Fluid resuscitation is a cornerstone of sepsis management, yet determining the optimal fluid volume for individual patients, especially those with HF, presents a formidable challenge [25, 26]. Our results suggest that the traditional “one-size-fits-all” approach to fluid resuscitation in sepsis may not be appropriate for patients with ADHD. The high BNP levels and the prevalent use of diuretics observed in our study population are indicative of a more active HF process, which may respond differently to fluid management strategies.

Our data indicate that an initial fluid resuscitation volume of 10–15 mL/kg within the first 3 h may be optimal for patients with sepsis and ADHD, particularly in those with worsened cardiac function. This finding is significant as it suggests a tailored approach to fluid management that takes into account the acute changes in cardiac status that can occur in the context of sepsis. It is important to note that for fluid volumes within 15–20 mL/kg, there were no statistically significant differences in in-hospital mortality, 30-day, and 1-year mortality rates compared to the 10–15 mL/kg range. However, it is crucial to acknowledge that this higher range significantly increases the risk of endotracheal intubation. Subgroup analysis revealed that patients classified as NYHA III and IV experienced the greatest benefits from fluid resuscitation within the 10–15 mL/kg range. This range notably outperformed other ranges in terms of reducing mortality rates, as well as intubation rates. Our analysis found no significant association between fluid volume and ICU admission, suggesting that clinicians may lean towards ICU admission for such patients regardless of the administered fluid volume.

Individualized fluid management in sepsis improves outcomes in patients with septic shock [11, 27]. The recommendation of 30 mL/kg fluid resuscitation originates from the early goal-directed therapy (EGDT) trial conducted by Rivers et al. in 2001 [28]. Subsequently, both the Surviving Sepsis Campaign (SSC) [1, 4] and Severe Sepsis and Septic Shock Early Management Bundle (SEP-1) have endorsed this approach. From 2014 to 2015, three independent, government-funded, multicenter randomized controlled trials were conducted: early septic shock in the USA (ProCESS) [7], sepsis resuscitation assessment in Australia (ARISE) [29], and sepsis programmed management in the UK (ProMISe) [9]. The average fluid volume administered to patients prior to randomization in these studies was also around 30 mL/kg, which led to the adoption of 30 mL/kg fluid resuscitation volume in routine clinical practice. However, there has been significant debate regarding the optimal fluid volume for sepsis management [30,31,32]. Increasing evidence suggests that the fluid administration during the EGDT era, which followed a permissive intravenous fluid strategy, resulted in suboptimal treatment outcomes [10, 33, 34]. This has raised persistent questions about the appropriateness of the 30 mL/kg fluid resuscitation volume [35]. Additionally, the guidelines do not mention a specific rehydration scheme for septic patients at risk of volume overload, and there is no randomized controlled study for reference. Individualized fluid resuscitation may be necessary for patients at risk of volume overload, as the fixed fluid rehydration amount of 30 mL/kg may not be suitable for all sepsis subgroups [12]. The CLOVERS trial demonstrated that early administration of vasopressors to reduce fluid resuscitation volumes did not decrease the 90-day mortality rate in patients with septic hypotension [34]. In line with these findings, our study emphasizes the importance of tailored fluid management strategies in sepsis, particularly for patients with comorbid heart failure, where the optimal initial fluid resuscitation volume of 10–15 mL/kg within the first 3 h may offer the best clinical outcomes.

Although some studies have shown that the resuscitation protocol of 30 mL/kg for patients with HF does not increase adverse outcomes [15, 16, 36], there is concern in practice that fluid administration at this volume may worsen the condition of these patients [6, 12, 17, 37]. Consequently, compliance with the SSC standard protocol is less likely [12, 16]. Studies have indicated that patients with HF often do not adhere to the 30 mL/kg rehydration protocol, and non-compliance with the rapid infusion protocol has been shown not to increase the mortality rate [6, 12, 16,17,18]. This suggests that it may be unreasonable to provide a one-size-fits-all rehydration volume of 30 mL/kg for patients with HF [12]. However, there is currently no research exploring the optimal initial fluid resuscitation volume range for septic patients with HF. Instead, studies have primarily compared different outcomes using a 30 mL/kg threshold.

