Introduction
Type 1 diabetes is associated with a substantial reduction in life span [
1] and a threefold increase in the rate of cardiovascular disease compared with individuals without diabetes and remains a common cause of end-stage renal disease and loss of vision [
2]. Poor glycaemic control as indicated by HbA
1c is a key determinant of such complications and lowering HbA
1c reduces complications and prolongs survival rate [
3]. However, achieving good levels of control remains a challenge in all countries. In an international study of type 1 diabetes in 19 countries in 2014, most people with type 1 diabetes had higher than recommended levels of HbA
1c [
4]. Of those aged 15 years and more, median levels of HbA
1c were highest in Scotland.
As has happened to varying extents in other high-income countries, the publicly funded National Health Service (NHS) Scotland has employed several important changes to improve glycaemic control in type 1 diabetes in recent years [
5]. Provision of insulin pumps has increased from 8.4% to 34.4% in those under 18 years and from 2.5% to 8.3% in adults between 2011 and 2016. In addition, during this period, policies were instituted to enhance access to early structured education and provision of psychological interventions and there was a slight expansion of continuous glucose monitoring (CGM). A national survey showed that the proportion of individuals with type 1 diabetes who achieved HbA
1c ≤58 mmol/mol (7.5%) in Scotland slightly improved from 21.5% in 2013 to 24.5% in 2016 [
6]. This survey reports the overall population HbA
1c annually but does not test whether year-on-year changes represent significant trends or random fluctuations and does not explore detailed trends by age, sex or socioeconomic strata. Therefore, we analysed a nationwide diabetes register in Scotland enriched for patient characteristics and repeated measurements of HbA
1c to assess whether the significance of trends was beyond random fluctuations and to measure their consistency across age group, sex and socioeconomic strata. We sought evidence of whether healthcare innovations have had any impact on HbA
1c in this high-income country.
Discussion
We present trends in glycaemic trajectories in type 1 diabetes for a national population across a 13 year period, showing a small overall improvement in HbA
1c levels between 2004 and 2016 of about 4 mmol/mol and a corresponding decrease in the percentage of those with poor glycaemic control. Nonetheless, by 2016, more than one-third of all those with type 1 diabetes still had poor glycaemic control and most did not achieve HbA
1c targets, particularly those in late adolescence/early adulthood. That noted, the largest improvement in control was seen in the two youngest age groups, which if such improvements are sustained over time is encouraging, given some evidence that those who develop diabetes younger also have the highest risks for adverse cardiovascular outcomes [
14]. However, we also found large socioeconomic differentials in HbA
1c that did not alter in this time period.
We do not know which aspects of diabetes management may have altered HbA
1c during this period. However, the timing and larger reduction in younger people from 2012 is consistent with an impact of age-specific policy changes. A major policy change to quickly increase provision of insulin pumps in Scotland was introduced in 2011, and initially was mostly targeted towards children [
6]. Although a recent study in England and Wales did not suggest any benefit in HbA
1c reduction with insulin pumps compared with multiple daily insulin injections in children and adolescents [
15], other studies have reported improvements [
16]. Apart from insulin pump policies, the larger improvement in children may have reflected other differences in services between paediatrics and adult clinics (e.g. the introduction of insulin pump therapy was accompanied by more widespread application of structured education). However, there have also been a number of focused initiatives among the 12 paediatric clinics that care for all of the paediatric population with type 1 diabetes in Scotland in this time period, including ensuring comprehensive education on carbohydrate counting and dynamic insulin dosing from diagnosis, regular meetings of the 12 leads from these centres at which key metrics on glycaemic control and policies are reviewed. Over the time period studied, there was no improvement in BMI or smoking, which show some relationship with HbA
1c.
Although findings in older age groups were less marked, there were reductions in the prevalence of poor glycaemic control in all age groups. Apart from insulin pumps, the benefit of other measures to improve glycaemic control remain unclear. The Scottish Government introduced a 2 year funded study to support Psychology in Diabetes, Psychology and Diabetes (PiD-PaD) to improve self-management of diabetes [
17] but this psychology support is still not widely available. Structured education, on the other hand, was recently shown to cost-effectively improve glycaemic control with or without insulin pump for adults [
18]. Other potential contributing factors include the increasing availability of SCI-Diabetes data in 2011, allowing health centres to compare achievement of glycaemic control. In March 2014, the first national comparison of HbA
1c data for the 0–18 years age range appeared and the data have been discussed at the National Paediatric Diabetes Multidisciplinary Team annual meeting since then. Starting in 2014, there was a national campaign to standardise and tighten glycaemic targets for individuals with type 1 diabetes [
17] and in January 2016, a national Scottish meeting set several key core targets, which have been cascaded across Scotland with the use of ‘Know your HbA
1c charts’. At present, we are unable to assess these specific measures across the datasets but future studies may look into types of insulin and change between regimens, as well as the emerging expansion of flash glucose monitoring (FGM). These trend data from this high-income country are encouraging, yet they also emphasise that even in such a resource-rich setting, wherein the NHS is free at the point of delivery and there is a concerted national policy, there remains an enormous challenge in achieving HbA
1c targets levels in most individuals with type 1 diabetes. It is worth noting the persistently poor glycaemic control in those aged 19–24 years. While transition from paediatric to adult care is rightly considered important, these data suggest a significant problem possibly initiated in but extending beyond the transition/transfer period. Strategies to improve control in this vulnerable age group must address issues of healthcare disengagement, including new models of care, greater accessibility and wider availability of services such as clinical psychology.
