Abstract Background The purpose of patient monitoring is to identify when immediate assessment to prevent morbidity and death is needed. Aim To determine the optimal description of level of consciousness, using death within 24 h as a surrogate marker for when a patient needs to be assessed immediately. Methods Observational non-interventional study: the level of consciousness of acutely ill medical patients to a low resource Ugandan hospital was assessed daily using a modified version of the Richmond Agitation and Sedation Scale, Results 2579 daily observations (3.4 per admission) were performed on 757 patients; while in hospital 43 (5.7%) patients died, 42 of whom died within 24 h of their last observation. The inability to answer questions using speech had the highest uncorrected odds ratio for 24 h mortality (OR 14.4, 95% CI 7.2–29.3), the highest odds ratio when corrected for age and sex (OR 12.3, 95% CI 6.1–24.8), and highest Youden number (0.545). Conclusion The inability to answer questions using speech was the level of consciousness assessment that best predicted 24 h mortality. As it is simple and unambiguous, it may help in the efficient identification of patients who need an immediate full clinical assessment. Key messages What is already known on this topic: Level of consciousness is frequently assessed by the AVPU score (Alert, responds to Verbal stimulus, responds to Painful stimulus and Unresponsive). However, AVPU has low rates of inter-rater agreement, and nurses may not agree on how to distinguish between “Alert” and “Voice.” Moreover, there are few reports of its ability to accurately identify patients who need to be seen immediately. What this study adds: This study found the ability to answer questions using speech was the consciousness level within the AVPU or the Richmond Agitation and Sedation Scale that best predicted death within 24-hours and, hence, the need to be seen immediately. How this study might affect research, practice, and policy: More research is needed to determine the monitoring variables that most efficiently bring time-dependent treatments to patients while they are still effective. The inability to answer questions promises to be a rapid, easy, unambiguous and efficient method of identifying patients with a reduced level of consciousness who need immediate attention.
Muhumuza et al. (Wed,) studied this question.