Table 2.

Results of stepwise regression analyses: variables associated with hospital standardised mortality ratios at 1% significance level and tabulated in order of selection.

Variable	Regression coefficient (95% CI)*	P value	Mean (SD)
Model A: all cases (emergencies and electives) in 183 hospitals (adjusted R2=0.65)
Percentage of cases admitted as emergency	0.58 (0.41 to 0.75)	<0.001	61.4 (8.5)
No of hospital doctors per 100 hospital beds in 1994-5	−0.47 (−0.64 to −0.30)	<0.001	25.4 (8.0)
No of general practitioners per 100 000 population†	−0.67 (−1.05 to −0.30)	<0.001	54.6 (3.4)
Standardised admissions ratio‡	−0.15 (−0.23 to −0.06)	0.001	107 (14.4)
Percentage of live discharges to home	1.61 (0.71 to 2.50)	0.001	98.0 (1.5)
Percentage of cases with comorbidity§	1.51 (0.47 to 2.55)	0.005	4.2 (1.4)
NHS facilities per 100 000 population‡	−1.12 (−1.92 to −0.32)	0.007	3.43 (1.6)
Model B: emergencies only in 183 hospitals (adjusted R2=0.50)
No of hospital doctors per 100 hospital beds in 1994-5	−0.51 (−0.65 to −0.38)	<0.001	25.4 (8.0)
Percentage of cases with comorbidity¶	4.51 (3.11 to 5.92)	<0.001	1.7 (0.8)
Percentage of live discharges to home	1.24 (0.74 to 1.73)	<0.001	97.0 (2.3)
NHS facilities per 100 000 population‡	−1.62 (−2.27 to −0.97)	<0.001	3.4 (1.6)
Standardised admissions ratio‡	−0.12 (−0.19 to −0.04)	0.003	107 (14.4)

^*Coefficients represent expected change in hospital standardised mortality ratios for a given change in the associated variable. For example, an increase of 10% in numbers of hospital doctors would increase average doctor:bed ratio by 2.54, from 25.4 to 27.94. From model A, this increase would be expected to result in a fall of −1.19 in hospital standardised mortality ratios (2.54×−0.47). 

^†According to Office of National Statistics. 

^‡For health authority where hospital located. 

^§Percentage comorbidity of bronchopneumonia or heart failure or fracture of neck of femur. 

^¶Percentage comorbidity of bronchopneumonia or malignant neoplasm.