Overall, 538 citations were included after deduplication and the specific search strategies can be found in the supplementary material. Additional 57 citations we found via hand search, which resulted in overall 595 hits. No prospective or controlled studies were found by the systematic literature search and one of the studies had a control group [13]. Data from eight retrospective studies (and seven indications) were used in the analysis of clinical effectiveness and safety. Two studies were found on the intervention of anterior cruciate ligament (ACL) reconstruction [14, 15], two on cataract surgery [16, 17], and one for each of the following interventions: meniscectomy, thyroidectomy, primary hip arthroscopy, open carpal tunnel release, and rotator cuff repair [13, 18-20].
Study characteristics
While two studies were single center analyses [13, 17], six studies were analyses of health care databases [14-16, 18-20]. Seven studies were conducted in the US [13-18, 20] and the eighth study was conducted in the Netherlands [19]. Information about study sponsors was not disclosed in five studies [13-15, 17, 18], two studies were funded by the National Institutes of Health/National Institute for Arthritis and Musculoskeletal and Skin Diseases [19, 20], and one study was funded by the National Eye Institute [16]. Two studies did not report on conflict of interests (COI) [15, 18], four studies reported that none of the authors had COI [14, 16, 17, 19], and two studies reported COI of one of their authors [13, 20]. The dates of data collection in all the studies were between 1997 and 2015.
All the studies gathered data on the outpatient setting and while four studies analyzed the VOR from the perspective of surgeons [16, 17, 19, 20], one analyzed it from the hospital perspective [14], and three from the perspectives of both surgeons as well as hospitals [13, 15, 18]. Follow-up time was not reported in five studies [13, 15-18], it was ten years in [20], six months in [19], and 30 days in [14].
Patient characteristics
The analysis of primary hip arthroscopy included 7,836 patients and 8,267 procedures that were performed by 295 surgeons in 137 centers [20]. The analysis of carpal tunnel release included 1,345 patients/procedures (712 patients not followed-up) performed by 17 surgeons in 11 centers [19]. The analysis of rotator cuff repair included 9,973 patients (961 not followed-up) [18], ACL reconstruction included 45,262 patients (14,050 not followed-up) [14, 15], and the analysis of meniscectomy included 123,012 patients (72,585 not followed-up) [15]. The number of procedures, surgeons, or surgical centers was not reported in the three studies above. For the single center analysis of thyroidectomy, 109 outpatient and 51 inpatient patients were included with 35 and 26 procedures respectively [13]. For the analysis of cataract surgery, 2,289,307 patients were included (200,520 not followed-up) with 3,280,966 procedures conducted by 22,877 surgeons in an unclear number of centers (except for four surgeons that were part of a single center analysis [17]). Because of the retrospective nature of the studies, loss to follow-up was not reported.
Surgeon as well as hospital volume was categorized into low, medium, and high (very high in one study [20]) and the thresholds differed with interventions. The low volume threshold ranged from six to 411 interventions, while the high (or very high) threshold ranged from 30 to 1,336 interventions per year. Inclusion and exclusion criteria were heterogeneous as they varied with interventions. Co-interventions were reported in four studies [13, 14, 18, 19] and the mean age ranged from 29,4 to 73 years. Study characteristics and results of included studies are displayed in Table 2 and Table 3.
Clinical effectiveness
Concerning cataract surgery, inverse VOR was observed [16]. The surgeon volume rates ranged from 1-50 (1), 51-200 (2), 201-500 (3), 501-1,000 (4) and ≥1,001 (5). The number of cases per surgeon was inversely correlated with the adverse event of posterior capsule rupture (PCR), where PCR and vitreous loss (VL) rate were 3.75% for low volume and 0.29% for very high volume surgeons [17]. The relative risk (RR) for endophtalmitis was 4-fold between low and very high volume surgeons [16]. After adjusting for risk factors the RR in category (1) was still 3.8-fold higher than in the reference category (5).
Concerning thyroidectomy, there was no VOR observed, but it was suggested that thyroidectomy is safe also in low volume centers as in the single low volume center, no cases of readmission occurred [13]. The surgeon volume amounted to ten thyroid surgical cases per year on average, while the hospital volume averaged of 20 thyroid surgical cases per year.
Concerning rotator cuff repair, VOR was observed. The surgeon volume rates ranged from low (<15), medium (15-30), to high (≥30), while hospital volume rates ranged from low (<75), medium (75-200), to high (≥200). Patients of low volume surgeons were 2.8 time more likely to have nonroutine disposition at discharge, while low volume hospitals were 2.1 times more likely to discharge patients with nonroutine dispositions. Surgeon-related mean operating time was 10 minutes shorter and hospital-related mean operating time was 6 minutes shorter for high volume compared to low volume surgeons/hospitals. Length of stay (LOS) was 2.3 times longer for low volume surgeons and 0.5 times for low volume hospitals compared to high volume surgeons/hospitals [18].
Concerning hip arthroscopy, VOR was observed. Surgeon yearly volume rates ranged from low (<102), medium (102-164), high (164-340), to very high (≥340). The survival rates for very high volume surgeons were 11.1-24.9% higher than for low volume and the hazard ratio for reoperation (with reference value of low volume) was 0.17 for very high volume surgeons [20].
Concerning open carpal tunnel release, VOR was not observed. Surgeon yearly volume rates ranged from low (6-44), medium (47-71), to high (75-163). BCTQ score did not vary with volume at all while the difference on the VAS scale was 1 point (out of 100) between low and high volume surgeons (18 vs. 19 points) [19]. Such difference is below the threshold of the minimal clinically important difference [21].
Concerning ACL reconstruction, VOR was observed. The surgeon volume rates ranged from low (<25), medium (25-75), to high (≥75) and hospital volume rates ranged from low (<125), medium (125-300), to high (≥300). The odds ratio for hospital based acute care within 30 days (with reference of low volume hospitals) was 0.47 for high volume hospitals [14]. Low volume surgeons were 4.5 times more likely and low volume hospital 3.33 times more likely to have nonroutinely discharged patients compared to high volume surgeons [15]. Furthermore, low volume surgeons had a 27 minutes longer and low volume hospitals 21 minutes longer mean operating time than high volume surgeons/ hospitals [15].
Concerning meniscectomy, VOR was observed. The surgeon volume rates ranged from low (<75), medium (75-175), to high (≥175) and the hospital volume rates ranged from low (<600), medium (600-1,200), to high (≥1,200). The low volume surgeons were 2.8 times and low volume hospitals were 8 times more likely to have nonroutinely disposed patients at discharge than high volume surgeons/hospitals [15]. In terms of mean operating time, both low volume surgeons and low volume hospitals had a longer mean operating time by 19 minutes compared to high volume surgeons/hospitals [15].
Safety
The only safety related data reported were without its relationship to surgeon/hospital volume. In the hip arthroscopy study, 0.2% of patients experienced procedural complication at 30 days post intervention – the complications were: myocardial infarction ileus, pneumonia, sepsis, mechanical complication, hardware failure, deep vein thrombosis/pulmonary embolism, wound infection, dislocation/iatrogenic instability, major bleed [20]. In the carpal tunnel release study, 1.6% of patients experienced procedural complications such as wound infections, wound dehiscence [19].
In the thyroidectomy study [13], 19 of the 160 patients experienced complications. Complications included transient hypercalcemia (5%), temporary vocal cord paralysis (2.5%), post-operative seromas requiring aspiration (1.9%), post-operative hematoma requiring aspiration (1.25%), bilateral vocal cord paralysis (0.63%), and inadvertent transection of the recurrent laryngeal nerve (0.63%).