Evaluation of demographic, clinical characteristics, and the prescribing practice of guideline-directed medical therapy among chronic heart failure outpatients in multidisciplinary clinics in a large tertiary hospital: a retrospective audit of 1186 patients from 12 years

Daya Ram Parajuli (  dayaram.parajuli@ inders.edu.au ) Flinders University Rural Clinical School https://orcid.org/0000-0002-9188-8265 Sepehr Shakib University of Adelaide School of Medical Sciences: The University of Adelaide Adelaide Medical School Joanne Eng-Frost Flinders Medical Centre Ross McKinnon Flinders Medical Centre Gillian Caughey South Australian Health and Medical Research Institute Dean Whitedhead University of Tasmania

clinic appointments or had incomplete data sets. Data were collected as part of routine clinical practice. The follow-up of patients varied depending on the date of rst presentation in either clinic.

Variables and outcomes
The outcome variables include the demographic, clinical characteristics, comorbidities and use of GDMT in CHF patients between two clinics. These outcome variables were measured between HFrEF, HFmrEF and HFpEF categories (demographics and clinical characteristics) as well as by MACS compared to GCHFS clinics (medication utilization only). The age, weight, systolic blood pressure (SBP), diastolic blood pressure (DBP), heart rate (HR), number of medications used, serum creatinine, hemoglobin, mean cell volume (MCV), and comorbidities were measured per patient. The SBP, DBP and HR are the four consecutive readings at rest, ve minutes apart, and average of last three readings. The data utilized were from the last clinic appointment. The hemoglobin, MCV and creatinine are the last values before rst presentation to clinic (which would usually represent the last values before hospital discharge) and the weight was measured at rst appointment.

Outcome measurements
The LVEF value of <40% for HFrEF, 40-49 % for HFmrEF, and ≥50% for HFpEF was considered for comparison of demographic, clinical characteristics and comorbidities whereas LVEF value <40% for HFrEF and ≥40% for HFpEF was considered for the evaluation of GDMT. It is important to note that the evaluation of GDMT was only considered for HFrEF and HFpEF in our analysis. It is clinically signi cant because, for the duration of this study, there were no separate guidelines for HFmrEF patients in the hospital where this study was conducted. To perform the evaluation of GDMT, a guideline was developed based on Australian and European guidelines in the management of CHF (see supplementary data). Patients data were reviewed for the type of medications prescribed, doses used and contraindications due to patient characteristics.
Classi cations for each group of medications was performed by two independent investigators (DRP and JEF) and checked for discrepancies as per the developed guidelines, with disagreements resolved by consensus with a third investigator (SS). It was determined that we were unable to determine the use of evidence-based therapies in CHF patients if they do not visit the MACS clinics at least twice. Therefore, for the comparison of the use of EBT, only patients who had ≥2 visit in MACS clinics were included. Polypharmacy was categorized into three groups: non-polypharmacy (0-4 drugs), polypharmacy (5)(6)(7)(8)(9) drugs) and hyperpolypharmacy (≥ 10 drugs) as de ned by Onder et al 2012 [31].

Study size
During the study period, there was a total of 1186 CHF patients who attended the outpatient clinics and met our eligibility criteria. For the evaluation of EBTs, an individual data of 359 patients from MACS clinic, and 369 patients from the GCHFS clinic were available.

Statistical Analysis
Statistical analysis was performed using IBM SPSS Statistics for Windows (Version 25.0.0.1. Armonk, NY: IBM Corp). Results are presented as frequency and percentages for categorical variables and median (IQR) for continuous variables. The normal distribution of the numeric variables was con rmed using Shapiro-Wilk test (P>0.05). Median differences between two clinics was evaluated using Mann-Whitney U test for comparison of demographic and clinical characteristics and use of EBTs between two clinics whereas Kruskal-Wallis test was used for similar comparison among ejection fraction groups. Univariate and multivariate binary logistic regression was performed to determine the predictors of evidence-based practice. Nagelkerke R 2 was used to establish the amount of variance explained by the model. Univariate binary logistic regression was performed to determine the important variables to be included in multivariate analysis. Independent variables which showed a value < 0.25 in univariate analysis were included in the multivariate analysis. Probability values of p <0.05 were chosen to indicate a statistically signi cant difference.

