Vigilance in Cardiac Telemetry Monitoring: Performance Outcomes and Effects on Operators' Cognitive and Affective States
TimeThursday, April 152:00pm - 3:00pm EDT
DescriptionBackground: Past research shows that vigilance performance declines because of the diminishment in available attentional resources. This diminishment in attentional resources, known as the vigilance decrement, has its foundational origin within resource theory, or the direct-cost model of vigilance (Mackworth, 1948; Parasuraman, 1979; Warm, Parasuraman, & Matthews, 2008). According to resource theory, event rate, target type, source complexity, and sense modality are all factors that influence the manifestation of the decrement. In healthcare, the vigilance decrement results in errors that potentially influence life-or-death outcomes in many high-level care environments such as cardiac telemetry floors and intensive care units. In cardiac telemetry monitoring, clinicians are asked to monitor a visual display for an extended period of time while scanning visual stimuli (i.e., patients’ electrocardiogram) for rare, but critical anomalies in cardiac performance.
Objectives: This study aimed to examine how vigilance performance in a cardiac telemetry monitoring task is influenced by task difficulty. The secondary objective is to investigate the effects of the vigil on psychological (i.e., mental workload and stress) and physiological functioning (i.e., heart rate variability (HRV)). The hypotheses focus on the manifestation of the vigilance decrement as characterized by fewer correct detections and false alarms, and its associated changes in perceived levels of workload and stress, as indicated by objective physiological functioning (decreased HRV) and subjective responses to validated questionnaires (Dundee Stress State Questionnaire (DSSQ) and NASA Task Load Index (NASA-TLX)).
Methods: 38 participants were recruited for this study from the California State University, Long Beach (CSULB) School of Nursing and Psychology Department. Prior to completing the study, participants were required to meet the following criteria: (1) read and sign the informed consent, (2) not meet any of the exclusion criteria, (3) be willing wear physiological data collection equipment, and (4) complete a demographics questionnaire. The participants were asked to remain seated at a desktop computer for one hour and thirty minutes while searching and verbally stating which patient the target event (the word “CAUTION”) appears. The pre-recorded and de-identified EKG videos were created by Noa Segall and her team of researchers at Duke University Medical Center in Durham, N.C. with support from the Agency for Healthcare Research and Quality (AHRQ; Segall, et al., 2015). Two levels of recorded EKG output were used for the vigilance task: low workload condition (8 patients) and high workload condition (16 patients). The target events were presented at pseudo-random intervals to prevent target signals from appearing on neighboring patients consecutively. The word “CAUTION” was presented for 5 second intervals before disappearing (i.e., the length of one EKG wave cycle). The independent variable of interest is mental workload, which was manipulated by the number of patients the participant is expected to monitor. The dependent variables were vigilance performance, subjective reports of stress and workload, and physiological functioning. Vigilance performance was measured via correct detections (hits) and false alarms. Cognitive and affective states was measured using participants’ subjective ratings of workload via the NASA-TLX (Hart & Staveland, 1988) and an abbreviated version of the DSSQ (Matthews et al., 1999; 2002; 2013). Physiological arousal was measured via heart rate variability (HRV) with the data collected from the BioPac MP160 data collection system (BioPac Systems Inc., Aero Camino Goleta, CA, US) and Acknowledge software (version TBD, BioPac Systems Inc., Aero Camino Goleta, CA, US). A mixed model ANOVA was performed on the correct detections and false alarms. The mixed model ANOVA was required because mental workload was a between-subjects variable, while duration of the vigil was a within-subjects variable. For all analyses, a probability value of p < .05 was assumed. Appropriate post hoc tests were used to further assess the results depending on the significance of the statistical analyses. The number of correct detections measured in the two workload conditions was compared against the time spent on vigil. A 2 (MENTAL WORKLOAD: High vs. Low) × 6 (PERIODS ON WATCH) mixed model ANOVA was conducted with repeated measures on the second factor. The cardiac measure (HRV difference scores) of the two workload conditions will be compared against the time spent on vigil. A 2 (MENTAL WORKLOAD: High vs. Low) × 6 (PERIODS ON WATCH) mixed model ANOVA was conducted with repeated measures on the second factor. The cognitive and affective states (as quantified by the NASA-TLX and DSSQ) was compared across the workload conditions. All questionnaire results were analyzed with a between subjects (MENTAL WORKLOAD: High vs. Low) ANOVA.
Results: As predicted, participants in the Low Workload Condition detected significantly more critical signals compared to the High Workload Condition. In addition, participants in the High Workload Condition also committed significantly more false alarms compared to participants in the Low Workload condition. The hypotheses for cognitive and affective states as it relates to difference scores on the DSSQ were not statistically significant, however, participants in both workload conditions reported high levels of workload when NASA=TLX scores were compared to the 50th percentile for monitoring and visual search tasks as reported by Grier (2015). The results for the cognitive and affective states support the findings of the Extended-U model (Hancock & Warm, 1989; Hancock, Ross, & Szalma, 2007)