Date

2024-11-01

Protocol

Protocol changed

Item Changed

Intervention(s) planned

Previous Version

The study will be conducted at three public hospitals in Southern Province, Sri Lanka: Teaching Hospital Karapitiya (THK), District General Hospital Matara (DGM), and District General Hospital Hambantota (DGH). The study population includes patients hospitalized in the adult medical wards in these three hospitals who meet a case definition for lower respiratory tract infection (LRTI). As per the stepped wedge cluster randomized design, all clusters start in the control group. Therefore, a set time interval of 3-6 months is given before starting the eCDST. There are 9 clusters, total of 18 wards Randomization to the intervention will occur at the cluster level at set time points. There will be one interventional arm, which will include the use of the eCDST for diagnosing and managing LRTI. Randomization will be conducted according to random number generator table. The study design is a stepped- wedge, cluster randomized study. The intervention is an electronic clinical decision support tool (eCDST) which physicians can access on their mobile phones. The eCDST uses machine learning techniques and real time surveillance data on influenza and SARS-Cov-2, patients’ clinical features, and point of care (POC) tests to provide a diagnosis and treatment recommendation regarding antibacterial use. The tool has been internally validated within a bio repository of approximately 550 patients with lower respiratory tract infections who were hospitalized in Sri Lanka. At the beginning of the study, all wards will start with usual care, when physicians will be asked to manage patients with LRTI as per their standard procedures. At set time points (3-6 months), a subset of wards will be randomized to switch over to intervention. On the intervention wards, physicians will be asked to use the eCDST when managing patients with LRTI. POC testing as advised by the eCDST will be conducted by either clinical staff or trained research staff. If performed by research staff, results will be delivered on paper within 6 hours of testing to at least 2 primary clinicians on the team (including the consultant/ attending-level physician or senior registrar). THK, DGM, and DGH do not have an electronic medical record system. Clinicians will be asked to input the results from diagnostic testing and to follow the treatment recommendations as per the eCDST, but will be advised that the final decision regarding prescription of antimicrobials is entirely at their discretion. The eCDST will consist of an electronic application that physicians will be asked to download onto their smart phones. The eCDST uses machine learning techniques to take into account real-time incidence of influenza and SARS-CoV-2 in Sri Lanka, as well as epidemiological and clinical variables that are predictive of LRTI etiology based on a bio repository of LRTI data that we have created in Sri Lanka. Clinicians will be asked to respond to questions about a patient’s presenting signs and symptoms, and based on responses, may be advised to perform POC tests for influenza, SARS-CoV-2, and potentially Streptococcus pneumoniae and PCT. Based on inputs into the eCDST, the tool will provide a probability of etiology, including of influenza, SARS-CoV-2, other viral infection, bacterial infection, non- infectious condition, or indeterminate condition. Clinicians will then be provided with a treatment recommendation (use of antibacterials or oseltamivir) based on the most likely diagnosis. POC testing as advised by the eCDST will be conducted by either clinical staff or trained research staff. If performed by research staff, results will be delivered on paper within 6 hours of testing to 2 primary clinicians on the team (including the consultant/ attending-level physician or senior registrar). THK, DGM, and DGH do not have an electronic medical record system. Clinicians will be asked to input the results from diagnostic testing and to follow the treatment recommendations as per the eCDST, but will be advised that the final decision regarding prescription of antimicrobials is entirely at their discretion. eCDST refers to the arm in which physicians are using the electronic app to manage patients with LRTI. In the usual care group physicians are not using this electronic app. Standard care will be provided for the usual/control group of patients. Case definitions: 1. Have evidence of new acute respiratory illness (less than 14 days of symptoms), as indicated by at least one of the following: a. New cough or sputum production b. Chest pain c. Dyspnea or tachypnea (respiratory rate greater than 20 breaths/minute) d. Abnormal lung examination 2. Have evidence of acute infection, as indicated by at least one of the following: a. Self-reported fever or chills b. Documented fever greater than or equal to 38 Degrees of Celsius (100.4 Degrees of Fahrenheit ) c. Documented hypothermia less than 35.5 Degrees of Celsius (95.9 Degrees of Fahrenheit) d. Leukocytosis (white blood cell count greater than 10,000/mm3) e. Leukopenia (white blood cell count less than 3000/mm3) f. New altered mental status.

