According to the per protocol analysis as presented inTable 2, after 12 weeks RMDQ significantly decreased (p < 0.05)and back flexibility during extension significantly increased(p < 0.001) in the yoga group. However, these variables did notsignificantly differ from baseline in the control group. After 12weeks, the mean changes from baseline in RMDQ (p < 0.05) andback flexibility (p < 0.01) in the yoga group were significantlygreater than those in the control group.Serum TNF- level maintained in the yoga group while itincreased significantly (p < 0.01) in the control group (Table 2)showing the larger increase for the control group compared to theyoga group (p < 0.001). The CRP levels were not significantlychanged over 12 weeks in both groups.In the intention-to-treat analysis about back pain, flexibility,and inflammatory factors (Table 3), smaller mean changes frombaseline in both groups were present compared with those by theper-protocol analysis, but overall results of statistical analyseswere not different from those of the per-protocol analysis.Outcomes of stress variables by the per protocol analysis arepresented in Table 4. After 12-week yoga intervention, the totalSOSI scores significantly decreased in the yoga group (p < 0.05).The yoga group significantly improved on five of the ten SOSIsubscales: Gastrointestinal symptoms (p < 0.05); Muscle tension(p < 0.05); Habitual patterns (p < 0.05); Depression (p < 0.05);Emotional irritability (p < 0.01). In the control group, the totalscores and all subscales of SOSI did not significantly changeover 12 weeks. However, differences between groups in themean changes from baseline in these SOSI scores did not reach