Welfare impact of the Lady Health Workers programme in Pakistan.
Farooq, Shujaat ; Durr-e-Nayab ; Arif, G.M. 等
1. INTRODUCTION
With the year 2015 fast approaching, Pakistan is not likely to
achieve most of the health targets set in the Millennium Development
Goals [Pakistan (2010)]. High levels of child and maternal mortality and
child malnutrition are among the major health challenges facing the
country. Along with this enhanced vulnerability for children and women
there is also an economic divide in the society because these health
challenges are more profound for the poor segment of the population than
for the better off. Another divide is between the rural and urban
populations due to concentration of health facilities in urban centres
of the country. The high cost of dealing with health issues adversely
affects the poor and rural population, lowering their productivity and
limiting their lifetime achievements. Without substantially improved
health outcomes it is impossible to break out of the cycle of poverty
[OECD (2003)].
The government of Pakistan has taken several initiatives to improve
the health status of the population, particularly women and children,
and the Lady Health Workers (LHW) (1) programme is one such initiative.
The LHW programme was launched in 1994 with the core objective of
reducing poverty by providing essential primary health care services to
people at large and hence also improving the national health indicators.
The programme also envisaged to contribute to the overall health sector
goals of improvement in maternal, new-born and child health, provision
of family planning services, and integration of other vertical health
promotion programmes.
The performance of the LHW programme was evaluated by the Oxford
Policy Management (OPM) in 1999 and 2008-09. According to their 1999
report, the LHW programme has had a positive impact on the health
outcomes in its catchments areas.
These outcomes include childhood vaccination rates, reversible
methods of contraception (especially in rural areas), antenatal
services, provision of iron tablets to pregnant women, child growth
monitoring and control of childhood diarrhoea among the lower income and
poor households. The OPM 2008-09 evaluation report stated that the LHWs
have played a substantial role in preventive and promotive care and in
delivering some of the basic curative care to the communities, along
with providing referral to emergency and tertiary care [OPM (2009)].
The evaluations of the LHW programme, however, did not carry out an
in-depth analysis of the distributional impact of the programme. Health
and poverty nexus is well documented and the literature shows that a
family's wellbeing is strongly tied to the physical health of its
members (WHO, 2003). An effective intervention in the health sector
improves the delivery of health services, which impacts positively the
health status of a population. This improvement in the health status
affects the well-being of the people by enabling them to take benefit of
the available economic opportunities more efficiently.
The present paper aims to: analyse whether the LHWs serve the poor
and the vulnerable disproportionately; examine the contribution of the
LHW programme in improving child and maternal health; and analyse the
poverty reduction impact of the LHW programme. To achieve these
objectives the paper is organised into five sections. The next section
presents a very brief review of the health and poverty situation in
Pakistan. It also outlines the main features of the LHW initiative. The
data source and methodology used in the paper are discussed in section 3
followed by investigating whether the LHW programme has served the poor
disproportionally in section 4. The health seeking behaviour of the
beneficiaries (women visited by the LHWs) and non-beneficiaries (women
not visited by the LHWs) is examined in section 5. Section 6 explores
the impact of the LHW programme on the health outcomes of women and
children and their poverty status. The final section presents the
conclusions of the study and draws some policy recommendations.
2. HEALTH, POVERTY AND THE LHW INITIATIVE: A BRIEF REVIEW
Child and maternal health are considered important summary
indicators of the development of a country. MDGs 4 and 5 are related to
child and maternal health. Goal 4 is to reduce child mortality while
goal 5 is to improve maternal health. Pakistan has made some
improvements in the indicators related to these goals but the progress
remains slow and unsatisfactory. Table 1 presents data on child and
maternal health indicators, covering the 1990-91 to 2008-09 period along
with the MDG targets for 2015. Pakistan lags behind in achieving the
goals for two important indicators of child health. The first goal is to
reduce under-five mortality to 52 deaths per 1000 by 2015 from its
current level of 94 deaths per 1000. The second goal is reduction in
infant mortality to 40 deaths per 1000 live births from the current
level of 75 deaths per 1000 live births. It seems difficult to attain
both these goals by 2015. The performance for immunisation of children
and reduction in diarrhoea cases can, however, be considered
satisfactory (Table 1). The performance of indicators related to
maternal health shows that while Pakistan has made significant progress
in reducing maternal mortality from 533 maternal deaths per 100,000 live
births in 1990-91 to 276 in 2006-07 (Table 1), the achievement of the
target of 140 by 2015 seems difficult in such a short time. Similarly,
despite an improvement in the proportion of women using contraceptives,
receiving antenatal care services and delivering by skilled birth
attendants, the progress is slow in achieving the targets set for the
year 2015. A considerable decline in total fertility rate from 5.4 in
1990-91 to 3.8 in 2008-09 is not sufficient to achieve the target of
replacement level fertility (2.1 births per women) by 2015.
Table 1 also presents data on poverty trends and the MDG target for
2015. If we look at the findings of the PPHS we see a fluctuating trend
in poverty incidence, with poverty decreasing during the period 2001 to
2004 and increasing in 2010 from what it was in 2004 [Arif and Shujaat
(2014)]. This concurs with the erratic poverty trends shown in Pakistan
during the last five decades. While the poverty was very high in the
1960s (40 percent), it declined in the 1970s, and the declining trends
continued in the 1980s, reaching to a level of only 18 percent in
1987-88. Poverty, however, began to rise again in early 1990s till the
beginning of the new millennium when the headcount ratio was about 35
percent. In addition to the decline in economic growth the inflows of
foreign remittances, which are believed to be one of the major factors
reducing poverty during the 1970s and 1980s, also declined markedly
during the 1990s. There was a sharp decline in poverty during the first
half of the last decade, from 34.5 percent in 2000-01 to 22.3 percent in
2005-06, This declining trend continued and poverty dropped to a low
level of 12.4 percent in 2010-11 (2) (Table 1). In recent times the
economy of Pakistan has been facing severe challenges with a declining
rate of economic growth, double-digit inflation--particularly food
inflation, power shortage, soaring oil prices and poor law and order
situation. But the inflows of foreign remittances, which played a major
role in poverty decline in the past, have increased to more than US$ 10
billion per annum. Irrespective of the poverty estimates for the more
recent period, historical trends show instability in poverty reduction.
The strategy of poverty reduction in Pakistan on the one hand has
focused on sustained high economic growth and on the other hand it gives
equal importance to income transfers as well as investment in human
capital by improving health and education indicators. In health sector
initiatives (3) the LHW programme is unique in terms of its objectives,
coverage and provision of services to women and children. The core
objective of the programme is reduction in poverty by providing
essential primary health care services to mothers, new-borns and also to
improve child health, provision of family planning services, and
integration of other vertical health promotion programmes. It began with
the strength of a little over 30,000 LHWs in the mid-1990s and over the
years it has expanded to a strength of over 100,000 LHWs currently
deployed in all districts of the country. The selection criteria for a
LHW include: female should preferably be married; be permanent resident
of the area for which she is recruited; has minimum 8 years of schooling
preferably matriculate; should be between 20 to 50 years; preference
will be given to women with past experience in community development and
willingness to carry out the services from home. Rural areas and the
communities living in urban slums across the entire country are the
targeted areas/communities of the LHW programme. The coverage of LHW
programme is reported as 83 percent in 2008-09, according to the 2013-14
Pakistan Economic Survey. Although a large number of LHWs are stationed
in each district of the country, the programme, however, does not exist
in hard to reach areas of some districts. The main constraints for
non-coverage are nonfunctional health facilities and unavailability of
women meeting the selection criteria set for recruitment as LHWs
[Pakistan (2011)].
The LHWs provide services to communities at their doorstep. They
also act as a liaison between a community and the formal health system
and ensure support from NGOs and other departments. The LHWs coordinate
with other maternal and health care providers (i.e. midwives,
traditional birth attendants and local health facility) in the community
for appropriate antenatal and postnatal services. The LHWs are also
responsible for making nutritional interventions such as anaemia
control, growth monitoring, accessing common risk factors causing
malnutrition and nutritional counselling. LHWs also provide treatment
for common diseases, for which they are provided with inexpensive drug
kits.
