Comparison of Least Absolute Shrinkage and Selection Operator and Maximum Likelihood Estimators to Establish Determinants of Immunization in Trans-Nzoia County

Abstract

The client factors that influence under-five child guardian compliance to the immunization schedule are interlinked based on household characteristics, socioeconomic status, and maternal health practices. An incentive to motivate the mothers to prioritize their child’s health practices especially on vaccination, works perfectly towards the achievement of full immunization coverage. A randomly sampled study carried out within Weonia Location–Trans Nzoia County in March 2014 with target population of under-five children showed the vital role an incentive innovation plays towards immunization coverage. Multinomial logistic regression model was used to analyze the determinant of partial or none-immunized and the parameters estimated using the maximum likelihood estimator (MLE) and the shrinkage estimator-Least Absolute Shrinkage and Selection Operator (LASSO). The shrinkage estimator method gave a sparse model that was easy to interpret and increased the estimated predictability accuracy. Maternal health practices and access to a motivating intervention are significant factors that ensure a guardian’s compliance to their child immunization.

Full text

Vol. 4, No. 1, Summer 2015 © 2012 Published by JSES.
COMPARISON OF LEAST ABSOLUTE SHRINKAGE AND SELECTION
OPERATOR AND MAXIMUM LIKELIHOOD ESTIMATORS TO ESTABLISH
DETERMINANTS OF IMMUNIZATION IN TRANS-NZOIA COUNTY
1
Sheilla Aoko OTIENOa, Benson Munyali WAMALWAb, Nelson Owuor ONYANGOc,
Joseph Antony Makoteku OTTIENOd, Victor ONGOMAe
Abstract
The client factors that influence under-five child guardian compliance to the
immunization schedule are interlinked based on household characteristics,
socioeconomic status, and maternal health practices. An incentive to motivate
the mothers to prioritize their child’s health practices especially on
vaccination, works perfectly towards the achievement of full immunization
coverage. A randomly sampled study carried out within Weonia Location–
Trans Nzoia County in March 2014 with target population of under-five
children showed the vital role an incentive innovation plays towards
immunization coverage. Multinomial logistic regression model was used to
analyze the determinant of partial or none-immunized and the parameters
estimated using the maximum likelihood estimator (MLE) and the shrinkage
estimator-Least Absolute Shrinkage and Selection Operator (LASSO). The
shrinkage estimator method gave a sparse model that was easy to interpret
and increased the estimated predictability accuracy. Maternal health
practices and access to a motivating intervention are significant factors
that ensure a guardian’s compliance to their child immunization.
Keywords: Immunization, Logistic regression, LASSO, MLE
JEL Classification: C50, C51, C52
Authors’ Affiliation
a - University of Nairobi, Department of Mathematics, sheillac901@gmail.com
b - University of Nairobi, Department of Chemistry, benson.wamalwa@uonbi.ac.ke
c - University of Nairobi, Department of Mathematics, onyango@uonbi.ac.ke
d - University of Nairobi, Department of Mathematics, joseph.otieno@uonbi.ac.ke
e - South Eastern Kenya University, Department of Meteorology, victor.ongoma@gmail.com
* The authors acknowledge the co-operation of people of Weonia Location in Trans-Nzoia County where data
was sourced from and to the workmates for support during the study period. Special appreciation goes to the
sponsor; Grand Challenges Canada, thank you for your material support in seeing this project a success.