The administration of fluids to septic patients with HF requires a careful balance to avoid both inadequate resuscitation and fluid overload, which are both associated with negative outcomes in septic shock. Our study suggests that administering 10–15 mL/kg of fluid within the first 3 h can help mitigate the risks of under-resuscitation and over-resuscitation. Powell et al. found that administering a 30 mL/kg intravenous infusion within the first 6 h to patients with septic shock and pre-existing HF (HFrEF < 40%) can lead to a decrease in in-hospital mortality [37], which may be similar to our study where patients received 10–15 mL/kg intravenous infusion within 3 h. Although the SSC [1, 4] and SEP-1 guidelines recommend initial fluid resuscitation based on body weight, it is unclear whether this is actual weight, predicted weight, or ideal weight (although the latest 2021 guidelines mention ideal weight, the original source could not be found). Even when following these guidelines, estimating patient weight is often necessary in cases of septic shock, and the empirical administration of 30 mL/kg of fluid may not be reliable. In our study, the decision to use IBW was made to standardize fluid resuscitation volumes across a diverse patient population, including those with varying degrees of obesity and underweight conditions. By using IBW, we aimed to minimize the risk of fluid overload, which is a significant concern in critically ill patients, especially those with HF.

Our optimal liquid volume range was 10–15 mL/kg, which is inconsistent with previous studies [15, 16, 36]. This discrepancy may be attributed to the predominantly severe HF patients in our study cohort. As the severity of HF escalates, the volume of fluid resuscitation required tends to decrease. Therefore, our recommended fluid volume ranges are particularly applicable to patients with severe HF. For patients with milder HF, fluid requirements may exceed 15 mL/kg. Unfortunately, a precise estimate for this subgroup could not be determined due to the limited sample size. Although the 15–20 mL/kg fluid resuscitation range did not statistically differentiate in-hospital mortality in the NYHA III and IV subgroups and the BNP > 4000 subgroup, it significantly increased the rate of tracheal intubation. In the HFrEF < 50 subgroup, our study did not find a correlation between fluid volume and intubation, which may be due to the inherently high intubation rate within this subgroup and the limited sample size that did not show statistical significance.

This study has several strengths. Firstly, we utilized a generalized additive model [24], which takes into account the nonlinear relationship between fluid volume and outcomes, allowing for exploration of the optimal fluid volume range. Secondly, we gathered comprehensive data for all outcome events. Thirdly, although this study is retrospective, the inclusion criteria for septic patients strictly adhered to the Sepsis 3.0 definition. This means that even patients treated before 2016 were classified according to the latest definition. The diagnosis of HF was also based on clearly documented NYHA classifications in EMR, with all ambiguous cases excluded. Finally, our study exclusively focused on patients who were considered for standard care and fluid resuscitation management. Patients who were under comfort care or hospice care protocols were not included in this analysis to ensure that our findings reflect the impact of fluid resuscitation in patients who are candidates for aggressive treatment measures, and this research is based on real-world data.

Several limitations should be acknowledged in this study. Firstly, being a retrospective study, there is a possibility of unrecorded data in medical records, making it challenging to fully account for potential biases that could affect the analysis. Secondly, despite our efforts to consider all possible factors, there may still be unadjusted variables that could impact the results [38]. Thirdly, it is difficult to assess the specific reasons behind patients not receiving fluids, whether due to concerns about fluid overload or other valid reasons. Fourthly, our patient population includes individuals from both the emergency department and the wards, and the timing of sepsis recognition by different physicians or the duration of pre-admission sepsis may vary, potentially influencing patient outcomes. Fifthly, when collecting data on fluid volumes, we solely focused on crystalloid solutions administered for resuscitation purposes, excluding maintenance fluids intended for hypoperfusion correction. Sixth, we recognize that the distinction between chronic and acute HF is not always clear-cut, and our study’s findings should be interpreted in the context of a patient population that is likely experiencing an acute exacerbation of their HF. Future research should continue to explore the nuances of fluid management in this complex patient population, with the goal of developing evidence-based guidelines that can improve outcomes for patients with sepsis and HF. Another limitation of our study is the sample size constraint, which limited our ability to perform subgroup analyses for HF categories with EF < 50% as recommended by ACC/AHA guidelines, potentially affecting the generalizability of our findings to more diverse HF populations. Seventh, we have acknowledged the limitations related to the lack of more detailed clinical indicators of acute HF (i.e., pulmonary edema, echo showing cardiogenic shock, peripheral edema/JVP). We recognize that the absence of such data may limit our ability to fully characterize the acute decompensation status of our patients. Finally, our study is its conduct within a single tertiary care center, which may constrain the generalizability of our findings to other healthcare settings or diverse patient populations.