SlideShare ist ein Scribd-Unternehmen logo
1 von 12
Downloaden Sie, um offline zu lesen
Health Policy 77 (2006) 221–232
Evaluating program effects on institutional delivery in Peru
Michael J. McQuestiona,∗, Anibal Velasquezb
a Johns Hopkins Bloomberg School of Public Health, Population and Family Health Sciences,
615 N Wolfe Street, E-4142, Baltimore, MD 21205, USA
b Consultant, PHRPlus, Antequera 777 Piso 8, San Isidro, Lima, Peru
Abstract
We evaluate the joint effects of two targeted Peruvian health programs on a mother’s choice of whether to deliver in a public
emergency obstetric care (EmOC) facility. The national maternal and child health insurance, or SMI Program, provided delivery
care coverage to Peru’s poorest households beginning in 1998. During 1996–2002, Proyecto 2000 sought to improve the quality
of EmOC and increase utilization of public EmOC facilities in the districts reporting the highest maternal and neonatal mortality
levels. Our data come from the Proyecto 2000 endline evaluation, which sampled 5335 mothers living in the catchment areas
of 29 treatment and 29 matched control EmOC facilities. Using propensity scoring and two quality of care indices, we find
significantly higher quality of care in Proyecto 2000 treatment facilities. Using variance components logistic models, we find a
mother enrolled in the SMI Program was more likely to have delivered her last child in a public EmOC, controlling for household
constraints. Residence in a Proyecto 2000 treatment area did not significantly affect the choice. A cross-level interaction term
was insignificant, indicating the two program effects were independent.
© 2005 Elsevier Ireland Ltd. All rights reserved.
Keywords: Quality of care; Evaluation; Developing countries; Safe motherhood
1. Introduction
This study examines two very different efforts
to increase institutional delivery in Peru. During
1992–1997, Peru implemented large-scale health sec-
tor decentralization reforms. The reforms were criti-
cized for widening health disparities, particularly in
hospital utilization [1]. Peru’s DHS III (1996) and DHS
IV (2000) surveys documented a relative decline in
physician-assisted deliveries among rural and poorly
∗ Corresponding author. Tel.: +1 410 502 6037;
fax: +1 410 955 2303.
E-mail address: mmcquest@jhsph.edu (M.J. McQuestion).
educated women over the period. To correct this, the
Peruvian Ministry of Health initiated a series of tar-
geted maternal and child health interventions, two of
which we evaluate. The first intervention was Proyecto
2000, a USAID-funded effort begun in 1996 in the 12
of Peru’s 25 departmentos reporting the highest mater-
nal mortality levels. Proyecto 2000 aimed to increase
the proportion and quality of institutional deliveries,
thereby reducing maternal mortality and improving
birth outcomes. The project began with mass media,
health education and social mobilization efforts pro-
moting delivery in the nearest public emergency obstet-
ric care (EmOC) facility. Its emphasis, however, was
on improving the quality of services on offer. The sec-
0168-8510/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.healthpol.2005.07.007
222 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232
ond intervention was the Maternal and Child Health
Insurance (SMI) Program, launched in 1998. The SMI
Program covered most maternal and child health costs,
including institutional delivery in public EmOC facil-
ities [2]. It was a means-tested program in that only
households in the poorest wealth quintile were eligi-
ble to participate. By 2000, this program was reaching
about 50% of eligible households in two pilot regions
[3], and the following year it was extended nationwide.
The two programs thus incorporated different target-
ing strategies. Proyecto 2000 targeted high-risk dis-
tritos, specific EmOC facilities and their surrounding
communities while the SMI Program directly targeted
the country’s poorest households. Did either program
increase EmOC utilization? In this study we use quasi-
experimental data to probe this question. We model a
woman’s choice of where she delivered her last baby,
conditional on exposure to these two programs.
2. Background
2.1. Recent perinatal health trends
Demographic data show perinatal health in Peru
improved over this period. The country’s neonatal
death rate fell from 27 to 18 deaths per 1000 live
births during the 1990s [4]. Peru’s estimated mater-
nal mortality ratio also fell, from 265/100,000 live
births in 1990–1996 to 185 in 1994–2000 [5], yet
it remained third highest among 14 Latin American
countries reporting in 1999 [6]. Maternal and neona-
tal mortality are largely influenced by two factors: a
woman’s decision whether or not to utilize institutional
delivery care and the quality of that care. High-quality
EmOC can prevent an estimated one-third of mater-
nal deaths [7], and 40–62% of neonatal deaths [8].
Regardingmaternalbehaviors,the1996DHSIIIsurvey
showed that 55% of women who had given birth in the
previous5yearsdidsoathome.Another38%usedpub-
lic health care facilities and 5% used private delivery
facilities [9]. Over the succeeding 5 years, the propor-
tion of home deliveries fell to 47%, the public sector’s
share rose to 48% and the proportion using private
facilities stayed at about 5% [10]. There are no com-
parable EmOC quality of care estimates, however, a
recent qualitative study ranked Peru second of 13 Latin
American countries evaluated in terms of maternal and
neonatal program effort [11,12]. It is thus plausible that
the observed perinatal health improvements were due
to increased institutional deliveries that in turn resulted
from program improvements.
There were other important factors affecting mater-
nal and perinatal health in Peru over this period. The
country’s per capita GNP grew by a mean 2.4% per
annum during the 1990s [13], an improvement over
the chaotic 1980s. Total fertility rates declined from 4.8
in 1986 to 2.1 in 2000, lengthening birth intervals and
reducing the proportion of high-parity births [4]. These
changing background forces may have been more deci-
sive health behavioral determinants that the program
effects we attempt to elucidate.
2.2. Maternal health risk factors
In Peru, as elsewhere, it is the poorest, most remote
and most socially excluded women who least use
maternal health services [14], and are at highest risk
of maternal, perinatal and post-perinatal mortality
[15,16]. A 2000 survey in Peru’s Ayacucho Depart-
ment, for example, found that only about one-fourth of
women with complications were delivered in adequate
EmOC facilities [17]. In Peru’s DHS IV survey some
83% of women identified at least one barrier to access-
ing local maternal health services. Expense was the
leading problem, followed by lack of female caregivers
[10]. Other cultural factors act as barriers to EmOC uti-
lization, particularly among the 47% of Peruvians who
do not speak Spanish as their first language. Reports
of discrimination and mistreatment by health work-
ers are commonplace [18,19]. The DHS data suggest
that more high-risk women chose to utilize the public
EmOC facilities over this period.
2.3. National SMI Program
TheFujimoriAdministrationinstitutedtheSMIPro-
gram in 1998. It was Peru’s first attempt to subsidize
preventive and maternal care for low-income pregnant
women, mothers and children ages 0–4 years. Many
saw it as an attempt to restore basic health rights that
had been infringed by decentralization. In 2001, the
program was supplanted by a national Integral Health
Insurance Plan, which offered a wider gamut of tar-
geted benefits to low-income Peruvians of all ages.
Until 1998, any woman could have accessed any public
M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 223
EmOC facility where she had to pay fees for service on
a sliding scale. The targeted insurance programs elim-
inated these fees for the eligible poor. However, by
2001 many eligible households were still not enrolled
in the program. Peru’s public health system still lacked
the infrastructure and level of performance needed to
extend MCH services to all those eligible. Production
levels remained exceedingly low. The median num-
ber of consultations that year was less than three per
day in half of the Ministry of Health’s peripheral PHC
facilities. To date there has been no comprehensive
evaluation of these targeted insurance efforts [20].
2.4. Proyecto 2000
2.4.1. Phase I
We describe Proyecto 2000 in greater detail because
it generated the data we analyze. Proyecto 2000 was
implemented by a team of Ministry of Health and exter-
nal expert consultants. The team sought to make the
Ministry’s EmOC services culturally acceptable and to
ensure that the facilities delivered high-quality care. A
hallmark of Proyecto 2000 and other Safe Motherhood
projects is an emphasis on making services “woman-
friendly”. An EmOC facility is woman-friendly if: (a)
it is easily accessible and convenient to use; (b) high-
quality services are offered; (c) local cultural beliefs
and social norms are incorporated into treatment proto-
cols and (d) confidentiality is guaranteed, information
is shared and clients’ choices are respected [21]. The
Proyecto 2000 team worked at facility and commu-
nity levels to accomplish these aims. At baseline, team
members and Regional Ministry of Health educators
gathered and analyzed qualitative data on mothers’
perceptions and preferences regarding pregnancy and
childbirth. They used these data to mount a multime-
dia Safe Motherhood campaign in the treatment areas.
In addition, expert trainers trained 3692 community-
based traditional birth attendants (promotoras), and
EmOC staff formally engaged newly constituted com-
munity health committees (Comites Locales de Admin-
istracion en Salud) in their catchment areas. Facility
inputs included physical plant improvements, retrain-
ing of 409 facility-based providers, incorporation of
local birthing practices into clinical protocols and the
introduction of a continuous quality of care (“autoeval-
uacion”) model in some 89 public hospitals and health
centers. In brief, the autoevaluacion model incorpo-
rated the Donabedian continuous quality of care [22],
and the McCarthy and Maine maternal mortality deter-
minants frameworks [23]. The autoevaluacion instru-
ment included a battery of detailed indicators regarding
essential obstetric and neonatal care, physical facilities,
patient interaction and management. The expectation
was that greater autonomy and participation in the
self-appraisal process would stimulate improved staff
performance, and the resulting improved quality of care
would generate more institutional deliveries as client
satisfaction improved. All facilities were expected to
attain quality of care improvements sufficient to merit
formal certification by expert evaluators. These 89
facilities comprised the original treatment arm.
2.4.2. Midterm evaluation (2000)
As of 1998, 72 treatment facilities were still active
in the program, all of which had attained formal qual-
ity of care certification as high-quality perinatal care
centers [24]. By October 2000, the number of active
treatment facilities had fallen to 60. At that time a
midterm evaluation was carried out. An external evalu-
ation team examined a random sample of 37 treatment
facilities. They also identified a group of 37 similar
EmOC facilities not exposed to the project to serve as
a comparison group. The control facilities were drawn
from six Ministry of Health districts (DISAs) with ser-
vice population characteristics (literacy, contraceptive
prevalence, use of institutional delivery services, mal-
nutrition and poverty levels) similar to the Proyecto
2000 areas. The control facilities had received only rou-
tine Ministry of Health supervision over the period. The
mid-term evaluation was entirely facility-based. Expert
observersusedstandardizedchecklistsandinstitutional
record reviews to assess the quality of EOC on offer.
They found evidence of improved quality of care and a
relative increase in the numbers of institutional deliver-
ies in the treatment group facilities as compared to the
control facilities (Table 1). Additionally, the observers
interviewed samples of prenatal clients. They found
users of treatment facilities were more knowledgeable
about pregnancy, more satisfied with their experiences
and more likely intended to deliver their babies in that
treatment area facility [25].
2.4.3. Phase II (2001–2002)
During Phase II, Proyecto 2000 inputs were con-
centrated on the 31 treatment facilities judged to have
224 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232
Table 1
Selected EOC facility indicators, Proyecto 2000
Variable Control facilities Treatment facilities
Mean S.D. Mean S.D.
1997
Institutional
births
1463 1937 1486 845
Prop births
<2.5 kg
0.07 0.08 0.10 0.09
Prop births
c-section
0.24 0.13 0.19 0.08
2000
Institutional
births
1052 1434 1542 681
Prop births
<2.5 kg
0.07 0.04 0.09 0.07
Prop births
c-section
0.24 0.10 0.23 0.11
Ob-gyns 7.92 7.61 7.50 5.37
Births/ob-gyn 140 59 284 152
2002
Autoevaluacion
scorea
26.80 8.30 51.20 14.40
EmOC capacity
score
69.30 8.90 72.19 7.18
a First principal component of nine-factor index.
performed best in Phase I. Project supervisors regu-
larly visited these facilities to ensure the autoevalua-
ciones were performed in each facility each quarter.
Project data show the autoevaluaciones were in fact
implemented. Of the 29 treatment establishments that
participated to endline (2002), all carried out at least
two autoevaluaciones, 25/29 carried out three, 13/29
carried out four and 3/29 carried out five. The auto-
evaluacion scores reported by the facilities increased
with each round (Fig. 1). These data indicate the institu-
tional Proyecto 2000 interventions were implemented
and suggest the interventions could have been strong
enough to improve the quality of EmOC services on
Fig. 1. Autoevaluacion scores by evaluation round, 2000–2001,
Proyecto 2000.
offer. Our task is thus to disentangle two distinct treat-
ment effects, one operating through the health sys-
tem, the other directly on household health production.
We expect the two effects will be synergetic: insured
women in high-quality EmOC catchment areas ought
to be the most likely to use that facility.
With these points in mind, we model the probability
a Peruvian mother chose to deliver her youngest child
at the nearest public EmOC facility, conditional on the
qualityofcareatthatfacility,herhouseholdconstraints,
SMI Program participation, and whether her commu-
nity and facility participated in Proyecto 2000.
3. Data and methods
3.1. Facility data
The Proyecto 2000 evaluators collected a second
round of endline evaluation data in mid-2002 and it is
these data we analyze in the present paper. The Phase
II treatment group included all 19 Phase I hospitals
and a subset of 12 Phase I health centers. The eval-
uators selected a new control group, consisting of 15
of the Phase I control establishments and 14 additional
establishments. As in Phase I, the 14 new control facili-
ties were purposively selected from six newly matched
DISAs that were unexposed to the project. Expert teams
again evaluated essential obstetric care in the EmOC
facilities using the same extensive standard checklist
used in the midterm evaluation. They also evaluated the
quality of services using the autoevaluacion instrument
itself. Thirdly, they collected selected service indica-
