Rotator cuff tear is a common disorder and there is a lack of knowledge of appropriate treatment and consequences of different treatment modalities. The overall aim of this thesis was to examine short- and long-term results of rotator cuff tear treatment.

In Paper I we did a retrospective 21 to 25-year follow-up of a consecutive series of patients with partial and full-thickness rotator cuff tears, treated with acromioplasty without cuff repair. The cuff status had been documented in a specific perioperative protocol in all patients at the index operation. We did x-ray, ultrasonography and clinical scores with Constant score and Western Ontario Rotator Cuff index (WORC) at follow-up. We identified 111 patients with either a partial or a full-thickness tear, but at follow-up 21 were deceased and 11 were too ill from medical conditions unrelated to their shoulder. Out of the remaining 78 eligible patients, 69 were examined (follow-up rate 88 %) and they had a mean age at the index operation of 49 years (range 19-69 years). Forty-five had a partial tear and 24 a full-thickness tear at the index operation. At follow-up, 74% of patients with full-thickness tear had cuff tear arthropathy grade 2 or more according to the arthropathy classification of Hamada (grade 1 to 5) and 87% had developed tear progression (i.e. a larger tear). Corresponding numbers in those with a partial tear was 7 % arthropathy and 42 % tear progression, and the differences between the full-thickness group and the partial tear group was significant for both outcome measures (P<0.001 for both analyses). In those with arthropathy, the mean Constant score was 47 (standard deviation [SD], 23), the mean age and gender-adjusted Constant score 62 (SD, 27) and the mean WORC 58 % (SD, 26). Patients with a partial tear at follow-up had mean Constant score and WORC within the normal range. In multivariable analysis with logistic regression, having a full-thickness tear at the index operation was a risk factor for arthropathy (odds ratio [OR] 37.8; 95% confidence interval [CI], 8.2-175.0) and for tear progression (OR 6.09; 95% CI, 1.41-26.29).

In Paper II we examined the contralateral shoulder in the same patients as in paper I and with the same methodology. Sixty-one patients were examined and 38 had had a partial tear at the index operation 21-25 years ago and 23 a full-thickness tear. The overall rate of contralateral full-thickness tears was 50.8 %, which is higher than the 16-35 % rate found in previous studies of newly diagnosed cuff patients. The rate of contralateral full-thickness tear ranged from 13.6 % in patients with a partial tear in the index shoulder at follow-up, to 90 % in patients with a full-thickness tear and arthropathy in the index shoulder. There was a significant correlation regarding conditions between shoulders in the same patient, with a Spearman coefficient of 0.72 for the number of ten-dons with a full-thickness tear, 0.31 for Hamada grade of arthropathy and 0.65 for Constant score. The number of tendons with a full-thickness tear in the index shoulder at follow-up was a risk factor for a contralateral full-thickness tear (OR 3.28; 95% CI, 1.67-6.44) in a multi-variable logistic regression model. We also found that cuff tear arthropathy was significantly more common in patients who had undergone an acromioplasty (P<0.001), a finding which is not confirmatory but may generate a hypothesis.

Paper III addressed 17 to 20-year results after operation with a synthetic interposition graft for irreparable cuff tears. We used X-ray, ultrasonography and clinical scores at follow-up. We identified a consecutive series of 13 patients, one of whom was deceased at follow-up. Ten of the remaining 12 participated in a complete follow-up and 2 did only x-ray examination. Nine out of 12 (75 %; 95% CI, 43-95 %) had cuff tear arthropathy Hamada grade 2 or more in the index shoulder at follow-up. The mean Constant score was 46 (SD, 26) and the mean WORC 59 % (SD, 20). Seven out of 12 had contralateral cuff tear arthropathy, and the difference in frequency of arthropathy between shoulders was not statistically significant (P=0.667).

In Paper IV we tested whether early repair of small cuff tears, involving mainly supraspinatus, would give a superior clinical result com-pared to physiotherapy without repair in a prospective randomised trial with 12 months follow-up. We used Constant score as the primary out-come, and WORC, EQ-VAS and Numerical Rating Scale for pain (NRS) as secondary outcomes. We also aimed at assessing the rate of tear progression in unrepaired shoulders and the healing rate in repaired shoulders by Magnetic Resonance Imaging (MRI) performed at 12 months. With a high grade of follow-up (100 % for 12 months Constant score and 95 % for 12 months MRI), the repair group had a 12 months median Constant score of 83 (Quartile range [QR], 25) and the conservative group 78 (QR, 22). This between-group difference in medians of 4.5 (95% CI,-5 to 9; P=0.68) was not statistically significant and we did not detect any significant differences in the secondary outcomes at 12 months. The retear rate was 6.5 % in repaired patients and 29 % of unrepaired patients had a tear enlargement >5 mm.

The results in this thesis indicate that patients with small, traumatic, full-thickness tears of mainly supraspinatus have no clinical benefit of early surgical repair compared to physiotherapy alone, but in the long-term, patients with full-thickness tears have an increased risk of tear progression, cuff tear arthropathy and low clinical scores. These results are especially important in the treatment decision of repair or not in younger patients. Having a full-thickness tear is also a risk factor for having a contralateral cuff tear, a phenomenon that underlines the importance of endogenous factors in the development of rotator cuff tears. If a cuff tear is not repairable to bone, the addition of a synthetic inter-position graft does not seem to prevent cuff tear arthropathy.