IUSSP Conference 2013, Busan, S Korea Rural Health ...

26
Rachana Patel, PhD (IUSSP 2013) Page 1 IUSSP Conference 2013, Busan, S Korea Rural Health Facility and Institutional Birth: A study in Composite Index formation and spatial modeling Session 208: Spatial approaches to estimation of demographic rates Date: 29-08-2013 at 15:30pm-17:00pm 1. INTRODUCTION Maternal and child health programs in India have undergone various stages of planning and intervention for strengthening rural health services in order to facilitate institutional delivery (MOHFW 2005). It is well established that giving birth under the care and supervision of trained health-care (especially at health institution) providers promotes child survival and reduces the risk of maternal mortality (Tsui et al. 1997; WHO 2004a, 2005). Both, child mortality (especially neonatal mortality) and maternal mortality remain high in India and seven out of every 100 children born in India die before reaching age one (Dyson et al. 2004); and approximately five out of every 1,000 women who become pregnant die of causes related to pregnancy and childbirth (MOHFW 2005). Institutional birth has been increasing over the period but still much below the desired level. DLHS III estimates that national average for institutional births was 47 percent during 2007-08. Yet more than 50 percent of births in India continue to take place at home, most of them without the assistance of any trained health worker (DLHS 2007-08), threatening the lives of both mother and child. Also, there is a wide gap in the proportion between rural and urban. The proportion of births delivered at health institution in urban area is 71 percent while only 38 percent in rural area. There is clear evidence of high inter-state variations. The estimates of institutional birth in the weaker States in the north and central India are very high compared to southern and western region States. The proportion of births delivered in institution is 68 percent in rural area of the major southern states together (Maharashtra, Andhra Pradesh, Tamilnadu, Kerala and Karnataka), while only 31 percent in northern states (DLHS, 2007-08). Several studies have stressed the importance of access to health services as a factor affecting the utilization of services (Kumar et al. 1997, Nathan J at. al. 2004, Amy J Kesterton et. al. 2010). Availability and quality of healthcare services is yet an important aspect for encouraging healthcare utilization, particularly public health facilities, as evident from the fact that programs which integrate quality as well as access to services enhance client satisfaction, leads to greater utilization (Shelton and Davis 1996; Koenig and Khan 1999). Rani et al. (2008) have noted that

Transcript of IUSSP Conference 2013, Busan, S Korea Rural Health ...

Page 1: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 1  

IUSSP Conference 2013, Busan, S Korea

Rural Health Facility and Institutional Birth: A study in Composite Index formation

and spatial modeling

Session 208: Spatial approaches to estimation of demographic rates Date: 29-08-2013 at 15:30pm-17:00pm

1. INTRODUCTION

Maternal and child health programs in India have undergone various stages of planning and

intervention for strengthening rural health services in order to facilitate institutional delivery

(MOHFW 2005). It is well established that giving birth under the care and supervision of trained

health-care (especially at health institution) providers promotes child survival and reduces the

risk of maternal mortality (Tsui et al. 1997; WHO 2004a, 2005). Both, child mortality (especially

neonatal mortality) and maternal mortality remain high in India and seven out of every 100

children born in India die before reaching age one (Dyson et al. 2004); and approximately five

out of every 1,000 women who become pregnant die of causes related to pregnancy and

childbirth (MOHFW 2005). Institutional birth has been increasing over the period but still much

below the desired level. DLHS III estimates that national average for institutional births was 47

percent during 2007-08. Yet more than 50 percent of births in India continue to take place at

home, most of them without the assistance of any trained health worker (DLHS 2007-08),

threatening the lives of both mother and child. Also, there is a wide gap in the proportion

between rural and urban. The proportion of births delivered at health institution in urban area

is 71 percent while only 38 percent in rural area. There is clear evidence of high inter-state

variations. The estimates of institutional birth in the weaker States in the north and

central India are very high compared to southern and western region States. The proportion of

births delivered in institution is 68 percent in rural area of the major southern states together

(Maharashtra, Andhra Pradesh, Tamilnadu, Kerala and Karnataka), while only 31 percent in

northern states (DLHS, 2007-08).

Several studies have stressed the importance of access to health services as a factor affecting the

utilization of services (Kumar et al. 1997, Nathan J at. al. 2004, Amy J Kesterton et. al. 2010).

Availability and quality of healthcare services is yet an important aspect for encouraging

healthcare utilization, particularly public health facilities, as evident from the fact that programs

which integrate quality as well as access to services enhance client satisfaction, leads to greater

utilization (Shelton and Davis 1996; Koenig and Khan 1999). Rani et al. (2008) have noted that

Page 2: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 2  

poor quality of antenatal care is likely to reduce its utilization and as far as ante-natal care is

concerned the quality of service in southern states is superior to than those of northern states in

India. In addition to expanding health-care facilities and infrastructure, India's family welfare

program has been emphasizing outreach programs, including home visits, mobile clinics, and

community-based delivery systems, as mechanisms to increase both the quantity and quality of

services (MOHFW 2005).

Information on the spatial distribution of adequate infrastructure at the public health facility and

service utilization in a district/county is of interest to policymakers and researchers for a number

of reasons. First, it can be used to quantify suspected regional disparities in public infrastructure

standards and identify which areas are falling behind in the process of health improvement even

in the presence of government special health program (NRHM) in all over the focused states.

Second, it facilitates the targeting of programs, such as available and easy access to health center,

and infrastructure aid, whose purpose is to improve utilization from the end users. In many

countries, the main sources of information on spatial patterns of utilization are national/state

household surveys. Geographic objective could be most efficient when the geographic units are

quite small, such as a village or district. The only household information usually available at this

level of disaggregation is national household and health surveys of country. The staggered

economy and huge population demand have had great repercussions on India's health system.

With the exception of few southern regions, and a few urban areas, there is a marked shortage of

equipment and qualified personnel for meeting the need of maternal care. The country had an

estimated 61 allopathic doctors per 1,00,000 population and of the total available doctors 52

percent were from southern states of Andhra Pradesh, Goa, Karnataka, Travancore-Cochin,

Maharashtra and Tamilnadu while MCI Delhi contributed only 5 percent (Medical Council of

India MCI, 2007). The quality of healthcare undoubtedly depends on health facility adequacy for

infrastructure, manpower, equipments, and stock of essential drugs.

