FormalPara Key Summary Points

Why carry out this study?

The REWIND cardiovascular outcomes trial demonstrated the benefit of the glucagon-like peptide-1 receptor agonist dulaglutide in reducing major adverse cardiovascular (CV) events in patients with type 2 diabetes (T2D) aged ≥ 50 years with established or at risk of CV disease.

This study uses real-world clinical data to describe the characteristics of patients with T2D in routine primary care in France based upon key inclusion criteria of the REWIND study to assess the proportion of T2D patients in France eligible for REWIND according to their CV profile.

What was learned from the study?

Electronic medical records from the IQVIA database were used to access data from patients with T2D in primary care in France between 2009 and 2019.

Patient characteristics, such as age, glycated hemoglobin and body mass index were comparable between the study population and REWIND participants.

Approximately 60% of the study population fulfilled key REWIND CV inclusion criteria; 12.4% of patients were aged  ≥ 50 years with established CV disease, 9.7% were aged ≥ 55 years with subclinical vascular disease and 44.7% of patients were aged ≥ 60 years with ≥ 2 CV risk factors.

Introduction

Cardiovascular disease (CVD) persists as a major comorbidity for patients with type 2 diabetes (T2D) [1]. Several glucagon-like peptide-1 receptor agonists (GLP-1 RA) have demonstrated a beneficial cardiovascular (CV) effect in patients with T2D and established CVD or cardiovascular (CV) risk factors [2]. However, GLP-1 RA cardiovascular outcomes trials (CVOTs) mainly included patients with established CVD [3,4,5,6,7], with the exception of the REWIND study which also included a substantial proportion of patients with no history of established CVD [2, 8]. The REWIND CVOT demonstrated the benefit of the GLP-1 RA dulaglutide in reducing major adverse CV events in patients with T2D aged ≥ 50 years with or without established CVD [9]. Key inclusion criteria were (1) a baseline glycated hemoglobin (HbA1c) ≤ 9.5%; (2) receiving up to two oral glucose-lowering agents (GLA) with or without basal insulin; (3) and aged ≥ 50 years with established CVD, aged ≥ 55 years with subclinical CVD or aged ≥ 60 years with ≥ 2 CV risk factors [10]. During the median follow-up of 5.4 years, dulaglutide treatment demonstrated a significantly lower risk of the primary composite endpoint of non-fatal myocardial infarction, non-fatal stroke or CV death compared to placebo (hazard ratio 0.88; 95% confidence interval 0.79–0.99) [9, 11]. Despite the considerable impact of CVOTs, limited data are available on the generalizability of results due to restrictive inclusion and exclusion criteria in all CVOTs conducted with antidiabetic drugs. A USA-based study examined the extent to which GLP-1 RA CVOTs were generalizable to the overall T2D population [8]. The results suggested that the characteristics of the patients with T2D included in the REWIND study were the most similar to those in the US adult T2D population than those of other GLP-1 RA CVOTs. In France, the ENTRED public health survey is periodically conducted on a national representative sample of people living with diabetes in order to accurately characterize their health status, the quality of care they received and the evolution trend of these parameters. The most recent assessment from 2007 indicated that a majority of patients with T2D had a very high risk of CVD, but the proportion of patients who would have fulfilled the REWIND CV inclusion criteria is unknown [9, 11, 12].

The aim of the current study was to describe similarities and differences in patient characteristics between a cohort of patients with T2D followed in routine primary care in France and the T2D population included in the REWIND study, and to define the prevalence of established CVD in the French cohort using similar definitions to those used in the REWIND study.

Methods

Database

A retrospective, observational study was conducted in France using IQVIA’s Electronic Medical Records (EMR) database contained in the Longitudinal Patient Data (LPD) IQVIA tool. The LPD tool contains anonymized prescription, patient and medical information from Patient Management software used by primary care physicians (general practitioners [GPs]) and specialists. The database is considered representative of day-to-day clinical practices as it details patient information and physicians’ care without any interventional purpose. The panel of physicians is representative of the primary care physician according to three criteria known to influence prescribing: age, gender and geographical distribution. For this study, only the GP panel was used (N = 1200) as the population of interest were patients with T2D, and these patients are mainly followed by GPs in routine care. Additionally, no unique identifier existed to link patient data from GPs and specialists and, therefore, not including patients treated by specialists prevented a duplication of patient data. Data collected included sociodemographic information, medical information (including T2D diagnosis details, comorbidities, history of CVD) and T2D treatment details. Diagnoses of T2D, CVD comorbidities, and other comorbidities were coded based on the International Classification of Diseases (ICD)-10 system. This study is based on de-identified data collected from a healthcare database. According to the French law no submission to CESREES (Ethics and Scientific Committee for Research, Studies and Evaluations in the Field of Health) was required. This study complied with the reference method MR004 of CNIL (French data privacy agency). As required, the study was recorded for information on the French Health Data Hub website.