30
1. Introduction
Immunization is effective in the reduction of childhood mortality towards the
achievement of the Millennium Development Goal (MDG4); the reduction of under-five
mortality rates by two-thirds in 2015 (UNICEF, 2005). Immunization is enshrined as one of
the utmost medical accomplishment that has succeeded in saving more lives than any other
health care intervention in the 20th century (Wiysonge et al., 2009).
Children are more vulnerable to all kinds of hazards as compared to adults because they
are dependent on their parent/guardians/caregivers to provide for their daily needs and care
especially health care. Therefore, the relationship between vaccination coverage and care
taker’s motivation and willingness to seek childhood vaccinations still need to be explored
and studied (Holte et al., 2012). Since immunization is the most effective (and cost-effective)
means of reducing morbidity, disability and mortality among children, it has to be the
principle message to every mother and child caretaker (Ibnouf et al., 2007).
Immunization for the under-five child and infants against preventable diseases is a cost-
effective public health intervention to improve the child’s health. Recent estimates suggest
that approximately 34 million children are not completely immunized, with almost 98% of
them residing in developing countries (Kumar et al., 2010). The determinants for none/partial
under-five child immunization within the scheduled time significantly revolve around access
to funds to facilitate the whole process. Studies conducted earlier in Kenya pointed several
socio-demographic factors associated with full vaccination, among them: socioeconomic
status, religion, maternal occupation, parents’ education, maternal age and ethnicity (Maina et
al., 2013; Moisi et al., 2010; Kamau and Esamai, 2001).
The delivery of vaccines later in the schedule after the infant stage of a child’s life and
achieving 100 percent complete immunization coverage among under-five children is a great
challenge in the country with particular reference to the study area. A study conducted in the
rural areas of the Nyanza and Western Provinces in Kenya showed that approximately 79.4%
of children aged 12 to 23 months were fully vaccinated; however, timeliness of vaccination
was not assessed (Kawakatsu and Honda, 2012; Calhoun et al., 2014). An analysis of the
determinants of partial/ incomplete immunization coverage among under-five children would
be essential to establish an effective empowerment mechanism to the community to ensure all
children are immunized against preventable disease within the scheduled time.
A review of the value of an agricultural intervention to motivate guardians to comply with
the immunization schedule and the determinants of none-compliance using the Least
Absolute Shrinkage and Selection Operator (LASSO) based on a multinomial logistic
regression is thus very essential in providing the guidelines towards the achievement of 100%
immunization coverage within the right time.
The main objective of the study was to identify the determinants of partial/incomplete and
none- immunization for the under-five children and develop mechanism compatible to the
community toward the achievement of full immunization.

Sheilla Aoko OTIENO, Benson Munyali WAMALWA, Nelson Owuor ONYANGO, Joseph Antony Makoteku OTTIENO, Victor ONGOMA -
COMPARISON OF LEAST ABSOLUTE SHRINKAGE AND SELECTION OPERATOR AND MAXIMUM LIKELIHOOD ESTIMATORS TO
ESTABLISH DETERMINANTS OF IMMUNIZATION IN TRANS-NZOIA COUNTY
31
2. Data and Methodology
A survey conducted in Weonia Location in Trans-Nzoia County in March 2014 to collect
data on under-five child immunization practices and influencing factors that determine the
compliance to the vaccination schedule by guardians/mothers. The questionnaire were
administered on a one on one interview basis for the individual mothers of the under five
children by the research assistants. The target population was all children under the age of
five years within the study area and their guardians.
Data analysis was conducted using the multinomial Logistic Regression Model (MLR) to
evaluate the significance of the various none-compliance determinants in the research area.
The logit equations of the MLR form a comparison the log odds of each of the non-reference
response variables to the categorical variable of choice (Shakhawat et al., 2012), logit
(Equation 1) and the unique category probability in Equation 2;
(1)
and
The likelihood function interpreted as the joint probability of the observed outcomes
expressed as a function of the chosen regression model (Dietz et al., 2005). The model
coefficients are unknown quantities and are estimated by maximizing their probabilities and
the likelihood function given by Equation (3).
The maximization process to estimate the coefficients is accomplished by getting the log
of the likelihood function, log-likelihood (Equation 4).
(4)
The first and second derivatives of the log-likelihood function with respect to beta
equated to zero are used to obtain the Maximum Likelihood Estimates (MLE) of the model.
Least Absolute Shrinkage and Selection Operation (LASSO) perform variable selection
and coefficient shrinkage simultaneously. LASSO minimizes the log-likelihood of the MLR
model subject to constrain. The penalty term in the LASSO estimator shrinks some
coefficients while setting others exactly to zero as given by Equation 5:
(5)

