Estimating productivity levels in primary medical services across clinical commissioning groups in England and the impact of the COVID-19 pandemic: a data envelopment analysis

Objectives To assess the relative productivity of primary medical services in England and the impact of the COVID-19 pandemic on productivity levels.Setting Primary medical services for 59 million patients (98% of the population in England), in 101 clinical commissioning groups (CCGs), across two time periods: period 1, pre-pandemic, April to December 2019 and period 2, pandemic, April to December 2020.Methods We use data envelopment analysis (DEA) to assess relative productivity with four input measures (the number of full-time equivalent general practitioners, nurses, other direct patient contact staff and administrators), and five output measures (face-to-face appointments, remote consultations, home visits, referrals to secondary care and prescriptions). Our units of analysis were CCGs. DEA assigns an efficiency score to a CCG, taking a value between 0 and 100%, by benchmarking it against the most productive CCGs. We use Tobit regression to examine the association between productivity and other factors.Results The mean bias-corrected efficiency score of primary medical services in CCGs was 92.9% (interquartile range 92.0% to 95.7%) in period 1, falling to 90.6% (interquartile range 86.8% to 95.2%) in period 2. In period 1, CCGs with a higher proportion of registered patients aged over 65 years, higher levels of deprivation, lower levels of disease prevalence, higher nurse to GP ratios and higher GP to other direct patient contact staff ratios, achieved statistically significantly higher general practice efficiency scores (p < 0.05). In period 2, only the ratio of GP to other direct patient contact staff was associated with efficiency scores (p > 0.05).Conclusions Our analysis indicates only modest geographic variation in productivity of primary medical services when measured at the level of clinical commissioning groups and a small reduction in productivity during the pandemic. Further work to establish relative productivity of individual GP practices is warranted once sufficient data on appointment rates by GP practice is available.