3. DATA AND METHODOLOGY
This study adopts a mixed approach by combining qualitative and
quantitative methods to accomplish its objectives. The main reason for
combining these approaches is that the latter is best suited to measure
levels and changes brought by an intervention and for drawing inferences
from observed statistical relations between those changes and other
covariates. The quantitative analysis is, however, less effective in
understanding processes--that is, the mechanisms by which a particular
intervention triggers a series of events that ultimately result in the
observed impact. (4) For the quantitative part the study uses a
multipurpose panel dataset generated by the Pakistan Institute of
Development Economics (PIDE) in August-December, 2010, named as the
Pakistan Panel Household Survey (PPHS) covering both rural and urban
areas in 16 districts of the country. The districts are: Attock,
Hafizabad, Faisalabad, Vehari, Bahawalpur and Muzaffargarh in Punjab;
Badin, Mirpur Khas, Nawabshah and Larkana in Sindh; Dir, Mardan and
Lakki Marwat in Khyber Pakhtunkhwa (KP); and Loralai, Khuzdar and Gwadar
in Balochistan. The 2010 PPHS is the third round of the panel survey.
The first and seconds rounds, named as the Pakistan Rural Household
Survey (PRHS), were carried out in 2001 and 2004 respectively only in
rural areas [for more details see Nayab and Arif (2014)]. A health
module was included in 2001 and 2010 rounds of the panel survey. This
study has used these two datasets; but for the impact analysis, it has
relied primarily on the 2010 PPHS. (5)
The units of analysis are the ever married women in the
reproductive ages (1549 years) and children under-five in the survey
sample as they mainly comprise the target population of the LHW
programme. In the 2010 PPHS as well as in the 2001 round, women in the
sampled households were asked whether their household was visited by an
LHW in three months preceding the survey and if yes what was the
frequency of her visit. Based on LHW visits, two methods have been
adopted to divide the sampled women and children into two broad
categories: the beneficiaries, and the non-beneficiaries. The first
method uses the household level data where the beneficiaries are those
households that were visited by the LHWs during the reference period;
and the non-beneficiaries include those households that were not visited
by the LHWs. The second method relates to the village level LHW visits
where the beneficiaries are those villages where LHWs on average have
visited 20 percent or more of the households during the reference
period; and the non-beneficiaries are those villages where on average
less than 20 percent of the households were visited by the LHWs. The
PPHS 2010 survey did not have the relevant community level information
and since LHWs are deployed at the village level the second method was
devised to overcome this shortcoming.
Using the household and village level visits of LHWs, the
quantitative analysis is carried out in three steps. First, it examines
whether the LHWs serve the poor more than the rich. For this purpose a
simple analysis of calculating the proportions of beneficiaries (women)
by income quintile and the level of their educational attainment is
carried out. A multivariate analysis is also carried out with a binary
dependent variable--the beneficiaries (or visited by LHWs=l) and
non-beneficiaries (not visited=0):
P([X.sub.i]) = Prob ([D.sub.i] = 1|[X.sub.i]) = E(D|[X.sub.i]) ...
(1)
Where
P ([X.sub.i]) = F(h ([X.sub.i]))
F (h ([X.sub.i])) can be the normal or the logistic cumulative
distribution
[D.sub.i] = 1 if beneficiary and 0 otherwise (non-beneficiary)
[X.sub.i] is a vector of pre-treatment characteristics.
The second step relates to the investigation of the health seeking
behaviour of the beneficiary and non-beneficiary women, focusing on the
use of contraceptives, antenatal care, place of delivery of last birth,
and child immunisation. Lastly, in the third step the paper estimates
the impact of the LHW programme on maternal and child health related
indicators and poverty level by the method of propensity-score matching
(PSM) developed by Rosenbaum and Rubin (1983).
However, it is not straightforward to compute the welfare impact of
the LHW programme for the non-beneficiary sample. Taking the mean
outcome of the non-beneficiary women as an approximation is not
advisable as the beneficiaries and non-beneficiaries usually differ in
socio-economic characteristics even in the absence of the programme, and
such a process could lead to a selection bias [Kopeinig (2008)]. The PSM
is one of the possible solutions to solve this selection bias problem
with the idea to find a comparison group that looks like the beneficiary
group in all aspects except one the comparison group does not benefit
from the programme [Ravallion (2003)].
In the PSM analysis, the beneficiaries of the LHW programme (women
as well as children) are the "treated units" while the
non-beneficiaries are "non-treated units". Beneficiaries are
matched to the non-beneficiaries on the basis of the propensity score by
meeting the two conditions. The first condition is the balancing of
pre-treatment variables given the propensity score, if p (X) is the
propensity score, then;
[D.sub.i] = [X.sub.i]| p([X.sub.i]). ... (2)
If the balancing hypothesis is satisfied, the pre-treatment
characteristics must be the same for both the beneficiary and
non-beneficiary groups. In other words, for a given propensity score,
exposure to benefit (or treatment) is a randomised experiment and,
therefore, beneficiary and non-beneficiary should be on average
observationally identical. The second condition is the un-confoundedness
given the propensity score. Suppose that assignment to beneficiaries is
un-confounded i.e.
[Y.sub.1],[V.sub.0] = [D.sub.i]|[X.sub.i]
= [D.sub.i]|p[X.sub.i]. ... (3)
When the assignment to beneficiaries is un-confounded conditional
on the variables before benefit (or treatment), assignment to
beneficiaries is un-confounded given the propensity score.
Using the Equation 1, first the propensity scores are calculated
through the logistic regression and then the Average Treatment on the
Treated (ATT) effects based on the propensity scores (Rosenbaum and
Rubin, 1983) are estimated as:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)
Where:
[Y.sub.1i] is the potential outcome if the individual is treated
(beneficiary), and
[Y.sub.0i] is the potential outcome if the individual is not
treated (non-beneficiary).
In order to make the working sample comparable, it has been
restricted to only those units with probabilities that lie within the
region known as the common support, that is, the area where there are
enough of both, control and treatment observations, to proceed with
comparison [Dehejia (2005)]. The PSM method has been applied on the PPHS
2010 micro dataset to analyse the impact of the LHW programme on
maternal and child health and poverty. For poverty impact of the LHW
programme, the consumption approach has been used by inflating the
official poverty line for 2010. (6)
A qualitative analysis was carried out to complement the
quantitative analysis of the present study. For this purpose fieldwork
was conducted in 10 localities of 8 selected districts of Pakistan
covering all the provinces. To cover the regional differences as much as
possible districts were selected to include the variations that may
exist in the functioning of the LHWs across the country. Only rural
areas were selected for the qualitative part of the study on the premise
that the LHWs programme is mainly rural based and has a more important
role to play in the rural areas than in urban areas. The selected
districts were: Attock (North Punjab), Hafizabad (Central Punjab),
Rajanpur (South Punjab), Mardan and Swabi (KP), Turbat (Balochistan) and
Badin and Mirpur Khas (Sindh). The qualitative analysis is focused on
four main areas of investigation regarding the LHWs programme including
coverage; delivery; advocacy; and hindrances/ suggestions for
improvement.
Two villages from each of the above mentioned districts were
selected, one having an LHW programme and the other being without it.
The latter was selected for the sake of comparison and to see how the
absence of the programme affected the community. The qualitative
information used in each of the selected villages is as follows:
(1) Villages with LHW Programme
(2) Focus group of women (beneficiary/non-beneficiary)
(3) Interview with LHWs
(4) Villages Without LHW Programme
(1) Focus Group of Women
Interviewers from the local areas, knowing local languages, were
hired to conduct both the FDGs and aforementioned interviews. Their
minimum qualification was masters and preference was given to those who
had previous field experience, especially to those who had the knowledge
of the LHW programme. Two one-day workshops were conducted in Islamabad
and Karachi to train the interviewers for the fieldwork. Interviewers
from Punjab and KP were given training in Islamabad while those working
in Sindh and Balochistan were trained in Karachi. The main purpose of
the qualitative fieldwork and its questions was explained to the
interviewers during the training. They were also made to understand the
functioning of the programme and the interview techniques used in the
field. The field notes were analysed for this study by the authors
themselves (see guides used in the field in Annex 2).