32
3. Results and Discussion
Table 1 shows the tabulation of the household size based on the respondent guardian's
age, approximately, 71.1 % household size was made of 2 - 6 members. The average
reproductive age for the women in the sampled population was 25-29 years of age (111
women), had a household size of 5-6 members.
Table 1: Cross Tabulation of the Household Size Based on the Respondent Guardian's
Age.
The most defaulted vaccines were measles and PCV, while BCG had the highest rate of
compliance compared to the other vaccines. The dropout rates in under-five child vaccine
indicate that the chance of dropping out of the schedule was on increase from one prior
vaccine (dose) to the next (Table 2). The chance of dropping out on DPT1 after the BCG
vaccine were 7.6% likely to happen compared to 34.1% for dropping out on measles given
that one got BCG vaccine. The negative signs for the dropout rates indicate that the vaccine
ought to have been received prior to the particular reference vaccine apart from the PCV3 and
measles case since a few number of children access PCV3 vaccine dose given that most of
them delay within the immunization schedule for over a year.
Table 2: Under-Five Child Vaccine Dropout Rate (%).
Vaccine
BCG
DPT1
DPT2
DPT3
OPV1
OPV2
OPV3
PCV1
PCV2
PCV3
Measles
BCG
0
7.6
19.2
26.3
8.7
19.2
27
17.8
27.9
35.8
34.1
DPT1
0
0
12.5
20.3
1.2
12.5
20.
11
21.9
30.5
28.7
DPT2
0
0
0
8.9
-12.9
0
9.6
-1.7
10.7
20.5
18.5
DPT3
0
0
0
0
-24
-9.7
0.8
-11.6
2.1
12.8
10.5
OPV1
0
0
0
0
0
11.5
20
10
21
29.7
27.8
OPV2
0
0
0
0
0
0
9.6
-1.7
10.7
20.5
18.5
OPV3
0
0
0
0
0
0
0
-12.5
1.3
12.1
9.8
PCV1
0
0
0
0
0
0
0
0
12
22
19.8
PCV2
0
0
0
0
0
0
0
0
0
11
8.6
PCV3
0
0
0
0
0
0
0
0
0
0
-2.6
Figure 1 presents the attributed to non-compliance to the immunization schedule by the
respondent mothers/guardians. Most mothers/guardians of the under-five who missed some
contact vaccines on time and this was attributed to ignorance and laziness at 36 % level. Lack
of vaccine fee was cited by approximately 15.6% of the respondents. Only 1% and 3% of the
Respondent Guardian’s age (Years)
Household size/members
15-19
20-24
25-29
30-34
35-39
40-44
>45
Total
2-4
41
96
70
10
9
6
8
240
5-6
8
41
111
45
18
5
11
239
7-8
0
2
43
46
18
8
13
130
>9
1
5
7
11
9
7
25
65
Total
50
144
231
112
54
26
57
674

Sheilla Aoko OTIENO, Benson Munyali WAMALWA, Nelson Owuor ONYANGO, Joseph Antony Makoteku OTTIENO, Victor ONGOMA -
COMPARISON OF LEAST ABSOLUTE SHRINKAGE AND SELECTION OPERATOR AND MAXIMUM LIKELIHOOD ESTIMATORS TO
ESTABLISH DETERMINANTS OF IMMUNIZATION IN TRANS-NZOIA COUNTY
33
respondents attributed their failure to complete the immunization to lack of lack of vaccine at
the medical facility and long distance to the facility respectively.
The largest percentage of default for the partially immunized children was within
households of 5-6 members for a period of 1-6 months this was attribute to the mothers’
ignorance and laziness.
Households with more than nine members had a high level of partial immunization
compared to the other household sizes (Figure 2).
Figure 1: Reasons Attributed to Non-Compliance to the Immunization Schedule by the
Respondent Guardians.
Figure 2: Under-Five Child Immunization Status with Reference to the Household Size.

34
The observation in Figure 2 is in agreement with a study by Calhoun et al., (2014) and
Kamau and Esamai, (2001). Calhoun et al., (2014) carried out to establish the determinants
and coverage of vaccination in children in western Kenya from a 2003 cross-sectional survey.
A comparison of the MLE and LASSO estimates shows that, the LASSO estimate are of a
little bit lower value to the MLE values, increasing their interpretability with an exception of
the land size and incentive estimates which were higher for the lasso estimate given the MLE
estimates (Table 3).
The main determinants of a parents/guardian’s compliance to the child’s immunization
schedule based on the LASSO estimator were the household size, the family’s source of
health information, wealth index-land size, maternal health practices (ANC), access to an
incentive, the child’s place of birth and the guardian/parent’s marital status (Table 3). ANC
practices, as a determinant to the compliance to immunization was a common factor to this
study finding, similar observations were made by Mutua et al., (2011). Donsa (2013) also had
similar results concerning the significance of funds/wealth index to the level of compliance to
the immunization schedule as determined by the LASSO estimators. Yet for the maximum
likelihood, in addition to the LASSO estimates: the other determinants were the religious
affiliation of the family, the guardian’s education level, and occupation. A study by Payne et
al., (2013) also had similar findings on religion, education level, and awareness and
occupation effects to the child’s immunization status.
The LASSO estimators are in general agreement with most observations in previous
studies including Donsa (2013) and Calhoun et al., (2014). The statistic accuracy is in
consistent agreement with Ibrahim et al., (2011); the study simulated fixed and random
effects in a general class of mixed effects models using Maximum Penalized Likelihood
(MPL) estimation along with the smoothly Clipped Absolute Deviation (SCAD) and adaptive
least absolute shrinkage and selection operator (LASSO) penalty functions. It was noted that
LASSO penalty functions using estimate performed best and had significantly less estimation
error than the MLE.
Table 3: The Multinomial Logistic Regression Estimates; MLE and LASSO
Comparison (α=0.05)
Dependent variable
Estimates
MLE
LASSO
Household size
0.6667
0.162
Radial distance
-12.3375
0
Source of health information
1.7171
0.069
Religion
-0.3190
0
Land size
-0.7659
-0.11
ANC
0.7591
0.097
Incentive
0.7402
0.124
Education level
1.4785
0
Occupation
0.3072
0
Place of birth
0.9353
0.117
Marital status
0.1547
0.116

