Table 6.
Subgroup analyses for dispensing outcome
| Variable* | No (I:C) | Rate (95% CI) | Subgroup specific adjusted rate ratio (95% CI)‡ |
P value§ | |
|---|---|---|---|---|---|
| Intervention (I)† | Control (C)† | ||||
| S5. Proportion of locums: | |||||
| <25% | 54:41 | 0.147 (0.129 to 0.168) | 0.157 (0.137 to 0.179) | 0.937 (0.904 to 0.971) | 0.005 |
| ≥25% | 48:58 | 0.156 (0.135 to 0.180) | 0.174 (0.151 to 0.200) | 0.958 (0.925 to 0.993) | |
| S6. Proportion of nurses: | |||||
| <17% | 81:72 | 0.154 (0.137 to 0.173) | 0.165 (0.147 to 0.186) | 0.921 (0.896 to 0.947) | <0.001 |
| ≥17% | 62:74 | 0.149 (0.128 to 0.172) | 0.149 (0.129 to 0.173) | 1.051 (1.021 to 1.081) | |
| S7. Past dispensing rate: | |||||
| <18% | 68:80 | 0.112 (0.098 to 0.129) | 0.110 (0.096 to 0.126) | 1.031 (1.001 to 1.063) | 0.56 |
| ≥18% | 76:69 | 0.182 (0.165 to 0.199) | 0.194 (0.177 to 0.213) | 0.992 (0.967 to 1.018) | |
| S8. Practice site¶: | |||||
| 1 site | 95:90 | 0.149 (0.129 to 0.172) | 0.160 (0.139 to 0.185) | 0.928 (0.903 to 0.954) | <0.001 |
| ≥2 sites | 34:37 | 0.163 (0.141 to 0.188) | 0.158 (0.137 to 0.182) | 1.028 (0.990 to 1.067) | |
| S9. Follow-up completed before covid-19 pandemic: | |||||
| Yes | 31:26 | 0.171 (0.138 to 0.212) | 0.163 (0.131 to 0.202) | 0.994 (0.955 to 1.036) | 0.39 |
| No | 113:123 | 0.148 (0.130 to 0.167) | 0.152 (0.134 to 0.172) | 1.012 (0.991 to 1.035) | |
| S10. Level of deprivation: | |||||
| Low (rank** ≥14 387) | 72:74 | 0.150 (0.134 to 0.167) | 0.155 (0.139 to 0.173) | 0.955 (0.929 to 0.981) | 0.004 |
| High (rank** <14 387) | 72:75 | 0.154 (0.132 to 0.178) | 0.150 (0.129 to 0.174) | 1.061 (1.032 to 1.090) | |
CCG=clinical commissioning group; CI=confidence interval.
All subgroup analyses were pre-specified in statistical analysis plan,20 although outcomes S8-S10 were introduced after trial started.
Dichotomised on basis of median for continuous subgroups.
Rates taken from random effects Poisson regression, incorporating CCG as random effect.
Random effects Poisson regression, adjusting for baseline dispensing rate and incorporating CCG as random effect.
Taken from likelihood ratio test comparing random effects Poisson models with/without interaction term included, treating subgroup of interest as continuous variable where possible.
Taken from practice follow-up questionnaire.
Index of deprivation ranks every neighbourhood in England from 1 (most deprived area) to 32 844 (least deprived area).