Participants
A total of 1186 patients were included: 725 patients were in the MACS clinic and 461 patients were in the GCHFS clinic. After excluding 74 patients in the MACS clinic and 54 from the GCHFS clinic who did not have echocardiography, the remaining 651 patients in the MACS clinic and 407 patients in the GCHFS clinic were included to compare their demographics and clinical characteristics by clinics and strati ed by ejection fraction (Figure 1). Two patients from the MACS clinic and thirty-eight patients from the GCHFS clinic were excluded due to incomplete data sets). For the evaluation of EBTs, individual data for 489 HFrEF patients and 239 of the HFpEF patients were reviewed for the type of medications prescribed, doses used and contraindications due to patient characteristics. The ow diagram of the study is illustrated in Figure 1.

Descriptive data
Comparison of the differences in demographics, clinical characteristics, and comorbidities by ejection fractions A comparison of the demographics and clinical characteristics among the CHF patients strati ed by ejection fraction is illustrated in Table 1 and Figures 2, Figure 3, and Figure 4. The prevalence of HFrEF, HFmrEF and HFpEF was 56%, 13% and 31%, respectively (p<0.001) ( Table 1). The median age of patients in HFpEF and HFmrEF was signi cantly greater (p < .001) than that of the HFrEF cohort. There was no signi cant difference in the distribution of weight, DBP and HR, serum creatinine and MCV, among HFpEF, HFmrEF and HFrEF group of patients (Table 1). In contrast, a statistically signi cant difference (p < .001) was observed for median SBP and number of medications used among HFpEF, HFmrEF and HFrEF group of patients. Hemoglobin level was highest for HFrEF followed by HFmrEF and HFpEF group. The prevalence of hypertension, AF, osteoarthritis, anemia, and asthma for HFmrEF patients lies between HFrEF and HFpEF whereas there was a highest prevalence of IHD in HFmrEF followed by HFrEF and HFpEF patients (p < .001) Table 1. But the prevalence of other comorbidities was similar between HFrEF, HFmrEF and HFpEF patients.
Comparison of the differences in demographics, clinical characteristics, and comorbidities by clinics Demographic and clinical characteristics of CHF patients compared by clinics are illustrated in Table 2.
MACS clinic patients were signi cantly older (p < .001), less likely to be male, had a signi cantly higher SBP (p < .001) and DBP (p < .05) compared to GCHFS clinic patients. There was also a signi cant difference in the age group category between the two clinics. For patients with >80 years, the MACS clinic had a much higher prevalence of older patients compared to GCHFS clinic. However, weight and HR were similar between the two clinics. The number of medications used was signi cantly higher in MACS patients (p < .001) compared to GCHFS patients. There also exist signi cant differences in polypharmacy and hyperpolypharmacy, and their prevalence between the two clinics, respectively.
There was no difference in prevalence of IHD, AF, hyperlipidemia, CRF, solid cancer and gout between the two clinics. The prevalence of major comorbidities was signi cantly more common in MACS patients compared to GCHFS patients, respectively. The prevalence of patients with multiple comorbidities was statistically signi cantly higher in the MACS clinic patients compared to the GCHFS clinic patients ( Table  2).
Comparison of the use of medications in chronic heart failure patients between clinics with heart failure with reduced and preserved ejection fractions The MACS clinic had similar rates for the guideline-based prescriptions regarding appropriate use of ACEIs/ARBs (68.4% v. 72%) as well as the rate of appropriate use of the MTD of ACEIs/ARBs (46.3% v. 52%) compared with the GCHFS clinic patients in HFrEF patients. A signi cantly lower rate of appropriate use of β-blockers (83.1% v. 91.1%), MTDs of β-blockers (31.5% v. 47.3 %), and MRAs (32.1% v. 62.2%) were observed in the MACS clinic patients compared to the GCHFS clinic patients ( Table 3). The use of target dose of ACEIs/ARBs was similar, but signi cantly lower in the MACS clinic compared to those in the GCHFS clinic for the use of β-blockers. Further, the MACS clinic patients had similar rates of prescription for diuretics, but a signi cantly higher prescription for digoxin in chronic AF (82.5% v. 58.5%) in HFrEF patients.
For patients with HFpEF, a signi cantly higher prescriptions of ACEIs/ARBs (70.5% v. 56.2%) in MACS clinic, but a signi cantly lower prescriptions of β-blockers (54% v. 68.5%), MRAs (30.1% v. 48%), furosemide and anticoagulation for AF were observed in the MACS clinic patients compared to those in the GCHFS clinic (Table 4). However, a similar prescription rate for digoxin was seen between two clinics.