4. HAS THE LHW PROGRAMME SERVED THE POOR DISPROPORTIONATELY?
Table 2 sets out the data on the proportion of women visited by the
LHWs by quintile. It shows that the LHWs are certainly not covering only
the poor. As can be seen from Table 2, 50 percent of the poorest women
(quintile1) reported visit by an LHW as compared to 54 percent of the
5th (richest) quintile. From these figures it might be inferred that the
LHWs do not select their clients on the basis of their wealth or
economic status. This notion, however, is negated when we look at the
figures broken down by urban-rural residence. While the LHWs reach out
more to the poor (59 percent) than to the rich (44 percent) households
in the urban areas, the trend is reversed for the rural areas (see Table
2). Regarding the level of educational attainment, Table 2 shows that in
rural areas, LHWs visit slightly more the literate and educated women,
but, in urban areas more illiterate women are visited by the LHWs than
the literate/educated women, though the difference is small.
The socio-demographic and economic characteristics at village level
are given in the appendix, as Annex 3, in which the villages have been
divided into four categories according to the percentage of households
visited by LHWs. The four categories are based on the proportion of
households visited by LHWs in a village. The categories are: not visited
by LHW; below 20 percent of the households visited; below 50 percent of
the households visited; and 50 percent and above households visited.
Except for the number of children in the household there is no
consistent pattern of LHWs' allocation at the village level at the
various cut-off points, as can be seen by the trend shown by average
literacy, household size, poverty, landless and livestock less
households in a village (Annex 3).
The results of the multivariate analysis (logistic regression)
using the equation 1, where the dependent variable is 1 for the
beneficiaries and 0 for the non-beneficiaries, are presented in Table 3.
In model 1, the beneficiaries' status is defined at individual
level where the dependent variable is 1 if ever married women in the
reproductive ages (15-49 years) are visited by the LHWs in three months
preceding the survey and 0 otherwise. In model 2, the
beneficiaries' status is defined at village level where the
dependent variable is 1 if 20 percent or above of the households in a
village are visited by the LHWs and 0 otherwise. Model 1 shows that the
LHWs are more likely to visit women aged 26-35 years and less likely to
visit those who have passed their prime reproductive ages (i.e. 36-49
years) compared to women in the reference category of 15-25 years. This
variable is not significant in model 2. The literacy level of the women
and that of the head of the household are not statistically significant
regarding visits by the LHWs in model 1, but model 2 shows that literate
women are significantly less likely to be visited by LHWs (Table 3). On
the contrary, the household size has a significant positive impact on
the LHWs' visits as an increase of one member in the household
raises the probability of an LHW visit by 1.05 times in both the models.
In model 1, the presence of a child has a positive and statistically
significant impact on the LHW's visit- an important finding with
reference to the influence of the LHW programme on women and child
health (Table 3).
The effect of land ownership (in acres) and livestock (in numbers)
on an LHW visit is mixed in both the models as the impact of only large
animal is significant in model I but both the land and livestock are
statistically significant in model 2. Cemented structure (paced) of
houses show a significant negative association with the visit of an LHW
only in model 1 whereas houses with mixed structure have a positive
relationship with the LHW visits in both the models, as can be seen in
Table 3. The significant coefficient of urban or rural residence shows
that women in urban areas are more likely to be visited by LHWs compared
to the rural women in both models. Relative to the reference category
(the non-beneficiary women in Punjab), the women of two provinces, Sindh
and KP, are more likely to be visited by LHWs while the women in
Balochistan are less likely to be visited (see Table 3). Based on this
multivariate analysis, it can be safely concluded that women are not
generally selected by LHWs on the basis of their economic status as they
tend to serve all women and children, and there are no major differences
in the results of two models.
The findings conform to the qualitative research carried out to
complement the quantitative data. In-depth and focus group interviews
(FGD) done in all three districts of Punjab, namely Attock, Hafizabad
and Rajanpur, show that the LHW programme is serving people of all
segments of the population whether they are poor or non-poor. However,
most of the people who approach the LHWs themselves for consultancy or
medicine are poor as the affordability factor is a major issue for them
while seeking medical assistance. In Sindh and KP a similar trend was
found among the beneficiaries of the LHWs visits--LHWs in these
provinces provided services irrespective of beneficiaries' economic
standing. Beneficiaries in both districts of Sindh (Badin and Mirpur
Khas) and KP (Mardan and Swabi) mentioned that the LHWs of their areas
give equal importance to all the people. From the standpoint of both
beneficiaries and LHWs, in the district Turbat of Balochistan, the
programme is mainly targeting poor people. The unanimous view of the
interviewed LHWs was that for them everyone was equal and they are there
to serve everyone, whether they are poor or rich.
4. HEALTH SEEKING BEHAVIOUR OF BENEFICIARIES AND NON-BENEFICIARIES
Has the LHW programme influenced the health seeking behaviour of
women? The two rounds of the panel data, carried out in 2001 and 2010,
include a comprehensive health module, which includes the use of
contraception by married women, antenatal care during the last
pregnancy, and the use of ORS for diarrhoea among children. The use of
contraceptives among the beneficiary and non-beneficiary women is
reported in Table 4, which also shows information on the proportion of
women using modern methods of contraception. Overall, 35 percent of the
sampled women reported using 'any method' for contraception.
There is a difference between rural and urban areas. More urban women
use contraceptives than their rural counterparts. Difference can also be
seen in the contraceptive behaviour of the beneficiary and
non-beneficiary women. The beneficiary women have a CPR of 39 percent
while non-beneficiary women have CPR of 32 percent. This difference,
however, according to data presented in Table 3, is mainly in rural
areas where the contraceptive prevalence rate is 37 percent among the
beneficiary women as compared to 27 percent among the non-beneficiary
women. The use of modern methods is also higher among the beneficiary
women than among the non- beneficiary women, particularly in rural
areas. There is a marked improvement in the CPR from 2001 to 2010 period
showing a positive contribution of LHWs in the use of family planning
practices (Table 4).
The data on antenatal care are presented in Table 5 for two
periods, 2001 and 2010. As compared to three-quarters of the beneficiary
women in 2010, two-thirds of the non-beneficiary women received
antenatal care during the last birth, indicating positive impact of LHW
programme on women's health. The impact, however, is evident only
in rural areas. Irrespective of the LHW visit, approximately 80 percent
of urban women received antenatal care during the last birth. There is
an improvement among the beneficiary women between 2001 and 2010 in
rural areas in receiving antenatal care and a decline in giving birth at
home while there is no corresponding increase among the non-
beneficiaries women. There is a modest increase between 2001 and 2010 in
the proportion of beneficiary women who received tetanus injection
during the last pregnancy whereas a considerable decline has been
witnessed among the non-beneficiary women.
No major difference is found in the incidence of illness and
diarrhoea between children belonging to beneficiary and non-beneficiary
women (Table 6). However, in case of diarrhoea the use of ORS was higher
among the former than the latter. The use of traditional medicines
during diarrhoea illness was higher among children living in
non-beneficiary households. Child immunisation is universal but it is
slightly higher among the children of the beneficiary women than among
the children of the non-beneficiary women (Figure 1). In the 2010 PPHS,
while examining the health seeking behaviour during the illness of
children, the respondents were also asked about the first health service
provider consulted for treatment. The role of LHWs was negligible in
such consultation because LHWs may not be authorised to prescribe
medicine but may advise for the treatment of some diseases like
diarrhoea.
The qualitative part of the study supports the findings of the
household survey data and gives more information about variations across
the provinces. When the LHWs were interviewed regarding the kind of
services they offer, they said that they were performing all the
services that were part of their duties and responsibilities including
family planning services, child vaccination, advice on ORS making,
antenatal care, and basic information about hygiene. Some of the LHWs in
Hafizabad and Attock districts said that they give a practical
demonstration if the community does not understand their verbal
explanation, particularly in the case of ORS making.
In general women were satisfied with these services. This
satisfaction, however, was not universal as some women also showed
dissatisfaction for the services of the LHWs in their areas, as one
woman complained:
"Whenever she visits us she asks about family planning
services, or that if any woman is pregnant here? She does not tell us
anything else".
(A women in FGD held in Attock district).
The FGDs held in areas with no LHW programme came up with
interesting results. There was almost a consensus that women want the
LHW programme in their villages. The non-beneficiaries mentioned that
they have to go to private clinics for check-ups, but that is not
feasible for them as private facilities are expensive. They reported
that their children also do not get proper vaccination, as the polio
vaccination teams do not visit their village frequently. Women in such
areas had to opt for traditional birth attendants (dais) for deliveries,
and also seek family planning services from them which are not always
safe.
