Table 3. Subgroup analysis.

	Co-efficient	95% CI
Lower	Upper
2015–2016 QP in bottom 80% of prescribers		-0.116	-0.125	-0.106
Effect of 2015–2016 QP in top 20% of prescribers		-0.200	-0.210	-0.187
Top 20% of prescribers		0.309	0.302	0.316
Months since QP		0.013	0.013	0.014
Season	Winter	Ref	Ref	Ref
	Spring	-0.046	-0.049	-0.044
	Summer	-0.148	-0.151	-0.145
	Autumn	-0.124	-0.126	-0.121
Comorbidities prevalence per 100 patients	Respiratory diseases	0.013	0.011	0.015
	Diabetes (<3.93%)	0.097	0.089	0.106
	Diabetes (≥3.93% to ≤7.59%)	0.015	0.012	0.017
	Diabetes (≥7.60% to ≤11.28%) before QP	0.016	0.012	0.021
	Diabetes (≥7.60% to ≤11.28%) after QP	0.007	0.002	0.011
	Diabetes (>11.28%)	0.041	0.030	0.052
GPHC per 10 000 patients, spline terms	GPHC1 (<4.91) before QP	0.015	0.013	0.017
	GPHC1 (<4.91) after QP	0.007	0.005	0.009
	GPHC2 (≥4.91 to ≤9.80) before QP	-0.007	-0.009	-0.005
	GPHC2 (≥4.91 to ≤9.80) after QP	-0.005	-0.007	-0.003
	GPHC3 (≥9.81 to ≤14.72)	0.005	-0.000	0.010
	GPHC4 (>14.72)	-0.007	-0.022	0.008
Benzodiazepine anxiolytics prescription		0.093	0.089	0.097
Benzodiazepine hypnotics prescription		0.118	0.115	0.121

Effects of diabetes are per 1% higher within each spline term. Effects of GPHC are per 1 higher per 10 000 patients within each spline term. GPHC = GP headcount. QP = Quality Premium.