In the present study, we evaluated the association between BMI groups, categorized according to WHO classification, with WOMAC and EQ5D preoperatively (baseline) and at different time intervals. There were no significant differences in self-reported preoperative pain, function, stiffness, or quality of life measures across all BMI groups. Our results indicate that by the end of 12 months all patients, regardless of their BMI, had improvement in pain, stiffness, physical function, and quality of life, and the magnitude of improvement was similar across all BMI groups.
The evidence of the impact of BMI on TKA surgical outcomes is conflicting. Some studies suggest there is an association between obesity and pain, functional recovery, and quality of life following TKA [9, 10, 12, 13, 15], while others suggest no association [11, 14, 17, 37]. The variation in findings may be related to differences in the overall health status of the cohorts, use of different cut-offs for BMI, lack of control for confounding factors such as age and sex, and the small sample size [19, 20].
Recently, a study in the U.S. population by Collins and colleagues  examined the association between BMI groups, using the recommended WHO classification, and WOMAC subscale of function. They demonstrated that subjects with higher BMI had worse preoperative WOMAC pain and function than patients with normal BMI. Studies by Baker et al. in the U.K. demonstrated that patients in higher BMI groups (assessed in groups of BMI of < 25, 25 to 39.9, and \(\ge\)40 kg/m2) also had significantly (p<.01) worse preoperative WOMAC total and EQ5D scores (p < .001)   than patients with normal BMI. We also observed that patients with higher BMI had poorer function, pain, and total scores at baseline, but these differences between BMI groups were not significant. The average baseline scores in our study compared to the U.S. population  were higher for pain (53.4 vs. 40.8) and function (54.1 vs. 42.5); whereas, the average baseline WOMAC total score was lower in our study compared to the U.K. population (54.2 vs. 63.2) . Indeed, our patients had worse preoperative pain and function compared to U.S. patients, but better preoperative health status compared to the U.K patients. This discrepancy may be due to different health care systems in the U.S. and U.K. where different indication criteria and algorithms/cut-offs are used to guide the appropriateness of TKA [39–41].
Collins and colleagues  reported that patients in higher BMI groups experienced greater (p < .001) improvement in pain and function from baseline to 3 months after TKA compared to the lower BMI groups, but all groups had similar levels of pain and function at 24 months. We observed a greater improvement in pain from baseline to 3 months postoperatively in patients with higher BMI. However, all BMI groups attained similar level of pain reduction at 12 months. Baker et al.  reported that the average change for WOMAC total score from baseline to 12-month following TKA was similar across different BMI groups. Giesinger et al.  used WHO classification to categorize patients, with the Oxford Knee Score (OKS) used to measure self-reported pain and function, and the EQ-5D-3L used to measure general health status. They found no influence of BMI on postoperative self-reported pain, function, or general health scores. Our results were in line with the previous studies demonstrating that all patients received the same benefit from TKA regardless of their BMI [17, 37, 38, 43], and most of the improvement occurred by 3 months postoperatively .
Similar to other studies [17, 37, 38, 43], at the end of the study period, all BMI groups experienced statistically significant and clinically meaningful improvement in pain, function, stiffness, and total WOMAC score as well as EQ5D index. Despite substantial improvement in pain and function after TKA across all BMI groups, at the end of 12 months, our patients experienced worse pain and function than patients in Collins et al. study . This may be explained by the worse baseline pain and function of our participants compared to Collins et al. study, as preoperative health status affects the postoperative outcomes .
We examined the PROMS preoperatively, pre- to 3 months postoperatively, as well as 3 to 12 months after TKA, and the findings of our study offer insight into the association between obesity and TKA outcomes in Albertans following TKA. In this study, we were able to examine the impact of different levels of BMI groups on PROMS using the WHO classification of BMI.
Strengths and Limitations
The strength of this study includes a large sample of patients (N = 7,714) with WOMAC total score and 3 subscales recorded at baseline and 3 and/or 12 months postoperatively. We have also categorized patients into 5 groups based on the WHO classification of BMI, which helped us to evaluate a clear relationship between each of the BMI groups with TKA outcomes. BMI records in our dataset were not self-reported, which provide more reliable results. Our analysis also had limitations inherent to retrospective studies. WOMAC was used to measure lower extremity physical function, which has been reported to have limited ability to accurately predict change in function . Individuals who did not have weight and height records were excluded from the study, though there were no significant differences in patient-reported outcomes between the cohort that was excluded and those included in the cohort studied (Supplementary: Table 1.S and Table 2.S). In this study, BMI has been used as a measure of obesity, however, BMI does not provide us with information about the body’s fat distribution as well as body composition . Further studies using methods such waist to hip ratios or sophisticated methods such as dual-energy X-ray absorptiometry (DEXA) are recommended to evaluate the association between fat distribution and body composition with TKA outcomes in patients who have undergone surgery. Patients who had baseline and at least 1 follow-up visit (postoperative month 3 or 12) for WOMAC and EQ5D questionnaires were included in the study. Patients with missing follow up questionnaires were also included in the analysis as linear mixed effect models allow for missing data and are robust to determining estimates in presence of missing data .