PRESCRIPTION OF REMOTE PATIENT MANAGEMENT BASED ON BIOMARKERS
20220260589 · 2022-08-18
Assignee
Inventors
- Tobias Stubbe (Hennigsdorf, DE)
- Jan Wiemer (Hennigsdorf, DE)
- Jan Kunde (Hennigsdorf, DE)
- Stefan Gehrig (Hennigsdorf, DE)
- Friedrich Köhler (Berlin, DE)
Cpc classification
G01N33/74
PHYSICS
G01N2800/325
PHYSICS
G01N2333/58
PHYSICS
International classification
Abstract
The invention relates to a method for determining whether a subject diagnosed with a cardiovascular disease should be prescribed a remote patient management, the method comprising measuring particular biomarkers in a sample from said patient. The invention therefore relates to a method for therapy guidance, stratification and/or monitoring of a remote patient management for a patient diagnosed with a cardiovascular disease, comprising providing at least one sample of a patient, determining a level of at least one biomarker selected from the group consisting of proADM, proBNP and proANP or fragment(s) and comparing said level of the at least one biomarker to one or more reference values, wherein said level is indicative of prescribing or not prescribing a remote patient management for said patient. In some embodiments a low benefit level of the at least one biomarker is indicative of not prescribing a remote patient management, whereas in some embodiments a high benefit level of the at least one biomarker is indicative of prescribing a remote patient management. In some embodiments the cardiovascular disease is heart failure, in particular a chronic heart failure that has led to a hospitalization within the last 12 months.
Claims
1. Method for therapy guidance, stratification and/or monitoring of a remote patient management for a patient diagnosed with a cardiovascular disease, the method comprising: providing at least one sample of said patient, determining a level of at least one biomarker selected from the group consisting of pro adrenomedullin (proADM), pro brain natriuretic peptide (proBNP) and/or pro atrial natriuretic peptide (proANP) or fragment(s) thereof in said at least one sample, comparing said level of the at least one biomarker or fragment(s) thereof to one or more reference values, wherein said level of the at least one biomarker or fragment(s) thereof is indicative of prescribing or not prescribing a remote patient management for said patient.
2. Method according to claim 1, wherein a low benefit level of the at least one biomarker or fragment(s) thereof is indicative of not prescribing a remote patient management and/or wherein a high benefit level of at least one biomarker or fragment(s) thereof is indicative of prescribing a remote patient management.
3. Method according to claim 2, wherein the low benefit level of the at least one biomarker or fragment(s) thereof is indicative of not prescribing a remote patient management for a time period of at least 10 days, preferably at least 30 days, 60 days, 90 days, 150 days, 180 days, 270 days or 365 days and/or wherein a high benefit level of at least one biomarker or fragment(s) thereof is indicative of prescribing a remote patient management for a time period of at least 10 days, preferably at least 30 days, 60 days, 90 days, 150 days, 180 days, 270 days or 365 days.
4. The method of claim 2, wherein the low benefit level of proADM or fragment(s) thereof is below a reference value ±20% or less and/or wherein the high benefit level of proADM or fragment(s) thereof is above a reference value ±20% or more, wherein the reference value is selected from a range of values from 0.75 nmol/L to 1.07 nmol/L, more preferably wherein the reference value is 1.07 nmol/L, 0.98 nmol/L, 0.91 nmol/L, 0.86 nmol/L, most preferably 0.75 nmol/L.
5. The method of claim 2, wherein the low benefit level of proBNP or fragment(s) thereof is below a reference value ±20% or less and/or wherein the high benefit level of proBNP or fragment(s) thereof is above a reference value ±20% or more, wherein the reference value is selected from a range of values from 237.6 pg/ml to 1595.8 pg/ml, more preferably wherein the reference value is 1595.8 pg/ml, 1402.95 pg/mol, 1107.9 pg/mol, 609.4 pg/ml, most preferably 237.6 pg/ml.
6. The method of claim 2, wherein the low benefit level of proANP or fragment(s) thereof is below a reference value ±20% or less and/or wherein the high benefit level of proANP or fragment(s) thereof is above a reference value ±20% or more, wherein the reference value is selected from a range of values from 106.9 pmol/L to 248.3 pmol/L, more preferably wherein the reference value is 248.3 pmol/L, 235.6, 186.2 pmol/L, 158.5 pmol/L, most preferably 106.9 pmol/L.
7. The method of claim 1, wherein the cardiovascular disease is a heart failure.
8. The method of claim 1, wherein the cardiovascular disease is a heart failure and the patient has been hospitalized within the last 12 months as a result of a heart failure.
9. The method of claim 1, wherein the cardiovascular disease is a heart failure with an elevated risk of an adverse outcome, preferably selected from the group consisting of acute decompensation and/or death.
10. The method of claim 1, wherein determining a level of proADM or fragment(s) thereof comprises determining a level of MR-proADM, wherein determining a level of proBNP or fragment(s) thereof comprises determining a level of NT-proBNP in the sample, and/or wherein determining a level of proANP or fragment(s) thereof comprises determining a level of MR-proANP.
11. Method for therapy guidance, stratification and/or monitoring of a remote patient management for a patient having been diagnosed with a cardiovascular disease according to claim 1, the method additionally comprising: determining at least one clinical parameter, wherein the at least one clinical parameter is preferably age, weight, body mass index, gender, ethnic background, blood creatinine, left ventricular ejection fraction (LVEF), right ventricular ejection fraction (LVEF), NYHA class, state of medical treatment, blood pressure (systolic/diastolic), heart rate, heart rhythm by electrocardiogram (ECG), peripheral oxygen rate (SpO2), self-rated health status (scale) or a parameter indicating renal function, such as creatinine clearance rate and/or a glomerular filtration rate (GFR). wherein the at least one clinical parameter and the level of the at least one biomarker or fragment thereof are indicative of prescribing or not prescribing a remote patient management for said patient.
12. Method according to claim 11, wherein the at least one clinical parameter is a parameter indicating renal function, preferably a creatinine clearance rate and/or a GFR.
13. The method of claim 1, wherein a first sample is isolated from the patient at a first time point and a second sample is isolated from the patient at a second time point, wherein the absolute difference, the ratio and/or the rate of change of the level of the at least one biomarker or fragment(s) thereof in regards to the first and second time point is indicative of prescribing or not prescribing a remote patient management for said patient.
14. The method of claim 1, wherein the sample is selected from the group consisting of a blood sample, such as a whole blood sample, a serum sample or a plasma sample, a saliva sample and/or a urine sample.
15. Kit for carrying out the method of claim 1, comprising: detection reagents for determining of at least one biomarker selected from the group consisting of proADM, proBNP and proANP or fragment(s) thereof in a sample from a patient, and reference data, such as reference values for determining whether a level of the at least one biomarker is indicative of prescribing or not prescribing a remote patient management, in particular reference data for a low benefit level of the at least one biomarker and a high benefit level of the at least one biomarker, wherein said reference data is preferably stored on a computer readable medium and/or employed in the form of computer executable code configured for comparing a determined at least one biomarker value with the reference values, optionally, detection reagents for determining the level of at least one additional biomarker or fragment(s) thereof, in a sample from a patient and/or means for determining at least one clinical parameter, preferably age, weight, body mass index, gender, ethnic background, blood creatinine, left ventricular ejection fraction (LVEF), right ventricular ejection fraction (LVEF), NYHA class, state of medical treatment, blood pressure (systolic/diastolic), heart rate, heart rhythm by electrocardiogram (ECG), peripheral oxygen rate (SpO2), self-rated health status (scale), or a parameter indicating renal function, preferably a creatinine clearance rate and/or a glomerular filtration rate (GFR) and reference data, such as reference values for determining whether a level of the at least one additional biomarker or fragment(s) thereof and/or the at least one clinical parameter is indicative of prescribing or not prescribing a remote patient management, wherein said reference data is preferably stored on a computer readable medium and/or is in the form of computer executable code configured for comparing the determined levels of said at least one biomarker or fragment(s) thereof and/or said at least one clinical parameter with the reference values.
Description
BRIEF DESCRIPTION OF THE FIGURES
[0417]
[0418]
[0419]
[0420]
[0421]
[0422]
[0423]
[0424]
[0425]
[0426]
[0427]
[0428]
[0429]
[0430]
[0431]
EXAMPLES
[0432] The present invention is further described by reference to the following non-limiting examples. The examples describe non-limiting and practical embodiments, presented for further illustration of the invention.
Methods of the Examples
[0433] Study Design and Participants
[0434] The present study is part of the Telemedical Interventional Management in Heart Failure II (TIM-HF2), randomized, controlled trial investigating the impact of telemedicine on unplanned cardiovascular hospitalisations and mortality in heart failure. Details on the method of the trial and partly some of the results have been published in Koehler et al. 2018a and Koehler et al. 2018b, which are hereby incorporated by reference.
[0435] The TIM-HF2 trial was a prospective, randomised, controlled, parallel-group, unmasked (with randomisation concealment), multicentre trial with pragmatic elements introduced for data collection (ClinicalTrials.gov Identifier: NCT01878630). The trial was done in Germany, and patients were recruited from 200 university, local, and regional hospitals, and cardiology and general practitioner (GP) practices. In total, 113 sites located in 14 metropolitan areas with more than 200 000 inhabitants and/or with a medical university (i.e. Berlin, Dresden, Hamburg, Stuttgart, Frankfurt am Main, Leipzig, Hannover), and in 11 rural areas in Germany (namely: Brandenburg, Bavaria, Thuringia, Saxony, Saxony-Anhalt, Hesse, Baden-Württemberg, Lower Saxony, Mecklenburg-Western Pomerania, North Rhine-Westphalia, Saarland) were included. Forty-three sites were hospitals, 10 sites were university hospitals, and 60 sites were local cardiologist practices. In addition, 87 general practitioners (GPs) collaborated in the study by screening and following up their patients
[0436] Patients were eligible for inclusion if they had been admitted to hospital for worsening heart failure within 12 months before randomisation, were in functional New York Heart Association class II or III, had a left ventricular ejection fraction of 45% or lower (or if more than 45%, were being treated with oral diuretics). Patients were excluded if they had major depression (ie, PHQ-9 score >9), were on haemodialysis, or had been admitted to hospital for any reason within 7 days before randomisation. In addition, patients with a left ventricular assist device or those who had undergone coronary revascularisation or cardiac resynchronisation therapy implantation within 28 days before randomisation were excluded, as were those who were scheduled for coronary revascularisation, transcatheter aortic valve implantation, mitral clip implantation, or cardiac resynchronisation therapy implantation 3 months after randomisation. The inclusion and exclusion criteria are summarized in Table 1.
[0437] The TIM-HF2 trial was designed, implemented, and overseen by an independent steering committee. This report was prepared and submitted for publication by the steering committee. An independent data safety monitoring board reviewed safety data on an ongoing basis. The clinical endpoint committee, masked to study group assignment, adjudicated all deaths and hospitalisations using prospectively defined criteria in the clinical endpoint committee charter. The adjudicated data were used for outcomes regarding hospitalisations and deaths.19 The study complied with good clinical practice in accordance with the Declaration of Helsinki and the laws and regulations applicable in Germany. Written approval from the appropriate ethics committees was obtained.
[0438] Patients provided written informed consent, granting permission for the telemedical centre to contact their health insurance company to cross check the hospital admissions reported by the investigators with those on file in the health insurance records. This process was approved by the German Federal Social Insurance Office and done for patients in both study groups.
[0439] Randomisation and Masking
[0440] Potentially eligible patients were screened for eligibility, and those agreeing to participate and who provided written informed consent were then screened and had baseline measurements and assessments done. Eligible and willing patients were randomly assigned (1:1) using a secure web-based system to either remote patient management plus usual care (remote patient management group) or to usual care alone (usual care group). To ensure a balance of important clinical covariates between the two study groups, we used Pocock's minimisation algorithm with 10% residual randomness. Randomisation was concealed but neither participants nor investigators were masked to group assignment in this open trial.
[0441] Procedures and Remote Patient Management
[0442] The remote patient management intervention comprised the following: a daily transmission of bodyweight, systolic and diastolic blood pressure, heart rate, analysis of the heart rhythm, peripheral capillary oxygen saturation (SpO2) and a self-rated health status (scale range one to five) to the telemedical centre; a definition of a patient's risk category using the baseline and follow-up visit biomarker data in combination with the daily transmitted data; patient education; and co-operation between the telemedical centre, and the patient's GP and cardiologist.
