Blog     Github

Objectives

Kenya ranks 52nd overall in global child mortality, and Kisumu has suffered from particularly high rates of child mortality and maternal deaths. To better understand the reasons for this, Kisumu was selected as a CHAMPS site that was officially launched on May 10, 2017, alongside the ongoing Kenya Mortality Study (KMS).

Running CHAMPS and KMS has required the collaboration of many partners, including the Kenya Mortality Research Institute (KEMRI), Jaramogi Oginga Odinga Teaching and Referral Hospital (JOORTH), the Centers for Disease Control and Prevention (CDC), Emory University, Task Force for Global Health, QED (https://qed.ai), and the community health volunteers in Manyatta and Karemo.

Both CHAMPS and KMS aim to determine underlying causes of death. When deaths occur in the catchment areas, they are reported by community health volunteers to the mortality surveillance team on the phone. Members of this team then visit the parents in person to confirm eligibility and request consent for these studies. If consent is granted, the bodies are transported to labs where they undergo a wide battery of analyses, including minimally invasive tissue sampling, PCR, HIV/TB/Malaria, and more. The clinical histories of mother and child are also catalogued. In the end, all data is reviewed by medical panels that carefully deliberate on the cause of death, and the conclusions are sent to the parents, to help bring closure to these events. Data from these studies is also used to compare the effectiveness of different methodologies, such as verbal autopsy vs. full autopsy, and anonymized data will help national organizations determine future health policy.

Data Processing Pipeline

CHAMPS data is collected and managed by the data processing pipeline illustrated below, developed by QED in collaboration with data liaisons from the CDC. (Note: This map is interactive, including pop-out explanations and hyperlinks.) Rather than building or using a single application to implement the studies, we built several modular applications with clearly segregated responsibilities, connected by clean interfaces.

CHAMPS Data Processing Diagram
  • Data collection is handled using KoBoToolbox.
  • Case management is handled by Progress.
  • Dashboard analytics are auto-updated at the CHAMPS dashboard.
  • Data exports are hosted for restricted audiences at CHAMPS data.
  • Server-side data integrity checks are periodically executed, and clinicians are e-mailed digests of any errors found.
  • All medical forms are stored in a versioned repo (mirrored from Git).
  • RosettaForm converts the Program Office’s forms (REDCap dictionaries) into XLSForms that are ODK-compatible. This enables the Kenya site to use its own technologies while still satisfying the PO’s constraints.

Below are screenshots and more details for selected applications.

Data Collection

Data collection is handled by KoBoToolbox, an open-source ODK-compliant system for deploying complex forms on both web and mobile platforms.

Progress

Progress allows the Kenya site to track the progress of each patient in CHAMPS and KMS through an interactive dashboard of checkmarks. Gray boxes are tests that are not required, and white boxes are required. All skip logic in the CHAMPS business process model can be encoded through a GUI. Progress effectively augments data collection systems with case management capability in a non-invasive manner, requiring no modifications to KoBoToolbox.

Dashboards

Monitoring and evaluation analytics for the Kenya site are computed automatically each day at the CHAMPS dashboard, with no human intervention required. This includes site-specific indicators, BMGF results-framework indicators, and visualizations conceived by local staff.

Data Portal

Each day, data is automatically reshaped and packaged to meet the formatting specifications of various consumers, such as CHAMPS PO, KMS, DeCoDE panelists, and local data administrators, and then served through our portal.

null
null

Acknowledgments

We thank all our partners for allowing us the opportunity to work on this meaningful project and welcome continued collaborations in the future.

Kenya Medical Research Institute
Centers for Disease Control and Prevention
null