Purpose: This study was conducted to investigate relationship between delirium, risk factors on delirium, and patient prognosis based on Donabedian’s structure–process–outcome model. Methods: This study utilized a path analysis design. We extracted data from the electronic medical records containing delirium screening data. Each five hundred data in a delirium and a non-delirium group were randomly selected from electronic medical records of medical and surgical intensive care patients. Data were analyzed using SPSS 20 and AMOS 24. Results: In the final model, admission via emergency department (Β=.06, p=.019), age over 65 years (Β=.11, p=.001), unconsciousness (Β=.18, p=.001), dependent activities (Β=.12, p=.001), abnormal vital signs (Β=.12, p=.001), pressure ulcer risk (Β=.12, p=.001), enteral nutrition (Β=.12, p=.001), and use of restraint (Β=.30, p=.001) directly affecting delirium accounted for 56.0% of delirium cases. Delirium had a direct effect on hospital mortality (Β=.06, p=.038), hospital length of stay (Β=5.06, p=.010), and discharge to another facility (not home) (Β=.12, p=.001), also risk factors on delirium indirectly affected patient prognosis through delirium. Conclusion: The use of interventions to reduce delirium may improve patient prognosis. To improve the dependency activities and risk of pressure ulcers that directly affect delirium, early ambulation is encouraged, and treatment and nursing interventions to remove the ventilator and drainage tube quickly must be provided to minimize the application of restraint. Further, delirium can be prevented and patient prognosis improved through continuous intervention to stimulate cognitive awareness and monitoring of the onset of delirium. This study also discussed the effects of delirium intervention on the prognosis of patients with delirium and future research in this area.