European Journal of Heart Failure (2015) 17, 187–195 doi:10.1002/ejhf.216
Efficiently screening heart failure in patients with type 2 diabetes Leandra J.M. Boonman-de Winter1,2,*, Frans H. Rutten1, Maarten J. Cramer3, Marcel J. Landman4, Nicolaas P.A. Zuithoff1, Anho H. Liem5, and Arno W. Hoes1 1 Julius
Center for Health Sciences and Primary Care, University Medical Center Utrecht, The Netherlands; 2 Center for Diagnostic Support in Primary Care (SHL-Groep), Department of Scientific Research, Etten-Leur, The Netherlands; 3 Department of Cardiology, Division of Heart and Lungs, University Medical Center Utrecht, The Netherlands; 4 Cardiologist, Leusden, The Netherlands; and 5 Department of Cardiology, Sint Franciscus Hospital, Rotterdam, The Netherlands Received 16 September 2014; revised 30 October 2014; accepted 7 November 2014 ; online publish-ahead-of-print 30 December 2014
Aims
Our aim was to develop a screening tool for heart failure in patients with type 2 diabetes. ..................................................................................................................................................................... Methods A total of 581 consecutive patients from 21 primary care practices in The Netherlands with type 2 diabetes, in whom and results the diagnosis of heart failure (HF) was not known, underwent an extensive diagnostic assessment, including medical history taking, physical examination, ECG, and echocardiography. The presence or absence of HF was established by a panel of two cardiologists and one general practitioner following the guidelines on HF of the European Society of Cardiology. In 161 patients, HF was considered present. A model based on the medical history and symptoms had a good discriminative value for detecting or excluding HF [C-statistic after bootstrapping 0.80; 95% confidence interval (CI) 0.76–0.83]. Adding signs improved the C-statistic to 0.82 (95% CI 0.79–0.86). A diagnostic screening score based on the clinical model had good discriminative properties applying a cut-off of 3 points (24.7% risk of HF) with sensitivity 70.8%, specificity 79.0%, negative predictive value 87.6%, and positive predictive value 56.4%. ECG and natriuretic peptides both had independent added value beyond the clinical model and increased the C-statistic to 0.86 (95% CI 0.83– 0.89). With a 20% threshold, the net reclassification of adding ECG and NT-proBNP to the clinical model was only 0.06. ..................................................................................................................................................................... Conclusions A decision aid based on items from the clinical assessment is useful for screening HF in older patients with type 2 diabetes and to pre-select for echocardiography. ..................................................................................................................................................................... Trial NL2271704108.
registration:
.......................................................................................................... Heart failure • specificity
Type 2 diabetes •
Screening •
Introduction Type 2 diabetes (T2DM) enhances the risk of cardiovascular disease. The increased risk of ischaemic heart disease is well known. Previous studies showed that heart failure (HF) is also common in patients with T2DM.1 – 5 A recent screening study including echocardiographic tissue Doppler measurements showed that the majority of uncovered cases of HF had a preserved EF.5 Echocardiography is the key investigation to establish the diagnosis of HF,6 but in many countries accessibility to this
..............................
Keywords
Diagnosis •
Primary care •
Sensitivity and
diagnostic facility is limited to primary care where most cases may be detected. Assessment of symptoms and signs followed by measurements of natriuretic peptides and ECG has been shown to be useful to select cases for echocardiography and increases the likelihood of detecting unsuspected HF.7,8 Available diagnostic models for detecting non-acute HF in suspected patients have some clinical variables in common: (i) a history of ischaemic heart disease, especially prior myocardial infarction: (ii) symptoms and/or signs of fluid retention; (iii) a laterally displaced or broadened and sustained apical beat; and (iv) male sex.9 – 14
*Corresponding author. Center for Diagnostic Support in Primary Care (SHL-Groep), Department of Scientific Research, Bredaseweg 165, 4872 LA Etten-Leur, The Netherlands. Tel: +31 76 5029192, Fax +31 76 5029191, E-mail: [email protected]
© 2014 The Authors European Journal of Heart Failure © 2014 European Society of Cardiology
188
