A total of 2,461 patients were included in our study, of whom 1,402 had ischemic stroke. The mean age was 70.55 ± 14.59 years, with 1,221 (49.61%) being female. The average SHR was 1.06 ± 0.33. The length of hospital stay was 6.99 days (IQR: 3.92–13.71), whereas the duration in the ICU was 3.09 days (IQR: 1.74–6.17). Patients were categorized into four quartiles based on their SHR levels: Q1 (0.227–0.85), Q2 (0.85–0.994), Q3 (0.994–1.197), and Q4 (1.197–29.28). As illustrated in Table 1, when compared to the Q1 group, patients in the Q4 group exhibited faster heart rates and respiratory rates but lower systolic, diastolic, and mean blood pressures; however, their SpO2 levels were higher. Additionally, both the hospital length of stay and ICU length of stay were notably prolonged for the Q4 group compared to the Q1 group (all P < 0.001). Notably, the mortality risk was markedly increased in the Q4 group compared to other groups: in-hospital mortality rates were 7.31%, 9.62%, 13.53%, and 27.33% respectively across the quartiles (P < 0.001); similarly, in-ICU mortality rates were 3.73%, 6.36%, 8.53%, and 18.00% respectively (P < 0.001). The baseline characteristics of patients with ischemic stroke and those with non-ischemic stroke are detailed in Supplementary Tables 2 and 3, respectively.

The baseline characteristics of patients in the survival and mortality groups during hospitalization are summarized in Table 2. Patients in the mortality group were generally older, though there was no significant difference in gender distribution. Of particular note, the SHR level was markedly higher in the mortality group (1.25 ± 0.42) than in the survival group (1.02 ± 0.30), with a statistically significant difference (P < 0.001).

We utilized Kaplan-Meier survival analysis curves to examine the incidence of in-hospital and ICU mortality across the four SHR quartile groups (Fig. 2), revealing that patients in the Q4 group had a significantly higher risk of both in-hospital and ICU mortality (log-rank P < 0.001). To further investigate the relationship between SHR and mortality, we conducted Cox proportional hazards regression models (Table 3). The results demonstrated that each standard deviation (SD) increase in SHR was associated with a 45% higher risk of in-hospital mortality in the unadjusted model (HR 1.45, 95% CI 1.34–1.57, P < 0.001), which attenuated to 34% in the partially adjusted model (HR 1.34, 95% CI 1.23–1.45, P < 0.001) and 35% in the fully adjusted model (HR 1.35, 95% CI 1.23–1.48, P < 0.001). When SHR was categorized as an ordinal variable, patients in the Q4 group showed a 178% higher risk of mortality compared to the Q1 group in the unadjusted model (HR 2.78, 95% CI 2.00-3.87, P < 0.001), which reduced to 137% in the partially adjusted model (HR 2.37, 95% CI 1.70–3.32, P < 0.001) and 152% in the fully adjusted model (HR 2.52, 95% CI 1.75–3.63, P < 0.001), with a significant trend towards increasing risk as SHR increased (P for trend < 0.001).

Kaplan-Meier survival analysis curves for all-cause mortality. A In-hospital mortality; B ICU mortality

Similar trends were observed for ICU mortality: each SD increase in SHR corresponded to HR of 1.43 (95% CI 1.30–1.58, P < 0.001) in the unadjusted model, 1.33 (95% CI 1.19–1.47, P < 0.001) in the partially adjusted model, and 1.37 (95% CI 1.21–1.54, P < 0.001) in the fully adjusted model. Patients in the Q4 group exhibited significantly higher ICU mortality risks compared to the Q1 group, with HR of 2.66 (95% CI 1.69–4.18) in the unadjusted model, 2.21 (95% CI 1.40–3.49, P < 0.001) in the partially adjusted model, and 2.65 (95% CI 1.57–4.47, P < 0.001) in the fully adjusted model, all showing a significant upward trend (P for trend < 0.001).

