Five-year trajectories of HbA1c by age, sex, ethnicity and deprivation in adults with newly diagnosed type 2 diabetes: Observational study in England

Abstract

The burden of diabetes mellitus (DM) is increasing worldwide, putting significant pressure on healthcare systems. Diabetes is often managed in primary care and includes a significant proportion of adults needing treatment. HbA1c is considered the gold standard for monitoring overall glycaemic level control: while guidelines generally recommend maintaining HbA1c levels below 52 mmol/mol (7%) for most individuals with diabetes, personalized management strategies are advocated.¹ However, the regular monitoring of HbA1c in real-world settings differs from that in controlled clinical trials.

This study investigates five-year HbA1c trajectories in newly diagnosed individuals with type 2 diabetes (T2DM) in routine primary care clinical practice in relation to age, sex, ethnicity and socioeconomic deprivation. We aim to identify disparities in HbA1c trajectories arising from these demographic factors, and to inform the development of more personalized and equitable approaches to diabetes care by identifying subgroups that may require targeted interventions or additional strategies for better diabetes glycaemic management.

We conducted this retrospective observational study, with approval from the Independent Scientific Advisory Committee (ISAC protocol No. 20_122), following the RECORD guidelines (checklist in the Supplementary material).

From an initial sample of 788 069 individuals, we excluded 237 421 without linkage to hospital records and 9944 with missing data on ethnicity and deprivation, leaving 540 704 individuals for the analyses reporting 3 886 451 measurements of HbA1c (1 396 464 in GOLD and 2 489 987 in Aurum database) during the 5 years of follow-up. The mean and median HbA1c were 51.0 (SD 36.4) mmol/mol and 48 (IQR: 41.0–58.5) mmol/mol, respectively. The median age at diagnosis was 62 (IQR: 52–71) (Table 1). There were fewer females (244 760 [45.3%]) than males. The largest proportion of participants (64 732; 12.0%) fell into the second most deprived group (ninth decile), while the smallest proportion (42 611, 7.9%) into the first decile. The majority of the cohort was of White ethnicity (49.9%), followed by Others/Unknown (37.4%), South Asian (8.2%) and Black (4.6%) ethnicity.

Across all groups, HbA1c levels showed an initial decrease until the first year from the diagnosis and an increase thereafter (Figure 1). The largest differences were observed for age and ethnicity: in terms of age, those between 18 and 30 years had an average HbA1c of 62 (95% CI: 61, 63) mmol/mol (results reported in % in Supplementary text) at diagnosis, 56 (55, 58) at 1 year and 64 (60, 67) at 5 years; corresponding estimates in the oldest group (≥80 years) were 52 (51, 52), 46 (45, 46) and 47 (47, 48). In White and Others/Unknown ethnicities, the trajectories of HbA1c were similar: at diagnosis, 1- and 5-year values were 55 (55, 55), 48 (48, 48) and 52 (52, 52) mmol/mol, respectively, in White; and 54 (54, 54), 48 (47, 48) and 51 (51, 52) in Others/Unknown ethnicity. The levels were similar, albeit higher for Black (56 [56, 57], 51 [51, 51] and 55 [54, 56] mmol/mol at diagnosis, 1 and 5 years, respectively) and South Asian (56 [55, 56], 52 [52, 53] and 56 [55, 57]) ethnicities. Variations were minimal across sex or deprivation levels (Figure 1).

We examined interactions between time and then two demographic factors in separate linear regression models. When combined with time, we found significant interactions with age and sex, age and deprivation, sex and ethnicity (p < 0.001) yet found no interactions between age and ethnicity (p = 0.130) and sex and deprivation (p = 0.180) (see Table S1).

While previous studies have summarized HbA1c trajectories in relation to outcomes of glucose-lowering therapies,^{6, 7} we detailed continuous five-year fluctuations in HbA1c levels across patients' age, sex, deprivation and ethnicity using large, real-world data. HbA1c decreased from diagnosis of T2DM to 12 months; however, there was a concerning reversal from 1 to 5 years post-diagnosis seen across all groups. The initial HbA1c improvement (downward trend) may reflect the positive impact of early interventions like medication and lifestyle modifications, and patient adherence. These patterns signify the importance of sustained, long-term disease management, highlighting the need for regular monitoring of HbA1c levels and possibly compelling modifications in the treatment strategy to maintain optimal blood glucose control over time.

