Send to

Choose Destination
BMJ Open. 2019 Sep 6;9(9):e028278. doi: 10.1136/bmjopen-2018-028278.

Analysis of English general practice level data linking medication levels, service activity and demography to levels of glycaemic control being achieved in type 2 diabetes to improve clinical practice and patient outcomes.

Author information

Department of Diabetes and Endocrinology, Salford Royal NHS Foundation Trust, Salford, UK.
The School of Medicine and Manchester Academic Health Sciences Centre, The University of Manchester, Manchester, UK.
Res Consortium, Andover, UK.
Clinical Biochemistry, Walsall Healthcare NHS Trust, Walsall, UK.
Clinical Biochemistry, University Hospitals of North Midlands, Stoke on Trent, Staffordshire, UK.
Warwick Medical School, University of Warwick, Coventry, UK.



Evaluate relative clinical effectiveness of treatment options for type 2 diabetes mellitus (T2DM) using a statistical model of real-world evidence within UK general practitioner practices (GPP), to quantify the opportunities for diabetes care performance improvement.


From the National Diabetes Audit in 2015-2016 and 2016-2017, GPP target glycaemic control (TGC-%HbA1c ≤58 mmol/mol) and higher glycaemic risk (HGR -%HbA1c results >86 mmol/mol) outcomes were linked using multivariate linear regression to prescribing, demographics and practice service indicators. This was carried out both cross-sectionally (XS) (within year) and longitudinally (Lo) (across years) on 35 indicators. Standardised β coefficients were used to show relative level of impact of each factor. Improvement opportunity was calculated as impact on TGC & HGR numbers.


Values from 6525 GPP with 2.7 million T2DM individuals were included. The cross-sectional model accounted for up to 28% TGC variance and 35% HGR variance, and the longitudinal model accounted for up to 9% TGC and 17% HGR variance. Practice service indicators including % achieving routine checks/blood pressure/cholesterol control targets were positively correlated, while demographic indicators including % younger age/social deprivation/white ethnicity were negatively correlated. The β values for selected molecules are shown as (increased TGC; decreased HGR), canagliflozin (XS 0.07;0.145/Lo 0.04;0.07), metformin (XS 0.12;0.04/Lo -;-), sitagliptin (XS 0.06;0.02/Lo 0.10;0.06), empagliflozin (XS-;0.07/Lo 0.09;0.07), dapagliflozin (XS -;0.04/Lo -;0.4), sulphonylurea (XS -0.18;-0.12/Lo-;-) and insulin (XS-0.14;0.02/ Lo-0.09;-). Moving all GPP prescribing and interventions to the equivalent of the top performing decile of GPP could result in total patients in TGC increasing from 1.90 million to 2.14 million, and total HGR falling from 191 000 to 123 000.


GPP using more legacy therapies such as sulphonylurea/insulin demonstrate poorer outcomes, while those applying holistic patient management/use of newer molecules demonstrate improved glycaemic outcomes. If all GPP moved service levels/prescribing to those of the top decile, both TGC/HGR could be substantially improved.


hba1c outcome; prescribing; primary care; type 2 diabetes

Free full text

Conflict of interest statement

Competing interests: None declared.

Supplemental Content

Full text links

Icon for HighWire
Loading ...
Support Center