The document summarizes the results of a baseline study conducted as part of the Salud Mesoamérica Initiative, which aims to improve health indicators in Central American and Mexican countries. Key findings from household and health facility surveys in multiple countries are presented. Dried blood spot samples were also collected and tested to estimate measles immunization coverage, identifying gaps between reported vaccination and presence of antibodies. Health facility characteristics associated with discrepancies included lack of internet access and inconsistent receipt of requested vaccine supplies. The study highlights opportunities to strengthen vaccination programs and better measure coverage through biomarkers.
3. Outline
• Introduction
• Design and data collection
• Selected baseline results
• Conclusion and next steps
• DBS for effective coverage
3
4. Mesoamerica health initiative
• Initiative funded by BMGF, Fundación Carlos Slim and the
Spanish Government
• Funds managed by the Inter-American Development Bank
(IDB)
• Public-private partnership
• Countries: Panama, Costa Rica, Nicaragua, Honduras, El
Salvador, Guatemala, Belize, Mexico
• Purpose: to improve health indicators through specific
interventions using results-based financing (RBF)
• Duration: five years
4
5. Outline
• Introduction
• Design and data collection
• Selected baseline results
• Conclusion and next steps
• DBS for effective coverage
5
6. Evaluation Activities
• Household survey
o Household characteristics
o Expenditure and health expenditure
o Health services utilization (women and children)
o Family planning, reproductive history
o Child health, breastfeeding and immunizations
• Physical measures
• Dried blood spots (DBS)
• Water quality (Panama)
6
7. Evaluation Activities
• Facility survey:
o Questionnaire administered to facility manager
o Physical observation
─ Equipment and inputs
─ Review of registries to detect stock-outs
o Medical record review
─ Record quota according to the characteristics of the unit
» Antenatal care
» Deliveries
» Maternal and neonatal complications
─ Selection of medical records from MoH records when possible
7
8. Information by country
Measure PAN CR NIC* HON* ESV GTM* BEL MEX*
HH Q X X X X X X
Hb X X X X X X
Ht & wt X X X X X X
DBS X X
Water
quality
X
HF Q X X X X X X X
HF obs X X X X X X X
MRR X X X X X X X
School
survey
X
LQAS X
* Countries with a control group in addition to the intervention group
8
9. Sample
Households Women Children Health
facilities
Belize * 351 311 39
Costa Rica 41 schools
El Salvador 3,625 4,730 3,328 60
Guatemala 4,414 5,829 5,211 93
Honduras 2,999 3,537 2,993 90
Mexico 5,410 6,945 6,262 90
Nicaragua 2,070 2,810 2,221 64
Panama 1,664 2,353 2,169 38
*LQAS
9
10. Outline
• Introduction
• Design and data collection
• Selected baseline results
• Conclusion and next steps
• DBS for effective coverage
10
18. MMR coverage among children
vaccinated at SM2015 facilities, by stock
18
*Children ages 12-59 months with any vaccinations whose caregiver reported visiting
a SM2015 health facility for vaccinations
Facility stock MMR vaccination for age*
Facilities with MMR vaccine in stock
at time of survey
75.4% (67.5, 83.3%)
Facilities with MMR vaccine not in
stock at time of survey
53.8% (39.7, 68.0%)
19. Outline
• Introduction
• Design and data collection
• Selected baseline results
• Conclusion and next steps
• DBS for effective coverage
19
20. Conclusions
• Wide disparities in health profile
• Wide variation in facilities’ capacities and functioning
• Management of facilities is a main driver
• Lots of missed opportunities
• Culture is crucial
20
21. Measurement Challenges
• Household survey
o Self-reporting and expectations
• Health facility survey
o Are we measuring the quality of care or quality of record-keeping?