Age, last clinic SBP, last clinic DBP, AF, anaemia, IHD, CRF, COPD, any cognitive impairment, any solid cancer, any CVA, falls, osteoarthritis, GORD, peripheral vascular disease (PVD), gout, ≥3 comorbidities and any thyroid disease being the signi cant predictors in the univariate analysis, were included in multivariate analysis of ACEIs/ARBs use (data not shown). Nagelkerke R 2 showed that the above variables used in the multivariate binary logistic analysis model could explain 26.4% in predicting the practice of ACEIs/ARBs use. Age, anaemia, CRF, gout and GORD were the negative predictors whereas, SBP was a positive predictor for the use of ACEIs/ARBs in HFrEF patients in the multivariate analysis (Table 5). Similarly, age, AF, IHD, CRF, COPD, any solid cancer, osteoarthritis, GORD, gout and presence of ≥3 comorbidities being the signi cant predictors in the univariate analysis, were included in multivariate analysis of the MTD use of ACEIs/ARBs (data not shown). The model explained 13.4% (Nagelkerke R 2 ) in predicting the practice of MTD of ACEIs/ARBs. Age and CRF were signi cant negative predictors of the use of MTD of ACEIs/ARBs in the multivariate analysis (Table 5).
Age, gender, HR, COPD, any solid cancer, gout, any anemia, IHD, any cognitive impairment, osteoporosis and any thyroid diseases being the signi cant predictors in the univariate analysis, were included in multivariate analysis of β-blockers use (data not shown). The model explained 12.9% (Nagelkerke R2) in predicting the use of f β-blockers. HR and gout were the signi cant negative predictors, but IHD is a signi cant positive predictor for the use of β-blockers in HFrEF patients in the multivariate analysis (Table  5). Age, gender, HR, COPD, any solid cancer, gout, any anemia, IHD, any cognitive impairment, osteoporosis and any thyroid diseases being the signi cant predictors in the univariate analysis, were included in multivariate analysis of MTD of β-blockers (data not shown). The model explained 12.9% (Nagelkerke R2) in predicting the use of MTD of β-blockers. HR and gout were the signi cant negative predictors, but IHD was the signi cant positive predictor for the use of MTD of β-blockers in HFrEF patients in the multivariate analysis (Table 5).
Age, Gender (male), last clinic SBP, last clinic DBP, last clinic postural BP, AF, anemia, CRF, hypertension, any cognitive impairment, any solid cancer, hyperlipidemia, falls, osteoarthritis, osteoporosis, PVD and ≥3 comorbidities being the signi cant predictors in the univariate analysis, were included in multivariate analysis of MRA use (data not shown). The model explained 26.4% (Nagelkerke R2) in predicting the use of MRA. Age and SBP were the signi cant negative predictors for the use of MRAs in HFrEF patients in the multivariate analysis ( Table 5).
Predictors of use of ACEIs/ARBs, β-blockers and MRAs in heart failure with preserved ejection fraction (EF>40) patients. Gender (male), hypertension, CRF, CVA, COPD, cognitive impairment, gout, and falls were signi cant predictors (p <.25) in the univariate analysis being the signi cant predictors in the univariate analysis, were included in multivariate analysis of ACEIs/ARBs (data not shown). The model explained 18.5% (Nagelkerke R2) in predicting the use of ACEIs/ARBs. CRF, and cognitive impairment were the signi cant negative predictors, but hypertension and COPD were the signi cant positive predictor for the use of ACEIs/ARBs in the multivariate analysis of the HFpEF patients (Table 5).
Hypertension, last clinic HR, last clinic low heart rate (HR<60), anemia, IHD, diabetes, COPD, cognitive impairment, hyperlipidemia, osteoarthritis and GORD being the signi cant predictors in the univariate analysis, were included in multivariate analysis of β-blockers use (data not shown). The model explained 30.1% (Nagelkerke R2) in predicting the use of β-blockers. HR, COPD and GORD were the signi cant negative predictors, but IHD was the signi cant positive predictor for the use of β-blockers in the multivariate analysis of the HFpEF patients (Table 6).
Gender, hypertension, AF, IHD, diabetes, CRF, asthma, hyperlipidemia, osteoporosis, and low standing SBP being the signi cant predictors in the univariate analysis, were included in multivariate analysis of MRAs use in HFpEF patients (data not shown). The model explained 15.5% (Nagelkerke R2) in predicting the use of MRAs. Only the low standing SBP was a signi cant positive predictor for the use of MRAs in the multivariate analysis (Table 6).

