[As supplied by authors]

The dependent variable used in this study has a skewed distribution. Seven of the sixty practices achieve maximum points. We investigated the sensitivity of our results through estimation of a negative binomial regression of the number of points lost, where the number of points lost is 650 minus the quality score. Negative binomial regression takes account of skewed nature and integer values of dependent variable without requiring us to omit practices with zero points lost (i.e. maximum points). These results are shown in an additional table along with mean values and ranges of the possible determinants of quality and the coefficients and associated confidence intervals from the regression models. The coefficients in the negative binomial model are of the opposite sign to those in the linear regression model since they represent the estimated effect on points lost rather than gained. The statistical significance of all variables is higher in the negative binomial regression model than in the linear model with the exception of the Global Sum payment per capita.

Additional table: Mean values and ranges of possible determinants of quality and coefficients from regression models

Possible determinant of quality scores	Mean [Range]	Univariate Models Linear regression coefficient [C.I.]	Multivariate model Linear regression coefficient [C.I.]	Multivariate model of �points lost� Negative binomial regression coefficient [C.I.]
Population characteristics
Material deprivation¹	23.4 [7.5, 45.0]	1.23 [0.06, 2.40]	1.24 [0.07, 2.40]	-0.026 [-0.049, -0.004]
Standardised chronic illness rate²	102 [74, 159]	0.53 [-0.02, 1.08]	-	-
Proportion of population aged over 64 years	0.13 [0.05, 0.25]	-375 [-837, 88]	-	-
Urban-rural location
Located in small town³	0.33 [0, 1]	-12.4 [-40.8, 16.0]	-	-
Located in rural area³	0.13 [0, 1]	-39.9 [-113.5, 33.9]	-	-
GP characteristics
Female proportion of GPs	0.31 [0, 1]	-16.5 [-117.7, 84.6]	-	-
Mean age of GPs (years)	44 [33, 58]	-4.1 [-6.9, -1.3]	-	-
Clinical team size
Small clinical team (>4 & <6 WTEs)⁴	0.28 [0, 1]	61.6 [18.9, 104.2]	42.8 [10.5, 75.2]	-0.75 [-1.35, -0.16]
Medium clinical team (>6 & <10 WTEs)⁴	0.35 [0, 1]	70.5 [28.7, 112.3]	48.1 [15.4, 80.7]	-1.25 [-1.97, -0.53]
Large clinical team (>10 WTEs)⁴	0.07 [0, 1]	81.8 [41.0, 122.5]	58.1 [24.7, 91.5]	-2.44 [-3.20, -1.67]
Clinical team composition
Non-Principal proportion of clinical team	0.07 [0, 0.51]	114.7 [-28.0, 257.3]	210.2 [73.1, 347.8]	-3.88 [-5.90, -1.86]
Nurse proportion of clinical team	0.20 [0, 0.43]	25.2 [-194.5, 244.9]	-	-
Practice characteristics
Royal College accreditation	0.52 [0, 1]	35.2 [5.3, 65.2]	-	-
Training practice	0.22 [0, 1]	39.2 [19.6, 58.8]	-	-
17C (salaried) contract	0.15 [0, 1]	-1.4 [-42.0, 39.3]	-	-
Ex-fundholding practice	0.37 [0, 1]	31.7 [6.9, 56.4]	26.3 [6.75, 45.8]	-0.87 [-1.37, -0.36]
Income from other sources
Global Sum payment (�, per capita)	68.0 [49.2, 103.9]	-2.53 [-4.78, -0.29]	-2.39 [-4.43, -0.35]	0.032 [0.001, 0.065]

[C.I.]=[95% Confidence Interval] �-� indicates variable not selected by stepwise variable selection procedure. ¹ Source: Scottish Index of Multiple Deprivation 2004. ² Source: Scottish Census 2001. ³ Source: Scottish Executive Urban-Rural Classification. Reference Group: Located in urban settlements. ⁴ WTEs=Whole Time Equivalents. Reference Group: Very small clinical team (<4 Whole-Time Equivalents).