Paid maternal leave and childhood vaccination uptake

advertisement
PAID MATERNITY LEAVE AND CHILDHOOD
VACCINATION UPTAKE:
A MULTILEVEL ANALYSIS IN 20 LOW-AND-MIDDLE-INCOME
COUNTRIES
Mohammad Hajizadeh, Jody Heymann, Erin Strumpf, Sam Harper, Arijit Nandi
Presented by: José M. Mendoza Rodríguez
Sep. 4-6 2014, Oslo, Norway
2
Background and research question





Vaccine-preventable diseases continue to be a public health problem in LMICs:
o In 2013, 21.8 million children under age 1 did not receive DPT3 vaccine
worldwide (WHO 2014).
o 70% of these children live in 10 countries: DR Congo, Ethiopia, India,
Indonesia, Kenya, Mexico, Nigeria, Pakistan, Viet Nam and South Africa.
o About 29% of deaths in children 1-59 months of age are vaccine preventable.
Vaccination uptake is not universal even in countries where vaccinations are free Trunz et al. (2006), Soares (2007).
Some vaccine preventable diseases are also re-emerging due to low immunization
rates in middle-and-high-income countries - Glatman-Freedman & Nichols (2012).
Key factors influencing childhood vaccination include parental education, maternal
age, household living conditions, financial factors, place of residence, availability of
vaccines, distance to clinics, transportation, and mass media campaigns.
Studies from both low and high income countries have indicated ‘conflicting work
schedules’ as a barrier to the immunization of children. – McCormick et al. (1997)
3
Background and research question




Paid maternity leave -- “leave that the country guarantees employed women in
connection with the birth of a child” (Heymann et al. 2011) - can provide
households with the opportunity to vaccinate their children without trading off
income generation. – Daku et al. (2012)
Few studies have examined the effect of maternity leave policies on vaccination:
o Berge et al. (2005) found a negative impact of early return to work on
diphtheria, pertussis, and tetanus (DPT) and Polio vaccinations in the US.
o Tanaka (2005) did not find an association between duration of maternal leave
and vaccination uptake in OECD countries.
o An ecological study on 185 countries showed that paid maternity leave was
associated with higher childhood vaccination rates – Daku et al. (2012)
Effect of national paid maternity leave policies on individual-level vaccination
outcomes has not been evaluated in low- and middle-income countries (LMICs).
How do paid maternity leave policies impact the probability of vaccination uptake
in LMICs?
4
Data – Vaccination uptake
 The individual-level outcomes comprise the uptake of:
o Bacillus Calmette-Guérin (BCG)
o Diphtheria, pertussis and tetanus (DPT) - three doses
o Polio – three doses
 Childhood immunization data comes from the Demographic and
Health Surveys (DHS), collected for all children under 5 through
vaccination cards or verbal reporting of mothers.
 Constructed a representative panel of child observations occurring in
selected LMICs between 2001 and 2008. Selection of countries was
determined by the availability of at least two DHS surveys between
2001 and 2011.
o Child observations were allocated to each year based on the
child’s year of birth.
5
Data - Vaccination uptake
 Panel: 277,787 child observations in 20 DHS countries over the 2001-
2008 period
Country
Honduras
Nepal
Uganda
Bangladesh
Armenia
Cambodia
Colombia
Rwanda
Senegal
Zimbabwe
Malawi
Tanzania
Lesotho
Ghana
Kenya
Madagascar
Nigeria
Philippines
Bolivia
Egypt
DHs survey years 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
2011 2005
2011 2006
2011 2006
2011 2007 2004
2010 2005
2010 2005
2010 2005
2010 2005
2010 2005
2010 2005
2010 2004
2010 2004
2009 2004
2008 2003
2008 2003
2008 2003
2008 2003
2008 2003
2008 2003
2008 2005
 These data were merged to longitudinal information on paid
maternity leave for each country
6
Data – Vaccination uptake
 We excluded all births that occurred less than four months prior
to the survey to allow each child a follow-up period of at least
four months to receive the vaccinations recorded by the DHS.
Daku et al. (2012)
7
Data – Paid Maternity Leave

