The prevalence of implantable cardioverter-defibrillator (ICD) and heart failure patients eligible for remote monitoring in Germany

Abstract

Background

In Germany, contemporary data on the prevalence of patients with an implantable cardioverter-defibrillator (ICD) is lacking. Recently, ICD patients with heart failure (HF) fulfilling pre-defined criteria by the G-BA (Federal Joined Committee) are eligible for remote monitoring (RM) reimbursement. This investigation aims to evaluate the prevalence of HF patients with an ICD meeting these criteria.

Methods

Annual national quality assurance data from all German hospitals on newly implanted ICDs, New York Heart Association (NYHA) class and left ventricular ejection fraction (LVEF) between 2010 and 2021 were obtained to build a prevalence model. The number of ICD patients eligible for RM was calculated by applying the main G-BA inclusion criteria.

Results

The ICD prevalence increased continuously from 2010 to 2017 (202.637 patients in 2017) and decreased with a lower rate until 2022. The model calculated an ICD prevalence of 190.698 patients in 2022 of which an estimated 120.941 ICD patients with HF were eligible for RM. This reflects approximately 63% of the actual total estimated ICD patient population in Germany.

Conclusions

The model identified a large patient population currently eligible for RM. To our knowledge, this is the first study providing information on the size of this ICD patient population with HF in Germany. With only a fraction of eligible patients currently receiving RM, these findings may facilitate future planning, resource calculations and scale-up of RM. The building of a specific infrastructure focussing on efficient use of resources reflects a mandatory prerequisite for successful RM implementation.

Background

Heart failure (HF) represents a common disease with more than 500.000 newly diagnosed patients in Germany in the year 2021 [1]. Due to the high morbidity and mortality, HF accounts for a relevant burden of the healthcare system. In 2020, HF was the most common primary diagnosis in patients being treated during a hospital stay [2]. The remote monitoring (RM) of HF patients reflects an innovative technique which was accepted as a new method in 2020 by the Gemeinsamer Bundesausschuss (G-BA) in the treatment of patients in the outpatient sector [3].

The most important building block of RM for HF is the close cooperation between the primary treating physician (PTP) and the telemonitoring centre (TMC). Roles and responsibilities as well as the requirements for RM are described in detail in the G-BA decision document [3]. Briefly, the role of the PTP consists of providing guideline-based HF care and RM guided therapy. TMCs on the other hand are responsible for the daily monitoring, which includes daily data transmission and capture, analysis, the interpretation, and filtering of incoming alarms as well as notifying the PTP in case of actionable alarms. Additionally, a TMC is required to provide documentation for quality assurance purposes and must have approval from the Kassenärztliche Vereinigung (KV).

Evidence from randomised controlled trials (RCTs) showed that multiparameter monitoring of HF patients with implantable cardioverter defibrillators (ICD) or with cardiac resynchronisation therapy systems (CRT-D) can lead to a significant reduction in mortality and cardiovascular morbidity [4, 5]. A recent meta-analysis by Zito et al. (2023) which included 4.869 patients confirmed the clinical benefit of multiparameter guided implant-based RM compared to standard clinical management [6].

Standard reimbursement for RM of HF patients with statutory health insurance based on the Einheitlicher Bewertungsmaßstab (German Uniform Evaluation Standard, EBM) was introduced on January 1st, 2022 [7]. Additionally, in January 2022, reimbursement for privately insured patients was introduced [8]. According to the decision of the G-BA, RM for HF can be used for patients either with implanted cardiac devices or in combination with non-invasive devices. Invasive RM using implanted cardiac devices consist of implantable cardioverter defibrillators (ICD) including cardiac resynchronisation therapy defibrillators (CRT-D) as well as CRT pacemakers (CRT-P). Data on the number of ICD patients (prevalence) and the subgroup of ICD patients eligible for RM for HF is of major importance to implement this new method in the daily clinical routine to ensure guideline-directed medical therapy. While data on the annual number of patients receiving a newly implanted ICD are readily available [9–11], information on the total number of patients living with an ICD in Germany is still lacking. The aim of the current investigation was to estimate the prevalence of ICD carriers in Germany and the proportion of these patients meeting the G-BA criteria for RM for HF.

