In this real-world cohort study of people with type 2 diabetes who initiated treatment with metformin, the use of FDCs was associated with a modestly lower rate of heart failure than the LDC regimen was, and 47% of this was mediated by medication adherence. This association was also observed in people aged 65 years and older, but was attenuated with further matching for diabetes duration or when drugs were matched at ATC code level instead of drug classes. No associations were observed for myocardial infarction, atrial fibrillation, unstable angina, stroke, or kidney impairment.
The observed cardiovascular protection of FDC therapy on heart failure could be partly explained by differences in medication adherence, since poor adherence could lead to an increased risk of CVD events [29]. The FDC initiators in this study had higher adherence. FDCs reduce pill burden and simplify treatment, which may improve patients’ consistency in taking medications as prescribed.
The larger association of FDCs with heart failure in people aged 65 years and older, could be explained by age-related alterations in pharmacokinetics and pharmacodynamics [30]. Older adults more frequently experience declined kidney function, leading to reduced drug clearance and prolonged plasma drug concentrations. Additionally, the higher prevalence of multimorbidity and polypharmacy in older adults may further influence drug absorption and excretion, potentially leading to increased drug exposure when taking FDCs.
Analysis stratified on drug class suggested that FDCs of metformin with DPP4i may be associated with a lower rate of heart failure. One trial has shown metformin and sitagliptin combination associated with improved glucose control compared to metformin monotherapy, which may indirectly reduce the risk of complications [31]. However, another study reported that the addition of saxagliptin to existing antidiabetic therapy was associated with an increased risk of hospitalization for heart failure (HR = 1.27; 95% CI 1.07, 1.51) [32]. Notably, that study did not include a direct comparison with metformin monotherapy, although metformin accounted for approximately 70% of concomitant antidiabetic use throughout the trial. Another important concern is around the 40% concomitant use of sulfonylureas, as sulfonylureas have been associated with elevated cardiovascular risk. Finally, saxagliptin in that study was not administered as an FDC. Therefore, it was an LDC-versus-LDC study. Consequently, the study provides only limited direct inference for our findings.
In analyses stratified on drug class, the observed differences may be attributable to factors inherent to the FDC itself. FDCs are designed to control drug release kinetics and enhance the physical and chemical stability of each component within the gastrointestinal environment. FDCs often incorporate specialized excipients or matrix systems designed to prevent the degradation of labile compounds in the presence of others, a limitation seen in LDCs [33]. This approach is useful, as several antidiabetic agents in this study, such as empagliflozin and saxagliptin, are labile under hydrolytic or oxidative stress conditions [34, 35]. Additionally, co-granulation, modified release layers, or enteric coatings can yield measurable differences in bioavailability and therapeutic performance. However, since our study did not include pharmacokinetic or bioavailability data, formulation-related mechanisms should be considered speculative in explaining our findings. Finally, in additional analyses in populations with stratification by drug classes pre-matching, no associations were observed.
There were no differences in acute myocardial infarction, ischaemic stroke, unstable angina, or atrial fibrillation between the FDC and LDC treatment groups. To date, only a few SGLT2i (e.g., canagliflozin) have been recommended by regulatory agencies to reduce the risk of major adverse cardiovascular events in people with known atherosclerotic CVD, but not in people at risk of those diseases [36]. In terms of cardiovascular events, the major protective effect of SGLT2i observed is on heart failure [36]. However, published research has not shown evidence of beneficial cardiovascular effects of DPP4i [37].
As yet, SGLT2i has shown kidney-protective effects, but there is insufficient evidence supporting the use of DPP4i [37]. This study has observed no associations between FDC and the kidney outcome (eGFRcr < 30 ml/min/1.73m2). This may be due to the suboptimal capture of kidney events, the short follow-up duration (median: 3.8 years), and the generally healthy status of the kidneys at baseline in both groups, resulting in low power. Because only 185 renal events accrued, statistical power was limited. Using Schoenfeld’s approximation (two-sided α = 0.05, 1:1 allocation), with 185 events the minimum detectable effect at 80% power was around HR = 0.66. Thus, the null finding for the renal outcome likely reflects limited power and not necessarily absence of effect. Further studies on people with impaired kidney function and longer follow-up periods are encouraged.