tors routinely reported by each facility to the Ministry
of Health. We used these data to derive two EmOC
quality of care measures, which we described below.
3.2. Household data
To assess changes in local utilization patterns and
measure SMI Program participation, the Proyecto 2000
evaluators carried out a household survey in all treat-
ment and control facility service areas. The survey
instrument incorporated selected items from Peru’s
DHS III and DHS IV survey questionnaires [26],
particularly household characteristics, birth histories
and pregnancy-related behaviors. Sampling procedures
were similar to those used in the DHS. Peru’s 1993 cen-
sus of households provided the sampling frame. Within
M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 225
each Proyecto 2000 catchment area, census tracts were
listed and selected at random. Within each tract house-
holds were selected systematically following cardinal
directions from the approximate center of each cluster.
Ten women who had given birth in the previous 5 years
were surveyed in each cluster. The measures we derive
from these household-level data described below. We
merged the facility and household data to make a hier-
archical dataset consisting of 5335 women nested in
420 clusters in 58 facility catchment areas.
3.3. Facility quality of care measures
Our two quality of care and basic EmOC capac-
ity measures, along with other facility indicators, are
shown in Table 1. The first measure (autevaluacion
score) is the first principal component (eigenvector)
from a factor score analysis of nine items from the
autoevaluacion checklist. The nine items were: blood
is routinely filtered, an incinerator is present, there
are generic versus proprietary drugs are in the phar-
macy, there is an up-to-date list of all drugs dispensed,
patients receive health educational messages, patient
satisfaction is measured, remedial activities to improve
patient satisfaction were implemented, there is a local
community advisory committee, staff meets at least
every 3 months, feedback on performance is given at
thosemeetings.WecomputedCronbach’salphaandthe
Kaiser–Meyer–Olkin measure of sampling adequacy
[27] for these nine items. The resulting coefficients
were, respectively, 0.70 and 0.65 (results not shown).
We conclude the nine items are tapping a common
underlying construct but we note that 0.80 is the con-
ventional “gold standard” for both measures [27].
The second measure, EmOC capacity score, is the
percent score on a battery of 711 items the evaluators
used to assess the technical capacity of a facility to deal
with obstetric emergencies. The evaluators grouped the
indicators into nine categories: human resources, pre-
natal and obstetric equipment, radiology, pharmacy,
delivery room equipment, neonatal care unit, maternity
ward, operating room and blood bank.
As Table 1 shows, treatment facilities scored higher
on both the autoevaluacion quality of care index and
EOC capacity score. This apparent improvement could
be a true difference due to the Proyecto 2000 inputs
or it could be an artifact of the non-random match-
ing of treatment and control facilities, attrition or other
sources of bias. To explore this further we used four
of the routinely reported EmOC facility indicators to
compute a propensity score for the assignment pro-
cess. The aim of propensity scoring is to make assign-
ment “strongly ignorable” by blocking observations
on observables [28,29]. The outcome is the dummy
variable indicating assignment to treatment or control
group. The covariates we used are: number of obstetri-
cians and gynecologists on staff, number of maternal
deathsin2000,numberofcaesariansectionsperformed
in 2000 and the proportion of all deliveries performed
outside of the facility. We generated a balanced score
with matched pairs of facilities falling into eight blocks
(results not shown). We then used the propensity score
to generate three alternative non-parametric treatment
effects estimates for each quality of care measure.
3.4. Household measures
We control for several household risk factors in our
models. Maternal education is a positive predictor of
maternal behaviors in Peru [14]. Other important fac-
tors include maternal age, number of births and socioe-
conomic status [30]. Maternal educational attainment
is coded using terciles, where 1 = no or primary educa-
tion, 2 = some secondary and 3 = completed secondary
and higher. Another dummy variable is coded one for
women who have had three or more live births, zero
otherwise. To control for household wealth we use the
Filmer–Pritchett method [31], wherein weights from
principal components are applied to a list of household
assets, scores are summed and ranked and each house-
hold is assigned to one of five wealth quintiles. We add
additional dummy variables to control for whether the
last child was born in 1998, 1999, 2000 or 2001. We
use a binary dummy variable to indicate whether or not
the household participates in the SMI Program.
As shown in Table 2, the characteristics of Proyecto
2000 sample households were broadly comparable
across treatment and control areas. Only ethnicity var-
ied: treatment area women were less likely to be Span-
ish speakers. Delivery patterns also appear similar
across the study arms. Four of every five women in both
treatment and control areas delivered their last babies in
some kind of institution. Though the matching appears
adequate, the Proyecto 2000 sample is not a nationally
representative sample. Table 2 shows the same indica-
tors computed from Peru’s DHS IV survey. The DHS
226 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232
Table 2
Sample characteristics and maternal health indicators, women giving birth in previous 5 years, Peru 1996–2002
Variable Proyecto 2000 t-Test DHS IV
Control facilities Treatment facilities Matched to P2000 Full sample
Mean S.D. Mean S.D. Mean S.D. Mean S.D.
Last birth institutional 0.83 0.37 0.82 0.38 0.77 0.21 0.53 0.50
Last birth prenatal care 0.91 0.29 0.91 0.29 0.87 0.11 0.78 0.42
Mothers characteristics
Age (years) 26.64 6.62 27.67 7.10 −5.38a 29.32 1.41 29.21 7.05
No. live births 2.57 1.70 2.66 1.88 2.92 0.60 3.51 2.47
Educational level
Primary 0.21 0.40 0.23 0.42 −2.37a 0.31 0.15 0.45 0.50
Secondary 0.44 0.50 0.44 0.50 0.38 0.15 0.32 0.47
Superior 0.35 0.48 0.32 0.47 0.26 0.13 0.14 0.34
Union status
Married 0.38 0.48 0.38 0.49 0.42 0.16 0.41 0.49
Consensual 0.48 0.50 0.47 0.50 0.43 0.16 0.47 0.50
Divorced/separated/widow 0.15 0.36 0.15 0.36 0.08 0.05 0.07 0.25
Rural origin 0.39 0.49 0.36 0.48 0.26 0.19 0.40 0.49
Non-Spanish speaker 0.02 0.15 0.08 0.27 −9.30a 0.10 0.17 0.22 0.42
Households
Electricity 0.93 0.25 0.90 0.30 −4.05a 0.80 0.20 0.52 0.50
Safe water 0.86 0.35 0.81 0.39 4.64a 0.97 0.12 0.81 0.39
Durable floor 0.53 0.50 0.56 0.50 −2.40a 0.44 0.11 0.53 0.50
Safe toilet 0.59 0.49 0.62 0.49 −2.09a 0.75 0.19 0.60 0.49
n 2514 2821 5826 13832
a Significant at p < 0.05 level.
IV sample is a nationally representative weighted sam-
ple drawn from 589 of the 1789 distritos enumerated
in Peru’s 1993 household census. We used the distrito
identifiers to match the DHS IV and Proyecto 2000 data
(n = 68 matched distritos). The Proyecto 2000 sample
is somewhat better educated, more likely to be Spanish-
speaking and living at a slightly higher socioeconomic
level than the DHS subsample from the same distri-
tos. Compared to the national DHS sample, women in
the Proyecto 2000 distritos were more intensive mater-
nal health service users, better educated, more likely to
speak Spanish and less likely to have households with
electricity. Accordingly, all inferences we will make
are limited to the Proyecto 2000 sample data.
3.5. Behavioral model
As mentioned, we estimate a facility-level Proyecto
2000 treatment effect using propensity scoring. Here
we describe our behavioral model, which includes indi-
cator variables that control for the effects of both pro-
grams. We interpret their slopes as indirect treatment
estimates. Given the heteroscedastic treatments and the
many suspected unobserved variables that could have
affectedmothers’deliverychoices,wefittwo-levelran-
dom effects models of the form:
yij = πij + εij
logit(πij) = β0j + β1Xij + β2Tij + β3Iij
+β4Pj + β5IijPj
β0j = δ0j + δ01z1j + κj
εij ∼ N(0, 1), cov(Xij, Pj, Iij, Tij, εij) = 0
κj ∼ N(0, σ2
κ ), cov(z1j, κj) = 0
cov(εij, κj) = 0
In this model πij is the probability mother i in EmOC
facility service area j chose institutional delivery yij,
and εij is an individual error term. β1 is a parame-
M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 227
ter measuring individual effects due to household and
individualcovariatesXij.β2 measurestheeffectoftime,
specified as dummy variables for child i’s birth year Tij.
β3 controls for SMI Program participation, indicated
by Ii, which is coded one for participants, zero for non-
participants. β4 controls for being in a Proyecto 2000
treatment facility area, indicated by Pj, a dummy vari-
able coded one for Proyecto 2000 treatment distritos,
zero otherwise. We include β5 to capture any cross-
level interaction between the two treatments. This term
also adjusts for the possibility the insurance program
was not uniformly implemented across the Proyecto
2000 areas. β0j is a random facility-level intercept,
δ0j and δ01 are parameters, z1j is a dummy variable
for facility and kj is a facility-level random effect. If
the variance of kj, denoted as σ2
k , is significant, then
we know there are unobserved variable effects which
might otherwise have biased the fixed effect parameters
in a conventional model.
To fit the behavioral model we must make several
assumptions. We assume that each mother is influenced
solely by her own EmOC facility. We further assume
that all mothers in the Proyecto 2000 treatment areas
were equally exposed to the treatments and that access
to the nearest Ministry of Health EmOC facility did not
differ between treatment and control areas. This was
not the case for the SMI Program, which was not uni-
formly implemented and was means-tested. Although
we lack any SMI Program data, we assume that all eli-
gible households with access did enroll. Because the
subsidy was conditioned on use of the nearest Ministry
EmOC facility we assume that any SMI Program effect
contributed fully to the likelihood of our outcome.
We estimate a series of nested multilevel models
using Stata’s gllamm program [32]. The program uses
a maximum likelihood algorithm with adaptive quadra-
ture to model latent variables as random effects. One
advantage of gllamm over other multilevel programs
is that it generates log-likelihood statistics useful for
comparing model fits. All standard errors are estimated
using the Huber–White sandwich estimator to adjust
for the clustered survey design effect [33].
4. Results
4.1. Institutional model
Our institutional treatment effects are shown in
Table 3. The estimates include population-average
treatment effects (ATE) produced by radius matching
and nearest neighbor matching algorithms. Following
Imbens (2003), we also estimate the within-sample
ATE. For comparison, we report the slope of a simple
OLS model with the treatment dummy the sole regres-
sor. The two indicators are measured on different met-
rics so their treatment effect estimates are not directly
comparable. What we expect are consistent estimates
for each indicator. Inferences are based on Wald tests.
With the outcome the EmOC capacity score, popula-
tion and sample ATE estimates were all significant.
With the autoevaluacion factor score as the outcome,
Table 3
Facility-level treatment effects estimates, conditioned on propensity scorea, Proyecto 2000
EmOC capacity (n = 52 facilities) Autoevaluation (n = 55 facilities)
Coefficient S.D. (Pairs) Coefficient S.D. (Pairs)
OLS slope 12.2** 2.8 0.55** 0.28
Radius matching ATTb 11.7** 2.6 (26t,18c) 0.5 0.43 (6t,16c)
Nearest neighbor
Random draw ATT 12.1** 2.7 (26t,11c) 0.55* 0.32 (26t,11c)
Equal weights ATT 12.1** 2.7 (26t,11c) 0.55 0.36 (26t,11c)
Sample ATEc 12.8** 2.9 (43t,43c) 0.65** 0.27 (n = 41t,41c)
a Propensity score variables: number of ob-gyns, no. maternal deaths 2000, no. caesarian sections 2000, proportion of all cases delivered in
facility in 2000.
b ATT, average treatment effect on the treated.
c ATE, average treatment effect.
* Significant at 0.10 < p < 0.05 level.
** Significant at p < 0.05 level.
228 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232
we find a significant sample ATE (0.65) but only one
populationaveragetreatmenteffect,anditismarginally
significant. The EmOC capacity score is apparently a
more sensitive quality of care measure. We conclude
the estimates are robust and that Proyecto 2000 inputs
did improve the quality of care in the EmOC treatment
facilities. Our main interest, however, is in measuring
any health behavioral impacts and assessing whether
they are linked to facility quality of care improvements,
to the provision of MCH insurance or a combination of
the two.
4.2. Behavioral model
Our behavioral model results are shown in Table 4.
We show exponentiated slopes (odds ratios) to ease
interpretation. In Model 1, the reference household
model, covariate effects are signed as expected. The
more educated and wealthier the woman, the more
likely she delivered in the EmOC facility. Those who
do not speak Spanish and have had three or more live
births are less likely to choose institutional delivery.
The dummy variables for birth years 1999, 2000 and
2001 capture unmeasured variables that are associated
with EmOC delivery. Those net effects are positive
compared with those in 1997 and 1998, the omitted
categories. There is a significant random effect, indi-
cating that women’s decisions to deliver in the public
EmOC facility are correlated in some catchment areas
more than in others due to omitted variables that jointly
affect their behaviors.
Model 2 results show that living in a Proyecto 2000
area has no significant effect on delivery choice. Model
3, in contrast, shows that the odds of institutional deliv-
ery for women covered by the SMI Program were twice
the odds for women not covered. Controlling for insur-
ance removes upward biases on the highest education
and wealth dummies. The difference in log-likelihoods
shows that Model 3 is also a significantly better-fitting
model than Models 1 or 2. Model 4 includes an inter-
action term between the highest wealth quintile and
the insurance dummies. These better-off households
were ineligible for coverage and the negative interac-
tion term captures this fact. Controlling this interac-
tion further decreases the direct effects of being in the
wealthiest quintile. The most dramatic effect, however,
is a seven-fold increase the odds of EmOC delivery for
the insured women. This pattern is consistent with the
fact only the poorest households were eligible for the
SMI Program. In Model 5 we add a cross-level interac-
Table 4
Two-level logistic regression delivery models, exponentiated effects, Proyecto 2000
Variablea Coefficient (S.E.)
Model 1 Model 2 Model 3 Model 4 Model 5
Non-Spanish speaker 0.38** (0.07) 0.38** (0.07) 0.36** (0.07) 0.36** (0.07) 0.36** (0.07)
Some secondary education 2.69** (0.27) 2.69** (0.27) 2.59** (0.26) 2.61** (0.27) 2.61** (0.27)
Complete secondary or more 5.63** (0.83) 5.63** (0.83) 4.81** (0.72) 4.75** (0.71) 4.76** (0.71)
Three or more live births 0.66** (0.06) 0.66** (0.06) 0.64** (0.06) 0.64** (0.06) 0.64** (0.06)
