Prescription patterns of SGLT2 inhibitors in patients admitted to a tertiary health service with heart failure with reduced ejection fraction: an opportunity to improve

Abstract

Background

Sodium‐glucose co‐transporter 2 inhibitors (SGLT2is) improve symptoms, reduce hospitalisations and increase survival in patients with heart failure with reduced ejection fraction (HFrEF). This study evaluated SGLT2i prescribing compliance in hospitalised patients with HFrEF at an Australian tertiary hospital.

Methods

A 10‐month retrospective review of patients treated for HFrEF (left ventricular ejection fraction ≤40%) was conducted. Patients were grouped as already on SGLT2i pre‐admission, initiated during admission, or not prescribed. Demographics, renal function, prescribing barriers, and post‐discharge outcomes were analysed.

Results

Of 142 patients, 27% were already taking an SGLT2i; of the rest, 46% were initiated during admission and 54% were not prescribed. Initiation rates were higher under cardiology (67% vs. 36%, P = 0.012). Non‐prescription was associated with older age, 80 versus 69 years for already prescribed (P=0.001) and 72 years for initiated during admission (P=0.007), and lower creatinine clearance, 48 mL/min versus 64 mL/min for already prescribed (P 0.049) and 61 mL/min for initiated during admission (P=0.047). Only one patient had an absolute contraindication; relative contraindications, mainly acute kidney injury (65%) and acute illness (35%), were documented in 25% of non‐prescribed cases. Just 9% of patients without inpatient initiation had discharge recommendations for community commencement. Thirty‐day readmission and mortality rates were numerically lower in patients prescribed SGLT2is, though differences were not statistically significant.

Conclusion

SGLT2is remain under‐prescribed in hospitalised patients with HFrEF, with advanced age and renal impairment influencing non‐prescription despite limited true contraindications. Implementing standardised prescribing protocols, reassessing relative contraindications prior to discharge and improving communication with primary care may enhance uptake of this guideline‐recommended therapy.

Background

Sodium‐glucose co‐transporter 2 inhibitors (SGLT2is) were initially introduced as a treatment option for type 2 diabetes due to their glucose‐lowering effect mediated by inhibiting the sodium‐dependent glucose co‐transporter 2 in the early proximal tubule. ¹ Research over the past decade has highlighted their broader benefits, particularly in managing heart failure (HF). The DAPA‐HF (Dapagliflozin and Prevention of Adverse Outcomes in Heart Failure), EMPEROR‐Reduced (Empagliflozin Outcome Trial in Patients With Chronic Heart Failure With Reduced Ejection Fraction), and EMPEROR‐Preserved (Empagliflozin Outcome Trial in Patients With Chronic Heart Failure With Preserved Ejection Fraction) trials demonstrated that the risk of worsening HF or death from cardiovascular causes was lower among those who received an SGLT2i, regardless of their diabetic status. ² , ³ , ⁴ The DAPA‐HF and EMPEROR‐Reduced trials further demonstrated improvement in quality of life and symptom relief in these patients. The EMPULSE (Empagliflozin in Patients Hospitalized With Acute Heart Failure Who Have Been Stabilized) trial, which assessed the utility of empagliflozin in hospitalised patients with decompensated de novo or chronic HF, showed a 36% greater benefit in the composite primary outcome of mortality and morbidity from HF. ⁵

SGLT2is are also known to provide renal benefits by slowing the progression of kidney disease in patients with HF. According to the DAPA‐CKD (Dapagliflozin and Prevention of Adverse Outcomes in Chronic Kidney Disease) trial, the composite of a sustained decline in the estimated glomerular filtration rate (eGFR) of at least 50%, end‐stage kidney disease or death from renal causes was 44% less compared with the placebo group. ⁶ Similarly, the composite of death from cardiovascular causes or hospitalisation for HF was 29% less in the dapagliflozin group.

The aforementioned studies underscore the robust evidence supporting the use of SGLT2is in patients with HF, showcasing their multifaceted benefits in reducing hospitalizations, improving survival, enhancing symptoms, protecting renal function and providing favourable metabolic effects. Due to this growing body of evidence, SGLT2is have been incorporated into HF management guidelines as a key component of standard therapy. ⁷

Despite the robust body of evidence supporting these benefits, SGLT2is remain underutilised in the reduction of cardiovascular risk. DISCOVER (Effectiveness of Diabetes Treatment Approaches in Routine Clinical Practice), a multinational study in 37 countries, reported that 8.7% of patients were started on an SGLT2i at enrolment, and a modest yet noteworthy increase of only up to 16.1% was noticed at the end of 3 years of follow‐up. ⁸ This underutility is reported to be even more pronounced in co‐morbid patients. A Canadian observational study noted only 20% of patients with diabetes and atherosclerotic cardiovascular disease were prescribed SGLT2is, and a history of HF and kidney disease independently lowered this rate. ⁹ Similar results have been reported by other US and UK studies. ¹⁰ , ¹¹

Patients admitted to a hospital provide a unique opportunity to optimise their therapies. This study aims to evaluate the prescription patterns of SGLT2is for patients diagnosed with HF with reduced ejection fraction (HFrEF) who were admitted to an Australian tertiary hospital.