[0443] Home Telemonitoring System
[0444] The telemonitoring system, which was installed in the patient's home within 7 days after randomisation, was a multicomponent system:
[0445] The system used is based on a Bluetooth system with a digital tablet (Physio-Gate® PG 1000, GETEMED Medizin-und Informationstechnik AG) as the central structural element to transmit vital measurements from the home of the patient to the TMC at the Charité—Universitätsmedizin Berlin. Four measuring devices are part of the system: a 3-channel ECG device to collect a 2 min or streaming ECG measurement (PhysioMem® PM 1000 GETEMED Medizin-und Informationstechnik AG), a device to collect peripheral capillary oxygen saturation (SpO2; Masimo Signal Extraction Technology (SET®), a system to collect blood pressure (UA767PBT, A&D Ltd.) and a body weighing scales (Seca 861, seca GmbH & Co KG). Each device is equipped with a Bluetooth chip and connected to the digital tablet. The TMC software used is ‘Fontane’ (eHealth Connect 2.0, T-Systems International GmbH), which was specifically developed for use in the TIM-HF2 study. The key innovation of Fontane is a novel self-adapting TMC middleware, which consists of three key components: [0446] An algorithm for the transmitted patient data to identify critical values or missing data, which allows for an immediate identification of the patients requiring immediate (medical) attention, [0447] Telecommunication software for a direct communication between TMC staff, patients, GPs, and local cardiologists, as well as [0448] Electronic health records for all relevant medical information (e.g. medication plan; reports about previous hospitalisation; laboratory data).
[0449] Patients were also provided with a mobile phone to be used to contact the telemedical centre directly in emergency situations. The mobile phone allowed (DORO Easy 510/Doro HandlePlus 334 gsm, Doro A B) to call the TMC directly in case of emergency. In such situations, it is also possible to initiate a live ECG stream using the ECG device. The tablet uses the mobile network to transmit the patient data automatically in an encrypted manner (GSM-encryption via VPN-Tunnel) to a central server of the TMC in Berlin provided by project partner Deutsche Telekom AG. The combination of measurements and personal data with distinct information codes are only executed at a server at the Charité—Universitätsmedizin Berlin. To ensure patient safety, it is required a priori that the average transmission time to get the data to the TMC must be <90 s. The availability of the mobile network connection is provided by the provider Deutsche Telekom AG. The complete data collection process, transmission and processing is done in strict compliance with state-of-the-art confidentiality and technical standards as agreed with and certified by the relevant data protection officer. For authentication of the individual measurements, all data transmissions incorporated unique device identification information. A service level agreement with the technical provider is concluded for first and second level support and corresponding service and escalation concepts.
[0450] Remote Patient Management
[0451] During the telemonitoring system installation process, certified nurses provided patient training on the system and initiated a heart failure patient education programme; the latter was continued monthly by structured telephone interviews with the patient. The monthly telephone interviews were an integral part of the remote patient management intervention. Combined with the daily data transmissions to the telemedical centre, the patient's clinical and symptomatic status and concomitant medications were assessed, in addition to adherence to the remote patient management intervention and other social and technical issues, which were discussed between the patient and the telemedical centre nurse. Using the wireless system with a digital tablet the data from the patient's was transmitted home to the centre. This was done by using the mobile phone network (secured via a virtual private network tunnel) and transmission of patient data was set at a fixed time daily.
[0452] The telemedical centre provided physician-led medical support and patient management 24 h a day, Monday to Sunday, for the entire study period using the Fontane system, a CE-marked telemedical analysis software (T-Systems International GmbH, Frankfurt, Germany). Algorithms were programmed and implemented in this system which guided patient management and allowed the telemedical centre physicians to act promptly (eg, concomitant medication change, initiation of an ambulatory assessment by a home physician, or to hospitalise the patient) and to piroritise high-risk patients.
[0453] Patients were categorised as low or high risk using the combination of mid-regional pro-adrenomedullin (MRproADM) values and the patient transmitted data.
[0454] At the baseline visit and at each follow-up visit, biomarkers are taken and analysed by an independent laboratory. The results are sent to the CTC and the TMC. According to defined cut-off values for mid-regional pro-adrenomedullin (MR-proADM), patients are risk categorised as follows: low risk patients (MR-proADM≤1.2 nmol/L) and high risk patients (MR-proADM>1.2 nmol/L). High risk patients were primarily followed by TMC physicians (‘doctors care’), and low risk patients by registered TMC nurses (‘nurse care’). The risk category was revaluated every 3 months using the MR-proADM results obtained at each follow-up visit.
[0455] The prioritization in regard to the transmitted data were managed according to the criteria shown below, wherein physicians and nurses prioritized the workload and workflow so that patients presenting with any of the data cut-off limits are managed with priority: [0456] Bradycardia, heart rate<50 b.p.m [0457] Tachycardia, heart rate>100 b.p.m. [0458] Ventricular tachycardia [0459] New-onset atrial fibrillation [0460] PQ interval>200 ms [0461] QRS duration ≥120 ms [0462] QTc interval>460 ms [0463] SpO2<94% [0464] Body weight (weight gain >1 kg in 1 day, >2 kg [0465] in 3 days; >2.5 kg in 8 days) [0466] Blood pressure systolic: <90 or >140 mmHg; diastolic <40 or >90 mmHg [0467] Self-rated health status (grades from 1-very good to 5-very bad): deterioration of about 2 grades starting from 1, or grade 4 or 5)
[0468] The Fontane system also enabled direct communication between the telemedical centre staff and the patient, and the patient's GP and local cardiologists, all of whom were involved in the management of the patient. Via the Fontane system, the telemedical centre created a study-specific electronic patient file, which was accessible by both the telemedical centre staff and patient's care provider.
[0469] Patients in both study groups were followed up for at least 365 days and up to 393 days after randomisation. All patients were seen by their treating cardiologist at the screening and baseline visit and at the final study visit; the latter was done on day 365 (28-day time window) after randomisation. In between, patient visits were scheduled at 3, 6, and 9 months, and were undertaken by the patient's GP or local cardiologist. At all visits, data were collected in a case report form which included vital signs and bodyweight, and patients were asked about the occurrence of hospital admissions since the last study contact. The study flow is shown in
[0470] To avoid contact information and data collection bias, given the daily contact with patients in the remote patient management group, a quality control system was implemented to ensure the accurate and completed reporting of hospital admissions in both the remote patient management plus usual care and usual care groups. This process required the cooperation of patients, investigators, and the patients' respective health insurance companies. The accuracy of data concerning hospital admissions was confirmed using data from the health insurance companies, and a cross check was done with the hospital admissions reported by the investigators.
[0471] The RPM intervention consisted thus in particular of the following elements: [0472] A daily transfer of body weight, blood pressure (systolic/diastolic), heart rate, analysis of the heart rhythm as derived from a 2 min 3-channel electrocardiogram (ECG), peripheral capillary oxygen saturation (SpO2) and a self-rated health status (scale range 1-5) [0473] Identification of a patient risk category using the baseline and follow-up visit biomarker values [0474] Patient education, and Cooperation between the telemedical centre (TMC), the patient's GP and cardiologist (‘doc-to-doc telemedical scenario’) with respect to patient management.
[0475] Patients randomised to the UC group were followed in accordance with the current standards (i.e. ESC guidelines for HF management) at the discretion of their treating physicians (Ponikowski et al. 2016).
[0476] In addition patients allocated to the RPM group undergo a daily structured review of their concomitant medications based on the transmitted data. In consent with the study site physicians, the TMC physicians will optimise concomitant treatments as appropriate to achieve the following targets: [0477] Heart rate<75 b.p.m. for patients in sinus rhythm. [0478] Blood pressure control: systolic <140 mmHg and diastolic <90 mmHg. [0479] Patients with new-onset atrial fibrillation: use of anticoagulant therapy as a long-term treatment and antiarrhythmic therapy. [0480] Patients in NYHA class II-IV: instigate the use of mineralocorticoid receptor antagonists where possible. The aim is to ensure that patients are prescribed the maximally tolerated doses to achieve these targets and, in addition, diuretic doses are adapted in case of weight gain and worsening symptoms.
[0481] The telemedical team informs the patients' GP or caring physician by telephone, fax or email about any new events or important clinical findings from the monthly telephone contact, contacts with the emergency doctor, or any intervention made to the patients' therapy as a result of measured telemedical vital parameters. The TMC preferably only advices the patient's primary physician—it is the latter who has the overall responsibility to instigate the medical management of the patients.
[0482] Study Outcomes
[0483] The primary outcome was the percentage of days lost due to unplanned cardiovascular hospital admissions or death from any cause, comparing remote patient management plus usual care to usual care alone during the individual patient follow-up time. The main secondary outcomes were all-cause mortality and cardiovascular mortality during the individual patient follow-up time plus 28 days after the last study visit, to a maximum of 393 days; percentage of days lost due to unplanned cardiovascular hospital admissions, and percentage of days lost due to unplanned heart failure hospital admissions; change in Minnesota Living with Heart Failure Questionnaire (MLHFQ) global score; and change in N-terminal prohormone brain natriuretic peptide (NT-proBNP) and MR-proADM between randomisation and the final study visit.
[0484] The main secondary outcomes, comparing RPM to Usual Care included: [0485] a) All-cause mortality during the individual patient follow-up time (+28 days of the final visit to a maximum 393 days) [0486] b) Cardiovascular mortality during the individual patient follow-up time (+28 days of the final visit to a maximum 393 days) [0487] c) Percentage of days lost due to unplanned cardiovascular hospitalisations during the individual patient follow-up time [0488] d) Percentage of days lost due to unplanned HF-hospitalisations during the individual patient follow-up time [0489] e) Change in MLHFQ-questionnaire global score between baseline and 365 days [0490] f) Change in levels of NT-proBNP and of MR-proADM between baseline and 365 days
[0491] The following recurrent event analyses were performed: [0492] a) Unplanned cardiovascular hospitalisations and cardiovascular mortality. [0493] b) Unplanned cardiovascular hospitalisations and all-cause mortality. [0494] c) Unplanned HF hospitalisations and cardiovascular mortality. [0495] d) Unplanned HF hospitalisations and all-cause mortality
[0496] Subgroup analyses is performed for the primary outcome to assess the consistency of intervention effects across the following subgroups: [0497] Metropolitan vs. rural area of medical care. [0498] Male vs. female. [0499] Above/below median age. [0500] LVEF 545% vs. LVEF>45%. [0501] NYHA functional class I/II vs. III/IV. [0502] Cardiac resynchronisation therapy (CRT) at baseline yes/no. [0503] Implantable cardioverter defibrillator (ICD) at baseline yes/no. [0504] MR-proADM at baseline ≤1.2 nmol/L vs. >1.2 nmol/L.
[0505] Statistical Analysis
[0506] Data for specific subgroups from the TIM-HF trial were used for sample size calculations. For the patient subgroup that mirrored the population it was intended to include in the TIM-HF2 trial, 19 days were lost due to all-cause death or unplanned cardiovascular hospital admissions at 12 months in the usual care group, and 12 days were lost for patients in the remote patient management group, which corresponds to a 38% reduction. With an estimated pooled SD of 48, it was calculated that 750 patients would be required in each group to detect this difference with a power of 80% and a two-sided a of 5%.
[0507] R (version 3.4.4) and Stata (version 14.2) were used for analyses. The primary and secondary efficacy analyses were performed on the full analysis set, in accordance with the intention-to-treat principle. The full analysis set consisted of all randomised patients who gave consent and began their assigned care.
[0508] Baseline characteristics were summarised as number of patients (%) for categorical variables and as mean (SD) for continuous variables; for all baseline laboratory tests, the median and IQR was used.
[0509] For the primary analysis of percentage of days lost due to all-cause death or unplanned cardiovascular hospital admission, the proportion of follow-up time lost due to death or unplanned cardiovascular hospitalisation was defined as the number of days lost divided by the intended follow-up. For patients who died, the number of days lost between the date of death and the date of intended follow-up plus the number of days spent in hospital for cardiovascular reasons were counted. For patients who completed the study as planned or who withdrew prematurely from follow-up, the fraction of follow-up time was defined as number of days lost (due to cardiovascular hospitalisation) divided by the follow-up time realised (ie, up to the censoring date). For the primary outcome, a permutation test was used to compare the weighted averages of the percentage of days lost between the two groups. The two-sided permutation test p value was calculated as the fraction of permutations, which had an absolute value of the test statistic at least as large as the observed test statistic, when we applied a mid-p correction in case of equality. For this analysis 2000 randomly drawn permutations were used.
[0510] Confidence intervals (CIs) were calculated using the method described by Garthwaite (Garthwaite P H. Confidence intervals from randomization tests. Biometrics 1996; 52: 1387-93), which is based on the Robbins-Monro method. In short, this method does a separate search for each endpoint of the CI by sequentially updating the estimates where the magnitude of steps is governed by the distance between the original test statistic and the test statistic for the permuted data, and the step number. Follow-up time was weighted using weighted arithmetic means, and annualised averages are presented.
[0511] In short, this method does a separate search for each endpoint of the CI by sequentially updating the estimates where the magnitude of steps is governed by the distance between the original test statistic and the test statistic for the permuted data, and the step number. Follow-up time was weighted using weighted arithmetic means, and annualised averages are presented.
[0512] All survival analyses were done on a time-to-first event basis. Cumulative incidence curves for all-cause mortality were constructed according to the Kaplan-Meier method and the differences between curves were examined by the log-rank statistic. A competing risk analysis was used for cardiovascular mortality to take into account that the event of interest could not occur because of another previous fatal event. Cox-proportional hazards regression models were used to estimate (cause-specific) hazard ratios (HRs). Event rates are expressed as the number of events per 100 patient years of follow-up, taking into account the censoring of follow-up data.
