Embed Size (px)
Citation preview
Making Time for Well-Baby Care: The Role of MaternalEmployment
Mary Kathryn Hamman
Published online: 13 August 2010
� Springer Science+Business Media, LLC 2010
Abstract The American Academy of Pediatrics recom-
mends children receive six well-baby visits between ages
1 month and 1 year, yet by age 14 months less than 10% of
infants have received all six visits. Cost sharing under
public and private insurance is very low. Low compliance
rates despite the low cost of care suggest other factors, such
as time costs, may be important. This paper examines the
relationship between maternal employment and receipt of
well-baby care. The Medical Expenditure Panel Survey
contains rich information on use of preventive care,
maternal employment, and other economic and non-
economic factors that may influence care decisions. Several
approaches, including a proxy variable strategy and
instrumental variables analysis, are used to attempt to
address the potential endogeneity of maternal employment
and examine the sensitivity of findings. Findings indicate
mothers who work full-time take their children to 0.18
fewer visits (or 9% fewer at the mean) than those who have
quit their jobs. Mothers with employer provided paid
vacation leave take their children to 0.20 more visits (or
9% more at the mean) than other working mothers. Time
appears to be an important factor in determining well-baby
care receipt. Policies that extend paid leave to more
employed women may improve compliance with pre-
ventive care recommendations.
Keywords Well-baby care � Preventive care � Maternal
employment � Paid leave
Introduction
The American Academy of Pediatrics (AAP) recommends
children receive six well-baby visits by age 12 months, in
addition to a newborn and a 3–5 day follow-up visit after
hospital discharge [1]. However, fewer than 10% of infants
meet recommendations. Visits require both time and
monetary resources. Yet, out of pocket costs are very low
for the insured, and insurance coverage rates are high
among infants. The fact that so few infants receive all
recommended care despite low costs suggests parents may
have trouble finding the time for visits.
One-third of mothers return to work within 3 months of
giving birth [2]. For these mothers, visits must either be
scheduled around work, or during paid or unpaid time off.
Therefore employed mothers, especially those working
full-time, have less time available for visits. Furthermore,
employed mothers without access to paid or unpaid leave
may have less control over their time than those who have
leave. Yet, additional household income may increase the
ability to pay for visits. Also, if time is scarce employed
mothers may actually find preventive care to be more
beneficial because it reduces the risk of time lost due to
illness in the future.
This paper examines the following questions: First, does
maternal employment compete with receipt of recom-
mended well-baby visits? Second, does access to employer
provided paid leave or unpaid leave under the Family and
Medical Leave Act (FMLA) increase receipt of recom-
mended visits? To address these questions, data from the
Medical Expenditure Panel Survey (MEPS) are analyzed.
Results indicate employment, and full-time employment in
particular, may compete with receipt of care but the neg-
ative relationship is at least partially offset if mothers have
access to paid vacation. This paper contributes to the
M. K. Hamman (&)
School of Human Resources and Labor Relations,
Michigan State University, 431 S. Kedzie Hall,
East Lansing, MI 48824, USA
e-mail: [email protected]
123
Matern Child Health J (2011) 15:1029–1036
DOI 10.1007/s10995-010-0657-9
understanding of preventive care receipt at a particularly
important point in the human developmental lifecycle.
The goals of well-baby care are to prevent infant mor-
tality and promote health. The anticipatory guidance pro-
vided during visits targets avoidable risks common in early
infancy and childhood. For example, in 1990 the Sudden
Infant Death Syndrome rate was 1.30 per 1,000 live births
[3]. Since proper sleep position guidance was publicized in
1992, the percentage of infants placed on their backs to
sleep increased from 13 to 72.8%, and the rate of SIDS fell
by over 50% [4]. Although parents could have received sleep
position guidance from many sources, a portion of each well-
baby visit is specifically devoted to age-appropriate pre-
ventive guidance, such as proper sleep position.
Well-baby visits also include health screenings. Early
diagnosis of existing conditions can improve prognosis.
For example, children who receive screening for strabis-
mus (commonly known as ‘‘lazy eye’’) or other sources of
limited visual acuity (amblyomia) and are diagnosed before
age 3 years have better long run visual acuity than children
diagnosed at age 3 years [5, 6]. Similarly, infants who are
diagnosed with congenital hearing loss have a higher
likelihood of developing speech than those whose hearing
loss goes undetected in infancy [7]. Screenings can occur in
other medical settings, but well-baby visits are specifically
designed to include screening. The visit schedule is set to
target critical points in a child’s development.
