Article Text
PDF
Abstract
Aims The aim of this study was to investigate a real-world heart failure (HF) cohort regarding (1) prevalence of known diabetes mellitus (DM), undiagnosed DM and pre-diabetes, (2) if hf treatment differs depending on glycaemic status and (3) if treatment of DM differs depending on HF phenotype.
Methods All patients who had received a diagnosis of HF at Umeå University Hospital between 2010 and 2019 were identified and data were extracted from patient files according to a prespecified protocol containing parameters for clinical characteristics, including echocardiogram results, comorbidities, fasting plasma glucose (FPG) and hemoglobin A1c (HbA1c) values. Patients’ HF phenotype was determined using the latest available echocardiogram. The number of patients with previous DM diagnosis was assessed. Patients without a previous diagnosis of DM were classified as non-DM, pre-diabetes or probable DM according to FPG and HbA1c levels using WHO criteria.
Results In total, 2326 patients (59% male, mean age 76±13 years) with HF and at least one echocardiogram were assessed. Of these, 617 (27%) patients had a previous diagnosis of DM. Of the 1709 patients without a previous diagnosis of DM, 1092 (67%) patients had either an FPG or HbA1c recorded, of which 441 (41%) met criteria for pre-diabetes and 97 (9%) met criteria for probable diabetes, corresponding to 19% and 4% of the entire cohort, respectively. Patients with HF and diabetes were more often treated with diuretics and beta blockers compared with non-DM patients (64% vs 42%, p<0.001 and 88% vs 83%, p<0.001, respectively). There was no difference in DM treatment between HF phenotypes.
Conclusions DM and pre-diabetes are common in this HF population with 50% of patients having either known DM, probable DM or pre-diabetes. Patients with HF and DM are more often treated with common HF medications. HF phenotype did not affect choice of DM therapy.
- HEART FAILURE
- Diabetes Mellitus
- Epidemiology
Data availability statement
Data are available on reasonable request. Request should be made to the corresponding author. Data is available in the form of deidentified spreadsheets.
This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Heart failure (HF) and diabetes mellitus or pre-diabetes are commonly coexisting disorders which negatively impact patient outcome.
Prevalence estimates for glycaemic disorders in HF patients differ between settings, and often come from selected populations, for example, trial participants.
WHAT THIS STUDY ADDS
This single-centre study encompassing an unselected HF population finds that 27% of patients had known diabetes mellitus, 4% of patients had probable undiagnosed diabetes and 19% of patients had pre-diabetes.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The high prevalence of glycaemic disorders warrants a generous approach to glycaemic testing in HF patients, and further studies on how to best manage pre-diabetes in HF.
Background
Heart failure (HF) and diabetes mellitus (DM) are two common disorders that individually increase mortality, morbidity and healthcare costs.1 2 Patients with both conditions have a worse prognosis than patients with either diagnosis alone.3 The prevalence of DM in HF is high, between 20% and 40%, depending on geographical region and population studied, with the higher numbers seen in hospitalised patients.4 5 DM is increasingly viewed, not only as a comorbidity to HF, but also as a driver of HF, especially HF with preserved ejection fraction (HFpEF).6
Pre-diabetes, defined as above normal blood glucose and below the diagnostic cut-off for DM,7 8 is also associated with worse outcomes in HF.9 The prevalence of pre-diabetes has been less thoroughly studied than DM, and is mostly based on clinical trials for HF medications.10–12 Some patients are also first diagnosed with DM at inclusion in clinical trials.13 14 However, patient populations in clinical trials are often not representative of the population at large.15 16 There is a need for updated data on the prevalence of pre-diabetes and diabetes in modern HF populations.
The aim of this study was to investigate a Northern European real-world HF population regarding:
Prevalence of pre-diabetes, undiagnosed DM and known DM.
If HF treatment differs depending on DM status.
If DM treatment differs depending on HF status.
Methods
Study population
We identified all patients who had received a diagnosis of HF (10h revision of the International Classification of Disease and Related Health Problems codes I50.X, I42.0, I46.6, I42.7, I42.9, I11.0, I13.0, I13.2) at the Heart Centre or the Department of Internal medicine at Umeå University Hospital, Sweden, between January 2010 and December 2019 and who lived within the catchment area of the hospital. The diagnosis could be from either a HF hospitalisation event or outpatient visit and includes both prevalent cases and new onset HF during the inclusion period. The hospital is the only one within its geographic catchment area and serves a mixed urban and rural population of about 150 000 people and has the only specialist cardiology clinic in the area.
