Figure 2.

Linear mixed-model derivation of unadjusted and adjusted mean differences with 95% CI for SPADI total score at three follow-up time points for: (A) injection comparison and (B) exercise comparison. The adjusted model was corrected for baseline SPADI scores, age, sex, shoulder problem duration and clinic location. The interaction between the interventions was investigated but was not significant at all time points and hence not presented in the table. The bars in the figure represent the CIs for the adjusted mean, not the differences. SPADI, Shoulder Pain and Disability Index.