Measles Elimination in the Americas:A Comparison of Countries With a One-
dose and Two-dose Routine Measles Vaccination Schedule
Adrianne Sever, MDEpidemic Intelligence Service Officer
Global Immunization Division
January 29, 2009
Presentation Overview
• Background• Study Objectives• Methods Overview• Results• Limitations• Discussion• Conclusions
PAHO Strategy• All Countries:
• “catch-up” campaigns—ages 1-14 yrs.• “keep-up” routine vaccination coverage
• High (≥95%) first-dose coverage at 12 months of age
• “follow-up” campaigns—ages 1-4 yrs.
• Some countries:• 2nd routine dose (MCV2) at 4-6 yrs.
Routine Measles Vaccination Schedule by Country 1992-2002
*El Salvador introduced MCV2 in 1999, but after the last measles case
Excluded from Analysis
1-Dose
2-Doses
Study ObjectivesUsing available data from the PAHO region:1) To identify if socioeconomic, demographic
and programmatic differences existed between countries with a one-dose and two-dose routine vaccination schedule (background)
Then:2) To assess if a 2nd routine dose had an effect
on time to elimination
Analytic Overview
Descriptive Analysis
Propensity Score-Matching
Survival Analysis
To evaluate characteristics of one and two-dose countries in the region
A method to reduce bias by identifying a subset of countries with similar
characteristics
To assess if time to elimination differed between countries with a one-dose and
two-dose schedule
Method Purpose
Study ObjectivesUsing available data from the PAHO region:1) To identify if socioeconomic, demographic
and programmatic differences existed between countries with a one-dose and two-dose routine vaccination schedule (background)
Then:2) To assess if a 2nd routine dose had an effect
on time to elimination
Methods: Objective 1General characteristics of countries
• Definitions:• “Two-dose country”: had a 2-dose routine
vaccination schedule prior to interruption of measles virus transmission
• “One-dose country”: had a one-dose routine vaccination schedule or a second routine dose was added after endemic virus transmission was interrupted
• Means and medians of one-dose and two-dose countries were compared using 95% CIs and the Kruskal-Wallis test (alpha level of 0.05)
VariablesVaccination Program
1999-2001**
• MCV1 Coverage• SIA Coverage• DTP 1 Coverage• DTP3 Coverage• Measles Dropout• DTP Dropout• Surveillance Quality
Demographic & Socioeconomic 2000*
• Population Density• Female Literacy• Gross National Income (GNI) per Capita• Proportion Urban Population• Percent GNP Spent on Healthcare• Infant Mortality• School Attendance• Aggregate Doctor and Nurse Ratio
*Data from the year 2000 or for the last year it was available**Coverage averaged over 1999-2001, or average of 3 campaigns
Comparison of Demographic and Socioeconomic Variables
2-Dose Countries (N=7)
1-Dose Countries (N=14)
Variable Mean (95% CI) Mean (95% CI)Infant mortality (per 1000 live births)*
17 (12-22) 33 (25-42)
Proportion urban population (%)* 76 (63-89) 60 (51-68)
Female literacy (%)* 94 (91-97) 80 (72-88)
GNI per capita (USD)* 473 (309-637) 335 (191-479)
Percent of GDP spent on health Care*
7 (6-9) 6 (6-7)
Net primary school enrollment (%)**
90 (79-100) 86 (75-98)
Doctor and nurse ratio (per 10,000 population)*
97 (86-108) 92 (81-104)
Population density(people per km2)*
37 (17-58) 86 (29-144)
•Source: PAHO Basic Indicators http://www.paho.org/English/SHA/coredata/tabulator/newTabulator.htm**Source: UNESCO database http://stats.uis.unesco.org/unesco/TableViewer/document.aspx?ReportId=143&IF_Language=eng
Comparison of Immunization Program Variables
2-Dose Countries(N=7)
1-Dose Countries(N=14)
Variable Mean (95% CI) Mean (95% CI)
MCV1 coverage 91 (85-97) 92 (88-96)
Measles SIA coverage 88 (80-95) 91 (89-94)
DTP1 coverage 95 (92-97) 93 (90-97)
DTP3 coverage 87 (80-94) 83 (74-92)
Measles dropout rate 4 (-3-11) 6 (1-11)
DTP dropout rate 7 (0-13) 12 (6-19)
Surveillance quality* 80 (62-99) 84 (75-94)
Source: Country reports to PAHO-WHO/UNICEF *Percent of sites reporting weekly to PAHO, 1999-2001
Summary of Descriptive Analysis
• High MCV1 and SIA coverage overall
• Two-dose countries:• Higher proportion urban population • Higher female literacy • Higher GNI per capita • Lower infant mortality
Study ObjectivesUsing available data from the PAHO region:1) To identify if socioeconomic, demographic
and programmatic differences existed between countries with a one-dose and two-dose routine vaccination schedule (background)
