This study demonstrated that AMI patients with elevated SI and mSI values at discharge experience significantly higher long-term mortality in Kaplan–Meier analysis. The survival curves indicate that the greatest difference in mortality was in the first year after hospital discharge, but the curves remained separated for the entire 5-year follow-up time. The differences in mortality were smaller in NSTEMI cases compared to STEMI cases. For the latter, the association between SI and long-term mortality remained significant even after adjustment for major confounders, which was not the case for NSTEMI cases.

Most previous studies have found that admission SI and mSI are strong predictors for early outcomes and are useful in emergency triage situations after AMI [7]. Especially for STEMI cases, our analysis extends this knowledge to the point of discharge, which addresses a significant gap in the present scientific literature. Previous research focused heavily on admission values, leaving the discharge situation largely unexplored, not only in AMI but also across multiple cardiovascular pathologies. However, the observed results partially align with previous findings by Alapati et al., who found that higher discharge heart rates were independently associated with higher 3-year mortality after AMI [10].

Comparisons between high and low SI groups revealed notable differences in key characteristics, such as age, comorbidities, and discharge medication, which may partially explain the observed differences in long-term mortality in Kaplan–Meier analysis. For example, patients in the high SI groups were significantly less likely to receive statins at discharge—a difference not directly attributable to blood pressure or risk of hypotension—suggesting that there are important underlying group differences that may have influenced the observed outcomes.

Apart from that, several physiological mechanisms may underlie the results seen in this study. The indices measured at discharge likely reflect a combination of post-intervention cardiovascular stress, medication effects, volume status, and subclinical dysfunctions that persisted throughout hospitalization [11, 12]. Clinical evidence indicates that AMI often causes chronically increased sympathetic and reduced parasympathetic activity, which can result in persistently elevated heart rates and the occurrence of arrhythmias and progressive heart failure [13,14,15]. Other studies have shown that persistent tachycardia, a driving factor for increased SI and mSI, is associated with pathological cardiac remodeling and an increased myocardial oxygen demand [16, 17]. Therefore, higher SI and mSI values during hospitalization and especially at discharge may reflect ongoing cardiovascular stress or dysfunction, which could be associated with increased mortality. Importantly, these considerations remain hypothetical, given that our observational analysis did not include measurements of biomarkers, autonomic nervous system activity, or myocardial remodeling.

The present analysis showed that SI and mSI were more predictive in STEMI patients than in NSTEMI patients. In a previous analysis based on data from the Augsburg Myocardial Infarction Registry, we reported that SI and mSI, measured at hospital admission, were significantly associated with long-term mortality in multivariable Cox regression models with a slightly better predictive value in NSTEMI cases compared to STEMI cases, which is the other way round compared to what we found in the present analysis [6]. The current findings might be explained through greater infarction severity and transmural ischemia in STEMI patients [18]. Contrarily, NSTEMI patients present a highly heterogenous group with higher comorbidity burdens and more complex cardiac profiles [19]. In NSTEMI cases, admission SI and mSI seem to be much more predictive compared to the corresponding discharge indices. For STEMI cases, discharge indices might provide (additional) predictive information regarding long-term mortality.

As previously mentioned, SI and mSI are resource-efficient tools, as they are derived from routine vital parameters collected during standard clinical monitoring. Especially in STEMI cases, these indices might be valuable measures for long-term mortality, specifically in the first year after the event (the Kaplan–Meier curves suggest the greatest effect in the early phase of the 5-year follow-up period). Integration of these indicators into discharge check lists could potentially support risk stratification and help to determine post-discharge care intensity and follow-up patterns in cooperation with primary care providers. Automated calculation of the indices within electronic health records systems would efficiently position them as additional information for assessing discharge readiness, complementing rather than replacing existing tools. Although scores like GRACE and TIMI remain more relevant in the present context of prognostic risk stratification, the emergence of artificial intelligence in medical contexts including AI-based early warning systems [20] opens new opportunities for incorporating additional indices. Including SI and mSI at discharge alongside other known predictors could enhance clinical decision-making without adding diagnostic or administrative complexity. The indicated implications suggest that SI and mSI could be implemented across different healthcare infrastructures. In resource-limited settings, such as small hospitals, even manual calculation from routinely obtained clinical parameters is an option. Conversely, hospitals with advanced IT infrastructures can implement automated calculations into decision-support systems.

Further research is necessary to externally validate the present findings. Investigating the dynamics of SI and mSI throughout the clinical stay as well as combining these indices with laboratory markers and ECG-derived features could provide additional insights into their predictive value for long-term mortality. Since SI and mSI measured at hospital admission as well as hospital discharge are associated with mortality after AMI and especially after STEMI, the analysis of whether the predictive strength of both indices could be transferred to the rehabilitation process and throughout later points of aftercare would be of interest. Additionally, experimental or imaging studies could help bridge the gap between the observed indices and underlying pathophysiological processes.

The strengths of this analysis primarily result from qualities of the Augsburg Myocardial Infarction Registry, which features a real-world setting, a large cohort, and long follow-up periods with consecutive and standardized data collection protocols. The population-based approach of the registry minimizes selection bias, ensuring a representative sample which carefully suggests transferability to other healthcare systems and aftercare structures. Furthermore, the extensive information collected for each patient enabled the adjustment for relevant confounders.

However, several limitations must be considered. First, the outcome in the present analysis was all-cause mortality, without differentiation between cardiac and non-cardiac causes. Second, changes in medication and therapeutic interventions during the follow-up period, especially the introduction of new medication classes, were not considered. Third, despite adjustment in the Cox regression models, residual and unmeasured confounding cannot be ruled out, particularly regarding factors such as ethnicity, socioeconomic status, or post-discharge medical care structures. In particular, there was no information on medication adherence, participation in rehabilitation programs, or lifestyle modification after discharge, all of which could have affected the relationship between the discharge SI/mSI and long-term mortality. Also, the models were not adjusted for measures of heart failure (e.g., left-ventricular ejection fraction), which may be relevant confounders. Moreover, no information on (high-sensitivity) cardiac troponin was available for this analysis. Finally, although all data were collected by medically trained staff, measurement errors or variations in the timing of measurements in relation to discharge could have influenced the accuracy and comparability of SI and mSI.