In Sindh, the LHWs reported a gradual change in the behaviour of
the local residents regarding maternal and child care, including
vaccination. The interviewed LHWs in Badin and Mirpur Khas districts
reported to be carrying out vaccination programme for children along
with telling the community about hygiene, family planning and maternal
health. They were satisfied with the changing attitude of the people,
like one of the LHW in Mirpur Khas said,
"They used to resist getting vaccinated but the community
agrees to get their children immunised now. Pregnant women are also now
ready to get vaccinated. It is our success and it is because of us that
this change is coming."
(An LHW in district Badin, Sindh)
Mixed results were found for the two districts of KP included in
the qualitative part of this study regarding the functioning of the LHW
programme. In district Swabi, the community shows a positive response
with the majority of the women satisfied with its functioning. Among the
most reported services delivered by the LHWs are registration of
pregnant women and new-born babies; frequent visits to expectant women;
and EPI vaccination. Since the LHWs reside in the villages, people have
access to them in time of need at their homes. In district Mardan,
however, the response in the FGD show that the community was not very
satisfied with the way the LHW programme was functioning. According to
the participants of the FGD, the LHWs of the area were not regular in
their visits. One respondent told that, "She is not performing her
duty well, the last time she visited us is one year ago". The LHWs
of the two districts of KP were also interviewed to know about the
services they were providing, and were found to narrate almost all the
duties assigned to them on paper. Regarding their irregular visits to
some of the areas they blamed the social milieu of the villages for it.
They said that female mobility is not easy in KP and in some areas it is
tougher than others.
"There are lots of problems in this area as people of this
area are not very cooperative. My in-laws do not allow me to visit
community on regular basis to deliver all the services I am supposed to
offer to the households. Women can, however, visit me in my home if they
are in need. They do come often for family planning methods and
medicines
(An LHW in district Mardan, Khyber Pakhtunkhwa)
In Balochistan, the vaccination of children against polio is one of
the major services delivered by LHWs. They perform their duties
efficiently and regularly and people report no complaints regarding this
task. Women in the district of Turbat complain that the LHW does not
provide them with any medicine. LHWs, on the other hand, reported that
they are not getting medicine supplies and people blame them for that.
Moreover they perceive that the LHWs are giving these medicines to their
relatives and friends only. One woman in the FGD said:
"She does not provide us with medicines. Whenever we go to her
she only has family planning pills and iron tablets and nothing
else".
(A woman in the FGD in district Turbat, Balochistan)
When women were asked about their accessibility to the LHWs, they
said that LHWs were accessible in their homes as well, even if they did
not visit, but they prefer going to the Rural Health Centre in that case
as the LHWs do not have medicine supplies at homes. The women argued
that if they have to go far to get medicines they can see a doctor there
as well.
Based on the above discussion one can conclude that the coverage of
the LHW programme is satisfactory and its scope is wide in terms of
advice for the health improvement of women and children. Regional
differences are, however, evident as the performance of the Programme in
Punjab and Sindh was reported to be better than in KP and Balochistan.
Security is one of the reasons for relatively poor performance in these
two provinces, along with the erratic supply of medicines hindering the
success of the programme. The qualitative study of the areas without an
LHW shows the need for enhancing the coverage of the programme to all
rural areas.
5. IMPACT ANALYSIS OF THE LHW PROGRAMME
For the impact analysis of the LHW programme, three sets of
variables related to the reproductive health of women, child health and
poverty status have been selected. The use of contraceptives, antenatal
care, vaccination (TT injection) and place of delivery represents
women's health outcomes while child immunization, illness, and
infant and child mortality are used for child health. The official
poverty line is used to see the welfare impact of the LHW programme.
Following the methodology given in Section 3, the propensity scores
and ATT effect are estimated by both the methods, which are
beneficiaries' status at the individual level and
beneficiaries' status at the village level. The results of equation
1 have been discussed in the previous section, showing that women are
not selected by the LHWs on the basis of their economic status, rather
the coverage seems to be universal within the target areas.
The results of Equation 4 are presented in Tables 7-9 with ATT
parameters under three measures, namely Nearest Neighbour (NN) Matching,
Kernel Matching, and Stratification Matching. The NN method matches each
treated unit (beneficiaries) with the controlled unit
(non-beneficiaries) that has the closest propensity score. The method is
usually applied with replacement in the control units. In the second
step, the difference of each pair of matched units is computed and
finally the ATT is obtained as the average of all these differences. (7)
In the Kernel and Local Linear methods, all the treated units
(beneficiaries) are matched with a weighted average of all non-treated
units (non-beneficiaries) using the weights which are inversely
proportional to the distance between the propensity scores of treated
(beneficiaries) and non-treated (non-beneficiaries). The stratification
matching method consists of dividing the range of variation of the
propensity score in a set of intervals (strata) such that within each
interval the treated (beneficiaries) and non-treated (non-beneficiaries)
units have the same propensity score on average [Rosenbaum and Rubin
(1983)]. Both types of standard errors, analytical and bootstrapped have
been reported in Tables 7-9, however, the Kernel matching method does
not estimate the standard error by default.
5.1. Impact of LHWs Programme on Women's Health
Table 7 presents the impact of the LHW programme on women's
health outcomes. The welfare impact has been calculated at the
individual and village levels. Table 7 shows that the ATT impact on the
use of contraceptives is only statistically significant at the
individual level by Kernel method with a welfare gain of 2.5 percent.
This positive effect reflects the contribution of the LHW programme in
enhancing the use of contraceptives by married women. As discussed
earlier, this is one of the focus areas for the LHWs, and even in the
FGDs some of the participant women complained about over emphasis of the
LHWs on contraceptive use.
Table 7 also shows a positive and significant ATT impact of the LHW
programme on the antenatal care under all the three measures in method 1
and two measures in method 2. Compared to the non-treated units
(non-beneficiary women) the treated units (beneficiary women) enjoy a
positive impact of 17.7 to 21.9 percentage points in method 1 and 8.3 to
12 percentage points in method 2. Both bivariate analysis and the FGDs
show positive contribution of the LHWs in antenatal care, particularly
in rural areas. The third column in Table 7 shows the results about
vaccination during the last pregnancy. The significant impact of the LHW
programme on this variable shows a positive gain through both the
methods and welfare measures, ranging from 10.6 percent to 22.9 percent.
However, the impact of the LHW programme on delivery in a hospital
is not statistically significant under all the three measures of ATT.
This probably reflects that the financial inability of the sampled women
to deliver in a hospital or impracticability of the distance involved to
travel to a hospital could be an obstacle. Preference of the women
themselves to deliver at home instead of at a health facility can not be
ruled out. These findings of the PSM analysis as well as the qualitative
analysis suggest that the LHWs have created goodwill in their target
areas and women do trust them for seeking advice regarding different
health issues.
5.2. Impact of LHW Programme on Child Health
The ATT effect of the LHW programme on the child health indicators
is computed on the basis of estimated propensity scores using the logit
regression (giving code 1 to children belonging to households and
villages visited by LHWs and 0 otherwise in model 3 and model 4,
respectively). The regression results presented in Table 8 do not show
any systematic preference for the LHWs, and the region and province
dummies seem to be the major differentiating factors. Children in Sindh
and KP provinces are more likely to be visited by LHWs than children in
Punjab while the likelihood of LHW visits reduces for the province of
Balochistan.
Table 9 presents the ATT effect of LHW programme on the child
health indicators by both the methods; at individual level and at
village level. The beneficiary children are more likely to be vaccinated
than the non-beneficiary children as indicated by all three measures of
ATT and by both methods. Child immunization campaigns comprise a major
work load for the LHWs nationwide. Because of their local residence and
good practices parents of the area seem to be relatively more willing to
immunise their children. The presence of LHW has a negative effect on
child illness but only under the stratification measure of ATT. Under
other two measures--Kernal and NN, the effect is not statistically
significant.
The impact of the LHW programme on infant and child mortality is
not statistically significant. The incidence of diarrhoea and
respiratory infection are the major causes of infant and child mortality
in Pakistan. The preventive role of LHWs has surely contributed in
reducing these causes of infant and mortality rate but their role has
not been great enough to reduce infant and child mortality in Pakistan.