Sheilla Aoko OTIENO, Benson Munyali WAMALWA, Nelson Owuor ONYANGO, Joseph Antony Makoteku OTTIENO, Victor ONGOMA -
COMPARISON OF LEAST ABSOLUTE SHRINKAGE AND SELECTION OPERATOR AND MAXIMUM LIKELIHOOD ESTIMATORS TO
ESTABLISH DETERMINANTS OF IMMUNIZATION IN TRANS-NZOIA COUNTY
35
The size of household was a significant factor in determining the immunization and
overall health condition of the child. Households with fewer children was strongly associated
with full vaccination as compared to their counterparts with more than six members, a similar
situation occurred for the families in either polygamous marriage or in parent-
separated/divorced families. Practically, mothers with lesser children may have more
attention to each child and may not need to organize child care for other children while
travelling to health facility for immunization thus making vaccine visits easier to uptake and
complete (Kamau and Esamai, 2001; Calhoun et al., 2014).
Religion has minimal impact on the immunization status of an under-five child. The same
observation on effect of religion on immunization is made by Sanou et al., (2009) in Burkina
Faso, the study noted insignificantly lower immunization rates compared to the rest.
Socio economic status of a family influences its health seeking behavior and hence the
child’s vaccination, the wealth index of a family measured in terms of the land size in acres
directly determined the immunization status of a child. The families with two acres and above
of land complied with the immunization scheduled more than their counterparts who owned
less than half an acre piece of land. Education level of the guardian had a direct impact on
child vaccination compliance since those who at least attended school attempted to
vaccinated their child, while those who had post-secondary education complied to the
schedule and within the right time. The same observations were made in Bangladesh by
Rahman and Obaida-Nasrin (2010) and by Odusanya et al., (2008) in Nigeria. The studies
found that those children with educated mothers or of higher wealth status were more likely
to be immunized.
Similar to land ownership given the agricultural economic background of the study
population, the form of occupation a parent engaged in, had a great impact on immunization,
in terms of time and availability of funds. Those who engaged in salaried employment and
were farmers on their own land complied more than those who engaged in casual labour form
of occupation.
4. Conclusion and Recommendations
Access to a motivating factor and adherence to maternal health practices were the
significance to the guardian’s compliance to their children immunization schedule. The
incentive played a significant role in the mothers’ activities since it was able to save on their
time, logistic of access to necessary daily needs and enabled them to forgo other activities for
their child immunization. The study recommends diversity of incentives to motivate more
mothers to avail their children for immunization on time. The same view is shared in
Bangladesh by Andrews-Chavez et al., (2012); policy makers should focus future
interventions to households with observed poverty related risk factors.
The factors such as the mother’s education, household size and its source on health
information, place of birth, wealth index, maternal health practices (ANC), greatly influence
compliance to the immunization schedule. In consistency with prior studies (e.g. Andrews-
Chavez et al., 2012; Ibnouf et al., 2007; Parashar, 2005; Kawakatsu and Honda, 2012), the