60–79th wealth quintile 2.23** (0.31) 2.24** (0.31) 2.10** (0.30) 2.13** (0.30) 2.13** (0.30)
80–100th wealth quintile 3.26** (0.61) 3.27** (0.62) 2.86** (0.54) 2.25** (0.46) 2.26** (0.46)
Born 1999 1.30** (0.15) 1.30** (0.15) 1.29** (0.15) 1.29** (0.15) 1.29** (0.15)
Born 2000 1.75** (0.22) 1.75** (0.22) 1.76** (0.22) 1.76** (0.22) 1.76** (0.22)
Born 2001 1.62** (0.20) 1.62** (0.20) 1.64** (0.20) 1.63** (0.20) 1.64** (0.20)
Insured 2.02** (0.25) 15.71** (14.52) 14.96** (13.98)
Insured 80–100th quintile* 0.34** (0.16) 0.34** (0.16)
P2000 treatment area 0.79 (0.25) 0.80 (0.26)
P2000 area insured* 1.08 (0.26)
Level-two random effect σ2
k 1.06** (0.27) 1.05** (0.26) 1.11** (0.28) 1.00** (0.28) 1.09** (0.24)
Log-likelihood −1797.4 −1797.1 −1780.1 −1777.0 −1776.8
n 5190 5190 5190 5190 5190
a Omitted categories: education secondary and beyond, Spanish speaker, one or two live births, lowest three wealth quintiles, born 1998, no
insurance.
* Significant at 0.10 < p < 0.05 level.
** Significant at p < 0.05 level.
M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 229
Fig. 2. Real and simulated posterior means Models 1–5.
tion term to test whether the two programs interacted to
affect EmOC probabilities. The interaction is insignif-
icant.
Comparisons of the models with their conventional
single-equation logit analogs show that, in each case,
the random effects specification improved model fit
(results not shown). To check whether the normality
assumption was met, we standardized and plotted the
posterior means from each model. All distributions
were near normal but somewhat negatively skewed.
The skewness was due to three clinics (two treatment,
one control) whose means were more than two standard
deviations below the sample means. We used gllamm’s
post-estimation command gllasim to resample the pos-
teriormeans.ResultsareshowninFig.2.Thesimulated
distributions were normal and no longer skewed. Fol-
lowing arguments by Longford [34], the three clinics
are thus not true outliers; their apparent outlier status is
a feature of the realized sample. We leave them in the
dataset and conclude that the models are robust.
5. Discussion and conclusions
5.1. Project impacts
Our results show that Proyecto 2000 improved the
quality of care on offer but did not directly increase the
probability of delivery in Ministry of Health EmOC
facilities. Nor was there an interaction between the
system-level Proyecto 2000 inputs and the household-
level SMI Program. Though they targeted the same
sub-population, each program operated independently.
The only behavioral impact we document is that of the
SMI Program. It shows, simply, that reducing out-of-
pocket costs increases EmOC utilization. The poorest
Peruvian women clearly benefited from the targeted
insurance program, however, the household risk fac-
tor effects remained consistently negative across the
models, indicating that neither program significantly
reduced socioeconomic or ethnic disparities in EmOC
utilization.
Behavioral impacts due to Proyecto 2000 may have
been too weak to be detectable or may have occurred
after the endline survey. As shown in the DHS data,
the share of births delivered in Ministry EmOC
facilities rose nationwide during the period. Looking
at our sample, we also see increasingly positive
period effects, represented by the slopes on the birth
year dummies in our models. The forces propelling
those increases were likely more decisive than any
attributable to Proyecto 2000. A lagged Proyecto 2000
treatment effect would be plausible for two reasons.
First, only about 40% of the women surveyed gave
birth during the project’s most intensive second phase
(2000–2002). Although the dummy variables for
230 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232
birth year capture a rising probability of EmOC use,
the majority of women interviewed may have been
unaware of any local improvements when they made
their birthing decisions, or any improvements made
may not have been noticeable. Second, delivery behav-
iors may be socially mediated. If so, the observation
period may have been too short for social learning
or other endogenous social processes to reach some
theoretical threshold level of women. The data did
not permit us to test this hypothesis, however, social
forces are one possible source of the consistent cor-
relation of birthing decisions within catchment areas
captured by the random effects. Future studies would
do well to explore these social aspects of maternal
behaviors.
5.2. Limitations
There are a number of methodological shortcomings
in this study. The relatively rich quasi-experimental
Proyecto 2000 data allowed us to estimate a treat-
ment effect for that program. However, we lacked
any kind of design for evaluating the SMI Program.
Strong designs are needed in order to evaluate such tar-
geted programs. A recent example was Gertler’s 2000
[35] evaluation of Mexico’s Progresa Program. In that
study,Gertleruseddifference-in-differencesestimators
and panel data from households in randomly sam-
pled treatment and control areas to show the program
increased school enrollment and health services utiliza-
tion and improved health outcomes. Had panel data
or even repeated cross-sectional data from the same
catchment areas been available we might have detected
household-level Proyecto 2000 treatment effects.
Our study also faced obvious sampling problems.
Attrition of the Proyecto 2000 facilities during Phase
I and the replacement of 14 of the original control
facilities with new ones at endline are likely sources
of sample selection bias. If the attriting EmOC facil-
ities were the stronger institutions then any treatment
effect would be underestimated. We lacked the data
necessary to assess this. The targeted nature of the two
programs presents another potential source of bias in
that the characteristics of facilities and households not
given treatment are likely to differ from those that did
receive treatments. We estimated Proyecto 2000 treat-
ment effects using a propensity score balanced on just
four observable covariates; many other, unmeasured
covariates could differ systematically across the two
groups. Regarding the SMI Program, the beneficiaries
we observed may differ from other potential beneficia-
ries in Proyecto 2000 catchment areas where the insur-
ance program had not yet been implemented. A more
general problem are background disturbances caused
by the constantly evolving mix of EmOC services many
Peruvian communities faced as public health services
decentralized and to some extent recentralized. In this
fluid policy environment, perceptions of EmOC qual-
ity, perhaps the legitimacy of public health services in
general, were in flux.
5.3. Policy implications
Peru’s SMI Program proved an effective means of
inducing high-risk women to use public EmOC facil-
ities. We document here its short-term impacts. They
show that cost is a significant barrier to many women.
However, such subsidized programs are generally fis-
cally unsustainable, particularly in poor countries. Fur-
ther, they may not be efficacious. The subsidies could
merely act as side payments for compliance and when
the subsidies end, the desired behavior, here use of
EmOC, may end too. The long-term sustainability
of targeted subsidy programs is an area where more
research is needed.
Proyecto 2000 sought to induce greater EmOC uti-
lization through more elaborate, technical strategies. It
theorized that improving institutional quality of care,
educating the public and working with communities
would be sufficient to induce behavioral change. We
lacked data on the latter but the data we did have
showed the first goal was achieved. Improved quality,
our results suggest, is not sufficient to change delivery
behaviors. Something else is needed. Recently, Gilson
[36] proposed a theory wherein trust, initially between
client and provider and later between community and
the state, is a necessary condition for communities to
become healthier. For this to happen people must per-
ceive the quality of care to be high and the public health
services to be legitimate. If out-of-pocket cost is a bar-
rier, then targeted subsidies may be warranted as an
interim measure. Studies elsewhere have shown even
the poorest people are willing to pay for health ser-
vices they value [37,38]. Though our interaction term
was insignificant, we encourage other researchers to
test this hypothesis.
M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 231
Acknowledgements
Technical assistance for Proyecto 2000 was pro-
vided by Pathfinder International with funding from
USAID-Peru, Contract 527-0366-C5049-00. Support
for our study was provided by the Bill and Melinda
Gates Institute for Population and Reproductive
Health, Department of Population and Family Health
Sciences, Johns Hopkins Bloomberg School of Public
Health.
References
[1] Arroyo-Laguna J. Greater autonomy for public hospitals in
Lima and the consequences of that on the use of health ser-
vices,1988–1997.PanAmericanJournalofHealth1999;6:301–
210.
[2] Ministry of Health. Lineamientos de pol´ıticas del sector salud
1995–2000: hacia un sector salud con eficiencia, equidad y cal-
idad. Lima: Ministerio de Salud; 1996.
[3] Pan American Health Organization. Core select data indicators.
Peru, 2004.
[4] MEASURE DHS+. Country statistics: Peru. Demographic and
Health Surveys, 2004. Website: http://www.measuredhs.com/
countries/country.cfm?ctry id=33 accessed 4/2004.
[5] Hill K, Abou-Zahr C, Wardlaw T. Estimates of maternal mor-
tality for 19/95. Bulletin of the World Health Organization
2001;79:182–93.
[6] Pan American Health Organization. Regional strategy for
maternal mortality and morbidity reduction, Document
CSP26/14. Washington, DC: Pan American Health Organiza-
tion; 2002.
[7] Murray SF, Davies S, Kumwenda Phiri R, Ahmed Y. Tools
for monitoring the effectiveness of district maternity referral
systems. Health Policy and Planning 2001;16(4):353–61.
[8] World Health Organization. Mother-baby package. Implement-
ing safe motherhood in countries. Document FHE/MSM/94.11.
Maternal health and safe motherhood programme. Geneva:
World Health Organization; 1994.
[9] Instituto Nacional de Estadistica e Informatica. Encuesta
demografica y de salud familiar 1996. Lima: Instituto Nacional
de Estadistica e Informatica, 1996. Accessed at: http://www.
paho.org/English/SHA/coredata/tabulator/newTabulator.htm
23 October 2004.
[10] Instituto Nacional de Estadistica e Informatica. Encuesta
demografica y de salud familiar 2000. Lima: Instituto Nacional
de Estadistica e Informatica; 2000.
[11] Ross JA, Campbell OMR, Bulatao R. The maternal and neona-
tal program effort index. Tropical Medicine and International
Health 2001;6:787–98.
[12] Futures Group. Herramienta para la incidencia politica en la
salud materna: MNPI (indice de esfuerzo de programa materno
y neonatal). Lima: Futures Group; 2002.
[13] United Nations Development Programme. Human Develop-
ment Reports: Human Development Indicators, 2003. Website:
http://www.undp.org/hdr2003/indicator/cty f PER.html
accessed 4/2004.
[14] Elo IT. Utilization of maternal health-care services in Peru:
the role of women’s education. Health Transition Review
1992;2:49–69.
[15] Vallenas G. Maternal mortality in Peru. Revista Peruana
de Poblacion 1993;3:33–56. Website: http://www.paho.org/
English/DD/AIS/cp 604.htm accessed 4/2004.
[16] de Meer K, Bergman R, Kusner JS. Socio-cultural determi-
nants of child mortality in southern Peru: including some
methodological considerations. Social Science and Medicine
1993;36:317–31.
[17] Averting Maternal Death and Disability Working Group on
Indicators. Using UN process indicators to assess needs in emer-
gency obstetric services: Pakistan, Peru and Vietnam. Inter-
national Journal of Gynaecology and Obstetrics 2002;78:275–
82.
[18] Meentzen A. Estrategias de desarrollo culturalmente adecuadas
para mujeres indigenas. Unpublished document. Washington,
DC: Unidad de Pueblos Indigenas y Desarrollo Comunitario.
Banco Interamericano de Desarrollo, 2000.
[19] Pan American Health Organization. Situacion de salud de los
pueblos indigenas de Peru. Document OPS/HSP/HSO/98.08.
Washington, DC: Pan American Health Organization; 1998.
[20] Francke BP, editor. Politicas de Salud 2001–2006. Diagnos-
tico y Propuesta 5. Lima: Consorcio de investigacion eco-
nomica y social, 2001. Available at: http://www.consorcio.org/
CIES/html/diag5.asp.
[21] Donnay F. Maternal survival in developing countries: what has
been done, what can be achieved in the next decade. Interna-
tional Journal of Gynaecology and Obstetrics 2000;70:89–97.
[22] Donabedian A. The definition of quality and approaches to
its assessment. Ann Arbor, Michigan: Health Administration
Press; 1980.
[23] McCarthy J, Maine D. A framework for analysing the deter-
minants of maternal mortality. Studies in Family Planning
1992;23:23–33.
[24] Benavides B, Seclen JE, Novata J, Velasquez A. Impacto del
mejoramiento continuo de la calidad en los servicios materno
perinatales del Peru. La experiencia del Proyecto 2000. Unpub-
lished document. Lima: Pathfinder Fund, 2000.
[25] Seclen J, Benavides B, Jacoby E, Velasquez A, Watanabe E.
Existe una relacion entre los programas de mejora de la calidad
y la satisfaccion de usuarias de atencion prenatal? Experiencia
en hospitals de Peru. Revista Panamericana de Salud Publica
2004;16:149–57.
[26] EASURE DHS+. Peru 2000 DHSIV Final Report. FR-120.
Calverton, MD, 2000.
[27] Cureton EE, D’Agostino RB. Factor analysis: an applied
approach. Hillsdale, NJ: Erlbaum; 1983.
[28] Rosenbaum PR, Rubin DB. The central role of the propensity
score in observational studies for causal effects. Biometrika
1983;70:41–55.
[29] Imben G. Nonparametric estimation of average treatment
effects under exogeneity: a review. Technical Working Paper
232 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232
293. Cambridge MA: National Bureau of Economic Research,
2003. http://www.nber.org/papers/T0294.
[30] AbouZahr CL, Wardlaw T. Antenatal care in developing
countries. Promises, achievements and missed opportunities.
Geneva: World Health Organization; 2003.
[31] Filmer D, Pritchett L. Estimating wealth effects without
expenditure data or tears: an application to educational
enrollments in states of India. Demography 2001;38:115–
32.
[32] Rabe-Hesketh S, Skrondal A, Pickles A. GLLAMM manual.
Paper 160. UC Berkeley Division of Biostatistics Working
Paper Series. The Berkeley Electronic Press, 2004. http://www.
bepress.com/ucbbiostat/paper160.
[33] Stata Corporation. Stata statistical software. Release 6.0. Col-
lege Station, TX: Stata Corporation; 1999.
[34] Longford NL. Simulation-based diagnostics in random-
coefficient models. Journal of the Royal Statistical Society A
2001;164(Part 2):259–73.
[35] Gertler P. Final report. The impact of Progresa on health.
Unpublished paper. International Food Policy Research Insti-
tute. Washington, DC: International Food Policy Research Insti-
tute, 2000.
[36] Gilson L. Trust and the development of health care as a social
institution. Social Science and Medicine 2003;56:1453–68.
[37] Acharya L, Cleland J. Maternal and child health services in rural
Nepal: does access or quality matter more? Health Policy and
Planning 2000;15:223–9.
[38] Ross G, Zeballos J, Infante A. La calidad y la reforma del sector
salud en America Latina y el Caribe. Revista Panamericana de
Salud Publica 2002;9:93–7.