5.3. Welfare Impact of the LHW Programme
Before presenting the findings of the PSM analysis regarding the
welfare impact of the LHW programme it is appropriate to discuss briefly
the changes in the poverty status of the households based on the panel
datasets. Figure 2 shows poverty statistics for rural and urban areas
for the year 2010 and 2001 when two rounds of the panel survey were
carried out. As noted earlier, this study uses the official poverty
line, inflating it in the year 2010.8 Two points are noteworthy from
this figure. First, there is no major difference between the beneficiary
and non-beneficiary samples in terms of their poverty status either in
2010 or in 2001 although rural poverty among the former is slightly
higher. Second, rural poverty between 2001 and 2010 period declined
sharply and it happened for both the beneficiary and non-beneficiary
samples. Since these simple poverty statistics are not sufficient to
gauge the welfare impact of the LHW programme we adopt the PSM
methodology that is well suited for the purpose.
Table 10 shows the estimated ATT on poverty status under the three
measures, namely NN, Kernal and stratification. The welfare impact of
the LHW programme is statistically significant under all these measures.
However, the impact varies across the three measures. At individual
level, it ranges from 4.1 to 5.3 percentage points with the lowest under
Kernel method and highest under the NN method while at village level,
the welfare impact ranges from 6.3 to 23.2 percentage points. The impact
is positive; the negative signs of the three measures show that the LHW
programme reduces the probability of being poor. Thus, the LHW
beneficiary women (and their households as well) are less likely to be
poor as compared to the non-beneficiary women who have similar
characteristics.
One logical question which is not under the scope of this study is
how the LHW programme has contributed to poverty reduction? Since the
poverty measure used in the PSM analysis is based on the consumption
approach the impact of the LHW programme would have been through an
increase in income and consumption of the beneficiary households. The
literature on health interventions and poverty suggests that an
improvement in women's health can lead to their higher
participation in the labour market which may in turn enhance their
well-being level. Has the LHW programme contributed in enhancing female
participation in the labour market? It depends on employment
opportunities for women and it requires an in-depth analysis. However,
the Labour Force Survey data do show an increase in female participation
in the labour market from 17 percent in 2001-02 to 27 percent in 2010-11
(LFS 2012). The LHW programme could be a contributory factor through
improving women health. Rural women, however, are working primarily as
unpaid family helpers (LFS 2012) and may not have control over the
resources earned through their engagement. Despite this, it would not be
wrong to presume that women participation as family helpers may be
viewed positively as it contributes to the household's strategy to
ensure food security and improve household well-being.
6. CONCLUSIONS
The government of Pakistan has taken several initiatives to improve
the poor health indicators in the country and the LHW programme is one
of such initiatives. With an aim to reduce poverty through an
improvement in the health status of population, particularly women and
children, the LHWs work at the grass root level. The LHWs are recruited
from the local communities to provide preventive health care services at
their door step. At present they are deployed in all districts of the
country and their services are available to more than half of the target
population.
In order to gauge the welfare impact of the LHW initiative, the
present study combines the quantitative and qualitative approaches. In
the quantitative analysis the multipurpose panel datasets, PRHS-2001 and
PPHS-2010, conducted by PIDE, are used. These datasets suit the
quantitative analysis because they have comprehensive modules on child
and maternal health and household consumption necessary for poverty
estimation. They also have a comprehensive module on the performance of
the LHWs. For the qualitative analysis, field work was conducted in ten
rural localities of the eight selected districts of Pakistan covering
all the four provinces.
The quantitative analysis of the panel datasets shows that slightly
more than half of the sampled women were visited by the LHWs during
three months preceding the survey. The analysis shows that the LHWs have
provided their services to all segments of society irrespective of their
income status. An improvement has been found in the health seeking
behaviour of the beneficiary women. The qualitative analysis supported
these findings.
The PSM methodology, that generates comparable samples of
beneficiaries and non-beneficiaries of the LHW programme, shows that the
LHW programme has a significant and positive impact on contraceptive
use, antenatal care and vaccination (TT) during pregnancy. The impact of
the LHW programme on child health has been evaluated by selecting four
indicators, which are child immunisation, child illness, and infant and
child mortality. A significant gain is observed in child vaccination and
child illness. However, the LHW programme does not show a significant
impact on infant and child mortality. The welfare impact of the LHW
programme in terms of reduction in poverty is found to be statistically
significant.
It appears from the findings of this study that the LHW programme
is a pro-poor initiative. Two factors probably have played key role in
its success, which are: recruitment of the LHWs from the communities
where they are assigned to work, and universalization of the programme
within the target areas--providing services to all women and children of
the covered areas.
Considering the positive impact the LHW programme has had on its
beneficiaries it is recommended that the programme may be extended to
all uncovered areas as well. This was also demanded by the
non-beneficiary women during the focus group discussions. Another factor
that can improve the effectiveness of the programme is enhanced training
of the LHWs and provision of medicines to them, especially in the
provinces of KP and Balochistan. Services provided by the LHWs include
family planning and antenatal and postnatal check-ups. Unfortunately,
irregular and delayed supply of medicines adversely affects their
functioning and creates mistrust among the LHWs and the women to whom
they provide the services.
In view the complaints of women at some of the study sites about
irregularity in the LHWs' visits, an effective supervision
mechanism is critical. Such a mechanism can help improve the service
delivery at the grass root level, further enhancing the positive impact
the programme has made. In order to sustain gains made by the programme
it should be made an integral component of the district health system
operating in the framework of Primary Health Care (PHC) and MNCH
programme. It will also help in formalising the service structure of the
LHWs, which is one of their long standing demands. Likewise, integration
with the PHC system will not only strengthen the LHW programme but also
help the recipients through a better referral system. As a result
everyone will benefit--the LHWs, the people and the health delivery
system at large.
APPENDIX
Annex 1
Households Covered in PRHS 2001 and PPHS 2010
PPHS 2010
PRHS 2001
Provinces (Rural only) Rural Urban Total
Pakistan 2721 2800 1342 4142
Punjab 1071 1221 657 1878
Sindh 808 852 359 1211
KP 447 435 166 601
Balochistan 395 292 160 452
Annex 2
Guide for Focus Group Discussion of Beneficiary
Wonien/Non-beneficiary Women in Area Having LHW Programme
* Knowledge about the program and source of knowledge- after an LHW
visited or before? If before, from whom/where?
* Frequency of LHW's visit.
* Any factors that hinder their visits ... weather/males/elders/any
other.
* Coverage of LHWs. Do they visit every household or the ones only
having women and children? Do you feel they are more inclined to visit a
certain kind of household than others (poor/vulnerable)?
* Kind of messages they give, (infant/child
health/immunisation/boiling water/ nutrition/antenatal
care/contraception/hand washing/hygiene/diarrhoea).
* Ease in understanding their given advice. Practicality in
following their advice and satisfaction level regarding it.
* Impact of their advice--any improvements in family health.
* Access to LHWs--ever approached them in case someone was ill in
the household or waited for their visit.
* For those who are not visited- have they ever tried making LHWs
visit them and reasons for non-visit in case they did not come to their
household.
Interview Guide for the Focus Group not having LHW Programme
* In the absence of LHWs, their source of fulfilling health needs,
including antenatal care/delivery/contraception.
* Any trouble accessing the health care service.
* Assess the contraception/immunisation rates, and who/where are
deliveries taking place.
* Questions judging their knowledge about hygiene/nutrition/boiling
water/ diarrhoea/ immunisation.
* Their knowledge about the LHW and MNCH programmes and desire to
have it in their village as well.
Interview Guide for LHWs
* Criteria to select the households they pick to visit.
* Frequency/regularity of the visits.
* Any hindrance in performing their duties.
* Access to the community when not visiting their homes themselves.
* Kind of advice given to the women they visit, and the method of
conveying the message--only verbally for do a demonstration as well.
* Availability of equipment/skills needed to perform their job.
* Satisfaction with their working conditions.
