Thomas Fritz Hansen

Using electronic patient records to discover disease correlations and stratify patient cohorts

Electronic patient records remain a rather unexplored, but potentially rich data source for discovering correlations between diseases. We describe a general approach for gathering phenotypic descriptions of patients from medical records in a systematic and non-cohort dependent manner. By extracting phenotype information from the free-text in such records we demonstrate that we can extend the information contained in the structured record data, and use it for producing fine-grained patient stratification and disease co-occurrence statistics. The approach uses a dictionary based on the International Classification of Disease ontology and is therefore in principle language independent. As a use case we show how records from a Danish psychiatric hospital lead to the identification of disease correlations, which subsequently can be mapped to systems biology frameworks.

Prevalence of infective endocarditis in patients with Staphylococcus aureus bacteraemia: the value of screening with echocardiography

AIMSnStaphylococcus aureus infective endocarditis (IE) is a critical medical condition associated with a high morbidity and mortality. In the present study, we prospectively evaluated the importance of screening with echocardiography in an unselected S. aureus bacteraemia (SAB) population.nnnMETHODS AND RESULTSnFrom 1 January 2009 to 31 August 2010, a total of 244 patients with SAB at six Danish hospitals underwent screening echocardiography. The inclusion rate was 73% of all eligible patients (n= 336), and 53 of the 244 included patients (22%; 95% CI: 17-27%) were diagnosed with definite IE. In patients with native heart valves the prevalence was 19% (95% CI: 14-25%) compared with 38% (95% CI: 20-55%) in patients with prosthetic heart valves and/or cardiac rhythm management devices (P= 0.02). No difference was found between Main Regional Hospitals and Tertiary Cardiac Hospitals, 20 vs. 23%, respectively (NS). The prevalence of IE in high-risk patients with one or more predisposing condition or clinical evidence of IE were significantly higher compared with low-risk patients with no additional risk factors (38 vs. 5%; P < 0.001). IE was associated with a higher 6 months mortality, 14(26%) vs. 28(15%) in SAB patients without IE, respectively (P < 0.05).nnnCONCLUSIONnSAB patients carry a high risk for development of IE, which is associated with a worse prognosis compared with uncomplicated SAB. The presenting symptoms and clinical findings associated with IE are often non-specific and echocardiography should always be considered as part of the initial evaluation of SAB patients.

Long-Term Outcome of Percutaneous Transluminal Septal Myocardial Ablation in Hypertrophic Obstructive Cardiomyopathy A Scandinavian Multicenter Study

Background— Single-center reports on percutaneous transluminal septal myocardial ablation (PTSMA) in patients with hypertrophic obstructive cardiomyopathy have shown considerable differences in outcome. Methods and Results— We report the long-term outcome of 313 PTSMA procedures performed in 279 patients with hypertrophic obstructive cardiomyopathy aged 59±14 years from 1999 to 2010 in 4 Scandinavian centers. Sixty-nine percent of patients had ≥1 comorbidity at baseline. The median (interquartile range) of left ventricular outflow tract gradient at rest was reduced from 58 mm Hg (34 to 89 mm Hg) at baseline to 12 mm Hg (8 to 24 mm Hg) at 1-year (P<0.001) and during Valsalva maneuver from 93 mm Hg (70 to 140 mm Hg) to 21 mm Hg (11 to 42 mm Hg) (P<0.001). The proportion of patients with syncope was reduced from 18% to 2% (P<0.001), and the proportion in New York Heart Association class III/IV was reduced from 94% to 21% (P<0.001). All treatment effects remained stable during the follow-up. New York Heart Association class III/IV at the most recent follow-up (2.9±2.6 years) was associated with diabetes mellitus (P=0.03), chronic obstructive pulmonary disease (P=0.02), and valve disease unrelated to hypertrophic cardiomyopathy (P<0.01). In-hospital mortality was 0.3%. The 1-, 5- and 10-year survival rates were 97%, 87%, and 67%, respectively (P=0.06 versus an age- and sex-matched background population) in all patients and 99%, 94%, and 88%, respectively (P=0.12) in patients aged <60 years (48±9 years, n=141). Age (hazard ratio, 1.07; 95% CI, 1.03 to 1.1) was the only predictor of survival. Conclusions— In this multicenter study, the in-hospital mortality after PTSMA was low despite considerable comorbidities. The hemodynamic and symptomatic effects were sustained long term. The long-term symptomatic outcome was associated with baseline comorbidities. The 10-year survival rate was comparable to that in an age- and sex-matched background population, and age was the only predictor of survival.