Additionally, we evaluated the potential non-linear relationship between SHR and mortality using RCS model. The results (Fig. 3) showed that in the fully adjusted model, the P values for non-linear relationships between SHR and both in-hospital and ICU mortality were not statistically significant (non-linear P > 0.05). However, threshold effects were observed (Supplementary Table 4): for in-hospital mortality, the SHR threshold point was identified at 0.77, beyond which each unit increase in SHR was associated with a significant rise in mortality risk (HR 2.41, 95% CI 1.77–3.30, P < 0.001). Similarly, for ICU mortality, the SHR threshold point was 0.79; when SHR exceeded this value, there was also a significant increase in mortality risk (HR 1.96, 95% CI 1.29–2.98, P = 0.002).

RCS analysis of SHR with all-cause mortality. A In-hospital mortality; B ICU mortality

The association between SHR and mortality remained significant in both ischemic and non-ischemic stroke patients (Supplementary Tables 5 and 6). This relationship was consistent across various subgroups, including different age groups and patients with hypertension, atrial fibrillation, renal disease, or chronic lung disease, with no significant interactions observed (Fig. 4). Interestingly, SHR demonstrated a more pronounced predictive value for in-hospital mortality in female patients compared to male patients (female HR 1.58, 95% CI 1.39–1.79 vs. male HR 1.13, 95% CI 0.98–1.30, p for interaction = 0.001). Notably, in patients without diabetes, SHR exhibited a more prominent predictive value: for in-hospital mortality (non-diabetic HR 1.52, 95% CI 1.35–1.70 vs. diabetic HR 1.01, 95% CI 0.86–1.18, P for interaction = 0.003) and for ICU mortality (non-diabetic HR 1.64, 95% CI 1.40–1.91 vs. diabetic HR 0.92, 95% CI 0.74–1.13, P for interaction = 0.001).

Subgroup forest plot for all-cause mortality. A In-hospital mortality; B ICU mortality

We evaluated the impact of adding SHR to existing scoring models on the prediction of in-hospital mortality by calculating the AUC. The results (Fig. 5) demonstrated that incorporating SHR consistently improved the predictive ability for in-hospital mortality across all examined scores: APSIII (from 0.707(0.678, 0.736) to 0.748(0.722, 0.775)), SAPSII (from 0.746(0.721, 0.772) to 0.780(0.757, 0.803)), OASIS (from 0.730(0.703, 0.758) to 0.766(0.740, 0.791)), and SOFA (from 0.678(0.647, 0.708) to 0.728(0.700, 0.756)). Specifically, as presented in Table 4, for the APSIII score, the C-index increased from 0.647 (95% CI 0.614, 0.680) to 0.695 (95% CI 0.665, 0.725) with the addition of SHR, accompanied by an IDI of 1.40 (95% CI 0.40, 3.00), and an NRI of 10.90 (95% CI 2.30, 16.90). For the SAPSII score, the C-index rose from 0.706 (95% CI 0.678, 0.734) to 0.738 (95% CI 0.712, 0.764), with an IDI of 1.20 (95% CI 0.20, 2.40), and an NRI of 5.90 (95% CI -2.20, 11.80). For the OASIS score, the C-index improved from 0.692 (95% CI 0.663, 0.722) to 0.729 (95% CI 0.702, 0.757), with an IDI of 1.70 (95% CI 0.50, 3.30), and an NRI of 7.70 (95% CI 0.90, 14.30). Lastly, for the SOFA score, the C-index enhanced from 0.625 (95% CI 0.591, 0.659) to 0.686 (95% CI 0.656, 0.716), with an IDI of 1.80 (95% CI 0.60, 3.60), and an NRI of 8.00 (95% CI 1.60, 15.50).

ROC curve analysis of the incremental effect of SHR on in-hospital all-cause mortality. A APSIII score + SHR; B SAPSII score + SHR; C OASIS score + SHR; DSOFA score + SHR