Our study shows that the percentage of patients with NICMP without an indication for ICD therapy according to the new Dutch guideline was indeed high: excluding the high-risk patient group, 74 patients out of 99 (75%) would not have had an ICD implanted, mainly because of a CRTP indication (57 out of 74, 77%). This new guideline therefore does indeed reduce significantly the number of ICD indications in this specific patient group. Nevertheless, and importantly, appropriate ICD therapy (both ATP and shocks) had a comparable incidence of 16% in the group with an indication versus 12% in the group without. The estimated annual risk for appropriate ICD therapy for ventricular arrhythmias of 2.4% is comparable with the incidence mentioned in the literature of between 2 and 4% in ICD for primary prevention (ICMP and NICMP patients, respectively) [22,23,24].

One of the main parameters in the 2023 Dutch guideline for ICD indication is the presence of fibrosis (or LGE) on CMR. In our single-centre retrospective study, of the group with an ICD indication according to the 2023 Dutch guideline 16 patients did not undergo CMR, so were without a clear indication. Because of the exploratory, retrospective nature of the study these patients were not excluded from the analysis so as not to underestimate the event rate in an already limited sample size. Interestingly the 9 with a clear indication because of LGE on CMR did not have ICD therapy during follow-up, but of the 16 without CMR, 4 patients did. In the total population studied, CMR was performed in 59 out of the 99 patients. Appropriate ICD therapy was seen in patients without CMR, in patients with LGE on CMR and even in patients without LGE on CMR. Our data could not confirm that the absence of LGE completely ruled out the risk of ventricular arrhythmias. Interobserver variability in the interpretation of the presence of LGE on CMR has to be close to zero using the sole criteria of presence of absence of LGE for an ICD indication.

Overtreating patients implies unnecessary exposure to the risk of complications, with a published complication rate of 1 out of 13 patients after ICD implantation [25]. Our data confirm this complication rate with 9% complications in the total group, mainly related to lead problems. Inappropriate ICD therapy was comparable in both groups and was mainly driven by supraventricular arrhythmias.

Improvements in treatment over the past years have significantly impacted the lives of heart failure patients as regards both prognosis and quality of life, but we are still not able to properly select patients who are at low risk for SCD. Our study results question the fact that the new Dutch guideline could completely or sufficiently rule out the risk of life-threatening ventricular arrhythmias in patients with NICMP, since the event rate of appropriate ICD therapy was comparable for patients with. and without an ICD indication. A better understanding of the mechanism of ventricular arrhythmias is needed in this population. More data are needed to understand and redefine ICD indications in this specific patient group with NICMP. A multicentre study to gather more retrospective data to enlarge the sample size could add valuable insights into the impact of the new Dutch guideline. The next step could be a prospective, randomised study to show the benefit and capability of the new Dutch guideline to properly exclude patients from ICD implantation. With an estimated and published annual risk of around 4% of ventricular arrythmias in patients with an ICD for primary prevention, a comparison to an acceptably low annual SCD risk should be made [22,23,24]. There is no consensus about this accepted annual risk, but 1% has been discussed. In a non-inferiority design and with the established 4% versus 1% annual SCD risk rate, with 80% power and 5% type I error, 424 patients per group would be needed in such a study. Several ongoing or soon to be started studies on ICD therapy, SCD risk and NICMP, such as the BRITISH, the CONTEMP-ICD and Dutch-ICD study will most likely show the beneficial effect of ICD on survival and provide valuable insights into specific clinical risk factors to further redefine ICD indication in this difficult patient group.

A recently published position paper dealt with risk assessment and ICD indication in NICMP. A more individualised approach was recommended, weighing arrhythmic risk, competing risks and possible risk modifiers to assess eligibility. The same steps, including CMR, genetic testing and ‘competing’ risk scores, as proposed in the 2023 Dutch guideline, were used. ICD therapy was not completely ruled out in patients eligible for CRTP. However, the authors concluded that probably only a small proportion should be considered for CRTD (with a probably high SCD risk) and that in the group of patients with NICMP without an indication for CRTP, a subgroup at high risk should have an ICD implanted. We are obligated to refine the SCD risk and define the patient at truly low risk [26]. The position paper puts current knowledge into perspective and warrants further research once again.

The retrospective study design is a limitation and lacks sufficient statistical power. ICD therapy could have been underestimated, since missing data is a concern in retrospective analysis. Since treatment strategies and ICD programming have changed over time, this could influence outcome when applied correctly.