Then:2) To assess if a 2nd routine dose had an effect
on time to elimination
Methods Used for Assessing if a Second Routine Dose had an Effect on Time to
Elimination
• Propensity Score Matching
• Survival Analysis
• Measles Susceptibility Estimates
Propensity Score Matching
Propensity Score Matching Methods• Why: Good approach for reducing selection bias
when you have a small sample size
• What: Method for equating covariates (potential confounders) of two comparison groups
• How: Model the conditional probability of exposure (one-dose vaccination schedule) given covariates
• Outcome: Model produces a score that represents a summary of covariates
• Then what: Conduct final analyses using countries with similar propensity scores
2 Propensity Score Models“partial model”—6 variables• MCV1 coverage • measles SIA coverage• measles SIA schedule• population density• percent urban population• surveillance quality
“ full model”—13 variables• MCV1 coverage • measles SIA coverage• measles SIA schedule• population density• percent urban population• surveillance quality• DTP 1 coverage• DTP3 coverage• female literacy• GNI per capita • % GNP spent on healthcare• infant mortality• school attendance
Propensity Score Matching: “Partial Model”
Country Propensity ScorePeru 0.66Brazil 0.66Cuba 0.66
Venezuela 0.48
Belize 0.82Bolivia 0.73DR 0.78Ecuador 0.73El Salvador 0.81Paraguay 0.83
Guatemala 0.87Haiti 0.89Honduras 0.89Nicaragua 0.89
Country Propensity ScoreChile 0.61
Panama 0.48Uruguay 0.50Argentina 0.43
Costa Rica 0.78
Colombia 0.36Mexico 0.35
1-dose countries 2-dose countries
Propensity Scores Summary
• “partial model” : 15 matched countries to include in final analysis
• “full model” : no matched countries• Propensity scores:
• One-dose countries: 0.61-0.92• Two-dose countries: 0.14-0.46
• Lack of matching was likely due to differences in socioeconomic and demographic variables
Survival Analysis
• Kaplan-Meier Curves
• Crude analysis (21 countries)
• Propensity score-matched analysis (15 countries)
• Time to elimination
• 1992 Start date
• 3 definitions
• Case-based
• Population-based
• Combination
Survival Analysis: Time to Elimination
Elimination Year Definitions
• Case-based--the year following the last endemic (non-imported) measles case up through 2002
• Population-based--the year that measles incidence of < 1 lab-confirmed cases/1 million was achieved and sustained
• Combination--the later of the two years of the definitions above
Crude Survival Analysis: Time to Measles Elimination
χ2 =0.07, p=0.79
Perc
ent o
f Cou
ntrie
s no
t A
chie
ving
Elim
inat
ion
Propensity Score-Matched Survival Analysis: Time to Elimination
χ2 =0.44, p=0.51
Perc
ent o
f Cou
ntrie
s no
t A
chie
ving
Elim
inat
ion
Measles Susceptibility
Measles Susceptibility Methods
• Estimated measles susceptibility for each country for birth cohorts under 15 years of age using:
• MCV1 coverage• SIA coverage x 3 campaigns• MCV2 coverage: assumed 0%, 50%, and 75%
coverage• Vaccine Efficacy: 85% in first year, 95% in
second year or older
0102030405060708090
100
Mean Measles Susceptibility by One-dose and Two-dose Countries
1-Dose Countries
N=14
2-Dose CountriesN=7
Pro
porti
on o
f per
son
unde
r 15
susc
eptib
le to
mea
sles
(%)
0% 50% 75%
Proposed Immunity Threshold
(<7% Susceptible)
MCV2 Coverage Assumptions
Question 2 Summary• “Partial model” matched 15 countries with
similar characteristics • Crude and propensity score-matched
survival analyses showed no difference in time to elimination
• Measles susceptibility after MCV1 and campaign doses was low for both groups and near the proposed herd immunity threshold
Limitations
• Ecologic analysis• Small number of observations (countries)• Could not reduce all possible bias• MCV2 coverage data not included in analysis • MCV2 age of introduction differs from other
regions (may not be applicable)
Discussion• The second routine measles dose had no
apparent impact on time to elimination• High first dose and campaign coverage• At threshold immunity level without MCV2• MCV2 given at 4-6 years of age, after most
children had received a second opportunity
• Campaigns occurred simultaneously throughout region
Public Health Implications
• In PAHO, where countries have high routine and campaign coverage, a second routine dose did not effect time to measles elimination when introduced at school entry.
