Evaluation of the Predictive Validity of a Social Triage Tool to Screen for Socially-at-risk Frail and Complex patients in Singapore

Background: In Singapore, under Yishun Health, National Healthcare Group (NHG), the Population Health and Community Transformation (PHCT) division has been providing post-hospital discharge care for frail and complex patients in their own homes through the Aging-in-Place Community Care Transitions (AIP-CCT) programme. Through an inter disciplinary team of health professionals, the AIP-CCT programme aims to address the medical, social and functional needs of patients, with a strong emphasis on self-management and patient/caregiver empowerment, while working with community partners and volunteers to address residents’ long-term needs. The AIP-CCT team designed a Social Triage Tool to assist nurses in screening patients for social risks. An expert panel of senior geriatrician, senior medical social workers and nurse clinicians reviewed the tool for content validity. Early identification and interventions of social needs can potentially improve health outcomes and reduce healthcare utilisations. This is a preliminary evaluation of the predictive validity of the Social Triage Tool in identifying social needs amongst frail and complex patients. Methods: The Social Triage Tool has four main domains – Family/Social Support, Mental Health, Compliance and Social/Emotional Impact – with each having a score range of 1 to 3. By summing up all domains, patients are placed in one of the three risk status bands – Low (4-7), Moderate (8-9), High (10-12). Additionally, a ‘Red flag’ category was included to help identify patients who need urgent social interventions. They include 1. Suicidal ideation/attempt, 2. Elder abuse/neglect, and 3. Family violence. 1591 patients were enrolled under the AIP-CCT programme from January 2022 to January 2023. The tool was administered to these patients as part of a first home visit assessment by the AIP-CCT nurse. Patients that scored between 8-12 (Moderate/High-risk) or were triaged as a ‘red flag’ case were considered screened positive for social needs by the tool. The ”Gold Standard” of social needs was defined as referrals to the Medical Social Worker (MSW) during the AIP-CCT episode after inter-disciplinary team discussions. A confusion matrix analysis was conducted to determine the tool’s positive predictive value (PPV), negative predictive value (NPV), sensitivity and specificity. Lastly, a logistic regression analysis was conducted to determine the statistical significance and odds ratio (OR) of the main domains only towards triaging outcomes. Results: The confusion matrix table showed that the tool had a PPV of 0.61 (95%CI [0.56,0.66]), NPV of 0.94 (95% CI [0.93,0.95]), sensitivity of 0.67 (95% CI [0.61,0.73]) and specificity of 0.93 (95% CI [0.91,0.94]). For the logistic regression, all four main domains were statistically significant (p≤0.5) in influencing triaging outcomes, with the following OR – Social/Emotional Impact=4.58 (95% CI [3.25,6.47]); Family/Social Support=3.97 (95% CI [2.94,5.35]); Compliance=1.78 (95% CI [1.32,3.29]); Mental Health=1.53 (95% CI [1.16,2.02]). Next Steps: Our Social Triage Tool is able to assist nurses to identify social needs in frail and complex patients, many of whom are high hospital utilizers. Future steps include refining the tool to improve the tool's sensitivity so that social needs can be identified early and interventions put in place, which may improve health outcomes and reduce hospital utilisations.