Table 2.

Regression analyses results from negative binomial regression for locum use at general practice level, Model A: over time (2018–2019), Model B: cross-sectionally (2019)a

CharacteristicModel A IRR (95% CI)Standard errorP-valueModel B IRR (95% CI)Standard errorP-value
Rurality (0 = urban, 1 = rural)1.250 (1.095 to 1.428)0.085<0.0011.300 (1.085 to 1.559)0.120<0.004
IMD 20191.002 (0.999 to 1.006)0.002<0.0961.005 (1.000 to 1.009)0.002<0.046
QOF practice performance1.005 (0.991 to 1.017)0.007<0.4791.009 (0.991 to 1.026)0.009<0.298
Single-handed practice4.611 (4.101 to 5.184)0.276<0.0014.618 (3.928 to 5.428)0.381<0.001
QOF morbidity burden1.384 (0.963 to 1.991)0.257<0.0791.255 (0.801 to 1.996)0.287<0.320
Percentage of female population0.967 (0.959 to 0.981)0.006<0.0010.970 (0.946 to 0.994)0.012<0.015
Proportion of practice population aged ≥65 years0.970 (0.950 to 0.984)0.009<0.0010.971 (0.958 to 0.988)0.007<0.001
Practice workload (total GP FTE/list size)1.001 (1.001 to 1.002)0.001<0.0011.001 (1.001 to 1.002)0.001<0.003
CQC ratings (reference group is outstanding services)Reference groupReference group
Inadequate2.108 (1.370 to 3.246)0.464<0.0012.687 (1.451 to 4.974)0.844<0.001
Requires improvement1.229 (0.949 to 1.592)0.163<0.1181.198 (0.822 to 1.744)0.229<0.346
Good1.343 (1.103 to 1.637)0.136<0.0031.267 (0.947 to 1.696)0.188<0.111
Year (reference year is 2018)Reference year
20191.055 (0.970 to 1.148)0.045<0.210
Constant0.041 (0.011 to 0.142)0.026<0.0010.020 (0.004 to 0.111)0.018<0.001
  • a Locum use is defined as practice aggregate FTE of locum doctors. QOF performance is measured as % achievement of the population across all QOF indicators. Coefficients can be interpreted as percentage change, for example, adjusted locum use in London was 0.45% lower than the East of England (Model A). CI = confidence interval. CQC = Care Quality Commission. FTE = full-time equivalent. IMD = Index of Multiple Deprivation. IRR = incidence rate ratio. QOF = Quality and Outcomes Framework.