Information on the spatial distribution of adequate infrastructure at the health facility,

community education, share of urban population and service utilization in a district/county could

be utilized for the spatial analysis. However, advances in health geography have improved our

understanding of the role played by geographic distribution of health services on access to health

services (Arcury et al. 2005; Luo 2004). Keeping the foregoing discussion in view in this paper

makes an attempt to evaluate adequacy and accessibility of health facilities in north India and

Page 3: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 3  

investigate how it determine the level of institutional delivery. The results of the study shall

provide key policy input for improving the level of institutional delivery and achieving stipulated

UN, Millennium Development Goals of reducing the maternal mortality ratio by three quarters

between 1990 and 2015 (WHO, 2012). First, this study deals with the distribution and inequality

in the public health infrastructure in the districts of EAG states however, apart from the

infrastructure there could be more impediments at district/county/village level hindering the

utilization for maternal care services provided by government health policy. Secondly, spatial

analysis assumes that the relationship between progress and utilization is homogenous and

uniform over space. Overlooking the spatial correlations may or may not bias the model results

depending on the magnitude of such correlations over time. The organization of the paper is as

follows. The next section outlines the health system structure in rural India and recent programs

of the government. This is followed by description of data sources, then a section on

methodology and ends with a section results and discussion.

2. RURAL HEALTH PROGRAMS AND INFRASTRUCTURE

Infrastructure means something that lies below or comes before the structure and is the end result,

or, in some sense, the aim of development and progress. In public health and social-studies,

broadly speaking, ‘infrastructure’ could be seen as all those activities and services whose

contribution to the socio-economic development is not the income generated within the sector

directly but the sustenance and support they provide to the progress and social development in the

society or community. In the view of that Government had launched National Rural Health

Mission (NRHM) in 2005 to strengthen the MCH program for rural area; under decentralization

scheme and Panchayati Raj for the primary health were included under the umbrella. National

Rural Health Mission (NRHM) has provided the opportunities to develop a standard for Sub

Centers (SC), Primary Health Centers (PHC) and Community Health Centers (CHC) in the

country. Under NRHM more emphasis has been given upon the Empowered Action Group (EAG)

states because of their poor health indicators. It is therefore expected that the quality and

standards of care provided by the PHCs in the EAG states will improve and more adequately

satisfy the IPHS standard to reach the level of the non-EAG states in health performance

indicator. The main aim of NRHM is to provide accessible, affordable, accountable, effective and

reliable primary health care, especially to poor and vulnerable sections of the population. It also

aims at bridging the gap in Rural Health Care through creation of a cadre of Auxiliary Nurse

Page 4: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 4  

Midwife (ANM) and Accredited Social Health Activists (ASHA) and improves hospital care,

decentralization of programme to district level to improve intra and inter-sectoral convergence

and effective utilization of resources. The Mission further seeks to build greater ownership of the

programme among the community through involvement of Panchayati Raj Institutions, NGOs and

others to progress more.

The health care infrastructure in rural areas has been developed as a three tier system and is based

on population norms as SC will cover 5,000 population in plain area and 3,000 in hilly area while

PHCs are supposed to cover 30,000 and 20,000 population respectively in plain and hilly area and

CHC will cover in plain and hilly area in 120,000: 80,000 ratio. To strengthen and to improve the

facilities in the existing rural health infrastructure under Reproductive and Child Health

Programme, the Government of India has assisted all the States in improving/ constructing labor

room, operation theatre and providing water/ electricity supply in CHCs/ PHCs etc. so that

essential and emergency obstetric services are improved. These SC, PHC and CHCs are the keys

of MCH program which need to look after in efficient way.

From a programmatic and policy perspective, connecting peoples’ perceptions of health services

and health care delivery system characteristics can contribute to our understanding of utilization

behavior in a more comprehensive manner. Majority of studies included environmental variables

which measured only urban-rural location, or region, which may be imprecise proxies for more

specific measures such as supply of services (Phillips, et al. 1998). Hence, characteristics such as

physician supply and availability of physicians in the community would be important contextual

variables to be considered within the health services utilization model (Andersen, et al. 1996).

Such decisions should be made after analysis and conscious deliberation.

3. DATA SOURCE AND OBJECTIVES

District Level Household Survey 2007-2008 (DLHS 3) data on health facility and village was

used for the purpose. DLHS-3 was a nationally representative survey of households and health

facility at district level. Facility questionnaires were designed to collect information on

manpower, medicines, equipments and infrastructure for all levels of health facilities. These are

Health Sub-Centre (HSC), Primary Health Centre (PHC), Community Health Centre (CHC) and

District Hospitals (DH). Further details of the survey design could be obtained from the country

report (DLHS- III, 2007-08).

Page 5: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 5  

District-level means of the characteristics were obtained from the Census 2011 and inserted into

this equation, generating estimates of rural service utilization for each of the rural districts from

the selected states. The earlier studies have some limitations like it did not generate unbiased

estimates of district-level utilization. Recent DLHS 2007-08 report and some studies utilizing

that data has estimated the maternal health service utilization in urban and rural areas of each of

the 34 states in India. Unlike the earlier utilization-mapping analysis, this study uses household-

level national data, spatial determinants of the utilization and estimates using the spatial

methods. This chapter uses the district-level institutional birth estimates from previous chapter to

investigate the extent to which variables may have an spatial effect on the incidence of

institutional birth in a district. It was decided not to carry out the analysis of geographic

determinants of institutional birth at the village level (PSU) for two reasons. First, the village-

level institutional birth estimates have large standard errors, indicating a large “noise”

component in these estimates. Second, some of the variables may be less accurate at the village

level due to less sample size. Interpolation at the district level is probably more reliable than

interpolation at the village level.