Weitere ähnliche Inhalte

Was ist angesagt?

Leveraging farmer field days to provide family planning and other health serv...
Leveraging farmer field days to provide family planning and other health serv...Leveraging farmer field days to provide family planning and other health serv...
Leveraging farmer field days to provide family planning and other health serv...Alexander Decker
 
National health programmes related to child health and welfare
National health programmes related to child health and welfareNational health programmes related to child health and welfare
National health programmes related to child health and welfareUniversity of Hyderabad
 
Population control and related population control programme
Population control and related population control programmePopulation control and related population control programme
Population control and related population control programmePinki Barman
 
Reproductive and child health program
Reproductive and child health programReproductive and child health program
Reproductive and child health programNabil Garry
 
State of Maternal and Children's Health and Nutrition During Pandemic and Cal...
State of Maternal and Children's Health and Nutrition During Pandemic and Cal...State of Maternal and Children's Health and Nutrition During Pandemic and Cal...
State of Maternal and Children's Health and Nutrition During Pandemic and Cal...KABAYAN Partylist
 
2020 global nutrition_report
2020 global nutrition_report2020 global nutrition_report
2020 global nutrition_reportCecilia Acuin
 
Family welfare programme Dr.chetan
Family welfare programme Dr.chetanFamily welfare programme Dr.chetan
Family welfare programme Dr.chetanDrChetanSharma5
 
Evolution of National Family Planning Programme (NFPP) and National Populatio...
Evolution of National Family Planning Programme (NFPP) and National Populatio...Evolution of National Family Planning Programme (NFPP) and National Populatio...
Evolution of National Family Planning Programme (NFPP) and National Populatio...Dr Kumaravel
 
National health mission (NHM)
National health mission (NHM)National health mission (NHM)
National health mission (NHM)anjalatchi
 
Strategies to strengthen Mission Indradhanush
Strategies to strengthen Mission IndradhanushStrategies to strengthen Mission Indradhanush
Strategies to strengthen Mission Indradhanushshayonisen2012
 
Stakeholder Consultation: State of Maternal Health and Nutrition in the PH
Stakeholder Consultation: State of Maternal Health and Nutrition in the PHStakeholder Consultation: State of Maternal Health and Nutrition in the PH
Stakeholder Consultation: State of Maternal Health and Nutrition in the PHKABAYAN Partylist
 
The Philippine Family Planning Program (DOCX)
The Philippine Family Planning Program (DOCX)The Philippine Family Planning Program (DOCX)
The Philippine Family Planning Program (DOCX)Ma Elena Oblino Abainza
 
RCH: Focus on lab investigations
RCH: Focus on lab investigationsRCH: Focus on lab investigations
RCH: Focus on lab investigationsVignesh Loganathan
 
Primary health care by Harsh RAstogi
Primary health care by Harsh RAstogiPrimary health care by Harsh RAstogi
Primary health care by Harsh RAstogiHarsh Rastogi
 

Was ist angesagt? (20)

Leveraging farmer field days to provide family planning and other health serv...
Leveraging farmer field days to provide family planning and other health serv...Leveraging farmer field days to provide family planning and other health serv...
Leveraging farmer field days to provide family planning and other health serv...
 
National health programmes related to child health and welfare
National health programmes related to child health and welfareNational health programmes related to child health and welfare
National health programmes related to child health and welfare
 
Population control and related population control programme
Population control and related population control programmePopulation control and related population control programme
Population control and related population control programme
 
Rch
RchRch
Rch
 
Reproductive and child health program
Reproductive and child health programReproductive and child health program
Reproductive and child health program
 
State of Maternal and Children's Health and Nutrition During Pandemic and Cal...
State of Maternal and Children's Health and Nutrition During Pandemic and Cal...State of Maternal and Children's Health and Nutrition During Pandemic and Cal...
State of Maternal and Children's Health and Nutrition During Pandemic and Cal...
 
2020 global nutrition_report
2020 global nutrition_report2020 global nutrition_report
2020 global nutrition_report
 
Family welfare programme Dr.chetan
Family welfare programme Dr.chetanFamily welfare programme Dr.chetan
Family welfare programme Dr.chetan
 
krithiga rmnch
 krithiga rmnch krithiga rmnch
krithiga rmnch
 
Evolution of National Family Planning Programme (NFPP) and National Populatio...
Evolution of National Family Planning Programme (NFPP) and National Populatio...Evolution of National Family Planning Programme (NFPP) and National Populatio...
Evolution of National Family Planning Programme (NFPP) and National Populatio...
 
Motherhood method 12 9-13
Motherhood method 12 9-13Motherhood method 12 9-13
Motherhood method 12 9-13
 
National health mission (NHM)
National health mission (NHM)National health mission (NHM)
National health mission (NHM)
 
Strategies to strengthen Mission Indradhanush
Strategies to strengthen Mission IndradhanushStrategies to strengthen Mission Indradhanush
Strategies to strengthen Mission Indradhanush
 
Stakeholder Consultation: State of Maternal Health and Nutrition in the PH
Stakeholder Consultation: State of Maternal Health and Nutrition in the PHStakeholder Consultation: State of Maternal Health and Nutrition in the PH
Stakeholder Consultation: State of Maternal Health and Nutrition in the PH
 
The Philippine Family Planning Program (DOCX)
The Philippine Family Planning Program (DOCX)The Philippine Family Planning Program (DOCX)
The Philippine Family Planning Program (DOCX)
 
Family planning
Family planningFamily planning
Family planning
 
RCH
RCHRCH
RCH
 
RCH: Focus on lab investigations
RCH: Focus on lab investigationsRCH: Focus on lab investigations
RCH: Focus on lab investigations
 
Primary health care by Harsh RAstogi
Primary health care by Harsh RAstogiPrimary health care by Harsh RAstogi
Primary health care by Harsh RAstogi
 
Rch programme
Rch programmeRch programme
Rch programme
 

Andere mochten auch

Pic Nic - abstract
Pic Nic - abstractPic Nic - abstract
Pic Nic - abstractSplit Eri
 
Foursquare- how to
Foursquare- how toFoursquare- how to
Foursquare- how toSplit Eri
 
Turn IT On Case Study
Turn IT On Case StudyTurn IT On Case Study
Turn IT On Case StudyNiki Dinsey
 
Ritual as the behavioral side of sanctification
Ritual as the behavioral side of sanctificationRitual as the behavioral side of sanctification
Ritual as the behavioral side of sanctificationPaweł Socha
 

Andere mochten auch (7)

Pic Nic - abstract
Pic Nic - abstractPic Nic - abstract
Pic Nic - abstract
 
Foursquare- how to
Foursquare- how toFoursquare- how to
Foursquare- how to
 
Trudy
TrudyTrudy
Trudy
 
Turn IT On Case Study
Turn IT On Case StudyTurn IT On Case Study
Turn IT On Case Study
 
Hdfc
HdfcHdfc
Hdfc
 
Intro to E-Rate
Intro to E-RateIntro to E-Rate
Intro to E-Rate
 
Ritual as the behavioral side of sanctification
Ritual as the behavioral side of sanctificationRitual as the behavioral side of sanctification
Ritual as the behavioral side of sanctification
 

Ähnlich wie Health policy program effects on institutional delivery in peru

reproductive and child health.docx
reproductive and child health.docxreproductive and child health.docx
reproductive and child health.docxSnehlata Parashar
 
reproductive and child health.docx
reproductive and child health.docxreproductive and child health.docx
reproductive and child health.docxSnehlata Parashar
 
national health progrmmes for children.pptx
national health progrmmes for children.pptxnational health progrmmes for children.pptx
national health progrmmes for children.pptxpayalgakhar
 
Reproductive and child health program
Reproductive and child health programReproductive and child health program
Reproductive and child health programHarsh Rastogi
 
reproductiveandchildhealthprogram-180721092426 (1).pdf
reproductiveandchildhealthprogram-180721092426 (1).pdfreproductiveandchildhealthprogram-180721092426 (1).pdf
reproductiveandchildhealthprogram-180721092426 (1).pdfJaenaVeronicaDimapil
 
reproductiveandchildhealthprogram-180721092426.pdf
reproductiveandchildhealthprogram-180721092426.pdfreproductiveandchildhealthprogram-180721092426.pdf
reproductiveandchildhealthprogram-180721092426.pdffogger2
 
Chapter 7Maternal, Infant, and Child HealthChapter Objec
Chapter 7Maternal, Infant, and Child HealthChapter ObjecChapter 7Maternal, Infant, and Child HealthChapter Objec
Chapter 7Maternal, Infant, and Child HealthChapter ObjecJinElias52
 
Nationalhealthprogrammes 130905012943-
Nationalhealthprogrammes 130905012943-Nationalhealthprogrammes 130905012943-
Nationalhealthprogrammes 130905012943-Raj Akhani
 
pdf national health programes .pdf
pdf national health programes .pdfpdf national health programes .pdf
pdf national health programes .pdfHananDar3
 
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...iosrphr_editor
 
Ssnc rch program for nursing students .pptx
Ssnc rch program for nursing students .pptxSsnc rch program for nursing students .pptx
Ssnc rch program for nursing students .pptxMilanHathaliya1
 
rch program in community health nursing.pptx
rch program in community health nursing.pptxrch program in community health nursing.pptx
rch program in community health nursing.pptxMilanAhir9
 
Reproductive and Child Health Services ppt.pptx
Reproductive and Child Health Services ppt.pptxReproductive and Child Health Services ppt.pptx
Reproductive and Child Health Services ppt.pptxVeereshDemashetti
 
Achieving polio eradication a review of helth communication evidence and le...
Achieving polio eradication   a review of helth communication evidence and le...Achieving polio eradication   a review of helth communication evidence and le...
Achieving polio eradication a review of helth communication evidence and le...Dr Lendy Spires
 
An Interrupted Time Series Multivariate Regression Analysis Evaluation of Sta...
An Interrupted Time Series Multivariate Regression Analysis Evaluation of Sta...An Interrupted Time Series Multivariate Regression Analysis Evaluation of Sta...
An Interrupted Time Series Multivariate Regression Analysis Evaluation of Sta...Whitney Bowman-Zatzkin
 
Nhp related to chn
Nhp related to chnNhp related to chn
Nhp related to chnPdianghun
 
innovations-in-nursing.9701471.powerpoint.pptx
innovations-in-nursing.9701471.powerpoint.pptxinnovations-in-nursing.9701471.powerpoint.pptx
innovations-in-nursing.9701471.powerpoint.pptxPreethi113391
 

Ähnlich wie Health policy program effects on institutional delivery in peru (20)

reproductive and child health.docx
reproductive and child health.docxreproductive and child health.docx
reproductive and child health.docx
 
reproductive and child health.docx
reproductive and child health.docxreproductive and child health.docx
reproductive and child health.docx
 
Family welfare programme
Family welfare programmeFamily welfare programme
Family welfare programme
 
Family welfare programme
Family welfare programmeFamily welfare programme
Family welfare programme
 
national health progrmmes for children.pptx
national health progrmmes for children.pptxnational health progrmmes for children.pptx
national health progrmmes for children.pptx
 
Mch services
Mch servicesMch services
Mch services
 
Reproductive and child health program
Reproductive and child health programReproductive and child health program
Reproductive and child health program
 
reproductiveandchildhealthprogram-180721092426 (1).pdf
reproductiveandchildhealthprogram-180721092426 (1).pdfreproductiveandchildhealthprogram-180721092426 (1).pdf
reproductiveandchildhealthprogram-180721092426 (1).pdf
 
reproductiveandchildhealthprogram-180721092426.pdf
reproductiveandchildhealthprogram-180721092426.pdfreproductiveandchildhealthprogram-180721092426.pdf
reproductiveandchildhealthprogram-180721092426.pdf
 
Chapter 7Maternal, Infant, and Child HealthChapter Objec
Chapter 7Maternal, Infant, and Child HealthChapter ObjecChapter 7Maternal, Infant, and Child HealthChapter Objec
Chapter 7Maternal, Infant, and Child HealthChapter Objec
 
Nationalhealthprogrammes 130905012943-
Nationalhealthprogrammes 130905012943-Nationalhealthprogrammes 130905012943-
Nationalhealthprogrammes 130905012943-
 
pdf national health programes .pdf
pdf national health programes .pdfpdf national health programes .pdf
pdf national health programes .pdf
 
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
 
Ssnc rch program for nursing students .pptx
Ssnc rch program for nursing students .pptxSsnc rch program for nursing students .pptx
Ssnc rch program for nursing students .pptx
 
rch program in community health nursing.pptx
rch program in community health nursing.pptxrch program in community health nursing.pptx
rch program in community health nursing.pptx
 
Reproductive and Child Health Services ppt.pptx
Reproductive and Child Health Services ppt.pptxReproductive and Child Health Services ppt.pptx
Reproductive and Child Health Services ppt.pptx
 
Achieving polio eradication a review of helth communication evidence and le...
Achieving polio eradication   a review of helth communication evidence and le...Achieving polio eradication   a review of helth communication evidence and le...
Achieving polio eradication a review of helth communication evidence and le...
 
An Interrupted Time Series Multivariate Regression Analysis Evaluation of Sta...
An Interrupted Time Series Multivariate Regression Analysis Evaluation of Sta...An Interrupted Time Series Multivariate Regression Analysis Evaluation of Sta...
An Interrupted Time Series Multivariate Regression Analysis Evaluation of Sta...
 
Nhp related to chn
Nhp related to chnNhp related to chn
Nhp related to chn
 
innovations-in-nursing.9701471.powerpoint.pptx
innovations-in-nursing.9701471.powerpoint.pptxinnovations-in-nursing.9701471.powerpoint.pptx
innovations-in-nursing.9701471.powerpoint.pptx
 

Mehr von Anibal Velasquez

Epidemiologia y administracion de salud
Epidemiologia y administracion de saludEpidemiologia y administracion de salud
Epidemiologia y administracion de saludAnibal Velasquez
 
El plan de evaluación y seguimiento de un proyecto
El plan de evaluación y seguimiento de un proyectoEl plan de evaluación y seguimiento de un proyecto
El plan de evaluación y seguimiento de un proyectoAnibal Velasquez
 
Revisión sistemática sobre el impacto de cocinas mejoradas
Revisión sistemática sobre el impacto de cocinas mejoradasRevisión sistemática sobre el impacto de cocinas mejoradas
Revisión sistemática sobre el impacto de cocinas mejoradasAnibal Velasquez
 
Recursos y utilitarios para evaluadores
Recursos y utilitarios para evaluadoresRecursos y utilitarios para evaluadores
Recursos y utilitarios para evaluadoresAnibal Velasquez
 
Gestion de la evidencia. Un nuevo enfoque de los sistemas de seguimiento y ev...
Gestion de la evidencia. Un nuevo enfoque de los sistemas de seguimiento y ev...Gestion de la evidencia. Un nuevo enfoque de los sistemas de seguimiento y ev...
Gestion de la evidencia. Un nuevo enfoque de los sistemas de seguimiento y ev...Anibal Velasquez
 
El Instituto Nacional de Salud de Perú 2010
El Instituto Nacional de Salud de Perú 2010El Instituto Nacional de Salud de Perú 2010
El Instituto Nacional de Salud de Perú 2010Anibal Velasquez
 
Retos de los Institutos Nacionales de Salud de UNASUR 2010
Retos de los Institutos Nacionales de Salud de UNASUR 2010Retos de los Institutos Nacionales de Salud de UNASUR 2010
Retos de los Institutos Nacionales de Salud de UNASUR 2010Anibal Velasquez
 
La reducción de la desnutrición crónica infantil como política de estado
La reducción de la desnutrición crónica infantil  como política de estadoLa reducción de la desnutrición crónica infantil  como política de estado
La reducción de la desnutrición crónica infantil como política de estadoAnibal Velasquez
 
Aportes para perú de la serie lancet 2013 de nutrición materna e infantil
Aportes para perú de la serie lancet 2013 de nutrición materna e infantilAportes para perú de la serie lancet 2013 de nutrición materna e infantil
Aportes para perú de la serie lancet 2013 de nutrición materna e infantilAnibal Velasquez
 
La técnica LQAS Lot Quality Assurance Sampling
La técnica LQAS Lot Quality Assurance SamplingLa técnica LQAS Lot Quality Assurance Sampling
La técnica LQAS Lot Quality Assurance SamplingAnibal Velasquez
 