• Culture and contraception
21
24. Linkage is Crucial (contraceptive coverage)
Common
indicator
• Modern contraceptive use among those in need
o Interruptions in contraceptive use reflect suboptimal coverage
Household Health facility
% users
%
interruptions
% with method
in stock
% with stock-out
in past 3 months
Male condom 10 15 93 14
Pill 3 4 87 35
Injectable 24 58 87 43
IUD 9 15 57 -
Implant 5 5 14 -
Emergency
contraception
0 0 29 -
25. Inconsistencies between household and
health facility surveys
Household
(n=400 women)
(%)
Health facility
(n=412 records)
(%)
Difference
(% points)
ANC in first trimester 47 8 39 *
Checks and
tests during
ANC
Blood
pressure
93 100 -7 *
Fundal
height
78 92 -14 *
Syphilis test 12 45 -33 *
Blood
glucose
35 53 18 *
* p< 0.05
25
26. Outline
• Introduction
• Design and data collection
• Selected baseline results
• Conclusion and next steps
• DBS for effective coverage
26
27. Effective coverage of measles immunization
• Immunization history from health card and caregiver recall for
6,204 children under five
• Dried blood spot (DBS) samples from 1134 children ages 12-
23 months
• ELISA test of DBS samples for presence of measles
antibodies
• Comparison of survey-based
and biomarker-based estimates
of measles immunization
coverage
27
28. Challenges associated with DBS
• Samples must be collected, transported and stored properly
• Measurement of serostatus typically done using whole blood or
serum samples
• Methods for eluting and analyzing DBS are still being developed
and standardized
• ELISA tests of DBS must be
validated, calibrated and checked
for consistency against known
standards
28
29. Validation
• Collected blood samples from 25 vaccinated and 25 unvaccinated children
• ELISA tests for each child from both serum and DBS
• Treated serum results as gold standard; measured adjustment and amount of error in
DBS results
• Calibrated DBS results for the whole sample using this transformation
y = 1.02x
R^2 = 0.89
0
2
4
6
8
DBS
antibody
result
0 2 4 6 8
Serum antibody result
29
30. Clear division of protected and
unprotected children
Serum
antibody
value
0
5
10
15
20
0 5 10 15
plate
serum_antibodies_validation_adj positive_cutoff
negative_cutoff
ELISA plate number
Individual DBS samples
Negative cutoff
Positive cutoff
30
31. Clear division of protected and
unprotected children
Serum
antibody
value
0
.5
1
1.5
2
0 5 10 15
plate
serum_antibodies_validation_adj positive_cutoff
negative_cutoff
ELISA plate number
Individual DBS samples
Negative cutoff
Positive cutoff
* Indeterminate results falling between the cutoffs were re-run
31
32. Clear division of protected and
unprotected children
0
10
20
30
Frequency
0 1 2 3
DBS antibody value
32
39. Household analysis: Multivariate results
Logistic regression on card-positive children on the likelihood that the child
lacks antibodies
Predictor OR CI
upper
CI
lower
Child age>=18 mo 0.71 0.45 1.11
Child female *0.65 0.43 0.99
Mom edu: literacy
course 0.70 0.13 3.78
Mom edu: primary 0.59 0.32 1.09
Mom edu: secondary 0.61 0.30 1.28
Mom aged 15-19 1.27 0.58 2.81
Mom aged 35-49 1.08 0.60 1.94
Log(household
expenditure) 0.82 0.60 1.12
Interview in Spanish 0.72 0.36 1.45
Usual vaccination
source is clinic 0.91 0.52 1.58
Predictor OR CI
upper
CI
lower
Log(time to usual
health facility) 0.98 0.75 1.27
Oportunidades 1.10 0.63 1.90
Jurisdiction group 2 1.01 0.28 3.62
Jurisdiction group 3 0.42 0.09 1.92
Jurisdiction group 4 0.71 0.21 2.44
Jurisdiction group 5 0.71 0.21 2.41
Urban residence 1.54 0.86 2.75
Altitude 500-999m 1.20 0.55 2.64
Altitude 1000-1499m 1.37 0.61 3.08
Altitude 1500-1999m 1.02 0.33 3.12
Altitude 2000-2400m 2.10 0.60 7.40
N = 565 * p <0.05 ** p <0.01 *** p <0.001
39
40. Mapping to health facilities
• Of 1134 children with DBS results, 192 matched to health
facilities we interviewed
• In this subset, card coverage is 71.2% and DBS coverage is
56.1%
• The caregivers of the 131 card-positive children reported
visiting 22 of the facilities we surveyed for vaccination services
• What characteristics of these health facilities predict that a
card-positive child will lack antibodies?
