Antidepressant Prescribing Insight
NHS BSA Open Data
PROTOTYPE
PHO
Regional variation

Where is mental health demand growing fastest?

Pick a region to compare its prescribing trend against the national average. Differences in growth rate tell us where mental health services are most strained.

Regional comparison

Click a region chip to highlight.

Growth ranking

RegionJan 2021Oct 2024Growth
December 2024 deep-dive

One month, 27 drugs, £16.6 million.

A close look at the most recent month for which we have a fully detailed drug-level breakdown across all 42 Integrated Care Boards.

Total items
across England
Total cost
drug only, ex. dispensing fees
Most prescribed
Sertraline
2.08M items, 27% of total
Most expensive
Venlafaxine
£3.83M total, 23% of cost

Volume vs cost ranking

Same drugs, ranked two ways. The gap between volume rank and cost rank reveals which medicines are expensive per dose.
#DrugItemsCost£ / item

Cost by region

Three insights for commissioners
1. Sertraline dominates volume (27% of all items) but is cheap per dose, accounting for 21% of cost.
2. Venlafaxine is the cost outlier: 6th in volume but 1st in cost. It is worth understanding why the per-item price is so high.
3. The Midlands has the highest absolute spend (£3.25M), reflecting both higher demand and a larger population base.
Predictive view

Looking six months ahead.

A simple linear-trend-plus-seasonality model fitted to 46 months of history. Useful for capacity planning, not for clinical decisions.

Model R²
items model fit
Trend slope
+items per month
Apr 2025 forecast
items prescribed nationally
YoY change
vs same month 2024

Six-month forecast

Solid line: actual. Dashed line: model prediction with ±5% band.
How is this calculated?
Linear regression on a time index plus eleven monthly dummy variables to capture seasonality. Fits 82% of the variance in items but only 34% of the variance in cost — cost is much more volatile because it is driven by patent expiries and procurement decisions rather than smooth demand.
Methodology and provenance

About PrescribeIQ.

Built as a prototype dashboard for the NHS DigData Step Up Career Challenge. Combines one month of detailed Excel data with four years of longitudinal Python analysis.

Data sources

  • Snapshot: Step Up Excel Data, December 2024 (1,120 rows, 27 drugs, 42 ICBs)
  • Trend: NHS BSA PCA Regional Summary, Jan 2021 to Oct 2024 (322 rows)
  • Drug detail: NHS BSA PCA Regional + Drug Summary (9,455 rows)
  • Provenance: NHSBSA Open Data Portal, official statistic

Methods

  • Aggregation: SUMIF / SUMIFS in Excel, groupby in pandas
  • Forecast: Linear regression with monthly dummies
  • Validation: Dec 2024 forecast was within 1.3% of actual
  • Charts: Chart.js in this app, matplotlib for static reports

Limitations

  • Item counts not normalised by population size
  • "UNIDENTIFIED" region grouped with England-level totals
  • Cost reflects drug only, excludes dispensing fees
  • Forecast is a trend model, not a causal one

Next iterations

  • Per-capita normalisation using ONS regional populations
  • Drug-level forecasting, not just national
  • Drill-through to ICB level on the regional map
  • SARIMA model for stronger forecast intervals