Results
Prescribing of anti-dementia drugs between 2009 and 2019
The number of prescription items dispensed in England for the pharmacological treatment of dementia (anti-dementia drugs, BNF Section 4.11) increased by approximately three folds from 24 prescription items per 1,000 populations in 2009 to 70.9 prescription items per 1,000 populations in 2019, reflecting an increase of 195.4% (table 1, figure 1).
Donepezil was the most commonly prescribed anti-dementia drug both in 2009 and 2019, accounting for 64.0% and 52.0% of all anti-dementia prescribing respectively. Of all anti-dementia drugs, highest increase was noted for prescribing of memantine which increased from 4.5% in 2009 to 34.7% in 2019. Prescriptions for galantamine decreased over the same period by 40% (table 1).
The results presented an overall decrease in the cost incurred from prescribing drugs for dementia, increasing by 17.5% from 2009 to 2011 but then drastically decreasing by 82.0% from 2011 to 2019. This resulted in an overall decrease in costs of 81.7% from 2009 to 2019 (table 1, figure 2).
The costs of prescribing rivastigmine peaked in 2012 at £342.69 per 1,000 populations, followed by a substantial reduction in costs by 88.8% in the following year between 2012 and 2013. Overall, the costs of memantine increased by 61.8% from 2009 to 2019, peaking in 2013 at £434.54 per 1,000 populations. A significant reduction in the costs of donepezil by 40.0% was observed between 2011 and 2012. Donepezil’s cost per item decreased by the largest proportion from 2019 to 2009 which decreased by 98.8% over the 10-year period (table 1, figure 2).
Prescribing of antipsychotics drugs: 2009 – 2019
The prescribing of all antipsychotic drugs showed a steady increase of 37.6% over the 10-year period (table 1, figure 3). Antipsychotic prescribing was consistently higher than prescribing of anti-dementia drugs, however, the rate of increase in prescribing for anti-dementia was higher in the study period. In particular, prescriptions of haloperidol and risperidone also increased from 2009 to 2019 by 13.2% (table 1, figure 3). Costs of prescribing of all antipsychotic drugs decreased by 71.4% from 2009 to 2019 (table 1, electronic supplemental material 2). Both antipsychotics and anti-dementia drugs presented an initial increase in costs from 2009 to 2011, followed by a decrease from 2011 to 2016. Haloperidol and risperidone showed a decrease in costs from £421.10 to £301.79 (-28.33%) per 1000 populations from 2009 to 2019 (electronic supplemental material 2).
Association between deprivation and prescribing
Prescribing in the least deprived areas was substantially higher than that in the most deprived areas from 2015 to 2019 (table 2).  The median (IQR) prescription items for the least deprived CCGs for all five observations was 75.68 (72.32-86.67) compared with 44.62 (43.03-44.85) items per 1,000 populations in the most deprived areas (p = 0.008). Similarly, the median (IQR) costs of prescribing per 1,000 populations in the most deprived areas across all five observations was £255.18 (233.48-320.74) compared with £625.87 (563.25-654.78) in the least deprived areas (p=0.008) (table 2). There was a larger observed variation in both items and cost data within the most deprived areas compared to the least deprived areas (table 2).
Variations across NHS England Commissioning Regions
Prescription items for dementia increased from 2015 to 2019 in every NHS England commissioning region (electronic supplemental material 3). Prescribing was consistently highest in the North East and Yorkshire region and lowest in the Midlands for the duration of the measured time period. In 2009, the North East and Yorkshire region, a total of 85.6 prescription items per 1,000 populations were prescribed compared with 41.2 per 1,000 populations in the Midlands. Similarly, in 2019, the North East and Yorkshire region prescribed 101.4 items per 1,000 population, 96% higher than the 51.8 items prescribed in the Midlands (table n) (p=0.015). A significant difference in the median costs of prescribing was also observed (p=0.001). A weak positive correlation between prevalence of dementia and prescription of anti-dementia drugs was observed (Pearson’s correlation coefficient 0.371, p= 0.413).