• Challenges:
o Limited statistical power
o Potentially biased subsample
40
41. Proportion of card-positive children without antibodies
By facility level
Facility level Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Ambulatory 68 31.2% 18.0% 44.4%
Basic 51 37.5% 21.3% 53.7%
Complete 12 38.7% 0.0% 81.7%
• In a logistic regression with dummies for each level, none are
statistically significant predictors
41
42. Proportion of card-positive children without antibodies
By referral status
Status Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Takes referrals 61 24.6% 8.4% 40.8%
Does not take referrals 70 44.2% 33.2% 55.3%
• In a logistic regression with single predictor, children attending
facilities that offer referral are 59% less likely to have “invalid”
vaccines (p=0.065)
42
43. Proportion of card-positive children without antibodies
By electricity power
Status Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Has enough electricity
to power all equipment 96 31.8% 18.2% 45.3%
Does not have enough
electrical power to run
all equipment 35 40.2% 26.3% 54.0%
• In a logistic regression with single predictor, children attending
facilities that have insufficient power for all equipment are 44% more
likely to have “invalid” vaccines, but the difference is not statistically
significant (p=0.630)
43
44. Proportion of card-positive children without antibodies
By generator ownership
Status Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Facility has generator 39 25.4% 5.7% 45.1%
Facility has no
generator 91 37.7% 25.9% 49.5%
• In a logistic regression with single predictor, the facility characteristic
was not statistically significant
44
45. Proportion of card-positive children without antibodies
By working fueled emergency generator
Status Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Facility has working
emergency generator
with fuel 32 41.8% 15.9% 67.7%
Facility does not have
working emergency
generator with fuel 99 31.8% 21.0% 42.7%
• In a logistic regression with single predictor, the facility characteristic
was not statistically significant (Odds ratio = 1.54, p = 0.427)
45
46. Proportion of card-positive children without antibodies
By internet access ***
Status Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Facility has internet 72 25.1% 11.7% 38.5%
Facility does not have
internet 59 48.2% 35.4% 61.0%
• In a logistic regression with single predictor, the facility characteristic
was statistically significant (Odds ratio = 0.36, p = 0.019)
46
47. Proportion of card-positive children without antibodies
By staff meetings
Status Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Facility has routine
meetings with staff
about medical topics 115 31.4% 20.4% 42.5%
Facility does not have
these meetings 15 49.6% 29.7% 69.6%
• In a logistic regression with single predictor, the facility characteristic
was not statistically significant (Odds ratio = 0.47, p = 0.059)
47
48. Proportion of card-positive children without antibodies
By days to fill vaccine orders
Status Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
1 day 64 26.7% 13.9% 39.4%
2 days 0
3 days 21 38.2% 8.1% 68.2%
4 days 21 48.7% 4.7% 92.7%
• In a logistic regression with single predictor, the facility characteristic
was not statistically significant (Odds ratio = 1.36, p = 0.096)
48
49. Proportion of card-positive children without antibodies
By actions taken during vaccine shortage
Procedure Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Special order 35 17.4% 0.0% 35.4%
Facility borrows from
other facility 40 47.1% 25.0% 69.2%
Nothing 34 38.1% 23.1% 53.1%
• In a logistic regression with single predictor, the facility characteristic
was not statistically significant
49