Sistema de Seguimiento, Evaluación y Gestión de la Evidencia del MIDIS 2013 2014
Sistema de Seguimiento, Evaluación y Gestión de la Evidencia del MIDIS 2013 2014Sistema de Seguimiento, Evaluación y Gestión de la Evidencia del MIDIS 2013 2014
Sistema de Seguimiento, Evaluación y Gestión de la Evidencia del MIDIS 2013 2014Anibal Velasquez
 
Evolución de la epidemiología
Evolución de la epidemiologíaEvolución de la epidemiología
Evolución de la epidemiologíaAnibal Velasquez
 
Guia metodologica del estudio de carga de enfermedad del Perú
Guia metodologica del estudio de carga de enfermedad del PerúGuia metodologica del estudio de carga de enfermedad del Perú
Guia metodologica del estudio de carga de enfermedad del PerúAnibal Velasquez
 
Fundamentos de un sistema de farmacovigilancia en malaria
Fundamentos de un sistema de farmacovigilancia en malariaFundamentos de un sistema de farmacovigilancia en malaria
Fundamentos de un sistema de farmacovigilancia en malariaAnibal Velasquez
 
Cómo incrementar coberturas de intervenciones efectivas en Supervivencia infa...
Cómo incrementar coberturas de intervenciones efectivas en Supervivencia infa...Cómo incrementar coberturas de intervenciones efectivas en Supervivencia infa...
Cómo incrementar coberturas de intervenciones efectivas en Supervivencia infa...Anibal Velasquez
 
Prioridades de investigacion en el Perú
Prioridades de investigacion en el PerúPrioridades de investigacion en el Perú
Prioridades de investigacion en el PerúAnibal Velasquez
 
Avances y retos en la construccion del sistema nacional de investigación en s...
Avances y retos en la construccion del sistema nacional de investigación en s...Avances y retos en la construccion del sistema nacional de investigación en s...
Avances y retos en la construccion del sistema nacional de investigación en s...Anibal Velasquez
 
Nuevos desafíos para los Institutos Nacionales de Salud
Nuevos desafíos para los Institutos Nacionales de SaludNuevos desafíos para los Institutos Nacionales de Salud
Nuevos desafíos para los Institutos Nacionales de SaludAnibal Velasquez
 
Ley marco de aseguramiento universal en el Perú
Ley marco de aseguramiento universal en el PerúLey marco de aseguramiento universal en el Perú
Ley marco de aseguramiento universal en el PerúAnibal Velasquez
 
Carga de enfermedad y paquete de aseguramiento en salud en Perú
Carga de enfermedad y paquete de aseguramiento en salud en PerúCarga de enfermedad y paquete de aseguramiento en salud en Perú
Carga de enfermedad y paquete de aseguramiento en salud en PerúAnibal Velasquez
 

Mehr von Anibal Velasquez (20)

Epidemiologia y administracion de salud
Epidemiologia y administracion de saludEpidemiologia y administracion de salud
Epidemiologia y administracion de salud
 
El plan de evaluación y seguimiento de un proyecto
El plan de evaluación y seguimiento de un proyectoEl plan de evaluación y seguimiento de un proyecto
El plan de evaluación y seguimiento de un proyecto
 
Revisión sistemática sobre el impacto de cocinas mejoradas
Revisión sistemática sobre el impacto de cocinas mejoradasRevisión sistemática sobre el impacto de cocinas mejoradas
Revisión sistemática sobre el impacto de cocinas mejoradas
 
Recursos y utilitarios para evaluadores
Recursos y utilitarios para evaluadoresRecursos y utilitarios para evaluadores
Recursos y utilitarios para evaluadores
 
Gestion de la evidencia. Un nuevo enfoque de los sistemas de seguimiento y ev...
Gestion de la evidencia. Un nuevo enfoque de los sistemas de seguimiento y ev...Gestion de la evidencia. Un nuevo enfoque de los sistemas de seguimiento y ev...
Gestion de la evidencia. Un nuevo enfoque de los sistemas de seguimiento y ev...
 
El Instituto Nacional de Salud de Perú 2010
El Instituto Nacional de Salud de Perú 2010El Instituto Nacional de Salud de Perú 2010
El Instituto Nacional de Salud de Perú 2010
 
Retos de los Institutos Nacionales de Salud de UNASUR 2010
Retos de los Institutos Nacionales de Salud de UNASUR 2010Retos de los Institutos Nacionales de Salud de UNASUR 2010
Retos de los Institutos Nacionales de Salud de UNASUR 2010
 
La reducción de la desnutrición crónica infantil como política de estado
La reducción de la desnutrición crónica infantil  como política de estadoLa reducción de la desnutrición crónica infantil  como política de estado
La reducción de la desnutrición crónica infantil como política de estado
 
Aportes para perú de la serie lancet 2013 de nutrición materna e infantil
Aportes para perú de la serie lancet 2013 de nutrición materna e infantilAportes para perú de la serie lancet 2013 de nutrición materna e infantil
Aportes para perú de la serie lancet 2013 de nutrición materna e infantil
 
La técnica LQAS Lot Quality Assurance Sampling
La técnica LQAS Lot Quality Assurance SamplingLa técnica LQAS Lot Quality Assurance Sampling
La técnica LQAS Lot Quality Assurance Sampling
 
Sistema de Seguimiento, Evaluación y Gestión de la Evidencia del MIDIS 2013 2014
Sistema de Seguimiento, Evaluación y Gestión de la Evidencia del MIDIS 2013 2014Sistema de Seguimiento, Evaluación y Gestión de la Evidencia del MIDIS 2013 2014
Sistema de Seguimiento, Evaluación y Gestión de la Evidencia del MIDIS 2013 2014
 
Evolución de la epidemiología
Evolución de la epidemiologíaEvolución de la epidemiología
Evolución de la epidemiología
 
Guia metodologica del estudio de carga de enfermedad del Perú
Guia metodologica del estudio de carga de enfermedad del PerúGuia metodologica del estudio de carga de enfermedad del Perú
Guia metodologica del estudio de carga de enfermedad del Perú
 
Fundamentos de un sistema de farmacovigilancia en malaria
Fundamentos de un sistema de farmacovigilancia en malariaFundamentos de un sistema de farmacovigilancia en malaria
Fundamentos de un sistema de farmacovigilancia en malaria
 
Cómo incrementar coberturas de intervenciones efectivas en Supervivencia infa...
Cómo incrementar coberturas de intervenciones efectivas en Supervivencia infa...Cómo incrementar coberturas de intervenciones efectivas en Supervivencia infa...
Cómo incrementar coberturas de intervenciones efectivas en Supervivencia infa...
 
Prioridades de investigacion en el Perú
Prioridades de investigacion en el PerúPrioridades de investigacion en el Perú
Prioridades de investigacion en el Perú
 
Avances y retos en la construccion del sistema nacional de investigación en s...
Avances y retos en la construccion del sistema nacional de investigación en s...Avances y retos en la construccion del sistema nacional de investigación en s...
Avances y retos en la construccion del sistema nacional de investigación en s...
 
Nuevos desafíos para los Institutos Nacionales de Salud
Nuevos desafíos para los Institutos Nacionales de SaludNuevos desafíos para los Institutos Nacionales de Salud
Nuevos desafíos para los Institutos Nacionales de Salud
 
Ley marco de aseguramiento universal en el Perú
Ley marco de aseguramiento universal en el PerúLey marco de aseguramiento universal en el Perú
Ley marco de aseguramiento universal en el Perú
 
Carga de enfermedad y paquete de aseguramiento en salud en Perú
Carga de enfermedad y paquete de aseguramiento en salud en PerúCarga de enfermedad y paquete de aseguramiento en salud en Perú
Carga de enfermedad y paquete de aseguramiento en salud en Perú
 

Kürzlich hochgeladen

METHODS OF ACQUIRING KNOWLEDGE IN NURSING.pptx by navdeep kaur
METHODS OF ACQUIRING KNOWLEDGE IN NURSING.pptx by navdeep kaurMETHODS OF ACQUIRING KNOWLEDGE IN NURSING.pptx by navdeep kaur
METHODS OF ACQUIRING KNOWLEDGE IN NURSING.pptx by navdeep kaurNavdeep Kaur
 
Glomerular Filtration rate and its determinants.pptx
Glomerular Filtration rate and its determinants.pptxGlomerular Filtration rate and its determinants.pptx
Glomerular Filtration rate and its determinants.pptxDr.Nusrat Tariq
 
Presentació "Real-Life VR Integration for Mild Cognitive Impairment Rehabilit...
Presentació "Real-Life VR Integration for Mild Cognitive Impairment Rehabilit...Presentació "Real-Life VR Integration for Mild Cognitive Impairment Rehabilit...
Presentació "Real-Life VR Integration for Mild Cognitive Impairment Rehabilit...Badalona Serveis Assistencials
 
LUNG TUMORS AND ITS CLASSIFICATIONS.pdf
LUNG TUMORS AND ITS  CLASSIFICATIONS.pdfLUNG TUMORS AND ITS  CLASSIFICATIONS.pdf
LUNG TUMORS AND ITS CLASSIFICATIONS.pdfDolisha Warbi
 
call girls in paharganj DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️
call girls in paharganj DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️call girls in paharganj DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️
call girls in paharganj DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️saminamagar
 
call girls in aerocity DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️
call girls in aerocity DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️call girls in aerocity DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️
call girls in aerocity DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️saminamagar
 
PERFECT BUT PAINFUL TKR -ROLE OF SYNOVECTOMY.pptx
PERFECT BUT PAINFUL TKR -ROLE OF SYNOVECTOMY.pptxPERFECT BUT PAINFUL TKR -ROLE OF SYNOVECTOMY.pptx
PERFECT BUT PAINFUL TKR -ROLE OF SYNOVECTOMY.pptxdrashraf369
 
SWD (Short wave diathermy)- Physiotherapy.ppt
SWD (Short wave diathermy)- Physiotherapy.pptSWD (Short wave diathermy)- Physiotherapy.ppt
SWD (Short wave diathermy)- Physiotherapy.pptMumux Mirani
 
Case Report Peripartum Cardiomyopathy.pptx
Case Report Peripartum Cardiomyopathy.pptxCase Report Peripartum Cardiomyopathy.pptx
Case Report Peripartum Cardiomyopathy.pptxNiranjan Chavan
 
PNEUMOTHORAX AND ITS MANAGEMENTS.pdf
PNEUMOTHORAX   AND  ITS  MANAGEMENTS.pdfPNEUMOTHORAX   AND  ITS  MANAGEMENTS.pdf
PNEUMOTHORAX AND ITS MANAGEMENTS.pdfDolisha Warbi
 
The next social challenge to public health: the information environment.pptx
The next social challenge to public health:  the information environment.pptxThe next social challenge to public health:  the information environment.pptx
The next social challenge to public health: the information environment.pptxTina Purnat
 
History and Development of Pharmacovigilence.pdf
History and Development of Pharmacovigilence.pdfHistory and Development of Pharmacovigilence.pdf
History and Development of Pharmacovigilence.pdfSasikiranMarri
 
POST NATAL EXERCISES AND ITS IMPACT.pptx
POST NATAL EXERCISES AND ITS IMPACT.pptxPOST NATAL EXERCISES AND ITS IMPACT.pptx
POST NATAL EXERCISES AND ITS IMPACT.pptxvirengeeta
 
97111 47426 Call Girls In Delhi MUNIRKAA
97111 47426 Call Girls In Delhi MUNIRKAA97111 47426 Call Girls In Delhi MUNIRKAA
97111 47426 Call Girls In Delhi MUNIRKAAjennyeacort
 
Informed Consent Empowering Healthcare Decision-Making.pptx
Informed Consent Empowering Healthcare Decision-Making.pptxInformed Consent Empowering Healthcare Decision-Making.pptx
Informed Consent Empowering Healthcare Decision-Making.pptxSasikiranMarri
 
Introduction to Sports Injuries by- Dr. Anjali Rai
Introduction to Sports Injuries by- Dr. Anjali RaiIntroduction to Sports Injuries by- Dr. Anjali Rai
Introduction to Sports Injuries by- Dr. Anjali RaiGoogle
 
Presentation on General Anesthetics pdf.
Presentation on General Anesthetics pdf.Presentation on General Anesthetics pdf.
Presentation on General Anesthetics pdf.Prerana Jadhav
 
Apiculture Chapter 1. Introduction 2.ppt
Apiculture Chapter 1. Introduction 2.pptApiculture Chapter 1. Introduction 2.ppt
Apiculture Chapter 1. Introduction 2.pptkedirjemalharun
 
Report Back from SGO: What’s New in Uterine Cancer?.pptx
Report Back from SGO: What’s New in Uterine Cancer?.pptxReport Back from SGO: What’s New in Uterine Cancer?.pptx
Report Back from SGO: What’s New in Uterine Cancer?.pptxbkling
 
April 2024 ONCOLOGY CARTOON by DR KANHU CHARAN PATRO
April 2024 ONCOLOGY CARTOON by  DR KANHU CHARAN PATROApril 2024 ONCOLOGY CARTOON by  DR KANHU CHARAN PATRO
April 2024 ONCOLOGY CARTOON by DR KANHU CHARAN PATROKanhu Charan
 

Kürzlich hochgeladen (20)

METHODS OF ACQUIRING KNOWLEDGE IN NURSING.pptx by navdeep kaur
METHODS OF ACQUIRING KNOWLEDGE IN NURSING.pptx by navdeep kaurMETHODS OF ACQUIRING KNOWLEDGE IN NURSING.pptx by navdeep kaur
METHODS OF ACQUIRING KNOWLEDGE IN NURSING.pptx by navdeep kaur
 
Glomerular Filtration rate and its determinants.pptx
Glomerular Filtration rate and its determinants.pptxGlomerular Filtration rate and its determinants.pptx
Glomerular Filtration rate and its determinants.pptx
 
Presentació "Real-Life VR Integration for Mild Cognitive Impairment Rehabilit...
Presentació "Real-Life VR Integration for Mild Cognitive Impairment Rehabilit...Presentació "Real-Life VR Integration for Mild Cognitive Impairment Rehabilit...
Presentació "Real-Life VR Integration for Mild Cognitive Impairment Rehabilit...
 
