A five-drug class model using routinely available clinical features to optimise prescribing in type 2 diabetes: a prediction model development and validation study
Authors' conclusion
We have developed a five-drug class model that uses routine clinical data to identify optimal glucose-lowering therapies for people with type 2 diabetes. Individuals on model-predicted optimal therapy had lower 12-month HbA1c, were less likely to need additional glucose-lowering therapy, and had a lower risk of diabetes complications than individuals on non-optimal therapy. With setting-specific optimisation, the use of routinely collected parameters means that the model is easy to introduce to clinical care in most countries worldwide.
Study context
- Population
- People with T2D aged 18–79 years initiating one of five drug classes, Jan 1, 2004–Oct 14, 2020, England (CPRD Aurum); plus 3 published randomised trials
- Participants
- 212,166 drug initiations
- Comparison
- Model-concordant vs model-discordant therapy
- Outcome
- Observed 12-month HbA1c benefit in model-concordant vs matched model-discordant groups
- Follow-up
- 12 months
Methodological detailsobservational data (CPRD Aurum), holdback validation by geographical region and calendar period, external validation in 3 randomised trials, matched (1:1) concordant vs discordant groups, Model developed from 100 107 drug initiations in CPRD using nine routine clinical features
- Observational data (CPRD Aurum)
- Holdback validation by geographical region and calendar period
- External validation in 3 randomised trials
- Matched (1:1) concordant vs discordant groups
- Model developed from 100 107 drug initiations in CPRD using nine routine clinical features
Performance
As reported in the abstract
Mean HbA1c benefit (geographical validation cohort)
5.3 mmol/mol
95% CI 4.9–5.7
n=12 373 matched pairs; lower HbA1c with model-concordant
Mean HbA1c benefit (temporal validation cohort)
5.0 mmol/mol
95% CI 4.3–5.6
n=4841 matched pairs; lower HbA1c with model-concordant
- Drug initiations of model-predicted optimal therapy
- 32 305 (15.2%) of 212 166
Further findings
| Outcome | aHR | |
|---|---|---|
| Glycaemic failure (confirmed HbA1c ≥69 mmol/mol) · 5 years | 0.62 (95% CI 0.59–0.64) | |
| All-cause mortality · 5 yearsSimilar risk | 0.95 (CI 0.83–1.09) | |
| MACE-HF outcomes · 5 years | 0.85 (CI 0.76–0.95) | |
| Renal progression · 5 years | 0.71 (CI 0.64–0.79) | |
| Microvascular complications · 5 years | 0.86 (CI 0.78–0.96) | |
| 0.512Adjusted hazard ratio (log scale) |
Calibration of predicted vs observed HbA1c differences
- Well calibrated in the three clinical trials and in pairwise comparisons of the five drug classes in CPRD
Original publication
Composed by AI from the published abstract and checked against it. How Briefs are made →