In this study, we demonstrated that higher hospital spending, represented as daily medical costs per person, was associated with lower in-hospital mortality and higher effective cost per survivor in patients with sepsis. Among the enrolled hospitals, those with a large number of beds and academic affiliations exhibited inconsistent results in the primary test.
The beneficial effects of hospital spending on mortality may be attributed to an abundance of medical resources. According to a previous report that enrolled patients with sepsis in U.S. hospitals, medical costs increased with higher mortality according to the severity level; however, the association between medical costs and mortality was not clearly demonstrated in this report or in other studies [24]. A cross-sectional study attempted to identify the relationship between hospital expenditure and mortality in patients with sepsis; however, no associations were found between higher medical expenses and better survival [10]. In other fields, greater resource utilization was not associated with an increase in mortality or other clinical outcomes in patients with acute myocardial infarction, chronic heart failure, or durable left ventricular devices, suggesting that high expenditures would not be efficient for hospitals striving for optimal medical allocation [8, 9, 21]. Conversely, only one study demonstrated that higher hospital spending intensity was associated with lower mortality in a mixed cohort of patients with acute myocardial infarction, chronic heart failure, hip fracture, and colon cancer [20]. In our study, we demonstrated a significant association between higher medical expenditure and decreased mortality in the sepsis population; however, the causal relationship remains to be ascertained owing to the limitations of this observational study. The divergent findings of previous studies can be attributed to disparate study designs, variations in healthcare infrastructure, and different pathologies. However, we must be cautious about the possibility that hospitals with lower mortality rates are more likely to treat patients who are eligible for therapeutic interventions based on their age and background characteristics.
To estimate the efficiency of hospital spending in healthcare, the cost-effectiveness is used as a representative economic indicator also in the medical field [25]. As sepsis requires substantial medical resources, including medications, artificial organ support, and experienced staff, we should be cautious about the cost-effectiveness of sepsis care to prevent deterioration in the balance of the fiscal budget [6]. Additionally, the optimal allocation of medical resources should be considered based on the cost-effectiveness of medical policies and treatments. Encouragingly, the cost-effectiveness of sepsis has gradually improved over the course of 8 years despite the increasing number of patients with sepsis and medical costs, possibly due to an increased rate of adherence to sepsis guidelines and advanced quality of sepsis management [4, 26, 27]. In this study, the effective cost per survivor was higher in the highest hospital spending group, suggesting that cost-effectiveness did not improve despite the lower mortality. As the median age of the highest hospital spending group was lower than that of the other groups, higher hospital spending might contribute to enhancing clinical outcomes in the working-age population. In future investigations, detailed analyses with quality-adjusted life years, estimating health-related quality of life, are expected to clarify promising tactics to enhance economic outcomes as well as improve clinical consequences in the sepsis population [28, 29].
Variations in medical costs among hospitals have been reported for several diseases [8, 14, 21, 22]. In the present study, we found some variations in the influence of the number of hospital beds, teaching hospitals, and census regions on in-hospital mortality. As the mortality of patients with sepsis was reportedly lower in hospitals with a higher case volume because of greater availability of medical resources and a larger number of experienced staff [30–33], a plausible explanation for the relationship between higher hospital spending and lower mortality could be a higher proportion of large hospitals. However, the results of the subgroup analysis denied the possibility of this speculation. Given that hospitals with a large number of beds potentially provide medical care that exceeds the standard of practice through more experienced medical staff and higher adherence to evidence-based practices [34–36], further improvement might be difficult to obtain with additional hospital spending. Similarly, teaching hospitals showed no correlation between the two variables, suggesting that hospitals with abundant medical resources would not yield disparate clinical outcomes, even if more substantial medical resources were allocated to sepsis management. Considering that hospitals with a small or medium number of beds demonstrated results consistent with the primary results, a small or medium number of hospital beds could improve patient outcomes through the additional allocation of medical resources. Other potential approaches for enhancing outcomes in these hospitals include encouraging adherence to sepsis protocols and optimal timing of transportation to large hospitals [5, 37]. Although we excluded transferred patients to analyze medical costs in our study, the timely transport of patients with sepsis would likely enhance their survival rate. Conversely, initial resuscitation would have a substantial impact on clinical outcomes, and medical staff should be cautious about the significance of initial resuscitation and the timing of transportation [38, 39].
Geographical variations in medical costs and quality of care have been reported for a range of conditions, including myocardial infarction, heart failure, surgery, trauma, and sepsis [15, 40–48]. With regard to trauma, the distance between the scene of the accident and the nearest hospital has been found to have a negative correlation with clinical outcomes [45, 46], as critical interventions such as transfusion, tracheal intubation, and surgical procedures are likely to be delayed in accordance with the arrival time. Likewise, clinical outcomes in patients with sepsis may worsen as the length of transportation increases [38, 39]. Given that early recognition and prompt intervention are vital for improving the survival of patients with sepsis [38, 39], geographical factors are strongly associated with early intervention. A previous report that examined regional disparities in access to plastic surgery in Japan revealed high coverage rates in the Kanto, Chubu, and Kansai areas [49], with regional variations similar to our findings. This regional disparity, potentially arising from distinctions in resident demographics and a gap in access to hospitals, could have affected the various relationships between in-hospital costs and mortality. A more comprehensive understanding of the differences among census regions can be obtained through a detailed analysis of the distribution of hospitals and medical resources.
This study has several limitations. First, the data were collected retrospectively. Second, a selection bias might have remained in the cohort despite the meticulous selection of patients and hospitals, which could represent the acute phase of sepsis treatment. Third, detailed medical costs for sepsis treatment were not collected from the database. Fourth, the medical spending investigated in this study differed from medical bills reimbursed on a bundled payment basis. Fifth, a prominent causal relationship was not shown in this study. Such a limitation may go beyond the design of observational studies, whereas randomized controlled studies attempting to show the positive effect of higher medical expenses on mortality would have some difficulties with their implementation in patients with sepsis. Sixth, long-term outcomes or parameters concerning the quality of life were not obtained from the database. As it would be highly challenging to extract such information from one database, future investigations should link other databases to the current database.