50. Proportion of card-positive children without antibodies
By receipt of number of vaccines ordered ***
Facility receives the
number of vaccines
they ordered
Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Always 49 41.2% 26.0% 56.4%
Usually 49 25.1% 7.9% 42.3%
Almost never 8 14.3% 0.0% 41.3%
• In a logistic regression with dummy predictors, the facility
characteristic was statistically significant for always receiving the
amount ordered versus almost never (Odds ratio=4.21, p=0.008)
50
51. Proportion of card-positive children without antibodies
By number of fridges
Number of fridges Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
None 0
1 fridge 57 41.4% 26.1% 56.7%
2 fridges 21 30.7% 0.0% 62.3%
3 fridges 28 19.7% 3.7% 35.7%
• In a logistic regression with dummy predictors, having 2-3 fridges
versus 1 was not statistically significant (odds ratio=0.43, p = 0.083)
51
52. Proportion of card-positive children without antibodies
By ice pack availability ***
Ice packs available
for vaccine transport
Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
None 3 67.7% - -
1 pack 15 54.1% 27.3% 80.9%
2 or more packs 99 30.8% 19.5% 42.0%
• In a logistic regression with dummy predictors, having 2+ ice packs
versus 0 was statistically significant (odds ratio=0.22, p<0.001)
52
53. Proportion of card-positive children without antibodies
By fridge temperature on day of survey ***
Any fridges out of
temp range
Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Yes 89 37.8% 26.2% 49.4%
No 12 5.5% 0.0% 16.6%
• In a logistic regression with single predictor, the facility characteristic
was statistically significant (odds ratio=10.5, p = 0.015)
• All fridges in temp range on the day of the survey (binary) was not a
significant predictor
53
54. Other factors that were not significant in
bivariate analyses (1)
• Facility has water
• Facility has electricity at all hours of the day
• Facility had interruptions in electricity in the past 7 days
• Facility is accredited by SSA – those without accreditation tended to have
more “invalid” vaccines
• Facility routinely stores vaccines– those that do not routinely store were more
likely to have “invalid” vaccines
• Vaccine ordering method (fixed time, as needed, other)
• Measles vaccine shortage in last 6 months– facilities with shortage were less
likely to delivery “invalid” vaccines
• Facility staff felt like they were going to run out of vaccines in past 3 months–
facilities not anticipating shortages were more likely to delivery “valid”
vaccines
• Number of fridges facility owns – more fridges was associated with less
invalid vaccines
54
55. Other factors that were not significant in
bivariate analyses (2)
• Number of thermal cold boxes owned
• Facility has {nurse, pediatrician, midwife, social worker, lab tech, community
health worker}; number of each type of personnel at facility
• Facility has nurse; number of nurses at facility
• Facility uses single use/reusable/auto-disable syringes for vaccines – use of
reusable syringes implies higher likelihood of invalid vaccines
• Vaccine registry book observed in child vaccination room
• Ownership, number and functioning of non-electric fridges and thermal cold
boxes
• Ownership, number and functioning of fridge thermometers
• Vaccine carriers observed (existence, number)
• Ice for vaccine carriers (existence, number)
• All fridges have temperature monitoring charts; any fridge does not have a
temperature monitoring chart
55
56. Other factors that were not significant in
bivariate analyses (3)
• Temperature recorded 2x daily all of last 30 days for every fridge; any fridge
without 2x daily monitoring in last 30 days
• All fridges at facility in correct temperature range for past 30 days; any fridge