Discussion
This study is a detailed analysis of demographics, clinical characteristics, comorbidities, and prescribing practice of GDMT in CHF outpatients in a large tertiary referral hospital in South Australia. The HFmrEF subjects resembled the HFpEF patients in terms of age, HR, SBP and having higher prevalence of polypharmacy whereas resembled with the HFrEF for the proportion of male distribution and prevalence of IHD. MACS clinic had similar rates of guideline-based prescribing of ACEIs/ARBs, their MTD and target doses, diuretics, and digoxin use in HFrEF, but signi cantly higher prescription of ACEIs/ARBs was found in HFpEF patients. There were signi cantly lower rates of prescription of β-blockers, MTD of β-blockers, target dose of β-blockers, and MRAs prescribed in the both the HFrEF and HFpEF groups in the MACS clinic, as compared to the GCHFS clinic. Additionally, signi cantly higher prescription for digoxin in chronic AF in HFrEF patients and signi cantly lower prescription of furosemide and anticoagulation for AF in HFpEF patients were observed in the MACS clinic patients compared to those in the GCHFS clinic.
The HFrEF, HFmrEF and HFpEF patients in this study were much older than the ESC Heart Failure Long-Term (ESC-HF-LT) registry [32]. HFpEF patients in the current study were 2 years older but one year younger for HFrEF group compared with the age of patients in the NSW (Australia) snapshot study [33]. AF prevalence in the current study was in ascending order with the increasing value of LVEF as found in the Swedish Heart Failure Registry [34]. In line with our results, similar ndings to the HFmrEF group resembling HFrEF for male gender, and IHD, were reported in an earlier studies [35,36]. The current study found a notable difference in demographics and comorbidities with the different cut-offs for EF. Based on above-mentioned results, our study showed intermediate demographic and clinical characteristics for HFmrEF category between HFpEF and HFrEF.
GDMT use was higher in the current study compared to the NSW HF snapshot study for the use of ACEIs/ARBs, β-blockers and MRAs in HFrEF patients [33]. Similar patterns of better use of ACEIs/ARBs, βblockers and MRAs were evident, but there were slightly lower rates of prescription of diuretics and digoxin observed in current study compared to an another Australian study on chronic HFrEF patients [37]. Importantly, the prescription of ACEIs/ARBs, β-blockers and MRA in current study were similar or even superior than to previous studies conducted in Australia  . Although our hypothesis was that the MACS clinic will have better practice of the ACEIs/ARBs, older age, anaemia and CRF being the signi cant negative predictors, patients received similar pattern of medicines to that of GCHFS clinic. Additionally, a greater number of contraindications for the use of ACEIs/ARBs and presence of polypharmacy were important factors to be considered in the MACS clinic compared to GCHFS clinic. Notably, gout and GORD are negative signi cant predictors for the utilization of ACEIs/ARBs in HFrEF patients which have not been reported before in the literature. Even though SBP was a signi cant positive predictor, and MACS clinic patients had higher SBP, impact of negative predictors and other variables as explained above was superior for the utilization of ACEIs/ARBs. Age and presence of CRF were signi cant negative predictors for the MTD use of ACEIs/ARBs.
Compelling evidence exists regarding underutilization of β-blockers and failure to up-titration in CHF patients including older age (>70 years) and presence of respiratory disease [44], hypotension and polypharmacy [45], concern of side effects, contraindications, and poor experience of GPs [46] low HR and poor adherence to prescriptions [47]. Patients under β-blockers may experience adverse effectsmortality and cardiovascular events associated with high resting HR, as described by Chen et al 2019 [48]. Nevertheless, the reluctance of the clinicians to prescribe β-blockers due to potential side effects need further investigation. In line with the previous ndings, HR was a signi cant negative predictor for the use of β-blocker, and HR and older age were the negative signi cant predictors for the MTD use of βblockers in our study. Other potential reasons for the lower utilization of β-blockers despite of presence of a pharmacist in the MACS clinic could be higher prevalence of polypharmacy, due to existing comorbidities [45] and presence of contraindication for their use [24, 46] as reported before. Patients were much older, and the prevalence of gout were signi cantly higher but IHD and HR were similar in HFrEF patients between two clinics. Notably, gout as a signi cant negative predictor and GORD as a signi cant positive predictor for the utilization of β-blocker were found in our study in HFrEF patients which have not been reported before in the literature. In certain instances, the underlying reason for the underutilisation of GDMT may also be unknown.
A critical reason behind the underutilisation of MRAs in HF is due to associated hyperkalaemia and the detrimental effect on renal function as reported earlier [49]. In contrast to previous study, renal function was not a signi cant predictor in our ndings. However, patient age and last clinic SBP were signi cant negative predictors for the utilisation of MRAs in the current study. There were more patients having contraindications for the use of MRAs in MACS clinic than GCHFS clinic patients, which had certainly some effect on lower prescription. Further research is needed to con rm other relevant reasons for example the occurrence of hyperkalaemia. Presence of digoxin use is more likely if patients have AF [37] as it improves morbidity in HF patients [50]. One key advantage of a pharmacist being on the MACS clinic was that a signi cantly higher number of patients received digoxin due to higher prevalence of chronic AF in the MACS clinic than in the GCHFS clinic in HFrEF patients.