In 2003, 8.9% of infant deaths that occurred after the
perinatal period and were not associated with congenital
abnormalities were the result of infectious, endocrine,
nutritional, or metabolic diseases [8]. Vaccines are available
for many infectious diseases. For some, treatment can be as
simple as administering vitamin supplements and maintain-
ing routine follow-up care. Well-baby visits include screen-
ings for many of these conditions, and routine physical
examinations may aid early diagnosis and treatment.
Despite the importance of these visits, most studies find
babies receive significantly less than the recommended
amount of care [9–11]. The literature includes many pos-
sible explanations for underutilization including inadequate
insurance coverage, racial disparities, lack of parental
education, and limited access to care [12–14]. Three pre-
vious studies examine the impact of maternal employment
on receipt of care. Colle and Grossman [15] find time costs
reduce the likelihood of receiving pediatric preventive
care, and the estimated effect is largest among children
whose mothers work. Their study also highlights the
importance of the price of care and household income.
Among children ages 0–15, Vistnes and Hamilton [16] find
the number of visits a child receives decreases as mother’s
hours worked per week and weeks worked per year
increase. Berger et al. [17] find mothers who return to work
early (within the first 12 weeks after giving birth) are 2.4%
points less likely to take their babies for any well-baby care
in the first year of life and their babies receive 0.20 fewer
visits on average. Findings on the impact of paid leave are
more mixed; Vistnes and Hamilton [16] find no evidence of
an effect of sick leave but Berger et al. [17] find positive
relationships between paid maternity leave duration and
well-baby care use. No previous study has considered the
influence of FMLA leave eligibility.
In summary, previous studies provide some evidence of
a negative relationship between maternal employment and
well-baby care. However, given the complex nature of
care-giving and maternal employment decisions, these
studies and the present study face challenges to identify
causal effects. The methods used to address these chal-
lenges can materially influence the results. For example, in
the broader literature, Blau et al. [18] show that after
accounting for endogeneity of maternal employment, there
is no evidence of a negative relationship between maternal
employment and child health. Of the well-baby care stud-
ies, Berger et al. [17] is the only study that attempts to
address the endogeneity problem. They use propensity
score matching to model mother’s return to work behavior.
Their propensity score matching technique yields results
that are very similar to their OLS estimates. The relevance
of this problem to the present study is discussed in the
results section alongside three new approaches to it.
Methods
This study investigates the following two hypotheses:
1. Maternal employment creates time constraints that
reduce the amount of well-baby care received.
2. Access to employer provided paid leave or FMLA
leave (at least partially) mitigates any negative rela-
tionship between maternal employment and receipt of
care.
It is possible that the true relationships are exactly the
opposite of those proposed here. Maternal employment
increases household income, which may in turn increase
the family’s ability to pay for well-baby care. Employed
mothers who have access to paid leave may see less value
in preventive care because they have time available to care
for a sick child. Alternatively, if paid time off is rationed,
employed mothers may feel it is better to save that time for
occasions when their child is sick. Given these competing
explanations, the significance of all estimated relationships
is evaluated using conventional two tail hypothesis tests.
The data used in the analysis are drawn from the 1996
through 2005 Household Component, Event Files, and
Conditions Files of the MEPS. Although the MEPS con-
sists of 2 year panels, the panel data structure was not
1030 Matern Child Health J (2011) 15:1029–1036
123
exploited because survey waves do not occur frequently
enough to match time varying covariates to specific visits.
Survey waves occur five times over 2 years; there are six
recommended well-baby visits in the first year of life alone
(not including the first visit which should occur within
1 week of the birth). So, the sample consists of all children
14 months and younger in the 1996 through 2005 MEPS
surveys (N = 3,536). Each child contributes one observa-
tion. Time varying covariates are measured as means
across all survey waves a child is observed.
Respondents (generally the mother) identify well-baby
visits when asked the main purpose of a visit. However,
even sick visits may include some well-baby care compo-
nents. Therefore, the analysis begins with an examination
of substitution between well-baby visits and other office
based visits.
