ABSTRACT
Pain management is often considered lower priority than many other aspects of health management in hospitals. However, there is potential for Quality Improvement (QI) teams to improve pain management by visualising and exploring pain data sets. Although dashboards are already used by QI teams in hospitals, there is limited evidence of teams accessing visualisations to support their decision making. This study aims to identify the needs of the QI team in a UK Critical Care Unit (CCU) and develop dashboards that visualise longitudinal data on the efficacy of patient pain management to assist the team in making informed decisions to improve pain management within the CCU. This research is based on an analysis of transcripts of interviews with healthcare professionals with a variety of roles in the CCU and their evaluation of probes. We identified two key uses of pain data: direct patient care (focusing on individual patient data) and QI (aggregating data across the CCU and over time); in this paper, we focus on the QI role. We have identified how CCU staff currently interpret information and determine what supplementary information can better inform their decision making and support sensemaking. From these, a set of data visualisations has been proposed, for integration with the hospital electronic health record. These visualisations are being iteratively refined in collaboration with CCU staff and technical staff responsible for maintaining the electronic health record. The paper presents user requirements for QI in pain management and a set of visualisations, including the design rationale behind the various methods proposed for visualising and exploring pain data using dashboards.
ABSTRACT
BACKGROUND: In-hospital pain services (IPS) are commonplace, but evidence of efficacy is inadequate, and patients' pain management in any hospital ward remains problematic. This service evaluation aimed to measure the effect of a contemporary IPS, its appropriate use and cost-efficacy. METHODS: Records of 249 adults reviewed by the IPS in an inner London Teaching Hospital over an 8-month period were analysed for demographic data, interventions, workload and change in pain intensity measured by numerical rating scale (NRS). Non-parametric tests were used to evaluate differences between initial and final NRS. Spearman's rank correlation analysis was used to create a correlation matrix to evaluate associations between all identified independent variables with the change in NRS. All strongly correlated variables (ρ > 0.5) were subsequently included in a binary logistic regression analysis to identify predictors of pain resolution greater than 50% NRS and improvement rather than deterioration or no change in NRS. Finally, referral practice and cost of inappropriate referrals were estimated. Referrals were thought to be inappropriate when pain was not optimised by the referring team; they were identified using a set algorithm. RESULTS: Initial median NRS and final median NRS were significantly different when a Wilcoxon signed-rank test was applied to the whole cohort; Z = -5.5 (p = 0.000). Subgroup analysis demonstrated no significant difference in the 'mild' pain group; z = -1.1 (p = 0.253). Regression analysis showed that for every unit increase in initial NRS, there was a 62% chance of general and a 33% chance of >50% improvement in final NRS. An estimated annual cost-saving potential of £1546 to £4558 was found in inappropriate referrals and patients experiencing no benefit from the service. DISCUSSION: Results suggest that patients with moderate to severe pain benefit most from IPS input. Also pain management resources are often distributed inefficiently. Future research is required to develop algorithms for easy identification of potential treatment responders.