The GEE regression

models were used to investigate the performance and characteristics of specialty tech support hospitals, including inpatient charges, LOS, readmission and mortality adjusting for patient-level and hospital-level confounders. Because the distributions of continuous dependent variables (inpatient charges and LOS) were skewed, we utilised log transformation in order to improve the distribution characteristics of the data. In addition, we ran the GEEs of the binary outcome variables for readmission within 30 days of discharge and mortality within 30 days of admission. In order to enhance case mix adjustment, we included the diagnosis and procedure code in each model. SAS V.9.2 (SAS Institute, Cary, North Carolina, USA) was

used for all calculations and analyses. As the data set does not have patient identification information, no ethics committee approval is required. Results A total of 645 449 patients nationwide were hospitalised for spinal disease during the study periods, and 17 specialty hospitals accounted for 45 649 (7.1%) of patients nationwide admitted for spine disease (table 1). Patients in spine specialty hospitals were aged and female, had undergone more surgical procedures, and had lower CCL scores. The increase in volume in 2012 compared with 2011 was greater than average in specialty hospitals as well as in conventional hospitals (total: 12.9% vs specialty 17.8%). Table 1 Characteristics of patients Table 2 shows the hospital characteristics analysed. Of the 823 hospitals in our study, there were 17 Ministry of Health and Welfare-designated spine specialty hospitals (2.1% of the total), which accounted for 7.1% of the total spinal procedures

performed nationwide during the study period. While none of these was a teaching hospital, they were located mainly in metropolitan areas, and their structural factors were greater in terms of number of beds (in 100 bed increments), specialists per 100 beds and nurses per 100 beds as well as bed occupancy rate as compared with hospitals in the small general Brefeldin_A hospital category. Although specialty hospitals are larger than small general hospitals in terms of structural factors, both types of hospitals fall within the same small hospital category in Korea. Clinical staff was greater in spine specialty hospitals than in mid-sized general hospitals. Furthermore, 11.8% of specialty hospitals were considered to be efficient compared with 6.8% of all hospitals. Table 2 Characteristics of hospitals Univariate analysis of outcome variables (see table 3) revealed that inpatient charges per case were lowest in spine specialty hospitals; however, per day charges were higher than in small and mid-sized general hospitals. LOS was 10.9 days per admission, which was comparable with tertiary research hospitals, but was much shorter than in small and mid-sized general hospitals.