LUNG TUMORS AND ITS CLASSIFICATIONS.pdf
LUNG TUMORS AND ITS  CLASSIFICATIONS.pdfLUNG TUMORS AND ITS  CLASSIFICATIONS.pdf
LUNG TUMORS AND ITS CLASSIFICATIONS.pdf
 
call girls in paharganj DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️
call girls in paharganj DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️call girls in paharganj DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️
call girls in paharganj DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️
 
call girls in aerocity DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️
call girls in aerocity DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️call girls in aerocity DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️
call girls in aerocity DELHI 🔝 >༒9540349809 🔝 genuine Escort Service 🔝✔️✔️
 
PERFECT BUT PAINFUL TKR -ROLE OF SYNOVECTOMY.pptx
PERFECT BUT PAINFUL TKR -ROLE OF SYNOVECTOMY.pptxPERFECT BUT PAINFUL TKR -ROLE OF SYNOVECTOMY.pptx
PERFECT BUT PAINFUL TKR -ROLE OF SYNOVECTOMY.pptx
 
SWD (Short wave diathermy)- Physiotherapy.ppt
SWD (Short wave diathermy)- Physiotherapy.pptSWD (Short wave diathermy)- Physiotherapy.ppt
SWD (Short wave diathermy)- Physiotherapy.ppt
 
Case Report Peripartum Cardiomyopathy.pptx
Case Report Peripartum Cardiomyopathy.pptxCase Report Peripartum Cardiomyopathy.pptx
Case Report Peripartum Cardiomyopathy.pptx
 
PNEUMOTHORAX AND ITS MANAGEMENTS.pdf
PNEUMOTHORAX   AND  ITS  MANAGEMENTS.pdfPNEUMOTHORAX   AND  ITS  MANAGEMENTS.pdf
PNEUMOTHORAX AND ITS MANAGEMENTS.pdf
 
The next social challenge to public health: the information environment.pptx
The next social challenge to public health:  the information environment.pptxThe next social challenge to public health:  the information environment.pptx
The next social challenge to public health: the information environment.pptx
 
History and Development of Pharmacovigilence.pdf
History and Development of Pharmacovigilence.pdfHistory and Development of Pharmacovigilence.pdf
History and Development of Pharmacovigilence.pdf
 
POST NATAL EXERCISES AND ITS IMPACT.pptx
POST NATAL EXERCISES AND ITS IMPACT.pptxPOST NATAL EXERCISES AND ITS IMPACT.pptx
POST NATAL EXERCISES AND ITS IMPACT.pptx
 
97111 47426 Call Girls In Delhi MUNIRKAA
97111 47426 Call Girls In Delhi MUNIRKAA97111 47426 Call Girls In Delhi MUNIRKAA
97111 47426 Call Girls In Delhi MUNIRKAA
 
Informed Consent Empowering Healthcare Decision-Making.pptx
Informed Consent Empowering Healthcare Decision-Making.pptxInformed Consent Empowering Healthcare Decision-Making.pptx
Informed Consent Empowering Healthcare Decision-Making.pptx
 
Introduction to Sports Injuries by- Dr. Anjali Rai
Introduction to Sports Injuries by- Dr. Anjali RaiIntroduction to Sports Injuries by- Dr. Anjali Rai
Introduction to Sports Injuries by- Dr. Anjali Rai
 
Presentation on General Anesthetics pdf.
Presentation on General Anesthetics pdf.Presentation on General Anesthetics pdf.
Presentation on General Anesthetics pdf.
 
Apiculture Chapter 1. Introduction 2.ppt
Apiculture Chapter 1. Introduction 2.pptApiculture Chapter 1. Introduction 2.ppt
Apiculture Chapter 1. Introduction 2.ppt
 
Report Back from SGO: What’s New in Uterine Cancer?.pptx
Report Back from SGO: What’s New in Uterine Cancer?.pptxReport Back from SGO: What’s New in Uterine Cancer?.pptx
Report Back from SGO: What’s New in Uterine Cancer?.pptx
 
April 2024 ONCOLOGY CARTOON by DR KANHU CHARAN PATRO
April 2024 ONCOLOGY CARTOON by  DR KANHU CHARAN PATROApril 2024 ONCOLOGY CARTOON by  DR KANHU CHARAN PATRO
April 2024 ONCOLOGY CARTOON by DR KANHU CHARAN PATRO
 