36
level of mother’s education is singled out as one of the most important factor in the uptake
and completion of child vaccination.
The factors identified in this study, especially incentives and mother’s literacy should be
factored in future immunization plans to increase its efficiency.
In conclusion, the shrinkage estimator method gave a sparse model that was easy to
interpret and increased the estimated predictability accuracy. The shrinkage estimator-
LASSO was a better estimated compared to the maximum likelihood estimator in terms of
interpretation and prediction of the multinomial logistic regression model, in agreement with
other studies (e.g. Ibrahim et al., 2011; Fan and Li, 2001; Steyerberg et al., 2000). The study
thus recommends for use shrinkage estimator-LASSO in similar studies with small and
complete data sets, especially with prespecified predictors.
References
Andrews-Chavez, J., Biswas, A., Gifford, M., Eriksson, C. and Dalal, K. (2012).
Identifying households with low immunization completion in Bangladesh. Health, 4(11) pp.
1088-1097, DOI:10.4236/health.2012.411166.
Calhoun, L.M., van Eijk, A.M., Lindblade, K.A., Odhiambo, F.O., Wilson, M.L.,
Winterbauer, E., Slutsker, L. and Hamel, M.J. (2014). Determinants and Coverage of
Vaccination in Children in Western Kenya from a 2003 Cross-Sectional Survey. American
Society of Tropical Medicine and Hygiene, 90(2), pp. 234-41, doi:10.4269/ajtmh.13-0127.
Dietz, K., Gail, M., Krickeberg, K., Samet, J. and Tsiatis, A., eds. (2005). Regression
Methods in Biostatistics; Linear, Logistic, Survival, and Repeated Measures Models. Spring
Street, New York, NY 10013, USA.
Donsa, L.D. (2013). An Examination of Mothers’ Socio-Demographic Factors Associated
with Incomplete Vaccination Status among Under-five Populations in Malawi. MPH thesis.
Georgia State University.
Fan, J. and Li, R. (2001). Variable Selection via Nonconcave Penalized Likelihood and its
Oracle Properties. Journal of the American Statistical Association, 96(456), pp. 1348- 1360.
Holte, J.H., Mæstad, O. and Jani, J.V. (2012). The decision to vaccinate a child: an
economic perspective from southern Malawi, Social Science Medicine, 75(2), pp. 384-391,
doi: 10.1016/j.socscimed.2012.03.015.
Ibnouf, A.H, Van den Borne, H.W. and Maarse J.A.M. (2007). Factors influencing
immunization coverage among children under five years of age in Khartoum State, Sudan.
South Africa Journal of Family Practice, 49(8), pp. 14-19.
Ibrahim, G.J., Zhu, H., Garcia I.R. and Guo, R. (2011). Fixed and Random Effects Selection
in Mixed Effects Models. Biometrics, 67, pp. 495-503, DOI: 10.1111/j.1541-0420.2010.01463.x.
Kamau, N. and Esamai, F.O. (2001). Determinants of immunization coverage among
children in Mathare Valley, Nairobi. East African Medical Journal, 78, pp. 590-594.
Kawakatsu, Y. and Honda, S. (2012). Individual-, family- and community-level
determinants of full vaccination coverage among children aged 12-23 months in western
Kenya. Vaccine, 30(52), pp. 7588-7593, doi: 10.1016/j.vaccine.2012.10.037.

Sheilla Aoko OTIENO, Benson Munyali WAMALWA, Nelson Owuor ONYANGO, Joseph Antony Makoteku OTTIENO, Victor ONGOMA -
COMPARISON OF LEAST ABSOLUTE SHRINKAGE AND SELECTION OPERATOR AND MAXIMUM LIKELIHOOD ESTIMATORS TO
ESTABLISH DETERMINANTS OF IMMUNIZATION IN TRANS-NZOIA COUNTY
37
Kumar, D., Anju, A. and Gomber, S. (2010). Immunization Status of Children Admitted
to a Tertiary-care Hospital of North India: Reasons for Partial Immunization or Non-
immunization. Journal of Health Population and Nutrition, 28(3), pp. 300-304,
PMCID: PMC2980896.
Maina, L.C., Karanja, S. and Kombich, J. (2013). Immunization coverage and its
determinants among children aged 12-23 months in a peri-urban area of Kenya. Pan African
Medical Journal, 14(3), doi: 10.11604/pamj.2013.14.3.2181.
Moisi, J.C., Kabuka, J., Mitingi, D., Levine, O.S. and Scott, J.A.G. (2010). Spatial and
socio-demographic predictors of time-to-immunization in a rural area in Kenya: is equit
attainable? Vaccine, 28, pp. 5725–5730, doi: 10.1016/j.vaccine.2010.06.011.
Mutua, M.K., Kimani-Murage, E. and Ettarh R.R. (2011): Childhood vaccination in
informal urban settlements in Nairobi, Kenya: Who Gets Vaccinated? BMC Public Health.
11(1), doi:10.1186/1471-2458-11-6.
Odusanya, O.O., Alufohai E.F., Meurice F.P. and Ahonkhai V.I. (2008). Determinants of
vaccination coverage in rural Nigeria. BMC Public Health, 8, 381, doi:10.1186/1471-2458-8-381.
Parashar, S. (2005). Moving beyond the mother-child dyad: Women’s education, child
immunization, and the importance of context in rural India. Social Science and Medicine, 61,
pp. 989-1000, doi:10.1016/j.socscimed.2004.12.023.
Payne, S., Townend, J., Momodou, J., Yamundow, L.J. and Beate, K. (2013). Achieving
comprehensive childhood immunization: an analysis of obstacles and opportunities in The
Gambia, Health Policy and Planning, 1-11. doi: 10.1093/heapol/czt004.
Rahman, M. and Obaida-Nasrin, S. (2010). Factors affecting acceptance of complete
immunization coverage of children under five years in rural Bangladesh. Saludpública de
México, 52, pp. 134-140, doi:10.1590/S0036-36342010000200005.
Sanou, A., Simboro, S., Kouyaté, B., Dugas, M., Graham, J. and Bibeau G. (2009).
Assessment of factors associated with complete immunization coverage in children aged 12 -
23 months: A cross-sectional study in Nouna district, Burkina Faso. BMC International
Health and Human Rights, 9, S10, doi:10.1186/1472-698X-9-S1-S10
Shakhawat, H., Ahmed, S.E. and Hatem, A.H. (2012). Model selection and parameter
estimation of a multinomial logistic regression model. Journal of Statistical Computation and
Simulation, doi:10.1080/00949655.2012.746347v.
Steyerberg, W.E., Eijkemans, J.C.M, Harrell Jr, E.F. and Habbema, D.F.J. (2000).
Prognostic modeling with logistic regression analysis: a comparison of selection and
estimation methods in small data sets. Statistics in Medicine, 19, pp. 1059-1079.
UNICEF (2005). Progress for children: A report card on immunization. [Online]
(http://www.unicef.org/progressforchildren/2005n3/PFC3_English2005.pdf) (Accessed 10
June 2014).
Wiysonge, C.S., Waggie Z., Rhoda L. and Hussey G. (2009). Improving communication
for immunization in Africa: contribution of the Vaccines for Africa website. Pan African
Medical Journal, Apr 14, 2:3, PMCID: PMC2984270.

