Time trends in surgery for non-traumatic shoulder disorders and postoperative risk of permanent work disability: a nationwide cohort study

Abstract

Objectives: Three previous randomized controlled trials found no overall difference in the effect of surgery compared with non-surgical treatment including exercise for patients with non-traumatic shoulder disorders. We assessed Danish temporal trends in the incidence of surgery for rotator cuff-related shoulder disorders, frozen shoulder, and osteoarthritis (OA) of the acromioclavicular (AC) joint, and the subsequent risk of permanent work disability.

Methods: Based on registers, we calculated annual incidence rates for the Danish population aged 18–63 years between 1996 and 2008. For a cohort of 19 264 first-time shoulder-operated patients who were in the labour market on admission, we used Cox proportional hazards models to assess risk of permanent work disability within 2 years after surgery.

Results: The annual incidence increased from 3.5 to 14.8 per 10 000, and 9.8% of the patients became permanently work disabled. No time trend was observed in this percentage. Formal education level was an important prognostic factor. Using ‘higher or medium-level education’ as a reference, the hazard ratio (HR) was 1.5 [95% confidence interval (CI) 1.3–1.8] for ‘vocational education and training’ and 2.0 (95% CI 1.8–2.3) for ‘low education level’. For ‘missing information on education’, the HR was 0.7 (95% CI 0.6–0.9).

Conclusions: We found a fourfold increase in surgery rates and a substantial risk of postoperative permanent work disability that remained constant over time. The risk was related to education level. These findings suggest that indications for surgery may need to be revisited and that attention should be given to supporting return to work, especially for blue-collar workers.

Acknowledgements

We thank Professor Johan Hviid Andersen for his valuable comments to an initial version of the manuscript and Data Manager Michael Victor Christensen for his preparation of the DREAM data for analysis. This work was supported by the Danish Rheumatism Association (grant no. R24-A707-B326).