This study is restricted to rural areas of eight socio-economically under developed states in north

India. The analysis is based on 8787 HSCs, 3269 PHCs and 5743 villages from 263 districts of

these states. All facility information was merged with village information. So the sample size was

5687 (villages) for preparation of facility indices at district level. Since, all EAG states comprise

263 districts so the districts are unit of analysis for the spatial analysis. Dependent variable is

institutional births and independent variables are categorized into two groups i.e. intermediate

(external) and direct (internal). All the indicators are computed from the DLHS-3 data and some

of the values like urban percentage, women literacy and SC/ST population were verified with the

census 2001 data. Household, women, village and facility file were used for calculating the

district level data.

Spatial proximity for accessibility studies has traditionally been defined through measures of

Euclidean distance where buffers around health centres and/or villages define travel thresholds

(McLafferty, 1988; Rosero-Bixby, 2004). Hospital choice also depends upon the services

available at the facility. The available studies till now are rarely discussed the correlation/spatial

association of availability of facility infrastructure itself. The present study tries to build on this

gap firstly, in order to examine the infrastructure availability at public health centre and its

Page 6: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 6  

adequacy for the institutional births. More specifically, we prepare the adequacy indices

separately at primary level health facility which are purposefully structured to promote the

utilization. Secondly, spatially weighted regression captures the spatial dependence with the

covariates.

4. METHODOLOGY

4.1 Health Facility adequacy indices:

The research tool employed in the present study is somewhat based on the scale and score

provided by Haddad, Fournier and Potvin (1998) to assess quality of healthcare services after

making adjustment for Indian setting and availability of data on facility information. Since, the

newly revised IPHS (PHC) has considered the services, physical infrastructure, manpower,

equipments and drugs so as to describe minimum assured services and the ideal level services

which the states shall try to achieve. Required infrastructure adequacy for the maternal care was

first aim to access facilities available at Health Sub-Centre (HSC) and Public Health centre

(PHC) using facility survey of DLHS 3, as these HSCs and PHCs are set-up in rural area to

facilitate decentralized government health program (NRHM, 2005) and to meet the maternal

health care need at the gross-root level. In the view to emphasize to take maternal care to the

door step by strengthening SHCs and PHCs, there is every need to scrutinize adequacy of these

facilities to ensure complete utilization. Essential equipments/instruments, manpower and drugs

etc required for birth delivery, was selected with the help of gynecologist in the institute (Dr.

Ambekar, IIPS, Mumbai). Availability of gynecologist, pharmacist, technician, nurses,

equipments for delivery, essential drugs, electricity, functional OT etc. are coded as 1 and 0

otherwise.

On basis of appropriate variables indices of adequacy of health facilities are prepared and

categorized into different sections for the districts. STATA version-10 software was used for

performing non-spatial statistical analysis while spatial analysis was done in GeoDa. The indices

were named as ‘manpower index,’ ‘physical Infrastructure index,’ ‘Essential equipments and

laboratory services index,’ and ‘essential drug index’ at PHC and similar at HSC with addition of

skilled ANM and ANM residing within 5km from the village.

Page 7: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 7  

4.1.1 Primary Health Center (PHC) and Health sub-Center (HSC) level indices: At the

village level PHCs should have proper infrastructure and adequate manpower, equipments and

essential drugs to provide all services related to maternal and child health care including three

major components of pre-natal, delivery and post natal care to the pregnant women. Separate

indices relating to functioning of PHCs and HSCs are outline in the following:

1. Manpower index (MI_P): Availability of trained health professionals at health center is

the primary requirement for institutional birth delivery. Study by López-Cevallos D F et.al.

(2010) provided evidence that density of public health practitioners was positively associated

with health care utilization in rural area. In the view of this composite index for manpower

required for delivery at health facility, was prepared. Manpower, required at health facility for

institutional delivery are medical officer, lady medical officer, staff nurse, pharmacist, lady

health veteran/health assistant, laboratory technician, auxiliary nurse midwife (ANM)/female

health worker, additional staff nurse. However, for the purpose of composite index of manpower

four essential personnel for institutional delivery namely availability of lady medical officer,

health assistant, pharmacist and any ANM are considered.

2. Physical infrastructure index (PII_P): Adequacy of physical infrastructure is crucial in

performing institutional delivery. This makes it important to construct an index of infrastructure

in accessing the implication of adequacy of health facility on maternal care. The items includes

in the construction of infrastructure index are: proper building for PHC, regular water supply,

regular electricity supply, functioning toilet, working phone, Boyler available, at least four bed

for patients, functional labor room, anesthesia, functional OT and communication facility.

3. Essential delivery care equipments/ laboratory services index (ELABI_P): Availability

of selected furniture, instruments, equipments and essential laboratory services required for natal

and delivery care at the health facility is considered for this index.

Selected furniture and instruments are: examination table, delivery table, OT table, bed side

screen, footstep, shadow less lamp light for OT/labor room, Macintosh for labor & OT table,

oxygen trolley with cylinder and flow meter, instrument trolley, sterilization instrument,

instrument cabinet, blood/saline stand, stretcher on trolley, stool for patients, wheel chair,

almirah/cupboard and separate dustbin for biomedical waste.

A number of equipments and kits should be available in the health facility for conducting

delivery. Additionally many storage system and instruments are necessary. Basic equipments

Page 8: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 8  

includes IUD insertion kit, normal delivery kit, equipment for assisted vacuum delivery and

forceps delivery, equipment for new born care and neonatal resuscitation, standard surgical set

(for minor procedures like episiotomy stitching), equipment for manual vacuum aspiration, baby

warmer/incubator and second, cold chain equipment comprises of ILR large/small, deep freezer

large/small, cold box and vaccine carrier. Additional lab requirement for Hb testing, reagent

strips for urine albumin and urine sugar analysis, rapid plasma regain (RPR) test kit for syphills

kit, reagent for peripheral blood smear examination for MP, residual chlorine in drinking water

testing strips, centrifuge, light microscope and binocular microscope.

Laboratory provisions for blood grouping, haemogram (TLC/DLC), diagnosis of RTIs/STDs

(with wet mounting, grams, stain etc.), sputum testing for TB, blood smear examination for

Malaria Parasite, urine (routine culture/sensitivity/microscopy ), rapid tests for pregnancy, rapid

plasma reagin (RPR) test for syphilis are considered in the construction of this index. All these

selected furniture, instruments, equipments, kits, cold storage devices and essential pathological

kits are included in the construction of equipments/lab services index (ELABI_P. Available

items are coded as 1 and 0 otherwise.