38
Appendix I
UNIVERSITY OF NAIROBI / GRAND CHALLENGES CANADA
BARCODES FOR IMPROVED CHILD VACCINATION AND FAMILY NUTRITION
QUESTIONNAIRE ON UNDER-FIVE CHILD IMMUNIZATION
………KENYA…………………………COUNTRY
UNDER-FIVE CHILD IMMUNIZATION HOUSEHOLD SURVEY/VALIDATION TOOL
IDENTIFICATION PAGE
COUNTY..................TRANS- NZOIA..............................................
DIVISION...........................................................................................
LOCATION.........................................................................................
SUB-LOCATION.................................................................................
VILLAGE...........................................................................................
NEAREST CLINIC ...............................…DISTANCE ............KM
CLUSTER.............................................................. CODE …………
Household code
Name of enumerators...........................................................................
Date of interview............Month....MARCH..........Year.......2014.......
Supervisor..............................................Signature...............................

Sheilla Aoko OTIENO, Benson Munyali WAMALWA, Nelson Owuor ONYANGO, Joseph Antony Makoteku OTTIENO, Victor ONGOMA -
COMPARISON OF LEAST ABSOLUTE SHRINKAGE AND SELECTION OPERATOR AND MAXIMUM LIKELIHOOD ESTIMATORS TO
ESTABLISH DETERMINANTS OF IMMUNIZATION IN TRANS-NZOIA COUNTY
39
CONSENT
BARCODES FOR IMPROVED CHILD VACCINATION AND FAMILY NUTRITION
Description/Purpose of the study
To eliminate persistent pocket areas of Kenya where children are not vaccinated or under
vaccinated, researchers will create a barcoded vaccination card redeemable for farm seeds
and fertilizer. Each time a child gets a vaccine, the card is taken to one of about 20,000 local
agro-vet outlets, where the barcode is scanned using an application on a camera-equipped
Smartphone. The farmer would then redeem an “agri-credit” for essential farm inputs.
This would powerfully incentivize parents to seek and adhere to their children’s
immunization schedule even when hard pressed financially to reach a distant vaccination
centre. This is a practical solution that would significantly boost small farm productivity and
incomes for poor household while safeguarding the general health of children in farming
villages through up-to-date immunizations.”
Research Site
Western Kenya
Research team
The research team is composed of: Dr. Benson Wamalwa (Principal Investigator), Dr.
Edward Muge (Project co-ordinator/logistician, Ms. Everlyne Munanga (Supervisor
Immunization staff), Ms. Caroline Aura (Supervisor Agri-business staff) and eight research
assistants.
Benefits of Participation in the Study
Participants will have their immunization cards affixed with a redeemable voucher. The value
of each voucher will be 2,000 Ksh worth of fertilizer or any agricultural seed type as per the
choice of the participant. The Researchers will obtain data to inform on reduced pockets of
non and under-immunization in the study area.
Archiving of specimens
N/A
Sharing of samples
N/A
Risks of participation
None
Confidentiality
All information obtained about you and the results of the research will be treated
confidentially. This information will be coded and kept under a password-protected database.
The study files will be kept electronically at the Department of Chemistry, University of
Nairobi, under the responsibility of Dr. Benson Wamalwa. Your participation and your
Child/children immunization results will not be shared with other medical personnel with
your identifying information. The results of this study maybe published, deposited on a
public database or communicated in other ways but it will be impossible to identify you.