Left bundle-branch block: the relationship between electrocardiogram electrical activation and echocardiography mechanical contraction

BACKGROUNDnThe relationship between myocardial electrical activation by electrocardiogram (ECG) and mechanical contraction by echocardiography in left bundle-branch block (LBBB) has never been clearly demonstrated. New strict criteria for LBBB based on a fundamental understanding of physiology have recently been independently published for both ECG and echocardiography. The relationship between the 2 modalities and the relation to cardiac resynchronization therapy (CRT) response was investigated.nnnMETHODSnSixty-six patients with LBBB by conventional criteria had a standard 12-lead ECG and 2-dimensional strain echocardiography performed before CRT implantation. Criteria for LBBB by echocardiography included early termination of contraction in one wall and prestretch and late contraction in opposing wall(s). New strict criteria by ECG included QRS duration ≥140 ms (men) or 130 ms (women), QS or rS in leads V1 and V2, and mid-QRS notching or slurring in ≥2 of leads V1, V2, V5, V6, I, and aVL. Response was defined as >15% decrease in left ventricular end-systolic volume after 6 months.nnnRESULTSnIn 64 of 66 patients, ECG analysis was possible. Echo and ECG readings for LBBB presence were concordant in 54 (84%) of 64. Thirty-seven (82%) of 45 patients with LBBB by strict ECG criteria responded to CRT, whereas only 4 (21%) of the 19 patients without LBBB responded (sensitivity 90% and specificity 65%). Thirty-six (95%) of 38 patients with concordance for the presence of LBBB responded to CRT. In patients with concordance for the absence of LBBB, 15 (94%) of 16 did not respond.nnnCONCLUSIONnFor the first time, a close relation has been demonstrated between electrical activation by ECG and mechanical contraction by echocardiography. These findings may help identify CRT candidates.

Long-Term Outcome of Percutaneous Transluminal Septal Myocardial Ablation in Hypertrophic Obstructive Cardiomyopathy

Background— Single-center reports on percutaneous transluminal septal myocardial ablation (PTSMA) in patients with hypertrophic obstructive cardiomyopathy have shown considerable differences in outcome. Methods and Results— We report the long-term outcome of 313 PTSMA procedures performed in 279 patients with hypertrophic obstructive cardiomyopathy aged 59±14 years from 1999 to 2010 in 4 Scandinavian centers. Sixty-nine percent of patients had ≥1 comorbidity at baseline. The median (interquartile range) of left ventricular outflow tract gradient at rest was reduced from 58 mm Hg (34 to 89 mm Hg) at baseline to 12 mm Hg (8 to 24 mm Hg) at 1-year (P<0.001) and during Valsalva maneuver from 93 mm Hg (70 to 140 mm Hg) to 21 mm Hg (11 to 42 mm Hg) (P<0.001). The proportion of patients with syncope was reduced from 18% to 2% (P<0.001), and the proportion in New York Heart Association class III/IV was reduced from 94% to 21% (P<0.001). All treatment effects remained stable during the follow-up. New York Heart Association class III/IV at the most recent follow-up (2.9±2.6 years) was associated with diabetes mellitus (P=0.03), chronic obstructive pulmonary disease (P=0.02), and valve disease unrelated to hypertrophic cardiomyopathy (P<0.01). In-hospital mortality was 0.3%. The 1-, 5- and 10-year survival rates were 97%, 87%, and 67%, respectively (P=0.06 versus an age- and sex-matched background population) in all patients and 99%, 94%, and 88%, respectively (P=0.12) in patients aged <60 years (48±9 years, n=141). Age (hazard ratio, 1.07; 95% CI, 1.03 to 1.1) was the only predictor of survival. Conclusions— In this multicenter study, the in-hospital mortality after PTSMA was low despite considerable comorbidities. The hemodynamic and symptomatic effects were sustained long term. The long-term symptomatic outcome was associated with baseline comorbidities. The 10-year survival rate was comparable to that in an age- and sex-matched background population, and age was the only predictor of survival.

