Quality of care has been systematically monitored in hospitals in high-income countries to ensure adequate care. However, in low- and middle-income countries, quality indicators are not readily measured. The primary aim of this study was to assess to what extent it was feasible to monitor the quality of intensive care in an ongoing health emergency, and the secondary aim was to assess a quality of care intervention (twinning project) focused on Intensive Care Unit (ICU) quality of care in public hospitals in Lebanon. We conducted a retrospective cohort study nested within an intervention implemented by the World Health Organization (WHO) together with partners. To assess the quality of care throughout the project, a monitoring system framed in the Donabedian model and included structure, process, and outcome indicators was developed and implemented. Data collection consisted of a checklist performed by external healthcare workers (HCWs) as well as collection of data from all admitted patients performed by each unit. The association between the number of activities within the interventional project and ICU mortality was evaluated. A total of 1679 patients were admitted to five COVID-19 ICUs during the study period. The project was conducted fully across four out of five hospitals. In these hospitals, a significant reduction in ICU mortality was found (OR: 0.83, P < 0.05, CI: 0.72-0.96). We present a feasible way to assess quality of care in ICUs and how it can be used in assessing a quality improvement project during ongoing crises in resource-limited settings. By implementing a quality of care intervention in Lebanon's public hospitals, we have shown that such initiatives might contribute to improvement of ICU care. The observed association between increased numbers of project activities and reduced ICU mortality underscores the potential of quality assurance interventions to improve outcomes for critically ill patients in resource-limited settings. Future research is needed to expand this model to be applicable in similar settings.
Read full abstract