4. Essential drug index (EDI_P): Essential drugs, namely availability of antiallergics and

drugs used in anaphylaxis, anti-hypertensive , anti-diabetics, anti-anginal, anti-tubercular, anti-

leprosy, anti-filarials, anti-bacterials, anti-helminthic, anti-protozoal, antidots, solutions

correcting water and electrolyte imbalance and essential obstetric care drugs are considered in

the development of essential drug index.

4.2 Reliability test of Indices: Cronbach’s Alpha (Inter-Item Reliability):

Reliability of the health facility adequacy indices, discussed in the preceding section are tested

by Cronbach’s alpha. Its value ranges between 0 and 1. The closer is the Cronbach’s alpha

coefficient to 1.0 the greater is the internal consistency of the items included in the index. The

size of alpha is determined by both the number of items in the scale and the mean inter-item

correlations.

Based upon the formula α= rk / [1 + (k -1)r] where k is the number of items considered and r is

the mean of the inter-item correlations the size of alpha is determined by both the number of

items in the scale and the mean inter-item correlations. George and Mallery (2003) provide the

following rules of thumb: “_ > 0.9 – Excellent, _ > 0.8 – Good, _ >0 .7 – Acceptable, _ >0 .6 –

Page 9: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 9  

Questionable, _ > 0.5 – Poor, and _ < 0.5 – Unacceptable” (p. 231). While increasing the value

of alpha is partially dependent upon the number of items in the scale, it should be noted that this

has diminishing returns. It should also be noted that an alpha of 0.8 is probably a reasonable

goal. It should also be noted that while a high value for Cronbach’s alpha indicates good internal

consistency of the items in the scale, it does not mean that the scale is uni-dimensional and factor

analysis is a method to determine the dimensionality of a scale.

4.3 Principle component analysis (PCA):

Principle component analysis (PCA) is used to examine the structure of the relationship among

items included in the construction of the above health facility adequacy indices. Prior to running

the factor analysis, the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and the

Bartlett’s test of sphericity were performed. An “eigen value greater than 1” criterion was

employed for determining the number of factors. In order to obtain more interpretable results

solution, Varimax rotation was used to rotate the solution. This caused the loadings to be

distributed among the selected factors making it easier to interpret results. STATA version 10

software was used for principle component analysis. Later, PCA scores obtained from the

infrastructures variables at PHC and HSC so that to create 5 adequacy quintiles.

4.4 Spatial autocorrelation:

According to Anseline and Bera (1998), spatial autocorrelation can be loosely defined as the co-

incidence of value similarity with location similarity.

i. Moron’s statistics: The Morons’ scatter plot provides a tool for visual exploration of

spatial autocorrelation (Anseline 1996, 2002). This statistic is used to quantify the degree of

spatial autocorrelation present in the data set across all the districts. Univariate and bivariate

Moran’s I will give the spatial structure in terms of spatial autocorrelation (SAC). Pearson

coefficient measure of SAC is given as

 ∑

Where zj is standardized variable of interest at location i. wij is weight matrix C = and N is

number of spatial unit. Negative (positive) values indicate negative (positive) spatial

Page 10: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 10  

autocorrelation. Values ranges from -1 (indicating perfect dispersion) to +1 (perfect correlation)

a zero indicates a random spatial pattern (Moron, 1948).

ii. Local Indicators of spatial association (LISA) statistics: Univariate and bi-variate LISA

statistics will be used for the purpose which measures the extent of spatial non-stationarity and

clustering to its neighborhood values.

Where observations zi, zj are in deviations from the mean from ith location to jth location, and the

summation over j is such that only neighboring values j Є Ji are included. For ease of

interpretation, the weights wij may be in row standardized form, though this is not necessary, and

by convention, wii = 0.

4.5 Spatial Weighted Regression Analysis (SWR):

The spatial regression analysis carried out in this study to estimate the institutional delivery as a

function of variables representing socio-economic development, accessibility and adequacy

indices in the districts. As discussed above, we are also interested in examining the geographic

determinants of institutional birth. The dependent variable in this analysis is, itself, an imputed

value, so special care must be taken in interpreting the results, but Elbers, Lanjouw, and Lanjouw

(2004) show that the basic results are essentially the same as they would be with a “true”

measure of institutional birth. Estimation strategy is done with ordinary least-squares (OLS)

model will be estimated with all exogenous variables included; later tests for the two types of

spatial dependence will be performed and lastly, either the spatial error or the spatial lag model

will be used to re-estimate the model using generalized least squares. Spatial weights were

adopted that are proportional to the inverse distance between the geographic centers of the

districts. The spatial lag dependence model can be written as follows:

    ε  with ε λ  ε υ

where yi is the dependent variable for location i,

σ is the spatial autoregressive coefficient,

wij is the spatial weight reflecting the proximity of i and j,

yj is the dependent variable for location j,

,jj

ijii zwzI

Page 11: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 11  

Xi is a row vector of explanatory variables for location i,

β is a column vector of coefficients, and εi is the error term for location i.

The spatial weights matrix w describes the degree of proximity between each pair of spatial

observations. Usually it is a binary variable based on whether the two locations are contiguous or

a continuous variable based on some function of the distance between the two locations. If the

regression analysis is carried out without adjustment for spatial lag dependence, the estimated

coefficients will be biased and inconsistent (Anselin 1988).

The second type of problem is spatial error dependence. When there is spatial error

dependence, ordinary least squares regression coefficients will be unbiased but not efficient (the

standard errors will be larger than they would be if all information were used). This model can be

written as follows:

    λ    ε u

where yi is the dependent variable for location i,

Xi is a row vector of explanatory variables for location i,

β is a column vector of coefficients,

εi is the error term for location i,

λ is the spatial error autoregressive coefficient,

wij is the spatial weight reflecting the proximity of i and j, and

ui is the uncorrelated portion of the error term for location i.