Methodology

A retrospective observational study was conducted on patients who were discharged from Gold Coast University Hospital with a diagnosis of decompensated HF between 1 September 2023 and 30 June 2024. Patients with decompensated HF were screened based on the International Classification of Diseases, Tenth Revision (ICD‐10) diagnostic codes I11.0 I13.0, I13.2, I50.0, I50.1 and I50.9 as included in hospital integrated electronic medical records (IEMRs). Records of these patients were screened manually for inclusion and exclusion criteria and recruited in the study. Patients with a left ventricular ejection fraction of ≤40% were included in this study, as determined by transthoracic echocardiograms (TTE) performed before or during the admission of these patients. Patients whose TTE was not available or performed after admission were excluded from this study. De‐identified data on patient demographics, comorbidities, current medications, hospitalisation details including prescription of an SGLT2i and reasons for non‐prescription, and time to re‐admission (≤30 days) were collected from the IEMR. Creatinine clearance was estimated using the Cockcroft‐Gault formula for ideal body weight.

The primary outcome of compliance in prescribing an SGLT2i was assessed by evaluating whether an SGLT2i was prescribed for these patients at discharge. Data were analysed using STATA 17 statistical package. Patients were grouped into three categories: patients who already had an SGLT2i prescribed (SGLT2i—already prescribed), patients who were initiated on an SGLT2i while being an inpatient (SGLT2i—initiated inpatient) and patients who were not prescribed an SGLT2i (SGLT2i—not prescribed). The SGLT2i—not prescribed group was compared independently with the SGLT2i—already prescribed and the SGLT2i—initiated inpatient groups for demographic and clinical parameters. Chi‐square test was used to compare categorical variables. Independent‐sample t test was used to compare continuous variables after assessing for parameter distribution and homogeneity of variance. Kaplan–Meier curves were generated to illustrate time to readmission and time to death across groups and a log‐rank test was performed to assess the significance of differences.

The study was approved by the Gold Coast Human Research Ethics committee (approval number EX/2024/QGC/111456).

Results

Initial screening for ICD‐10 codes returned 548 eligible patients with HF admissions during the study period. Of these, 304 had an ejection fraction >40% and 102 patients did not have documented echocardiography reports and hence were excluded; 142 patients were recruited. Of the recruited patients, 94 patients had an echo performed while an inpatient and 48 had previous echocardiography done on average 262 days (standard deviation (SD)=239 days) before admission. Of the patients recruited, 39 (27%) had an SGLT2i already prescribed, while among the remaining patients, 47 (46%) had an SGLT2i initiated during their admission and 56 (54%) did not have an SGLT2i prescribed (Fig. 1).

Figure 1.

Participant flow diagram. SGLT2i, sodium‐glucose co‐transporter 2 inhibitor.

Of the SGLT2i—not prescribed group, only one patient had an absolute contraindication (anaphylaxis to SGLT2i). Of the rest, a relative contraindication was documented in only 14 (25%) patients, which included acute kidney injury (65%) and acute illness (35%). Only 9% had a general practitioner recommendation to commence an SGLT2i in the community.

The average age was 74 years (SD=14 years) and 61% were male. The demographic and clinical characteristics of these patients are summarised in Table 1.

Table 1.

Demographic and clinical characteristics

	Overall study population	Already prescribed	Initiated inpatient	Not prescribed*
n	142	39	47	41
Demographic characteristics
Age (years)	74 (SD=14)	69 (SD=15) P=0.001	72 (SD=13) P=0.007	80 (SD=13)
Male (%)	61	77 P=0.020	57 P=0.694	54
ATSI (%)	0.7	2.6 P=0.228	0	0
RACF resident (%)	7	10.3 P=0.713	4.3 P=0.686	7
Prescription data
β‐blockers (%)	83	92	87	73
ACEI/ARB (%)	51	44	62	51
ARNI (%)	27	54	30	7
MRA (%)	57	76	55	44
Clinical characteristics
Ejection fraction (%)	28 (SD 9)	26 (SD=10) P=0.081	27 (SD=9) P 0.222	29 (SD=8)
Diabetes (%)	32	54 P=0.045	23 P=0.553	29
Hypertension (%)	77	59 P=0.029	43 P=0.066	61
Creatinine clearance (mL/min)	57 (SD=36)	64 (SD=44) P=0.049	61 (SD=30) P=0.047	48 (SD=33)

Note: The P value reported is compared with the SGLT2i not prescribed group.