Throughout the analysis, the dependent variable is a
count of on-time well-baby visits received. Both the number
and timing of visits are important. To incorporate both
aspects in the dependent variable measure, all visits are
assigned to recommended care intervals (RCI). An RCI is a
time period defined to contain the age at which a visit is
supposed to occur under the AAP guidelines. To illustrate
the construction of RCIs for each visit, Fig. 1 displays a
histogram of the ages at which well-baby visits were
received among infants in the MEPS sample. The horizontal
axis highlights the ages at which the AAP recommends
visits. The peaks indicate actual timing of visits corresponds
closely with recommendations. When the recommended
ages for visits are more than 1 month apart, the bins to the
right of each recommended age generally contain more
visits than the bins to the left. Thus, each RCI is defined to
begin with the month in which the visit is recommended and
continue until the month of the next recommended visit.
The only exception to this coding rule is the 1 month visit;
infants who receive a visit before 1 month are considered to
have received an on-time 1 month visit even if they do not
receive another visit at age 1 month. An infant can receive
future on-time visits even if he did not receive on-time visits
in the past. Given the construction of the dependent vari-
able, throughout the analysis regression coefficients can be
interpreted as an estimated marginal change in the number
of on-time visits received.
Not all children in the sample are observed for all rec-
ommended visits in the first year. Since some children are
born during the survey they exit the survey before they
reach age 14 months; others are born before and enter the
survey after age 0 months. The dependent variable is
defined so that children can only contribute on-time visits
for whole RCIs observed. For example, if a child exits the
survey at age 13 months, the dependent variable will not
count any visits he received during the RCI corresponding
to the 12 month visit because he exited the survey before
that RCI was completed. All specifications control for the
first RCI observed and the length of time (in months)
encompassed by the first and last RCI observed.
Throughout the analysis, maternal wages proxy for the
monetary value of time. Yet, wages are unobserved for non-
employed mothers. To address this issue, past observed
wages for mothers who worked at any time during the survey
are used in place of actual wages. Wages are predicted for
mothers who are non-employed throughout the survey.
Similarly, out of pocket cost data is missing for visits
not received. Missing costs are predicted using the average
cost of all previous office based visits and insurance cov-
erage information. Cost sharing for well-baby care under
Medicaid was eliminated in 1996. Thus, any missing prices
for Medicaid recipients are coded as 0.
All regressions control for whether or not the child has a
usual care provider and the child’s insurance coverage,
race, and subjective health status reported by the household
respondent (usually the mother). Mother’s education,
marital status, and age are included, as are the number of
children in the household by age group, the presence of
adults other than the parents, the region of residence, and
urban or rural residence.
Results
Before analyzing the relationships of interest, it is neces-
sary to confirm the well-baby care measure includes all
care received. As stated previously, households identify the
main purpose of each doctor visit received. Yet a child may
receive some components of well-baby care during a visit
where the main purpose is to treat an illness. If the measure
used in the analysis includes only visits identified as
Fig. 1 Distribution of ages at which visits were received. Note The
x-axis denotes the ages at which the AAP recommends visits. SourceAuthor’s calculation using Medical Expenditure Panel Survey
1996–2005
Matern Child Health J (2011) 15:1029–1036 1031
123
well-baby care and less healthy infants actually receive a
non negligible amount of well care during sick visits, the
measure will systematically underestimate care received
for this group. Among the 647 infants observed for all six
RCIs, the correlation between well-baby visits and sick
visits is 0.15. So it seems even if infants do receive some
well care during sick visits, it is not enough to supplant
routine well-baby visits.
The MEPS also allows parents to identify the main
purpose of a visit as a general check-up, immunization, or
postnatal care visit. All of these designations could be
construed by the parent as equivalent to well-baby care. In
the data, the correlation between well-baby visits and these
other types of preventive care is -0.49. So, there is some
evidence of substitution between well-baby visits and these
other types of preventive care visits. Thus, the remainder of
the analysis will include visits identified as general check-
ups, immunizations, or postnatal care delivered to the child
as well-baby visits.
Figure 2 presents a histogram of the total number of on-
time well-baby visits received among the 647 infants
observed for all six recommended care intervals. The
majority of the sample received less than the recommended
6 visits, and approximately half received fewer than 3
visits. The population weighted mean number of visits
received is 3.68.
Table 1 presents descriptive statistics and cross-
tabulations for key variables. In these simple descriptive
statistics, average household income and mother’s hourly
wage are higher among infants who receive more visits.
