Trauma Follow-Up Prediction (Project 2: Aim 2)

NCT ID: NCT05464017

Last Updated: 2022-07-21

Basic Information

• Recruitment Status: NOT_YET_RECRUITING
• Clinical Phase: NA
• Total Enrollment: 852 participants
• Study Classification: INTERVENTIONAL
• Study Start Date: 2024-06-30
• Study Completion Date: 2025-12-31

Brief Summary

Approximately 9% of the world's deaths, more than 5 million deaths annually, are due to injury. In high-income countries, where the epidemiology and outcomes of traumatic injury are well characterized, trauma primarily affects young, productive members of the population and is associated with significant long-term disability. In sub-Saharan Africa (SSA) countries like Cameroon, injured people face multiple obstacles to trauma care, including potentially lifesaving follow-up care after hospital discharge. The Investigators' community-based survey of 8,065 patients in South west Cameroon found that 34.6% of injured respondents did not seek immediate formal care after injury, and another 9.9% only sought formal care after alternative means, such as consultation with traditional medicine practitioners.

In Cameroon, for the 65.4% of injured people who seek formal care after injury,5 therapeutic itineraries can be complex, often involving poorly supported referrals to other facilities or transitions away from formal care. As a result, formal systems of care fail to retain trauma patients for follow-up care, a missed opportunity as these patients have already overcome significant financial and personal challenges to seek initial care for their injuries. Consequently, discharged trauma patients who may benefit from follow-up care often delay care until advanced complications develop.

The objective of this study is to evaluate a machine learning optimized phone-based screening tool that predicts which trauma patients are most likely to benefit from follow-up care. A Cluster randomized trial controlled trail will be carried out in 10 hospitals in Cameroon involving 852 trauma patients. The control group shall use the existing standard mHealth screening tool while the intervention shall use the optimized version of the mHealth screening tool (intervention) using the machine learning approach. Patients shall be followed up over a 6 months period to determine the proportion of trauma post discharge patients that need follow up care using mobile phone.

Detailed Description

The technological convergence of mHealth and machine learning provides an unprecedented opportunity to transform injury care in SSA, particularly for disadvantaged populations. The ubiquity of mobile phones and the advent of mHealth provides a novel opportunity to improve injury care in SSA. Given high levels of mobile phone penetration in Cameroon (85% to 95%) and elsewhere in SSA, the investigators designed and piloted an mHealth, phone-based 7-item screening tool for trauma patients to predict the need for in-person follow-up care after discharge. If effective, this approach could efficiently identify the subset of patients most likely to benefit from follow-up care, which is more feasible, scalable, and cost-effective than blanket advice for post-discharge care. The investigators found that phone follow-up is feasible and acceptable and a validation study revealed good correlation of the screening tool with an independent, in-person exam.

Investigators will build upon their prior research and use data science to improve, implement and evaluate the mHealth screening tool, with the ultimate objective of reducing the crippling burden of injury. This will be achieved by leveraging on machine learning, which has demonstrated promise in optimizing trauma care and trauma systems.The novel combination of mHealth and machine learning provides a powerful opportunity to transform access to health care for those least likely to receive it. Building on existing knowledge, the investigators hypothesize that a data-adaptive, machine-learning approach to outcomes prediction could radically improve survival and reduce morbidity after injury in SSA.

Investigators will apply a machine learning approach to adaptively optimize the mHealth triage tool, improving the phone call timing and algorithm that predicts the need for follow-up care via a cluster randomized controlled trial. This will be achieved using SuperLearner for prediction and cross-validated targeted maximum likelihood estimation (CV-TMLE) for variable importance, using the trauma registry, contact attempt, and screening survey data collected in Aim 1. The overall goal is to improve the mHealth tool's prediction of vulnerable patients needing follow-up care after discharge. This study shall be conducted over an 18-months period; enrollment in 6 months and follow-up participants for 12 months. Investigators will evaluate the impact of the optimized approach in a randomized study in 10 hospitals with 852 injury patients with the primary outcome of the Glasgow Outcomes Scale-Extended (GOSE)24,25 score at 3 months.

Conditions

Injury Traumatic; Injuries

Study Design

• Allocation Method: RANDOMIZED
• Intervention Model: PARALLEL
• Primary Study Purpose: PREVENTION
• Blinding Strategy: SINGLE

Study Groups

Standard mHealth screening tool

This is a tested standard phone screening tool which determines the need for in-person follow-up after a patient has been discharge. Consenting trauma patients will be contacted via mobile phone at 0.5, 1, 3, and 6 months post-discharge by a research assistant to complete the screening which will guide whether or not the patient should seek follow-up care based on the number of flagged responses to ≥1 question on the 7-item screening survey.

