Public Health Computing
Dept. Epidemiology & Public Health
Tablet computers and Open Data Kit for
electronic data capture
Nicolas Maire, Sven Meury
Andreas Steiner
20 March 2015 2
Open Data Kit (ODK) ODK extensions
ODK@SwissTPH
Outline
20 March 2015 3
Electronic data capture as a replacement for paper
questionnaires
Guidance for interviewer
Skip logic
Input constraints
Adequate data storage (1-N relationships)
Potential cost savings
Printing, copying, shipping, archiving paper Data entry, cleaning
Requirement by some funders
Near-time availability of data
Secure and central storage
20 March 2015 4
A dynamic field…
Episurveyor (Palm, J2ME), openXData (J2ME), EpiCollect (iOS, Android),…
Open Data Kit (ODK)
Open source software suite (Univ. Washington, Google.org)
Large user community, active developer community (including Swiss TPH) Open standard compliant
Android phones or tablets
Technology options
Open Data Kit: Architecture and Workflow
ODK Build/ XLSForm ODK Aggregate ServerPublish form UploadData
ODK Collect
Download Form
Questionnaire design
Excel as a master form definitionIntuitive and simple, but powerful through optional definitions Transformation to renderable XForm
20 March 2015 7
Case study: FEVERKIN (2011)
– Cross sectional urban malaria survey in the city of Kinshasa, DRC
– Possession and use of bed nets
– 7 days training / 2 days pre-testing / 30 days data collection
– 8 teams (1 interviewer and 1 lab technician each)
– >2500 HH interviewed
– GPS coordinates of all HH
20 March 2015 8
20 March 2015 9
20 March 2015 10
Feedback from project teams
Challenges
Requires infrastructure (mobile devices and server) Requires more careful planning
Dependence on expert IT support
Advantages
Fast and efficient way to collect data Better quality data
What’s the evidence?
Some:
Studies comparing costs, time/delay, reliability of various electronic
data capture methodologies indicate advantages or non-inferiority
Shortcomings:
Lack of good studies to compare data quality against traditional
paper-based forms
ODK validation study
• EMIC (Explanatory ModelInterview Catalogue) interview for influenza vaccine acceptance
• 3 teams of 2 conducting 96 interviews with both paper and tablet devices
• Field work in Pune, India • Debriefing assessment of
interviewers’ experience in
administering the interview with given device
• Tablet-paper discrepancy analysis with reference to authoritative
Recurring roles for each of the field research
assistants over the course of the study
This cycle of respective roles in each interview repeats for the two field research assistants on each team using paper (P) or tablet (T) device, and functioning as interviewer (Lead, L) or follower (coder only, F).
Results
• (Discoverable) error rates comparable for tablet and paper (2%)
• Interviewers indicate either no preference, or prefer tablet
• Tablets more cost-effective if sample size > 100
• Additional objective: to acquire experience with qualitative data, using available features of hand writing recognition
ODK and qualitative research
Workflow integration with analysis software possible (MAXQDA), with room for further improvements
Data capture a challenge
Strategies for direct digitization of notes (eliminating paper)
Strategies for more efficient processing of audio recordings (e.g. timestamps, automated transcription)
Beyond the paper-questionnaire analogy
Additional data types (GPS, barcodes, diagnostic devices, time stamps…) Support for system-wide unique ids
Near-time quality control (openhds & su2)
Monitoring and planning (su2 & ODK_planner)
ODK: beyond cross-sectional surveys
• Longitudinal follow-up requires extension of the ODK system (at present)
• In particular, system-wide, referencable ids are a key requirement • Our (and others) approach is integration of ODK with other software
Health and Demographic Surveillance Systems (HDSS)
openHDS
A set of field and computing operations to handle the longitudinal follow-up of well-defined entities or primary subjects (individuals, households, and residential units) and all related demographic and health outcomes within a clearly circumscribed geographic area
46 HDSS sites for INDEPTH over 46 years
In 24 countries Each with unique health systems www.indepth-network.org
Primary goal can be routine
surveillance in the absence of a vital registration system
Platform to provide essential
denominator data for epidemiological and demographic research
An operational challenge, in particular the data management
A functional data collection and management platform is a key requirement: OpenHDS
Open Data Kit: Architecture and Workflow
ODK Build/ XLSForm ODK Aggregate ServerPublish form UploadData
ODK Collect
Download Form
openHDS/ODK integration
ODK Build XLSForm ODK Aggregate Server Publish form OpenHDS Mobile Download Form Sync Data openHDS Download/ Update Location Information Upload DataSOLARMAL HDSS experience
Rusinga Island
First update completed 2 days before 2nd annual SOLARMAL workshop Data presented at workshop
BUT:
Su2: the surveillance surveillant
Demoodk_planner
History
Bagamoyo, Tanzania : paper-based forms
Long delays for “data capture”, difficult overview Started using ODK for all data entry
Missing parts: implemented odk_planner
1. Overview ODK/odk_planner 2. TB-Dar study 3. TB-Pharm study 4. Availability 5. Live demo Outline
odk_planner : Overview
Build forms
Aggregate
ODK Collect
MySQL Database
odk_planner
odk_planner
Study planning Data review Data Upload
SMS reminder
odk_planner : Why? Planning!
TB-Dar study
• Tuberculosis cohort
• 500 cases per year, + controls • 21 forms, complicated flow
• Doctors, field workers, secretaries, … • X-rays, CRFs, lab results, …
ODK Aggregate
odk_planner : Why? Automation!
“no show” : automatic SMS reminder Pharmacy TB Clinic Study center Patient referral Data Dataodk_planner : How?
1) Find it
2) Read
• Open source code
• Extensive documentation
3) Apply
• TPH : Experimental add-on service for ODK instances, just ask • Use it, extend it !
odk_planner : Live demo
http://figur.li/demo Username: demo Password: ODKTPH Non-admin features •Colored overview •See some data•Download/upload forms •Help
Swiss TPH: requesting ODK hosting and support
• Order by mail to [email protected]• Meaningful subject (e.g. Application for ODK server) • Order as early as possible
Required information
• Project name• Project name abbreviation (up to 12 alphanumeric characters) • Cost center
• Starting date of the project • Expected end date of project
What is included in the service
• An URL https://odk.tph.unibas.ch/ODK<yourproject> to send your data to • A trusted SSL certificate to keep the data transfer securely
• Unlimited server space
• Daily backup of your server based ODK data • Setup and maintenance of your ODK instance
Costs
• for setup
• maintenance per year
Process
• Ticket is reviewed (check for completeness) • Approval request to cost center owner
• Database setup by IT
• ODK instance setup by Public Health Computing group
• Default 2 users: Form Manager + Data collector
• Invoice
Tablets
Separate order by mail to [email protected]
• Requirements of the tablets (display size, mobile broadband module 3G/4G) • Specific model (normally only available for short time)
Advisory capacity
• Depending on needs and capacity
- Advisory support for form design and field processes - Support for data manipulation/export for data analysis
Coming soon: [email protected]
Acknowlegdements
A. Di Pasquale, R. Mitra, J. Giduthuri, M. Weiss, L. Fenner, J. Hella ODK core developers