Furthermore, 61% of mothers who took their children to 4
or more visits are employed as compared to 53% of all
mothers. There are no notable differences in usual hours of
work between the mothers in each group, but those who
took their children to more visits seem to be more likely to
have paid vacation or sick leave offered through their
employer. This may be because the leave enables mothers
to take their babies to visits, or simply because mothers in
more advantaged households are more likely to have jobs
with better leave policies. Multivariate analysis is needed
to distinguish between these two explanations.
Table 2 contains the baseline results of multivariate
analysis. Out of pocket cost of care, mother’s wage, and
household income variables all have the predicted signs.
Yet, the estimated relationship between mother’s wages
and visits received is small and not statistically significant.
Therefore, this estimate does not support or refute
Hypothesis 1. The own-price elasticity indicates a 1%
increase in the monetary cost of a visit would lead to a
0.1% reduction in visits received. At the mean of 3.68, this
would be a reduction of 0.04 visits.
Fig. 2 Distribution of total on-time visits received. Note Includes
infants ages 0–14 months and observed across all RCIs only.
SourceAuthor’s calculation using Medical Expenditure Panel Survey
1996–2005
Table 1 Descriptive statistics and cross tabulations for key variables
Sample mean Sample mean conditional
on 4 or more visits received
Out of pocket cost $8.69 (19.23) $11.81 (20.88)
Wage $11.49 (8.08) $13.27 (8.83)
Income (thousands) $38.36 (33.77) $43.52 (35.63)
Employed 0.53 (0.47) 0.61 (0.45)
Usual weekly hours worked (conditional on working) 34.75 (11.38) 35.08 (12.34)
Access to paid sick leave (conditional on working) 0.56 (0.49) 0.63 (0.47)
Access to paid vacation (conditional on working) 0.63 (0.47) 0.67 (0.46)
Access to FMLA leave (conditional on working) 0.36 (0.47) 0.36 (0.46)
N 3,536 656
N employed 1,654 336
Standard deviations in parentheses
1032 Matern Child Health J (2011) 15:1029–1036
123
Rather than relying on wages to measure the time cost of
care, it may be better to directly enter measures of maternal
employment into the regression equation. These specifi-
cations are presented in Table 3. One prefers these speci-
fications if we believe mothers have difficulty taking the
marginal hour off work or incur costs other than the lost
wages, such as animosity from a supervisor or coworkers.
Adding employment status to the model yields a negative,
but insignificant, coefficient. However, this estimate is
likely biased due to endogeneity.
To address the endogeneity problem, three approaches
are employed. The first is a proxy variable strategy.
Whether or not a mother has ever worked for pay is used as
a proxy for unobserved maternal tastes for paid employ-
ment. These unobserved tastes may also correlate with
factors that lead a woman to take her child to well-baby
visits such as diligence, ambition, and an ability to
understand medical guidelines. The results in Column II
indicate that women who have ever worked outside the
home take their children to 0.38 more visits than those who
have never worked outside the home, but those who are
currently working take their children to 0.13 fewer visits
than those who had worked but quit their jobs.
The second strategy used to address potential endoge-
neity is instrumental variables (IV) estimation. The instru-
ment for maternal employment is a binary indicator for
health insurance held in the mother’s own name before the
baby was born. Controlling for the baby’s current health
insurance coverage and health status, mother’s previous use
of employer provided insurance should have no impact on
current receipt of care. It should, however, relate to the
opportunity cost of returning to work. Indeed, in the first
stage regression mothers who held employer provided
insurance in their own name before the birth worked during
36% more RCIs (P = .00). Unfortunately this instrument is
missing for children who enter the sample after birth, so
observations fall from 3,529 to 1,122. Column III of
Table 3 shows the IV estimate is negative and large relative
to Columns I and II, but not statistically significant
(P = .35). However, if the proxy variable strategy is used
with the limited sample of 1,122, the sign pattern is the
same as in Column II, but maternal employment is insig-
nificant. Therefore, it seems lack of statistical power is a
larger problem than inefficiency of the IV estimate.
