Sodium-glucose co-transporter-2 (SGLT2) inhibitors were firstly authorized for the treatment of type 2 diabetes mellitus, but subsequently dapagliflozin and empagliflozin also for the treatment of heart failure with reduced ejection fraction based on their favorable cardiovascular outcomes (Mascolo et al., 2022b). Indeed, SGLT2 inhibitors have also shown pleiotropic effects able to determine cardiovascular and renal protection besides the hypoglycemic effect (Mascolo et al., 2021; Mascolo et al., 2022a; Mascolo et al., 2022b).

Despite their clinical benefits, these drugs have also been associated with safety concerns in different cardiovascular trials, some of which have already been implemented in the summary of product’s information by Regulatory Agencies (Mascolo et al., 2022b). Among safety warnings, the risk of fractures and falls, which can sometime be related with each other, has been characterized by conflicting evidence (Ye et al., 2018). Specifically, canagliflozin was firstly associated with an increased risk of bone fractures in the Canagliflozin Cardiovascular Assessment Study (CANVAS) (Neal et al., 2017). However, this risk was not confirmed in other large randomized clinical trials (RCTs) of canagliflozin or other SGLT2 inhibitors (Zinman et al., 2015; McMurray et al., 2019; Perkovic et al., 2019; Wiviott et al., 2019; Heerspink et al., 2020). Moreover, no difference in the risk of fractures between SGLT2 inhibitors and dipeptidyl peptidase-4 (DPP4) inhibitors or glucagon-like peptide 1 (GLP-1) receptor agonists was found in a population-based cohort study conducted on US data related to older adults (Zhuo et al., 2021). Another observational, retrospective, pharmacovigilance study conducted on data from the US Food and Drug Administration Adverse Event Reporting System (FAERS) showed no association between SGLT2 inhibitors alone or combined with other glucose-lowering medications and the risk of fractures (Zhao et al., 2021). The risk of falls with SGLT2 inhibitors has instead been less investigated. The US cohort study found a lower risk of falls when SGLT2 inhibitors were compared with DPP4 inhibitors, while no difference was observed when compared with GLP-1 receptor agonists (Zhuo et al., 2021). Another pharmacovigilance safety study conducted only on FAERS data also found no difference for fall when SGLT-2 inhibitors were compered to other non-insulin antidiabetic agents (Goldman et al., 2023). Generally, the evidence suggested that the increased rates of neuropathy and orthostatic hypotension observed with SGLT2 inhibitors can facilitate the onset of accidental falls and then fractures (Kohan et al., 2014). However, a prospective multicenter study found that SGLT2 inhibitors might also significantly modulate the cardiac autonomic neuropathy dysfunction in type 2 diabetes mellitus patients with vasovagal syncope leading to a reduction of vasovagal syncope recurrence in these patients (Sardu et al., 2022). For DPP4 inhibitors, a study found a higher reporting of events related to the gastrointestinal tract, pancreas, malignancies, infection, musculoskeletal disorders, general disorders, hypersensitivity, and skin reactions, but no safety signal was observed for fractures events (Huang et al., 2020). Moreover, the SAVOR-TIMI 53 trial did not find any difference between the DPP4 inhibitor saxagliptin and placebo (Mosenzon et al., 2015), while a meta-analysis found a reduced risk with DPP4 inhibitors (Monami et al., 2011). Considering the conflicting and meagre evidence available, especially for the risk of falls, and the lack of European pharmacovigilance studies, we decided to conduct a Real-World safety study to compare Individual Case Safety Reports (ICSRs) of falls and bone fractures between SGLT2 inhibitors and DPP4 inhibitors by using the European Pharmacovigilance database (EudraVigilance, EV).

During the study period, a total of 507 ICSRs reported at least one event of fall or fractures with SGLT2 inhibitors. The most reported SGLT2 inhibitor was canagliflozin (N = 188; 36.9%), followed by empagliflozin (N = 176; 34.5%), and dapagliflozin (N = 143; 28.0%). Only 1 (0.2%) ICSR reported the combination of two SGLT2 inhibitors (dapagliflozin and empagliflozin) and it was excluded from the analyses due to the low number. Most ICSRs were related to patients aged 65–85 years (N = 223; 44.0%) and mainly reported in females (N = 291; 57.4%). Most ICSRs were reported by HCPs (N = 319; 62.9%), and came from the non-EEA (N = 447; 88.2%). The SGLT2 inhibitor was the only suspected drug reported (N = 385; 75.9%) and no concomitant drug (N = 226; 44.6%) was reported in most ICSRs. A total of 362 ICSRs reported at least one event of fall and fractures, with sitagliptin as the most reported DPP4 inhibitors (N = 211; 58.3%), followed by linagliptin (N = 74; 20.4%), vildagliptin (N = 46; 12.7%), saxagliptin (N = 17; 4.7%), and alogliptin (N = 12; 3.3%). Only two ICSRs (0.5%) reported both sitagliptin and saxagliptin as suspected drugs. Characteristics of ICSRs for each SGLT2 inhibitor and for DPP4 inhibitors were reported in Table 1. A flowchart of the data selection process was reported in Figure 1. Therapeutic indications were shown in Table 1.