[0513] Sensitivity analyses for mortality outcomes examined the robustness of the results using the full analysis set of all patients censored at day 393 after randomisation as defined in the statistical analysis plan. We analysed data for number of hospitalisation events by negative binomial models. For continuous variables such as the MLHFQ global score, changes in group means of both study groups at 12 months were compared by ANCOVA models adjusting for the baseline value. The biomarker test results were analysed using a log scale and ANCOVA models.
[0514] Compliance with the daily data transmissions to the telemedical centre was defined as the number of days between the day when the first data transmission was sent to the telemedical centre up to the end of the patient's individual follow-up minus any day when the patient was admitted to hospital for any reason. A statistical test of interaction was done to assess whether the effect of the remote patient management on the primary outcome was consistent across the prespecified subgroups. Interaction tests for the subgroup analyses were done by adding the interaction term in the corresponding models.
Example 1: Benefit of Remote Patient Management for Patients Suffering of HF
[0515] Between Aug. 13, 2013, and May 12, 2017, 1571 patients were randomly assigned (796 to remote patient management plus usual care and 775 to usual care only, of which 765 in the remote patient management group and 773 in the usual care group were included in the full analysis set;
[0516] The mean age of all patients was 70 years (SD 10), and 70% were men.
[0517] For patients randomly assigned to receive remote patient management, 743 (97%) were at least 70% compliant with the daily transfer of data to the telemedical centre. Additionally, all patients were contacted within 24 h of missing data transmissions. Survival status was known for all patients up to the maximum follow-up for each patient (ie, up to day 393 after randomisation).
[0518] 265 (35%) of 765 patients in the remote patient management group and 290 (38%) of 773 in the usual care group were admitted to hospital for an unplanned cardiovascular reason or died. The percentage of days lost due to unplanned cardiovascular hospital admissions or all-cause death was statistically reduced in patients allocated to remote patient management (4.88%, 95% CI 4.55-5.23) as compared with usual care (6.64%, 95% CI 6.19-7.13; ratio 0.80, 95% CI 0.65-1.Math.00; p=0-0460; table 2). Patients assigned to remote patient management lost a weighted average of 17.8 days per year compared with 24.2 days per year for patients assigned to usual care for this outcome.
[0519] The rate of all-cause death was 7.9 per 100 person-years of follow-up in the remote patient management group and 11.3 per 100 person-years of follow-up in the usual care group (HR 0.70, 95% CI 0.50-0.96; p=0.0280; table 4;
[0520] Patients assigned to remote patient management lost fewer days than the usual care group for unplanned hospital admissions due to worsening heart failure (mean 3.8 days per year [95% CI 3.5-4.1] vs 5.6 days per year [5.2-6.Math.0], respectively). The percentage of days lost for this outcome for the remote patient management and usual care groups was 1.04% (95% CI.0.Math.96-1.11) and 1.53% (1.43-1.64), respectively (ratio 0.80, 95% CI 0.67-0.95; p=0.0070). Comparing remote patient management with the usual care group, similar results were obtained for the sensitivity analysis done for all-cause mortality (ratio 0.74, 95% CI 0.54-1.02; p=0.0633).
[0521] The percentage of days lost due to unplanned cardiovascular hospital admissions was 1.71% (95% CI 1.59-1.83) for the remote patient management group and 2.29% (2.13-2.45) for the usual care group (ratio 0.89, 95% CI 0.74-1.07; p=0.208).
[0522] The change from baseline in the Minnesota Living with Heart Failure Questionnaire (MLHFQ) global score at 12 months, was not statistically different between the remote patient management and usual care group (table 5).
[0523]
[0524] 2251 unplanned hospital admissions were reported and classified by the clinical endpoint committee (appendix p 4). Of these hospitalisations, 262 (14 in the remote patient management group and 248 in the usual care group) were identified during the cross-check verification procedure with health insurance records. 1,026,078 vital parameters were transmitted to the telemedical centre (a median of 1421 per patient [range 6-3962]); table 6 provides a summary of the data transmitted and actions taken.
Example 2: Prognostic Ability of Biomarker Determination for Prescribing or not Prescribing Remote Patient Management
[0525] On the study population and collected data from the TIM-HF II trial as described above and in Koehler et al. 2018a and Koehler et al. 2018b further analysis was perform to assess the prognostic ability of the biomarkers proADM, proBNP and/or proANP to predict the benefit of prescribing or not prescribing a remote patient management.
[0526] Methods for Biomarker Analysis
[0527] As detailed above, in TIM-HF II trial a total of 1538 patients (median age: 73, IQR: 64-78; 70% male) were randomly assigned either to RPM (N=765) or SOC (N=773). The terms SOC (standard of care) and usual care group (UC) are used interchangeably herein. Patients had study visits every 90 days over one year of follow-up time. For the analysis of the biomarkers, blood was drawn at baseline and at every visit.
[0528] NT-proBNP, MR-ANP and MR-proADM was assessed concerning (i) their association with the primary endpoint of the original trial % lost days due to unplanned cardiovascular hospitalization or all-cause death, (ii) their association with the pre-specified endpoint all-cause death or re-hospitalization for heart failure considered to be decisive for RPM patient selection, and (iii) their predictive performance when used in combination for the identification of patients who can safely be excluded from RPM for the next 90 days (no all-cause death or re-hospitalization for heart failure).
[0529] For the latter, we gained statistical power by pooling all repeated observations of biomarkers and subsequent 90-days follow-up for all patients under SOC (as expected no trend over repeated measurements observed). After accounting for renal insufficiency and its association with biomarker levels via stratification by eGFR (CKD-EPI formula) or GFR based on Cockroft-Gault, we calculated biomarker cutoffs for safe rule-outs (100%, 99%, 95% sensitivity) and their hypothetical performance in terms of saved RPM effort (% excluded patients from RPM) as well as in terms of efficacy of RPM for the included patients (reduction of risk of event, number-needed-to-be-treated NNT). Due to the unique available dataset for RPM patients, the exclusion algorithm could further be evaluated with detailed data on emergencies, telephone calls and medication.
[0530] Results of Biomarker Analysis for MR-proADM, ProBNP or proANP within 90 Days from Baseline
[0531] For an endpoint of “All-cause death or unplanned CV-hospitalization due to acute decompens. within 90 days from baseline” an ROC analysis and analysis of the performance biomarker cutoffs in UC patients was conducted as a benchmark.
[0532]
[0533] Patients with higher levels of NT-proBNP or MR-proADM at baseline visit of TIM-HF2 had a higher rate of lost days and were more likely to develop adverse events during the year of the trial. This indicated their potential for the allocation of RPM.
[0534] Table 7-9 summarize the cutoff analysis for the biomarkers proADM, proBNP and proANP. Particularly, preferred cutoffs for establishing a low benefit and high benefit levels for prescribing or not prescribing relate to cutoffs for which a sensitivity of 100% or 95% is reached.
[0535] As shown in Table 7 for MR-proADM a cutoff value of 0.75 nmol/L yields a sensitivity of 100% and 0.86 nmol/L a sensitivity of 95%.
[0536] As shown in Table 8 for NT-proBNP a cutoff value of 237.6 pg/mL yields a sensitivity of 100% and 609.4 pg/mL a sensitivity of 95%.
[0537] As shown in Table 9 for MR-proANP a cutoff value of 106.9 pmol/L yields a sensitivity of 100% and 158.5 pmol/L a sensitivity of 95%.
[0538] Results of Biomarker Analysis when Combined with Determining the GFR within 90 Days from Baseline
[0539] Particular, well results were achieved when using an NT-proBNP- and MR-proADM—in combination with determining a parameter indicating renal function, such as a GFR.
[0540] An NT-proBNP- and MR-proADM-based algorithm stratified by eGFR<60 and eGFR>=60 could identify a notable portion of patients who remained event-free for 90 days already under SOC, while, on the other hand, identifying critical SOC patients with event with perfect or near-perfect sensitivity. For 100% (99%, 95%) sensitivity, 9% (15%, 25%) of SOC patients could hypothetically be excluded from RPM. For comparison, at 100% sensitivity, using NT-proBNP alone or in combination with eGFR would only have excluded safely 0.8% or 2.7% of patients, respectively.
[0541] For the exclusion algorithm with 100% sensitivity, the risk of suffering from an event within 90 days for SOC patients was 11% for patients hypothetically assigned to RPM and 0% for patients kept for SOC. By comparing these sub-population between study arms (again, pooling all observations of biomarkers and subsequent 90-day follow-ups; no trend of biomarkers over repeated measurements observed), it was shown that RPM could significantly decrease the risk of suffering from an event among the hypothetically included patients from 11% to 8%. Further, NNT was lowered from 42 (all patients) to 34 (hypothetically included patients). Finally, true RPM patients who would have been excluded from RPM via the presented algorithm had a significantly lower chance of suffering from emergencies, received less medication changes and communicated less with RPM doctors than those who would have been included. This further suggested that the identified low-risk patients were less in need of RPM than the identified high-risk patients, based on a characterization by biomarkers.
[0542] Table 10 summarizes the results and beneficial impact for a stratification of biomarker cutoffs by a GFR based upon CKD-EPI.
[0543] Table 11 and 13 summarize the results and beneficial impact for a stratification of biomarker cutoffs by a GFR based upon Cockroft-Gault.
[0544] Particular well results are achieved when using a combination of proADM, proBNP and GFR. As seen in the columns 6 and 9 of both tables, in such a case twice as many or three times as many patients may be safely excluded from a remote management in comparison to only using proBNP and GFR without proADM for prescribing or not prescribing a remote patient management.
[0545] Results of Biomarker Analysis within One Year from Baseline
[0546] Furthermore the predictive power of the biomarkers were evaluated with regard to adverse events within one year after randomization and sample isolation. NT-proBNP and MR-proADM were used at baseline. The primary endpoint was lost days due to all cause death or CV hospitalization, the secondary endpoint was all cause death within one year after randomization.
[0547] The primary analysis of quintiles of the two biomarkers with respect to the prediction of the primary endpoint shows that the incidence increased in the SOC group from 1.4% lost days (MR-proADM≤0.75 nmol/L; lowest quintile) to 17.6% in the highest quintile. In the RPM group values were similar at the lowest quintile (1.4%) and 12.1% at the highest quintile (p=0.21 versus SOC). The treatment effect (ratio % lost days RPM vs. SOC) increased from 0.98 to 0.69 (p for interaction 0.29). NT-proBNP had similar prognostic power with the treatment effect ranging from 0.87 at the lowest to 0.63 at the highest quintile (p for interaction 0.33). Findings for the secondary endpoint (all-cause death) were similar. Accordingly, a trend towards lower % lost days and risk of death, as well as smaller benefits from RPM for patients with lower biomarker levels was observed. Building on this observation, a biomarker-based RPM patient selection algorithm was explored with the aim to reduce the rate of patients recommended for RPM.
[0548] Once the biomarkers were combined to identify patients who had no events (all-cause death) and therefore could not profit from RPM, 13.7% of patients with MR-proADM<0.69 nmol/L or NT-proBNP<125.1 ng/L could have been excluded in the SOC group. If miss rates of <2.5% or <5% are accepted, the relative share of excluded patients rose to 16.9% or 25.3% respectively. In all three scenarios the hazard ratio for the beneficial treatment effect of RPM remained at 0.71 significant (p<0.05) and was mutually the same as in the primary study (HR=0.70).
[0549] Thus, by using biomarker based selection of patients, the number needed to treat to prevent one death, could be lowered from 28 to 26 (sensitivity 100%, no event missed) to a minimum of 23 (sensitivity 95.5%, MR-proADM<0.75 nmol/L, NT-proBNP<383.3 ng/L).
[0550] Table 12 summarizes the results and the
[0551] In conclusion, the use of MR-proADM and NT-proBNP alone or in combination allows a safe, more precise, effective and therefore cost-saving allocation of patients with heart failure to remote patient management. The number needed to treat by RPM to save one life could be lowered from 28 in the original study to 23 using the biomarkers approach.
Example 3: Prognostic Ability of a Combination of ProBNP and ADM for Prescribing or not Prescribing Remote Patient Management
[0552] On the study population and collected data from the TIM-HF II trial as described above and in Koehler et al. 2018a and Koehler et al. 2018b a further statistical analysis was performed to determine particularly beneficial scenarios of using proADM and/or proBNP to predict the benefit of prescribing or not prescribing a remote patient management.
[0553] Outcomes and Biomarker Analysis
[0554] In brief, as outcomes the primary study endpoint was “percentage of days lost due to unplanned cardiovascular (CV) hospitalization or due to all-cause death during the individual follow-up time”. The regular individual follow-up period covered 365 days after randomization in all patients plus the varying time to the final study visit, which should have taken place within 4 weeks after day 365. The secondary endpoints include all-cause mortality (a), CV mortality (b), % lost days due to unplanned CV hospitalizations (c) and % lost days due to hospitalization with worsening heart failure (d).