It would be interesting to evaluate whether other countries have achieved greater gains over this period. The Diabetes-Patienten-Verlaufsdokumentation (DPV) database in Germany and Austria showed that despite substantial improvements in pump availability and other care aspects expected to improve HbA
1c, HbA
1c actually increased between 2002 and 2011 before falling thereafter [
19]. Data from the Swedish National Diabetes Registry reveal that HbA
1c increased by 2 mmol/mol (2.3%) between 2007 and 2012 and decreased afterwards until 2017 [
20]. Data from the USA show that mean HbA
1c levels were 66 mmol/mol (8.2%) in individuals enrolled into the T1D Exchange Clinic Network in 2010–2012, rising to 68 mmol/mol (8.4%) in the same individuals in 2013–2014 [
21]. This increase was greatest among those aged 13–17 and 18–26 years. Also in the USA, the National Health and Nutrition Examination Survey showed an increasing proportion of individuals with HbA
1c ≤53 mmol/mol (7%) from 1999–2002 to 2003–2006 in adults with any diabetes, followed by a plateau until 2011–2014 [
22]. Although mean HbA
1c values in individuals with type 1 diabetes in England seemed to be stable between 1998 and 2013 [
23], the percentage of individuals achieving HbA
1c ≤58 mmol/mol (7.5%) decreased from 28.7% in 2009–2010 to 27.0% in 2011–2012 before increasing to 30% in 2016–2017 in England and Wales [
24]. Increased cost sharing may have explained the plateau in glycaemic control attainment in the USA, whereas population changes may have contributed towards the temporary increases in HbA
1c in other countries. Changes in care process may also play a role as HbA
1c attainment varies across centres in Germany, Austria, England, Wales, USA, Sweden, Denmark and Norway [
25].
We noted an increase in HbA
1c between 2010 and 2012 in most age groups, both sexes, all socioeconomic strata and all health boards. Of note, the denominator population in Scotland in our data was fairly stable during this time and such increase was seen across all age groups. A potential explanation for this increasing trend was the policy to adopt IFCC units (mmol/mol) to replace the conventional DCCT unit (%) for HbA
1c measurements. From June 2009, a dual reporting method with both the DCCT units and IFCC units was used in Scotland during a short adaptation period for both clinicians and patients before fully transitioning to IFCC units from October 2011 [
26]. The impact of this change, particularly on patient care, remains unclear. Similar increases in mean HbA
1c that coincided with the IFCC standardisation have been reported in Sweden [
20]. Therefore, our 2010–2012 findings may well have resulted from biases related to the method of HbA
1c reporting rather than real increases per se.
Despite the encouraging improvement in population HbA
1c, our data showed that there are large persistent unchanging socioeconomic inequalities in HbA
1c across all age groups. In 2016, HbA
1c in the most-deprived residential category was around 8 mmol/mol (2.9%) higher compared with HbA
1c in the least-deprived category. To put this into context, the DCCT trial data suggest that a relative difference of 10% in HbA
1c may lead to a difference of 30–60% in microvascular complications of diabetes [
27]. Although these data do not allow us to determine the cause of the differential we observed, we previously reported (in a subset of one-third of adults with type 1 diabetes in Scotland) that those living in more-deprived areas had a lower frequency of injections of insulin per day, lower pump use, lower numbers of glucose monitoring per day and were less likely to use carbohydrate counting [
28] and by inference were less likely to have received structured education. Correspondingly, in other countries, HbA
1c has been reported to be higher among people of lower social class and lower educational attainment [
29]. This may contribute to the socioeconomic inequalities of complications in type 1 diabetes, such as diabetic retinopathy and foot ulceration [
30]. Our findings therefore prompt the need to ensure the achievement of adequate glycaemic control equally across the spectrum of socioeconomic status. It is particularly important to ensure that recent innovations expected to improve glucose management in diabetes in future, such as CGM and FGM, and widening coverage of pump availability, reach all of those in need across socioeconomic strata.
We also noted that sex differences in HbA
1c levels persisted over time, with better glycaemic control in men than in women. Higher HbA
1c in girls, compared with boys, at time of first diagnosis with type 1 diabetes have been reported [
31]. The higher HbA
1c levels in women may underestimate the true difference, since anaemia, more common in women is expected to lower HbA
1c levels [
32]. The magnitude of these sex differences is slight in comparison with the magnitude of the socioeconomic differences.
The strength of our study lies in the population-based data (99.5% coverage) with repeated measures of HbA1c for over a decade, which allowed us to estimate long-term glycaemic trends. A limitation is our use of an area-based rather than individual measures of socioeconomic status. In addition, we do not yet have sufficient individual-level data on new insulin delivery systems and other innovations, including flash monitoring and structured education in diabetes management, to enable a direct assessment of the impact within person before and after changes in treatment. This will be the subject of future research when the information becomes available.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.