Abbreviations: ACE, angiotensin‐converting enzyme inhibitor; ARB, angiotensin II receptor blocker; ARNI, angiotensin II receptor blocker and neprilysin inhibitor; ATSI, Aboriginal and Torres Strait Islander; MRA, mineralocorticoid receptor antagonist; RACF, residential aged care facility; SGLT2i, sodium‐glucose co‐transporter 2 inhibitor.

^*After excluding patients with absolute/relative contraindications (n = 15).

Patients who did not have an SGLT2i prescribed were significantly older compared with the ones who had it already prescribed (80 vs 69 years, P=0.001) and initiated inpatient (80 vs 72 years, P=0.007). Patients who were older than 65 years were less likely to have an SGLT2i already prescribed (61% vs 88%, P=0.007) or initiated during admission (70% vs 88%, P=0.045) (Fig. 2).

Figure 2.

SGLT2i prescription based on age. (A) Age distribution vs SGLT2i prescription pattern. (B) Age >65 years vs SGLT2i prescription pattern. SGLT2i, sodium‐glucose co‐transporter 2 inhibitor.

A significantly higher proportion of males were taking an SGLT2i at the time of admission compared with the SGLT2i—not prescribed group (77% vs 54%, P=0.02). There was no gender‐based difference when an SGLT2i was initiated as inpatient.

Estimated creatinine clearance was lower in patients who did not have an SGLT2i prescribed compared with the those who had it already prescribed (48 vs 64 mL/min, P=0.049) and thsoe who had it initiated during admission (48 vs 61 mL/min, P=0.047). Patients with moderate to severe renal impairment (creatinine clearance <45 mL/min) were compared with others. Patients who had creatinine clearance <45 mL/min were more likely to have an SGLT2i not prescribed at the time of admission (61% vs 38%, P=0.033) or initiated during admission (61% vs 36%, P=0.013) (Fig. 3).

Figure 3.

SGLT2i prescription based on renal function. (A) Creatinine clearance vs SGLT2i prescription pattern. (B) Creatinine clearance <45 mL/min vs SGLT2i prescription pattern. SGLT2i, sodium‐glucose co‐transporter 2 inhibitor.

Patients who had diabetes had a significantly higher rate of SGLT2i already prescribed (54%, P=0.045) at presentation. Of the other diabetic patients, only 41% (P=0.553) had an SGLT2i initiated during their admission.

Dapagliflozin was the most commonly (79%) prescribed SGLT2i, followed by empagliflozin. Co‐prescribed medications included β‐blockers (83%), angiotensin‐converting enzyme inhibitors (ACEIs)/angiotensin receptor II blockers (ARBs) (51%), angiotensin receptor neprilysin inhibitors (27%) and mineralocorticoid receptor antagonists (MRAs) (57%).

The cardiology team initiated SGLT2i significantly more than other specialties (67% vs 36%, P=0.012). There was no difference in prescription rates between general medicine and other sub‐specialties (37% vs 29%, P=0.678).

Within 30 days of discharge, 27% of patients were re‐admitted. This was higher, but not statistically significant, in patients who were not taking an SGLT2i (33%), compared with the patients who were already taking an SGLT2i (26%, P=0.471) and initiated on one during admission (23%, P 0.271).

The median time for re‐admission within 30 days of discharge was 15 days for patients who were already taking SGLT2i or initiated on one during admission and 14 days for patients who were not taking an SGLT2i. There was no difference in these re‐admission times based on log‐rank testing (χ²=1.27, P=0.529). Kaplan–Meier curves for re‐admission times are depicted in Figure 4.

Figure 4.

Kaplan–Meier curves for re‐admission times within 30 days post‐discharge.

At the time of recruitment of cases to this retrospective study, 30 patients (21%) had died. Median follow‐up time of this retrospective cohort was 11 months (interqartile range: 8–13 months). Mortality rate was higher (29%), but not statistically significant, in patients who did not have an SGLT2i prescribed compared with the ones who had SGLT2i already prescribed (21%, P=0.374) and the ones who had SGLT2i initiated during their admission (13%, P=0.051).