Study Design

The study inclusion period extended from 1 January 2019 to 31 December 2019. Patients aged ≥ 18 years with previously or newly diagnosed T2D were identified. Eligible patients had at least one visit with T2D diagnosis and/or at least one prescription of a glucose-lowering agent during the inclusion period. If not diagnosed during the inclusion period, patients were required to have a record of T2D diagnosis within the preceding 10 years (from 1 January 2009). A 24-month pre-index period from 1 January 2017 to 1 January 2019 was included to assess patient and clinical characteristics. A 3-month follow-up window from 31 December 2019 was included to assess laboratory results.

Patient Criteria

Patients included in the study were aged ≥ 18 years at the index date and fulfilled diagnosis or prescription, and medical history requirements detailed in Study Design. Patients were excluded if they had a diagnosis of type 1 diabetes among back-data prior to the index date or had gestational diabetes or were pregnant in the 9 months before or at the index date.

Patients aged ≥ 50 years at the index date were identified as a subgroup of interest due to the REWIND CV inclusion criteria [10]. Among these patients, two distinct groups were categorized following the REWIND enrollment criteria: (1) those aged ≥ 50 years with a history of established CVD and (2) those with high CV risk but without a history of established CVD—either aged ≥ 55 years with subclinical vascular disease or aged ≥ 60 years with multiple CV risk factors. A history of established CVD was defined as prior myocardial infarction, ischemic stroke, presence of coronary revascularization, occlusion of stenosis of precerebral arteries or carotid artery and/or hospitalization for unstable angina. See Electronic Suplementary Material (ESM) Table S1 for inclusion criteria of the REWIND study and of the current study with proxies used in the current analysis.

Comorbidities of interest were identified within the database with all back-data period available (i.e., up to 10 years), except for obesity which was identified within the 24 months prior to the index date. Comorbidities were identified from physician reports in the database. However, to address potential underreporting of the comorbidities hypertension and dyslipidemia, additional analyses were done to identify patients with hypertension and/or prescription of blood pressure-lowering medication, and those with hyperlipidemia and/or prescription of lipid-lowering medications.

Statistical Analysis

Descriptive analysis was used. For the baseline characteristics analysis, the REWIND population was described using published data only [9, 10]. Presented characteristics included age, sex, body mass index (BMI), duration of diabetes, HbA1c and other key laboratory and diagnostic variables.

Proxies for key REWIND CV criteria were applied to the general T2D study population, as described in Study Design. Continuous variables were described by number of values, mean and standard deviation (SD), or median with 1st and 3rd quartile. Categorical variables were described by the number of values, the frequency and the percentage of each modality. All results were presented with one decimal. All analyses were performed using SAS® enterprise guide 6.1 (SAS 9.4) software (SAS Institute, Cary, NC, USA).

Results

Of the T2D primary care population in the EMR database, 87,670 patients with a diagnosis of T2D and/or at least one prescription of a glucose-lowering agent were identified during the inclusion period. Of these, 63,927 (72.9%) satisfied the inclusion and exclusion criteria (Fig. 1) and were classified as the general T2D study population. Most baseline characteristics of the study cohort were comparable with those of the REWIND participants (Table 1). However, some differences were observed between the populations, such as diabetes duration and comorbidities.

Fig. 1
figure 1

Flow chart of patient inclusion in analysis cohort

Full size image
Table 1 Baseline characteristics of total type 2 diabetes study cohort, those aged ≥ 50 years fulfilling key REWIND cardiovascular criteria and REWIND participants
Full size table

The mean (± SD) age of the study cohort was 67 (± 11.7) years, and 59,313 (92.8%) were aged ≥ 50 years, 58.2% were male, mean HbA1c was 7.1% and mean BMI was 30.4 kg/m2 (± 5.8). The most frequent comorbidities reported by the physician were hypertension (63.8% of patients) and hyperlipidemia (31.4%). An additional analysis considering diagnosis and prescriptions has shown that 78.2% of the general T2D population had hypertension and 63.7% had hyperlipidemia. The mean (± SD) age of REWIND participants was 66.2 (± 6.5) years and 53.7% were male; mean HbA1c was 7.3% and mean BMI was 32.3 (± 5.7) kg/m2.