* Perception about their performance
* Suggestions for improvement
Annex 3
Average Socio-demographic and Economic Characteristics
at Village Level by the Status of LHW Visit
No visit <20% visit
Overal Rural Urban Overal Rural Urban
Characteristics 1 only only 1 only only
Literacy of
Head (%) 38.6 16.1 59.9 37.2 18.7 56
Household Size
(numbers) 7.4 8.6 6.3 7.1 8 6.3
Married Female
(15-49) in
household (%) 16.8 16.4 17.1 17.2 17 17.4
Children (under 5)
in household (%) 7.4 7.1 7.7 7.6 7.7 7.4
Poverty (%) 20.7 23.4 17.2 16 7 18.8 13.9
Landless
Households <%) -- 48.2 -- -- 52.9 --
Livestock less
Households (%) -- 34.4 -- -- 33 --
Number of Villages 31 17 14 44 24 20
<50 % visit 50 and above % visit
Overal Rural Urban Overal Rural Urban
Characteristics 1 only only 1 only only
Literacy of
Head (%) 41.7 33.6 57.3 45.3 43.1 50.5
Household Size
(numbers) 7 7.3 6.4 7.9 8.1 7.4
Married Female
(15-49) in
household (%) 17.4 17.6 17.1 16.5 17.1 15.1
Children (under 5)
in household (%) 8.7 9.4 7.5 9.3 9.9 8.1
Poverty (%) 21 24.6 13.2 20.6 21.3 18.8
Landless
Households <%) -- 61.4 -- -- 52.1 --
Livestock less
Households (%) -- 39 -- -- 29.3 --
Number of Villages 86 54 32 129 87 42
Source: Authors' computation from the micro
datasets of 2010 PPHS.
REFERENCES
Ajakaiye, O. and G. Mwabu (2007) The Demand for Reproductive Health
Services: An Application of Control Function Approach. Paper presented
to the 12th Annual Conference of the African Econometric Society.
University of Cape Town, South Africa.
Arif, G. M. and Shujaat Farooq (2011) Poverty Inequality and
Unemployment in Pakistan. Backgroung study for IDB group MCPS Document
for Pakistan. DhulQa'dah 1432H.
Caliendo, Marco and Marco Kopeinig (2008) Some Practical Guidance
for the Implementation of Propensity Score Matching. Journal of Economic
Survey 22:1, 31-72.
Dehejia, Rajeev (2005) Practical Propensity Score Matching: A Reply
to Smith and Todd. Journal of Econometrics 125, 355-364.
Ilafeez, A., B. K. Mohamud, M. R. Shiekh, S. A. 1. Shah, and R.
Jooma (2011) Lady Health Workers Programme in Pakistan: Challenges,
Achievements and the Way Forward. Journal Pakistan Medical Association
61:3, 210-215.
Heckman, J. and J. Smith (1999) The Pre-programnie Earnings Dip and
the Determinants of Participation in a Social Programme: Implications
for Simple Programme Evaluation Strategies. Economic Journal 109:457,
313-348.
Ministry of Health (2011) www.mnch.gov.pk/dp.php last retrieved on
March 2011.
Ministry of Health (2011) www.mnch.gov.pk/strategies.php last
retrieved on March 2011.
Nayab and G. M. Arif (2012) Pakistan Panel Household Survey: Sample
Size, Attrition and Socio-demographic Dynamics. Pakistan Institute of
Development Economics, Islamabad. (Poverty and Social Dynamics Paper
Series, PSDPS-1).
OPM (2009) Lady Health Worker Programme, External Evaluation of the
National Programme for Family Planning and Primary Health Care, Final
Report. Oxford Policy Management.
Organisation of Economic Cooperation and Development (OECD) (2003)
Poverty and Health in Developing Countries: Key Actions, Policy Brief,
OECD Observer.
Oxford Policy Management (OPM) (2002) Lady Health Worker Programme,
External Evaluation of the National Programme for Family Planning and
Primary Health Care, Final Report. Oxford Policy Management.
Planning CommissionPakistan, Government of (2010) Pakistan
Millennium Development Goals Report 2010. Islamabad: Planning
Commission, Centre for Poverty Reduction and Social Policy Development
Islamabad, Government of Pakistan.
Pakistan, Government of (2011) Economic Survey of Pakistan 2010-11.
Islamabad: (2011) Economic Advisor Wing. Government of Pakistan.
Pakistan, Government of (2011) Economic Survey of Pakistan 2010-12.
Islamabad: (2012) Economic Advisor Wing. Government of Pakistan.
PC-1 (2003) National Programme for Family Planning and Primary
Health Care 'The Lady Health Workers Programme" 2003-2008.
Government of Pakistan. Ministry of Heath.
PC-1 (2006) National Maternal Newborn and Child Health (MNCH)
Programme. Government of Pakistan. Ministry of Heath.
Ravallion, Martin (2003) Assessing the Poverty Impact of an
Assigned Programme. In Francois Bourguignon and Luiz Pereira da Silva
(eds.) The Impact of Economic Policies on Poverty and Income
Distribution: Evaluation Techniques and Tools. New York: Oxford
University Press.
Rosenbaum, Paul and Donald Rubin (1983) The Central Role of the
Propensity Score in Observational Studies for Causal Effects. Biometrika
70, 41-55.
Smith, J. and P. Todd (2005) Does Matching Overcome LaLonde's
Critique of Nonexperimental Estimators? Journal of Econometrics 125:1-2,
305-353.
United Nations (2011) The Millennium Development Goals Report 2011.
United Nations.
United Nations (2012) The Millennium Development Goals Report 2012.
United Nations.
World Bank (2011) World Development Indicators. Washington, D.C.
World Health Organisation (WHO) (2003) Poverty and Health: A
Strategy for the African Region. WHO, Regional Office for Africa,
Brazzaville.
(1) LHWs provide primary healthcare services, including disease
prevention, cure and rehabilitative services, and family planning, in
rural areas and urban slums, with more than 75 percent of the population
served by LHWs living in rural areas. LHWs reside in the locality they
serve and their homes are called health houses. Each LHW covers
approximately 200 houses, which is an average of over 1200 individuals.
LHWs are supposed to visit each household in their assigned area at
least once a month. Each LHW is attached to a government health
facility, from which they receive training, a small allowance, and
medical supplies. Provincial and district coordinators monitor and
supervise the LHWs. The average annual salary of LHWs is $343 who are
not allowed to engage in other paid activities. 110,000 currently
deployed. Target is 150,000.
(2) However, this figure has been reported in the 2013-14 Pakistan
Economic Survey as an interim indication of poverty situation in the
country.
(3) The health programme includes Expanded Programme on
Immunisation, AIDS Control Programme, Malaria Control Programme,
National T.B. Control Programme, National Programme for Family Planning
and Primary Health Care, National Programme for Prevention and Control
of Blindness, National Maternal Newborn and Child Health Programme,
Cancer Treatment Program, Drug Abuse, Dengue Epidemic and Control
Program and Food and Nutrition Programmes [Pakistan (2012)].
(4) Vijayendren Rao and Michael Woolcock; Integrating Qualitative
and Quantitative Approaches in Programme Evaluation.
(5) For the sample size, see Appendix Table 1.
(6) The 2010 PPHS has a comprehensive consumption expenditure
module. For more detail, see Arif and Farooq (2012).
(7) The NN method may face the risk of bad matches if the closest
neighbour is far away. Such risk can be avoided by imposing a tolerance
level on the maximum propensity score distance (caliper). Hence, caliper
matching is one form of imposing a common support condition where bad
matches can be avoided and the matching quality rises. However, if fewer
matches can be performed, the variance of the estimates increases
[Caliendo and Kopeining (2008); Smith and Todd (2005)].
(8) For more detail on poverty line, see Arif and Farooq (2012).
Shujaat Farooq <
[email protected]> is Assistant Professor
at the Pakistan Institute of Development Economics, Islamabad.
Durr-e-Nayab <
[email protected]> is Chief of Research at the
Pakistan Institute of Development Economics, Islamabad. G. M. Arif
<
[email protected]> is Joint Director at the Pakistan Institute
of Development Economics, Islamabad.