[0555] For the purpose of biomarker analysis, levels of NT-proBNP and MR-proADM at baseline were used. Whole blood was collected by venipuncture during study visits. NT-proBNP was measured with the chemiluminescence immunoassay Roche NT-proBNP (Roche Diagnostics GmbH, Mannheim, Germany) which has a measuring range of 5 to 35.000 pg/ml and functional assay sensitivity of 50 pg/ml (manufacturer information, package insert). MR-proADM was measured with the immunofluorescent assay B.Math.R.Math.A.Math.H.Math.M.Math.S MR-proADM KRYPTOR (B.Math.R.Math.A.Math.H.Math.M.Math.S GmbH, Hennigsdorf, Germany). The MR-proADM assay has a total measuring range of 0.05 to 100 nmol/L with a functional assay sensitivity of 0.25 nmol/L.
[0556] Statistics
[0557] Association of NT-proBNP and MR-proADM with Endpoints
[0558] Linear and Cox proportional hazards regressions were used to test the association of % lost days due to unplanned CV hospitalization and time to all-cause death, respectively, with both biomarkers. For modelling, biomarker levels were log-transformed. Also, % lost days was log-transformed after imputing a value of 0.1% for patients who had 0.0% lost days, which is in line with previous analyses. Models including both biomarkers were compared to models only including NT-proBNP as a predictor variable to assess the significance of adding MR-proADM (by F-test and likelihood ratio test, respectively).
[0559] Complementarily, average % lost days due to unplanned CV hospitalization (using an average weighted by individual follow-up time, in line with previous analyses of the trial) was calculated, as well as the rate of all-cause death for both the SOC and RPM group for all quintiles of both biomarkers. For each quintile, a p-value for the effect of RPM vs. SOC on these endpoints was calculated using permutation tests (% lost days due to unplanned CV hospitalization) and Cox proportional hazards regressions (all-cause death). Also a p-value for the interaction of quintile and RPM was calculated to explore how strong the evidence was that the effect of RPM differed across the biomarker quintiles.
[0560] Selection Scenarios for Patients Recommended for RPM
[0561] To identify criteria for the subgroup of the original population for which RPM could be recommended on the basis of additional biomarker assessment, only those patients were used that had been assigned to the SOC study arm. Hence, these served as the benchmark population for the derivation and evaluation of biomarker guidance. The reason is that only in this group it can be assumed that any clinical endpoint has not been prevented by RPM.
[0562] For the primary endpoint, an event was defined as having at least 30 lost days out of 365 days of follow up, i.e., as having a rate of at least 8.2% lost days. In the following description of this example, this scenario will be called “30 lost days/year”. With this approach, the vast majority of patients who died (82 out of 89 in the SOC group), but also all patients who out of one year spent a month or more in hospitals due to unplanned CV admission are included. With this binary measure, only those deaths which were relatively far away in time from the baseline biomarker measurement (at least 11 months for a patient who was followed up for one year) were categorized as no event.
[0563] The biomarker-based selection scenarios for patients recommended for RPM were further optimized for high safety. Different scenarios were explored for both endpoints with desired sensitivities of 100%, 98% and 95%, meaning that the defined biomarker cutoffs should not miss more than 0%, 2% or 5% of patients with an event during follow-up time. To make use of the information carried by both biomarkers, patients were recommended for RPM if they had at least a certain level of NT-proBNP and, simultaneously, at least a certain level of MR-proADM.
[0564] Note that, theoretically, for all scenarios with sensitivity <100%, there is more than one possible combination of biomarker cutoffs that realizes the desired sensitivity. The analysis of the current example, remained restricted to those cases where both biomarkers on their own achieved identical sensitivity. This is equivalent with assuming that both biomarkers should be weighed equally in their joint use for patient selection (instead of assuming that one should be more permissive with high levels of NT-proBNP than with levels of MR-proADM, or vice versa).
[0565] Evaluation of Scenarios Regarding Efficacy of RPM
[0566] The basic intention of biomarker guidance for RPM is to assign those patients to the intervention who will profit most from it, and rather exclude those who will not. Consequently, it was important for all six scenarios (two endpoints (% lost days due to unplanned CV hospitalization; all-cause death) and three desired sensitivities) to show how the endpoints of TIM-HF2, and hence the efficacy of RPM in heart failure, were affected by retrospectively reducing the original population to the subpopulation recommended to RPM via biomarkers. Effect estimates and p-values were calculated for the subpopulation who satisfied the recommendation criteria (i.e., whose NT-proBNP and MR-proADM surpassed the respective thresholds) in full accordance with all statistical procedures in the original trial population in the earlier publication (Koehler et al. 2018a).
[0567] In brief, for ratios of % lost days between treatments (primary endpoint and secondary endpoints c and d), the geometric mean of % lost days in the RPM group was divided by the geometric mean of % lost days in the SOC group. This was preceded by imputing a value of 0.1% lost days for patients who had 0.0% lost days. The corresponding p-values were calculated via permutation tests with 2000 permutations and the difference in means as test statistic. For the time-to-event analyses for all-cause death (secondary endpoint a) and CV death (secondary endpoint b), Cox proportional hazards regressions were used to estimate hazard ratios, confidence intervals and p-values. Note for the interpretation of the results that upward changes in p-values are at least to some degree mechanical since in the reduced subpopulations the statistical tests have less power to detect an effect at the same significance level. For the latter endpoint, non-CV deaths were treated as regular censoring events (end of follow up). Further, a Kaplan-Meier curve and a log-rank test were provided for the endpoint all-cause death for one of the patient selection scenarios.
[0568] Further Analysis for the Most Efficient Scenario: Patient Characteristics and RPM Interventions
[0569] For the most efficient of the evaluated scenarios of double-biomarker guidance (reducing the population recommended for RPM by the most) which was the one with 95% sensitivity regarding ≥30 lost days/year, patient characteristics at baseline were compared between the group recommended for RPM and the group not recommended for RPM. Also, patient characteristics at baseline were compared between the SOC and RPM group for the patients recommended for RPM (i.e., for the groups which were the basis for the endpoint calculations described in the previous subsection).
[0570] The availability of unique electronic health record data and cross-verified emergency data for the patients who were in the original RPM group allowed further comparisons. The TMC employed various interventions during the course of the telemedical treatment and registered the occurrence of emergencies for the RPM patients. Thus, these RPM data allowed to compare (i) the rate of emergencies, (ii) the average medical effort in time spent by TMC doctors between those patients who would have been recommended to RPM in this retrospective scenario, in line with their original random group assignment in the trial, and those who would not have been recommended to RPM in this scenario, against their original group assignment in the trial. For the calculation of effort in time, all interventions by the TMC documented in the electronic patient record were categorized into medical and non-medical related actions. Average duration of each action was calculated via empirical values and values from the electronic patient record. Emergencies and medical effort was visualized following Allen et al. (2019).
[0571] Further, by summing up the time of medical and non-medical effort across all patients who would not have been recommended to RPM in the retrospective scenario, but were assigned to RPM in the original trial, the effort that could have been saved by employing the explored biomarker guidance could roughly be estimated.
[0572] All statistical analyses were conducted and documented by scripts using R version 3.5.1 (R. Core Team 2018), a language and environment for data processing, statistical computing and graphics. Cox proportional hazards models were computed with package survival 2.42.3 (Terry et al. 2000).
[0573] Results for Using a Combination of proBNP and ADM for the Prescription of a Remote Patient Management
[0574] Association of NT-proBNP and MR-proADM with Endpoints
[0575] Linear and Cox proportional hazards regression models showed that both, NT-proBNP and of MR-proADM, were significantly associated with % lost days due to unplanned CV hospitalization and time to all-cause death (all p<0.001). In model comparisons, linear and Cox proportional hazards models that included both biomarkers performed significantly better than models that included only NT-proBNP as a predictor (all p<0.001). The raw data on the association of biomarkers with the primary endpoint is shown in
[0576] In line with this, the primary analysis of quintiles of the two biomarkers and their association with the event rates for the primary endpoint showed that for MR-proADM, % lost days increased in the SOC group from 1.4% (MR-proADM≤0.75 nmol/L; lowest quintile; Table 14A) to 17.6% (MR-proADM up to the measured maximum of 7.8 nmol/L; highest quintile) across quintiles. In the RPM group, this trend was similar (1.4% to 12.1%). The same tendency of higher risk for suffering from events with higher biomarker levels can also be seen for NT-proBNP (Table 14A), as well as for the endpoint all-cause death (Table 14B).
[0577] Furthermore, the effect that RPM had on these two endpoints tended to increase across quintiles. For example, NT-proBNP had prognostic power for the treatment effect of RPM on % lost days, as patients in the lowest quintile (with ≤487.9 pg/ml) had slightly more % lost days in RPM than in SOC, but for those patients in the highest quintile (with 3701.2-35000 pg/ml), % lost days were markedly reduced in the SOC group. The same trend held for MR-proADM, as well for the endpoint all-cause death.
[0578] Selection Scenarios for Patients Recommended for RPM
[0579] The cutoffs for selection which patients should not be recommended for RPM ranged between 125.1 and 413.7 pg/ml for NT-proBNP and 0.63 and 0.75 nmol/L for MR-proADM, depending on the desired safety (sensitivities 100%; 98%; 95%) and patient selection criterion (at least 30 lost days/year due to unplanned CV hospitalization or all-cause death; all-cause death), as shown in Table 15.
[0580] The lower the desired sensitivity, the higher were the critical biomarker cutoffs and the higher was the proportion of patients that would not be recommended for RPM. Also, the lower the desired sensitivity, the higher was the positive predictive value (PPV) of the biomarker guidance. For example, if we allowed that 5% of the patients who suffered from 30 out of 365 lost days or more did not receive RPM (95% sensitivity), 21.5% of patients who would be recommended for RPM by the biomarker combination would experience this event. For comparison, the rate of this event in the full SOC group in the original trial population was 16.4%. This increase in PPV was associated with a reduction of the population who was recommended to RPM by 27.0%.
[0581] This scenario of biomarker guidance is shown in
[0582] The reduction of the population recommended for RPM by the joint use of NT-proBNP and MR-proADM allowed to exclude event-free patients with rather high levels of NT-proBNP, who nevertheless had rather low levels of MR-proADM (lower right quadrant in
[0583] Comparing a single vs. the explored double-biomarker guidance, the superiority of using two biomarkers was in particular obvious for high safety: for 100% sensitivity regarding 30 out of 365 or more lost days (all-cause death, respectively), NT-proBNP alone would reduce the population with RPM recommended by 3.4% (3.4%) of patients, reaching 4.0% (3.9%) specificity. This is more than three times higher when using a combination of NT-proBNP and MR-proADM, where the population with RPM recommended would safely be reduced by 10.8% (13.9%) of patients, reaching 12.9% (15.7%) specificity.
[0584] For 95% sensitivity regarding 30 out of 365 or more lost days, NT-proBNP alone would reduce the population with RPM recommended by 23.4% of patients (
[0585] Note that for all sensitivities <100%, these numbers depend on the exact combination algorithm that is used to integrate the information of both biomarkers. For example, following the approach explored in the example where both biomarkers are weight equally (see Methods), in the scenario of 95% sensitivity for all-cause death, NT-proBNP alone would reduce the population recommended for RPM by 32.3% (cutoff 383.3 pg/ml), while for the double-biomarker guidance in this scenario only a reduction by 25.6% was found (Table 15). The exclusion rate could be increased further to 36.0% by additionally employing MR-proADM (with single marker sensitivity at 100%) in the selection procedure.
[0586] Evaluation of Scenarios Regarding Efficacy of RPM
[0587] Table 15 shows the effect of biomarker based reduction of the randomized groups with respect to the primary and all secondary endpoints. The effects on the endpoints remained mainly significant, specifically for the most efficient scenario and the endpoint all-cause mortality. More notably, since p-values are affected by a biomarker-based reduction of the sample, effect estimates remained highly similar to the results of the original trial, indicating efficacy of RPM for most endpoints. This also highlighted by the Kaplan-Meier curve for the latter scenario and the endpoint all-cause death (see
[0588] Further Analysis for 95% Scenario: Patient Characteristics and RPM Interventions
[0589] For the biomarker guidance scenario with 95% sensitivity regarding the endpoint 30 lost days/year (i.e., ≥8.2% lost days), the patient characteristics of the identified subgroups is described. The biomarker cutoffs for this scenario were as follows: only patients were recommended for RPM who had both a level of NT-proBNP 413.7 pg/ml and a level of MR-proADM 0.75 nmol/L (see Table 16).
[0590] With this, related to the primary endpoint, with a sensitivity of 95%, n=1098/1538 (71.4%) patients remained in the biomarkers selected data set. The Table 16a shows the characteristics of these patients by randomized treatment groups and the Table 16b the characteristics of selected versus not selected patients. Data in the Table 16a confirm that despite biomarkers based selection, both randomized treatment groups remain comparable without significant differences. Table 16b shows that the biomarkers selected patients were at higher risk with lower LVEF and higher NYHA categories.