Methods Study population Twenty-one primary care practices from the South-West of The Netherlands agreed to participate in this study conducted between February 2009 and March 2010. All patients with T2DM who were enlisted in these practices were eligible, including those who received co-operative care by a medical specialist. Those already known to have a cardiologist’s diagnosis of HF were excluded. The study protocol and data on the prevalence of HF were published previously.5,15
Diagnostic investigations All 581 patients underwent a single 1.5 h standardized diagnostic assessment at the cardiology outpatient department of the Admiraal de Ruyter hospital in Goes, including history taking, physical examination, ECG, and echocardiography. A standardized questionnaire was used to assess the medical history, symptoms, duration of diabetes, smoking history, and medication use. Patients were asked to bring their medication boxes to assess current medication use.16 Physical examination was performed in a standardized manner and included a single blood pressure measurement in the supine position, measurement of the jugular venous pressure, and palpation of the apex beat in both the supine and lateral decubital position. A standard 12-lead ECG was recorded and classified by an experienced cardiologist (A.H.L.) according to the Minnesota coding criteria.16 The ECG results were dichotomized into abnormal and normal. The following ECG findings were considered as abnormal: ST and/or T wave abnormalities, abnormal Q waves, tachycardia (heart rate >100 b.p.m.) or bradycardia (heart rate 95 g/m2 in women and >115 g/m2 in men.6,19 The LVEF was calculated using Simpson’s rule (disc summation method)20 or the single plane area method21 in 522 patients (89.8%). In 46 patients (7.9%), adequate tracing was not possible, and LVEF was estimated by the 2D visual estimate method (‘eyeballing’).22 In 13 patients (2.2%), the image quality did not allow any estimation of the LVEF. An LVEF ≤45% was considered as reduced. Patients with an E/e’ ≥15 were classified as having diastolic dysfunction.23 When the E/e’ was between 8 and 15, a combination of the indexed volume of the left atrium, the mitral inflow (E deceleration time and E/A), and pulmonary venous flow (S/D) was used to classify the presence or absence of diastolic dysfunction. We also considered diastolic dysfunction present in patients with AF who had enlarged left atrial (LA) volumes and in patients with echocardiographic LVH combined with either enlarged LA volume, abnormal mitral inflow, or pulmonary venous flow profile.6 The LA volume was indexed for body surface area.24 As cut-off values for a definitely increased LA volume index, we used 34 mL/m2 .25 Echocardiograms and ECGs were interpreted blinded to the clinical data.
Definition of heart failure The diagnosis of HF was based on the recommendations of the HF guidelines of the European Society of Cardiology.6 Our expert panel, consisting of two cardiologists and one experienced general practitioner with specific expertise in HF, used all available diagnostic information, except the NT-proBNP values, to decide on the presence or absence of HF. The reproducibility of such a consensus method has been shown to be high.26 Heart failure was classified into heart failure with reduced ejection fraction (HF-REF), heart failure with preserved ejection fraction (HF-PEF), or ‘isolated’ right-sided HF. For HF-REF, patients had to have an LVEF ≤45%. For HF-PEF, patients had to have echocardiographic evidence of diastolic dysfunction, and/or LA enlargement or LVH, and an LVEF >45%.23 ‘Isolated’ right ventricular failure was classified if the calculated systolic pulmonary artery pressure (sPAP) was >50 mmHg in the presence of a normal LVEF, and in the absence of substantial diastolic dysfunction.6 In order to classify a patient as a HF patient, the echocardiographic abnormalities listed above had to be accompanied by indicative symptoms (e.g. breathlessness, ankle swelling, and fatigue) and signs (e.g. elevated jugular venous pressure, pulmonary crepitations, and displaced apex beat) of HF.27 An exception was made for patients on diuretics; then, ‘typical’ signs were not obligatory for the diagnosis. The same expert panel re-assessed a random sample of 63 (10.7%) participants, blinded to the first classification. In five cases, the diagnosis of the presence or absence of HF did not correspond to the first classification (Cohen’s kappa 0.82, SD 0.08). Three cases of HF-PEF were re-classified as diastolic dysfunction. One individual initially classified as having diastolic dysfunction and one classified as normal were reclassified as having HF-PEF. There was no disagreement with the re-assessment in those classified initially as having LV systolic dysfunction and HF-REF.