Our country-level exposure was the legislated length of paid maternity
leave in full time equivalent (FTE) weeks for each country-year (2001-2008).
 For each country and year, we recorded the legislated length of leave
available to mothers only.
 We calculated the length of paid maternity leave in FTE weeks by
multiplying the legislated length of leave by the wage replacement rate.
 Data sources
 National labour legislation
 The Social Security Programs Throughout the World database (SSPTW)
 Other sources:
o
o
o
International Labour Organization’s Maternity Protection Database
Council of Europe Family Policy Database
International Review of Leave Policies and Related Research
8
Data – Paid Maternity Leave (FTE weeks): 2000-2008
25
20
FTE/WEEKS
15
10
5
0
2000
2001
2002
2003
2004
2005
2006
2007
2008
9
Data – Control variables
We accounted for potential confounding by:
 Individual-level covariates, obtained from the DHS:
o socio-demographic characteristics of the mother and household
(e.g., mother’s education and urban/rural area of residence)
o relevant birth characteristics (e.g., gender, mother’s age at birth,
birth order, and number of children)
o attendance of skilled health personnel
 Country-level covariates, World Bank's WDI and Global Development
Finance (GDF):
o per capita GDP (PPP)
o per capita total health expenditures
o per capita government health expenditure
o female labor force participation
10
Summary Statistics – All variables
Variable
Mean
Std. Dev.
BCG
0.89
0.31
DPT1
0.86
0.34
DPT2
0.83
0.38
DPT3
0.76
0.43
Polio1
0.91
0.29
Polio2
0.86
0.35
Polio3
0.74
0.44
9.93
3.44
GDP/cap-log
6.43
0.77
Total health expenditure/cap-log
4.55
0.78
Government health expenditure/cap-log
3.75
0.90
Female labor-force participation
52.81
18.59
Mother's education/ year
5.29
4.50
Household size
6.65
4.02
Urban
0.30
0.46
Male
0.51
0.50
Female (ref.)
0.49
0.50
Birth order # 1
0.26
0.44
Birth order # 2
0.22
0.41
Birth order # 3 and above (ref.)
0.52
0.50
Mother's age at birth - 19 and below
0.12
0.33
Mother's age at birth - 20 to 39 (ref.)
Mother's age at birth - 40 and above
Other
0.83
0.05
0.38
0.22
0.50
0.50
Outcome Variables
Exposure variable
FTE/week
Control variables
Country-level covariates
Household-level covariates
Birth characteristics
Attendance of skilled health personnel
11
Summary Statistics – Vaccination Rates
0.95
0.9
0.85
0.8
0.75
0.7
0.65
2001
2002
BCG
2003
DPT1
2004
DPT2
2005
DPT3
2006
Pol1
2007
Pol2
2008
Pol3
12
Empirical strategy: regression with fixed effects
 Linear Probability Regression Model (LPM) of the form:
𝑌𝑖𝑗𝑡 =∝ +𝛿𝑀𝐿𝑗𝑡−1 +
𝛽𝑘 𝑋𝑘𝑖𝑗𝑡 + 𝜑𝐶𝑗 + 𝛾𝑇𝑡 + 𝜀𝑖𝑗𝑡 ,
𝑘
o
𝑌𝑖𝑗𝑡 : binary vaccination outcome (receiving BCG, DPT, Polio)
o
𝛿 : effect of 1 additional FTE week in paid maternity leave 𝑀𝐿𝑗𝑡
o
𝑋𝑘𝑖𝑗𝑡 : set of individual, household and country-level covariates
o
𝑗 indexes country, 𝑖 indexes child, and 𝑡 indexes year of birth
 We included fixed effects for country (𝐶𝑗 ) and year (𝑇𝑡 ) to control for
unobserved time-invariant confounders that vary across countries,
and any temporal trends in vaccination shared across countries
 Incorporated respondent-level sampling weights and robust standard
errors to account for clustering
13
Effect of 1 additional FTE week of paid maternity leave on BCG vaccination
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Model 7
0.0162 (0.0006)
0.0048 (0.0009)
0.0033 (0.0009)
0.0022 (0.0008)
0.0017 (0.0008)
0.0018 (0.0008)
0.0016 (0.0008)
-0.1255 (0.0199)
-0.08 (0.026)
-0.1035 (0.0253)
-0.1054 (0.0253)
-0.123 (0.0252)
Total health expenditure/cap-log
0.0121 (0.0108)
0.0138 (0.0104)
0.0135 (0.0104)
0 (0.0103)
Government health expenditure/cap-log
-0.0057 (0.008)
-0.0005 (0.0077)
-0.0003 (0.0077)
0.0065 (0.0076)
0.0005 (0.0005)
0.0007 (0.0004)
0.0007 (0.0004)
0.0001 (0.0004)
0.014 (0.0003)
0.0137 (0.0003)
0.011 (0.0003)
Household size
-0.0008 (0.0003)
-0.0008 (0.0003)
-0.0007 (0.0003)
Urban
0.0212 (0.0031)
0.0205 (0.0031)
0.001 (0.003)
-0.0007 (0.0014)
-0.0013 (0.0014)
Birth order # 1
0.0092 (0.002)
-0.0018 (0.002)
Birth order # 2
0.0062 (0.0017)
0.0019 (0.0017)
19 and below
-0.0293 (0.0028)
-0.0245 (0.0027)
40 and above
0.0007 (0.0038)
-0.0009 (0.0038)
FTE
Country-level covariates
GDP/cap-log
Female labour-force participation
0.0013 (0.0005)
Household-level covariates
Mother's education
0.0147 (0.0003)
Birth characteristics
Gender/Male
Birth order
Mother's age at birth
Other
Attendance of skilled health personnel
0.0911 (0.0025)
Sample weights
Y
Y
Y
Y
Y
Y
Y
Clustered SEs
Y
Y
Y
Y
Y
Y
Y
Country FE
Y
Y
Y
Y
Y
Y
Time trend
Y
Y
Y
Y
Y
Y
Note: Bold indicates statistical significance at 5 percent or less. Standard errors in parentheses. Findings are robust to the use of Poisson
regression models and an alternate measure of maternity leave based on the ILO convention
paid maternity leave: LPM results
.
14
Overall Results – Full models
Effect of 1 additional FTE week of maternity leave on
probability of vaccination
BCG
0.16
3.35
DPT1
DPT2
3.24
DPT3
2.98
Polio1
0.01
Polio2
0.17
Polio3
-1
-0.5
-0.06
0
0.5
1
1.5
2
2.5
3
3.5
4
Percentage points
Note: 95% confidence interval bars included. Findings are robust to the use of Poisson regression models and an alternate measure of
maternity leave based on the ILO convention
15
Conclusions





