Filip Haegdorens, The effect of a rapid response system on patient outcomes in Belgian acute hospitals
Promotoren: Prof. dr. Peter Van Bogaert, Prof. dr. Koenraad Monsieurs, Universiteit Antwerpen
Prof. dr. Christiaan Vrints, Prof. dr. Bart Van Rompaey, juryleden Universiteit Antwerpen
Prof. dr. Sean Clarke, New York University (United States of America), Dr. Christian P. Subbe, Bangor University (United Kingdom), externe juryleden
Deterioration of hospitalised patients is often missed, misinterpreted, and mismanaged and can eventually result in death. Rapid response systems (RRSs) provide a framework to detect and interpret in-hospital clinical deterioration, to enhance communication between clinicians, and to initiate a response in a timely manner. The main goal of RRSs is to prevent progressing patient deterioration and subsequently to reduce preventable serious adverse events improving the quality of care. A considerable body of research exists studying different aspects and types of RRSs. However, the quality of evidence in relation to patient outcomes is poor and there is no consensus on what the most effective strategy is to prevent serious adverse events. The general aim of this doctoral study is to investigate the effect of a rapid response system on patient outcomes using an evidence-based afferent limb strategy while ensuring an adequate and timely medical response by existing hospital resources. Furthermore, we studied the predictive performance of the national early warning score (NEWS), which is a track-and-trigger system to detect patient deterioration using vital signs. Because of the importance of adequate staffing levels for patient surveillance, we additionally investigated which nurse staffing levels are needed to provide safe care and to minimise patient’s unexpected death.
We defined new outcome measures, more sensitive to care processes including nursing practice, to be used in research concerning the deteriorating patient on the general ward. We confirmed an association between higher adherence to a nurse observation protocol and lower patient unexpected death and mortality after cardiopulmonary arrest or unplanned intensive care admission adjusted for patient comorbidity and age. Furthermore, we have shown that an RRS improves nurses’ performance in observing patients without increasing the observation frequency in stable patients. We showed that our intervention increased the mean number of vital signs registered per observation. In an external validation study, we calculated the predictive performance of the NEWS and confirmed that it shows acceptable performance to predict unexpected death, cardiac arrest with CPR and unplanned ICU admission. However, the NEWS had a relatively high proportion of false positive scores. This implies that clinicians should be aware of the limitations of this scoring system to prevent increased workload associated with false positives. Lastly, we found an association between higher nurse staffing levels and lower patient mortality controlling for important confounders. We confirmed the relation between a higher proportion of bachelor’s degree nurses and patient mortality.