Patient Characteristics Based on Key REWIND CV Criteria

Of the total T2D study population, 37,965 (59.4%) were aged ≥ 50 years and fulfilled the REWIND CV criteria (Table 2). More specifically, 7922 (12.4%) of the total T2D study population were aged ≥ 50 years and had a history of established CVD (Table 2), which represents 20.9% of the T2D patients who fulfilled REWIND CV criteria (7922 of the 37,965). The majority of these patients, 7751 (97.8%), had one or two prior CVDs.

Table 2 Distribution of study cohort patients based on key REWIND CV risk criteria
Full size table

In addition, 30,043 (47%) of the total T2D study population fulfilled REWIND CV criteria without a history of established CVD (Table 2). Specifically, 6205 (9.7%) were aged ≥ 55 years with ≥ 1 subclinical vascular disease and 28,554 (44.7%) were aged ≥ 60 years with ≥ 2 CV risk factors. Patients could belong to both strata. Of the patients aged ≥ 60 years with CV risk factors, 26,927 (94.3%) had two or three CV risk factors.

Baseline characteristics of the population who satisfied REWIND CV criteria, those with established CVD and those without are presented alongside those from the REWIND study in Table 1. In the population fulfilling REWIND CV criteria, age was comparable between patients with or without established CVD, whereas gender and BMI were not, with more males in the population with established CVD and a higher BMI in the population without established CVD. The eGFR was similar regardless of CV profile although a higher proportion of patients with established CVD had an estimated glomerular filtration rate ≤ 60 ml/min/1.73 m2. A slightly higher percentage of patients with CVD had diagnosed albuminuria than those without CVD. HbA1c was similar between the two subgroups and patients with established CVD had lower low-density lipoprotein-cholesterol levels.

Treatment Patterns of Patients With and Without Established CVD Meeting REWIND CV Criteria

In the population fulfilling key REWIND CV criteria, 7590 (95.8%) and 28,906 (96.2%) of the patients with and without established CVD, respectively, received at least one GLA. In both groups, over 50% of patients received only one GLA (Fig. 2a). Metformin was the most prescribed GLA, while 18.8% of the patients with established CVD and 14.3% of the patients without established CVD received insulin (Fig. 2b). The most common treatment regimen was metformin alone, with 34.5% and 36.1% of the populations with and without established CVD receiving metformin alone, respectively (ESM Fig. 1). GLP-1 RAs were prescribed to 6% of patients, regardless of their CVD status (Fig. 2b).

Fig. 2
figure 2

Antidiabetic treatment patterns of patients who satisfied key REWIND CV criteria (N = 37,965) receiving at least one antidiabetic treatment. a The percentage of patients receiving 1, 2, 3, or ≥ 4 antidiabetic treatments in the total population, the population with established CVD and the population without established CVD. b Types of oral or injectable antidiabetic treatments that the total population, and patients with and without established CVD received. AGIs Alpha glucosidase inhibitors, CVD cardiovascular disease, DPP-4  dipeptidylpeptidase-4, GLP-1 glucagon like peptide-1, SUs sulfonylureas

Full size image

Among the patients who fulfilled key REWIND criteria, 91.1% and 87.3% of patients with and without established CVD, respectively, received blood pressure medications. Statin use was also prevalent among the populations, with 70.5% and 62.6% of patients with and without established CVD, respectively, prescribed a statin. Fibrate use was low, with 2.4% and 6.8% of patients with and without established CVD, respectively, receiving fibrate medications (ESM Fig. 2).

Discussion

This study described patients with T2D followed by GPs in routine clinical practice in France and put them in the perspective of patients included in the REWIND study cohort, applying key REWIND inclusion criteria to examine patients’ CV risk profiles. While many baseline characteristics were comparable, others, including duration of diabetes, the presence of hypertension and albuminuria, differed between the two cohorts. After applying REWIND CV inclusion criteria, some baseline characteristics of the study cohort notably changed, such as mean duration of diabetes, the percentage of patients with hypertension and eGFR < 60 ml/min/1.73 m2. The percentage of patients with albuminuria was low compared to the REWIND cohort (2.4% of the total T2D population fulfilling key REWIND criteria versus 35% of the REWIND study cohort), but the prevalence presented here may be biased as albuminuria: (1) was recorded as a categorical variable within the LPD database, which did not allow differentiating the effective absence of albuminuria and missing values; (2) may have been underreported by physicians in the database; and (3) is not often assessed in routine practice. Indeed, the ENTRED 2007 report found that only 28% of patients had a dosage of albuminuria or proteinuria within the previous year, and albuminuria status was not reported for 39% of patients, which does not allow us to assess the prevalence of albuminuria in French patients with T2D [11, 12]. In addition, in an effort to encourage the monitoring of renal function in patients with T2D and improve quality of care, the National Health Insurance Fund (Caisse Nationale d’Assurance Maladie) has recently included an annual albuminuria threshold as a quality indicator within the French primary care pay-for-performance program [13].

