Escalation of Care in Surgery: a Systematic Risk Assessment to Prevent Avoidable Harm in Hospitalized Patients
We read with interest the recent article by Johnston et al1 examining pathways inherent to the escalation of surgical patient care. This welcome addition introducing ethnographic observations to mainstream medical research acknowledges that study questions involving human factors cannot be answered by traditional empirical methods of data collection and analysis. The comprehensive breakdown of a complex system such as the escalation process forms a good basis for future work and analysis into the escalation of care whilst the use of the Healthcare Failure Mode Effect Analysis (HFMEA) framework increases the study's validity and allows it to build on the experiences of the field to tackle the subject matter without having to create a new approach. We would like, however, to comment on a few limitations of this study.
First, the researchers achieved their stated aim of a systematic risk assessment and clearly demonstrated how they had carried out the various steps. However, the limited number of observation hours and number of staff observed is likely to have underestimated process failures. It would have been helpful to have clarification of the criteria used to determine which events, results or patient observations were used to identify a patient deteriorating and therefore what “deteriorations” were deemed appropriate for escalation of care. This is essential to allow future comparative work.
Observer bias (so-called “Hawthorne effect”2) is to some extent unavoidable and inevitably influences the behavior resulting in underestimation of the degree of escalation failures. The steps in the escalation process are overall well presented. The influence of other patients in the hospital setting alerting clinical staff to a change in another patient's condition is not mentioned in this article but is likely to be a significant contributing factor. This additional pathway of escalating care will perhaps become less common with the current trend toward single patient rooms in newer hospital design and could potentially lead to missed opportunities for early intervention.
The inter-professional group consensus methodology, used throughout the article, is based on the subjective views of the participants and as such is inherently open to bias, which the authors have rightly sought to minimize. However, in the determination of hazard scores for each failure mode, the consequences of failing to observe a step in the escalation pathway are unpredictable and do not necessarily reflect true outcomes. This is a crucial area as flawed data at this stage will affect the findings in subsequent steps.
We recognize that the results from phase 4 are the consensus of the modified HFMEA group and not the authors. However, we would disagree with some of the distinctions between controlled and uncontrolled failures such as the nurse failing to assess the patient correctly, deemed a controlled failure given the access to vital signs recording equipment, in contrast to a nurse failing to notice a patient is deteriorating or record the vital signs accurately which was thought to be an uncontrolled failure due to lack of specialty training or lack of appropriate technology. The recommendations for improvement in this specific example are also not valid—an electronic chart will not improve a nurse failing to measure vital signs correctly and may indeed introduce an additional potential hazard. Having existing controls in place does not reduce failure if they are not properly used or implemented, which highlights the influence of human factors. Individual failures are not institutional and yet they could explain almost every hazardous failure, the majority of which are errors of process, not escalation.