Identification of Typical Left Bundle Branch Block Contraction by Strain Echocardiography Is Additive to Electrocardiography in Prediction of Long-Term Outcome After Cardiac Resynchronization Therapy

BACKGROUNDnCurrent guidelines suggest that patients with left bundle branch block (LBBB) be treated with cardiac resynchronization therapy (CRT); however, one-third do not have a significant activation delay, which can result in nonresponse. By identifying characteristic opposing wall contraction, 2-dimensional strain echocardiography (2DSE) may detect true LBBB activation.nnnOBJECTIVESnThis study sought to investigate whether the absence of a typical LBBB mechanical activation pattern by 2DSE was associated with unfavorable long-term outcome and if this is additive to electrocardiographic (ECG) morphology and duration.nnnMETHODSnFrom 2 centers, 208 CRT candidates (New York Heart Association classes II to IV, ejection fraction ≤35%, QRS duration ≥120 ms) with LBBB by ECG were prospectively included. Before CRT implantation, longitudinal strain in the apical 4-chamber view determined whether typical LBBB contraction was present. The pre-defined outcome was freedom from death, left ventricular assist device, or heart transplantation over 4 years.nnnRESULTSnTwo-thirds of patients (63%) had a typical LBBB contraction pattern. During 4 years, 48 patients (23%) reached the primary endpoint. Absence of a typical LBBB contraction was independently associated with increased risk of adverse outcome after adjustment for ischemic heart disease and QRS width (hazard ratio [HR]: 3.1; 95% CI: 1.64 to 5.88; p < 0.005). Adding pattern assessment to a risk prediction model including QRS duration and ischemic heart disease significantly improved the net reclassification index to 0.14 (p = 0.04) and improved the C-statistics (0.63 [95% CI: 0.54 to 0.72] vs. 0.71 [95% CI: 0.63 to 0.80]; p = 0.02). Use of strict LBBB ECG criteria was not independently associated with outcome in the multivariate model (HR: 1.72; 95% CI: 0.89 to 3.33; p = 0.11. Assessment of LBBB contraction pattern was superior to time-to-peak indexes of dyssynchrony (p < 0.01 for all).nnnCONCLUSIONSnContraction pattern assessment to identify true LBBB activation provided important prognostic information in CRT candidates.

Global longitudinal strain is not impaired in type 1 diabetes patients without albuminuria: the Thousand & 1 study

OBJECTIVESnThe purpose of this study was to investigate if systolic myocardial function is reduced in all patients with type 1 diabetes (T1DM) or only in patients with albuminuria.nnnBACKGROUNDnHeart failure is a common cause of mortality in T1DM, and a specific diabetic cardiomyopathy has been suggested. It is not known whether myocardial dysfunction is a feature of T1DM per se or primarily associated with diabetes with albuminuria.nnnMETHODSnThis cross-sectional study compared 1,065 T1DM patients without known heart disease from the outpatient clinic at the Steno Diabetes Center with 198 healthy control subjects. Conventional echocardiography and global longitudinal strain (GLS) by 2-dimensional speckle-tracking echocardiography was performed and analyzed in relation toxa0normoalbuminuria (nxa0= 739), microalbuminuria (nxa0= 223), and macroalbuminuria (nxa0= 103). Data were analyzed in univariable and multivariable linear regression models adjusted for confounding factors including conventional risk factors, medication, and systolic and diastolic dysfunction. Investigators were blinded to degree of albuminuria.nnnRESULTSnMean age was 49.5 years, 52% men, mean glycated hemoglobin 8.2% (66 mmol/mol), mean body mass index 25.5 kg/m(2), and mean diabetes duration 26.1 years. In unadjusted analyses, GLS differed significantly between T1DM patients and control subjects (pxa0= 0.02). When stratified by degrees of albuminuria, the difference in GLS compared with control subjects wasxa0-18.8 ± 2.5% versusxa0-18.5 ± 2.5% for normoalbuminuria (pxa0= 0.28), versusxa0-17.9 ± 2.7% for microalbuminuria (pxa0= 0.001), and versusxa0-17.4 ± 2.9% for macroalbuminuria (pxa0< 0.001). Multivariable analyses, including clinical characteristics, diastolic and systolic dysfunction, and use of medication, did not change this relationship.nnnCONCLUSIONSnSystolic function assessed by GLS was reduced in T1DM compared with control subjects. This difference, however, was driven solely by decreased GLS in T1DM patients with albuminuria. T1DM patients with normoalbuminuria have systolic myocardial function similar to healthy control subjects. These findings do not support the presence of specific diabetic cardiomyopathy without albuminuria.

