Review of Prediction Model of Pre-discharge MRMI in Stroke with Higher Accuracy

Abstract Description

Proposed Topic (Most preferred): :

Clinical Safety and Quality Service II (Projects aiming to enhance clinical safety and outcomes, clinical governance / risk management)

Authors (including presenting author) :

Wong HL(1), Chung PH(1), Lau FO(1)

Affiliation :

(1)Physiotherapy Department, Tai Po Hospital

Introduction :

Since 2019, Accelerated Stroke Ambulation Program (ASAP) was started to develop in Physiotherapy Department in Tai Po Hospital (TPH). Stroke Registry, a longitudinal stroke database with a clinical prediction model of Modified Rivermead Mobility Index (MRMI) gain, is one of the vital parts which brought the program success. Taken the predicted MRMI gain as an indicator, we can monitor the progress of each stroke patient proactively after initial assessment. This promoted a goal-oriented approach and timely supported the therapists and patients. With the program’s success, the stroke outcome data might change these few years and it is necessary to review the prediction model with updated data to enhance its accuracy.

Objectives :

To enhance accuracy of prediction model of pre-discharge MRMI in stroke for outcome monitoring and clinical decision support.

Methodology :

TPH stroke data after ASAP from Oct 2019 to Sep 2023 (n=1700) were analyzed with SPSS. To replace the old formula, standard multiple regression model was used this time. Age, premorbid Modified Functional Ambulatory Category (MFAC) and admission MRMI were variables to predict the pre-discharge MRMI.

Result & Outcome :

Compared to the old model with Pearson correlation of 0.386, the standard multiple regression model generated a much higher correlation coefficient (R=0.893) which showed a strong correlation between the predicted and actual values of the pre-discharge MRMI. The adjusted R2 was 0.798. This means that age, premorbid MFAC and admission MRMI explained 79.8% of the variability of pre-discharge MRMI. All the individual slope coefficients were statistically significant with p<0.001 and the whole prediction model was also statistically significant, F(3, 1696) = 2235.276, p<0.0005. The regression equation was “Predicted Pre-discharge MRMI = 11.313 - (0.141 x Age) + (0.902 x Premorbid MFAC) + (0.909 x Admission MRMI)”. This new prediction model enhanced the prediction accuracy for better stroke outcome monitoring and precise clinical decision.