Methods

Data sources

Accredited hospitals are obliged to provide structured quality records including the annual number of ICD and CRT-D new implantation procedures, the combined number of device changes, number of system changes and device explantations. Published reports based on annual national quality assurance data [9–11] on newly implanted cardiac implantable electronic devices (CIEDs), New York Heart Association (NYHA) functional class and left ventricular ejection fraction (LVEF) covering the years 2010–2021 were used to construct an Excel-based model (Microsoft Corporation, Microsoft Excel Version 2023). Data on device explantations, device changes or system changes was not used in the model. The quality assurance data used refers to implantations carried out in the inpatient setting. It can be assumed that the number of new ICD and CRT-D implantations in the outpatient sector is very low due to a lack of regular reimbursement of these procedures during the observation period [11]. Data analysis was restricted to single-chamber and dual-chamber ICDs and CRT-Ds. Subcutaneous ICD (S-ICDs) were excluded because, in comparison with conventional ICD systems, patients are younger and differ in the underlying cardiac disease [12, 13]. Additionally, there were no S-ICD implantations reported for years 2010–2014.

Data on the size of the ICD patient population implanted before the model start year was derived from a prevalence model developed by Smala et al. (2011) which estimated the CIED prevalence in Germany and the United Kingdom between 2005 and 2015 [14] and all-cause mortality rates for ICD implanted patients were calculated based on published literature [15].

Model overview

The prevalence model comprises a time horizon from 2010 to 2022. To account for the patient population who received an ICD before the model start (2010), data from a previously published model by Smala et al. (2011) [14] was integrated to estimate the number of ICD patients alive in 2010 (personal communication). This model estimated a total ICD population of 114.133 in 2010, hereafter referred to as the “start cohort”.

Due to a 2-year delay in published implantation data, ICD implantation numbers for 2022 were projected using the average percent change in ICD implantations over the period 2015–2021 (Table 1). Because the implantation numbers showed a declining trend after 2015, the period 2015–2021 was selected to ensure this trend was accurately reflected in the model.

Table 1.

Model inputs and sources

Model Inputs	Base Case Value	Source
Annual all-cause mortality rate for ICD patients	11,15%	a Based on Köbe et al. [15]
Size of the start cohort	114.133	Smala et al. [14] personal communication
Percentage of ICD patients with NYHA class II/III	84,56%	b Based on data for years 2014–2021 (range 83,39% − 86,13%) [9–11]
Percentage of patients with LVEF < 35%	75%	Based on the 75th percentile reported by IQTIG over 2016–2021 [9,10]
Average % change in ICD implantation numbers (used to project implants for the year 2022)	-7,12%	c Based on IQTIG data for years 2021–2015 [9,10]

^a Calculated based on 3.100 patients in the ICD group and 331 deaths, and 1.284 patients in the CRT-D group and 158 deaths. The total number of deaths/total number of patients results in an overall 1-year mortality rate of 11,15% for the combined ICD and CRT-D groups. ^b Proportion of patients with NYHA II/III and ICD/CRT-D devices. Calculated by summing up % of patients with NYHA class II and III for every year and taking the average over 2021–2014. ^c Calculated by summing up the yearly % change and dividing by the number of years in the period 2021–2015

ICD prevalence

ICD prevalence was defined as the cumulative number of patients alive in a certain year. For each implantation year, starting in 2010, the number of patients alive in the following years over the model time horizon (2010–2022) was calculated by applying a constant year-on-year all-cause mortality rate, based on a large device registry study which reported 1-year mortality rates in newly implanted ICD patients in Germany [15]. The same year-on-year all-cause mortality rate was applied to the start cohort. Using this method, a life table was constructed, describing the number of patients alive for each ICD implantation year over the model time horizon.

For each year, the cumulative number of ICD patients (prevalence) was then calculated by taking the patients in the start cohort still alive in that year, adding patients still alive which were newly implanted with an ICD in all preceding years and adding patients newly implanted with an ICD in the current year (e.g. total cumulative number of ICD patients in 2012 = patients in the start cohort and alive in 2012 + patients implanted in 2010 and 2011 and still alive in 2012 + patients receiving an ICD implant in 2012).

ICD patients eligible for remote monitoring for heart failure

To estimate the ICD patient population eligible for RM for HF, the two main disease-related G-BA inclusion criteria (NYHA class II/III and LVEF < 40%) were applied sequentially:

1. the number of ICD patients with HF (NYHA class II/III) was calculated by multiplying the cumulative total number of ICD patients in each year with the average % of ICD patients with NYHA class II/III.