Data collection
The patients were identified in the hospitals electronic health records system (NCS Cross). Clinical parameters were extracted from the patients’ electronic health records using a prespecified protocol encompassing lab results, echocardiogram data, ECG results, comorbidities, medication and known diabetes status. DM was categorised as either Type 1 DM (T1DM) or T2DM, according to ICD 10 codes. Coronary artery disease was defined as either a previous myocardial infarction or the presence of at least one stenosis of 50% or more in a major coronary vessel. The latest available value for each parameter was used regardless of if it was from an inpatient or outpatient setting, including echocardiogram results. HF phenotype was categorised according to left ventricle EF (LVEF) as HF with reduced EF (HFrEF, EF ≤40), HF with mildly reduced ejection (HFmrEF, EF 41%–49%) or HFpEF (EF ≥50%).17 Lab, ECG and echocardiogram data were available regardless of which department had ordered them, including primary care. New York Heart Association (NYHA) class was not routinely stated in patient files and could not be assessed.
Diabetes and pre-diabetes classification
Data on HbA1c and fasting plasma-glucose (FPG) were recorded if there was a value from within 1 year from the otherwise latest available lab result. It was not possible to ascertain in every case if a plasma glucose measurement was fasting or not. Due to this uncertainty, we put more emphasis on HbA1c levels when classifying patients’ DM status. With that in mind, we categorised patients’ DM status using the following criteria, based on WHO recommendations18 19:
Known DM: Presence of a clinical diagnosis of DM.
Probable DM: HbA1c≥48 mmol/mol, or FPG≥7.0 mmol/L in the absence of a HbA1c-value.
Pre-diabetes:
Impaired fasting glucose (IFG) defined as FPG≥6.1 mmol/L if HbA1c <48 mmol/mol, or FPG 6.1–6.9 mmol/L in the absence of a HbA1c value.
Intermediate hyperglycaemia defined as HbA1c 42–47 mmol/mol.
Non-DM: No previous diagnosis of DM and not meeting criteria for probable DM or pre-diabetes.
Unknown: No previous diagnosis of DM and no glycaemic tests available.
Statistical analysis
Extracted data on patients who were alive 1 January 2020 were analysed using SPSS V.27. Continuous parameters are presented as means with SDs. Continuous variables without normal distribution are presented as medians with IQR. Categorical variables are presented with frequencies and percentage. When analysing differences between glycaemic profiles, each group was compared with the non-DM group using Pearson’s χ2 test. When analysing differences between HF phenotypes, a test of homogeneity was performed between all groups. If that was significant, pairwise comparisons between groups were made. The null hypothesis used was that there are no significant differences between groups. If a group had less than five patients, Fisher’s exact test was used instead. A p<0.05 was considered significant.
Results
A total of 2433 patients with a diagnosis of HF were alive 1 January 2020. Of these, 107 did not have an echocardiogram recorded and no EF available and were excluded. Of the 2326 patients with known EF, a majority (59%) were men, 617 (27%) had known DM, 26 with T1DM and 591 with T2DM. A total of 1084 patients (47%) had HFpEF, 561 patients (24%) had HFmrEF and 681 patients (29%) HFrEF. There were differences in many clinical characteristics between men and women. See table 1 for clinical characteristics for the cohort as a whole as well as for men and women, respectively.
Clinical characteristics
Of the 1709 patients without DM, 577 (34%) had both a HbA1c and an FPG recorded within the inclusion period, while 274 (16%) patients had an FPG but no HbA1c, and 247 (14%) patients had a HbA1c but no FPG recorded. 617 (36%) of the patients without previously known DM had neither an FPG nor HbA1c recorded. When applying the stated criteria for DM status on the 1092 patients without previous DM diagnosis with available FPG or HbA1c tests, 97 (9%) patients were deemed to have probable DM and 447 (41%) patients pre-diabetes (of which 216 IFG, 322 intermediate hyperglycaemia and 91 patients both). DM status did not differ significantly between HF phenotype (p=0.32, figure 1). Clinical characteristics of patients by DM status are available in online supplemental file 1.
Supplemental material
Distribution of patients according to heart failure classification and diabetes status. There were no statistically significant differences (p=0.32) in distribution of diabetes status between heart failure phenotypes. HFrEF, heart failure with reduced ejection fraction; HFmrEF, heart failure with mildly reduced ejection fraction; HFpEF, heart failure with preserved ejection fraction; known DM, patients with a diagnosis of diabetes mellitus; probable DM, patients with glycaemic test results at diabetic levels; non-DM, patients without a diagnosis of diabetes mellitus and normal glycaemic results; unknown, patients without available glycaemic data.
