A new paper in Diabetologia describes how a unique dataset of HbA1c measurements drawn from the UK Biobank reveals the extent of undiagnosed diabetes. The findings suggest that population-wide screening could lead to earlier clinical diagnosis, earlier intervention and, perhaps, fewer long-term complications. Dr Susan Aldridge reports.
Screening can detect undiagnosed diabetes earlier than either symptomatic or incidental diagnosis. This allows for earlier intervention – lifestyle change, medication or both – which may reduce the risk of later complications.
There are already diabetes screening programmes in many countries, which differ according to their eligibility criteria and population coverage. In England, there is the NHS Health Check, which was initiated in 2009 and provides screening for high-risk adults aged 40-74, while the American Diabetes Association (ADA) recommends that all US adults aged 35 or older be screened.
Studies on these programmes are important as they can tell us how many cases of undiagnosed diabetes are identified by screening. However, since there is a direct link between screening and diagnosis in these studies, they don’t measure how much earlier diagnosis by screening occurs compared with diagnosis in routine clinical care. This would reveal the reduction in the time people are living with undiagnosed diabetes that could be achieved by screening and help policymakers decide whether it is worth the investment.
The UK Biobank is a population-based cohort of over 500,000 UK residents aged 40-70 at enrolment (between 2006 and 2010). HbA1c is measured on enrolment, but the results are not fed back to participants or their clinicians. These measurements therefore represent a rich resource that uncouples screening from diagnosis because the Biobank is also linked to participants’ routine healthcare data up to 10 years post-enrolment. This allows assessment of the prospective time to diabetes diagnosis in routine care following an elevated (≥48 mmol/mol) HbA1c at enrolment among those who did not have a pre-existing diabetes diagnosis.
Dr Nicholas Thomas, Dr John Dennis and colleagues from the Exeter Centre of Excellence in Diabetes (EXCEED), UK, have used this UK Biobank data to estimate the reduction in time to diagnosis that could be achieved by implementing population-level HbA1c screening compared with routine care. They also looked at the participant characteristics associated with longer time to diagnosis.
Mining the UK Biobank
The study started with a cohort of 179,923 participants from the UK Biobank who had both an enrolment HbA1c measurement and available longitudinal primary care data up to 2016/2017. Of these, 13,077 proved to have a pre-existing diabetes diagnosis, leaving a study population of 166,846 participants.
Undiagnosed diabetes was defined as having an HbA1c measurement of ≥48 mmol/mol at enrolment. Time to diabetes diagnosis in routine care was calculated as the time between the HbA1c measurement and the date of a code for diabetes, prescription for glucose-lowering medication, HbA1c of ≥48 mmol/mol, fasting glucose of ≥7.0 mmol/l or random/two-hour postprandial glucose of ≥11.1 mmol/l in primary care medical records or records of a code for diabetes in secondary care medical records.
In total, 1% – 1703 – of the study group had undiagnosed diabetes, which is an extra 13% of diabetes cases relative to the 13,077 who had a pre-existing diagnosis. This gives us a rough idea of the possible extent of undiagnosed diabetes among the UK population. Compared with those in the study population with HbA1c <48 mmol/mol, those with undiagnosed diabetes were, on average, older, had a higher BMI, were more likely to be male, live in more deprived areas and self-report being of non-White ethnicity. They were similar in most respects to those with a pre-existing diagnosis, but were more likely to self-report being of non-White ethnicity (median 11.4% vs 9.8%) and had a slightly higher BMI (median 30.9 vs 30.1 kg/m2).
Time to diagnosis
So how did these participants with undiagnosed diabetes fare over the next few years? Most of them – 87.7% – did receive a clinical diagnosis of diabetes during the median of 7.3 years follow-up. During this time, HbA1c tended to increase – the median value for those who had it measured at diagnosis in primary care was 58.2 mmol/mol.
However, this increase wasn’t inevitable – 6.9% of the participants who had undiagnosed diabetes at enrolment in the Biobank subsequently had HbA1c below 48 mmol/mol recorded in primary care, but half of this group subsequently received a diabetes diagnosis.