2. to approximate the G-BA LVEF < 40% criterion, the number of ICD patients with NYHA class II/III was then multiplied with 0,75. This factor was chosen as a proxy because quality assurance statistics [9, 10] for % LVEF in ICD implanted patients has consistently shown that 75% of all LVEF values fell below 35% (75th percentile), which can serve as a proxy for the percentage of patients with LVEF < 35%.

All model inputs, calculations and input sources are summarised in Table 1.

Sensitivity analysis

To test the robustness of the model and estimate the impact of model parameter uncertainty on the model outcome (prevalence of ICD patients eligible for RM for HF), a one-way sensitivity analysis, varying inputs by ± 10% was conducted.

Results

The aggregated quality assurance report data [9–11] on new ICD implantations (ICD incidence) over 2010–2021, which formed the basis of the current prevalence model, showed that 25.582 ICDs were implanted in 2010. Implantation numbers peaked in 2015 (29.327 implants) and declined thereafter to 18.780 ICDs in 2021. The current model projected a total of 17.443 implantations for 2022.

Prevalence

Over the years 2010–2022, the ICD patient prevalence estimates gradually increased from 139.715 patients in 2010 to a peak prevalence of 202.637 patients in 2017 after which it declined (Fig. 1 grey area, Table 2).

Fig. 1.

Prevalence estimates for patients implanted with an ICD (grey area) and ICD patients eligible for RM for HF (blue area) over the model time horizon (2010–2022). For illustration purposes, the non-continuous annual prevalence estimates are represented as a smooth continuous curve instead of a bar chart

Table 2.

Model prevalence estimates of patients with an ICD and prevalence of ICD patients eligible for RM for HF according to main the G-BA criteria (2010–2022)

	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	2021	2022
ICD Patients	139.715	152.589	165.149	176.193	185.759	194.374	200.472	202.637	202.220	200.712	198.331	194.997	190.698
ICD patients eligible for RM	88.607	96.772	104.738	111.742	117.809	123.272	127.139	128.512	128.248	127.292	125.782	123.667	120.941

For 2022, it was estimated that a total of 190.698 patients would be living with an ICD. Of those, the model estimated that approximately 63% (120.941 patients) would have NYHA class II/III and a LVEF under 35%, thus making them eligible for RM for HF according to the main G-BA criteria (Fig. 1 blue area, Table 2).

One-way sensitivity analysis

The impact of varying single model inputs by ± 10% on the model outcome (number of ICD patients eligible for RM for HF) is displayed in the tornado graph (Fig. 2). The sensitivity analysis showed that variations in the % of ICD patients with NYHA class II/III and the % of patients with a LVEF < 35% had the largest impact on the model outcome, with increasing inputs leading to an increase in outcome values. This was followed by the annual all-cause mortality rate, which showed a reduction in the model outcome at higher mortality rates. The size of the start cohort, which ranked 4th in terms of impact (< 2% change from base case), showed that a larger start cohort led to an increase in the model outcome. The model outcome was least sensitive to variations in the average % change in ICD implantation numbers (< 0.1% change from base case, data not shown).

Fig. 2.

The graph displays the one-way sensitivity analysis, representing the impact on the prevalence of ICD patients eligible for RM for HF when varying single model inputs by ± 10%. The vertical line represents base-case analysis (120.941 patients) and the horizontal axis represents the variation in the model outcome. The horizontal boxes represent the range of model outcomes, varying the inputs by ± 10%

Discussion

Following the recent reimbursement decisions, RM for HF has become available for a large group of patients with implanted cardiac devices in Germany. However, the exact size of the patient population is unknown. To fill this information gap, inform RM planning and implementation, the current study aimed to estimate the prevalence of ICD patients in Germany as well as the proportion of these patients eligible for reimbursement according to the G-BA criteria. We utilised ICD implantation data together with all-cause mortality rates to calculate the ICD prevalence over 2010–2022 and calculated the number of patients eligible for RM by applying the main G-BA criteria.