Each additional FTE week of paid maternity leave was associated with:
o 0.16 percentage point increase in the probability of BCG vaccination
o 3.35, 3.24 and 2.98 percentage points increase in the probability of DPT1/2/3
o Findings were robust to inclusion of country- and household-level covariates,
birth characteristics, and attendance of skilled health personnel.
Findings suggest that there is a benefit to increasing paid maternity leave even for
vaccines scheduled soon after birth such as BCG in LMICs.
No evidence of effect on the probability of receiving all three doses of the Polio
vaccine, after adjustment for potential confounders.
Similar to Berge et al. (2005) and an ecological study by Daku et al. (2012), our
results indicate that paid maternity leave was positively associated with higher
immunization rates for BCG and all three doses of DPT.
Policy Implication: Maternity leave policies may represent an important mechanism
for removing barriers to improved vaccination coverage and child health.
16
Limitations





Use of mothers' recall for determination of child vaccination status when
vaccination cards were not available: recall bias
o These data may still be valid: Valadez & Weld (1992), AbdelSalam &
Sokal (2004)
Our maternity leave variable is calculated based on the legislated maternity
leave and does not account for other leave (i.e., parental leave).
Employment history of women around time of birth not available in DHS.
Time-varying covariates in our analysis are subject to measurement error
because they are taken at the time of interview and assigned to all prior
births (e.g., mother’s education).
Potential unmeasured confounding: it is possible that another policy or
program that influenced vaccination coverage coincided with changes in
maternity leave policy.
17
Thank you
http://machequity.com
Email: jose.mendozarodriguez@mcgill.ca
18
Appendix
Vaccination and Maternity Leave Data
19
Data – Vaccination uptake
 Mothers surveyed in the DHS are asked to provide vaccination
information concerning live births over the previous 59 months
20
Data – Paid Maternity Leave
Rwanda Labour Code, Article 68:
• Upon delivery, every employed woman has the right to suspend her
job for a period of 12 consecutive weeks, of which at least 2 weeks
are taken before the presumed date of delivery and 6 weeks
afterwards
• The employer cannot give the employed woman a notice of lay off
during her maternity leave
• The employed woman has the right, during the period of contract
suspension, at the charge of the employer, and until the instauration
of a social security system that assumes the full responsibility of the
matter, to 2/3 of the salary she received before suspending her job
FTE weeks of leave = 12 * 2/3 = 8
paid maternity leave: LPM results
.
21
Effect of 1 additional FTE week of paid maternity leave on DPT1
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Model 7
0.033
(0.0011)
0.034
(0.0011)
0.0336
(0.0011)
0.0336
(0.0011)
0.0335
(0.0011)
0.1199
(0.0208)
0.054
(0.0257)
0.0801
(0.0113)
0.0377
(0.0089)
0.0053
(0.0005)
0.0315
(0.0253)
0.0815
(0.011)
0.0428
(0.0086)
0.0055
(0.0005)
0.0297
(0.0252)
0.0812
(0.011)
0.043
(0.0086)
0.0054
(0.0005)
0.0141
(0.0252)
0.0694
(0.0109)
0.0489
(0.0086)
0.0049
(0.0005)
0.0133
(0.0003)
-0.0009
(0.0004)
0.013
(0.0033)
0.013
(0.0003)
-0.0009
(0.0004)
0.0124
(0.0033)
0.0105
(0.0003)
-0.0008
(0.0004)
-0.005
(0.0032)
-0.0006
(0.0014)
-0.0011
(0.0014)
0.0096
(0.0022)
0.0076
(0.0019)
-0.0002
(0.0022)
0.0038
(0.0019)
-0.0248
(0.0029)
0.0064
(0.004)
-0.0206
(0.0028)
0.005
(0.0039)
DPT1 (First Dose)
0.0214
(0.0006)
FTE
0.0318
(0.001)
Country-level covariates
GDP/cap-log
Total health expenditure/cap-log
Government health expenditure/cap-log
Female labor-force participation
0.006
(0.0005)
Household-level covariates
Mother's education
Household size
Urban
0.0136
(0.0003)
Birth characteristics
Gender/Male
Birth order
Birth order # 1
Birth order # 2
Mother's age at birth
19 and below
40 and above
Other
Attendance of skilled health personnel
0.0807
(0.0027)
Note: Bold indicates statistical significance at 5 percent or less. Standard errors in parentheses. Findings are robust to the use of Poisson
regression models and an alternate measure of maternity leave based on the ILO convention
Download