Previous studies comparing the eligibility of patients with T2D in routine clinical care with major phase 3 randomized controlled trials (including GLP-1 RA and sodium-glucose co-transporter-2 (SGLT2) inhibitor CVOTs) have shown limited eligibility of general populations based on trial inclusion–exclusion criteria and differences between baseline characteristics [14,15,16,17,18]. The proportion of patients in general populations meeting CVOT eligibility ranged from 16.6% of an English general T2D population for the LEADER CVOT to 59% of a pooled German, Dutch, Swedish and Norwegian general T2D population for the DECLARE-TIMI 58 CVOT [16, 18]. Similarly, a Scottish study that examined the proportion of patients with T2D in clinical care who met the eligibility criteria of seven major T2D randomized controlled trials which have been used to form diabetes treatment guidelines for glycemic control (ACCORD, ADVANCE, PROactive, RECORD, VADT, UKPDS 33 and UKPDS 34) found that a maximum of 51% of newly diagnosed patients were eligible for these trials, with age and renal impairment being two of the most common reasons for exclusion [15]. Additionally, heterogeneity between cohorts from different countries has been published [14, 18], necessitating studies to understand how different CVOT criteria apply to individual countries. While the baseline characteristics of the REWIND CVOT were suggested to be the most generalizable to patients with T2D in routine clinical practice in the USA from four major GLP-1 RA CVOTs [8], it was of interest to analyze the eligibility of the French patients to the REWIND study as participants were neither recruited to the REWIND trial nor studied in France [9]. The current study showed that a large proportion of the French study population were eligible for REWIND, with 59.4% of the general study population fulfilling REWIND CV criteria, including 12.4% with established CVD. The lower percentage of patients with established CVD than in the REWIND study (31.4%) [10] was expected as patients enrolled in CVOTs are usually highly selected for very high CV risk.

The prevalence of CVD in patients with T2D in routine clinical care varies depending on the study design and the region [19]. Among European studies [18, 20,21,22], the prevalence of CVD has been reported to range from 6.9 [22] to 44% [18], although different patient inclusion criteria and definitions of CVD may contribute to these differences. Among the French population in 2007, 26% of patients with T2D had a high risk of CVD and 16.7% of patients reported angina or myocardial infarction while 13.9% reported coronary revascularization [11]. The CAPTURE study, conducted over 10 months between 2018 and 2019, analyzed 659 French patients with T2D within a total cohort of 9823 patients across 13 countries; the results indicated a 34.2% prevalence of CVD among French patients, similar to the overall cohort [19]. This result is much higher than the current analysis in which 12.4% of patients had a history of CVD; however, there is a large difference in sample sizes between the studies, and the CAPTURE study population was recruited almost equally from primary care and specialists settings [11]. In another study conducted in 2011 on the National Health Insurance Cross-Schemes Information System (SNIIRAM) data, 26% of patients with diabetes in France received care for a cardio-neurovascular disease (coronary disease, ischemic stroke, heart failure, lower extremity arterial disease, cardiac arrythmia, valvular disease, acute pulmonary embolism or another CV condition) [23]. The same study showed that the average age of the French population with diabetes was 65 years and that 48% were women. While this slightly contrasts the current analysis in which the average age of patients was 67 years and 42% were women, the differences may be explained by the gap in study dates and the fact that the SNIIRAM study did not differentiate between patients with type 1 diabetes and T2D. Finally, the 2007 ENTRED study reported that 41% of patients were obese [11], compared to 47.6% in the current study population and 61.2% in REWIND study. Nonetheless, the current analysis is largely similar to reported results for French patients in 2007 and 2011 [19].