Despite stratification and matching, baseline cardiovascular disease differed across drug classes, especially among SGLT2i users, indicating prescribing bias by cardiovascular risk. This may cause confounding by indications, potentially masking the true benefits of FDC in high-risk patients, which could explain part of the attenuation in associations observed in the analyses stratified by drug class and with matching at ATC level.
This study contributes to the clinical management of type 2 diabetes by examining the role of FDCs in combination therapy, thereby clarifying a previously ambiguous area within current guidelines. To our knowledge, this is the first study that has specifically examined cardiorenal endpoints for FDC versus LDC therapy in type 2 diabetes using real-world data.
This study has several strengths, such as the large, real-world cohort drawn from routine clinical practice, which enhances the generalizability of the findings to broader populations. Additionally, the use of an active comparator provides a rigorous reference group, and a wide range of baseline characteristics and comorbidities were adjusted for to minimize confounding.
Despite these strengths, several limitations must be acknowledged. First, the exposure definition relied on prescription fill records, which do not provide information on the actual duration and doses. Second, there were imbalances in drug classes after matching, which could have biased the results of the main analysis. However, stratified analysis on each drug class showed possible negative associations with heart failure in DPP4i FDC users. Third, the sensitivity analysis using matching at ATC code level showed similar trends as those of the primary findings but an attenuation in associations. While matching at the ATC code level may offer a greater control over potential confounding by indication, matching at the drug class level is more aligned with the original aim of the study to compare the form of combination therapies (FDC versus LDC), independently from individual drug effect. Additionally, current clinical guidelines (1) recommend combination therapies based on drug classes (e.g., metformin in combination with SGLT2i) rather than specific drugs (e.g., dapagliflozin). Therefore, drug class-level matching enhances the clinical generalizability of our findings. Fourth, the proportion of days covered is a proxy for adherence, and this is only an approximation of true medication-taking behaviour [38]. Fifth, the follow-up time, although sufficient to observe differences in heart failure, may not have been long enough to detect divergences in outcomes such as chronic kidney disease progression or hard atherosclerotic events, which develop over many years. Sixth, the intention-to-treat approach did not reflect the drug-switching during the follow-up duration. In the matched population, 25.6% of FDC users switched to LDC or non-metformin monotherapy, compared to 8.5% of LDC users. There were also users who switched to a different treatment group and later returned to their original treatment. This imbalance in switching may dilute the differences between FDC and LDC therapies, potentially underestimating the true comparative effectiveness of FDC. However, the mediation analysis has partially offset this disadvantage, showing that being adherent is an important mediator. Seventh, GLP-1 receptor agonists were not included in this study because no FDCs containing GLP-1 RAs were available on the market at the time of data collection. Future research on GLP-1 RAs is required. Eighth, potential informative censoring like death and emigration may bias the results. However, the direction of the effect is unknown. All-cause mortality during follow-up was 8.0% in FDC users and 8.6% in LDC users. The emigration rate was 0.94% and 0.68% for FDC and LDC users respectively. Given these small and similar rates, any differential informative censoring is expected to be modest and unlikely to explain the results. Finally, this study was conducted within a specific healthcare system. Differences in FDC availability, prescribing practices, and government subsidies may limit the generalizability of the findings to other countries.
These findings suggest that the use of FDCs is associated with a lower rate of heart failure in people with type 2 diabetes, but not with a reduced risk of other cardiovascular outcomes. The association is partially mediated by improved adherence. This strategy could be particularly relevant for older adults or those facing polypharmacy. In health care systems with low FDC uptake, FDCs may serve as a practical tool to simplify treatment, improve adherence, and reduce heart-failure burden. Priority should be given to clinically equivalent FDCs for adults with type 2 diabetes who require dual oral therapy, especially older patients and those on polypharmacy, with parity (or better) reimbursement to reduce access barriers. Due to the neutral results for non-heart failure outcomes, policy implementation should proceed via evidence-guided pilots with prospective monitoring. While the results highlight a potential clinical advantage of FDC use, especially in routine clinical care settings, further studies are encouraged to focus on pragmatic trial designs, long-term kidney outcomes, and high-risk groups such as patients with impaired kidney function.