In this post hoc analysis of risk factors affecting the incidence of MI or stroke in a large cohort of people with T2D in Spain, baseline HbA1c levels < 6.5% were independently associated with lower risk of first MI and of first stroke. This result suggests that interventions to favor stringent HbA1c control (< 6.5%) could reduce the risk of major CV complications in people with T2D. Non-modifiable factors, such as male sex, prior CVD, and age ≥ 50 years were independently associated with increased risk of a first MI and of a first stroke, while an eGFR < 60 ml/min/1.73 m2 was also associated with increased risk of a first MI, but not of first stroke.

A relationship between lower HbA1c levels and reduced risk of macrovascular complications has been observed before in several long-term studies of people with T2D. A meta-analysis of several of these studies showed a 15% reduction in MI risk (HR 0.85 [95% CI 0.76–0.94]) for a 0.88% lower HbA1c [20]. A study of participants in the UK Prospective Diabetes Study (UKPDS) also revealed an up to 33% reduction in MI (p = 0.005) after a 10-year follow-up in patients with overweight receiving intensive therapy for glycemic control [21]. Further, a recent re-examination of the UKPDS data has shown that a sustained HbA1c level < 6.5% over a period of 5 years was associated with a significant reduction in MI risk [12]. Another study based on the Swedish National Diabetes Register of people with T2D (mean follow-up, 5.7 years) showed that HbA1c levels below 53 mmol/mol (7%) were associated with an acute MI of HR 0.84 (95% CI 0.75–0.93), and a stroke of HR 0.95 (95% CI 0.84–1.07) [5]. This study showed HbA1c levels were among the strongest predictors for acute MI risk. In Spain, a study of 11,003 people with uncontrolled T2D (HbA1c ≥ 6.5%) observed a linear and increasingly positive relationship between HbA1c levels and hospitalization due to coronary heart disease [10]. In this study, the relative risk for coronary heart disease and stroke hospitalization (comparing patients with and without uncontrolled diabetes) was 1.38 (95% CI 1.20–1.59) and 1.05 (95% CI 0.91–1.21), respectively.

The results of this post hoc analysis would suggest that stringent control of HbA1c levels could significantly help in reducing the risk of development of macrovascular complications later in life. Early intervention for stringent HbA1c control has been suggested before to reduce diabetes-related CV complications. For example, the UKPDS showed that one percentage unit lower HbA1c from the diagnosis of diabetes significantly lowered the risk of MI events 15 and 20 years later, compared with reducing HbA1c by the same amount from 10 years after diagnosis [12]. Also, the Diabetes & Aging Study found that, among patients with newly diagnosed diabetes and 10 years of survival, HbA1c levels ≥ 6.5% (≥ 48 mmol/mol) for the first year after diagnosis were associated with worse outcomes [11]. These results highlight the importance of early treatment intensification aimed at reaching the HbA1c target, as currently recommended by the ADA–European Association for the Study of Diabetes consensus, and of implementing strategies aimed at reducing the risk of CVD as early as possible after T2D diagnosis [15]. Although historically the treatment for T2D followed a step-by-step approach—where a new medication was added to the existing regimen to reach the glycemic target—current evidence justifies a more proactive approach, especially in those patients with established CVD, increased CVD risk, or long life expectancy [15, 22]. Early intensification can result in diabetes remission in some cases [23, 24]. Current guidelines suggest a HbA1c target for most adults, excluding pregnant women, of 53 mmol/mol (7%) or less [15]. A target of HbA1c < 7% can be reasonable if it can be accomplished in a safe manner without significant hypoglycemia or other treatment side effects, especially when using pharmacologic treatments, which are not associated with hypoglycemia risk [15].

In this post hoc analysis, the incidence of first MI or stroke was approximately two- to three-times higher in patients with prior CVD compared with patients without prior CVD. The molecular mechanisms involved in the damage caused by persistent hyperglycemia point to prolonged increases in reactive oxygen species production and altered secretion of inflammatory cytokines, among other processes [25,26,27]. Insulin resistance, hyperinsulinemia, and vascular calcification may lead to atherosclerosis and the formation of unstable plaques that ultimately can cause coronary events and stroke [27]. Given the added risk in people with T2D with prior CVD, current guidelines promote the early use of drugs with cardiorenal benefit in these patients, such as GLP-1 RAs or SGLT2is [15]. Although these drugs were originally introduced as glucose-lowering agents, they are now also recommended for organ protection based on the results from the CVOTs [9]. In this regard, it should be noted that although these drugs could have potentially affected the results of our study, the effects are likely to be very minor, as studies have shown a limited use of these drugs at the time of study data collection in Spain [28, 29]. In the population studied here, only 0.6% and 0.3% had prescriptions for GLP-1 RAs and SGLT2is, respectively, at index date (Table 1).

In the interpretation of the results, it should be noted that although the Cox regression analysis did not find a relationship between the occurrence of macrovascular events and other major CV risk factors such as hypertension or hyperlipidemia, the baseline control of these CV risk factors was not included in the model.

There are additional limitations that must be considered when interpreting the results of this study. Due to the retrospective design of our study, only associations but not cause and effect relationships could be explored. The IQVIA medical records database covered a population of 1.2 million patients at the time of the study, but this was based on voluntary participation of the treating physicians and may not be fully representative of the population with diabetes. However, the characteristics of patients in our study are similar to those reported in other studies with Spanish cohorts [30]. As with any database study, there was heterogeneity of data quality and the frequency of data capture and coverage for some key study-related parameters. Missing data were present for certain variables (e.g., BMI). Also, prescriptions and diagnosis were only collected in the database since 2008. As a consequence, a diagnosis date was not available in the database for patients diagnosed before 2008. Therefore, the effect of diabetes duration on CV outcomes in our study was evaluated as a dichotomous variable: newly diagnosed diabetes during the inclusion period vs. known T2D diagnosis before the index date. Consequently, time since diagnosis could not be calculated as this approach resulted in an underestimation of the time from diagnosis in this subpopulation. Further, it was not possible to identify repeated events under the same ICD-9 code. Analyses were conducted in patients with CVD “other than MI or stroke,” as applicable, which may have resulted in the underestimation of the incidence of MI or stroke in this subpopulation. Another database limitation was that death data were not available, as these patients were categorized as “lost to follow-up due to any reason,” and therefore evaluating the rate of cardiovascular or non-cardiovascular mortality in this population was not possible. It should also be considered that, at the index date when baseline data were collected, the cohort consisted of patients with heterogeneous stages of the disease and that HbA1c, or the presence of certain comorbidities, could change over the follow-up period. The results apply to the entire population with T2D, in which most patients were not newly diagnosed. Although it would be interesting to know the effect of stringent HbA1c control specifically in a population of patients with a recent diagnosis, only 29.2% of our sample was newly diagnosed. Being a small sample, the number of events would probably be insufficient to be able to perform this analysis. However, the purpose of this study was to reflect the regular patient population in routine practice at any point in time, providing the average risk of a real-life population of people with T2D.