Despite limitation of our analyses performed on pool spontaneous reports of adverse events, in this real-world safety pharmacovigilance study, we found a 34% lower reporting probability of fall with SGLT-2 inhibitors compared with DPP4 inhibitors. Specifically, a 44% lower reporting probability of fall was found for canagliflozin, followed by dapagliflozin (33%), and empagliflozin (23%) in spite of the different year of marketing authorization. Indeed, dapagliflozin was the first authorized SGLT-2 inhibitor (Year 2012) in Europe, followed by canagliflozin (Year 2013) and empagliflozin (Year 2014). Although European consumption data on SGLT2 inhibitors are lacking, canagliflozin seems the most prescribed SGLT-2 inhibitor followed by empagliflozin and dapagliflozin (Sangha et al., 2021).

In Europe, falls are related to at least 3.8 million/year of visits in the emergency departments for fall-related injuries in older adults, of which 1.4 million generally require hospitalization (Promotion EAfIPaS, 2022). Falls negatively affect functionality and quality of life especially in elderly patients (Coles et al., 2020). Indeed, we found a higher reporting probability of falls for the age group ≥65 years than <65 years in SGLT2 inhibitors cases. It may also be related to the high prevalence of diabetes mellitus in older adults. It was estimated that 30% of people aged more than 65 years is affected by diabetes mellitus, with 90% of them having the type 2 (Custódio et al., 2020). However, when comparing the reporting probability of falls between SGLT-2 inhibitors and DPP4 inhibitors in cases aged ≥65 years, no difference was observed. This could be related to the notoriously higher frailty of older patients affected by diabetes mellitus and other cardiovascular comorbidities (Ungar et al., 2021; Wiersinga et al., 2021). Generally, patients treated with SGLT-2 inhibitors or DPP4 inhibitors are vulnerable and characterized by many predisposing risk factors for falls besides the anti-diabetic medication, including the presence of cardiovascular comorbidities that often require treatments with hypotensive agents. Hypotension is associated with a high risk of falls (Jansen et al., 2016) due to the acute transient hypoperfusion in some cerebral area that can cause neurodegenerative lesions and gait/balance problems (van Poelgeest et al., 2023). Hypotension due to volume depletion is an uncommon (1/1.000 ≤ frequency <1/100) event with SGLT2 inhibitors (European Medicines Agency EMA, 2023a; European Medicines Agency EMA, 2023b) and seems more frequent in patients co-treated with diuretics (van Poelgeest et al., 2023). Moreover, the increased diuresis than can occur with both medicines can evolve to urinary incontinence (Strain et al., 2021), which is another independent risk factor for falls (Deandrea et al., 2010) that can negatively impact quality of life in elders (Wehling, 2013; Benetos et al., 2019). We found that about half of falls (N = 116/336; 52.68%) happened in patients treated with anti-hypertensive drugs and that the association with falls was lower in the absence of risk factors. Based on the potential to cause or worsen the risk of fall, the prescription of SGLT2 inhibitors with diuretics or other anti-hypertensive drugs should be considered with caution at the lowest dosage possible, especially in elderly patients, and volume status, blood pressure, and electrolytes should be monitored during follow-up visits (van Poelgeest et al., 2023). Concerns were also raised for the potential ability of SGLT2 inhibitors to harm bone metabolism through the modulation of calcium and phosphate homeostasis (Blau and Taylor, 2018). Bone fragility can cause fractures and then falls. On the contrary, DPP4 inhibitors were hypothesized to have protective effects on bone by promoting osteoblast differentiation and inhibiting osteoclast activity (Kalaitzoglou et al., 2019; Hygum et al., 2020).