Using the IMD-2010 index, we found minimal differences in HbA1c levels during follow-up between the ‘least’ and ‘most’ deprived groups, in contrast to previous findings of clearer associations between deprivation and HbA1c.⁸ When comparing HbA1c by sex, males generally had slightly higher levels of HbA1c than females throughout the follow-up period, but differences were negligible. Larger differences in HbA1c were observed across ethnic groups: while all groups showed a decrease in HbA1c levels in the first year post-diagnosis, individuals in the South Asian and Black ethnicity groups showed higher HbA1c from diagnosis compared with the White ethnic group, similar to previous findings.⁸ Further exploration could aid disease management and address any sequelae associated with suboptimal blood glucose management. Finally, our research found that hyperglycaemia at diagnosis and age are inversely associated with HbA1c at diagnosis, highest in the youngest group across and during the 5 years of follow-up, in line with previous findings.⁹ Particularly concerning was the observation that after 5 years of treatment, the youngest age groups (<40 years) showed higher blood glucose levels than at diagnosis. This finding may be related to the higher rate of complications associated with early-onset type 2 diabetes (EOT2D).¹⁰ It also highlights potential limitations in treatment planning, which are mainly based on existing T2D in older populations. These results underscore the need for specific tailored interventions by healthcare professionals for EOT2D.¹¹ Further research is needed to clarify the aetiology of the lower HbA1c results over 5 years in the elderly age groups, and if it could be contributing to lower quality of life because of unnecessary polypharmacy.¹² Moreover, we were surprised to see that there were limited differences in HbA1c across deprivation deciles, contrary to previously published literature.¹³ Further investigation is needed to understand this conflicting finding.

Overall, the comprehensive analysis performed in this study showed differential HbA1c levels in T2DM patients when investigated by age, sex, ethnicity and deprivation; albeit when interaction analyses were conducted, we found significant interactions between some of the different factors and time. One of the main strengths of this study is its use of large real-world primary care data to track HbA1c levels over a five-year period. We found that, despite the availability of new treatments over the period of analysis, HbA1c levels remained poor for up to 5 years across all age groups. The findings show how this key clinical biomarker for T2DM management differed by multiple demographic factors. Further research should explore the development of strategies for higher-risk groups such as EOT2D and Black populations. Targeted interventions and policies to reduce these disparities are needed.

Francesco Zaccardi, David Kloecker and Suping Ling designed the study. Francesco Zaccardi and Kamlesh Khunti acquired research funding. Francesco Zaccardi defined the clinical codes. Suping Ling extracted the data. Mohammad R. Ali and Atanu Bhattacharjee cleaned the data. Mohammad R. Ali, Francesco Zaccardi and Atanu Bhattacharjee analysed the data. Mohammad R. Ali, Francesco Zaccardi, Atanu Bhattacharjee and Gayathri Victoria Balasubramanian drafted the article. All authors contributed to the interpretation of the data, critically revised the article and approved the final version. Mohammad R. Ali, Francesco Zaccardi and Atanu Bhattacharjee had full access to all the data. Francesco Zaccardi is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data.

Kamlesh Khunti has acted as a consultant, speaker or received grants for investigator-initiated studies for AstraZeneca, Novartis, Novo Nordisk, Sanofi-Aventis, Lilly, Merck Sharp & Dohme, Boehringer Ingelheim and Bayer. Melanie Davies has acted as a consultant/advisor and speaker for Eli Lilly, Novo Nordisk and Sanofi, has attended advisory boards for Amgen, AstraZeneca, Biomea Fusion, Carmot/Roche, Sanofi, Zealand Pharma and Regeneron, and as a speaker for AstraZeneca and Boehringer Ingelheim. She has received grants from AstraZeneca, Boehringer Ingelheim and Novo Nordisk. Francesco Zaccardi has received consultancy fees from Daiichi Sankyo, Menarini and Servier for projects not related to this investigation. All other authors declare that there are no relationships or activities that might bias, or be perceived to bias, their work.