[0591] Further results were obtained based on the unique electronic health record data that was available for patients who were in the RPM group during the trial. For the scenario explored in the current example, the rate of emergencies in patients who would have been excluded from RPM was significantly lower than among those who had would have been recommended for RPM (median 0, IQR 0-1 vs. median 1, IQR 0-2; Wilcoxon rank sum test, p<0.001;
[0592] Based on the effort data that could be estimated from the electronic health records for each individual patient who was assigned to RPM in the original trial, a substantial amount of staff time could have been saved in this biomarker guidance scenario. The total estimated time of medical effort that was spent for patients in the trial's RPM group by the TMC was 4332 hours (appr. 5.7 hours per patient). The time of non-medical effort that was spent by the TMC was 356 hours (appr. 0.5 hours per patient). Of this, 1170 hours of medial effort and 99 hours of non-medical effort could have been saved if RPM had not been recommended to those patients who were below biomarker cutoffs that have been explored here (see Table 16a, b for detailed description of this subpopulation of patients). Note that these results are qualitatively similar for other biomarker guidance scenarios based on other desired sensitivities or the other endpoint (all-cause death).
[0593] Further results are related to patients selected with 95% sensitivity regarding the endpoint all-cause mortality.
[0594] In summary, in the present example of a biomarkers substudy of TIM-HF2 it was found that with the combination of NT-proBNP (cutoff <383.3 pg/ml) and MR-proADM (cutoff <0.75 nmol/L) 72.5% of the originally randomized patients could be selected who have the same benefit as the original cohort. Therefore, by a simple measurement of those two biomarkers at baseline, RPM indication could be personalized to a higher risk cohort. Of note, the biomarker test can be easily taken from any caregiver and requires no other pre-analytic procedures. The blood sample can be send by the standard postal service to a centralized core lab very cost effective as proven in the main study (Koehler et al 2018a).
[0595] Example 3 primarily analysed the prognostic effect of NT-proBNP and MR-proADM, which both are significantly associated with outcome. Comparing the two randomized groups, effects of treatment are not significantly predicted by both markers. Therefore, the approach to use these markers was primarily employed to identify patients who do not profit from RPM. This was done by analyzing the biomarkers in the SOC-group with no events. Cutoffs identified by this approach were applied to the randomized cohorts and the selected populations were compared for efficacy. Example 3 shows that the effect of RPM remained mainly the same. The approach enabled to personalize the recommendation for RPM and for the present example reduce the eligible cohort to 71.4% of the entire population and thus avoids efficiently unnecessary cost-intensive therapies.
TABLE-US-00005 TABLE 1 Inclusion and exclusion criteria for patient participation in the study Inclusion criteria Exclusion criteria Diagnosed with HF—NYHA Hospitalisation within the last class II or III 7 days before randomisation Echocardiographically determined Implanted cardiac assist system left ventricular ejection fraction ≤45% or >45% + oral diuretic prescribed Hospitalisation due to Acute coronary syndrome decompensated HF within the last within the last 7 days 12 months before randomisation before randomisation Depression score PHQ-9 < 10 High urgent listed for heart transplantation Written informed consent Planned revascularisation, obtained transcatheter aortic valve implantation, MitraClip and/or CRT implantation within 3 months after randomisation Revascularisation and/or CRT implantation within 28 days before randomisation Known alcohol or drug abuse Terminal renal insufficiency with haemodialysis Impairment or unwillingness to use the telemonitoring equipment (e.g. dementia, impaired self-determination, lacking ability to communicate) Existence of any disease reducing life expectancy to less than 1 year Age < 18 years Pregnancy Participation in other treatment studies or remote patient management programmes (register studies possible) Abbreviations: CRT, cardiac resynchronisation therapy; HF, heart failure; NYHA, New York Heart Association; PHQ, Patient Health Questionnaire.
TABLE-US-00006 TABLE 2 Study flow and assessments performed at different visits Final visit 3- 6- 9- (365 days or Month Month Month within +28 Screening Baseline visit visit visit days Informed consent and patient X X information Review inclusion/exclusion X X criteria Randomisation X Physical examination X X Registration medication X X Echocardiography X 12-channel ECG X X Laboratory tests: haemoglobin, X X X X X haematocrit, leucocytes, thrombocytes, sodium, potassium, creatinine Cardiac biomarkers: NT-proBNP, X X X X X MR-proADM, MR-proANP, procalcitonin Health questionnaires: MLHFQ, X EQ-5D-3 L, PHQ-9D, G9- EHFScBS Registration of events: X X X X hospitalisation, emergency, death Abbreviations: EQ-5D-3 L, EuroQo1-5 Dimensions-3 Levels; G9-EHFScBS, German 9-Item European Heart Failure Self-care Behaviour Scale; MLHFQ, Minnesota Living with Heart Failure Questionnaire; MR-proADM, mid-regional pro-adrenomedullin; MR-proANP, mid-regional pro-A type natriuretic peptide; NT-proBNP, N-terminal pro-B type natriuretic peptide; PHQ-9D, Patient Health Questionnaire nine questions in German
TABLE-US-00007 TABLE 3 Baseline characteristics: Remote patient management Usual care (n = 765) (n = 773) Age (years) 70 (11) 70 (10) Sex Male 533 (70%) 537 (69%) Female 232 (30%) 236 (31%) Living alone 213 (28%) 222 (29%) Living in a urban area vs rural area Rural 457 (60%) 458 (59%) Urban 308 (40%) 315 (41%) NYHA class I 3 (0%) 8 (1%) II 400 (52%) 396 (51%) III 359 (47%) 367 (47%) IV 3 (0%) 2 (0%) LVEF 41 (13) 41 (13) ≤45% 492 (64%) 509 (66%) >45% 273 (36%) 264 (34%) <40% 342 (45%) 328 (42%) 40-50% 228 (30%) 272 (35%) >50% 195 (25%) 173 (22%) Days between discharge 92 (81) 93 (82) of last heart failure hospital admission and randomisation ≤30 days 192 (25%) 198 (26%) 31-90 days 281 (36%) 276 (36%) >90 days 299 (39%) 291 (38%) Bodyweight (kg) 87 (21) 88 (21) Body-mass index (kg/m.sup.2) 30 (6) 30 (6) Blood pressure (mm Hg) Systolic 126 (19) 125 (20) Diastolic 74 (11) 74 (11) Pulse (beats per min) 73 (14) 72 (14) Primary cause of heart failure Ischaemic cause (coronary artery 301 (39%) 323 (42%) disease or myocardial infarction) Hypertension 128 (17%) 146 (19%) Dilated cardiomyopathy 176 (23%) 171 (22%) Other 160 (21%) 133 (17%) Cardiovascular risk factors Smoking status Unknown 24 (3%) 27 (3%) Non-smoker 378 (49%) 385 (50%) Former smoker 286 (37%) 304 (39%) Smoker 77 (10%) 57 (7%) Hyperlipidaemia Unknown 41 (5%) 39 (5%) No 306 (40%) 318 (41%) Yes 418 (55%) 415 (54%) Diabetes mellitus 347 (45%) 355 (46%) Medical history Coronary revascularisation (PCI) 262 (34%) 298 (39%) Coronary artery bypass surgery 134 (18%) 145 (19%) TAVI 23 (3%) 30 (4%) Mitral clip 26 (3%) 34 (4%) Cardiac surgery for valves 86 (11%) 71 (9%) Implantable cardioverter 222 (29%) 234 (30%) defibrillator Cardiac resynchronisation therapy 118 (15%) 122 (16%) Ablation of pulmonary veins 71 (9%) 52 (7%) Laboratory measurements Haemoglobin (mmol/L) 8 (7-9) 8 (8-9) Serum sodium (mmol/L) 140 (137-142) 140 (138-142) Potassium (mmol/L) 5 (4-5) 5 (4-5) Serum creatinine (μmol/L) 108 (87-141) 109 (88-148) Estimated GFR (mL/min per 60 (43-88) 60 (42-84) 1.73 m.sup.2 of body surface area, Cockcroft-Gault) NT-proBNP (pg/mL) 1407 1488 (626-3142) (594-3069) In patients with LVEF ≤45 1728 1798 (n = 1001) (798-3858) (786-3667) In patients with LVEF >45 1056 1035 (n = 537) (468-2042) (405-1985) MR-proADM (nmol/L) 1 (1-2) 1 (1-2) Concomitant treatment ACE inhibitors or ARBs 628 (82%) 641 (83%) ARN inhibitors 44 (6%) 47 (6%) β blockers 702 (92%) 711 (92%) Aldosterone antagonists 441 (58%) 405 (52%) Loop diuretics 717 (94%) 721 (93%) Thiazides 191 (25%) 185 (24%) Other diuretics 4 (1%) 1 (0%) Vitamin K antagonists 265 (35%) 272 (35%) Antiplatelet therapy 103 (13%) 130 (17%) NOACs 205 (27%) 208 (27%) Platelet aggregation inhibitors 266 (35%) 267 (35%) Lipid-lowering drugs 456 (60%) 453 (59%) Insulin 170 (22%) 170 (22%) Oral hypoglycaemic drugs 206 (27%) 185 (24%) Ivabradine 22 (3%) 43 (6%) Calcium antagonists 163 (21%) 188 (24%) Nitrates 37 (5%) 48 (6%) Digitalis glycosides 119 (16%) 133 (17%) Antiarrhythmic drugs 99 (13%) 98 (13%) Data are mean (SD) or n (%), median (IQR) for all laboratory tests. NYHA = New York Heart Association. LVEF = left ventricular ejection fraction. PCI = percutaneous coronary intervention. TAVI = transcatheter aortic valve implantation. GFR = glomerular filtration rate. NT-proBNP = N-terminal prohormone of brain natriuretic peptide. MR-proADM = mid-regional proadrenomedullin. ACE = angiotensin-converting enzyme. ARB = angiotensin-receptor blocker. ARN = angiotensin receptor-neprilysin. NOAC = novel oral anticoagulant.
TABLE-US-00008 TABLE 4 Primary and key secondary outcomes Ratio, Remote patient management (n = 765) Usual care (n = 773) remote patient Number Weighted Number Weighted management of patients average of patients average vs usual with event (95% Cl) with event (95% Cl) care (95% Cl) p value Percentage of 265 (35%) 4.88% 290 6.64% 0-80* 0.0460 days lost due (4.55-5.23) (38%) (6.19-7.13) (0.65-1.00) to unplanned cardiovascular hospitalisation or death of any cause Days lost . . . 17.8 days . . . 24.2 days . . . . . . per year (16.6-19.1) (22.6-26.0) All-cause 61 (8%) 7.86 89 11.34 0.70‡ 0.0280 mortality† (6.14-10.10) (12%) (9.21-13.95) Cardiovascular 39 (5%) 5.04 59 7.51 0.67‡ 0.0560 moriality† (3.68-6.90) (8%) (5.82-9.70) (0.45-1.01) .sup.* Ratio of the weighted average. .sup.† Measured during induvidual patient follow-up time plus 28 days after the study visit, to a maximum of 393 days. .sup.‡ Hazard ratio.
TABLE-US-00009 TABLE 5 Remote patient management (n = 465) Usual care (n = 773) Mean difference* Patients (n) Mean (95% Cl) Patients (n) Mean (95% Cl) (95% Cl) p value Quality of life Change in MLHFQ 649 −3.08 624 −1.98 −1.11 0.26 global score from baseline to 12 months† (−4.42 to (−3.34 to (−3.01 to −1.75) −0.61) 0.80) Biomarker values Change in NT- 654 −24.66% 628 −18.72% −7.31% 0.13 proBNP (pg/mL) (−29.68 to (−24.28 to (−16.03 to from baseline to 19.29) −12.75) 2.31) 12 months† In patients with 423 −34.30% 410 −27.16% −9.80% 0.11 LVEF ≤ 45% (−39.94 to (−33.51 to (−20.64 to 28.12) −20.20) 2.52) In patients with 241 −3.71% 218 −0.68% −3.04% 0.68 LVEF > 45% (−12.99 to (−10.73 to (−16.32 to 6.56) 10.49) 12.33) Change in MR- 665 8.44% 628 3.76% 4.50% 0.0084 proADM (nmol/L) (5.99 to (1.35 to (l.14 to from baseline to 10.94) 6.23) 7.98) 12 months† .sup.MLHFQ = Minnesota Living with Heart Failure Questionnaire. .sup.NT-proBNP = N-terminal prohormone of brain natriuretic peptide. .sup.LVEF = left ventricular ejection fraction. .sup.MR-proADM = mid-regional proadrenomedullin. .sup.* Mean difference in change in the remote patient management group vs change in the usuall care group. .sup.† Data obtained at final study visit performed at a maximum of 393 days after randomisation.
