occur for pressure pain thresholds; however, the differences tend to converge with age, with no gender difference in pressure pain thresholds at 50 to 70 years . We found the differences in pressure pain thresholds maintained in our population suggesting greater mechanical pain sensitivity of the deep tissue in women when compared to men when a chronic pain condition such as OA is present. This relationship of greater clinical and experimental pain in women has recently been shown in a chronic shoulder pain population .
The current study showed similar scores between women and men for depression, state anxiety, trait anxiety, pain catastrophizing, and perceived social support. It is often noted that women have a higher prevalence of depression . However, we noted no significant difference with around 15% of both women and men screening positive for depression. This prevalence rate is similar to prior studies in chronic pain populations [14, 66]. However, one study found that depression tendency in older Chinese patients with OA explained a portion of the gender differences in pain . The current study similarly, shows that depression explains a portion of physical activity levels in people with late-stage OA. Thus, depression may be related to not only pain, but also function in people with OA.
Sex differences in anxiety are controversial with some studies finding significant differences while others do not [68–70]. Some report that men with higher anxiety also have higher pain intensity  while other research suggests that this relationship is actually stronger in women . People with OA have higher anxiety than the general population, which is associated with higher pain intensity, worse symptoms, and greater healthcare utilization . However, based on the current study, these higher anxiety rates appear to occur similarly among women and men.
Pain catastrophizing has also shown a mixed relationship in pain research with some studies showing no gender differences [15, 72, 73] while others showing women have significantly higher pain catastrophizing . Prior work shows that pain catastrophizing may increase daily pain recall, but does not explain differences in experimental pain . The differences between studies could be due to studying different populations (younger vs older; experimental vs clinical pain) or using different measurement tools for catastrophizing. In people with OA, higher pain catastrophizing scores are associated with greater pain and disability [49, 55], pain 6 weeks after total knee replacement , and poor outcome 6 months after total knee replacement . The current study shows similar catastrophizing scores between women and men, and pain catastrophizing did not explain differences in physical activity levels. Thus, while pain catastrophizing is clearly a valuable construct that explains pain in people with OA, there was no sex differences observed in this population.
Our subjects were recruited from a large teaching hospital, which may include a different patient population than other clinical settings. There are many ways to analyze the differences between women and men. For our regressions explaining pain, function, and accelerometry, we did not perform further analyses to see if the predictor variables were different for women and men. That will be focused on and analyzed further as we follow this population through the preoperative to the postoperative period. The usage of accelerometry is beneficial to help in understanding actual levels of physical activity, but there are limitations in the validity of the METS equation as a subject performs higher levels of physical activity. Further research will focus on how the ActivPAL variables differ between our OA population and healthy controls.