Evaluating an Algorithm-Based Implementation Strategy to Improve HIV Care Outcomes

Harnessing Data Science to Improve HIV Care Continuum Outcomes: A Hybrid Type 2 Trial Evaluating a Machine-Learning Algorithm-Based Implementation Strategy

This study tests a strategy for helping Care Management Agencies prioritize patients with HIV (PWH) for outreach and support. Under the new strategy, care managers are given a list of highest-priority patients who have been identified by a computer algorithm as being at high risk of going to the emergency room in the next two weeks. This strategy is compared to traditional (standard of care) care management, in which care managers reach out to patients based on a set schedule and their clinical judgement (but not based on a computerized report). We are looking at whether the use of the computer report helps care managers reach the right patients at the right time, preventing them from having to go to the emergency room.

Study Overview

• Status: Enrolling by invitation
• Conditions: HIV (Human Immunodeficiency Virus)
• Intervention / Treatment: Other: predictive emergency room alerts (pERA); Other: Standard of care

Detailed Description

Comprehensive Care Management and Care Coordination (CCM/CC) is a medical case management intervention with demonstrated effectiveness in reducing ED visits and hospitalization for PWH, and improving both health outcomes (viral load, CD4 count) and retention in care. However, despite CCM/CC's effectiveness, there are persistent challenges to its implementation. This project is based on the scientific premise that the effectiveness of the CCM/CC intervention can be greatly improved by utilizing a data-driven implementation strategy that optimizes timely provision of CCM/CC services to the patients who need it most. Our community-based collaborator, Comprehensive Care Management Partners (CCMP) Health Home, has developed and validated a machine-learning algorithm that can reliably predict which of its PWH patients are most likely to visit the ED in the next two weeks. In this project, we will apply this algorithm as a targeted implementation strategy for CCM/CC, focusing service provision on the PWH who need it most, when they need it most. Our core hypothesis (supported by preliminary studies data) is that this "just-in-time" strategy for implementing a care management intervention will overcome both provider-level barriers to the provision of CCM/CC services and patient-level barriers to the receipt of HIV treatment and care. We will conduct a Hybrid 2 implementation-effectiveness trial, guided by the RE-AIM implementation science framework and the behavioral economics theory of Scarcity to collect rigorous data on the impact of this algorithm-driven implementation strategy on the reach, effectiveness, adoption, implementation and maintenance of the CCM/CC intervention

• Study Type: Interventional
• Enrollment (Estimated): 2600
• Phase: Not Applicable

Contacts and Locations

Study Locations

• United States
  • New York
    • New York, New York, United States, 10016
      • Community Care Management Partners Health Home

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

Study Population

All care management agencies that comprise the Community Care Management Partners (CCMP) Health Home and serve patient with HIV will be included. Randomization occurs at the level of the Care Management Agency. Data will be extracted from the CCMP electronic medical record on all PWH patients (approximately 2600) in the Health Home.

Description

Inclusion Criteria:

• Participants must be members of one of the Care Management Agencies that comprise the Community Care Management Partners (CCMP) Health Home
• Participants must be living with HIV

Exclusion Criteria:

• None, other than those listed above.

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Treatment
• Allocation: Randomized
• Interventional Model: Crossover Assignment
• Masking: None (Open Label)

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Active Comparator: Predictive Emergency Room Alerts (perA)Implementation Strategy; Refers to patients within Care Management Agencies that have been randomized to use the pERA implementation strategy to delivery CCM/CC during that study period.	Other: predictive emergency room alerts (pERA); pERA is a machine-learning algorithm-driven implementation strategy that identifies patients at higher risk of emergency room visits and alerts the care manager to follow-up with them.
Other: Standard of Care Implementation Strategy; Refers to patients in Care Management Agencies that have been randomized to use their standard of care implementation strategy to deliver CCM/CC during that study period.	Other: Standard of care; Care managers interact with patients according to their standard of care protocols

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
ER visits	Number of ER visits made by patients	Each 18 month cluster period (36 months total)
Hospitalizations	Number of days of Hospitalization	Each 18 month Cluster Period (36 months total)
Viral Suppression	Number of timepoints at which patient was virally suppressed	Each 18 month cluster period (36 months total)
CD4 Count	CD4 Level at each data collection timepoint	Each 18 month Cluster Period (36 months total)

Collaborators and Investigators

• Sponsor: Hunter College of City University of New York
• Investigators: Principal Investigator: Sarit A. Golub, PhD, MPH, Hunter College of The City University of New York

Study record dates

Study Major Dates

• Study Start (Actual): November 18, 2025
• Primary Completion (Estimated): February 1, 2029
• Study Completion (Estimated): August 1, 2029

Study Registration Dates

• First Submitted: December 9, 2025
• First Submitted That Met QC Criteria: December 9, 2025
• First Posted (Actual): December 12, 2025

Study Record Updates

• Last Update Posted (Actual): January 27, 2026
• Last Update Submitted That Met QC Criteria: January 23, 2026
• Last Verified: January 1, 2026

More Information

Terms related to this study

• Keywords: HIV; implementation science; emergency room; machine-learning algorithm
• Additional Relevant MeSH Terms: Blood-Borne Infections; Urogenital Diseases; Genital Diseases; Pathologic Processes; Disease Attributes; Immune System Diseases; Infections; RNA Virus Infections; Virus Diseases; Communicable Diseases; Sexually Transmitted Diseases, Viral; Sexually Transmitted Diseases; Lentivirus Infections; Retroviridae Infections; Immunologic Deficiency Syndromes; Slow Virus Diseases; HIV Infections; Pathological Conditions, Signs and Symptoms; Emergencies; Acquired Immunodeficiency Syndrome; Health Services Administration; Health Care Quality, Access, and Evaluation; Quality of Health Care; Quality Indicators, Health Care; Standard of Care
• Other Study ID Numbers: R01MH136903 (U.S. NIH Grant/Contract)

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO
• IPD Plan Description: State health department requirements regarding these electronic medical record data prohibit sharing IPD.

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No