not in correct temperature range in past 30 days
• MMR vaccine observed in stock
56
57. Other factors considered for which there
was no variation
• Pretty much all facilities have routine meetings about admin
and management of the facility
• Pretty much all facilities said the offer child services and
vaccine services
• Pretty much all facilities determine the amount of vaccine they
order, and determine this quantity in the same way for each
vaccine type
• National immunization scheme in child vaccination room
• Summary sheets for permanent records or immunization
records observed in child vaccination room
57
58. Multivariate regression version 1: Facility
characteristics
Characteristic Odds
Ratio
95% CI
lower
95% CI
upper
Basic facility 0.35 0.07 1.77
Complete facility 1.16 0.09 15.09
Facility has doctor 0.78 0.06 9.98
Electricity at all hours 2.72 0.28 26.91
Internet access 0.23 0.04 1.37
Routine staff meetings on
medical topics 1.22 0.12 12.81
Any fridges with out-of-range
temperature on day of survey **10.79 1.55 74.99
N = 101 This is a survey-weighed logistic regression
* p <0.10 ** p <0.05 *** p <0.01 **** p <0.001
58
59. Multivariate regression version 2: Facility
characteristics
Characteristic Odds Ratio 95% CI
lower
95% CI
upper
Facility takes referrals 0.81 0.02 37.42
Facility has doctor ****16100000 53821.51 4840000000
Number of auxiliary nurses 0.91 0.48 1.69
Electricity at all hours ***8627188 1163.15 64000000000
Internet access 0.53 0.01 38.16
Routine staff meetings on medical topics ****0.00 0.00 0.00
Days required to fill vaccine order 1.18 0.39 3.58
Number of ice packs observed 0.98 0.13 7.68
Any fridges with out-of-range temperature
on day of survey
0.97 0.00 13863.67
N = 98 This is a survey-weighted logistic regression
* p <0.10 ** p <0.05 *** p <0.01 **** p <0.001
59
60. Multivariate regression version 3:
household and health facility
Characteristic Odds
Ratio
95% CI
lower
95% CI
upper
Child age >=18 months 1.04 0.36 3.01
Mom edu: literacy 8.41 0.24 294.37
Mom edu: primary 0.91 0.22 3.70
Mom edu: secondary + 1.58 0.40 6.17
Log monthly household expenditure 0.74 0.24 2.35
Urban 1.47 0.48 4.52
Facility has doctor 0.49 0.05 4.55
Facility has electricity at all hours 1.04 0.15 7.16
Routine staff meetings on medical topics 0.89 0.08 9.43
Any fridges with out-of-range temperature on day of
survey
4.86 0.58 40.79
N = 96 This is a survey-weighted logistic regression
Nothing was significant
60
61. Timing of vaccination
• For 741 of the 1134 children, we have valid information on
dates of vaccines received
• Can look at seasonality and timing of card-documented
vaccines and the proportion that correspond to children with
antibodies
61
66. Where are the gaps?
66
Proportion of card-positive children lacking
antibodies, by municipality
N = 313
67. Investigation using household survey data
• Of 453 children with DBS results, 315 had health card
documentation of receipt of measles antibodies
• What are the individual, maternal, household, and community
characteristics of card-positive children who lack antibodies?
67
68. Significant bivariate characteristics
• Asset ownership (Odds ratio = 0.07, p=0.014)
• Anyone in household speaks an indigenous language (Odds
ratio = 3.80, p=0.033)
• Residence in municipality Rancho Grande (Odds ratio = 10.98,
p=0.027)
• Residence in department RAAN vs Jinotega (Odds ratio=7.65,
p<0.001)
• Altitude >1000m versus <500m (OR = 0.25, p=0.018)
68
69. Non-significant bivariate characteristics
• Child’s age
• Child’s gender
• Mother’s education
• Mother’s age
• Mother’s age at first live birth
• Marital status of mother
• Household size
• Household expenditure (log
continuous or categorical)
• Interview in Spanish
• Child received vaccines that
aren’t marked on health card
• Usual source of vaccines
(facility type)
• Travel time to usual health
facility (categorical, linear, log
linear)
• Health insurance type
• Urban/rural
69
70. Household analysis: Multivariate results
Logistic regression on 301 card-positive children on the likelihood that the child
lacks antibodies
Predictor OR CI
upper
CI
lower
Child age>=18 mo 1.40 0.77 2.52
Child female *0.46 0.22 0.95
Mom edu: primary **7.41 2.08 26.37
Mom edu: secondary 2.40 0.88 6.56
Mom edu: terciary 0.22 0.01 9.43
Mom aged 15-19 0.96 0.38 2.42
Mom aged 35-49 1.79 0.83 3.85
Household size 1.09 0.97 1.24
HH asset ownership **0.01 0.00 0.27
HH speaks
indigenous language
1.48 0.27 8.23
Predictor OR CI
upper
CI
lower
Time to HF 15-29 min *0.38 0.15 0.94
Time to HF 30-44 min 0.60 0.24 1.50
Time to HF 45-59min 1.39 0.23 8.43
Time to HF 60+ min 0.40 0.15 1.05
Madriz dept. 1.28 0.41 4.01
Matagalpa dept. 1.30 0.43 3.92
RAAN dept. *12.18 1.49 99.58
RAAS dept. 4.42 0.68 28.67
Urban **0.20 0.07 0.53
Altitude 500-999m 1.80 0.39 8.23
Altitude 1000+ m 0.66 0.13 3.47
N = 272 * p <0.05 ** p <0.01 *** p <0.001
70
71. Mapping to health facilities
• Of 453 children with DBS results, 135 matched to health
facilities we interviewed
• In this subset, card coverage is 80.5% and DBS coverage is
50.4%
• The caregivers of the 90 card-positive children reported visiting
20 of the facilities we surveyed for vaccination services
• What characteristics of these health facilities predict that a
card-positive child will lack antibodies?