Table 1
The Performance of Health Sector and Poverty Situation in Pakistan
1990- 2001- 2004-
Indicators 91 02 05
Poverty incidence 26.1 34.5 23.9
MDG indicators related to reducing infant and child mortality
<5 mortality
117 n/a n/a
Infant mortality rate
102 77 77
Proportion of fully immunized children
(12-23 months) 75 53 77
Proportion of 1 year children immunized
against measles 80 57 78
Proportion <5 suffered from diarrhoea 26 12 14
MDG indicators related to improve maternal health
Maternal Mortality Ratio * 533 350 400
Proportion of skilled birth attendance 18 40 48
Contraceptive prevalence rate
12 28 n/a
Total fertility rate 5.4 n/a n/a
Proportion made at least 1 antenatal
check-up (for births in last 3 years) 15 35 50
2005- 2006- 2007-
Indicators 06 07 08
Poverty incidence 22.3 n/a 17.2
MDG indicators related to reducing infant and child mortality
<5 mortality
n/a 94 n/a
Infant mortality rate
76 75 n/a
Proportion of fully immunized children
(12-23 months) 71 76 73
Proportion of 1 year children immunized
against measles 76 77 76
Proportion <5 suffered from diarrhoea 12 11 10
MDG indicators related to improve maternal health
Maternal Mortality Ratio * 380 276 n/a
Proportion of skilled birth attendance 35 37 40
Contraceptive prevalence rate
n/a 29.6 30.2
Total fertility rate n/a 41 3.85
Proportion made at least 1 antenatal
check-up (for births in last 3 years) 52 53 56
MDG
2010- Target
Indicators 11 2015
Poverty incidence 12.4 (a) 13
MDG indicators related to reducing infant and child mortality
<5 mortality 89
(2012-13) 52
Infant mortality rate 66
(2014) 40
Proportion of fully immunized children
(12-23 months) 81 >90
Proportion of 1 year children immunized
against measles 82 >90
Proportion <5 suffered from diarrhoea 11 <10
MDG indicators related to improve maternal health
Maternal Mortality Ratio * 260 140
Proportion of skilled birth attendance 47 >90
Contraceptive prevalence rate 35
(2012-13) 55
Total fertility rate 3.6 2.1
Proportion made at least 1 antenatal
check-up (for births in last 3 years) 58 100
Source: Government of Pakistan (2010), Pakistan Millennium
Development Goals Report, Planning Commission, GOP,
Islamabad.
Note: *--MMR estimates, like in most other places similar to
Pakistan, are very uncertain, with a wide range of error.
(a) : These figures may be considered interim indication of
poverty situation in the country, according to the Pakistan
Economic Survey 2013-14.
Table 2
Proportion of Beneficiary Women in 2001 and 2010 (%)
PPHS2010
PRHS 2001
Quintile Total Urban Rural Rural Only
Qi 50.2 59.3 47.0 13.2
Q2 53.3 41.2 58.7 15.1
Q3 55.7 54.6 56.2 14.9
Q4 53.9 54.5 53.6 21.8
Q5 54.0 44.1 57.0 21.4
Level of Educational Attainment
No education 52.2 53.0 51.9 16.1
Primary 60.8 48.5 67.8 25.7
Middle 58.3 50.0 65.4 27.7
Secondary 57.1 49.2 64.8 27.1
Higher 55.1 51.6 59.2 28.6
All 53.7 51.8 54.4 17.5
Source: Authors' computation from the micro datasets
of 2001 (PRHS) and 2010 PPHS.
Table 3
Determinants of Lady Health Workers ' Visits--Odd Ratio
Model 1
Correlates Odd Ratio Std. Error
Age of Woman (15-25 as reference)
26-35 1.154 ** 0.098
36-49 0.781 * 0.070
Literacy of woman (yes=1) 0.955 0.080
Literacy of household head (literate=l) 0.957 0.067
Household size 1.047 * 0.009
Presence of a child (yes=1) 1.301 * 0.160
Sex of household head (male=1) 1.361 ** 0.248
Land owned (acres) 0.997 0.003
Large animals owned (numbers) 1.065 * 0.014
Small animals owned (numbers) 0.993 0.009
Structure of House (Katcha as reference)
Pacca 1.064 0.097
Mix 1.212 * 0.114
Region (urban=l) 1.178 ** 0.100
Province (Punjab as reference)
Sindh (yes=1) 1.985 * 0.165
KP ((yes=1) 1.771 * 0.184
Balochistan (yes=1) 0.089 * 0.015
LR chi2 (12) 816.44
Log likelihood -2683.85
Pseudo [R.sup.2] 0.13
N 4,515
Model 2
Correlates Odd Ratio Std. Error
Age of Woman (15-25 as reference)
26-35 1.041 0.158
36-49 1.039 0.165
Literacy of woman (yes=1) 0.759 ** 0.110
Literacy of household head (literate=1) 0.979 0.129
Household size 1.046 * 0.016
Presence of a child (yes=l) 1.403 0.338
Sex of household head (male=l) 1.212 0.386
Land owned (acres) 0.986 * 0.005
Large animals owned (numbers) 1.150 * 0.039
Small animals owned (numbers) 0.945 * 0.013
Structure of House (Katcha as reference)
Pacca 0.551 * 0.095
Mix 1.628 * 0.307
Region (urban=l) 0.488 * 0.070
Province (Punjab as reference)
Sindh (yes=1) 8.372 * 2.046
KP ((yes=1) 2.229 * 0.496
Balochistan (yes=1) 0.014 * 0.003
LR chi2 (12) 1949.5 (16)
Log likelihood -1021.6017
Pseudo [R.sup.2] 0.4883
N 4,517
Source: Authors' computation from the micro datasets of
2010 PPHS.
* significant at 5 percent, ** significant at 10 percent.
Table 4
The Contraceptive Prevalence Rate by Status of LHW Visit and Region (%)
PPHS2010 PRHS 2001
Contraception Urban Rural Total Rural Only
Beneficiaries (Visited by LHW)
Using contraceptives 41.6 37.2 38.5 29.3
Using modern method 26.8 23.8 24.6 14.3
Non-beneficiaries
(No one Visited)
Using contraceptives 40.8 26.9 31.5 17.7
Using modern method 28.9 16.0 19.8 10.4
Source: Authors' computation from the micro datasets of
2001 PRHS and 2010 PPHS.
Table 5
Women Receiving Antenatal Care during the Last Pregnancy by
Status of LHW Visit and Region (%)
2010 2001
Antenatal Care Urban Rural Total Rural Only
Beneficiaries (Visited by LHW)
Received antenatal care 78.9 73.9 75.2 61.7
Received TT injections 83.9 83.4 83.5 80.6
Delivered at home 32.3 49.9 45.0 65.0
(Non-beneficiary) Not Visited
Received antenatal care 81.3 61.3 66.7 50.8
Received TT injections 69.0 46.8 54.1 66.1
Delivered at home 48.4 66.1 60.2 69.6
Source: Authors' computation from the micro datasets of
2001 PRHS and 2010 PPHS.
Table 6
Use of ORS for Diarrhoea by Status of LHW Visit and Region
ORS 2010 2001
Total Urban Rural Rural Only
Beneficiary (Visited by LHW)
ORS 51.08 61.22 48.35 51.32
Home-made fluids 9.09 4.08 10.44 3.95
Medicines 29 18.37 31.87 30.26
Traditional Medicine 5.63 8.16 4.95 5.26
None of the above 5.19 8.16 4.4 9.21
Total 100 100 100 100
Non-beneficiary (No one Visited)
