Arthroscopic Capsular Release Versus Manipulation under Anesthesia for Refractory Frozen Shoulder: A Systematic Review with Meta‐Analysis

Objective Frozen shoulder (FS) is a painful and debilitating condition affecting the shoulder joint. When patients fail to improve after conservative treatments, operative treatments including arthroscopic capsular release (ACR) and manipulation under anesthesia (MUA) are recommended. However, the comparison between these two interventions remains controversial. This study aimed to compare the efficacy and safety of ACR and MUA for refractory FS. Methods A systematic review and meta‐analysis was conducted following the Preferred Reporting Items for Systematic Review and Meta‐analyses (PRISMA) guidelines. PubMed, EMBASE, Cochrane Library, and Web of Science were searched for eligible studies until December 10, 2023. Meta‐analyses were conducted using Manager V.5.3.3. Pooled effect sizes were expressed as the weighted mean difference (WMD) or odds ratio (OR) with 95% confidence intervals (CIs). Results A total of eight comparative studies with 768 patients were included. Compared with MUA, ACR had statistically better Δ VAS (WMD, −0.44; 95% CI, −0.71 to −0.18; I 2 = 6%; p = 0.001) at over 12‐month follow‐up, which did not reach the minimal clinically important difference (MCID). Other outcomes regarding pain relief, function, and range of motion (ROM) improvements were not statistically different between the two groups at different follow‐up timepoints. Compared with the MUA group, the ACR group had a significantly higher rate of severe complications (OR, 4.14; 95% CI, 1.01 to 16.94; I2 = 0%; p = 0.05), but comparable rates of mild complications and additional intervention. Conclusions In treating refractory FS, ACR demonstrated comparable pain relief, functional and ROM improvements, rates of mild complications and additional intervention but a higher risk of severe complications to MUA during short‐term follow‐up periods. Notably, ACR exhibited statistically superior improvement in the long‐term pain relief compared to the MUA group, although it did not reach the MCID.