Researchers: Yuval Shahar¹

Background: To improve medical care for the aging population and to reduce its costs, the use of a clinical decision support system for both ensuring and assessing the quality of care, based on computerized clinical guidelines (GLs), may improve care, reduce efforts, save time, and enhance nursing staff capabilities.

Objectives: Implement and evaluate a decision support system for managing the medical treatment of pressure ulcers for geriatric patients admitted to the Herzfeld Geriatric Medical Center in the nursing wards. In the first stage, following the IRB committee’s instructions, we investigated the level of compliance of the clinical and nursing staff to the GLs by using retrospective data. In the second step, we will investigate the staff's compliance to system recommendations and alerts prospectively.

Method: The study included several steps:
1. Building the protocol for the treatment of pressure ulcers in a computerized manner;
2. Implementing a computational system that assesses the quality of care using the GL and the patient longitudinal database, including providing partial scores for partial performance;
3. Technical evaluation of the system, in order to check the applicability and usefulness of the system.
4. Functional evaluation of the system.
To do this, we selected 29 randomized patients. Two experienced geriatric-care nurses evaluated the performance of the medical staff for half of the patients using the new system and for half of them without the system. We compared the manual score the nurses gave in each stage to the system’s scores, and also the time it took the nurses to perform the assessment with and without the system. We report the results in detail for the first nurse.

Findings: There were no significant differences (P>0.05) across the values of most quality measures given by the nurse manually. Versus the values given by the system automatically. That is, in practice, the system appears to give a very similar rating to that of an experienced senior nurse. There were also no significant differences (P>0.05) in most measures when the nurse used the system, versus the scores given by an experienced knowledge engineer using the same system; i.e., by using the system after a short training, a senior nurse is able to find the correct values of the quality measures. Using the system significantly reduced the average assessment time it took for the nurse to score each patient across all quality measures, reducing the time, even at only the 14th patient evaluated by the nurse, from 1,039.29 ± 242.4 seconds to only 634.29 ± 303.87 seconds on average.

Conclusions: Decision-support systems such as the one developed and assessed here may empower the nursing staff, who will be able to manage more patients, and in a more correct fashion; most likely also in other types of disorders, while reducing treatment costs. We are now preparing the infrastructure for the prospective phase. Next, we would like to examine additional protocols, such as pain management and early detection of deterioration in a patient's condition.

Recommendations: We recommend connecting the system to the computerized medical record and running a retrospective quality control system as a routine and to produce all quality control reports with its help.
We also recommend a more comprehensive recommendation for the health care system: move on to the second phase of the project as soon as possible - that is, developing real-time therapeutic recommendation systems for both the nursing and medical staff.

Research end date: 10/2019