Median time to death was 3.1 months for patients who were already on an SGLT2i, 4.6 months for patients who were initiated on an SGLT2i during admission, and 2.2 months for patients who did not have an SGLT2i prescribed. This numerical difference was not statistically significant in log‐rank testing (χ² 2.14, P=0.343). Kaplan–Meier curves for time to death are depicted in Figure 5.

Figure 5.

Kaplan–Meier curves for time to death.

Discussion

SGLT2is, a paradigm‐shifting class of medications, remains underprescribed in HFrEF despite robust and growing evidence supporting their use. This single‐centre retrospective study aimed to examine the prescribing patterns of SGLT2i and evaluate their use in patients with HFrEF admitted to a tertiary healthcare facility in Australia.

Current literature suggests that although overall prescription rates of SGLT2is remain low, there is a promising upward trend. Population‐based studies from Canada and the United States have demonstrated up to a 50‐fold rise in SGLT2i uptake over the past decade. ¹² , ¹³ In Australia, national data indicate a 200‐fold increase in SGLT2i prescriptions between 2014 and 2022. ¹⁴ In our study, a comparatively higher 27% of patients with HFrEF were taking an SGLT2i at the time of hospital admission, and it was particularly favourable (54%) among patients who had concurrent diabetes. This comparatively higher uptake may, at least in part, be attributed to better affordability in Australia. For instance, a 30‐day supply of dapagliflozin can cost over AUD 400 in the United States, whereas under the Australian Pharmaceutical Benefits Scheme, it is available for approximately AUD 30. ¹⁵ , ¹⁶ Beyond financial accessibility, other literature reported factors that include limited clinician awareness, established prescribing habits, clinician concerns regarding risk–benefit trade‐offs and administrative or regulatory hurdles. ¹⁷ , ¹⁸

In our cohort, of the patients who were not taking an SGLT2i on admission, 46% had an SGLT2i prescribed during their admission. This is higher than what is reported in comparable single‐centre retrospective studies of 16% to 30%, highlighting persistent gaps in therapeutic optimisation. ¹² , ¹⁹ Notably, although not statistically significant, our data showed that patients with diabetes were less likely to be initiated on an SGLT2i during their admission—despite the well‐established efficacy, safety and tolerability of inpatient initiation in this population. ²⁰

Other guideline‐directed therapy including ACE/ARB/ARNI and β‐blockers had good uptake, although they did not reach 100%. Despite clear guideline indications backed by studies such as RALES (Randomized Aldactone Evaluation Study) and EMPHASIS‐HF (Eplerenone in Mild Patients Hospitalization and Survival Study in Heart Failure), uptake of MRAs in this cohort was suboptimal at 57%. ²¹ , ²² Though exploring causes for this was beyond the scope of this study, potential contributors to low uptake could include clinician concerns regarding hyperkalaemia and renal dysfunction, competing priorities during acute admissions, therapeutic inertia and fragmentation between inpatient and outpatient optimisation. In addition, the contemporary introduction of SGLT2is may have shifted prescribing focus, despite MRAs remaining a core component of quadruple therapy.

Patients admitted under cardiology were significantly more likely to have SGLT2i therapy initiated. This trend suggests a gap in prescribing practices across departments and highlights the need for institutional strategies to promote guideline‐concordant therapy across departments. Similar findings have been reported elsewhere, where being known to a cardiology service increased SGLT2i prescription rates. ¹² , ¹⁸ Nevertheless, even across specialties, prescribing remains lower than expected. Shin et al. reported that despite patients meeting one or more guideline‐based indications, overall prescription rates rarely exceeded 25%. ²³ Taken together, these findings suggest that despite clear indications and robust supporting evidence, SGLT2i therapy continues to be underutilised across a range of clinical settings and specialties.

The list of absolute contraindications to SGLT2i is short, with the main one being severe kidney impairment; others include type 1 diabetes, hypersensitivity and previous diabetic ketoacidosis. However, there are more relative contraindications, eg, genitourinary tract infections, volume depletion, foot infections, susceptibility to fractures, surgery and acute illness. In our cohort, absolute contraindications were exceedingly rare. Among patients who were not initiated on an SGLT2i, relative contraindications were infrequently documented, and discharge summaries rarely included recommendations for general practitioners to consider initiating therapy post‐discharge. These findings suggest that, in most cases, the omission of SGLT2i therapy was more likely attributable to clinical inertia or uncertainty rather than the presence of valid contraindications. Notably, among patients without any documented contraindications, increasing age and reduced renal function were significantly associated with lower rates of SGLT2i prescribing, highlighting potential areas of bias or caution that may not be evidence‐based.