Our study found that 41.5% of patients were given the recommended target dose for ACEIs/ARBs and 31% of patients received recommended doses of β-blockers. Overall, it is important to note that target dose prescribed in our study were superior than larger studies conducted in Europe [22,24], and in Asia [51]. The tolerability of speci c doses in individual patients with multiple comorbidities and polypharmacy in HF patients should be closely monitored rather than just an approach to reach the target doses [52,53], therefore, it is crucial that the emphasis for up-titration should be adopted based on an individualised dose approach. According to a systematic review, the widely recognised de nition of polypharmacy is a condition that requires the use of ve or more medications daily (range = 2 to 11) [54]. The mean number of medications used in the MACS patients was 11 ± 4, much higher than reported indicating that there was substantial polypharmacy in MACS clinic patients. Lower GDMT use in HFrEF patients due to underlying contraindications have been reported previously [55]. The contraindications for use of ACEIs, MTDs of ACEIs, β-blockers and MTD of β-blockers were signi cantly higher in MACS clinic patients than in GCHFS clinic patients. It can be hypothesized that as many patients have contraindications, clinicians were more reluctant to prescribe GDMT due to lack of more extensive experience of appropriate dosing when patients do not have contraindications. These ndings highlighted that contraindications may be one potential reason for lower utilization of GDMT in the MACS clinic in the current study despite the pharmacist's active involvement.
Despite of some dilemma for understanding of the epidemiology, pathophysiology and lack of evidence for the effective management of HFpEF, expert groups have highlighted that the management of HFpEF has been partly addressed due to the possible bene ts of currently available medications [56]. In contrast to HFrEF patients, a signi cantly higher prescription of ACEIs/ARBs but the similar trend of signi cantly lower prescription of β-blockers and MRAs in the MACS clinic patients compared to GCHFS clinic in HFpEF patients was revealed. The higher prescription of ACEI/ARBs in HFpEF patients in MACS patients may be due to underlying left atrial hypertension and pulmonary hypertension as explained by Lam et al 2018 [57] due to more number of HFpEF patients in the MACS clinic. Hypertension and COPD are the signi cant positive predictors whereas CRF and cognitive impairment were the signi cant negative predictors for the utilization of ACEIs/ARBs in our study. In consistent to our study, a previous Australian study demonstrated a signi cantly lower prescription of ACEIs in HFpEF patients compared to those in HFrEF patients [58]. These ndings indicate that patients in the HFpEF category in our study were not over treated. It has also been reported that age is a strong predictor of the lower prescription of β-blockers in the elderly in HFpEF patients [59]. The presence of COPD, gout and last clinic HR were signi cant positive predictors for the lower use of β-blockers in HFpEF in the current. However, the presence of IHD was a signi cant positive predictor for the use of β-blockers. Again, the differential prevalence of these comorbidities between the MACS and GCHFS clinics explains why MACS patients have signi cantly lower prescriptions of β-blockers in our study. The low standing SBP, was associated with a higher prescription of MRAs in HFpEF patients. Indeed, effectiveness of currently available GDMT for HFpEF is still a controversy except a symptomatic management of underlying comorbidities.
A systematic review and meta-analysis revealed that the use of MRAs in HFpEF was associated with ADRs including hyperkalaemia and gynecomastia compared with HFrEF patients [60]. The exact bene ts of MRAs in HFpEF patients is still poorly understood [61]; therefore, the generalisation of the role of currently available medications may not be clinically relevant. The MACS clinic, being a holistic model of care, may have considered these ADRs in prescribing MRAs in HFpEF patients, which may be a potential reason for a signi cantly lower prescription of MRAs in the MACS clinic compared with in the GCHFS clinic. Some cases of inappropriate prescribing were also noticed; for example, two patients were on two β-blockers simultaneously, two patients were on both ACEIs and ARBs and one patient received the wrong dose for apixaban. Similarly, some patients were on contraindicated medications. The bene t of having a pharmacist in the multidisciplinary team is that pharmacists can easily detect cases of inappropriate prescribing and contraindicated medications under usage. The underutilization of β-blockers and MRAs in the MACS clinic in HFrEF and HFpEF patients can not only be generalized just with the presence of a pharmacist compared to GCHFS clinic as the model of practice in MACS clinic is shared decision making by clinicians and a pharmacist for the effective management of CHF outpatients.