The third and final technique for addressing endogeneity
requires fewer assumptions but is more descriptive. If
Table 2 Maternal employment and well-baby visits: derived demand
specification
Out of pocket cost -0.005? (0.003)
Out of pocket cost2 0.000* (0.000)
Ln (full income in 10 thousands) 0.094? (0.055)
Ln (wage) -0.027 (0.048)
Constant 0.474 (0.557)
Observations 3,536
R2 0.496
Own price elasticity -0.001? (0.001)
Income elasticity 0.002? (0.001)
Cross-price elasticity -0.001 (0.001)
Regressions control for usual care provider, insurance coverage by
type, race, subjective health status, mother’s education, mother’s
marital status, quadratic in mother’s age, number of children by age
group, presence of other adults in the household, region of residence,
urban, first RCI observed, and number of months in the sample
** Significant at 1%; * Significant at 5%; ? Significant at 10%
Robust standard errors in parentheses
Table 3 Maternal employment and well-baby visits: alternative specifications
(I) (II) (III) (IV)
Out of pocket cost -0.005? (0.003) -0.005? (0.003) -0.001 (0.005) -0.005? (0.003)
Out of pocket cost2 0.000* (0.000) 0.000* (0.000) 0.000 (0.000) 0.000* (0.000)
Ln (full income in 10 thousands) 0.090 (0.056) 0.072 (0.056) 0.112 (0.098) 0.067 (0.056)
Ln (wage) -0.026 (0.05) -0.020 (0.049) -0.088 (0.080) -0.009 (0.050)
Mother employed -0.028 (0.057) -0.134* (0.063) -0.285 (0.302)
Mother has ever worked for pay 0.377** (0.087) 0.374** (0.085)
Mother works part-time -0.061 (0.073)
Mother works full-time -0.180** (0.065)
Constant 0.421 (0.549) 0.774 (0.974) 0.380 (0.548)
Observations 3,529 1,122 3,529
R-Squared 0.500 0.357 0.501
Regressions control for usual care provider, insurance coverage by type, race, subjective health status, mother’s education, mother’s marital
status, quadratic in mother’s age, number of children by age group, presence of other adults in the household, region of residence, urban, first RCI
observed, and number of months in the sample
** Significant at 1%; * Significant at 5%; ? Significant at 10% Robust standard errors in parentheses
Matern Child Health J (2011) 15:1029–1036 1033
123
Hypothesis 1 is true, then we should find the relationship is
larger (in absolute value) for mothers working full time.
Importantly, these estimates are likely biased for reasons
explained. This technique is descriptive; it does not reduce
any bias. Yet, it does provide a different test of the
hypothesis of interest that relies on the relative magnitudes
of coefficients rather than the absolute magnitude. If biases
are similar in direction and magnitude, this technique may
be more valid than simple OLS or the other methods used.
Column IV presents the results. Estimates do indicate the
negative relationship is larger for mothers who worked
full-time than for those who worked part-time or did not
work.
Hypothesis 2 predicts that among continuously employed
mothers, those with access to paid or unpaid leave will take
their children to more visits. In addition to the control
variables used thus far, mother’s occupation, salaried or
hourly pay status, and job tenure are added to these
regressions. Instead of using part-time or full-time indictor
variables, usual hours worked per week is used because
leave policies are concentrated among full-time employees
and using a full-time indicator results in multicolinearity.
Column I of Table 4 indicates employed mothers with
continuous access to paid vacation leave take their children
to 0.21 more visits than those who do not have any access
to paid vacation leave. Although sick leave does not appear
to have a statistically significant separate effect, access to
paid vacation and to paid sick leave are highly correlated in
the sample (q = .65). Further analysis reveals they are
jointly significant (F = 3.68, P = .03). There is no evi-
dence of a positive relationship between access to FMLA
leave and receipt of care.
One concern with the foregoing results is paid leave
variables may be capturing the advantages of having a
‘‘good job’’ rather than easing of time constraints. As a
sensitivity check, offer of a pension plan was added to the
regression. A pension plan is also an attribute of a good job
but it should have no significant impact on well-baby care
otherwise. Column II presents the results. The point esti-
mate on pension is very small and insignificant, and the
coefficient estimate for paid vacation is relatively
unchanged. So the ‘‘good jobs’’ explanation is unlikely to
be the sole reason for the positive relationship between
paid leave and visits received.
Discussion
The driving question behind this analysis is: does maternal
employment compete with well-baby care? Most specifi-
cations indicate the relationship between full-time maternal
employment and receipt of care is negative and significant,
and the negative relationship holds across specifications
used to account for endogeneity. Mothers who work full-
time take their children to an estimated 0.18 fewer visits
than those who have quit their jobs. Each additional hour of
work is estimated to reduce visits received by -0.006
(P = .07). This implies children whose mothers work 30
hours per week receive approximately 0.06 fewer visits
than those whose mothers work 40 hours per week, which
is in keeping with the estimates for part-time and full-time
work in the full sample presented in Table 3.
