Rapid Mobile Phone-based survey (RAMP)in
KenyaPresented by
Kioko Kiilu (KRCS)
Jenny Cervinskas (IFRC)Nairobi, February 1st, 2011
Introduction to Rapid Mobile Phone-based (RAMP) survey
RAMP experience with KRCS volunteers◦ Site and project identification◦ Survey methods, training, fieldwork◦ Lessons learnt
Preliminary results
Plenary
Outline
To provide a survey methodology and operations protocol so that governments and NGOs can:◦conduct health surveys at reduced cost ◦with limited external technical assistance◦and achieve high standards of data quality
Dramatically decrease the time from data collection to having data available for decision making
RAMP: Purpose
Technical Reference Manual Standardized questionnaires for malaria Questionnaires designed on the internet using
EpiSurveyor Data collected using cell phones Training manual and tools adaptable to local settings Standard survey methods Rapid analysis and reporting of results
RAMP survey tool: Features
The traditional data cycle
Mobile technology can drastically reduce the time between data collection and action.
Weeks to Months (sometimes continuous)
Weeks to Months to Years
Weeks to Months to Years
Weeks to Months to Years (or never)
Weeks to Months (sometimes continuous)
Questionnaire Design
Data Entry
Data Analysis
Data Reporting
ACTION
Data Collection
The SMT data cycle EpiSurveyor has:
◦ eliminated the need for data entry and is now automating many analysis and reporting functions
◦ shortened the time and reduced the costs between collection and action
Anyone can create a username and password at www.episurveyor.org and start using these tools for free
Questionnaire Design
Data Entry
Data Analysis
Data Reporting
ACTION
Data Collection
Questionnaire design in Episurveyor (internet) Real-time data entry on cell phones Daily upload of data from cell phone over 2G cell
network to internet database Real-time data cleaning Real-time data analysis Rapid production of preliminary survey results bulletin
within 24 hours of last interview Rapid production of preliminary feedback survey
report in 72 hours
Cell phone-enabled innovations
Ongoing operational research project in malaria
Hard-to-reach areas/Long data cycle
Mobile network coverage
Project: Home Management of Malaria (HMM) in Malindi district, Coast province
Site and project identification
1st stage: standard probability-proportional-to-estimated-size (PPES) selection of PSUs◦ Sampling frame: 106 villages of the HMM project
2nd stage: segmentation of PSU; choose 1 segment using PPES
SRS to choose 10 households Precision:
+/- 6% for each key indicator from household questions +/- 3% using roster/individual data
30 PSUs, 10 households per PSU, 1500 persons, all ages
Survey methods
Household questionnaire◦ Usual household characteristics (wealth asset questions, distance
to health facility, etc.)◦ Summary questions (innovation)
Duplicated nearly all key indicators that are in the person & net register
Eg., no. of persons: all ages & children <5 yo No. of any nets, ITNs No. of persons/children <5 yo slept under ITN last night
Person roster Net roster
◦ Number of persons that slept under each net
Three survey instruments
HMM volunteers (Interviewers)
HMM Coaches /MOH Public Health Officers (PHOs) (Supervisors)
Training – 4 days (January 19-22, 2011)
Recruitment and training
Content ◦ Cellphone basics◦ Questionnaires◦ Informed consent◦ Interview techniques◦ Field procedures ◦ Field logistics/reporting◦ Supervisor training
Methodology◦ Presentations, role play, group discussion, demonstrations,
field tests (2)
Training content & methodology
Survey teams: ◦ 6 teams
1 Team supervisor and 2-4 interviewers/team)
Survey supervisory team (KRC, IFRC, WHO, MOH, DataDyne):◦ Planning, logistic & financial
responsibilities, field support, daily “quality” rounds, and remote monitoring of data quality
Field work (January 24th-28th, 2011)
Morning briefing (“quality round”)
Meeting with community leaders, reviewing sketch
maps, segmentation, selection of HHs
Conduct interviews at HH level
Supervisor will send data to server
Debriefing at day’s end with support team in Malindi
Data cleaning and analysis
A day’s schedule
Data entry: worked well, all teams were able to collect data using the cellphone and send to server
Survey conducted with reasonable adherence to correct field procedures
KRC volunteers were able to prepare the sketch maps, carry out segmentation, and apply SRS to select HHs
Preliminary results were available within 24 hrs. of the return of the last team from the field
Lessons learnt
Results: key indicators, HH questionnaire
HH ownership at least one
ITN
Access, % pop. with access to
ITN
Use, all persons
Use, children <5 years
Use in children <5y, given at least 1 ITN
Of <5y fever cases, treated
ACT
Of <5y fever cases, treated ACT within 24
hr
Of <5y fever cases, blood
taken (testing)
0
10
20
30
40
50
60
70
80
90
100
78
68
55
65
7977
69
14
Percentage
Access: Two-thirds of ITNs to reach universal coverage are present. Gap is 32%.
Key indicators
Target population 68 753
Persons per net 2.47
ITNs needed 27 835
Survey-estimated ITNs in HH of target pop
18 931 (68%)
ITN/LLIN need/gap 8 904 (32%)
Results: High percentage of ITNs are being used. Use gap is due to insufficient ownership of ITNs
Key indicators Point estimate
% ITNs that were slept under last night 87%
% ITNs that were hung last night 86%
ITN use, all ages 55%
ITN use, <5 yo 65%
* 47% of nets had 3 or 4 persons sleeping under them
ITN use by age group
<1y 1 2-4 5-9 10-14 15-24 25-44 45-59 60+0
10
20
30
40
50
60
70
80
90
100
Age groups (years)
Age in months Cumulative %
<12 months 28
12-23 months 59
24-36 months 80
Age of ITNs
* 88% of nets were LLINs
Number of persons sleeping under a single net last night
%, nets
1 person 15
2 persons 39
3 persons 32
4 persons 15
Key results from roster-only data
Treatment & diagnosis, <5 yo
Key indicators %
Treated ACT, <5 yo 77
Treated ACT within 24 hours, <5 yo 69
Received finger/heal stick for blood 14
- Denominator for all indicators was % of children <5y with fever in the previous two weeks
Using the cellphones
No major problems: all cellphones were operational
No calls to the Datadyne “hotline”
Data entry: worked well
Data was sent to the server by all teams, every day
Daily/immediate upload of data if 2G/GPRS available
Potential difficulties: initial connection of cell phone to
data network
So, does the RAMP “work”?
Conducted by secondary-school graduates with no previous survey experience
Survey was completed within two weeks◦ 1 week training, 4.5 days field work
Daily data cleaning accomplished Preliminary survey results bulletin finished within
24 hours Preliminary report finished within 72 hours Provided excellent management information on
the key indicators
Cost component USD
Local operational costs (e.g., personnel costs: avg. $40 per interviewer & supervisor/day * 20 persons * 10 days, training hall, stationary)
13 429
Phones + accessories 5 416
Transport (drivers, fuel) 3 950
Total 22 795
Approx Kshs 1.7m
Analysis + Report Free (WHO)
Costs
• Kenya Red Cross Volunteers• Kenya Ministry of Public Health and Sanitation• IFRC• Datadyne• WHO• Kenya Bureau of Statistics
A special thanks to the survey team and the many families who agreed to be interviewed for this survey
Acknowledgements