40
Disclosure of potential economic gain
There are no potential economic gains that the researchers will receive from using your
child/children immunization data.
Basis of participation
You are free to consent or refuse to give consent for your participation in this study. You are
also free to withdraw your consent to participate in the study at any given point in time. Your
choice to consent or not consent to this study will in no way affect your relationship with
University of Nairobi or the other stakeholders involved in this project.
Obtaining additional Information
You are free to seek clarity or ask any questions at any point in time in the course of the
study. If you desire to get more information concerning the study, feel free to call or sms Dr.
Benson Wamalwa @ +254729903792, or Dr. Edward Muge @ +254716059466 or Ms.
Everlyne Munanga @ +254702365996 or Ms. Caroline Aura @ +254724511323.
CONSENT
BARCODES FOR IMPROVED CHILD VACCINATION AND FAMILY NUTRITION
I have read the information stated above and have had the opportunity to ask questions
regarding the study. I therefore consent to:
 Participate in this study
 My child/children immunization card be affixed with the barcode sticker
 My child/children immunization records be used in the study
 Withdraw my participation in the study after prior discussion with the research team
member.
Name………………………Signature………………….Date…………………
I, the undersigned, have fully explained the relevant details of this study to the patient.
Name………………………Signature………………….Date…………………
Witness……………………Signature:…………………Date…………………….

Sheilla Aoko OTIENO, Benson Munyali WAMALWA, Nelson Owuor ONYANGO, Joseph Antony Makoteku OTTIENO, Victor ONGOMA -
COMPARISON OF LEAST ABSOLUTE SHRINKAGE AND SELECTION OPERATOR AND MAXIMUM LIKELIHOOD ESTIMATORS TO
ESTABLISH DETERMINANTS OF IMMUNIZATION IN TRANS-NZOIA COUNTY
41
SECTION 1: UNDER-FIVE SOCIO-DEMOGRAPHIC INFORMATION
HOUSEHOLD SCHEDULE (H/H – House Hold) Respondent code ………………..
SECTION 2: IMMUNIZATION OF UNDER-FIVE CHILDREN
201. Access vaccinations received and indicate in the relevant space
<5’s name/code
Place of birth
Date of birth
Age-months
Sex
Birth order
Clinic card
Date first seen
Date of last
vaccination
Drop out period
BCG
DPT 1
DPT 2
DPT 3
OPV 1
OPV 2
OPV 3
PCV 1
PCV 2
PCV 3
Measles
Vitamin A
Date format-------DD/MM/YY
Sex: Female---- F Male------M
Vaccine reception: Not Given-------0 Given -------- 1
No BCG scar ----- 2 BCG redone ------ 3 Not applicable ---- 7

202. Find out the reason for defaulting on immunization
203. Find out the reason for the time taken after birth to visit the health facility for child
vaccination
204. Were all of your elderly (above 5 years of age)
immunized?
205. Who/what is the family’s source of information on immunization (Health care)
Source
Tick where appropriate
Health facility
CHW
Community
Relatives
Media
Others (specify)
206. What time does it take to walk to the nearest health facility for vaccination (treatment)?
Less than 30 minutes
30 – 1 hour
More than 1 hour
207. When do you take your child for vaccination?
Time of vaccination
Tick where appropriate
On the indicated T.C.A date
During outreaches
When the child is sick
Others (specify)
SECTION 3: FOOD SECURITY AND NUTRITION
301. Do you/ the family own any of the following?
Property
Tick where appropriate
Land
A phone
Television
Radio
Livestock
Others (specify)
302. What size of land do you own? (Tick where appropriate)
Land size
Owned land
Portion of land cultivation
leased land
Less than quarter an acre(< 0.25)
Quarter An acre (0.25)
Half an acre (0.5)
An acre (1)
Two acre (2)
Others (specify)
303. What farm products do you plant, /preferred, when do you plant, harvest and in what
quantities?
Farm product
Farming practiced
Planting season
Harvest season
Quantity in 80kg bags
Cash crops
Stable crops
Subsistence farming
Livestock farming
Yes
No
Do not know