In this case, using ordinary least squares to estimate the model does not yield biased coefficients,

but the estimates of the coefficient are not efficient and the standard t and F tests will produce

misleading inference (Anselin 1988). In order to test for the presence of spatial autocorrelation,

Moran’s I is frequently used:

I = (x – μ)′W(x – μ)/(x – μ)′(x – μ)

Where x is a column vector of the variable of interest,

μ is the mean of x, and W is the weighting matrix.

This statistic is simply the correlation coefficient between x at one point in space and the

weighted average of the values of x nearby. In order to test whether there is spatial lag

Page 12: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 12  

dependence or spatial error dependence, the Lagrange multiplier is used to test the statistical

significance of the spatial autocorrelation coefficient (λ) in the two models. Anselin (1988)

shows that the model with the larger coefficient (λ) is likely to be the appropriate model.

Whenever spatial error or spatial lag dependence is indicated, special types of generalized least-

squares (GLS) regression models need to be applied. In the case of spatial error dependence, the

spatial error model is appropriate, whereas in the case of spatial lag dependence, the spatial lag

model would be used. The independent variables are listed in Table 1 below.

Table 4.1 Variables descriptionExogenous variables (indirect) Endogenous variables (direct) Percentage of literate women Percentage with lowest quintile adequacy at

PHC (exclude doctors) Percentage urban Percentage with doctor

Percentage SC/ST Percentage women with at least 3 Ante-Natal Care (ANC)

average population covered by SC/PHC Average distance to nearest health centre providing ANC care

Percent women received conditional cash transfer (JSY beneficiary)

Average distance to nearest health centre providing delivery care

Percentage with all weather road connectivity to health centre

Percent lowest quintile population

5. RESULTS AND DISCUSSIONS

5.1 Reliability test of Indices:

Table1 5.1 The first subscale with Cronbach alpha 0.72 included 13 items related to ‘manpower

index (MP_P): adequate availability of health personnel at PHC. The second subscale, ‘physical

Infrastructure index’ (PII_P) with Cronbach alpha 0.79 comprised eighteen items: building for

PHC services, regular water, electricity supply, functional toilet, communication mode

(telephone, vehicle on road), adequacy of beds for patients, clean hospital premises, and proper

disposal of waste etc. The third subscale, ‘Essential delivery care equipments/ laboratory services

index (ELABI_P):’ with Cronbach alpha 0.92, included forty eight variables which include

availability of selected furniture, required instrument, equipments and essential laboratory

services for the delivery care. The fourth subscale ‘essential drug index’ (EDI with Cronbach

alpha 0.80 contained availability of thirteen essential drugs based on record of stock register.

Similarly indices were prepared for HSC, and same method of test scale was performed for

reliability. It had an overall Cronbach’s alpha value of 0.82 that ranged from 0.34 to 0.87. The

Page 13: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 13  

reliability was highest for ‘Physical infrastructure/ equipments index’ (PIEI_HS) (0.87) and

lowest for ‘Manpower index’ (MI_HS)’ i.e.0.34.

Table 5.1: Summary statistics and degree of reliability for adequacy indices

Indices of health facility adequacy Min Max Mean Std. Dev.

Cronbach's alpha(k)

PHC (N=3269)

Manpower index (MI_P) -3.4514 4.2674 5.47E-09 1.7781 0.72 (13)Physical infrastructure index (PII_P) -5.6156 4.1519 3.19E-09 2.0400 0.79(18)

Essential delivery care equipments/ laboratory services index (ELABI_P)

-6.5327 7.7423 -5.78E-09 3.2241 0.92(48)

Essential drug index (EDI_P) -4.6589 3.0733 -1.11E-08 2.0549 0.80(13)

HSC(N=8787) ANM residing in village or within 5km

0.0000 1.0000 0.664732 0.4721 -

Manpower index (MI_HS) -2.6386 4.1639 1.07E-08 1.2236 0.34(7)Physical infrastructure/ equipments index (PIEI_HS)

-3.4392 4.4586 4.90E-09 2.0602 0.87(19)

Essential drug index (EDI_HS) -2.5123 4.0351 -2.15E-08 2.0169 0.76(12)

5.2 Adequacy indices score and distribution by states:

For the selected adequacy indices at PHC and HSC Eigen value was obtained and an “Eigen

value greater than 1” criterion was employed for determining the number of factors. In order to

obtain more interpretable results solution, Varimax rotation was used to rotate the solution. This

caused the loadings to be distributed among the selected factors making it easier to interpret

results. Factor loadings of 0.5 or greater on a factor were regarded as significant. The factor

analysis of the selected items scale on the basis of principal component extraction by using

Varimax rotation converged with iterations. The following figure fig 5.1 (a, b, c, d, e and f)

shows the score plot of Eigen value and probability plot for the selected component based on

PCA factor 1.

a) Adequacy indices:

Page 14: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 14  

Fig 5.1: Scree-Plot and normal plot of the scores of health facility (PHC) adequacy indices: a, b for health personnel; c, d for physical infrastructure. Normal plot of the scores of health facility (PHC) adequacy indices: e for essential equipments and laboratory services; f plot is availability of essential drugs for the maternal care.

 

.51

1.5

22.

53

Eig

enva

lues

0 5 10 15Number

95% CI Eigenvalues

Scree plot of eigenvalues after pca: health personnel adequacy at PHCHealth personnel score plot at PHC (using PCA) with normal curve

0.1

.2.3

.4D

ens

ity

-4 -2 0 2 4Scores for component 1

01

23

4E

igen

valu

es

0 5 10 15Number

95% CI Eigenvalues

Scree plot of eigenvalues after pca: physical infrastructure adequacy at PHC physical infrastructure score plot at PHC (using PCA) with normal curve

0.1

.2.3

De

nsity

-6 -4 -2 0 2 4Scores for component 1

0.0

5.1

.15

De

nsity

-10 -5 0 5 10Scores for component 1

Essential equipments/lab service score plot at PHC (using PCA) with normal curveEssential drugS score plot at PHC (using PCA) with normal curve