TABLE-US-00010 TABLE 6 Selected interventions of TMC physicians and nurses in the remote management group Number of Median (range) interventions per patient Evaluation of patient-transmitted vital parameters* 1 026 078 1421.0 (6-3962) Patient case review by TMC physicians and nurses 38 694 36.0 (0-273) Monthly structured telephone interview 9189 12.0 (1-13) TMC initiated contact with patient for evaluation of key vital 4324 4.0 (0-37) parameters TMC initiated contact with patient after discharge, physician 6037 7.0 (1-27) appointment, and for validation of medication list TMC initiated medication changes 3546 3.0 (0-57) TMC initiated scheduled 3-month medical report sent to patient's 2812 4.0 (0-4) local physician (GP or cardiologist) TMC physician and patient telephone consultations 1535 1.0 (0-40) TMC initiated contact with health-care professionals 863 0.0 (0-21) Patient home heart failure education including caregivers 765 1.0 (1-1) TMC initiated emergency department visits 30 NA TMC initiated unplanned cardiovascular hospital admissions 57 NA TMC initiated unplanned non-cardiovascular hospital admissions 13 NA *Vital parameters are bodyweight, blood pressure, self-rated health status, and electrocardiogram including peripheral capillary oxygen saturation. TMC = telemedicine centre. GP = general practitioner. NA = not applicable; only the total number is known, and not the median per patient.
TABLE-US-00011 TABLE 7 Cutoff Analysis for proADM Cutoff Sensitivity Specificity NPV PPV PredNeg PredPos 0.75 nmol/L 100 18.8 100 12.7 16.8 83.2 0.86 nmol/L 95.1 30.2 98.1 13.8 27.5 72.5 0.91 nmol/L 91.4 35.7 97.2 14.4 32.9 67.1 0.98 nmol/L 86.4 42.7 96.4 15.1 39.6 60.4 1.07 nmol/L 80.2 51.2 95.6 16.2 47.8 52.2
TABLE-US-00012 TABLE 8 Cutoff Analysis for proBNP Cutoff Sensitivity Specificity NPV PPV PredNeg PredPos 237.6 pg/ml 100 8.9 100 11.5 7.9 92.1 609.4 pg/ml 95.1 27.9 98 13.5 25.5 74.5 1107.9 pg/ml 90.1 42.9 97.4 15.7 39.5 60.5 1402.95 pg/ml 85.2 51.8 96.7 17.2 47.9 52.1 1595.8 pg/ml 80.2 55.9 96 17.7 52.1 47.9
TABLE-US-00013 TABLE 9 Cutoff Analysis for proANP Cutoff Sensitivity Specificity NPV PPV PredNeg PredPos 106.9 pmol/L 100 9.9 100 11.5 8.9 91.1 158.8 pmol/L 93.8 23.9 97 12.6 22.1 77.9 186.2 pmol/L 90 33.7 96.7 13.7 31.2 68.8 235.6 pmol/L 85 45.9 96.3 15.5 42.7 57.3 248.3 pmol/L 80 49.9 95.5 15.7 46.7 53.3
TABLE-US-00014 TABLE 10 Stratification of biomarker cutoffs by a GFR based upon CKD-EPI % % patients patients Event excluded excluded rate of Event from RPM from RPM excluded rate of w/o Arm Total ADM ADM BNP BNP with SNP, pats included using only sensitivity Cutoff Cutoff Cutoff Cutoff ADM and (100- pats BNP and (%) (GFR<) (GFR>=) (GFR<) (GFR>=) GFR NPV) (PPV) GFR GFR 50 strata 100 0.98 0.53 237.6 42.4 6.4 0 10.5 2 99 0.98 0.62 237.6 140.2 15.1 1 11.4 7.6 95 1.06 0.68 273.7 281.6 23.4 1.9 12.3 19.5 90 1.1 0.73 564.4 459.8 33.9 3 13.4 30.7 86 1.13 0.77 714.9 526.2 39.3 3.6 13.9 41.9 81 1.24 0.86 903.2 774.6 51.1 3.7 18.3 50 GFR 60 strata 100 0,97 0.53 237.6 42.4 9.3 0 10.9 2.7 99 0.98 0.57 273.7 137.7 15.4 1 11.5 7.9 95 1.05 0.63 402.6 266.7 24.9 1.8 12.5 18.5 90 1.07 0.7 585.2 358.1 33 3 13.3 30 86 1.1 0.73 842.2 460.1 38.2 3.5 13.8 38.9 80 1.14 0.78 1082 590.8 46.5 4.2 14.8 50
TABLE-US-00015 TABLE 11 Stratification of biomarker cutoffs by a GFR based upon Cockroft-Gault % patients % patients excUed excluded from Event from RPM RPM rate Event w/o ACM Total ADM ADM BNP BNP with BNP, excluded rate using only sensitivity Cutoff Cutoff Cutoff Cutoff ADM and (100- included BNP and (%) (GFR<) (GFR>=) (GFR<) (GFR>=) GFR NPV) PPV GFR GFR 50 strata 100 0.62 0.53 937.6 42.4 4.5 0 10.3 2.2 99 0.84 0.57 273.7 137.7 11.6 1.3 11 7.4 95 1.05 0.63 454.4 203.9 21.7 2.1 12 17.1 89 1.08 0.7 842.2 298.4 30.8 3.4 12.7 30.5 86 1.14 0.73 1007 457.1 3.7 13.8 80 1.23 0.75 1364 502.2 45.4 4.3 14.5 45.6 GFR 60 strata 100 0.62 0.53 137.7 42.4 4.2 0 10.3 1.5 99 0.75 0.53 203.9 42.4 6.6 1.1 10.5 2.5 95 0.92 0.63 402.6 273.6 24.9 2.1 12.4 16.5 90 0.97 0.68 585.2 298.4 31.3 3.1 12.9 30 85 1.05 0.7 935.5 457.1 40.9 3.6 14.2 38.8 80 1.08 0.73 1268 502.2 46.3 4.2 14.7 43.4
TABLE-US-00016 TABLE 12 Exclusion based on SOC outcomes MR- NT- SOC patients RPM vs. SOC outcomes Sensi- proADM proBNP not for RPM- Follow- tivity cutoff cutoff recommended Hazard Biomarker Endpoint up (%) [nmol/L] [ng/L] for RPM [%] Ratio p-value NNT combined All-cause 1 year 100 0.69 125.1 13.7 0.71 0.0423 26 Death combined All-cause 1 year 97.7 0.72 145.4 16.9 0.71 0.0467 26 Death combined All-cause 1 year 95.5 0.75 383.3 25.3 0.71 0.0425 23 Death
TABLE-US-00017 TABLE 13 Gain in Gain in specificity specificity when when using using MR- Optimized MR- Optimized NT- Specificity when GFR in proADM in proADM proBNP cutoff using GFR strata addition to addition to NT- cutoff for 100% for 100% in addition to both proBNP and sensitivity sensitivity biomarkers biomarkers GFR GFR >= 30 0.75 383.3 28.3% +4.5%-points +11.8%-points GFR < 30 1.14 237.6
TABLE-US-00018 TABLE 14A Average % lost days due to unplanned CV hospitalization for the different biomarker quintiles for SOC and RPM subgroups. P-values for the treatment effect, as well as for the interaction of treatment with quintile are presented. Lost days Lost days A Lost days p-value p-value Biomarker Quintiles* SOC RPM SOC-RPM SOC vs. RPM Interaction NT-proBNP 0-488 0.95 1.56 −0.62 0.40 [pg/ml] 488-1099 2.38 2.74 0.36 0.56 1099-1880 5.09 2.21 2.88 0.05 0.32 1880-3701 7.61 5.33 2.29 0.36 3701-35000 17.28 12.84 4.45 0.11 MR- 0-0.75 1.35 1.39 −0.04 0.91 proADM 0.75-0.95 4.05 1.29 2.76 0.05 [nmol/L] 0.98-1.20 3.47 3.60 −0.13 0.57 0.68 1.20-1.58 6.66 6.78 −0.12 0.56 1.58-7.82 17.64 12.13 5.51 0.21 .sup.* upper limits are included in respective Quintiles, lower limits not
TABLE-US-00019 TABLE 14B Rate of all-cause death for the different biomarker quintiles for SOC and RPM subgroups. P-values for the treatment effect, as well as for the interaction of treatment with quintile are presented. Eventrate Eventrate Hazard ratio p-value p-value Biomarker Quintiles* SOC (%) RPM (%) SOC/RPM SOC vs. RPM Interaction NT-proBNP 0-488 1.9 2.6 1.35 0.70 0.71 [pg/ml] 488-1099 4.6 5.1 1.10 0.86 1099-1880 5.9 3.2 0.56 0.30 1880-3701 14.7 8.4 0.55 0.10 3701-35000 30.5 20.4 0.68 0.09 MR- 0-0.75 2.1 2.3 1.11 0.89 0.40 proADM 0.75-0.95 5.8 1.3 0.21 0.05 [nmol/L] 0.98-1.20 7.2 6.0 0.85 0.73 1.20-1.58 12.4 10.8 0.90 0.76 1.58-7.82 29.9 20.3 0.67 0.08 .sup.* upper limits are included in respective Quintiles, lower limits not
TABLE-US-00020 TABLE 15 Biomarker-based RPM recommendation scenarios and corresponding RPM effects for primary and secondary endpoints. Endpoint for selection: ≥ 30 lost days/year all-cause death Endpoint Sensitivity: 100% 98% 95% 100% 98% 95% Selection Cutoff MR- 0.63 0.72 0.75 0.69 0.72 0.75 scenarios proADM of patients [nmol/L] recommended Cutoff NT- 125.1 145.4 413.7 125.1 145.4 383.3 for RPM proBNP [pg/ml] Sensitivity (%) 100 97.64 95.28 100 97.73 95.45 Specificity (%) 12.87 19.68 31.42 15.7 18.98 28.36 Excluded (%) 10.75 16.83 27.04 13.87 17.04 25.59 NPV (%) 100 97.69 97.13 100 98.45 97.94 PPV (%) 18.43 19.31 21.45 13.5 13.69 14.89 Evaluation Days lost Average 8.68 7.75 7.41 8.53 7.75 7.64 of RPM due to SOC (%) intervention unplanned Average 6.28 5.95 5.52 6.35 5.95 5.80 effect: cardiovascular RPM (%) RPM vs. hospital Ratio 0.78 0.83 0.80 0.80 0.83 0.81 SOC for admissions p-value 0.082 0.146 0.052 0.089 0.146 0.106 subset of or all- patients cause death recommended All-cause Hazard ratio 0.68 0.72 0.70 0.71 0.72 0.71 for RPM death 95%-Cl low 0.48 0.51 0.50 0.50 0.51 0.52 (time to 95%-Cl high 0.96 1.00 0.97 0.99 1.00 0.99 event) p-value 0.027 0.048 0.031 0.043 0.048 0.044 Cardiovascular Hazard ratio 0.64 0.69 0.67 0.68 0.69 0.68 death 95%-Cl low 0.42 0.45 0.44 0.45 0.45 0.45 (time to event) 95%-Cl high 0.99 1.04 1.00 1.03 1.04 1.03 % days p-value 0.042 0.071 0.051 0.066 0.071 0.065 lost due to Average 2.77 2.52 2.41 2.73 2.52 2.46 unplanned SOC (%) cardiovascular Average 2.11 2.00 1.87 2.10 2.00 1.94 hospital RPM (%) admissions Ratio 0.90 0.92 0.89 0.91 0.92 0.91 p-value 0.341 0.455 0.250 0.394 0.455 0.349 % days Average 1.90 1.73 1.63 1.87 1.73 1.68 lost due to SOC (%) unplanned hospital admissions due to worsening heart failure
TABLE-US-00021 TABLE 16a Characteristics of biomarkers based selected patients by randomized treatment groups. All patients RPM patients SOC patients *p Variable class N % N % N % value Sex female 337 30.7 168 31.5 169 30 0.637 Age [years] mean (SD) 72.6 (9.2) 72.9 (9.1) 72.3 (9.3) 0.296 NYHA class I 1 0.1 0 0 1 0.2 0.730 II 501 45.6 239 44.8 262 46.5 0.730 III 592 53.9 293 54.9 299 53 0.730 IV 4 0.4 2 0.4 2 0.4 0.730 Living in a rural 630 57.4 313 58.6 317 56.2 0.456 urban area vs rural area Living alone 317 28.9 153 28.7 164 29.1 0.929 LVEF 40 < 50 254 23.1 112 21 142 25.2 0.200 ≥50 336 30.6 173 32.4 163 28.9 0.200 EF < 40 508 46.3 249 46.6 259 45.9 0.200 Discharge from last heart 727 66.2 359 67.2 368 65.2 0.556 failure admission and randomization <3 months Primary cause Dilated cardio- 219 19.9 105 19.7 114 20.2 0.181 of heart failure myopathy Hypertension 184 16.8 81 15.2 103 18.3 0.181 Ischaemic 491 44.7 236 44.2 255 45.2 0.181 Other 204 18.6 112 21 92 16.3 0.181 Current smoker 75 6.8 42 7.9 33 5.9 0.333 Hyperlipidemia 632 57.6 313 58.6 319 56.6 0.288 Diabetes mellitus 537 48.9 265 49.6 272 48.2 0.687 Coronary revascularisation 434 39.5 201 37.6 233 41.3 0.360 Coronary artery bypass surgery 228 20.8 114 21.3 114 20.2 0.697 TAVI 45 4.1 19 3.6 26 4.6 0.421 Mitral clip 55 5 24 4.5 31 5.5 0.647 Prior cardiac valve surgery 126 11.5 68 12.7 58 10.3 0.258 ICD 356 32.4 168 31.5 188 33.3 0.550 CRTd 202 18.4 97 18.2 105 18.6 0.580 ACE inhibitors or ARBs 882 80.3 427 80 455 80.7 0.826 ARN inhibitors 75 6.8 34 6.4 41 7.3 0.636 13 blockers 1019 92.8 491 91.9 528 93.6 0.341 Aldosterone antagonists 575 52.4 292 54.7 283 50.2 0.152 Loop diuretics 1056 96.2 514 96.3 542 96.1 1.000 Vitamin K antagonists 445 40.5 217 40.6 228 40.4 0.992 Antiplatelet therapy 169 15.4 76 14.2 93 16.5 0.341 NOACs 306 27.9 148 27.7 158 28 0.966 Digitalis glycosides 218 19.9 100 18.7 118 20.9 0.403 Antiarrhythmic drugs 144 13.1 71 13.3 73 12.9 0.933 Median (IQR) MR-proADM [mmol/L] 1.22 1.23 1.21 0.598 (0.97-1.65) (0.96-1.68) (0.97-1.63) NT-proBNP [pg/ml] 1980 1918 2048 0.650 (1107-3927) (1075-4119) (1146-3779)