Yet, the findings indicate the negative impact of full-
time work may be at least partially offset if the job pro-
vides paid leave. FMLA leave, however, does not appear to
offset any of the negative relationship. This is likely
because very few women are both eligible for FMLA leave
and work for covered employers when they have a child. In
the MEPS sample, only 36% of women were both eligible
for FMLA and working for covered employers. Among
those women, the fact that the law does not require any
remuneration may further limit its use.
In all, a difference of 0.18 visits between mothers
working full-time and those who have quit their jobs seems
quite small given that the average child receives only 3.68
Table 4 Impact of access to paid and unpaid leave among employed
mothers
(I) (II)
Out of pocket cost -0.002 (0.004) -0.002 (0.004)
Out of pocket cost2 0.000 (0.000) 0.000 (0.000)
Ln (full income in 10
thousands)
0.136? (0.076) 0.133? (0.077)
Ln (wage) -0.051 (0.054) -0.051 (0.054)
Mother’s usual hours
of work
-0.006? (0.004) -0.006? (0.004)
Mother’s employer offers
paid vacation
0.206* (0.101) 0.200* (0.102)
Mother’s employer offers
paid sick days
0.072 (0.096) 0.082 (0.099)
Mother is eligible for
FMLA leave
-0.100 (0.084) -0.104 (0.085)
Mother’s employer offers
a pension plan
0.018 (0.089)
Constant 0.493 (1.003) 0.466 (1.004)
N 1,474 1,461
R-Squared 0.550 0.550
Regressions control for usual care provider, insurance coverage by
type, race, subjective health status, mother’s education, mother’s
marital status, quadratic in mother’s age, number of children by age
group, presence of other adults in the household, region of residence,
urban, first RCI observed, and number of months in the sample.
Benchmark occupational category is other occupations
** Significant at 1%; * Significant at 5%; ? Significant at 10%
Robust standard errors in parentheses
1034 Matern Child Health J (2011) 15:1029–1036
123
of 6 recommended visits. However, if mothers who quit
their jobs to care for their families are also those who
experience the greatest employment related constraints
then this figure may underestimate conflicts between work
and care. Unfortunately this data contains too few obser-
vations to rigorously analyze systematic attrition, but as a
first step Table 5 contains descriptive characteristics of
jobs among mothers who quit to care for their family.
The differences in access to leave between women
who quit and those who did not are quite striking. Only
43.21% of mothers who quit have access to paid vaca-
tion time across all RCIs employed as compared to
64.61% of mothers who did not quit. Similarly, 33.09%
of mothers who quit had paid sick leave as compared to
57.56% of those who did not. Thus, it seems women
with the worst jobs are no longer working after their
children are born. This possibility deserves further con-
sideration; it not only means that the estimates in this
study likely understate the true extent of conflict between
maternal employment and care but also that any study of
maternal employment and child outcomes is subject to
this systematic attrition.
The analysis in this study should be considered in light
of the following limitations. First, as illustrated in the
descriptive statistics in Table 5 and discussed above, there
appears to be systematic attrition from the labor force that
is not accounted for in the analysis. The attrition likely
positively biases the reported estimates and thus the esti-
mates understate the true negative relationship between
maternal employment and receipt of well-baby care. Sec-
ond, given the infrequency of survey waves in the MEPS,
analysis was based on cross-sectional variation and time
varying covariates were averaged across RCIs. Thus, the
timing of changes in paid or unpaid leave access could not
be precisely tied to specific RCIs. Clearly the ability to use
longitudinal variation of this sort, provided there are
enough changes in leave eligibility for identification, would
be helpful. Third, the paid leave measures available in the
MEPS are quite blunt; they simply indicate whether or not
the mother’s employer provides either paid vacation or sick
leave but do not measure how much leave she has per year
or how much remains. Since many women must use their
paid leave as part of their maternity leave, it is possible
those who say their employer provides leave actually have
no leave left to use. This should negatively bias the
reported coefficients; the true impact of paid leave is likely
larger than the estimates imply. Finally, although three
approaches were used to address the likely endogeneity of
maternal employment, the first two approaches require
fairly strong assumptions to arrive at unbiased estimates
and the third is purely descriptive. Thus, the magnitude of
point estimates should be extrapolated to underlying causal
relationships with caution. Nonetheless, it is encouraging
that these approaches do yield similar results.