0.1

.2.3

De

nsity

-4 -2 0 2 4Scores for component 1

a b

dc

ef

Page 15: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 15  

The score provided to health centers were categorized into quintiles and further analysis was

done accordingly to link with its utilization for MCH care. Table 5.2 explains distribution of

health centers in EAG states by the quintiles. Maximum PHCs with lowest health personnel (Q1)

was found in Uttar Pradesh (37%) and Uttaranchal (25%) followed by MP (21%) and maximum

PHCs with highest health personnel adequacy (Q4+Q5) was found in Orissa (60%) and Bihar

(59%). Further, maximum PHCs with least adequate physical infrastructure was found in state

of Orissa (46%) followed b by UP (26%) while the highest equipped (Q1+Q5) states are

Jharkhand (69%) and UTT (59%). Adequacy for required furniture, equipments, instruments for

delivery care and essential laboratory services are found to be least in state of Orissa (37%) and

UP (33%) followed by Bihar (34%) while the highest equipped states are Rajasthan (38%) and

Jharkhand (32%). Least drugs availability of drugs was found in state of Orissa and Bihar while

the highest was found in UTT and Jharkhand.

Surprisingly, Bihar and UP is the state where almost 50 percent ANM residing in village or

within 5km range and highest percentage of ANM (had training in MCH care and birth attendant,

accessible to village) which could motivate to safe delivery if the socially-economic-cultural

environment of those region could not support women for institutional delivery while the least

adequacy (physical infrastructure, essential drugs) for other facility at HSC was found in same

states.

Table 5.2: Percent distribution of EAG states by level (quintile) of selected infrastructure adequacy at PHC

a) Physical Infrastructure index at PHC States Q1 Q2 Q3 Q4 Q5 Total PHC

UTT n 4 11 17 17 30 79 % 5.06 13.92 21.52 21.52 37.97 100 RAJ n 24 87 170 251 145 677 % 3.55 12.85 25.11 37.08 21.42 100 UP n 201 149 164 135 116 765 % 26.27 19.48 21.44 17.65 15.16 100 BH n 103 87 51 58 118 417 % 24.7 20.86 12.23 13.91 28.3 100 JH n 8 18 21 26 76 149 % 5.37 12.08 14.09 17.45 51.01 100 OR n 208 118 60 24 42 452 % 46.02 26.11 13.27 5.31 9.29 100 CHH n 42 81 69 27 25 244 % 17.21 33.2 28.28 11.07 10.25 100 MP n 58 86 100 102 83 429 % 13.52 20.05 23.31 23.78 19.35 100 b) Essential equipments and laboratory services index at PHC

States Q1 Q2 Q3 Q4 Q5 Total PHC UTT n 2 23 34 10 11 80

Page 16: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 16  

% 2.5 28.75 42.5 12.5 13.75 100RAJ n 9 32 126 253 258 678 % 1.33 4.72 18.58 37.32 38.05 100UP n 256 265 85 48 113 767 % 33.38 34.55 11.08 6.26 14.73 100BH n 157 82 108 96 23 466 % 33.69 17.6 23.18 20.6 4.94 100JH n 6 6 28 62 47 149 % 4.03 4.03 18.79 41.61 31.54 100OR n 168 90 95 48 55 456 % 36.84 19.74 20.83 10.53 12.06 100CHH n 18 60 62 48 56 244 % 7.38 24.59 25.41 19.67 22.95 100MP n 38 96 116 89 90 429 % 8.86 22.38 27.04 20.75 20.98 100

Factor analysis technique was employed to examine the structure of the relationship among

variables representing the adequate infrastructure dimensions of healthcare services in EAG

states. Prior to running the factor analysis, the Kaiser-Meyer-Olkin (KMO) measure of sampling

adequacy and the Bartlett’s test of sphericity were performed. The generated score of KMO was

0.82 and highly significant Bartlett’s test of sphericity supported the appropriateness of using

factor analysis to explore the underlying structure of perceived quality of healthcare services.

5.3 Adequacy Inequality by States:

This, segment of the analysis identified the number of districts with lowest adequacy and

inequality was computed as the relative deviation from the average EAG adequacy. LQ helps to

measure the inequality in the adequacy at PHCs and HSCs and find from the average of overall

EAG estimates. Below average adequacy from the average EAG was calculated as follows:

proportion of lowest (Q1) equipped health centre was observed in the districts and 20 percent

c) Essential drugs index at PHC states Q1 Q2 Q3 Q4 Q5 Total

PHC UTT n 4 12 21 23 20 80 % 5.00 15.00 26.25 28.75 25.00 100 RAJ n 48 125 218 164 123 678 % 7.08 18.44 32.15 24.19 18.14 100 UP n 100 231 149 151 136 767 % 13.04 30.12 19.43 19.69 17.73 100 BH n 149 57 42 90 128 466 % 31.97 12.23 9.01 19.31 27.47 100 JH n 14 18 29 50 38 149 % 9.4 12.08 19.46 33.56 25.5 100 OR n 256 48 49 40 63 456 % 56.14 10.53 10.75 8.77 13.82 100 CHH n 26 54 46 61 57 244 % 10.66 22.13 18.85 25 23.36 100 MP n 58 108 103 79 81 429 % 13.52 25.17 24.01 18.41 18.88 100

Page 17: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 17  

and more were counted in lowest infrastructure based on assumption of 20 percent cut-off since

average EAG infrastructure quintiles is distributed over equal share of 20 percent in all Q1, Q2,

Q3, Q4 and Q5.

Table 5.3: spatial concentration

Table 5.3 Hence UP and Bihar is the state where maximum number of districts having lowest

adequacy (Q1) of all category of infrastructures at PHCs and HSCs. Followed by Orissa, where

districts with lowest adequacy only at PHCs was observed. Result explored that inequality in the

distribution and pattern of adequacy in the districts. Location quotient for the districts was

calculated as the relative deviation from the median adequacy of EAG. It was observed that most

of the districts from Rajasthan, western UP, Bihar, south western Orissa, Bihar, Jharkhand and

western MP have below average health personnel adequacy at PHC while the quite similar

pattern was found for Physical infrastructure and availability of equipments/laboratory services.