TABLE-US-00022 TABLE 16b Characteristics of biomarkers based selected patients versus not selected patients All patients Not RPM RPM *p Variable class N % N % N % value Sex female 467 30.4 130 29.7 337 30.7 0.743 Age [years] mean (SD) 70.3 (10.5) 64.6 (11.3) 72.6 (9 2) <0.001 NYHA class I 11 0.7 10 2.3 1 0.1 <0.001 II 795 51.8 294 67.1 501 45.6 <0.001 III 725 47.2 133 30.4 592 53.9 <0.001 IV 5 0.3 1 0.2 4 0.4 <0.001 Living in a urban rural 915 59.6 285 65.1 630 57.4 0.007 area vs rural area Living alone 434 28.3 117 26.7 317 28.9 0.432 LVEF 40 < 50 372 24.2 118 26.9 254 23.1 0.004 ≥50 494 32.2 158 36.1 336 30.6 0.004 EF < 40 670 43.6 162 37 508 46.3 0.004 Discharge from last heart 956 62.2 229 52.3 727 66.2 <0.001 failure admission and randomization <3 months Primary cause of Dilated cardio- 347 22.6 128 29.2 219 19.9 <0.001 heart failure myopathy Hypertension 274 17.8 90 20.5 184 16.8 <0.001 Ischaemic 624 40.6 133 30.4 491 44.7 <0.001 Other 291 18.9 87 19.9 204 18.6 <0.001 Current smoker 134 8.7 59 13.5 75 6.8 <0.001 Hyperlipidemia 832 54.2 200 45.7 632 57.6 <0.001 Diabetes mellitus 701 45.6 164 37.4 537 48.9 <0.001 Coronary revascularisation 559 36.4 125 28.5 434 39.5 <0.001 Coronary artery bypass surgery 279 18.2 51 11.6 228 20.8 <0.001 TAVI 53 3.5 8 1.8 45 4.1 0.031 Mitral clip 60 3.9 5 1.1 55 5 0.002 Cardiac surgery for valves 157 10.2 31 7.1 126 11.5 0.013 ICD 456 29.7 100 22.8 356 32.4 <0.001 CRTd 240 15.6 38 8.7 202 18.4 <0.001 ACE inhibitors or ARBs 1269 82.6 387 88.4 882 80.3 <0.001 ARN inhibitors 91 5.9 16 3.7 75 6.8 0.024 13 blockers 1411 91.9 392 89.5 1019 92.8 0.042 Aldosterone antagonists 845 55 270 61.6 575 52.4 0.001 Loop diuretics 1436 93.5 380 86.8 1056 96.2 <0.001 Vitamin K antagonists 536 34.9 91 20.8 445 40.5 <0.001 Antiplatelet therapy 233 15.2 64 14.6 169 15.4 0.760 NOACs 413 26.9 107 24.4 306 27.9 0.191 Calcium antagonists 350 22.8 98 22.4 252 23 0.860 Digitalis glycosides 252 16.4 34 7.8 218 19.9 <0.001 Antiarrhythmic drugs 197 12.8 53 12.1 144 13.1 0.651 Median (IQR) MR-proADM [mmol/L] 1.1 (0.8-1.46) 0.69 (0.6-0.84) 1.2 (0.97-1.7) <0.001 NT-proBNP [pg/ml] 1436 359 1980 <0.001 (605-3097) (200-983) (1107-3927) SD, standard deviation; .sup.*p-value RPM versus SOC; LVEF, left ventricular ejection fraction
TABLE-US-00023 TABLE 17 Median and interquartile range (IQR) of LVEF across biomarker quintiles in the TIM-HF2 population. NT-proBNP concentration was correlated with LVEF (Spearman ρ = −0.30, ρ < 0.001), while MR-proADM was not (Spearman ρ = −0.01, ρ = 0.70). Biomarker Quintile * LVEF median (IQR) NT-proBNP 0-488 45 (40-55) [pg/ml] 488-1099 45 (35-55) 1099-1880 40 (30-50) 1880-3701 40 (30-50) 3701-35000 32 (25-42) MR-proADM 0-0.75 40 (30-48) [nmol/L] 0.75-0.95 44 (32-54) 0.98-1.20 40 (30-51) 1.20-1.58 40 (30-52) 1.58-7.82 40 (30-50) * upper limits are included in respective quintiles, lower limits not
TABLE-US-00024 TABLE 18 Biomarker-based recommendation scenarios with different safety levels (Table part 1: “Selection scenarios of patients recommended for RPM”) and evaluation of the RPM intervention effect (Table part 2: “Evaluation of RPM intervention effect: RPM vs. SOC for subset of patients recommended for RPM”) for more strict binary event definitions of the primary endpoint % days lost due to unplanned cardiovascular hospital admission or all-cause death used to derive cutoffs for biomarker based selection of patients with RPM recommendation 15 lost days/year, 1 lost days/year); analogous to Table 15 (see above, 30 lost days/year). Endpoint for selection: Targeted ≥15 lost days/year ≥1 lost days/year Endpoint sensitivity: 100% 98% 95% 100% 98% 95% Part 1: Cutoff NT- 0.58 0.63 0.70 0.48 0.58 0.66 proBNP Selection [nmol/L] scenarios of Cutoff MR- 125.1 145.4 383.3 52.7 112.4 214.2 proADM patients [pg/ml] recommended Sensitivity (%) 100 98.8 94.7 100 97.9 94.5 for RPM Specificity (%) 10.3 14.4 27.8 2.7 10.8 20.3 Excluded (%) 8.0 11.5 22.9 1.7 7.5 14.8 NPV (%) 100 97.8 94.9 100 89.7 86.0 PPV (%) 23.8 24.5 26.9 38.1 39.6 41.4 Part 2: Days lost due to Average 7.21 7.43 8.28 6.76 7.17 7.66 Evaluation unplanned SOC (%) of RPM cardiovascular Average 5.30 5.51 6.19 5.01 5.23 5.79 intervention hospital RPM (%) effect: admissions or all- Ratio 0.79 0.80 0.80 0.80 0.78 0.82 RPM vs. SOC cause death p-value 0.042 0.068 0.098 0.044 0.033 0.099 for subset All-cause death of patients (time to event) recommended Hazard ratio 0.69 0.70 0.70 0.70 0.69 0.72 for RPM (0.50- (0.50- (0.50- (0.51- (0.49- (0.52- 0.96) 0.97) 0.98) 0.97) 0.95) 1.00) p-value 0.027 0.032 0.035 0.032 0.024 0.050 Cardiovascular Hazard ratio 0.66 0.67 0.67 0.68 0.66 0.68 death (95%-Cl) (0.44- (0.44- (0.45- (0.45- (0.44- (0.46- (time to event) 0.99) 1.01) 1.02) 1.01) 0.99) 1.03) p-value 0.045 0.054 0.059 0.056 0.041 0.066 Days lost due to Average 2.36 2.44 2.68 2.23 2.35 2.50 unplanned SOC (%) cardiovascular Average 1.80 1.89 2.12 1.70 1.77 1.97 hospital RPM (%) admissions Ratio 0.88 0.90 0.91 0.88 0.87 0.91 p-value 0.20 0.28 0.38 0.20 0.16 0.35 Days lost due to Average 1.59 1.65 1.84 1.49 1.58 1.70 unplanned SOC (%) hospital Average 1.11 1.17 1.31 1.03 1.09 1.23 admissions due RPM (%) to worsening Ratio 0.80 0.80 0.80 0.79 0.79 0.80 heart failure p-value 0.007 0.016 0.034 0.003 0.006 0.016
REFERENCES
[0596] Abraham W T, Adamson P B, Bourge R C, Aaron M F, Costanzo M R, Stevenson L W, Strickland W, Neelagaru S, Raval N, Krueger S, Weiner S, Shavelle D, Jeffries B, Yadav J S, CHAMPION Trial Study Group. Wireless pulmonary artery haemodynamic monitoring in chronic heart failure: a randomised controlled trial. Lancet 2011; 377:658-666. [0597] Adamson P B, Ginn G, Anker S D, Bourge R C, Abraham W T. Remote haemodynamic-guided care for patients with chronic heart failure: a meta-analysis of completed trials. Eur J Heart Fail 2017; 19:426-443. [0598] Albrich W C, Ruegger K, Dusemund F, et al. (2013) Biomarker-enhanced triage in respiratory infections: a proof-of-concept feasibility trial. Eur Respir J; 42(4):1064-1075. [0599] Allen, M., Poggiali, D., Whitaker, K., Marshall, T. R., & Kievit, R. A. (2019). Raincloud plots: a multi-platform tool for robust data visualization. Wellcome Open Research, 4 (https://doi.org/10.12688/wellcomeopenres.15191.1). [0600] Anker S, Koehler F, Abraham W. Telemedicine and remote management of patients with heart failure. Lancet 2011; 378:731-739. [0601] Andrès E., Talha S., Zulfiqar A A, Hajjam M, Ervé S, Hajjam J, Gény B, Hajjam E, Hassani A. (2018) Current Research and New Perspectives of Telemedicine in Chronic Heart Failure: Narrative Review and Points of Interest for the Clinician. J Clin Med. Dec. 13, 2018; 7 (12). [0602] Böhm M, Drexler H, Oswald H, Rybak K, Bosch R, Butter C, Klein G, Gerritse B, Monteiro J, Israel C, Bimmel D, Käab S, Huegl B, Brachmann J; OptiLink H F Study Investigators. Fluid status telemedicine alerts for heart failure: a randomized controlled trial. Eur Heart J 2016; 37:3154-3163. [0603] Chioncel O, Lainscak M, Seferovic P M, et al. Epidemiology and one-year outcomes in patients with chronic heart failure and preserved, mid-range and reduced ejection fraction: an analysis of the ESC Heart Failure Long-Term Registry. Eur J Heart Fail 2017; 19: 1574-85. [0604] Christopher Adlbrecht, Martin Hülsmann, Guido Strunk, Rudolf Berger, Deddo Mörtl, Joachim Struck, Nils G. Morgenthaler, Andreas Bergmann, Johannes Jakowitsch, Gerald Maurer, Irene M. Lang, and Richard Pacher (2009) European Journal of Heart Failure (2009) 11, 361-366 [0605] Cockcroft, D. W., & Gault, M. H. (1976). Prediction of creatinine clearance from serum creatinine. Nephron, 16, 31-41. [0606] Cowie M R, Anker S D, Cleland J G F, et al. Improving care for patients with acute heart failure: before, during and after hospitalization. ESC Heart Fail 2014; 1: 110-45. [0607] Cowie M R, Bax J, Bruning N, et al. e-Health: a position statement of the European Society of Cardiology. Eur Heart J 2016; 37: 63-66. [0608] Francis G S, Felker G M, Tang W H. (2016) A test in context: critical evaluation of natriuretic peptide testing in heart failure. J Am Coll Cardiol; 67:331-7. [0609] Gheorghiade M, Zannad F, Sopko G, Klein L, Piña I L, Konstam M A, Massie B M, Roland E, Targum S, Collins S P, Filippatos G, Tavazzi L; International Working Group on Acute Heart Failure Syndromes. International working group on acute heart failure syndromes. Acute heart failure syndromes: current state and framework for future research. Circulation. 2005; 112(25):3958-68. [0610] Gowda S., Prakash B. Desai, Shruthi S. Kulkarni, Vinayak V. Hull, Avinash A. K. Math, and Sonal N. Vernekar Markers of renal function tests, N Am J Med Sci. 2010 April; 2(4): 170-173. [0611] Hartmann O, Schuetz P, Albrich W C, Anker S D, Mueller B, Schmidt T (2012). Time-dependent Cox regression: serial measurement of the cardiovascular biomarker proadrenomedullin improves survival prediction in patients with lower respiratory tract infection. Int J Cardiol; 1 61 (3):166-173. [0612] Hindricks G, Taborsky M, Glikson M, Heinrich U, Schumacher B, Katz A, Brachmann J, Lewalter T, Goette A, Block M, Kautzner J, Sack S, Husser D, Piorkowski C, Sogaard P; IN-TIME Study Group. Implant-based multiparameter telemonitoring of patients with heart failure (IN-TIME): a randomised controlled trial. Lancet 2014; 384:583-590. [0613] Hofmann R, Voeller H, Nagels K, Bindl D, Vettorazzi E, Dittmar R, Wohlgemuth W, Neumann T, Stork S, Bruder O, Wegscheider K, Nagel E, Fleck E. First outline and baseline data of a randomized, controlled multicenter trial to evaluate the health economic impact of home telemonitoring in chronic heart failure—CardioBBEAT. Trials 2015; 16:343. [0614] Hunt S A, Abraham W T, Chin M H, Feldman A M, Francis G S, Ganiats T G, et al. 2009 focused update incorporated into the ACC/AHA 2005 Guidelines for the Diagnosis and Management of Heart Failure in Adults: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines: developed in collaboration with the International Society for Heart and Lung Transplantation. Circulation. 