Although there are biologically plausible mechanisms underlying the onset of these adverse events with SGLT-2 inhibitors, we found a lower reporting probability of falls and no effect on the reporting probability of fractures (except for the slight increase with canagliflozin not confirmed in the sensitivity analysis) when these drugs were compared with DPP4 inhibitors. Clinical studies on this outcomes are inconsistent. The CANVAS trial found a higher rate of fractures (26%) with canagliflozin than placebo (Neal et al., 2015) (Hazard Ratio, HR, 1.26; 95%CI, 1.04-1.52). On the contrary, the Canagliflozin and Renal Outcomes in Type 2 Diabetes and Nephropathy trial and other large RCTs, as well as meta-analyses, did not find an increased risk of fractures (B et al., 2015; Ruanpeng et al., 2017; McMurray et al., 2019; Perkovic et al., 2019; Wiviott et al., 2019; Heerspink et al., 2020; Lou et al., 2020). Moreover, in accordance with our results, observational studies did not show an increased risk of fractures with SGLT2 inhibitors compared with DPP4 inhibitors (HR, 1.11; 95%CI, 0.96-1.28) (Adimadhyam et al., 2019) or GLP-1 receptor agonists (HR, 0.98; 95%CI, 0.75-1.26; HR, 1.11; 95%CI, 0.93-1.33) in relatively young (mean: 55–61 years) (Ueda et al., 2018; Fralick et al., 2019) or older patients (compared with DPP4 inhibitors: HR, 0.90; 95% CI, 0.73-1.11; or compared with GLP-1 receptor agonists: HR, 1.00; 95%CI, 0.80-1.25) (Zhuo et al., 2021). Finally, a US pharmacovigilance study found no signal for the reporting of fractures with canagliflozin (ROR: 0.49; 95%CI 0.42-0.57), dapagliflozin (ROR, 0.54; 95%CI, 0.43–0.67), and empagliflozin (ROR, 0.42; 95%CI 0.33-0.52) (Zhao et al., 2021). For the risk of fall, a pharmacovigilance studies conducted on the US FAERS found no difference in the reporting when SGLT-2 inhibitors were compered to other non-insulin antidiabetic drugs (ROR in adults, 1.16; 95%CI, 0.98-1.37 and ROR in older adults, 1.19; 95%CI, 0.89-1.59) (Goldman et al., 2023). According to our findings, instead, an observational study found a lower risk of falls with SGLT-2 inhibitors than DPP4 inhibitors (HR, 0.82; 95%CI, 0.77-0.87) (Zhuo et al., 2021). However, as also stated by study’s Authors, DPP4 inhibitors users were older and frailer than SGLT-2 inhibitors users before matching, thus suggesting that even after adjustment for many measured factors (such as frailty status and age) residual confounding factors may be present such as the presence of mild cognitive impairment (Zhuo et al., 2021). Therefore, in our study, we cannot exclude a higher frailty for cases with DPP4 inhibitors since the 67% of falls occurred in presence of other predisposing factors compared to the 53% with SGLT-2 inhibitors. Moreover, the adjustment in concomitant medications that can occur at the beginning of treatment with SGLT-2 inhibitors (such as deprescribing diuretics or insulin) (Li et al., 2019; van Poelgeest et al., 2023) might also justify the decreased risk of falls. Further studies may specifically clarify whether the initiation of SGLT-2 inhibitors is associated with deprescribing. Furthermore, falls and fractures are typically caused by the interaction of multiple confounders, such comorbidities (musculoskeletal, cardiovascular, metabolic, degenerative disorders, etc.) and drug therapies (cardiovascular, psychotropic medications, etc.) (Berry and Miller, 2008), which are difficult to measure with spontaneous reporting data.

In the analysis for gender, we found a higher reporting probability with female than male in accordance with the CANVAS and an observational study that found a high rate of fractures in female but not associated with a statistically significant increased risk (Watts et al., 2016; Zhuo et al., 2021). Furthermore, we need to consider in spite of the higher initiation of SGLT-2 inhibitors in male that female patients are more prone to develop adverse events with SGLT-2 inhibitors (Di Mauro et al., 2022), but also in general independently from the drug class due to hormone-related pharmacokinetics changes.

The main strength is the use of the large EV database that allows to detect a wide range of real-life safety cases related to the treatment with SGLT2 inhibitors. The use of pharmacovigilance data is indeed not expensive and helpful to better characterize the drug safety profile. However, several limits need to be identified for the analysis of EV data. Firstly, safety reports may not be reported to national competent authorities and therefore in EV (underreporting), thus implying that not all safety cases that occurred in real-life are contained in EV. There are several reasons for failing to recognize an adverse events as linked to a drug. The common reasons for HCPs are the feeling of guilt, ignorance, lethargy, inadequate risk perception, diffidence, insufficient training, and lack of awareness about pharmacovigilance (Li et al., 2022). Another limitation is the quality of information reported that can be non-homogeneous and incomplete. Many cases may lack of information on age, drug dosing, time to onset, comorbid conditions, therapeutic indication, and concomitant drugs, thus hindering the identification of risk factors. Indeed, our descriptive analysis for risk factors may underestimate the real frailty among cases since many comorbidities or drug therapies may not be reported due to lack of time or laziness in completing the report form. This is a vary important limitation considering that falls and fractures can be caused by the interaction of multiple confounders. Moreover, another limitation is the impossibility to obtain data regarding the entirety of ICSRs reported in the EV database, thus impeding a proper evaluation of safety signals. Indeed, in contrast to other pharmacovigilance databases, we cannot retrieve all ICSRs on falls and bone fractures from the online EV (www.adrreports.eu), making impossible the computation of disproportionality analyses on all ICSRs. Furthermore, the exact denominator of patients exposed to SGLT2 inhibitors cannot be evaluated in EV, not allowing any estimate in terms of incidence. The total number of events for a drug can be used instead as a denominator for disproportionality analyses. Finally, in the absence of consumption and prescription data, we cannot exclude that differences in the reporting were related to a differential use.

In this European safety study, SGLT-2 inhibitors were associated with a lower reporting probability of falls compared with DPP4 inhibitors. Results were consistent across sex and age categories. Our results provide real-world evidence on the safety profile of SGLT-2 inhibitors. Generally, prior to initiate any treatment, clinicians must consider predisposing risk factors for potential adverse events before developing or adjusting a treatment pharmacological plan that includes SGLT2 inhibitors. Future researches and post-marketing analyses involving these drugs will help to understand their long-term safety.