Non-DM patients were less often treated with beta blockers and loop diuretics compared with patients with pre-diabetes or known DM and tended to be less often treated with ACE inhibitors or angiotensin receptor blocker. Patients with probable DM had the highest rate of loop diuretic and MRA treatment. There were no differences between groups regarding treatment with angiotensin receptor blocker-neprilysin inhibitor (ARNI) or digitalis. Sodium-glucose cotransporter 2 (SGLT2)-inhibitor treatment only occurred in probable DM (1%) and known DM (12%) (table 2).
Heart failure treatment according to diabetes status
Of the 617 patients with known diabetes, 301 were treated with metformin, 245 with insulin, 106 with dipeptidyl peptidase four inhibitors, 75 with SGLT2 inhibitors, 47 with glucagon-like peptide one receptor analogues and 38 with sulphonylureas (including repaglinide). There were no differences in choice of DM treatment across HF phenotypes (table 3).
Diabetes treatment according to HF phenotype
Discussion
In this real-world HF population, one-fourth of patients had previously known DM. Of the patients without previous DM with available glycaemic tests, 1/10 had probable DM and 4/10 pre-diabetes. Diabetes and pre-diabetes prevalence and treatment did not differ between HF phenotypes. DM and pre-diabetes patients tended to have higher rates of diuretics and beta blockers prescribed.
There are three ways to diagnose pre-diabetes and diabetes—FPG, HbA1c and through an oral glucose tolerance test (OGTT).20 An OGTT is the most sensitive and specific test,21 22 but is not routinely used due to being more time- and resource intensive than the alternatives. FPG has the lowest sensitivity but good specificity. HbA1c has low specificity and sensitivity compared with an OGTT, but higher sensitivity than FPG.21 In this study, we have used both FPG and HbA1c-levels to characterise patients’ glycaemic status. The two measurements were available in roughly equal number of patients, but more patients were found to have pre-diabetes with HbA1c than with FPG, indicating that it is more sensitive than FPG in this population as well. Patients who were identified with each method were only partially overlapping. Previous prevalence estimates for pre-diabetes in HF patients have mainly been based on HbA1c levels alone10 11 or FPG alone,9 23 and have been performed in different settings. The work of Gotsman et al9 is the most similar to this study, but only used FPG to diagnose pre-diabetes. The proportion of known DM was higher than in our study (49.5% vs 27%), but the proportion of pre-diabetes was lower (12% vs 19%). One reason for that discrepancy could be that the prevalence of pre-diabetes is underestimated if only FPG or HbA1c is used. A generous approach to testing with FPG and HbA1c in HF patients would likely identify more cases of pre-diabetes, with the addition of an OGTT in selected patients.
Patients with known DM and pre-diabetes were more often treated with beta blockers and diuretics compared with non-DM patients. This might indicate that these patients have more symptomatic HF compared with non-DM patients, which would be in line with the worse prognosis seen in HF patients with DM.17 24
The usage of SGLT2-inhibitors was generally low, at only 3%. This reflects that the data used for this study covers the time up to December 2019, which was before SGLT2-inhibitors were approved for HF treatment without DM. If this study was repeated today, SGLT2 inhibitor treatment rates would likely be markedly higher.
Use of ARNI was low at only 6% of all patients, corresponding to 22% of the number of HFrEF patients, but did not differ between glycaemic groups. The prescription rate in our healthcare region was still above average compared with national rates,25 and in line with international findings26 27 that ARNI implementation is slow. There could be many reasons for low ARNI uptake, including clinical inertia, strict prescription criteria and patient factors. Old age and renal impairment have been implicated as reasons for low usage of ARNI and other guideline directed medical treatment,28 and our cohort has those factors compared with drug trial populations.15 16 There have been efforts to increase use of ARNI and other HF therapies with targeted programmes,29 including at our centre,30 but still only a minority of patients receive ARNI.
Our cohort was reasonably balanced in regard to sex, with 59% men and 41% women. However, women were significantly older with more HFpEF, less HFrEF and worse renal function, and less use of MRA, ARNI and SGLT2-inhibitors, which is commonly seen in HF cohorts.31 In this cohort, lower use of medications recommended for HFrEF could partly be explained by a lower rate of HFrEF, and lower use of SGLT2-inhibitors due a lower rate of DM, but women are often undertreated in HF studies.31 A previous study by Norberg et al32 in this cohort found that the main determinants for lower medication rate was older age, worse renal function and lower body weight, and not sex in and of itself.