Model derivation and development We evaluated variables known from the literature to predict the presence of HF.10,11,28 Additionally, variables from our own cohort with a univariable P-value 75 years Male sex Median duration of diabetes in years Current smoker Medical history Ischaemic heart disease Prior myocardial infarction Angina pectoris PCI/CABG Atrial fibrillation Hypertension TIA or stroke Asthma or COPD Medication Diuretics Mineralocorticoid antagonists ACE inhibitors or ARBs Beta-blockers Lipid-lowering agents Oral antidiabetics Insulin
71.6 32.5 53.4 5.5 13.8
74.5 49.7 47.8 6.3 14.9
70.5 26.0 55.5 5.1 13.3
1.08 (1.05–1.11) 2.82 (1.93–4.11) 0.74 (0.51–1.06) 1.02 (0.99–1.05) 1.14 (0.68–1.91)
Efficiently screening heart failure in patients with type 2 diabetes Leandra J.M. Boonman-de Winter1,2,*, Frans H. Rutten1, Maarten J. Cramer3, Marcel J. Landman4, Nicolaas P.A. Zuithoff1, Anho H. Liem5, and Arno W. Hoes1 1 Julius
Center for Health Sciences and Primary Care, University Medical Center Utrecht, The Netherlands; 2 Center for Diagnostic Support in Primary Care (SHL-Groep), Department of Scientific Research, Etten-Leur, The Netherlands; 3 Department of Cardiology, Division of Heart and Lungs, University Medical Center Utrecht, The Netherlands; 4 Cardiologist, Leusden, The Netherlands; and 5 Department of Cardiology, Sint Franciscus Hospital, Rotterdam, The Netherlands Received 16 September 2014; revised 30 October 2014; accepted 7 November 2014 ; online publish-ahead-of-print 30 December 2014
Aims
Our aim was to develop a screening tool for heart failure in patients with type 2 diabetes. ..................................................................................................................................................................... Methods A total of 581 consecutive patients from 21 primary care practices in The Netherlands with type 2 diabetes, in whom and results the diagnosis of heart failure (HF) was not known, underwent an extensive diagnostic assessment, including medical history taking, physical examination, ECG, and echocardiography. The presence or absence of HF was established by a panel of two cardiologists and one general practitioner following the guidelines on HF of the European Society of Cardiology. In 161 patients, HF was considered present. A model based on the medical history and symptoms had a good discriminative value for detecting or excluding HF [C-statistic after bootstrapping 0.80; 95% confidence interval (CI) 0.76–0.83]. Adding signs improved the C-statistic to 0.82 (95% CI 0.79–0.86). A diagnostic screening score based on the clinical model had good discriminative properties applying a cut-off of 3 points (24.7% risk of HF) with sensitivity 70.8%, specificity 79.0%, negative predictive value 87.6%, and positive predictive value 56.4%. ECG and natriuretic peptides both had independent added value beyond the clinical model and increased the C-statistic to 0.86 (95% CI 0.83– 0.89). With a 20% threshold, the net reclassification of adding ECG and NT-proBNP to the clinical model was only 0.06. ..................................................................................................................................................................... Conclusions A decision aid based on items from the clinical assessment is useful for screening HF in older patients with type 2 diabetes and to pre-select for echocardiography. ..................................................................................................................................................................... Trial NL2271704108.
registration:
.......................................................................................................... Heart failure • specificity
Type 2 diabetes •
Screening •
Introduction Type 2 diabetes (T2DM) enhances the risk of cardiovascular disease. The increased risk of ischaemic heart disease is well known. Previous studies showed that heart failure (HF) is also common in patients with T2DM.1 – 5 A recent screening study including echocardiographic tissue Doppler measurements showed that the majority of uncovered cases of HF had a preserved EF.5 Echocardiography is the key investigation to establish the diagnosis of HF,6 but in many countries accessibility to this
..............................
Keywords
Diagnosis •
Primary care •
Sensitivity and
diagnostic facility is limited to primary care where most cases may be detected. Assessment of symptoms and signs followed by measurements of natriuretic peptides and ECG has been shown to be useful to select cases for echocardiography and increases the likelihood of detecting unsuspected HF.7,8 Available diagnostic models for detecting non-acute HF in suspected patients have some clinical variables in common: (i) a history of ischaemic heart disease, especially prior myocardial infarction: (ii) symptoms and/or signs of fluid retention; (iii) a laterally displaced or broadened and sustained apical beat; and (iv) male sex.9 – 14
*Corresponding author. Center for Diagnostic Support in Primary Care (SHL-Groep), Department of Scientific Research, Bredaseweg 165, 4872 LA Etten-Leur, The Netherlands. Tel: +31 76 5029192, Fax +31 76 5029191, E-mail: [email protected]
© 2014 The Authors European Journal of Heart Failure © 2014 European Society of Cardiology
188
