BMJ Open. 2026 Sep 22;16(9):e122364. doi: 10.1136/bmjopen-2026-122364.
ABSTRACT
OBJECTIVES: To assess differential changes in prescribing uptake of dapagliflozin and empagliflozin relative to canagliflozin in England following the July 2023 expansion of heart failure guideline recommendations.
DESIGN: Controlled interrupted time series (ITS) analysis using segmented regression with Newey-West SEs and month fixed effects to account for seasonality.
SETTING AND PARTICIPANTS: National monthly primary care prescribing data in England (January 2019-December 2025). Aggregate prescribing data were analysed at the drug level.
PRIMARY AND SECONDARY OUTCOME MEASURES: Primary outcome: monthly utilisation measured in defined daily doses (DDDs).
SECONDARY OUTCOMES: monthly prescription items, market share of total SGLT2 inhibitor DDDs and net ingredient cost (NIC).
RESULTS: Overall utilisation increased for all three monotherapy SGLT2 inhibitors, but prescribing trajectories diverged markedly following the intervention. Dapagliflozin's market share rose from 33.2% to 55.9%, while empagliflozin declined to 41.0% and canagliflozin to 3.1%. The controlled ITS model showed no immediate step change at July 2023. However, dapagliflozin demonstrated a significant differential postintervention slope increase relative to canagliflozin (β=0.042, 95% CI 0.035 to 0.049; p<0.001), corresponding to an additional 4.3% monthly growth. Empagliflozin showed a smaller but significant relative increase (β=0.015, 95% CI 0.008 to 0.022; p<0.001). Counterfactual analysis estimated 68.4 million excess dapagliflozin DDDs and 21.2 million excess empagliflozin DDDs over the postintervention period. Results were consistent across prespecified breakpoint and formulation sensitivity analyses; however, an exploratory NIC-adjusted analysis attenuated the dapagliflozin differential slope by 33%, indicating sensitivity to concurrent price changes.
CONCLUSIONS: SGLT2 inhibitor uptake in English primary care has been highly heterogeneous. The marked acceleration in dapagliflozin prescribing from July 2023, against stable background trends for canagliflozin, is temporally consistent with expanded heart failure guideline recommendations. However, several important limitations preclude definitive causal attribution. First, because this study analysed aggregate prescribing data without patient-level indications, we cannot directly attribute observed changes to heart failure rather than to diabetes, chronic kidney disease or other factors. Second, the substantial concurrent reduction in dapagliflozin's NIC (from £1.38 to £0.35 per DDD) represents a competing explanation; exploratory adjustment for NIC attenuated the dapagliflozin estimate by approximately one-third, though this magnitude cannot be interpreted as a causal attribution because NIC may be endogenous. The observation that empagliflozin, which experienced a more modest price change, showed a smaller differential effect is hypothesis-generating and requires confirmation with methods that can separate guideline from price effects. Third, the ecological design inherently limits causal inference. These findings highlight that class-wide guideline updates do not translate into uniform prescribing and underscore the importance of monitoring drug-specific adoption patterns while considering the complex interplay of clinical, economic and policy factors.
PMID:42772884 | DOI:10.1136/bmjopen-2026-122364