While pre-diabetes is a risk factor for adverse outcomes in HF, the benefit of identifying and treating pre-diabetes in HF is unclear. Treatment of other comorbidities in HF patients, including DM, is considered important, and choice of DM therapy should be influenced by the presence of HF.17 20 Specifically, SGLT2-inhibitors are effective both in the treatment of DM and HF (regardless of LVEF or DM status), as well as chronic kidney disease,14 17 20 33 which are all conditions with high overlap. Considering the well-established effects and strong recommendation for SGLT2-inhibitors in HF,34 many patients with concomitant pre-diabetes and DM, both known and unknown, will receive SGLT2-inhibitor treatment as an added benefit. This extra glucose-lowering effect could necessitate adjustment in other glucose-lowering therapy. Due to high overlap with ischaemic heart disease and HF, identification of DM is important in HF patients so that treatment can be individualised with, for example, GLP1 receptor agonists in eligible patients.35 Since pre-diabetes and probable DM is common in HF patients, it is reasonable to have a generous approach to glycaemic testing in HF patients.
There is evidence for lifestyle and drug interventions in pre-diabetes to prevent progression to T2DM and to lower the risk of cardiovascular disease,36 37 but there is lacking evidence if the same interventions are effective against pre-diabetes in patients with established HF, and what the clinical benefit of those interventions are. This should be an avenue of further research.
Limitations
This study is limited by being a single-centre study based on data from patient records, so the results might not be generalisable to other populations or settings. The wide patient selection criteria used has both advantages and drawbacks. The intent and the advantage of this method is that as many HF patients as possible could be included so that results are applicable to the full span of HF patients encountered in the clinic. The drawbacks include that the patient cohort is heterogeneous with regard to HF type, aetiology, duration and clinical setting, which makes the results less applicable to specific HF subpopulations. Using retrospective data from patient files meant that we did not have complete or current data on all patients. Specifically, FPG and HbA1c were missing on many patients. Additionally, some recorded FPG values could be from non-fasting patients due to missing or erroneous labelling in patient files, which is why we relied more on HbA1c levels than FPG to classify a patient as having probable DM. Of patients without previous DM with available tests, half had some level of dysglycaemia. The rate in the whole population might be lower due to bias in which patients were tested, but our results indicate that blood glucose abnormalities are very common, despite limitations. Prevalence numbers are also affected by which criteria are used. The American Diabetes Association (ADA) recommend lower limits than the WHO,20 38 so more patients would have been classified as having pre-diabetes if we had used the ADA criteria.
NYHA class was another parameter that was rarely stated in patient files. This limits us from analysing our material regarding symptomatic severity and how it correlated to glycaemic class and choice of treatment.
Another limitation is that we used the latest available echocardiogram to classify patients’ HF phenotype. Patients with HF with recovered EF would therefore be classified as HFmrEF or HFpEF, instead of HFrEF as the European Society of Cardiology guidelines recommend.17
Conclusions
In this Northern European real-world HF population, the prevalence of known DM was 27%. Of patients without known DM with available FPG or HbA1c-measurements, 9% had probable DM and 41% pre-diabetes, corresponding to 4% and 19% of the entire cohort, respectively. The prevalence of DM, pre-diabetes or normoglycaemia or choice of DM treatment did not differ between HF phenotypes. Patients with pre-diabetes and DM were more often treated with diuretics and beta blockers than normoglycaemic patients.
Data availability statement
Data are available on reasonable request. Request should be made to the corresponding author. Data is available in the form of deidentified spreadsheets.
Ethics statements
Patient consent for publication
Ethics approval
This study complies with the Declaration of Helsinki. The Regional Ethical Review Board in Umeå, Sweden, has approved this study, ethics approval IDs 2015/419-31, 2016-233-32M and 2020-01588.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors KL conceived of and planned the study. KL, EH and HN acquired data for the study. EH drafted the manuscript. All credited authors participated in analysis, interpretation, and critical revision of the manuscript, and approved of its publication. All credited authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. KL is the guarantor for the study.
Funding This article has been supported by AstraZeneca
Competing interests KL has received lecturer grants from Orion Pharma, Novartis and AstraZeneca. Helena Norberg has received lecturer grants from Pfizer. EH has received lecturer grants from Orion Pharma. MB and SS report no conflicts of interest.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.