Acknowledgments The author gratefully acknowledges helpful
comments from Peter Berg, Steven Haider, Stacy Dickert-Conlin,
Dale Belman, Mark Roehling, Stuart Low, and participants at the
2009 Midwest Economics Association and Population Association of
America conferences.
References
1. Committee on Practice and Ambulatory Medicine, Bright Futures
Steering Committee. (2007). Recommendations for preventive
pediatric health care. Pediatrics, 120, 1376.
2. Klerman, J. A., & Leibowitz, A. (1990). Child care and women’s
return to work after childbirth. The American Economic Review,80, 284–288.
3. Hauck, F. R., & Tanabe, K. O. (2008). International trends in
sudden infant death syndrome: Stabilization of rates requires
further action. Pediatrics, 122, 660–666.
4. National Institute of Child Health and Human Development.
(2009). SIDS: ‘Back to Sleep’ Campaign. http://www.nichd.nih.
gov/sids/. Accessed: 10 July 2009.
5. Williams, C. K. N., Harrad, R. A., Sparrow, J. M., & Harvey, I.
(2002). Amblyopia treatment outcomes after screening before or
at age 3 years: Follow up from randomized trial. British MedicalJournal, 324, 1549–1551.
6. Daw, N. W. (1998). Critical periods and amblyopia. Archives ofOphthalmology, 116, 502–505.
7. Kaye, C. I. (2006). Newborn screening fact sheets. Pediatrics,118, 934–963.
8. Hoyert, D. L., Kung, H., & Smith, B. L. (2005). Deaths:
Preliminary data for 2003. National Vital Statistics Reports, 53, 15.
9. Byrd, R. S., Hoekelman, R. A., & Auinger, P. (1999). Adherence
to AAP guidelines for well-child care under managed care.
Pediatrics, 104, 536–540.
10. Maisels, M. J., & Kring, E. (1997). Early discharge from the
newborn nursery–effect on scheduling of follow-up visits by
pediatricians. Pediatrics, 100, 72–74.
11. Ronsaville, D. S., & Hakim, R. B. (2000). Well child care in the
United States: Racial differences in compliance with guidelines.
American Journal of Public Health, 90, 1436–1443.
12. Yu, S. M., Bellamy, H. A., Kogan, M. D., Dunbar, J. L.,
Schwalberg, R. H., & Schuster, M. A. (2002). Factors that
influence receipt of recommended preventive pediatric health and
dental care. Pediatrics, 110.6:1–8.
Table 5 Descriptive analysis of jobs held by mothers who quit work
to care for family
Quit to care
for family
Continuously
employed
% Had paid vacation
in all RCIs employed
43.21% (4.59) 64.61% (1.20)
% Had paid sick leave
in all RCIs employed
33.09% (4.45) 57.56% (1.25)
% FMLA eligible in all
RCIs employed
7.70% (2.69) 35.94% (1.24)
N 172 2,344
Survey weights used to obtain population estimates. Standard errors
in parentheses
Matern Child Health J (2011) 15:1029–1036 1035
123
13. Mustin, H. D., Holt, V. L., & Connell, F. A. (1994). Adequacy
of well-child care and immunizations in U.S. infants born in
1988. The Journal of the American Medical Association, 272,
1111–1115.
14. Moore, P., & Hepworth, J. T. (1994). Use of perinatal and infant
health services by Mexican-American medicaid enrollees. TheJournal of the American Medical Association, 27, 297–304.
15. Colle, A. D., & Grossman, M. (1978). Determinants of pediatric
care utilization. The Journal of Human Resources, 13, 115–158.
16. Vistnes, J. P., & Hamilton, V. (1995). The time and monetary
costs of outpatient care for children. The American EconomicReview, 85, 117–121.
17. Berger, L. M., Hill, J., & Waldfogel, J. (2005). Maternity leave,
early maternal employment and child health and development in
the US. The Economic Journal, 115, F29–F47.
18. Blau, D. M., Guilkey, D. K., & Popkin, B. M. (1996). Infant
health and the labor supply of mothers. The Journal of HumanResources, 31, 90–139.
1036 Matern Child Health J (2011) 15:1029–1036
123