Next Version

The study will be conducted at three public hospitals in Southern Province, Sri Lanka: National Hospital Galle (NHG), District General Hospital Matara (DGM), and Base Hospital Balapitiya (BHB). The study population includes patients hospitalized in the adult medical wards in these three hospitals who meet a case definition for lower respiratory tract infection (LRTI). As per the stepped wedge cluster randomized design, all clusters start in the control group. Therefore, a set time interval of 3-6 months is given before starting the eCDST. There are 9 clusters, total of 18 wards Randomization to the intervention will occur at the cluster level at set time points. There will be one interventional arm, which will include the use of the eCDST for diagnosing and managing LRTI. Randomization will be conducted according to random number generator table. The study design is a stepped- wedge, cluster randomized study. The intervention is an electronic clinical decision support tool (eCDST) which physicians can access on their mobile phones. The eCDST uses machine learning techniques and real time surveillance data on influenza and SARS-Cov-2, patients’ clinical features, and point of care (POC) tests to provide a diagnosis and treatment recommendation regarding antibacterial use. The tool has been internally validated within a bio repository of approximately 550 patients with lower respiratory tract infections who were hospitalized in Sri Lanka. At the beginning of the study, all wards will start with usual care, when physicians will be asked to manage patients with LRTI as per their standard procedures. At set time points (3-6 months), a subset of wards will be randomized to switch over to intervention. On the intervention wards, physicians will be asked to use the eCDST when managing patients with LRTI. POC testing as advised by the eCDST will be conducted by either clinical staff or trained research staff. If performed by research staff, results will be delivered on paper within 6 hours of testing to at least 2 primary clinicians on the team (including the consultant/ attending-level physician or senior registrar).NHG, DGM, and BHB do not have an electronic medical record system. Clinicians will be asked to input the results from diagnostic testing and to follow the treatment recommendations as per the eCDST, but will be advised that the final decision regarding prescription of antimicrobials is entirely at their discretion. The eCDST will consist of an electronic application that physicians will be asked to download onto their smart phones. The eCDST uses machine learning techniques to take into account real-time incidence of influenza 2 in Sri Lanka, as well as epidemiological and clinical variables that are predictive of LRTI etiology based on a bio repository of LRTI data that we have created in Sri Lanka. Clinicians will be asked to respond to questions about a patient’s presenting signs and symptoms, and based on responses, may be advised to perform POC tests for influenza, SARS-CoV-2, and potentially Streptococcus pneumoniae and PCT. Based on inputs into the eCDST, the tool will provide a probability of etiology, including of influenza other viral infection, bacterial infection, non- infectious condition, or indeterminate condition. Clinicians will then be provided with a treatment recommendation (use of antibacterials or oseltamivir) based on the most likely diagnosis. POC testing as advised by the eCDST will be conducted by either clinical staff or trained research staff. If performed by research staff, results will be delivered on paper within 6 hours of testing to 2 primary clinicians on the team (including the consultant/ attending-level physician or senior registrar).NHG, DGM, and BHB do not have an electronic medical record system. Clinicians will be asked to input the results from diagnostic testing and to follow the treatment recommendations as per the eCDST, but will be advised that the final decision regarding prescription of antimicrobials is entirely at their discretion. eCDST refers to the arm in which physicians are using the electronic app to manage patients with LRTI. In the usual care group physicians are not using this electronic app. Standard care will be provided for the usual/control group of patients. Case definitions: 1. Have evidence of new acute respiratory illness (less than 14 days of symptoms), as indicated by at least one of the following: a. new cough or sputum production b. Chest pain c. Dyspnea or tachypnea (respiratory rate greater than 20 breaths/minute) d. Abnormal lung examination 2. Have evidence of acute infection, as indicated by at least one of the following: a. Self-reported fever or chills b. Documented fever greater than or equal to 38 Degrees of Celsius (100.4 Degrees of Fahrenheit) c. Documented hypothermia less than 35.5 Degrees of Celsius (95.9 Degrees of Fahrenheit) d. Leukocytosis (white blood cell count greater than 10,000/mm3) e. Leukopenia (white blood cell count less than 3000/mm3) f. New altered mental status