ICD implantations declined from 25.582 in 2010 to 18.780 in 2021, which can be partly explained by the advent of new improved medical therapies (e.g. Sodium–glucose cotransporter 2, SGLT-2 inhibitors) for the treatment of HF. In a previous study, we evaluated the impact of the COVID-19 pandemic on the number of newly implanted ICDs in Germany [16], showing that ICD implantations initially decreased during the early pandemic period (maximal reduction of 23,2% in April 2020 compared to April 2019) and then recovered to levels seen during the pre-pandemic period.

The ICD prevalence declined slowly after a peak in 2017. This can be explained by the fact that more ICD patients left the patient pool due to mortality than are replenished by new implantations post 2017.

ICD patient population eligible for remote monitoring for heart failure

The current model identified a large ICD population, encompassing 190.698 patients in Germany, of which two third (120.941 patients) could benefit from RM for HF according to criteria set by the G-BA. To the best of our knowledge, this is the first study to estimate contemporary ICD prevalence and the prevalence of ICD patients eligible for RM for HF in Germany, thus filling an important information gap. Only one study estimated the prevalence of CIED patients (including ICDs) in Germany [14], however, the model by Smala and colleagues did not provide estimations beyond 2015.

In 2022, 120.941 patients could have benefitted from RM. We believe this number to be conservative because other CIED types were not included in the analysis (S-ICD and CRT-P). Moreover, the ICD population eligible for RM was calculated by applying the proportion of patients with LEVF < 35% while the G-BA criteria is slightly higher (LEVF < 40%). This may have resulted in an underestimation of the actual patient population size.

Planning and infrastructure

With regards to planning aspects, the estimated size of the potential patient population helps health insurance providers, service providers, TMCs and PTPs to allocate resources and plan ahead.

Despite the well-defined infrastructure requirements specified by the G-BA, TMC accreditation by the responsible KV, close cooperation between TMC and PTP, daily transmission of device data and standard operation procedures (SOPs) and reaction times, the infrastructure necessary for scaling up RM in this patient population is not yet in place. Recommendations for making RM systems more efficient are mentioned in the 2023 Expert Consensus Statement on Practical Management of the Remote Device Clinic position paper [17] and include, but are not limited to, artificial intelligence (AI) tools, alert-based management, alert programming and continuous data transmission.

Regional differences in Germany

Remote monitoring methods would likely be of benefit especially in areas that are characterised by a low number of healthcare providers per population and with greater distances between patient and healthcare providers. But, in contrast, there are examples where TMCs are concentrated in large metropolitan areas (e.g. Hamburg, München) that already have a high density of healthcare providers. At the moment, there is no comprehensive information on the number of TMCs and their geographic distribution. Since 2024, there is a mandatory reporting requirement for TMCs to provide data on their RM for HF activities [18, 19]. Accordingly, the Kassenärztliche Bundesvereinigung (KBV) will be able to aggregate data on a national level, which may include the analysis of the regional distribution of TMCs. However, it is not yet clear if and when these data will become available to the general public.

Strength and limitations

The current model is based on 12 years of ICD implantation data derived from quality assurance reports, covering virtually all ICD implantations performed in the inpatient setting across Germany. In addition, the proportion of patients with NYHA class II/III and the proportion of patients with LVEF < 35% that served as model inputs were derived from the same quality assurance reports, showing a stable trend with little or no year-on-year variation (see Table 1).

As the source for the one-year all-cause mortality after ICD or CRT-D implantation, we used the analysis of the German DEVICE registry [15]. DEVICE is a real-world registry which enrolled patients with high comorbidity rates (e.g. patients with chronic kidney disease) and therefore reported higher one-year all-cause mortality rates compared to RCTs, which often exclude these patients. The higher mortality rate used in our model would reflect the real-world setting of patients implanted with ICDs in Germany. However, since the DEVICE registry study is based on data collected from 2008 to 2015, mortality rates might have been lower in recent years (e.g. 2016–2022) due to new medical therapies for HF. Also, the proportion of patients with LVEF < 35% was slightly higher in the DEVICE registry study population (approximately 78%, based on 2.195 ICD patients with LVEF ≤ 35% and 1.213 CRT-D patients with LVEF ≤ 35%), compared to the 75% used in the current study population. This may also contribute to the overestimation of annual mortality in the current model.