2009; 119 (14). [0615] Jougasaki M, Burnett Jr J C (2000). Adrenomedullin: potential in physiology and pathophysiology. Life Sci; 66:855-72. [0616] Koehler F, Koehler K, Deckwart O, Prescher S, Wegscheider K, Winkler S, Vettorazzi E, Polze A, Stangl K, Hartmann O, Marx A, Neuhaus P, Scherf M, Kirwan B A, Anker S D. Telemedical Interventional Management in Heart Failure II (TIM-HF2), a randomised, controlled trial investigating the impact of telemedicine on unplanned cardiovascular hospitalisations and mortality in heart failure patients: study design and description of the intervention. Eur J Heart Fail. 2018 October; 20(10):1485-1493. doi: 10.1002/ejhf.1300. Epub Sep. 19, 2018b. [0617] Koehler F, Winkler S, Schieber M, Sechtem U, Stangl K, Böhm M, de Brouwer S, Perrin E, Baumann G, Gelbrich G, Boll H, Honold M, Koehler K, Kirwan B A, Anker S D. Telemedicine in heart failure: pre-specified and exploratory subgroup analyses from the TIM-HF trial. Int J Cardiol 2012a; 161:143-150. [0618] Koehler F, Winkler S, Schieber M, Sechtem U, Stangl K, Böhm M, Brouwer S, Perrin E, Baumann G, Goetz G, Boll H, Honold M, Koehler K, Kirwan B A, Anker S, “Telemedicine in heart failure: Pre-specified and exploratory subgroup analyses from the TIM-HF trial”, Int J Cardiology 2012b, 161, 3 143-150. [0619] Koehler F, Winkler S, Schieber M, Sechtem U, Stangl K, Böhm M, Boll H, Baumann G, Honold M, Koehler K, Gelbrich G, Kirwan B A, Anker S D; Telemedical Interventional Monitoring in Heart Failure Investigators. Impact of remote telemedical management on mortality and hospitalisations in ambulatory patients with chronic heart failure: the Telemedical Interventional Monitoring in Heart Failure study. Circulation 2011; 123:1873-1880. [0620] Koehler, Friedrich & Koehler, Kerstin & Deckwart, Oliver & Prescher, Sandra & Wegscheider, Karl & Kirwan, Bridget-Anne & Winkler, Sebastian & Vettorazzi, Eik & Bruch, Leonhard & Oeff, Michael & Zugck, Christian & Doerr, Gesine & Naegele, Herbert & Störk, Stefan & Butter, Christian & Sechtem, Udo & Angermann, Christiane & Gola, Guntram & Prondzinsky, Roland & Stangl, Karl (2018a). Efficacy of telemedical interventional management in patients with heart failure (TIM-HF2): a randomised, controlled, parallel-group, unmasked trial. The Lancet. Volume 392, ISSUE 10152, P1047-1057, Sep. 22, 2018a. [0621] Koehler F., Koehler K., Deckwart O., Prescher S., Wegscheider K., Winkler S., Vettorazzi E., Polze A., Stangl K., Hartmann O., Marx A., Neuhaus P., Scherf M., Kirwan B A, Anker S D (2018b) Telemedical Interventional Management in Heart Failure II (TIM-HF2), a randomised, controlled trial investigating the impact of telemedicine on unplanned cardiovascular hospitalisations and mortality in heart failure patients: study design and description of the intervention. Eur J Heart Fail. 2018 October; 20(10):1485-1493 [0622] Kojima H, Tsujimoto T, Uemura M, et al. (1998) Significance of increased plasma adrenomedullin concentration in patients with cirrhosis. J Hepatol; 28:840-6. [0623] Krstic, Danijela & Tomić, Nenad & Radosavljević, Branimir & Avramović, Nataša & Dragutinovic, Vesna & Radojević Škodrić, Sanja & Colović, Mirjana. (2016). Biochemical Markers of Renal Function. Current medicinal chemistry. 23: 2018-2040. [0624] Levey A. S., Inker L. A. Assessment of Glomerular Filtration Rate in Health and Disease: A State of the Art Review, Clin Pharmacol Ther. 2017 September; 102(3):405-419. [0625] Levey, A. S., et al. (2009). A new equation to estimate glomerular filtration rate. Annals of internal medicine, 150(9), 604-612. [0626] Marco Metra, M D John R Teerlink, M D The LANCET: Volume 390, ISSUE 10106, P1981-1995 (2017). [0627] McMurray J J, Adamopoulos S, Anker S D, Auricchio A, Böhm M, Dickstein K, Falk V, Filippatos G, Fonseca C, Gomez-Sanchez M A, Jaarsma T, Køber L, Lip G Y, Maggioni A P, Parkhomenko A, Pieske B M, Popescu B A, Rønnevik P K, Rutten F H, Schwitter J, Seferovic P, Stepinska J, Trindade P T, Voors A A, Zannad F, Zeiher A; Task Force for the Diagnosis and Treatment of Acute and Chronic Heart Failure 2012 of the European Society of Cardiology, Bax J J, Baumgartner H, Ceconi C, Dean V, Deaton C, Fagard R, Funck-Brentano C, Hasdai D, Hoes A, Kirchhof P, Knuuti J, Kolh P, McDonagh T, Moulin C, Popescu B A, Reiner Z, Sechtem U, Sirnes P A, Tendera M, Torbicki A, Vahanian A, Windecker S, McDonagh T, Sechtem U, Bonet L A, Avraamides P, Ben Lamin H A, Brignole M, Coca A, Cowburn P, Dargie H, Elliott P, Flachskampf F A, Guida G F, Hardman S, lung B, Merkely B, Mueller C, Nanas J N, Nielsen. [0628] Melillo P and Pecchia L. “A Preliminary Model to Choose the Most Appropriate Target Population for Home Monitoring Telemedicine Interventions Basing on the Best Available Evidence”, COMPUTER OF IMAGES AND PATTERNS (CAIP) 2014; 406-408. [0629] Morgan J M, Kitt S, Gill J, McComb J M, Ng G A, Raftery J, Roderick P, Seed A, Williams S G, Witte K K, Wright D J, Harris S, Cowie M R. Remote management of heart failure using implantable electronic devices. Eur Heart J 2017; 38:2352-2360. [0630] Mukoyama et al, (1991) Brain natriuretic peptide as a novel cardiac hormone in humans. Evidence for an exquisite dual natriuretic peptide system, atrial natriuretic peptide and brain natriuretic peptide. J Clin Invest; 87: 1402-12 [0631] Ong M K, Romano P S, Edgington S, Aronow H U, Auerbach A D, Black J T, De Marco T, Escarce J J, Evangelista L S, Hanna B, Ganiats T G, Greenberg B H, Greenfield S, Kaplan S H, Kimchi A, Liu H, Lombardo D, Mangione C M, Sadeghi B, Sadeghi B, Sarrafzadeh M, Tong K, Fonarow G C; Better Effectiveness After Transition-Heart Failure (BEAT-HF) Research Group. Effectiveness of remote patient monitoring after discharge of hospitalised patients with heart failure: The Better Effectiveness After Transition-Heart Failure (BEAT-HF) randomized clinical trial. JAMA Intern Med 2016; 176:310-318. [0632] O W, Orn S, Parissis J T, Ponikowski P; ESC Committee for Practice Guidelines. ESC guidelines for the diagnosis and treatment of acute and chronic heart failure 2012: The Task Force for the Diagnosis and Treatment of Acute and Chronic Heart Failure 2012 of the European Society of Cardiology. Developed in collaboration with the Heart Failure Association (HFA) of the ESC. Eur J Heart Fail. 2012; 14(8):803-69. Erratum in: Eur J Heart Fail. 2013; 15(3):361-2. e391-479. [0633] Pocock S J, Ariti C A, McMurray J J, Maggioni A, Køber L, Squire I B, Swedberg K, Dobson J, Poppe K K, Whalley G A, Doughty R N; Meta-Analysis Global Group in Chronic Heart Failure: Predicting survival in heart failure: a risk score based on 39 372 patients from 30 studies. Eur Heart J. 2013 May; 34(19):1404-13. [0634] Ponikowski P, Voors A A, Anker S D, Bueno H, Cleland J G, Coats A J, Falk V, Gonzalez-Juanatey J R, Harjola V P, Jankowska E A, Jessup M, Linde C, Nihoyannopoulos P, Parissis J T, Pieske B, Riley J P, Rosano G M, Ruilope L M, Ruschitzka F, Rutten F H, van der Meer P. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: The Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC). Developed with the special contribution of the Heart Failure Association (HFA) of the ESC. Eur J Heart Fail 2016; 18:891-975. [0635] Ponikowski P, Voors A A, Anker S D, Bueno H, Cleland J G, Coats A J, Falk V, Gonzalez-Juanatey J R, Harjola V P, Jankowska E A, Jessup M, Linde C, Nihoyannopoulos P, Parissis J T, Pieske B, Riley J P, Rosano G M, Ruilope L M, Ruschitzka F, Rutten F H, van der Meer P. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: The Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC). Developed with the special contribution of the Heart Failure Association (HFA) of the ESC. Eur J Heart Fail 2016; 37(27):2129-2200. [0636] Pousset F, Masson F, Chavirovskaia O, et al. (2000) Plasma adrenomedullin, a new independent predictor of prognosis in patients with chronic heart failure. Eur Heart J; 21:1009-14. [0637] Spethmann S., Prescher S., Dreger H., Nettlau H., Baumann G., Knebel F., Koehler F. Electrocardiographic monitoring during marathon running: a proof of feasibility for a new telemedical approach. Eur J Prev Cardiol. 2014 November; 21 (2 Suppl):32-7. doi: 10.1177/2047487314553736. [0638] Sudoh, T., Kangawa K., Minamino N. and Matsuo H. (1988) A new natriuretic peptide in porcine brain, Nature; 332: 78-81. [0639] R Core Team (2018). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. [0640] T. Sudoh BBRC 1989; 159: 1427-34. [0641] Terry M. Therneau, Patricia M. Grambsch (2000). Modeling Survival Data: Extending the Cox Model. Springer, New York. ISBN 0-387-98784-3 [0642] Tzikas S, Keller T, Ojeda F M, et al. (2013) M R-proANP and M R-proADM for risk stratification of patients with acute chest pain. Heart; 99:388-95. [0643] van Riet E E, Hoes A W, Wagenaar K P, Limburg A, Landman M A, Rutten F H. Epidemiology of heart failure: the prevalence of heart failure and ventricular dysfunction in older adults over time. A systematic review. Eur J Heart Fail 2016; 18: 242-52. [0644] Van Spall H G C, Rahman T, Mytton O, Ramasundarahettige C, Ibrahim Q, Kabali C, Coppens M, Brian Haynes R, Connolly S. Comparative effectiveness of transitional care services in patients discharged from the hospital with heart failure: a systematic review and network meta-analysis. Eur J Heart Fail 2017; 19:1427-1443. [0645] Vuolteenaho O., Ala-Kopsala M, Ruskoaho H. (2005) BNP as a biomarker in heart disease. Adv Clin Chem; 40:1-36. [0646] Wild P S, Schnabel R B, Lubos E, et al. (2011) Midregional proadrenomedullin for prediction of cardiovascular events in coronary artery disease: results from the AtheroGene study. Clin Chem; 58:226-36. [0647] Xiang R, Li L, Liu S “Meta-analysis and meta-regression of telehealth programmes for patients with chronic heart failure” J Telemedicine and Telecare 2013; vol. 19, 5 249-259. [0648] Yancy C W, Jessup M, Bozkurt B, Butler J, Casey D E Jr, Drazner M H, Fonarow G C, Geraci S A, Horwich T, Januzzi J L, Johnson M R, Kasper E K, Levy W C, Masoudi F A, McBride P E, McMurray J J V, Mitchell J E, Peterson P N, Riegel B, Sam F, Stevenson L W, Tang W H W, Tsai E J, Wilkoff B L. 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013; 62:e147-239.