Methods Study population Twenty-one primary care practices from the South-West of The Netherlands agreed to participate in this study conducted between February 2009 and March 2010. All patients with T2DM who were enlisted in these practices were eligible, including those who received co-operative care by a medical specialist. Those already known to have a cardiologist’s diagnosis of HF were excluded. The study protocol and data on the prevalence of HF were published previously.5,15
Diagnostic investigations All 581 patients underwent a single 1.5 h standardized diagnostic assessment at the cardiology outpatient department of the Admiraal de Ruyter hospital in Goes, including history taking, physical examination, ECG, and echocardiography. A standardized questionnaire was used to assess the medical history, symptoms, duration of diabetes, smoking history, and medication use. Patients were asked to bring their medication boxes to assess current medication use.16 Physical examination was performed in a standardized manner and included a single blood pressure measurement in the supine position, measurement of the jugular venous pressure, and palpation of the apex beat in both the supine and lateral decubital position. A standard 12-lead ECG was recorded and classified by an experienced cardiologist (A.H.L.) according to the Minnesota coding criteria.16 The ECG results were dichotomized into abnormal and normal. The following ECG findings were considered as abnormal: ST and/or T wave abnormalities, abnormal Q waves, tachycardia (heart rate >100 b.p.m.) or bradycardia (heart rate 95 g/m2 in women and >115 g/m2 in men.6,19 The LVEF was calculated using Simpson’s rule (disc summation method)20 or the single plane area method21 in 522 patients (89.8%). In 46 patients (7.9%), adequate tracing was not possible, and LVEF was estimated by the 2D visual estimate method (‘eyeballing’).22 In 13 patients (2.2%), the image quality did not allow any estimation of the LVEF. An LVEF ≤45% was considered as reduced. Patients with an E/e’ ≥15 were classified as having diastolic dysfunction.23 When the E/e’ was between 8 and 15, a combination of the indexed volume of the left atrium, the mitral inflow (E deceleration time and E/A), and pulmonary venous flow (S/D) was used to classify the presence or absence of diastolic dysfunction. We also considered diastolic dysfunction present in patients with AF who had enlarged left atrial (LA) volumes and in patients with echocardiographic LVH combined with either enlarged LA volume, abnormal mitral inflow, or pulmonary venous flow profile.6 The LA volume was indexed for body surface area.24 As cut-off values for a definitely increased LA volume index, we used 34 mL/m2 .25 Echocardiograms and ECGs were interpreted blinded to the clinical data.
Definition of heart failure The diagnosis of HF was based on the recommendations of the HF guidelines of the European Society of Cardiology.6 Our expert panel, consisting of two cardiologists and one experienced general practitioner with specific expertise in HF, used all available diagnostic information, except the NT-proBNP values, to decide on the presence or absence of HF. The reproducibility of such a consensus method has been shown to be high.26 Heart failure was classified into heart failure with reduced ejection fraction (HF-REF), heart failure with preserved ejection fraction (HF-PEF), or ‘isolated’ right-sided HF. For HF-REF, patients had to have an LVEF ≤45%. For HF-PEF, patients had to have echocardiographic evidence of diastolic dysfunction, and/or LA enlargement or LVH, and an LVEF >45%.23 ‘Isolated’ right ventricular failure was classified if the calculated systolic pulmonary artery pressure (sPAP) was >50 mmHg in the presence of a normal LVEF, and in the absence of substantial diastolic dysfunction.6 In order to classify a patient as a HF patient, the echocardiographic abnormalities listed above had to be accompanied by indicative symptoms (e.g. breathlessness, ankle swelling, and fatigue) and signs (e.g. elevated jugular venous pressure, pulmonary crepitations, and displaced apex beat) of HF.27 An exception was made for patients on diuretics; then, ‘typical’ signs were not obligatory for the diagnosis. The same expert panel re-assessed a random sample of 63 (10.7%) participants, blinded to the first classification. In five cases, the diagnosis of the presence or absence of HF did not correspond to the first classification (Cohen’s kappa 0.82, SD 0.08). Three cases of HF-PEF were re-classified as diastolic dysfunction. One individual initially classified as having diastolic dysfunction and one classified as normal were reclassified as having HF-PEF. There was no disagreement with the re-assessment in those classified initially as having LV systolic dysfunction and HF-REF.
Model derivation and development We evaluated variables known from the literature to predict the presence of HF.10,11,28 Additionally, variables from our own cohort with a univariable P-value 75 years Male sex Median duration of diabetes in years Current smoker Medical history Ischaemic heart disease Prior myocardial infarction Angina pectoris PCI/CABG Atrial fibrillation Hypertension TIA or stroke Asthma or COPD Medication Diuretics Mineralocorticoid antagonists ACE inhibitors or ARBs Beta-blockers Lipid-lowering agents Oral antidiabetics Insulin
71.6 32.5 53.4 5.5 13.8
74.5 49.7 47.8 6.3 14.9
70.5 26.0 55.5 5.1 13.3
1.08 (1.05–1.11) 2.82 (1.93–4.11) 0.74 (0.51–1.06) 1.02 (0.99–1.05) 1.14 (0.68–1.91)