Most of the below adequacy clustered was observed in the districts of eastern UP, Jharkhand,

Bihar and some part of Orissa. while the, very uneven pattern was observed for the adequacy at

HSCs however, some districts of eastern UP and western Orissa have shown below average

adequacy of HSC indices.

5.4 Results from spatial Autocorrelation: Adequacy and Delivery care

a) Moron’s I and LISA: Univariate

Following maps shows LISA cluster and significance map generated from spatial software

GeoDa for all the districts taking institutional births percentage as the georefence values of the

Table 5.3 Number of districts by lowest concentration (Q1) of public health facilities in EAG states states

Facility adequacy at PHC Facility adequacy at HSC Total Districts

Bihar LM LPI LEI LD LS_ANM (<40 %)

ANM_ far (>40%)

LPI_H LD_H N

UTT 6 1 1 2 10 1 5 2 13 Raj 3 2 2 1 31 9 0 0 32 UP 41 32 38 13 34 40 61 51 70 BH 9 26 28 29 16 30 33 35 37 Jh 1 1 2 3 16 3 0 0 22 OR 10 22 14 24 28 0 0 0 30 CHH 10 7 1 2 14 2 0 0 16 MP 15 11 7 11 43 25 3 3 45 Note: LM: Lowest manpower, LPI: Lowest physical infrastructure; LEI: Lowest equipment and instruments; LD: Lowest drugs; LS_ANM: Lowest Skilled ANM, ANM_far: ANM residing far from village; LPI_H: Lowest physical infra at HSC; LD_H: Lowest drugs at HSC.

Page 18: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Pa

 

districts.

cluster m

significan

significan

region th

Correlati

adequacy

Univariat

and HSC

autocorre

was foun

required

(I=0.66)

Fig 5.2: Physical and low-l

(1) 

(3) 

tel, PhD (IUSSP

These maps

map for iden

nce of degre

nce and clus

he Local M

ion matrix an

y indices.

te Moron’s

C. Bivariate

elation with

nd maximum

for materna

at 5 percent

Univariate LISInfrastructure;

low locations s

P 2013)  

s are genera

ntifying loca

ee of spatial

ster maps ar

Moron statisti

nd Moron’s

statistics sho

Moron’s an

its estimate

m for phys

al care (I=0

level of sign

SA cluster ma; 3) Essential suggest clusteri

I

ated using co

al clusters an

autocorrlati

re generated

ic is compu

, LISA has

ows the auto

nd LISA ha

es to the nei

ical infrastr

0.50) while a

nificance.

ap of facility aequipments/labing of similar v

I=0.48, p<0.0

I=0.50, p<0.05

ontiguity ba

nd spatial o

ion is of per

d after 999 p

uted and tes

captured the

ocorrelation

ave captured

ighboring di

ructure at P

at HSC, dru

adequacy indicboratory servicvalues.

(2)

(4)

05

sed rook we

outliers of a

rmutations se

permutations

sted for sig

e correlation

in the select

d the one w

istricts. Spat

PHC (I=0.51

ug adequacy

ces at PHC byces; 4) Essent

eight. GeoD

geo-refrenc

elected. The

s for each lo

nificance by

n and autoco

ted adequacy

way/two way

tial autocorr

1) and essen

y was highly

y districts: 1) Hial drug adequ

Pa

a creates a L

ced variable.

e univariate L

ocation. For

y randomiza

orrelation be

y indices at

y correlation

relation (Fig

ntial equipm

y auto-corre

Health personnuacy. The high

I=0.51, ap<0.05

I=0.42, p<0.05

ge 18 

LISA

. The

LISA

each

ation.

ween

PHC

n and

g 5.2)

ments

elated

nel; 2) h-high

5

Page 19: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 19  

High-high regions shows the positive spatial association with its neighboring values of districts

in institutional births while low-low regions have positive autocorrelation from own and

neighboring low values of institutional births. High-high association are mostly found in the

districts of eastern Rajasthan, weatern MP and some of south-east Orissa. It could be concluded

that about higher inter-district variation and the improved health care utilization for institutional

births in these states. None of the high-high regions are located in UP, Uttaranchal, Chhattisgarh,

Bihar and Jharkhand part where low-low part are observed. Mostly low-low part are observed in

the district of Chhattisgarh, Uttarakhand and UP.

However, very few spatial ouliesr are identified in low-high and high-low regions because of

their inverse association. Low are surrounded by high values and vise-versa. Low-high outliers

are dispersed across total of 13 districts from Orissa, Bihar, Uttaranchal, Rajasthan and MP while

high-low are located in only a distict of Uttaranchal.

Fig 5.3 Moron’s scatter plot and significance map for institutional births in EAG: 1)LISA cluster Map; 2)

Box-plot; 3) LISA Significance map; 4)Moron’s scatter plot

1 2

3 4

Page 20: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Pa

 

LISA sig

these ide

are signi

significan

significan

b) Moro

Bivariate

the select

indices h

adequate

with drug

Result sh

equipmen

districts

said that

between

found fo

institutio

Negative

(1) 

tel, PhD (IUSSP

gnificance m

entified local

ificnce at l

nt at 5% l

nce in the di

n’s and LIS

e LISA clust

ted infrastru

however, a s

e essential eq

gs availabilit

howed that t

nts at PHC

from Rajast

t they are co

institutional

or availabilit

nal birth wh

e spatial auto

P 2013)  

map fig 5.3

l clusters and

evel of 5%

level in wh

istricts of Ra

SA: Bivariat

ter maps for

ucture indice

strong spatia

quipments/la

ty there are s

the bivariate

infrastructur

than, Orissa,

orrelated. Fi

l births and

ty of equip

hile insignifi

ocorrelation w

I=-

(3) showed

d spatial out

%. Almost a

hich few ar

ajasthan, MP

te

r the infrastr

s. There was

al association

aboratory se

significantly

e distribution

re. Level of

, MP and so

ig 5.3 Bivar

d adequacy i

pments/ lab

ficant associa

was observe

-0.172, p<0.05

additional i

tliers. Nearly

all the spati

re found si

P, Orissa, Ch

ructure show

s weaker ass

n was obser

ervices at PH

y autocorrela

n of delivery

adequacy a

ome part fro

riate LISA

indices. Sign

services, ph

ation was fo

ed with the lo

(2)

information

y all detected

ially correla

ignificantly

hhattisgarh,

wed that spat

sociation wa

rved between

HC (I=0.44)

ated (I=0.71)

y care (insti

and utilizatio

om the Uttar

has confine

nificant pos

hysical infr

ound with th

owest conce

on the sign

d spatial clu

ated regions

associated

UP and Utta

tial auto cor

as found betw

n physical in

while infra

).

