Diabetes Distilled: To screen, or not to screen, that is the question

“All screening programmes do harm; some do good as well, and of these, some do more good than harm at reasonable cost.”
Sir Muir Gray, former Director of the UK National Screening Committee (Gray et al, 2008)

Population screening for type 2 diabetes is instinctively a good strategy to reduce the morbidity and mortality associated with uncontrolled hyperglycaemia and the associated costs to the NHS. However, the evidence base for type 2 diabetes screening at the population level is limited with regard to impact on health outcomes.

Previously, Shaw (2017) examined past evidence and found that, overall, there was little benefit directly related to centrally organised screening programmes, partly due to high levels of opportunistic screening in control groups. However, the author did highlight the influence of newer cardioprotective therapies for type 2 diabetes (i.e. SGLT2 inhibitors and GLP-1 receptor agonists), which have been found to reduce the risk of major adverse cardiovascular events (MACE) in people living with type 2 diabetes at high cardiovascular risk. If these therapies also demonstrated benefits in people with type 2 diabetes at lower cardiovascular risk, this might sway the argument for screening. Indeed, recent studies such as DECLARE-TIMI 58 (dapagliflozin) and REWIND (dulaglutide) have demonstrated improvements in MACE in cohorts with mixed cardiovascular risk (Wiviott et al, 2019; Gerstein et al, 2019).

In England, the NHS Health Check screens for type 2 diabetes in adults aged 40–74 years, whereas in the US, the US Preventive Services Taskforce (2021) recommends screening for prediabetes or type 2 diabetes in adults aged 35–70 years who have overweight or obesity.

The population-based, cohort study reviewed here used UK Biobank data to estimate the reduction in time to diagnosis of type 2 diabetes that could be achieved with population screening using HbA_1c in adults aged 40–70 years. In the UK Biobank, HbA_1c was checked at enrolment and this result was not relayed back to participants or their clinicians, thus allowing an opportunity to assess any improvement in time to a diagnosis of type 2 diabetes with HbA_1c screening.

A total of 179 923 Biobank participants had an HbA_1c measured at enrolment. Of these, 13 077 (7.3%) had a pre-existing diagnosis of type 2 diabetes, while 1703 (1.0%) had undiagnosed diabetes according to an HbA_1c diagnostic threshold of ≥48 mmol/mol (6.5%). Median HbA_1c was 51.3 mmol/mol in the latter group.

The median time to clinical diagnosis for those with undiagnosed type 2 diabetes was 2.2 years, with a median HbA_1c of 58.2 mmol/mol at diagnosis. Notably, females with lower HbA_1c results and BMI measurements at enrolment appeared to have the longest delay in diagnosing type 2 diabetes.

The authors conclude that HbA_1c screening at enrolment of the UK Biobank study would have diagnosed type 2 diabetes cases a median of 2.2 years earlier than they received a clinical diagnosis. Furthermore, the gender inequality in diagnosis requires further exploration.

The authors do acknowledge some limitations of this study, principally that the UK Biobank has previously been established not to be a representative UK cohort. Participants are from less deprived areas, predominantly of white ethnicity and tend to have better health outcomes than the general UK population.

In conclusion, this well-conducted cohort study provides compelling evidence that population screening with HbA_1c can effectively reduce the time to diagnosis of type 2 diabetes in middle-aged adults in the UK. This has important implications for us in primary care: should we consider a routine HbA_1c check for all new patients registering with our practices?