• Successful vaccination program implementation was more important than the underlying socioeconomics of a country in eliminating measles.
Future Research
• Apply similar methods using data from other regions
• Assess MCV2 impact on sustaining elimination
Acknowledgements
• CDCDr. Vance DietzDr. Jeanette RaineyDr. Elizabeth ZellDr. Karen HennesseyDr. Amra UzicaninDr. Steve CochiDr. Hardeep Sandhu
• PAHODr. Carlos Castillo-SolorzanoDr. Jon AndrusDr. Carolina DonavaroMs. Christina MarsigliMs. Pamela Bravo
Thank You!
Extra slides
Methods: Country Classification
Country Vaccination Policy MCV2 Date Case-based Elimination YrPanama Two-dose 1992 1996Costa Rica Two-dose 1992 2000Chile Two-dose 1992 2000Uruguay Two-dose 1992 2000Colombia Two-dose 1997 2002Argentina Two-dose 1998 2001Mexico Two-dose 1998 2001El Salvador One-dose 1999 1997Brazil One-dose 2004 2001Paraguay One-dose 2004 1999Cuba One-dose 2004 1996Belize One-dose 2005 1996Guatemala One-dose NA 1999Honduras One-dose NA 1998Nicaragua One-dose NA 1996Venezuela One-dose NA 2002Peru One-dose NA 2001Ecuador One-dose NA 1997Bolivia One-dose NA 2001Dominican Republic One-dose NA 2002Haiti One-dose NA 2002
Summary of Immunization Program Indicators
Indicator Mean (95% CI) Median (Range)
MCV1 Coverage 92 (89-95) 94 (73-99)
Measles SIA Coverage 90 (87-93) 93 (77-97)
DTP1 Coverage 94 (91-96) 96 (77-98)
DTP3 Coverage 84 (79-90) 88 (45-99)
Measles Dropout Rate 5.6 (1.7-9.4) 3.2 (-6.7-28.6)
DTP3 Dropout Rate 7.3 (5.8-14.9) 7.3 (0-41.6)
Surveillance 82 (74-90) 90 (44-100)
Summary of Socioeconomic and Demographic Indicators
Indicator Mean (95% CI) Median (Range)Population density (people per km2)* 70 (32-109) 41 (7.7-303)
Percent urban population* 65 (58-73) 60 (50-92)Literacy rate* 86 (81-92) 92 (50-98)GNI per capita* 383 (278-488) 364 (100-860)Percent of GDP spent on Health Care** 6.6 (6.0-7.2) 6.5 (4.4-10.1)Net Primary School Enrollment 87 (80-95) 90 (21-99)
Doctor and Nurse Ratio (per 10,000 population) 94 (86-102) 96 (27-107)
Infant Mortality Rate 28 (21-35) 28 (7.5-62)
*Data for the year 2000** Last available data (not collected yearly
Summary of comparison of indicators* versus last confirmed case
*Each indicator was broken down into specific categories
Indicator Categorization of Indicators
Mean MCV1 CoverageMean of 1999-2001 ( 1>=95%, 2=90-94%, 3=80-89%, 4=<80%)
SIA CategoryClassification of countries as to whether SIAs were held every 4-6 years, (1=4-6 yrs, 2= >6 yrs)
SIA CoverageVaccination coverage of SIAs (Average of 3 campaigns) (1>=95%, 2=90-94%, 3=80-89%, 4=<80%)
Population Density People per km2 for year 2000 (quartiles)
GNI per Capita Gross National Index per capita for the year 2000 (quartiles)
Proportion Urban Population Proportion Urban population for the year 2000 (quartiles)
Infant Mortality Infant Mortality for 2000 (per 1000 persons) (quartiles)
0
20
40
60
80
100
120
140
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
Year
Mea
sles
Cas
es p
er 1
mill
ion
popu
latio
n
0102030405060708090100
MCV
1 C
over
age
2-dose countries 1-dose countries 2-dose countries 1-dose countries
Measles Cases and MCV1 Coverage among countries with 1 or 2 doses* of MCV1, Region
of the Americas 1994-2002
*Over time, new countries introduced MCV2 at which point that country moved from one-dose to 2-dose country category
Methods: Question 2 Were any variables independent predictors of
elimination?