• Challenges:
o Limited statistical power
71
72. Non-significant bivariate characteristics (1)
• Facility level
• Facility receives referrals
• Facility has a {doctor, nurse,
pediatrician, nutritionist, auxiliary
nurse, social worker, lab tech,
community health worker,
internist, gynecologist, surgeon},
number of each personnel
• Facility has a nurse, number of
nurses
• Enough electricity to run all
equipment
• Electricity at all hours of the day
• Interruption in electricity in past 7
days
• Working fueled emergency
generator
• Water supply
• Internet access
• Routine staff meetings about
medical topics
• Routinely stores vaccines
• Vaccine ordering method (fixed
time, as needed, other)
• Days require to fill vaccine order
• Procedure during vaccine
shortage
• Frequency of receipt of number
of vaccines ordered
72
73. Non-significant bivariate characteristics (2)
• Reported shortage of measles
vaccine in past 6 months
• Staff anticipated vaccine shortage
in past 3 months
• Number of fridges
• Number of cold boxes for vaccine
transport
• Availability of ice packs for vaccine
transport
• National immunization scheme
observed in child vaccination room
• Summary sheets for permanent
records or immunization records
observed in vaccination room
• Vaccine registry book observed
• Vaccine storage equipment
availability
• Thermometers for storage
temperature monitoring
• Vaccine carriers observed
• Ice for vaccine carriers observed
• Any fridge outside of temperature
range on day of survey (there were
none)
• Every fridge has a temperature
monitoring chart
• Every fridge has temperature
recorded on chart for all of the last
30 days
73
74. Non-significant bivariate characteristics (3)
• Any fridge not in appropriate
temperature range in last 30 days
• MMR vaccine observed in stock
• MMR vaccine stock-out in last 1,
2, or 3 months
74
75. Proportion of card-positive children without antibodies
By facility department
Facility level Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Jinotega 60 13.9% 3.0% 24.7%
Madriz 11 19.0% 0.0% 55.5%
RAAN 12 74.2% 26.6% 100%
Matagalpa 7 35.4% 0.0% 100%
• In a logistic regression with dummies for each department, RAAN is
statistically significant versus Jinotega (Odds ratio = 17.8, p=0.006)
75
76. Proportion of card-positive children without antibodies
By generator ownership
Has a generator Number of
children
%
vaccines
invalid
95% CI
Lower
95% CI
Upper
Yes 29 56.1% 17.5% 94.6%
No 61 18.1% 5.5% 30.6%
• In a logistic regression with single predictor, the facility characteristic is
statistically significant (Odds ratio = 5.78, p=0.038)
76
77. Multivariate results: Facility
characteristics
Characteristic Odds
Ratio
95% CI
lower
95% CI
upper
Basic facility 0.86 0.05 13.72
Number of fridges 1.48 0.15 14.36
Number of cold boxes 1.01 0.86 1.18
Facility has generator 0.31 0.01 12.06
Facility has doctor 3.12 0.16 59.91
Madriz department (ref: Jinotega) 9.94 0.01 13959.19
RAAN department 10.17 0.19 543.85
Matagalpa department 0.10 0.00 9.21
N = 65 Survey-weighed logistic regression
* p <0.05 ** p <0.01 *** p <0.001
77
78. Multivariate results: Household and facility
characteristics
Predictor OR CI
upper
CI
lower
Child female 0.21 0.04 1.19
Mom edu: primary 0.89 0.04 17.70
Mom edu: secondary 0.23 0.03 2.16
HH asset ownership *0.00 0.00 0.04
Time to HF 15-29 min **0.03 0.00 0.30
Time to HF 30-44 min *0.15 0.03 0.88
Time to HF 45-59min (empty)
Time to HF 60+ min 0.08 0.00 1.63
Predictor OR CI
upper
CI
lower
Madriz dept. 3.13 0.04 275.97
Matagalpa dept. 1.59 0.05 45.89
RAAN dept. ***176.46 12.79 2435.32
Urban 0.09 0.00 2.37
Facility as generator 0.42 0.00 73.41
Facility has doctor 2.63 0.11 62.66
Basic facility 0.36 0.00 26.62
Complete facility 9.12 0.33 251.78
N = 76 * p <0.05 ** p <0.01 *** p <0.001
78
79. Timing of vaccination
• For 453 of the 315 children, we have valid information on dates
of vaccines received
• Can look at seasonality and timing of card-documented
vaccines and the proportion that correspond to children with
antibodies
79
80. Why the gap?
• Errors in caregiver recall of immunization
• Errors in health card documentation, possibly motivated by
conditions for cash transfer programs
• Dried blood spot ELISA test sensitivity
• Interruptions in cold chain
• Incorrect administration of the vaccine
• Limitations vaccine efficacy, even under ideal conditions
80
81. DBS methodological strengths
• Cut-point between protected and unprotected children is clear
• The ELISA test is made to detect antibody levels on the same
order of magnitude as the DBS samples
• External validation of ELISA test showed 99-100% sensitivity
and 91-100% specificity for serum samples with low
coefficients of variation (4.5-5.4%) across repeated samples
• Commercial ELISA kit has conservative cutoffs and is
approved by European FDA for use in clinical settings
• Positive controls in all runs were always clearly positive
81
82. Thank You
For more information:
mokdaa@uw.edu
www.healthmetricsandevaluation.org
82