Our model applied a constant annual mortality rate, however, a small proportion of implanted ICD patients would have received RM for HF according to the approaches similar to the G-BA requirements. A potential reduction in the all-cause mortality rate due to RM is not accounted for in our model. The number of ICD patients eligible for RM was calculated by sequentially multiplying the ICD population with the % of patients with NYHA II/III and the % of patients with LVEF < 35%. This introduces a potential source of model inaccuracy by simply using the product of both rates. For model simplicity and due to the lack of patient level data in our data source, we assumed that NYHA and LVEF were completely independent. However, it is likely that there is a correlation between NYHA and LVEF. Our estimated proportion fulfilling both conditions (63%) lies within the range of theoretically possible extremes: 75% and 59,6% (maximum and smallest possible overlap of patient groups with both characteristics). Given these extremes, our calculated proportion (63%) reflects a conservative estimate.

Due to the lack of ICD prevalence data, it was not possible to validate our model against published figures. However, with the expected publication of mandatory quality assurance data captured by the TMCs, validation of the estimated number of eligible ICD patients could be performed in the future.

Model robustness was evaluated by a one-way sensitivity analysis (Fig. 2). Varying the proportion of NYHA II/III and proportion of patients with LVEF < 35% had the largest impact on the model outcome. However, annual quality assurance report data showed that these documented parameters varied much less compared to the ± 10% variation used in the sensitivity analysis. The proportion of NYHA II/III patients varied less than 2% over 2014–2021 (mean 84,56%, range 83,39% – 86,13%) and the 75th percentile (proportion of patients with a LVEF < 35%) remained constant during 2016–2021. Given this low variation, it is therefore reasonable to assume that the changes in the model outcome are less pronounced compared to the sensitivity analysis results displayed in Fig. 2.

Conclusions

The present study demonstrates that a large ICD patient population exists who would definitively benefit from RM for HF. Knowing the size of the patient pool facilitates the definition and setting of specific care targets and goals.

From 2024 onwards, there is a mandatory reporting requirement of patient numbers with RM for HF (statutory health insured patients only) that are managed by the TMCs [18, 19]. With this information, the gap between the patient population eligible for RM for HF and the number of patients actually receiving RM can be quantified.

Remote device interrogation (RDI) is another less complex type of RM of ICDs (independent of clinical indication) which means that in principle most ICD patients could receive and benefit from one type of remote monitoring (RDI or RM for HF). The specific ICD reimbursement for RDI was introduced in the current reimbursement structure in Germany in 2017. Five years later (2021), and despite the COVID-19 pandemic, only 12% of ICD follow-up services were performed as RDI, i.e., a marginal increase from 5% in 2017 [20]. This is a relatively slow adoption compared, for example, to Austria. Indeed, data from the Austrian Pacemaker, ICD and loop recorder registry covering over 75% of all ICD implantations showed that 55% of the patients with follow-up documentation received their follow up via RDI in 2021 [21].

The formulation of future goals reflects a logical consequence of the results being reported in this study. Moreover, potential barrier hampering their fast adoption should be identified and addressed. For example, the administrative procedures and processes should be simplified to enable rapid implementation.

Abbreviations

CIEDs

Cardiac implantable electronic devices

CRT-D

Cardiac resynchronisation therapy defibrillator

CRT-P

CRT pacemaker

EBM

Einheitlicher Bewertungsmaßstab (German Uniform Evaluation Standard)

G-BA

Gemeinsamer Bundesausschuss (Federal Joined Committee)

HF

Heart failure

ICD

Implantable cardioverter-defibrillator

KBV

Kassenärztliche Bundesvereinigung

KV

Kassenärztliche Vereinigung

LVEF

Left ventricular ejection fraction

NYHA

New York Heart Association

RCT

Randomised controlled trial

RDI

Remote device interrogation

RM

Remote monitoring

SGLT-2

Sodium–glucose cotransporter 2

S-ICD

Subcutaneous ICD

SOPs

Standard operation procedures

TMC

Telemonitoring centre

Author contributions

Idea and concept: JOS, TH Methodology: TH, BG Calculation and data processing: TH, BG Writing and reviewing the draft: JOS, TH, BG.

Funding

Not applicable.

Data availability

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Declarations

Ethics approval

Not applicable. The research consists of an analysis of annually reported quality assurance data which are publicly available.

Consent for publication

b) Not applicable.

Competing interests

Boye Gricar and Tino Hauser are employees of Biotronik, Jörg Schwab received consulting grants from Biotronik.