Health policy program effects on institutional delivery in peru

  • 1. Health Policy 77 (2006) 221–232 Evaluating program effects on institutional delivery in Peru Michael J. McQuestiona,∗, Anibal Velasquezb a Johns Hopkins Bloomberg School of Public Health, Population and Family Health Sciences, 615 N Wolfe Street, E-4142, Baltimore, MD 21205, USA b Consultant, PHRPlus, Antequera 777 Piso 8, San Isidro, Lima, Peru Abstract We evaluate the joint effects of two targeted Peruvian health programs on a mother’s choice of whether to deliver in a public emergency obstetric care (EmOC) facility. The national maternal and child health insurance, or SMI Program, provided delivery care coverage to Peru’s poorest households beginning in 1998. During 1996–2002, Proyecto 2000 sought to improve the quality of EmOC and increase utilization of public EmOC facilities in the districts reporting the highest maternal and neonatal mortality levels. Our data come from the Proyecto 2000 endline evaluation, which sampled 5335 mothers living in the catchment areas of 29 treatment and 29 matched control EmOC facilities. Using propensity scoring and two quality of care indices, we find significantly higher quality of care in Proyecto 2000 treatment facilities. Using variance components logistic models, we find a mother enrolled in the SMI Program was more likely to have delivered her last child in a public EmOC, controlling for household constraints. Residence in a Proyecto 2000 treatment area did not significantly affect the choice. A cross-level interaction term was insignificant, indicating the two program effects were independent. © 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Quality of care; Evaluation; Developing countries; Safe motherhood 1. Introduction This study examines two very different efforts to increase institutional delivery in Peru. During 1992–1997, Peru implemented large-scale health sec- tor decentralization reforms. The reforms were criti- cized for widening health disparities, particularly in hospital utilization [1]. Peru’s DHS III (1996) and DHS IV (2000) surveys documented a relative decline in physician-assisted deliveries among rural and poorly ∗ Corresponding author. Tel.: +1 410 502 6037; fax: +1 410 955 2303. E-mail address: mmcquest@jhsph.edu (M.J. McQuestion). educated women over the period. To correct this, the Peruvian Ministry of Health initiated a series of tar- geted maternal and child health interventions, two of which we evaluate. The first intervention was Proyecto 2000, a USAID-funded effort begun in 1996 in the 12 of Peru’s 25 departmentos reporting the highest mater- nal mortality levels. Proyecto 2000 aimed to increase the proportion and quality of institutional deliveries, thereby reducing maternal mortality and improving birth outcomes. The project began with mass media, health education and social mobilization efforts pro- moting delivery in the nearest public emergency obstet- ric care (EmOC) facility. Its emphasis, however, was on improving the quality of services on offer. The sec- 0168-8510/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.healthpol.2005.07.007
  • 2. 222 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 ond intervention was the Maternal and Child Health Insurance (SMI) Program, launched in 1998. The SMI Program covered most maternal and child health costs, including institutional delivery in public EmOC facil- ities [2]. It was a means-tested program in that only households in the poorest wealth quintile were eligi- ble to participate. By 2000, this program was reaching about 50% of eligible households in two pilot regions [3], and the following year it was extended nationwide. The two programs thus incorporated different target- ing strategies. Proyecto 2000 targeted high-risk dis- tritos, specific EmOC facilities and their surrounding communities while the SMI Program directly targeted the country’s poorest households. Did either program increase EmOC utilization? In this study we use quasi- experimental data to probe this question. We model a woman’s choice of where she delivered her last baby, conditional on exposure to these two programs. 2. Background 2.1. Recent perinatal health trends Demographic data show perinatal health in Peru improved over this period. The country’s neonatal death rate fell from 27 to 18 deaths per 1000 live births during the 1990s [4]. Peru’s estimated mater- nal mortality ratio also fell, from 265/100,000 live births in 1990–1996 to 185 in 1994–2000 [5], yet it remained third highest among 14 Latin American countries reporting in 1999 [6]. Maternal and neona- tal mortality are largely influenced by two factors: a woman’s decision whether or not to utilize institutional delivery care and the quality of that care. High-quality EmOC can prevent an estimated one-third of mater- nal deaths [7], and 40–62% of neonatal deaths [8]. Regardingmaternalbehaviors,the1996DHSIIIsurvey showed that 55% of women who had given birth in the previous5yearsdidsoathome.Another38%usedpub- lic health care facilities and 5% used private delivery facilities [9]. Over the succeeding 5 years, the propor- tion of home deliveries fell to 47%, the public sector’s share rose to 48% and the proportion using private facilities stayed at about 5% [10]. There are no com- parable EmOC quality of care estimates, however, a recent qualitative study ranked Peru second of 13 Latin American countries evaluated in terms of maternal and neonatal program effort [11,12]. It is thus plausible that the observed perinatal health improvements were due to increased institutional deliveries that in turn resulted from program improvements. There were other important factors affecting mater- nal and perinatal health in Peru over this period. The country’s per capita GNP grew by a mean 2.4% per annum during the 1990s [13], an improvement over the chaotic 1980s. Total fertility rates declined from 4.8 in 1986 to 2.1 in 2000, lengthening birth intervals and reducing the proportion of high-parity births [4]. These changing background forces may have been more deci- sive health behavioral determinants that the program effects we attempt to elucidate. 2.2. Maternal health risk factors In Peru, as elsewhere, it is the poorest, most remote and most socially excluded women who least use maternal health services [14], and are at highest risk of maternal, perinatal and post-perinatal mortality [15,16]. A 2000 survey in Peru’s Ayacucho Depart- ment, for example, found that only about one-fourth of women with complications were delivered in adequate EmOC facilities [17]. In Peru’s DHS IV survey some 83% of women identified at least one barrier to access- ing local maternal health services. Expense was the leading problem, followed by lack of female caregivers [10]. Other cultural factors act as barriers to EmOC uti- lization, particularly among the 47% of Peruvians who do not speak Spanish as their first language. Reports of discrimination and mistreatment by health work- ers are commonplace [18,19]. The DHS data suggest that more high-risk women chose to utilize the public EmOC facilities over this period. 2.3. National SMI Program TheFujimoriAdministrationinstitutedtheSMIPro- gram in 1998. It was Peru’s first attempt to subsidize preventive and maternal care for low-income pregnant women, mothers and children ages 0–4 years. Many saw it as an attempt to restore basic health rights that had been infringed by decentralization. In 2001, the program was supplanted by a national Integral Health Insurance Plan, which offered a wider gamut of tar- geted benefits to low-income Peruvians of all ages. Until 1998, any woman could have accessed any public
  • 3. M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 223 EmOC facility where she had to pay fees for service on a sliding scale. The targeted insurance programs elim- inated these fees for the eligible poor. However, by 2001 many eligible households were still not enrolled in the program. Peru’s public health system still lacked the infrastructure and level of performance needed to extend MCH services to all those eligible. Production levels remained exceedingly low. The median num- ber of consultations that year was less than three per day in half of the Ministry of Health’s peripheral PHC facilities. To date there has been no comprehensive evaluation of these targeted insurance efforts [20]. 2.4. Proyecto 2000 2.4.1. Phase I We describe Proyecto 2000 in greater detail because it generated the data we analyze. Proyecto 2000 was implemented by a team of Ministry of Health and exter- nal expert consultants. The team sought to make the Ministry’s EmOC services culturally acceptable and to ensure that the facilities delivered high-quality care. A hallmark of Proyecto 2000 and other Safe Motherhood projects is an emphasis on making services “woman- friendly”. An EmOC facility is woman-friendly if: (a) it is easily accessible and convenient to use; (b) high- quality services are offered; (c) local cultural beliefs and social norms are incorporated into treatment proto- cols and (d) confidentiality is guaranteed, information is shared and clients’ choices are respected [21]. The Proyecto 2000 team worked at facility and commu- nity levels to accomplish these aims. At baseline, team members and Regional Ministry of Health educators gathered and analyzed qualitative data on mothers’ perceptions and preferences regarding pregnancy and childbirth. They used these data to mount a multime- dia Safe Motherhood campaign in the treatment areas. In addition, expert trainers trained 3692 community- based traditional birth attendants (promotoras), and EmOC staff formally engaged newly constituted com- munity health committees (Comites Locales de Admin- istracion en Salud) in their catchment areas. Facility inputs included physical plant improvements, retrain- ing of 409 facility-based providers, incorporation of local birthing practices into clinical protocols and the introduction of a continuous quality of care (“autoeval- uacion”) model in some 89 public hospitals and health centers. In brief, the autoevaluacion model incorpo- rated the Donabedian continuous quality of care [22], and the McCarthy and Maine maternal mortality deter- minants frameworks [23]. The autoevaluacion instru- ment included a battery of detailed indicators regarding essential obstetric and neonatal care, physical facilities, patient interaction and management. The expectation was that greater autonomy and participation in the self-appraisal process would stimulate improved staff performance, and the resulting improved quality of care would generate more institutional deliveries as client satisfaction improved. All facilities were expected to attain quality of care improvements sufficient to merit formal certification by expert evaluators. These 89 facilities comprised the original treatment arm. 2.4.2. Midterm evaluation (2000) As of 1998, 72 treatment facilities were still active in the program, all of which had attained formal qual- ity of care certification as high-quality perinatal care centers [24]. By October 2000, the number of active treatment facilities had fallen to 60. At that time a midterm evaluation was carried out. An external evalu- ation team examined a random sample of 37 treatment facilities. They also identified a group of 37 similar EmOC facilities not exposed to the project to serve as a comparison group. The control facilities were drawn from six Ministry of Health districts (DISAs) with ser- vice population characteristics (literacy, contraceptive prevalence, use of institutional delivery services, mal- nutrition and poverty levels) similar to the Proyecto 2000 areas. The control facilities had received only rou- tine Ministry of Health supervision over the period. The mid-term evaluation was entirely facility-based. Expert observersusedstandardizedchecklistsandinstitutional record reviews to assess the quality of EOC on offer. They found evidence of improved quality of care and a relative increase in the numbers of institutional deliver- ies in the treatment group facilities as compared to the control facilities (Table 1). Additionally, the observers interviewed samples of prenatal clients. They found users of treatment facilities were more knowledgeable about pregnancy, more satisfied with their experiences and more likely intended to deliver their babies in that treatment area facility [25]. 2.4.3. Phase II (2001–2002) During Phase II, Proyecto 2000 inputs were con- centrated on the 31 treatment facilities judged to have
  • 4. 224 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 Table 1 Selected EOC facility indicators, Proyecto 2000 Variable Control facilities Treatment facilities Mean S.D. Mean S.D. 1997 Institutional births 1463 1937 1486 845 Prop births <2.5 kg 0.07 0.08 0.10 0.09 Prop births c-section 0.24 0.13 0.19 0.08 2000 Institutional births 1052 1434 1542 681 Prop births <2.5 kg 0.07 0.04 0.09 0.07 Prop births c-section 0.24 0.10 0.23 0.11 Ob-gyns 7.92 7.61 7.50 5.37 Births/ob-gyn 140 59 284 152 2002 Autoevaluacion scorea 26.80 8.30 51.20 14.40 EmOC capacity score 69.30 8.90 72.19 7.18 a First principal component of nine-factor index. performed best in Phase I. Project supervisors regu- larly visited these facilities to ensure the autoevalua- ciones were performed in each facility each quarter. Project data show the autoevaluaciones were in fact implemented. Of the 29 treatment establishments that participated to endline (2002), all carried out at least two autoevaluaciones, 25/29 carried out three, 13/29 carried out four and 3/29 carried out five. The auto- evaluacion scores reported by the facilities increased with each round (Fig. 1). These data indicate the institu- tional Proyecto 2000 interventions were implemented and suggest the interventions could have been strong enough to improve the quality of EmOC services on Fig. 1. Autoevaluacion scores by evaluation round, 2000–2001, Proyecto 2000. offer. Our task is thus to disentangle two distinct treat- ment effects, one operating through the health sys- tem, the other directly on household health production. We expect the two effects will be synergetic: insured women in high-quality EmOC catchment areas ought to be the most likely to use that facility. With these points in mind, we model the probability a Peruvian mother chose to deliver her youngest child at the nearest public EmOC facility, conditional on the qualityofcareatthatfacility,herhouseholdconstraints, SMI Program participation, and whether her commu- nity and facility participated in Proyecto 2000. 3. Data and methods 3.1. Facility data The Proyecto 2000 evaluators collected a second round of endline evaluation data in mid-2002 and it is these data we analyze in the present paper. The Phase II treatment group included all 19 Phase I hospitals and a subset of 12 Phase I health centers. The eval- uators selected a new control group, consisting of 15 of the Phase I control establishments and 14 additional establishments. As in Phase I, the 14 new control facili- ties were purposively selected from six newly matched DISAs that were unexposed to the project. Expert teams again evaluated essential obstetric care in the EmOC facilities using the same extensive standard checklist used in the midterm evaluation. They also evaluated the quality of services using the autoevaluacion instrument itself. Thirdly, they collected selected service indica- tors routinely reported by each facility to the Ministry of Health. We used these data to derive two EmOC quality of care measures, which we described below. 3.2. Household data To assess changes in local utilization patterns and measure SMI Program participation, the Proyecto 2000 evaluators carried out a household survey in all treat- ment and control facility service areas. The survey instrument incorporated selected items from Peru’s DHS III and DHS IV survey questionnaires [26], particularly household characteristics, birth histories and pregnancy-related behaviors. Sampling procedures were similar to those used in the DHS. Peru’s 1993 cen- sus of households provided the sampling frame. Within
  • 5. M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 225 each Proyecto 2000 catchment area, census tracts were listed and selected at random. Within each tract house- holds were selected systematically following cardinal directions from the approximate center of each cluster. Ten women who had given birth in the previous 5 years were surveyed in each cluster. The measures we derive from these household-level data described below. We merged the facility and household data to make a hier- archical dataset consisting of 5335 women nested in 420 clusters in 58 facility catchment areas. 3.3. Facility quality of care measures Our two quality of care and basic EmOC capac- ity measures, along with other facility indicators, are shown in Table 1. The first measure (autevaluacion score) is the first principal component (eigenvector) from a factor score analysis of nine items from the autoevaluacion checklist. The nine items were: blood is routinely filtered, an incinerator is present, there are generic versus proprietary drugs are in the phar- macy, there is an up-to-date list of all drugs dispensed, patients receive health educational messages, patient satisfaction is measured, remedial activities to improve patient satisfaction were implemented, there is a local community advisory committee, staff meets at least every 3 months, feedback on performance is given at thosemeetings.WecomputedCronbach’salphaandthe Kaiser–Meyer–Olkin measure of sampling adequacy [27] for these nine items. The resulting coefficients were, respectively, 0.70 and 0.65 (results not shown). We conclude the nine items are tapping a common underlying construct but we note that 0.80 is the con- ventional “gold standard” for both measures [27]. The second measure, EmOC capacity score, is the percent score on a battery of 711 items the evaluators used to assess the technical capacity of a facility to deal with obstetric emergencies. The evaluators grouped the indicators into nine categories: human resources, pre- natal and obstetric equipment, radiology, pharmacy, delivery room equipment, neonatal care unit, maternity ward, operating room and blood bank. As Table 1 shows, treatment facilities scored higher on both the autoevaluacion quality of care index and EOC capacity score. This apparent improvement could be a true difference due to the Proyecto 2000 inputs or it could be an artifact of the non-random match- ing of treatment and control facilities, attrition or other sources of bias. To explore this further we used four of the routinely reported EmOC facility indicators to compute a propensity score for the assignment pro- cess. The aim of propensity scoring is to make assign- ment “strongly ignorable” by blocking observations on observables [28,29]. The outcome is the dummy variable indicating assignment to treatment or control group. The covariates we used are: number of obstetri- cians and gynecologists on staff, number of maternal deathsin2000,numberofcaesariansectionsperformed in 2000 and the proportion of all deliveries performed outside of the facility. We generated a balanced score with matched pairs of facilities falling into eight blocks (results not shown). We then used the propensity score to generate three alternative non-parametric treatment effects estimates for each quality of care measure. 3.4. Household measures We control for several household risk factors in our models. Maternal education is a positive predictor of maternal behaviors in Peru [14]. Other important fac- tors include maternal age, number of births and socioe- conomic status [30]. Maternal educational attainment is coded using terciles, where 1 = no or primary educa- tion, 2 = some secondary and 3 = completed secondary and higher. Another dummy variable is coded one for women who have had three or more live births, zero otherwise. To control for household wealth we use the Filmer–Pritchett method [31], wherein weights from principal components are applied to a list of household assets, scores are summed and ranked and each house- hold is assigned to one of five wealth quintiles. We add additional dummy variables to control for whether the last child was born in 1998, 1999, 2000 or 2001. We use a binary dummy variable to indicate whether or not the household participates in the SMI Program. As shown in Table 2, the characteristics of Proyecto 2000 sample households were broadly comparable across treatment and control areas. Only ethnicity var- ied: treatment area women were less likely to be Span- ish speakers. Delivery patterns also appear similar across the study arms. Four of every five women in both treatment and control areas delivered their last babies in some kind of institution. Though the matching appears adequate, the Proyecto 2000 sample is not a nationally representative sample. Table 2 shows the same indica- tors computed from Peru’s DHS IV survey. The DHS
  • 6. 226 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 Table 2 Sample characteristics and maternal health indicators, women giving birth in previous 5 years, Peru 1996–2002 Variable Proyecto 2000 t-Test DHS IV Control facilities Treatment facilities Matched to P2000 Full sample Mean S.D. Mean S.D. Mean S.D. Mean S.D. Last birth institutional 0.83 0.37 0.82 0.38 0.77 0.21 0.53 0.50 Last birth prenatal care 0.91 0.29 0.91 0.29 0.87 0.11 0.78 0.42 Mothers characteristics Age (years) 26.64 6.62 27.67 7.10 −5.38a 29.32 1.41 29.21 7.05 No. live births 2.57 1.70 2.66 1.88 2.92 0.60 3.51 2.47 Educational level Primary 0.21 0.40 0.23 0.42 −2.37a 0.31 0.15 0.45 0.50 Secondary 0.44 0.50 0.44 0.50 0.38 0.15 0.32 0.47 Superior 0.35 0.48 0.32 0.47 0.26 0.13 0.14 0.34 Union status Married 0.38 0.48 0.38 0.49 0.42 0.16 0.41 0.49 Consensual 0.48 0.50 0.47 0.50 0.43 0.16 0.47 0.50 Divorced/separated/widow 0.15 0.36 0.15 0.36 0.08 0.05 0.07 0.25 Rural origin 0.39 0.49 0.36 0.48 0.26 0.19 0.40 0.49 Non-Spanish speaker 0.02 0.15 0.08 0.27 −9.30a 0.10 0.17 0.22 0.42 Households Electricity 0.93 0.25 0.90 0.30 −4.05a 0.80 0.20 0.52 0.50 Safe water 0.86 0.35 0.81 0.39 4.64a 0.97 0.12 0.81 0.39 Durable floor 0.53 0.50 0.56 0.50 −2.40a 0.44 0.11 0.53 0.50 Safe toilet 0.59 0.49 0.62 0.49 −2.09a 0.75 0.19 0.60 0.49 n 2514 2821 5826 13832 a Significant at p < 0.05 level. IV sample is a nationally representative weighted sam- ple drawn from 589 of the 1789 distritos enumerated in Peru’s 1993 household census. We used the distrito identifiers to match the DHS IV and Proyecto 2000 data (n = 68 matched distritos). The Proyecto 2000 sample is somewhat better educated, more likely to be Spanish- speaking and living at a slightly higher socioeconomic level than the DHS subsample from the same distri- tos. Compared to the national DHS sample, women in the Proyecto 2000 distritos were more intensive mater- nal health service users, better educated, more likely to speak Spanish and less likely to have households with electricity. Accordingly, all inferences we will make are limited to the Proyecto 2000 sample data. 3.5. Behavioral model As mentioned, we estimate a facility-level Proyecto 2000 treatment effect using propensity scoring. Here we describe our behavioral model, which includes indi- cator variables that control for the effects of both pro- grams. We interpret their slopes as indirect treatment estimates. Given the heteroscedastic treatments and the many suspected unobserved variables that could have affectedmothers’deliverychoices,wefittwo-levelran- dom effects models of the form: yij = πij + εij logit(πij) = β0j + β1Xij + β2Tij + β3Iij +β4Pj + β5IijPj β0j = δ0j + δ01z1j + κj εij ∼ N(0, 1), cov(Xij, Pj, Iij, Tij, εij) = 0 κj ∼ N(0, σ2 κ ), cov(z1j, κj) = 0 cov(εij, κj) = 0 In this model πij is the probability mother i in EmOC facility service area j chose institutional delivery yij, and εij is an individual error term. β1 is a parame-
