Background: The HCAHPS survey is used nationally by health systems as a standard to measure the patient experience and identify opportunities for improvement. Historically, process improvement initiatives based on the survey are implemented retroactively, attempting to prevent negative experiences for future patients. However, given the increasing complexity of patient care, proactively addressing the identified concerns of individual patients will improve patient care and communication.Previously, a large observational study performed at our health system analyzed HCAHPS surveys and developed a machine learning algorithm to identify characteristics of patients who are more likely to report a negative experience with a sensitivity between 72 and 80%. Variables that contribute to this algorithm include prolonged length of stay, higher comorbidity index, pain score, and prior negative survey responses. By utilizing this predictive model and integrating it within the care model unit, the aim is to specifically target and perform interventions in real time to improve the patient experience.

Purpose: The purpose of the intervention is to positively impact the patient experience by implementing a patient experience predictive model within morning interdisciplinary rounds and afternoon huddles.

Description: The pilot was initiated on a high functioning medical floor identified with opportunities for improvement in doctor communication based on prior HCAHPS surveys. Leveraging a predictive model, a HIPAA compliant interface known as the “IDR tool” displays a purple flag (“Flag” patient) for those high-risk patients identified as more likely to rate the hospital negatively.During interdisciplinary rounds, the nurse manager reviews the IDR tool and informs the multidisciplinary team of Flag patients. These patients receive an initial tuck-in visit from the medical team. A patient care representative visits these patients to identify any concerns, and directly escalates them to the nurse manager and team. Team members update the tool to reflect the date of visits and note comments. Realtime feedback enables the staff to address concerns in a timely matter. Examples of escalations include patient or family questions specific for the medical team to clarify the care plan, dietary complaints, and questions about medications.The multidisciplinary team regroups during the afternoon and review over any concerns from Flag patients. Using processes that already exist on the floor, follow up appointments are made for patients, pharmacy education is provided, and home medications are delivered to the patient’s bedside. Prior to discharge, these patients are visited again and “tucked in”.Since the start of the pilot in late June 2021, improvements were noted in the HCAHPS survey results beginning in Quarter 3 (July through September 2021) when compared to prior quarters. Recommend the hospital improved in top box percentage points from 64.8% to 68.2%. Survey responses revealed an improvement in physician communication, and understanding of medications between 5-12%. The care transitions domain, which encompasses an understanding of health management and medications upon discharge also improved up to 5.9%. Team members found the intervention improved communication amongst the patients as well as within the team.

Conclusions: Efficient, quality patient care incorporating a predictive model can improve the overall patient experience as well as multidisciplinary team satisfaction.