Were V1, Ijaa W1, Amek N1, Chiteri E1, Obor D1, Onyango E1, Odhiambo F1,
Herbst K2 and Laserson KF1
Fingerprinting Individuals in the KEMRI/CDC Health and Demographic Surveillance System
(HDSS), Western Kenya, 2010
Herbst K2 and Laserson KF1
1. KEMRI/CDC Research and Public Health Collaboration
2. Africa Centre For Health and Population Studies, UKZN, South Africa
11th INDEPTH Scientific Conference (ISC)
October 24-27, 2011
Maputo- Mozambique
KEMRI/CDC HDSS Surveillance Area
Approximate area: 700 km2
Approximate population: 223,000
Households: 92,187
Background
� The KEMRI/CDC HDSS faces challenges in re-identification
of individuals resident in the HDSS due to:
� Name similarities
� High rate of migration
� Lack of unique identification� Lack of unique identification
� Resulting in double enumerations
� Currently, personal attributes e.g. names and age are used
to identify individuals and a 70 % match rate is achieved
using a tedious and time consuming process of name/age
matching
Background
� There is need for more secure and accurate identification
systems
� Use of biometrics such as fingerprints offers an effective
solution
� Human fingerprints are unique to each person� Human fingerprints are unique to each person
Objective
� To evaluate feasibility and effectiveness of fingerprinting
as a mode of individual identification
Methodology
� Fingerprint identification system was developed using
Graiule Biometrics Software Development Kit (SDK)
� Interviewers trained extensively
� Written informed consent obtained to collect fingerprint
� Collected fingerprint from HDSS residents and non-� Collected fingerprint from HDSS residents and non-
residents
� Infants (0-11 months) were enrolled by linking their details
to their parents or guardians fingerprints
Methodology
� Data collected from July to December, 2010
� A 2% random sample of individuals successfully enrolled
was selected for re-identification
� Fingerprint database used with an HIV sub study in HDSS -
to re-identify participantsto re-identify participants
� Fingerprints used to re-identify individuals at health
facilities (ongoing in two health facilities)
� We used matching algorithm threshold of 35 on the SDK
for identification
HDSS Community Interviewer Taking Fingerprints Of HDSS Participants
Participants Approached
Participants Percentage
N 178,970 100
Consented 168,151 94
Refusals 10,819 6
Participants Enrollment
Participants Percentage
N 168,151 100
Enrolled 155,048 92
Failed 13,103 8
Reasons For Failure To Enroll
26%
1%Failed to acquire
Fingerprint distorted
Software/Device failure
73%
N =13,103
Reasons for failure to acquire
•Failure to capture high or
medium quality fingerprint
image
•Dirty fingerprint area
•Dry or wet fingerprint area
Age Group Distribution Of Enrolled Participants
Age Group (Years) Participants Percentage
N 155,048 100
0 4,264 3
1-2 9,184 6
3-4 11,668 73-4 11,668 7
5-6 10,516 7
7-16 46,071 30
17-59 59,506 38
60+ 13,839 9
0 yrs were enrolled by linking their details with the parents or guardians
fingerprints
N – Total Participants
Failure Rate With Age Groups
Age Group (Years) Attempted
enrollment
Enrolled
(%)
Fail
(%)
N 168,151 155,048 13,103
0 4,275 4,264 (100) 11 (0)
1-2 12,795 9,184 (72) 3,611 (28)
N – Total Participants
3-4 12,483 11,668 (94) 815 (6)
5-6 11,357 10,516 (93) 841(7)
7-16 49590 46,071 (93) 3,519 (7)
17-59 61,810 59,506 (96) 2,304 (4)
60+ 15,841 13,839 (87) 2,002 (13)
0 yrs were enrolled by linking their details with the parents or guardians
fingerprints
Sampled Individuals For Re-Identification(2%)
3, 101
Sampled Individuals
694 2,407
Attempted to be re-identify Did not attempt to re-identify
469 (68%)
Re-identified
225 (32%)
Failed
Age Group Distribution Of Attempted And Not Attempted To Re-Identify
Variable Attempted to re-identify
(%)
Did not attempt to re-identify (%)
N 694 (22) 2,407 (78)
Age group
1-2 23 (3) 75 (3)
3-4 57 (8) 156 (6)
5-6 53 (8) 173 (7)
7-16 176 (25) 691 (29)
17-59 302 (44) 1,083 (45)
60+ 83 (12) 229 (10)
Gender
Male 268 (39) 1053 (44)
Female 426 (61) 1354 (56)
Sampled Individuals Attempted To Be Re-Identified
Age Group (Years) Attempted to
re-identify
Re-identified
(%)
Fa iled (%)
N 694 469(68) 225(32)
1-2 23 2 (9) 21 (91)
3-4 57 7 (12) 50 (88)
5-6 53 16 (30) 37 (70)
7-16 176 134 (76) 42 (24)
17-59 302 263 (87) 39 (13)
60+ 83 47 (57) 36 (43)
Fingerprint Data Use By HIV Sub Study
� Participants 13 years and above
� From March to September 2011
Participants Percentage
N 3,316 100
N- Total Participants
Re-Identified 2,578 78
Failed 738 22
Fingerprint Data Use By Hospital Surveillance
� Data collected from July - ongoing
� Current use mostly at adult ward
Participants Percentage
N 92 100
N – Total Participants
Re-Identified 79 86
Failed 13 14
Challenges
� Finding individuals at home required several revisits
� Difficulty in capturing good fingerprints
� Children <6 years
• Fingerprints not properly developed
• Small fingers • Small fingers
� Adults ≥60 years of age
• Worn-out fingerprints
� Training of interviewers
Conclusion
� Fingerprinting system is highly acceptable, with only 6%
refusal
� Fingerprint technology does not yield a higher matching
rate than the current identification system, though it is
much fastermuch faster
� Combining existing individual identification system with
fingerprinting is expected to improve the rate and
timeliness of obtaining linkages of household data with
health facility data
Recommendation
� Identify children <6years by linking to parents or
guardians
� Use other biometric methods like eye recognition for 60+
years individuals.
� Cleaning fingerprint area properly� Cleaning fingerprint area properly
� Flexible working time in order to enroll many individuals
� Extensive training of the interviewers
Acknowledgement
� HDSS Residents
� KEMRI/CDC Staff
� Ministry of Public Health and Sanitation
� INDEPTH Network
� Dr. Kobus Herbst
Thank You!
For more information please contact:
KEMRI/CDC
P.O. Box 1578
Kisumu, Kenya
E-mail: [email protected]
The findings and conclusions in this report are those of the authors and do not necessarily represent the official
position of the Centers for Disease Control and Prevention.