Prevalence of systolic and diastolic dysfunction in patients with type 1 diabetes without known heart disease: the Thousand & 1 Study

Aims/hypothesisHeart failure is one of the leading causes of mortality in type 1 diabetes. Early identification is vitally important. We sought to determine the prevalence and clinical characteristics associated with subclinical impaired systolic and diastolic function in type 1 diabetes patients without known heart disease.MethodsIn this cross-sectional examination of 1,093 type 1 diabetes patients without known heart disease, randomly selected from the Steno Diabetes Center, complete clinical and echocardiographic examinations were performed and analysed in uni- and multivariable regression models.ResultsThe mean (SD) age was 49.6 (15)u2009years, 53% of participants were men, and the mean duration of diabetes was 25.5 (15)u2009years. Overall, 15.5% (nu2009=u2009169) of participants had grossly abnormal systolic or diastolic function, including 1.7% with left ventricular ejection fraction (LVEF)u2009<u200945% and 14.4% with evidence of long-standing diastolic dysfunction. In univariable models, clinical characteristics associated with abnormal myocardial function were: age (per 10xa0years), OR (95% CI) 2.1 (1.8, 2.4); diabetes duration (per 10xa0years), 1.7 (1.4, 1.9); systolic BPu2009≥u2009140xa0mmHg, 2.7 (1.9, 3.8); diastolic BPu2009≥u200990xa0mmHg, 1.8 (1.0, 3.1); estimated (e)GFRu2009<u200960xa0ml min−1 1.73xa0m−2, 3.8 (2.5, 5.9); microalbuminuria, 2.0 (1.3, 3.0); macroalbuminuria, 5.9 (3.8, 9.3); proliferative retinopathy, 3.6 (2.3, 5.8); blindness, 10.1 (3.2, 31.6); and peripheral neuropathy, 3.8 (2.7, 5.3). In multivariable models only age (2.1 [1.7, 2.5]), female sex, (1.9 [1.2, 2.8]) and macroalbuminuria (5.2 [2.9, 10.3]) remained significantly associated with subclinical grossly abnormal myocardial function.Conclusions/interpretationSubclinical myocardial dysfunction is a common finding in type 1 diabetes patients without known heart disease. Type 1 diabetes patients with albuminuria are at greatly increased risk of having subclinical abnormal myocardial function compared with patients without albuminuria. Echocardiography may be particularly warranted in patients with albuminuria.

Metformin is associated with improved left ventricular diastolic function measured by tissue Doppler imaging in patients with diabetes

OBJECTIVEnTo examine the association between selected glucose-lowering medications and left ventricular (LV) diastolic function in patients with diabetes.nnnDESIGNnRetrospective cohort study (years 2005-2008).nnnMETHODSnEchocardiograms of 242 patients with diabetes undergoing coronary angiography were analyzed. All patients had an LV ejection fraction (LVEF) ≥20% and were without atrial fibrillation, bundle branch block, valvular disease, or cardiac pacemaker. Patients were grouped according to the use of metformin (n=56), sulfonylureas (n=43), insulin (n=61), and combination treatment (n=82).nnnRESULTSnMean age (66±10 years) and mean LVEF (45±11%) were similar across the groups. Mean isovolumic relaxation time (IVRT) was 66±31, 79±42, 69±23, and 66±29 ms in metformin, sulfonylureas, insulin, and combination treatment groups respectively (P=0.4). Mean early diastolic longitudinal tissue velocity (e) was 5.3±1.6, 4.6±1.6, 5.3±1.8, and 5.4±1.7 cm/s in metformin, sulfonylureas, insulin, and combination treatment groups (P=0.04). In adjusted linear regression models, the use of metformin was associated with a shorter IVRT (parameter estimate -9.9 ms, P=0.049) and higher e (parameter estimate +0.52 cm/s, P=0.03), compared with no use of metformin. The effects of metformin were not altered by concomitant use of sulfonylureas or insulin (P for interactions >0.4).nnnCONCLUSIONSnThe use of metformin is associated with improved LV relaxation, as compared with no use of metformin.

Comparison of Dyssynchrony Parameters for VV-Optimization in CRT Patients

Optimization of the interventricular delay (VV‐optimization) in cardiac resynchronization therapy (CRT) patients can be performed by evaluation of mechanical dyssynchrony. However, there is no consensus on which method to use. In this study, three conceptually different methods were evaluated.