Search

Purpose: The increasing use of advanced medical technologies to detect adverse events, for instance, artificial intelligence‐assisted technologies, has shown promise in improving various aspects within health care but may also come with substantial expenses. Therefore, understanding the potential economic benefits can guide decision‐making processes regarding implementation. We aimed to estimate the potential cost savings associated with reducing length of stay and avoiding readmissions within the framework of an artificial intelligence‐assisted vital signs monitoring system. Methods: We used data from Danish national registries and coarsened exact matching to estimate the difference in length of stay and probability of readmission among adult in‐hospital patients exposed to and not exposed to serious adverse events. We used these estimates to calculate the maximum potential savings that could be achieved by early detection of adverse events to reduce length of stay and avoid readmissions. Results: Patients exposed to serious adverse events during admission had 2.4 (95% CI: 2.4–2.5) additional hospital bed days and had 14% (95% CI 11%–17%) higher odds of readmissions compared with patients not exposed to such events. A base case scenario yielded maximum potential savings if one patient avoided a serious adverse event of EUR 2040 due to reduced length of stay and EUR 43 due to avoidance of readmissions caused by serious adverse events. Conclusion: Reductions in serious adverse events are associated with decreased healthcare costs due to reduced length of stay and avoided readmissions. Artificial intelligence‐assisted vital signs monitoring systems are one potential approach to reduce serious adverse events, however, the ability of this technology to reduce adverse events remains unclear. Comprehensive prospective analyses of such systems including the intervention and implementation costs are necessary to understand their full economic impact. [ABSTRACT FROM AUTHOR]

Over the last decade, the strain on the English National Health Service (NHS) has increased. This has been especially felt by acute hospital trusts where the volume of admissions has steadily increased. Patient outcomes, including inpatient mortality, vary between trusts. The extent to which these differences are explained by systems-based factors, and whether they are avoidable, is unclear. Few studies have investigated these relationships. A systems-based methodology recognises the complexity of influences on healthcare outcomes. Rather than clinical interventions alone, the resources supporting a patient's treatment journey have near-equal importance. This paper first identifies suitable metrics of resource and demand within healthcare delivery from routinely collected, publicly available, hospital-level data. Then it proceeds to use univariate and multivariable linear regression to associate such systems-based factors with standardised mortality. Three sequential cross-sectional analyses were performed, spanning the last decade. The results of the univariate regression analyses show clear relationships between five out of the six selected predictor variables and standardised mortality. When these five predicators are included within a multivariable regression analysis, they reliably explain approximately 36% of the variation in standardised mortality between hospital trusts. Three factors are consistently statistically significant: the number of doctors per hospital bed, bed occupancy, and the percentage of patients who are placed in a bed within four hours after a decision to admit them. Of these, the number of doctors per bed had the strongest effect. Linear regression assumption testing and a robustness analysis indicate the observations have internal validity. However, our empirical strategy cannot determine causality and our findings should not be interpreted as established causal relationships. This study provides hypothesis-generating evidence of significant relationships between systems-based factors of healthcare delivery and standardised mortality. These have relevance to clinicians and policymakers alike. While identifying causal relationships between the predictors is left to the future, it establishes an important paradigm for further research. [ABSTRACT FROM AUTHOR]