Sheilla Aoko OTIENO, Benson Munyali WAMALWA, Nelson Owuor ONYANGO, Joseph Antony Makoteku OTTIENO, Victor ONGOMA -
COMPARISON OF LEAST ABSOLUTE SHRINKAGE AND SELECTION OPERATOR AND MAXIMUM LIKELIHOOD ESTIMATORS TO
ESTABLISH DETERMINANTS OF IMMUNIZATION IN TRANS-NZOIA COUNTY
43
Poultry farming
Others (specify)
304. What amount of time do you spent on the farm and household chores in a day (probe):
Time
Land
preparation
Planting
Wedding
Harvesting
Post harvesting
Household chores
1-3 hrs.
4-6 hrs.
7-9 hrs.
>10 hrs.
305. Incentive given on immunization
Incentive
Tick where appropriate
Seed
Beans
Maize
Local vegetable
Others (specify)
Fertilizer
CAN
DAP
Urea
Vaccine
Clinic card
Others (specify)
SECTION 4: HEALTH SEEKING BEHAVIOUR
401. What do you consider as danger signs for a serious illness in under-five children? (Do
not read the alternative: probe and tick where appropriate)
Serious/ dangerous signs
Difficult / fast breathing
Repeated vomiting
Breast feeding/drinking poorly or not all
Not eating/drinking well
Blood in stool
High fever/temperature
Getting more sick/very sick
Not getting better
Convulsions
Unconscious/difficult to walk
Others (specify)
402. What action do you take on noting any of the above stated conditions? Visit:
Place
Tick where appropriate
Hospital
Health facility
Bought drug Chemist
No action
Traditional herbalist
CHW
Self-medication
Faith healing

[Type text]
44
403. What was the outcome of the action taken?
Feedback
Tick where appropriate
Recovery
Still sick on treatment
Still sick not on treatment
Others (specify)
SECTION 5: MATERNAL HEALTH (Circle where appropriate)
501. What is the mother to the child parity? ……………………children
502. a) Are your entire live birth present today?
b) If No, how many died? …………………….. Children
503. Do you use family planning?
If yes, which modes of family planning do you
used?
Family planning method
Trick where appropriate
Perception on the FP
Injection
Pills
Implants
Natural family planning
Condoms
504. What is her partner’s opinion on family planning?
For family planning
Against family planning
Do not know
505. Are you pregnant now?
506. If yes, how many months pregnant? ………………….months
503. At the time you become pregnant was it planned or not?
505. Have you had a miscarriage, abortion or stillbirth?
506. If the pregnancy was miscarried, aborted or ended in a still birth, when did the last such
pregnancy end?
Date………….Month…………………..Year………….
507. How many months pregnant were you when the last such pregnancy ended?
………………….months
508. Have you ever had any other pregnancies, which did not result in a live birth?
509. Did you attend any antenatal care during your last
pregnancy?
Yes
No
Do not know
Yes
No
Do not know
Yes
No
Do not know
Yes
No
Do not know
Yes
No
Do not know
Yes
No
Do not know

Sheilla Aoko OTIENO, Benson Munyali WAMALWA, Nelson Owuor ONYANGO, Joseph Antony Makoteku OTTIENO, Victor ONGOMA -
COMPARISON OF LEAST ABSOLUTE SHRINKAGE AND SELECTION OPERATOR AND MAXIMUM LIKELIHOOD ESTIMATORS TO
ESTABLISH DETERMINANTS OF IMMUNIZATION IN TRANS-NZOIA COUNTY
45
510. If yes, who attended to you for antenatal care during your last pregnancy?
511. How many months pregnant were you when you first received antenatal care?
…………months Do not know
512. How many times did you receive antenatal care during this pregnancy?
Once
Twice
Thrice
Four time
More than four time
Do not know
513. During this pregnancy were you given/did any of the following and how many times
was this done?
1= Given , 2= Not given
Number of times
Tetanus injection
Iron syrup
Malaria drugs
Blood pressure
Weight measured
Height measured
Urine test
Blood sample test
Others (specify)
514. Who assisted you during delivery?
515. How much did the newborn child weigh at birth?
……………..grams Do not know
516. Was the birth (NAME) registered?
Yes
No
Do not know
Health professional
Doctor
Nurse/midwife
Traditional Birth Attendant
Other persons
No one
Health professional
Doctor
Nurse/midwife
Traditional Birth Attendant
Friend/ Relative
Others (specify)
No one
Yes
No
Do not know