  • 7. M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 227 ter measuring individual effects due to household and individualcovariatesXij.β2 measurestheeffectoftime, specified as dummy variables for child i’s birth year Tij. β3 controls for SMI Program participation, indicated by Ii, which is coded one for participants, zero for non- participants. β4 controls for being in a Proyecto 2000 treatment facility area, indicated by Pj, a dummy vari- able coded one for Proyecto 2000 treatment distritos, zero otherwise. We include β5 to capture any cross- level interaction between the two treatments. This term also adjusts for the possibility the insurance program was not uniformly implemented across the Proyecto 2000 areas. β0j is a random facility-level intercept, δ0j and δ01 are parameters, z1j is a dummy variable for facility and kj is a facility-level random effect. If the variance of kj, denoted as σ2 k , is significant, then we know there are unobserved variable effects which might otherwise have biased the fixed effect parameters in a conventional model. To fit the behavioral model we must make several assumptions. We assume that each mother is influenced solely by her own EmOC facility. We further assume that all mothers in the Proyecto 2000 treatment areas were equally exposed to the treatments and that access to the nearest Ministry of Health EmOC facility did not differ between treatment and control areas. This was not the case for the SMI Program, which was not uni- formly implemented and was means-tested. Although we lack any SMI Program data, we assume that all eli- gible households with access did enroll. Because the subsidy was conditioned on use of the nearest Ministry EmOC facility we assume that any SMI Program effect contributed fully to the likelihood of our outcome. We estimate a series of nested multilevel models using Stata’s gllamm program [32]. The program uses a maximum likelihood algorithm with adaptive quadra- ture to model latent variables as random effects. One advantage of gllamm over other multilevel programs is that it generates log-likelihood statistics useful for comparing model fits. All standard errors are estimated using the Huber–White sandwich estimator to adjust for the clustered survey design effect [33]. 4. Results 4.1. Institutional model Our institutional treatment effects are shown in Table 3. The estimates include population-average treatment effects (ATE) produced by radius matching and nearest neighbor matching algorithms. Following Imbens (2003), we also estimate the within-sample ATE. For comparison, we report the slope of a simple OLS model with the treatment dummy the sole regres- sor. The two indicators are measured on different met- rics so their treatment effect estimates are not directly comparable. What we expect are consistent estimates for each indicator. Inferences are based on Wald tests. With the outcome the EmOC capacity score, popula- tion and sample ATE estimates were all significant. With the autoevaluacion factor score as the outcome, Table 3 Facility-level treatment effects estimates, conditioned on propensity scorea, Proyecto 2000 EmOC capacity (n = 52 facilities) Autoevaluation (n = 55 facilities) Coefficient S.D. (Pairs) Coefficient S.D. (Pairs) OLS slope 12.2** 2.8 0.55** 0.28 Radius matching ATTb 11.7** 2.6 (26t,18c) 0.5 0.43 (6t,16c) Nearest neighbor Random draw ATT 12.1** 2.7 (26t,11c) 0.55* 0.32 (26t,11c) Equal weights ATT 12.1** 2.7 (26t,11c) 0.55 0.36 (26t,11c) Sample ATEc 12.8** 2.9 (43t,43c) 0.65** 0.27 (n = 41t,41c) a Propensity score variables: number of ob-gyns, no. maternal deaths 2000, no. caesarian sections 2000, proportion of all cases delivered in facility in 2000. b ATT, average treatment effect on the treated. c ATE, average treatment effect. * Significant at 0.10 < p < 0.05 level. ** Significant at p < 0.05 level.
  • 8. 228 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 we find a significant sample ATE (0.65) but only one populationaveragetreatmenteffect,anditismarginally significant. The EmOC capacity score is apparently a more sensitive quality of care measure. We conclude the estimates are robust and that Proyecto 2000 inputs did improve the quality of care in the EmOC treatment facilities. Our main interest, however, is in measuring any health behavioral impacts and assessing whether they are linked to facility quality of care improvements, to the provision of MCH insurance or a combination of the two. 4.2. Behavioral model Our behavioral model results are shown in Table 4. We show exponentiated slopes (odds ratios) to ease interpretation. In Model 1, the reference household model, covariate effects are signed as expected. The more educated and wealthier the woman, the more likely she delivered in the EmOC facility. Those who do not speak Spanish and have had three or more live births are less likely to choose institutional delivery. The dummy variables for birth years 1999, 2000 and 2001 capture unmeasured variables that are associated with EmOC delivery. Those net effects are positive compared with those in 1997 and 1998, the omitted categories. There is a significant random effect, indi- cating that women’s decisions to deliver in the public EmOC facility are correlated in some catchment areas more than in others due to omitted variables that jointly affect their behaviors. Model 2 results show that living in a Proyecto 2000 area has no significant effect on delivery choice. Model 3, in contrast, shows that the odds of institutional deliv- ery for women covered by the SMI Program were twice the odds for women not covered. Controlling for insur- ance removes upward biases on the highest education and wealth dummies. The difference in log-likelihoods shows that Model 3 is also a significantly better-fitting model than Models 1 or 2. Model 4 includes an inter- action term between the highest wealth quintile and the insurance dummies. These better-off households were ineligible for coverage and the negative interac- tion term captures this fact. Controlling this interac- tion further decreases the direct effects of being in the wealthiest quintile. The most dramatic effect, however, is a seven-fold increase the odds of EmOC delivery for the insured women. This pattern is consistent with the fact only the poorest households were eligible for the SMI Program. In Model 5 we add a cross-level interac- Table 4 Two-level logistic regression delivery models, exponentiated effects, Proyecto 2000 Variablea Coefficient (S.E.) Model 1 Model 2 Model 3 Model 4 Model 5 Non-Spanish speaker 0.38** (0.07) 0.38** (0.07) 0.36** (0.07) 0.36** (0.07) 0.36** (0.07) Some secondary education 2.69** (0.27) 2.69** (0.27) 2.59** (0.26) 2.61** (0.27) 2.61** (0.27) Complete secondary or more 5.63** (0.83) 5.63** (0.83) 4.81** (0.72) 4.75** (0.71) 4.76** (0.71) Three or more live births 0.66** (0.06) 0.66** (0.06) 0.64** (0.06) 0.64** (0.06) 0.64** (0.06) 60–79th wealth quintile 2.23** (0.31) 2.24** (0.31) 2.10** (0.30) 2.13** (0.30) 2.13** (0.30) 80–100th wealth quintile 3.26** (0.61) 3.27** (0.62) 2.86** (0.54) 2.25** (0.46) 2.26** (0.46) Born 1999 1.30** (0.15) 1.30** (0.15) 1.29** (0.15) 1.29** (0.15) 1.29** (0.15) Born 2000 1.75** (0.22) 1.75** (0.22) 1.76** (0.22) 1.76** (0.22) 1.76** (0.22) Born 2001 1.62** (0.20) 1.62** (0.20) 1.64** (0.20) 1.63** (0.20) 1.64** (0.20) Insured 2.02** (0.25) 15.71** (14.52) 14.96** (13.98) Insured 80–100th quintile* 0.34** (0.16) 0.34** (0.16) P2000 treatment area 0.79 (0.25) 0.80 (0.26) P2000 area insured* 1.08 (0.26) Level-two random effect σ2 k 1.06** (0.27) 1.05** (0.26) 1.11** (0.28) 1.00** (0.28) 1.09** (0.24) Log-likelihood −1797.4 −1797.1 −1780.1 −1777.0 −1776.8 n 5190 5190 5190 5190 5190 a Omitted categories: education secondary and beyond, Spanish speaker, one or two live births, lowest three wealth quintiles, born 1998, no insurance. * Significant at 0.10 < p < 0.05 level. ** Significant at p < 0.05 level.
  • 9. M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 229 Fig. 2. Real and simulated posterior means Models 1–5. tion term to test whether the two programs interacted to affect EmOC probabilities. The interaction is insignif- icant. Comparisons of the models with their conventional single-equation logit analogs show that, in each case, the random effects specification improved model fit (results not shown). To check whether the normality assumption was met, we standardized and plotted the posterior means from each model. All distributions were near normal but somewhat negatively skewed. The skewness was due to three clinics (two treatment, one control) whose means were more than two standard deviations below the sample means. We used gllamm’s post-estimation command gllasim to resample the pos- teriormeans.ResultsareshowninFig.2.Thesimulated distributions were normal and no longer skewed. Fol- lowing arguments by Longford [34], the three clinics are thus not true outliers; their apparent outlier status is a feature of the realized sample. We leave them in the dataset and conclude that the models are robust. 5. Discussion and conclusions 5.1. Project impacts Our results show that Proyecto 2000 improved the quality of care on offer but did not directly increase the probability of delivery in Ministry of Health EmOC facilities. Nor was there an interaction between the system-level Proyecto 2000 inputs and the household- level SMI Program. Though they targeted the same sub-population, each program operated independently. The only behavioral impact we document is that of the SMI Program. It shows, simply, that reducing out-of- pocket costs increases EmOC utilization. The poorest Peruvian women clearly benefited from the targeted insurance program, however, the household risk fac- tor effects remained consistently negative across the models, indicating that neither program significantly reduced socioeconomic or ethnic disparities in EmOC utilization. Behavioral impacts due to Proyecto 2000 may have been too weak to be detectable or may have occurred after the endline survey. As shown in the DHS data, the share of births delivered in Ministry EmOC facilities rose nationwide during the period. Looking at our sample, we also see increasingly positive period effects, represented by the slopes on the birth year dummies in our models. The forces propelling those increases were likely more decisive than any attributable to Proyecto 2000. A lagged Proyecto 2000 treatment effect would be plausible for two reasons. First, only about 40% of the women surveyed gave birth during the project’s most intensive second phase (2000–2002). Although the dummy variables for
  • 10. 230 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 birth year capture a rising probability of EmOC use, the majority of women interviewed may have been unaware of any local improvements when they made their birthing decisions, or any improvements made may not have been noticeable. Second, delivery behav- iors may be socially mediated. If so, the observation period may have been too short for social learning or other endogenous social processes to reach some theoretical threshold level of women. The data did not permit us to test this hypothesis, however, social forces are one possible source of the consistent cor- relation of birthing decisions within catchment areas captured by the random effects. Future studies would do well to explore these social aspects of maternal behaviors. 5.2. Limitations There are a number of methodological shortcomings in this study. The relatively rich quasi-experimental Proyecto 2000 data allowed us to estimate a treat- ment effect for that program. However, we lacked any kind of design for evaluating the SMI Program. Strong designs are needed in order to evaluate such tar- geted programs. A recent example was Gertler’s 2000 [35] evaluation of Mexico’s Progresa Program. In that study,Gertleruseddifference-in-differencesestimators and panel data from households in randomly sam- pled treatment and control areas to show the program increased school enrollment and health services utiliza- tion and improved health outcomes. Had panel data or even repeated cross-sectional data from the same catchment areas been available we might have detected household-level Proyecto 2000 treatment effects. Our study also faced obvious sampling problems. Attrition of the Proyecto 2000 facilities during Phase I and the replacement of 14 of the original control facilities with new ones at endline are likely sources of sample selection bias. If the attriting EmOC facil- ities were the stronger institutions then any treatment effect would be underestimated. We lacked the data necessary to assess this. The targeted nature of the two programs presents another potential source of bias in that the characteristics of facilities and households not given treatment are likely to differ from those that did receive treatments. We estimated Proyecto 2000 treat- ment effects using a propensity score balanced on just four observable covariates; many other, unmeasured covariates could differ systematically across the two groups. Regarding the SMI Program, the beneficiaries we observed may differ from other potential beneficia- ries in Proyecto 2000 catchment areas where the insur- ance program had not yet been implemented. A more general problem are background disturbances caused by the constantly evolving mix of EmOC services many Peruvian communities faced as public health services decentralized and to some extent recentralized. In this fluid policy environment, perceptions of EmOC qual- ity, perhaps the legitimacy of public health services in general, were in flux. 5.3. Policy implications Peru’s SMI Program proved an effective means of inducing high-risk women to use public EmOC facil- ities. We document here its short-term impacts. They show that cost is a significant barrier to many women. However, such subsidized programs are generally fis- cally unsustainable, particularly in poor countries. Fur- ther, they may not be efficacious. The subsidies could merely act as side payments for compliance and when the subsidies end, the desired behavior, here use of EmOC, may end too. The long-term sustainability of targeted subsidy programs is an area where more research is needed. Proyecto 2000 sought to induce greater EmOC uti- lization through more elaborate, technical strategies. It theorized that improving institutional quality of care, educating the public and working with communities would be sufficient to induce behavioral change. We lacked data on the latter but the data we did have showed the first goal was achieved. Improved quality, our results suggest, is not sufficient to change delivery behaviors. Something else is needed. Recently, Gilson [36] proposed a theory wherein trust, initially between client and provider and later between community and the state, is a necessary condition for communities to become healthier. For this to happen people must per- ceive the quality of care to be high and the public health services to be legitimate. If out-of-pocket cost is a bar- rier, then targeted subsidies may be warranted as an interim measure. Studies elsewhere have shown even the poorest people are willing to pay for health ser- vices they value [37,38]. Though our interaction term was insignificant, we encourage other researchers to test this hypothesis.
  • 11. M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 231 Acknowledgements Technical assistance for Proyecto 2000 was pro- vided by Pathfinder International with funding from USAID-Peru, Contract 527-0366-C5049-00. Support for our study was provided by the Bill and Melinda Gates Institute for Population and Reproductive Health, Department of Population and Family Health Sciences, Johns Hopkins Bloomberg School of Public Health. References [1] Arroyo-Laguna J. Greater autonomy for public hospitals in Lima and the consequences of that on the use of health ser- vices,1988–1997.PanAmericanJournalofHealth1999;6:301– 210. [2] Ministry of Health. Lineamientos de pol´ıticas del sector salud 1995–2000: hacia un sector salud con eficiencia, equidad y cal- idad. Lima: Ministerio de Salud; 1996. [3] Pan American Health Organization. Core select data indicators. Peru, 2004. [4] MEASURE DHS+. Country statistics: Peru. Demographic and Health Surveys, 2004. Website: http://www.measuredhs.com/ countries/country.cfm?ctry id=33 accessed 4/2004. [5] Hill K, Abou-Zahr C, Wardlaw T. Estimates of maternal mor- tality for 19/95. Bulletin of the World Health Organization 2001;79:182–93. [6] Pan American Health Organization. Regional strategy for maternal mortality and morbidity reduction, Document CSP26/14. Washington, DC: Pan American Health Organiza- tion; 2002. [7] Murray SF, Davies S, Kumwenda Phiri R, Ahmed Y. Tools for monitoring the effectiveness of district maternity referral systems. Health Policy and Planning 2001;16(4):353–61. [8] World Health Organization. Mother-baby package. Implement- ing safe motherhood in countries. Document FHE/MSM/94.11. Maternal health and safe motherhood programme. Geneva: World Health Organization; 1994. [9] Instituto Nacional de Estadistica e Informatica. Encuesta demografica y de salud familiar 1996. Lima: Instituto Nacional de Estadistica e Informatica, 1996. Accessed at: http://www. paho.org/English/SHA/coredata/tabulator/newTabulator.htm 23 October 2004. [10] Instituto Nacional de Estadistica e Informatica. Encuesta demografica y de salud familiar 2000. Lima: Instituto Nacional de Estadistica e Informatica; 2000. [11] Ross JA, Campbell OMR, Bulatao R. The maternal and neona- tal program effort index. Tropical Medicine and International Health 2001;6:787–98. [12] Futures Group. Herramienta para la incidencia politica en la salud materna: MNPI (indice de esfuerzo de programa materno y neonatal). Lima: Futures Group; 2002. [13] United Nations Development Programme. Human Develop- ment Reports: Human Development Indicators, 2003. Website: http://www.undp.org/hdr2003/indicator/cty f PER.html accessed 4/2004. [14] Elo IT. Utilization of maternal health-care services in Peru: the role of women’s education. Health Transition Review 1992;2:49–69. [15] Vallenas G. Maternal mortality in Peru. Revista Peruana de Poblacion 1993;3:33–56. Website: http://www.paho.org/ English/DD/AIS/cp 604.htm accessed 4/2004. [16] de Meer K, Bergman R, Kusner JS. Socio-cultural determi- nants of child mortality in southern Peru: including some methodological considerations. Social Science and Medicine 1993;36:317–31. [17] Averting Maternal Death and Disability Working Group on Indicators. Using UN process indicators to assess needs in emer- gency obstetric services: Pakistan, Peru and Vietnam. Inter- national Journal of Gynaecology and Obstetrics 2002;78:275– 82. [18] Meentzen A. Estrategias de desarrollo culturalmente adecuadas para mujeres indigenas. Unpublished document. Washington, DC: Unidad de Pueblos Indigenas y Desarrollo Comunitario. Banco Interamericano de Desarrollo, 2000. [19] Pan American Health Organization. Situacion de salud de los pueblos indigenas de Peru. Document OPS/HSP/HSO/98.08. Washington, DC: Pan American Health Organization; 1998. [20] Francke BP, editor. Politicas de Salud 2001–2006. Diagnos- tico y Propuesta 5. Lima: Consorcio de investigacion eco- nomica y social, 2001. Available at: http://www.consorcio.org/ CIES/html/diag5.asp. [21] Donnay F. Maternal survival in developing countries: what has been done, what can be achieved in the next decade. Interna- tional Journal of Gynaecology and Obstetrics 2000;70:89–97. [22] Donabedian A. The definition of quality and approaches to its assessment. Ann Arbor, Michigan: Health Administration Press; 1980. [23] McCarthy J, Maine D. A framework for analysing the deter- minants of maternal mortality. Studies in Family Planning 1992;23:23–33. [24] Benavides B, Seclen JE, Novata J, Velasquez A. Impacto del mejoramiento continuo de la calidad en los servicios materno perinatales del Peru. La experiencia del Proyecto 2000. Unpub- lished document. Lima: Pathfinder Fund, 2000. [25] Seclen J, Benavides B, Jacoby E, Velasquez A, Watanabe E. Existe una relacion entre los programas de mejora de la calidad y la satisfaccion de usuarias de atencion prenatal? Experiencia en hospitals de Peru. Revista Panamericana de Salud Publica 2004;16:149–57. [26] EASURE DHS+. Peru 2000 DHSIV Final Report. FR-120. Calverton, MD, 2000. [27] Cureton EE, D’Agostino RB. Factor analysis: an applied approach. Hillsdale, NJ: Erlbaum; 1983. [28] Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41–55. [29] Imben G. Nonparametric estimation of average treatment effects under exogeneity: a review. Technical Working Paper
  • 12. 232 M.J. McQuestion, A. Velasquez / Health Policy 77 (2006) 221–232 293. Cambridge MA: National Bureau of Economic Research, 2003. http://www.nber.org/papers/T0294. [30] AbouZahr CL, Wardlaw T. Antenatal care in developing countries. Promises, achievements and missed opportunities. Geneva: World Health Organization; 2003. [31] Filmer D, Pritchett L. Estimating wealth effects without expenditure data or tears: an application to educational enrollments in states of India. Demography 2001;38:115– 32. [32] Rabe-Hesketh S, Skrondal A, Pickles A. GLLAMM manual. Paper 160. UC Berkeley Division of Biostatistics Working Paper Series. The Berkeley Electronic Press, 2004. http://www. bepress.com/ucbbiostat/paper160. [33] Stata Corporation. Stata statistical software. Release 6.0. Col- lege Station, TX: Stata Corporation; 1999. [34] Longford NL. Simulation-based diagnostics in random- coefficient models. Journal of the Royal Statistical Society A 2001;164(Part 2):259–73. [35] Gertler P. Final report. The impact of Progresa on health. Unpublished paper. International Food Policy Research Insti- tute. Washington, DC: International Food Policy Research Insti- tute, 2000. [36] Gilson L. Trust and the development of health care as a social institution. Social Science and Medicine 2003;56:1453–68. [37] Acharya L, Cleland J. Maternal and child health services in rural Nepal: does access or quality matter more? Health Policy and Planning 2000;15:223–9. [38] Ross G, Zeballos J, Infante A. La calidad y la reforma del sector salud en America Latina y el Caribe. Revista Panamericana de Salud Publica 2002;9:93–7.