Group Type: NO_INTERVENTION

No interventions assigned to this group

Optimized version of the mHealth screening tool (intervention) using the machine learning approach

This arm will receive an improvement to the mHealth triage tool using a machine learning approach. Patients will be called using the optimized tool at outcome timepoints (3 months, 6months and 12months). At each call, research assistants will complete the GOSE survey and the mHealth triage tool, entering call outcomes and patient responses directly into the mHealth system. If follow-up care is indicated, the research assistant will share that information with the patient and offer to schedule an appointment.

Group Type: EXPERIMENTAL

Optimized version of the mHealth screening tool (intervention) using the machine learning approach

Intervention Type: DEVICE

An improvement to the mHealth triage tool using a machine learning approach, optimizing the efficiency of call schedule and the prediction of which patients are most likely to benefit from follow-up care given data collected at the hospital through the Cameroon Trauma Registry, as well as post-discharge phone contact attempts and survey information. The backbone of the estimators is the ensemble machine learning algorithm the Superlearner, which has been applied to medical contexts, including injury and trauma. It is a theory-driven method based on cross-validation, which combines potentially many different learners (e.g., standard regression, tree regression, random forest, neural nets) such that the model chosen (a weighted average of the learners) is asymptotically equivalent to the so called "Oracle" - the learner that fits optimally for the data-generating distribution. Note, double-robust CV-TMLE versions of this estimator are available as the tmle3mopttx function in tlverse.

Interventions

Optimized version of the mHealth screening tool (intervention) using the machine learning approach

An improvement to the mHealth triage tool using a machine learning approach, optimizing the efficiency of call schedule and the prediction of which patients are most likely to benefit from follow-up care given data collected at the hospital through the Cameroon Trauma Registry, as well as post-discharge phone contact attempts and survey information. The backbone of the estimators is the ensemble machine learning algorithm the Superlearner, which has been applied to medical contexts, including injury and trauma. It is a theory-driven method based on cross-validation, which combines potentially many different learners (e.g., standard regression, tree regression, random forest, neural nets) such that the model chosen (a weighted average of the learners) is asymptotically equivalent to the so called "Oracle" - the learner that fits optimally for the data-generating distribution. Note, double-robust CV-TMLE versions of this estimator are available as the tmle3mopttx function in tlverse.

Intervention Type: DEVICE

Eligibility Criteria

Inclusion Criteria

1. Patients with acute traumatic injury i.e. within 2 weeks of presentation for care.

2. Trauma patients who are formally admitted to the hospital as in-patients.

3. Trauma patients who die upon arriving to the Emergency Departments or while admitted in the hospital.

4. Trauma patients who are transferred to other health facilities.

5. Trauma patients with indications for hospital admission (based on physicians' assessments) but leave against medical advice

6. Trauma patients who are kept under observation in the Emergency Department for over 24 hours

Standard mHealth Triage Tool Eligibility: The mHealth triage tool will be administered to the subset of patients included in the trauma registry who are admitted then discharged home after treatment.

Optimized version of the mHealth screening tool (intervention) Eligibility: The optimized version of mHealth screening tool will be administered to the subset of patients included in the trauma registry who are admitted then discharged home after treatment.

Exclusion Criteria

According to the World Health Organization (WHO) injury definition, the following will be excluded from the definition of "injury": "Whereas the above definition of an injury includes drowning (lack of oxygen), hypothermia (lack of heat), strangulation (lack of oxygen), decompression sickness or "the bends" (excess nitrogen compounds) and poisonings (by toxic substances), it does NOT include conditions that result from continual stress, such as carpal tunnel syndrome, chronic back pain and poisoning due to infections. Mental disorders and chronic disability, although these may be eventual consequences of physical injury, are also excluded by the above definition." Although included in the WHO definition, poisonings will be excluded from the CTR as these have been extremely rare events in the CTR to date and are not typically included in trauma registries in most other contexts.

Patients who are not formally admitted and discharged within 24 hours from the Emergency Ward will be excluded.

• Eligible Sex: ALL
• Accepts Healthy Volunteers: Yes

Sponsors

Fogarty International Center of the National Institute of Health

NIH

Sponsor Role: collaborator

University of California, Los Angeles

OTHER

Sponsor Role: collaborator

University of California, Berkeley

OTHER

Sponsor Role: collaborator

University of Buea

OTHER

Sponsor Role: lead

Responsible Party

Mefire Alain Chichom

Professor of Surgery

Responsibility Role: PRINCIPAL_INVESTIGATOR

Principal Investigators

Alain Chichom-Mefire, MD

Role: PRINCIPAL_INVESTIGATOR

University of Buea

Catherine Juillard, MD, MPH

Role: PRINCIPAL_INVESTIGATOR

University of California, Los Angeles

Central Contacts

Alain Chichom-Mefire, MD

Role: CONTACT

chichom.mefire@ubuea.cm

+237677530532

Fanny JN Dissak-Delon, MD, PhD

Role: CONTACT

fannynadia@gmail.com

+237697582185

Other Identifiers

U54TW012087

Identifier Type: NIH

Identifier Source: secondary_id

View Link

GRANT13254336 - Aim 2

Identifier Type: -

Identifier Source: org_study_id