The majority of patients in this study received metformin (75.9% and 79.2% of patients with and without established CVD, respectively), which was comparable to the REWIND study in which 81% of patients received metformin at baseline, and also in accordance with French guidelines for the management of hyperglycemia in T2D [24]. The percentages of patients receiving insulin were somewhat lower in the current study than in the REWIND study, with 18.8% and 14.3% of the populations with or without established CVD, respectively, and approximately 24% of REWIND participants receiving insulin at baseline. Dipeptidylpeptidase-4 (DPP-4) inhibitors were used by > 30% of patients fulfilling key REWIND criteria. In REWIND, patients were instructed to remain on their antihyperglycemic therapy, with the exception of patients taking a DPP-4 inhibitors or GLP-1 RAs at screening, who discontinued these therapies at the start of the run-in period [10]. Glinide usage was also higher in the current study, while sulfonylurea usage was higher in the REWIND study, and α-glucosidase inhibitor usage was similar between both. The differences in treatment patterns between the current and the REWIND studies could be due to geographical differences in patient characteristics, treatment guidelines and sources of participants recruitment (primary vs. tertiary care).

The pattern of glucose-lowering medication usage was comparable between subgroups of patients with and without established CVD, as defined by key REWIND criteria, which was consistent with the treatment pattern observed in the CAPTURE study. Despite the prevalence of CVD and CV risk factors among the general T2D study population, 6% of patients fulfilling REWIND CV criteria were prescribed GLP-1 RAs, regardless of their CV profile and frequently in addition to metformin alone or metformin and sulfonylureas. This percentage is lower than that described by the CAPTURE study which reported that 8.6% of participants worldwide and 24.7% of French participants were prescribed a GLP-1 RA with a similar usage across patients with and without CVD [19, 25]. This may be explained by the fact that of the 659 French participants in the CAPTURE study, 361 (54.8%) were followed by a GP and 298 by a specialist (45.2%) [19]; and GLP-1 RAs are more frequently prescribed by specialists than GPs (41.9 vs. 8.6%) [25]. Given that the current study only analyzed patients under GP care, this may account for the difference with CAPTURE data in GLP-1 RA prescriptions. However the absence of a difference in prescription according to patient’s CV profile was observed in both studies [26,27,28,29].

This study has several strengths. It used a large and robust cohort that is representative of the general French patient population. The data source collected information from a panel of 1200 GPs in a non-interventional way using pseudonymized, patient-level information that reflects routine clinical practice. The panel of contributing GPs was maintained as a representative sample of primary care physicians according to three criteria known to influence prescriptions: age, gender and geographical distribution. Whenever a physician left the panel, they were replaced by another one with a similar profile. Indeed, public health data have previously highlighted that 82% of patients with T2D are routinely followed-up by a GP only and that only 14% had an appointment with a private specialist physician throughout a 2-year period [12].

The study also has several limitations. There were a high number of missing values (as indicated in Table 1), and it was impossible to query the physicians for any missing information as no individual patient identifiers were available. The analyzed data did not capture data from patients managed exclusively by specialists. To apply REWIND CV criteria, some information was approximated with proxies using data available in the IQVIA LPD database for France. Certain study variables were underreported in the EMRs, including weight, laboratory values and smoking status (see Table 1; smoker status unknown, 98.7%). In some situations, proxies used a combination of several indicators and may have caused an overestimation. Due to Commission Nationale de l'Informatique et des Libertés (CNIL) legal requirements, only the previous 10 years could be studied, which may have affected some results, especially T2D duration. Within secondary outcomes, CV risk groups were not exclusive, meaning patients could belong to several groups, which leads to an overestimation of the proportion of patients in each stratum. In patients aged ≥ 55 years with subclinical vascular disease, documented myocardial ischemia by stress tests or imaging and ≥ 50% coronary carotid or lower extremity artery stenosis were included in the REWIND criteria; however, this information was not available in the current study, so some patients may have been incorrectly excluded. The absence of French patients in REWIND limits the transposability of the results to French patients with T2D but highlights the relevance of our study.

Conclusions

In conclusion, this observational study of 63,927 French patients with T2D showed that the mean age, HbA1c and BMI observed were comparable to the REWIND population. Almost 60% of the general study population our this observational study fulfilled REWIND criteria. A lower percentage of patients aged ≥ 50 years had established CVD in this cohort than that in the REWIND study. Despite their benefits for lowering CVD risk, also highlighted in international T2D management guidelines, GLP-1 RAs were infrequently prescribed, regardless of the participants’ established CVD status. Future studies should investigate if recent re-assessments of GLP-1 RAs impact the prescription of these drugs in T2D patients with and without a history of CVD.