Older age has consistently been identified as a barrier to prescribing SGLT2is. For example, Lanier et al. reported that each additional year of age was associated with a 3% decrease in the likelihood of SGLT2i initiation. ¹⁹ Similarly, Jeong et al. observed a negative correlation between age ≥65 years and SGLT2i prescribing in an outpatient HF cohort. ²⁴ This age‐related discrepancy may stem from clinician concerns regarding limited representation of older adults in major clinical trials, along with presumed increased susceptibility to adverse events such as volume depletion, hypotension, genital mycotic infections and hypoglycaemia in this population. ²⁵ However, emerging evidence suggests that these concerns may be overstated. A post hoc analysis of VERTIS‐CV (Evaluation of Ertugliflozin Efficacy and Safety Cardiovascular Outcomes Trial), which included a subgroup of elderly participants, demonstrated comparable cardio‐renal benefits, preservation of kidney function and overall safety across age groups. ²⁶ These findings underscore the importance of not withholding SGLT2i therapy based solely on age, particularly when clear clinical indications are present.

There was a statistically significant male predominance in the patients who were already taking an SGLT2i on admission in our cohort. Similar gender discrepancies in SGLT2i prescription have been identified in multiple other studies, where SGLT2is were noted to be underutilised in female patients with HF, diabetes and chronic kidney disease. ²⁷ , ²⁸

In our study, patients with impaired renal function were significantly less likely to be prescribed an SGLT2i. Although the glucose‐lowering efficacy of SGLT2i diminishes at an eGFR <45 mL/min, their cardiovascular and renal protective benefits remain substantial. ²⁹ A meta‐analysis of four major SGLT2i trials demonstrated that these agents significantly reduce the risk of progression to dialysis, kidney transplantation and death from renal causes, and also offer protection against acute kidney injury, suggesting that withholding SGLT2is based on impaired renal function alone is unjustified. ³⁰ Furthermore, there are ongoing trials assessing SGLT2i efficacy in dialysis patients. ³¹

Our study did not reach statistical significance in avoiding readmissions and mortality, likely stemming from limitations of sample size and follow‐up duration, yet numerically lower 30‐day readmission rates, mortality rates and time for such events in patients who had an SGLT2i prescribed on admission or initiated during admission echoes outcomes of major clinical trials. It has been demonstrated that just 12 days after initiation of empagliflozin there had been a 58% relative reduction in mortality, HF hospitalisation or urgent HF visits. ³² Such early benefits have further been reinforced by the SOLOIST‐WHF (Sotagliflozin in Patients With Diabetes and Recent Worsening Heart Failure) study, where it was demonstrated that in‐hospital or early post‐discharge initiation of sotagliflozin significantly lowered total number of deaths from cardiovascular causes and hospitalizations and urgent visits for HF compared with placebo. ³³ Similarly, in another study, dapagliflozin was shown to be associated with rapid reduction in the risk of cardiovascular death or worsening HF, with an even greater relative and absolute risk reduction noted in patients who were initiated on dapagliflozin during a recent hospitalisation. ³⁴

Overall, this study reinforces both the clinical opportunity and the implementation gap in SGLT2i prescribing. The hospitalisation period represents a critical window for medication optimization. Systems‐based strategies—such as clinical decision support tools, multidisciplinary care models and discharge planning protocols—may enhance uptake during this period. Addressing misconceptions around age and renal function, along with ensuring appropriate handover to primary care, could also improve long‐term outcomes.

There is a notable paucity of data regarding the initiation and prescribing patterns of SGLT2is in hospitalised patients with HF. This study contributes to addressing this gap by providing insights into real‐world prescribing behaviours in this context. However, this study was conducted at a single centre and over a limited timeframe, which may affect the generalisability of the findings. Nonetheless, retrospective single‐centre studies such as this remain valuable for identifying local practice gaps and generating hypotheses to inform larger, prospective investigations.

Conclusion

This study highlights the ongoing gap in prescribing SGLT2is among eligible hospitalised patients with HFrEF, particularly in a non‐cardiology setting. Although clinical contraindications were infrequently documented, many patients remained untreated at discharge, and follow‐up plans for initiation were rarely communicated. Older age and poor renal function were significantly associated with poor prescription rates. While no statistically significant differences in 30‐day readmission or mortality were observed, trends suggest potential clinical benefit associated with SGLT2i use. These findings underscore the need for improved implementation strategies to support guideline‐concordant therapy in the inpatient setting.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.