Abstract
Introduction
Overwhelming evidence has established obesity as a risk factor for osteoarthritis (OA) of the knee. Randomized clinical trials such as the Look AHEAD study have shown long term successful intentional weight loss with an intensive lifestyle intervention (ILI) in overweight and obese type 2 diabetics. Weight loss can also decrease knee pain in persons who have OA, but it is unknown if intentional weight loss can reduce the risk of TKR. To answer this question, data from the Look AHEAD study were examined to determine if intentional weight loss could reduce the risk of TKR.
Methods
Look AHEAD is a multicenter, randomized trial which began in August 2001 and follow-up continued for a median of 11.3 years at 16 academic centers. 5145 persons aged 45–76 with diabetes were randomized to either an ILI with reduced calorie consumption and increased physical activity designed or to diabetes support and education intervention (DSE). TKR events were ascertained every 6 months. Retrospective data of reported knee pain was assessed using the WOMAC knee pain questionnaire. Participants with partial TKR or revisions were excluded. Cox proportional hazard models were used to relate baseline BMI category (obese, Class I, Class II, or Class III obesity), baseline knee pain, and treatment group with TKR. Weight change category (lost<=5%, stable, gained>=2%) from baseline to year 1 follow-up by treatment assignment was also examined as a predictor of TKR after excluding TKR occurring prior to year 1.
Results
Out of the randomized participants, 2171 reported knee pain (43%) at the baseline visit (p=0.81). WOMAC knee pain score did not differ by random assignment (ILI:3.6±2.9, DSE:3.9±3.0, p=0.08). During follow up there were 631 TKRs reported by participants. TKR was more common in heavier (p<0.001), and older (p<0.001) participants and did not differ by randomization. Heterogeneity of treatment effect was observed with baseline knee pain (interaction p = 0.02), therefore analyses were stratified by presence or absence of knee pain at baseline. In persons without knee pain at baseline, there was a 29% reduced rate of TKR in the ILI group compared to the DSE group (HR[95%CI] 0.71[0.52,0.96]; Figure 1A). Whereas in persons with knee pain at baseline, there was a trend for the ILI to have an increased rate of TKR compared to DSE (1.11[0.92, 1.33]; Figure 1B). In both sets of analyses, obese participants had significantly higher hazard of TKR than overweight participants (No pain: Class I 1.78[1.04,3.05], Class II 2.27 [1.31,3.94], Class III 2.94[1.67,5.18]; With pain: Class I 1.70[1.12,2.59], Class II 2.42 [1.60,3.65], Class III 2.80[1.85,4.23]). When TKR incidence was examined by weight change at year 1 there was no difference in weight change category between randomization groups in persons with knee pain (interaction p=0.20) or without baseline knee pain (interaction p=0.87) (Figure 1C and D) or for the overall effect of year 1 weight loss category in either stratum (with knee pain p=0.26, without knee pain p=0.48). However, in persons without knee pain at baseline, ILI had a marginal reduction in hazard of TKR compared to DSE (0.71[0.49, 1.05]) but not in the group with baseline knee pain.
Conclusions
In persons with no knee pain at baseline and who were overweight or had Class I or II obesity, ILI seemed to reduce the risk of TKR compared to DSE. In contrast, persons with knee pain who gained weight at 1 year and were randomized to ILI had the highest risk of TKR. This suggests that weight loss to prevent TKR may be more effective prior to the development of significant knee pain. After the onset of knee pain however, low impact or non-weight bearing activity should be considered to avoid worsening knee symptoms.
For any figures or tables, please contact authors directly.