Median time to diagnosis varied slightly with the year the participants had enrolled in the Biobank – for 2008, it was 2.3 years; for 2009, 2.2 years; for 2010, it was significantly shorter at 1.8 years. The EXCEED team quote 2.2 years as median delay time to diagnosis based on this study population. Having a higher BMI and higher enrolment HbA1c were both associated with a shorter time to diagnosis, but there was no clear evidence of an association between age, ethnicity or socioeconomic status and time to diabetes diagnosis.
The value of HbA1c screening
This is the first study to use real-world clinical data to show how much earlier a diabetes diagnosis might be made by implementing a population-based screening programme. It shows that HbA1c screening at UK Biobank enrolment could have identified these cases of undiagnosed diabetes a median of 2.2 years earlier before receiving a clinical diagnosis in routine care.
This finding is similar to that of the Ely study of 40-65 year olds, which compared outcomes for participants randomised to screening at five-yearly intervals with outcomes for those receiving no screening. The reduction in time to diagnosis was 3.3 years.
Meanwhile, the prevalence of undiagnosed diabetes of 1% found in this study population is similar to the 1.4% estimate derived by modelling done by Diabetes UK, confirmed by a 2003-2005 UK National Screening Committee (NSC) pilot, which also showed a prevalence of 1.4%. The prevalence of undiagnosed diabetes in older populations has been found to be higher, with the NSC study showing a prevalence of 2.8% in those over 40 and, in the Ely study, 4.5%.
The finding that male sex, higher HbA1c and BMI >30 kg/m2 are associated with shorter time to diagnosis suggests that clinicians are more likely to screen men and those with obesity. The relative delay in diagnosis for women is interesting as it does not arise from differences in age, BMI, HbA1c deprivation or ethnicity. This association between sex and delayed diagnosis has not been observed in other studies, so maybe this is peculiar to UK primary care or to the UK Biobank cohort. The link with higher HbA1c and earlier diagnosis is easily explained by those with higher blood glucose being more likely to have diabetes symptoms earlier, prompting earlier diagnosis.
A key strength of this study was that the systematic baseline HbA1c assessment of UK Biobank participants was not fed back to either them or their clinicians, providing the EXCEED team with a unique dataset to evaluate the benefits of HbA1c screening on time to diagnosis compared with diagnosis in routine care. In this study, 87.7% of those with undiagnosed diabetes eventually received a clinical diagnosis of the condition, while only 6.9% reverted to an HbA1c below the diagnostic threshold.
A limitation is that the UK Biobank is not a representative UK cohort – previous studies show that participants have better health outcomes, are from less deprived areas and are more likely to be of White ethnicity than the general UK population. Given the known association between non-White ethnicity and higher social deprivation and increased risk of diabetes, the rate of undiagnosed diabetes is likely to be higher in the wider UK population than in the UK Biobank cohort.
In addition, those volunteering for the Biobank are likely to be more health conscious than average, as are volunteers in other research studies. This means they may have more healthcare appointments and are likely to be diagnosed sooner. This suggests that population-based screening initiatives could identify even more cases of undiagnosed diabetes and reduce time to diagnosis more than the 2.2 years seen in this study.
Finally, most of the cohort had not accumulated sufficient follow-up data to reliably evaluate the impact of delayed diagnosis upon diabetes complications. The UK Prospective Diabetes Study (UKPDS) showed that earlier intensive blood-glucose control reduces the risk of later complications. We would therefore expect earlier diagnosis to lead to fewer complications in the long term. However, thus far, trials of screening interventions have not shown a significant reduction in all-cause mortality. This is worthy of further exploration when more recent UK Biobank-linked primary care data become available.
To read this paper, go to: Young K, McGovern A, Barroso I, Hattersley A, Jones A, Shields B, Thomas N, Dennis J. The impact of population-level HbA1c screening on reducing diabetes diagnostic delay in middle-aged adults: a UK Biobank analysis. Diabetologia online 22 November 2022. https://doi.org/10.1007/s00125-022-05824-0
Any opinions expressed in this article are the responsibility of the EASD e-Learning Programme Director, Dr Eleanor D Kennedy.