Inadequate HbA1c reduction: add, switch or continue?

  • McGovern AP & al.
  • BMC Med
  • 12 Apr 2019

  • curated by Sarfaroj Khan
  • UK Clinical Digest
Access to the full content of this site is available only to registered healthcare professionals. Access to the full content of this site is available only to registered healthcare professionals.


  • Continuing and switching medication demonstrated modest improvement in glycated haemoglobin (HbA1c) over the following year, but neither approach improved HbA1c much beyond the baseline HbA1c prior to the first new therapy.
  • Adding another new therapy to an apparently ineffective new therapy was the only approach associated with clinically significant improvement in HbA1c.

Why this matters

  • At present, there is no clarity on the approach to follow when patients with type 2 diabetes (T2D) have had a limited response to a recently introduced medication.
  • Some guidelines suggest stopping apparently ineffective therapy, but no studies have addressed this issue.

Study design

  • Retrospective analysis of patients with T2D (n=55,530; baseline HbA1c >58 mmol/mol [7.5%]) starting a second or third ever glucose-lowering medication between 2004 and 2017.
  • In patients with limited response at 6 months (HbA1c fall
  • Funding: The MASTERMIND consortium is funded by the UK Medical Research.

Key results

  • Limited reduction in HbA1c was recorded in 21.9% (12,168) of patients with a mean increase of 2.5 mmol/mol (0.2%).
  • In patients with limited response, 8939 (73.5%) continued therapy, 1119 (9.2%) switched medication and 2110 (17.3%) added a new medication within the first year.
  • HbA1c response was slightly greater in patients who switched vs those who continued the same therapy (−6.8 mmol/mol [−0.6%]; 95% CI, −7.7 to −6.0 vs −5.1 mmol/mol [−0.5%]; 95% CI, −5.5 to −4.8).
  • Adding additional therapy to an apparently ineffective therapy was associated with higher HbA1c reduction (−12.4 mmol/mol [−1.1%]; 95% CI, −13.1 to −11.7).
  • Similar trends were seen in propensity score-matched subgroups.


  • Exclusion of people with incomplete follow-up data.
  • Information on the rationale for treatment decision was not available.