• Bivariate analysis• Variables were categorized
- A priori knowledge (immunization program)- Quartiles (socioeconomics and demographics)
• Outcome Categories: Early (<2000) vs. Late (≥2000)
•
Question 2: Results and Summary
• Surveillance quality (p=0.03) only variable significantly (p<0.05) associated with measles elimination
• Percent urban population (p=0.05) approached significance
Summary of comparison of indicators* versus last confirmed case
Fisher's Exact
Indicator (table probability) P-value
MCV1 Coverage 0.04 0.61
SIA Coverage 0.03 0.49
SIA Category 0.34 1.00
Population Density 0.03 0.78
Percent Urban Population 0.002 0.05
Infant Mortality Rate 0.03 0.57
GNI per Capita 0.005 0.17
*Each indicator was broken down into specific categories
Propensity Scores: 13 variable model
Demographic & Socioeconomic Variables
• Population density• Literacy rate• GNI per capita• Proportion urban population• % GNP spent on healthcare• Infant mortality• School attendance
Program Variables
•MCV1 coverage•SIA category•SIA coverage•DTP 1 coverage•DTP3 coverage•Surveillance
Propensity Scores- “Full Model”1-Dose Countries
Country Propensity ScoreVenezuela 0.61Cuba 0.77El Salvador 0.77Peru 0.78Belize 0.80DR 0.80Brazil 0.83Guatemala 0.88Nicaragua 0.88Haiti 0.89Bolivia 0.90Ecuador 0.92Honduras 0.92Paraguay 0.92
2-Dose CountriesCountry Propensity Score
Uruguay 0.14Mexico 0.18Argentina 0.21Colombia 0.26Panama 0.33Costa Rica 0.39Chile 0.46
Mean and median time to elimination* for 1-dose and 2-dose countries
*From 1992 to PAHO declaration of Measles Elimination in 2002
Doses # of countries Mean 95% CI Median Min:Max
1-Only 14 7.2 5.8-8.6 7.5* 4.0:10.0
2-Early** 4 7.0 3.8-10.2 8.0* 4.0:8.0
2-Late*** 3 9.3 7.9-10.7 9.0* 9.0:10.0
All 2-Dose 7 8.0 6.2-9.8 8.0**** 4.0:10.0
All 21 7.5 6.5-8.5 8.0 4.0:10.0
*X2 = 3.19, p-value = .20 (no statistical difference, using Kruskal-Wallis test)** Early introduction refers to countries that introduced in 1992***Late introduction refers to countries that introduced in 1997-1998****X2 = 0.24, p-value = .62 (no statistical difference, using Kruskal-Wallis test)
*1992 chosen as start point since year of first two-dose schedule in region
Survival Analysis: 1-dose vs. early and late 2-dose
0.00
0.25
0.50
0.75
1.00
1992 1994 1996 1998 2000 2002
1 dose measles schedule
2 dose measles schedule – early introducers2 dose measles schedule – late introducers
Time to Elimination
Time to elimination similar (X2 = 0.27, p-value =0.26)
Secondary Outcome: Susceptibility
Susceptibility under 15 years of age
Routine MCV Vaccination
Schedule Mean (%) (95% CI)
Median
1-dose countries 7.4 (5.5-9.2)
2-dose countries0% MCV2 coverage 5.9 (3.9-7.9) 5.5
50% MCV2 coverage 4.1 (2.3-5.8)* 3.575% MCV2 coverage 3.3 (1.5-5.0)* 2.5
7.1
* statistically significant
Acknowledgement
This project is a collaborative effort with PAHO, and we would like to thank Dr. Carlos Castillo-Solorzano, Dr. Jon Andrus, Dr. Carolina Danovaro, and Ms. Christina Marsigli for their contribution and help with the project.
Objectives…cont.
• This study was designed to address whether a second routine dose had added benefit in measles elimination
• While new questions have been raised, i.e., if a second routine dose aids in sustaining measles elimination, we did attempt to asses this
• IS THIS EVEN NEEDED??