Strength and limitations
The major strength of this study is that it includes a large number of real-life data of CHF patients (12 years data) from a large tertiary hospital. In addition, this study compared the similarities and differences of CHF patients for HFrEF, HFmrEF and HFpEF category although the comparison of GDMT use is only between HFrEF and HFpEF category. Only the patients who have echocardiography data to determine the left ventricular function to measure the ejection fraction were included in current analysis. We did have some limitations for this study. While the clinics were set up slightly differently, the main predictor was the presence of a pharmacist in the MACS service and no involvement of a pharmacist in the GCHFS clinic. The most important confounder in this study was the series of different guidelines for the management of CHF across the study duration. The quali cations, experiences, and expertise of the pharmacists in the MACS group and all the nurses and clinicians in both groups are the effect modi ers in this study. The main bias here was the referral bias, where different types of patients may be referred to the two different clinics. As there were no separate guidelines for HFmrEF patients in the hospital where this study was conducted, evaluation of GDMT was only considered for HFrEF and HFpEF in our analysis.
Due to the limitation of funding and study completion timeline, differences in hospitalizations and mortality rates were not evaluated although they are important clinical outcomes.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
No competing interests exist. Author's contributions DRP: study design, development of analysis protocol development, data coding, data analysis, and development of the manuscript draft; SS: analysis protocol development, study design, data custodian, and critically reviewed the manuscript; EFJ: second author on classi cation of medications as per the analysis protocol, critically reviewed the manuscript; RM: Supervised the whole work and critical review of the manuscript; CG: consultation for data analysis and critically reviewed the manuscript; DW: Supervised the whole work and critical review of the manuscript. The author(s) read and approved the nal manuscript.      Variable (s) entered on step 1: last clinic SBP, last clinic DBP, any anemia, CRF, gout, IHD, GORD, any solid cancer, any CVA, PVD, OA, falls, any cognitive impairment, COPD, AF and any thyroid. EF: ejection fraction; SBP: systolic blood pressure; DBP: diastolic blood pressure; CRF: chronic renal failure; IHD: ischemic heart diseases; GORD: gastroesophageal re ux diseases; CVA: cardiovascular accident; PVD: peripheral vascular disease; COPD: chronic obstructive pulmonary diseases; AF: atrial brillation; ACEIs/ARBs: angiotensin converting enzyme inhibitor/angiotensin receptor antagonists; MTD: maximum tolerated dose; MRAs: mineralocorticoid receptor blockers; HR: heart rate; COPD: chronic obstructive pulmonary diseases; IHD. Only the signi cant variables in multivariate analysis are shown.   Age distribution in heart failure with reduced, mid-range and preserved ejection fractions. Kruskal-Wallis test showed a signi cant difference of age distribution among three ejection fractions (p<.001). p<.05 was considered signi cant.

Figure 4
Polypharmacy distribution in heart failure with reduced, mid-range and preserved ejection fractions.
Kruskal-Wallis test showed a signi cant difference of polypharmacy distribution among three ejection fractions (p<.001). p<.05 was considered signi cant.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Supplementary.docx