ORS 42.74 53.85 41.35 44.71
Home-made fluids 6.84 7.69 6.73 6.83
Medicines 29.91 1 29.81 37.54
Traditional Medicine 11.97 7.69 12.5 7.85
None of the above 8.55 0 9.62 3.07
Total 100 100 100 100
Source: Authors' computation from the micro datasets of
2001 PRHS and 2010 PPHS.
Table 7
Average Treatment Effects of the LHW Programme on
the Reproductive Health of Women Aged 15-49 Years
Contraceptive Use Antenatal Care
Method (Yes=1) (Yes=1)
Method 1 (at individual level)
Nearest Neighbour
ATT 0.027 0.219
N. Treated 2548 2548
N. Control 1037 503
Standard Error 0.018 0.030
t-stat 1.474 7.246
St. Error Bootstrap 0.022 0.035
t-stat 1.223 6.276
Kernel
ATT 0.025 0.177
N. Treated 2548 2548
N. Control 1945 1945
St. Error Bootstrap 0.014 0.026
t-stat 1.711 6.710
Stratification
ATT 0.020 0.187
N. Treated 2548 2548
N. Control 1947 1947
Standard Error 0.014 0.017
t-stat 1.432 10.994
St. Error Bootstrap 0.014 0.022
t-stat 1.381 8.428
Method 2 (at village level)
Nearest Neighbour
ATT 0.058 0.077
N. Treated 3788 3788
N. Control 285 145
St. Error Bootstrap 0.064 0.122
l-stat 0.904 0.633
Kernel
ATT 0.025 0.083
N. Treated 3788 3788
N. Control 724 724
St. Error Bootstrap 0.036 0.046
t-stat 0.687 1.80
Stratification
ATT 0.006 0.120
N. Treated 3788 3788
N. Control 724 724
St. Error Bootstrap 0.046 0.068
t-stat 0.132 1.756
IT Injections Place of Delivery
Method (Yes=l) (Hospital=l)
Method 1 (at individual level)
Nearest Neighbour
ATT 0.135 0.070
N. Treated 2548 2548
N. Control 309 308
Standard Error 0.035 0.037
t-stat 3.883 1.895
St. Error Bootstrap 0.040 0.044
t-stat 3.347 1.608
Kernel
ATT 0.126 0.030
N. Treated 2548 2548
N. Control 1945 1945
St. Error Bootstrap 0.031 0.032
t-stat 4.037 0.326
Stratification
ATT 0.131 0.004
N. Treated 2548 2548
N. Control 1947 1947
Standard Error 0.016 0.018
t-stat 8.332 0.238
St. Error Bootstrap 0.026 0.038
t-stat 5.064 0.111
Method 2 (at village level)
Nearest Neighbour
ATT 0.229 -0.005
N. Treated 3788 3788
N. Control 118 118
St. Error Bootstrap 0.107 0.110
l-stat 2.137 -0.046
Kernel
ATT 0.117 0.090
N. Treated 3788 3788
N. Control 724 724
St. Error Bootstrap 0.063 0.088
t-stat 1.851 1.028
Stratification
ATT 0.106 0.048
N. Treated 3788 3788
N. Control 724 724
St. Error Bootstrap 0.061 0.133
t-stat 1.736 0.360
Source: Authors' computation from the micro datasets of 2010 PPHS.
Table 8
Determinants of Lady Health Worker Visits--Odd Ratio
Model 3
Correlates Odd Ratio Std. Error
Sex of child (male=1) 1.049 0.085
Number of children at home 0.921 * 0.037
Sex of household head (male=1) 1.316 0.312
Education of household head (in years) 1.010 0.009
Number of married women in the household 0.917 0.066
Household size 1.040 * 0.016
Land ownership (acres) 0.998 0.004
Large animals owned (numbers) 1.023 ** 0.013
Small animals owned (numbers) 1.008 0.012
Structure of House (Katcha as reference)
Pacca 1.214 ** 0.131
Mix 1.381 * 0.161
Region (urban=1) 1.521 * 0.167
Province (Punjab as reference)
Sindh (yes=1) 1.971 * 0.190
KP (yes=1) 4.523 * 0766
Balochistan (yes=1) 0.019 * 0.007
LR chi2 711.1 (15)
Log likelihood -1808.1813
Pseudo [R.sup.2] 0.164
N 3,333
Model 4
Correlates Odd Ratio Std. Error
Sex of child (male=1) 0.973 0.120
Number of children at home 0.993 0.061
Sex of household head (male=1) 0.189 * 0.138
Education of household head (in years) 1.018 0.014
Number of married women in the household 0.875 0.106
Household size 1.072 * 0.024
Land ownership (acres) 0.992 ** 0.004
Large animals owned (numbers) 1.088 * 0.023
Small animals owned (numbers) 0.964 * 0.013
Structure of House (Katcha as reference)
Pacca 0.529 * 0.086
Mix 3.115 * 0.641
Region (urban=1) 0.716 * 0.103
Province (Punjab as reference)
Sindh (yes=1) 6.088 * 1.166
KP (yes=1) 1.000 --
Balochistan (yes=1) 0.010 * 0.002
LR chi2 1929.68(14)
Log likelihood -938.60784
Pseudo [R.sup.2] 0.507
N 3,893
Source: Authors' computation from the micro
datasets of 2010 PPHS.
* significant at 5 percent, ** significant
at 10 percent.
Table 9
Average Treatment Effects of Propensity Score
Matching on Child Health Indicators
Immunization
Received Child Illness
Measures/ATT (Yes=1) (Yes=1)
Method 1 (at Individual Level)
Nearest Neighbour
ATT 0.066 -0.013
N. Treated 2157 2157
N. Control 643 642
Standard Error 0.025 0.031
t-stat 2.609 -0.411
St. Error Bootstrap 0.020 0.036
t-stat 3.290 -0.352
Kernel
ATT 0.072 -0.025
N. Treated 2157 2157
N. Control 1166 1166
St. Error Bootstrap 0.015 0.021
t-stat 4.690 -1.163
Stratification
ATT 0.063 -0.042
N. Treated 2157 2157
N. Control 2141 2141
Standard Error 0.015 0.021
t-stat 4.172 -1.980
St. Error Bootstrap 0.015 0.019
t-slat 4.275 -2.203
Method 2 (at Village Level)
Nearest Neighbour
ATT 0.121 -0.025
N. Treated 3146 3146
N. Control 244 246
St. Error Bootstrap 0.053 0.063
t-stat 2.283 -0.390
Kernel
ATT 0.103 0.034
N. Treated 3146 3146
N. Control 677 677
St. Error Bootstrap 0.038 0040
t-stat 2.726 0.845
Stratification
ATT 0.135 0.019
N. Treated 3138 3138
N. Control 685 685
St. Error Bootstrap 0.056 0.047
t-stat 2.386 0.411
Infant Mortality Child Mortality
Measures/ATT (Yes=1) (Yes=1)
Method 1 (at Individual Level)
Nearest Neighbour
ATT -0.001 -0.001
N. Treated 2157 2157
N. Control 650 650
Standard Error 0.001 0.001
t-stat -1.424 -1.424
St. Error Bootstrap 0.001 0.001
t-stat -1.288 -1.469
Kernel
ATT -0.001 -0.001
N. Treated 2157 2157
N. Control 1166 1166
St. Error Bootstrap 0.001 0.001
t-stat -1.230 -1.360
Stratification
ATT -0.001 -0.001
N. Treated 2157 2157
N. Control 2141 2141
Standard Error 0.001 0.001
t-stat -1.396 -1.396
St. Error Bootstrap 0.001 0.001
t-slat -1.263 -1.258
Method 2 (at Village Level)
Nearest Neighbour
ATT -0.001 -0.001
N. Treated 3146 3146
N. Control 262 262
St. Error Bootstrap 0.001 0.001
t-stat -1 111 -1.290
Kernel
ATT -0.001 -0.001
N. Treated 3146 3146
N. Control 677 677
St. Error Bootstrap 0.001 0.000
t-stat -1.147 -1.315
Stratification
ATT -0.001 -0.001
N. Treated 3138 3138
N. Control 685 685
St. Error Bootstrap 0.000 0.000
t-stat -1.169 -1.194
Source: Authors' computation from the micro datasets of
2001 PRHS and 2010 PPHS.
Table 10
Average Treatment Effects Under various Measures of
Propensity Score Matching on Poverty, PPHS 2010
Method
Nearest Kernel Strati-
ATT Neighbour fication
Method 1 (at Individual Level)
ATT -0.053 -0.041 -0.048
N. Treated (number of observation) 2548 2548 2548
N. Control (number of observation) 1153 1945 1947
St. Error Bootstrap 0.022 0.011 0.017
t-statistics -2.401 -3.630 -2.835
Method 2 (at Village Level)
ATT -0.232 -0.063 -0.114
N. Treated (number of observation) 3788 3788 3788
N. Control (number of observation) 313 724 724
St. Error Bootstrap 0.056 0.037 0.052
t-statistics -4.110 -1.690 -2.198
Source: Authors' computation front the micro
datasets of the 2010 PPHS.
Figure. 1. Proportion of Children Immunized by Status of
LHW Visit and Region
Visited by LHW 89.89 93.94 88.38 86.21
No one visited 82.78 84.87 82.16 72.43
Note: Table made from bar graph.
Fig. 2. The Incidence of Poverty Among the Beneficiary and
Non-beneficiary Rural Sample (%)
Beneficiaries 22.7 15.2
Non-beneficiaries 26.0 18.6
Source: Authors' computation from the micro-data of
2001 PRHS and 2010 PPHS.
Note: Table made from bar graph.