Mean Measles Susceptibility by one and two-dose countries
N=8
N=13
***
*Assuming 50% coverage for MCV2**Assuming 75% coverage for MCV2
Pro
porti
on o
f per
sons
und
er 1
5 su
scep
tible
to m
easl
es (%
)
Propensity Score Matching Methods• Why: Good approach for small sample size
or where there is no control group
• What: Estimates the probability of exposure given a set of possible covariates
• How: Model the conditional probability of exposure (one-dose vaccination schedule) given potential covariates
• Outcome: “Summary” score with balanced covariates to reduce bias
• Then what: Match 1-dose country and 2-dose country propensity scores +/- 5%; countries with matching propensityscores used in final analysis
0
20004000
60008000
10000
1200014000
16000
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Year
Num
ber o
f Mea
sles
Cas
esConfirmed Measles Cases, the
Americas, 1998-2007
Measles Elimination
Introduction of 2nd Routine Measles Vaccine Dose by Country
1990 Never92 98 04 0894 96 2000 02 06
ChilePanamaUruguayCosta Rica
Colombia
ArgentinaMexico
El Salvador
BrazilCubaParaguay
Belize
BoliviaNicaraguaHondurasPeruDREcuadorVenezuelaGuatemalaHaiti
*includes Caribbean countries
Case-Based Elimination Year
1990 92 94 96 98 2000 02 04 06 08
NicaraguaCubaPanamaBelize
Honduras
GuatemalaParaguay
UruguayCosta RicaChile
BoliviaPeruArgentina
El SalvadorEcuador
ColombiaHaitiDR
BrazilMexico
Venezuela
Propensity Score Matching
Matching Algorithms• Nearest neighbor matching• Caliper/Radius matching• Stratification/Interval matching• Kernel & local linear weights• Weighting
Assess balance between groups one matched
Results: Immunization Program Variables
2-Dose Countries 1-Dose Countries
Variable Mean (95% CI) Median (Range) Mean (95% CI) Median (Range)
MCV1 Coverage 91 (85-97) 94 (80-96) 92 (88-96) 94 (73-99)
Measles SIA Coverage 88 (80-95) 91 (77-97) 91 (89-94) 94 (77-97)
DTP1 Coverage 95 (92-97) 94 (92-98) 93 (90-97) 96 (77-98)
DTP3 Coverage 87 (80-94) 88 (74-98) 83 (74-92) 87 (44-99)
Measles Dropout Rate 4.3 (-2.9-11) 3.2 (-6.7-19) 6.2 (1.0-11) 3.1 (-5.7-29)
DTP Dropout Rate 6.7 (0.45-13) 3.3 (1-20) 12.2 (5.7-19) 10.3 (0-42)
Surveillance Quality* 80 (62-99) 90 (44-97) 84 (75-94) 91 (48-100)
Source: Country reports to PAHO-WHO/UNICEF *Percent of sites reporting weekly to PAHO, 1999-2001
Results: Socioeconomic and Demographic Variables
2-Dose Countries 1-Dose CountriesVariable Mean (95% CI) Median (Range) Mean (95% CI) Median (Range)Population density (people per km2)*
37 (17-58) 40 (13-77) 86 (29-144) 43 (7.7-303)
Proportion urban population (%)*
76 (63-89) 75 (56-92) 60 (51-68) 58 (36-87)
Female literacy (%)* 94 (91-97) 96 (89-98) 80 (72-88) 84 (48-97)
GNI per capita (USD)* 473 (309-637) 486 (206-747) 335 (191-479) 217 (100-860)
Percent of GDP spent on Health Care*
7.3 (5.7-8.9) 7.0 (5.1-10.1) 6.2 (5.7-6.8) 6.4 (4.4-7.7)
Net Primary School Enrollment (%)**
90 (79-100) 90 (67-99) 86 (75-98) 91 (21-98)
Doctor and Nurse Ratio (per 10,000 population)*
97 (86-108) 96 (72-107) 92 (81-104) 98 (27-104)
Infant Mortality (per 1000 live births)*
17 (12-22) 18 (9-23) 33 (25-42) 32 (7.5-62)
•Source: PAHO Basic Indicators http://www.paho.org/English/SHA/coredata/tabulator/newTabulator.htm**Source: UNESCO database http://stats.uis.unesco.org/unesco/TableViewer/document.aspx?ReportId=143&IF_Language=eng
Mean Measles Susceptibility by One-dose and Two-dose Countries
012
34567
89
10
1-Dose Countries
N=14
2-Dose CountriesN=7
Pro
porti
on o
f per
son
unde
r 15
susc
eptib
le to
mea
sles
(%)
0% 50%* 75%**
Estimated Herd Immunity Threshold
*Assuming 50% Coverage fro MCV2**Assuming 75% Coverage for MCV2