itutional birt

on and both

ranchal, ther

d the spatia

itive spatial

astructure a

he health ma

ntration of f

Pa

nificance lev

usters and ou

s were foun

at 1% leve

arakhand.

rrelation bet

ween some o

nfrastructure

astructure at

ths) and esse

are higher i

refore it cou

al autocorrel

l association

at PHC with

anpower at P

facility adequ

I=-0.034, p<

ge 20 

vel of

utliers

nd as

el of

tween

of the

e and

HSC

ential

in the

uld be

lation

n was

h the

PHC.

uacy.

<0.05

Page 21: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Pa

 

Fig 5.3: bdelivery*phyinfrastructure

5.5 Spati

a) Comp

OLS resi

spatial-la

Spatial la

significan

low value

(3) 

tel, PhD (IUSSP

bivariate LIysical infrae at PHC; 4) I

ial Regressio

arison of OL

idual and spa

ag model fo

ag-residuals

ntly clustere

es are mostly

Fig 5.4: R

P 2013)  

SA map:1)Istructure;3)InInstitutional d

on Model:

LS and Spat

atial- lag-res

r the regres

are highly c

ed in similar

y concentrat

Residual map f

I=-0.1

Institutional nstitutional del*percent lo

tial Lag Mo

sidual map a

ssion predict

correlated (0

states MP, R

ted in Uttara

for institutiona

102, p<0.05

delivery*equdelivery*per

owest concen

del

are shown in

ted and erro

0.6778). The

Rajasthan an

anchal, UP, J

al delivery: 1) O

(4)

uipments &rcent lowesntration of equ

n the fig 5.4

or values are

estimated h

nd few distric

Jharkhand an

OLS residual; 2

& lab servit concentra

uipments & la

(1 and 2). A

e generated

high-high res

cts of Orissa

nd some part

2)spatial Lag-r

Pa

ices; 2)Institation of pab services at

After adoptin

for the dist

siduals value

a while low-

t of Chhattis

residual

I=-0.112, p<0

ge 21 

tutional physical PHC

ng the

tricts.

es are

with-

sgarh.

0.05

Page 22: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 22  

Table 5.4: Comparison of OLS and Spatial-lag model (dependent institutional delivery=1) OLS model Spatial lag model

Indep Variables Coefficient (β)

p-value Coefficient (β)

p-value

Direct LOW_ADQ -0.074 0.0124 -0.071 0.0341 DOC_ADQ 0.069 0.0522 0.071 0.3594 3ANC_P 2.079 0.0038 2.960 0.0006 NEAR_ANC -0.113 0.0014 -0.102 0.0251 NEAR_DEL -0.119 0.0158 -0.131 0.0073 VILL_CONNCT 0.134 0.0660 0.111 0.0044 LOW_WI -4.297 0.0000 -3.051 0.0000 Indirect LIT_P 2.132 0.0010 2.782 0.0011 URBAN_P 1.238 0.0001 1.632 0.0332 SCST_P -0.084 0.0461 -0.084 0.0872 POP_COV -0.044 0.4023 -0.311 0.6720 JSY_P 5.0206 0.0000 6.261 0.0000 CONSTANT 22.594 0.0252 19.321 0.0021 N 263 263 Log-likelihood -877.762 -818.402 AIC 1779.52 1662.8

R2 adjusted 0.730647 0.790647 Lag coeff (lamda) 0.588293 Heteroskedasticity test (Breusch-Pagan)

28.72187 0.00250 10.06128 0.0031

Spatial dependence test (likelihood ratio test)

12.034 0.0002

Page 23: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 23  

5.6 Conclusions:

Inequality measure reveals that UP and Bihar is the state where maximum number of districts

having lowest adequacy of all indices of infrastructures at PHCs and HSCs. Followed by Orissa,

where districts with lowest adequacy only at PHCs was observed. However, districts of Orissa

have better adequacy at HSCs. Inequality in pattern of adequacy was captured very clearly

through the maps which clearly revealed that the most of the districts from Rajasthan, western

UP, Bihar, south western Orissa, Jharkhand and western MP have below average (EAG) of

health personnel adequacy at PHC and quite similar pattern was found for physical infrastructure

and availability of equipments/laboratory services. Though, very uneven pattern was observed

for the facility adequacy at HSCs. On the other hand if we look for the concentration of only

lowest adequate facility at PHCs then most of the districts belongs to eastern UP, Jharkhand,

Bihar and some part of Orissa.

Correlation matrix showed health personnel adequacy index was highly correlated with physical

infrastructure index at PHC. It could be said that availability of physical infrastructure are

supposed to have availability of health personal. Whereas, equipments/laboratory services at

PHC was highly correlated with adequate drugs at PHC and with physical infrastructure at PHC.

This could be easily concluded that adequate availability of one facility at PHC is very much

reflects the availability of other facility. Importantly, physical infrastructure at HSC (r=0.53)

was found to be correlated significantly with physical infrastructure at PHC (r=0.49) and ANM

residing within 5 km (r=0.57) which might be integrated program effect of these services at PHC

and PHC both.

Spatial results (univariate Moron’s and LISA) for infrastructure indices show the significantly

high spatial autocorrelation for adequacy indices at PHC and CHS in districts considering

Rook’s spatial weights to the neighboring districts. Additionally, bivariate results showed the

maximum autocorrelation between physical infrastructure and adequate essential equipments/lab

services at PHC (0.44). Equipments adequacy clustering is found significant in western districts

comprising districts of western Rajasthan, Middle MP and southern UP. Whereas, essential drugs

adequacy are clustered in south-west Orissa, Chhattisgarh, and eastern MP. However, manpower

adequacy is also clustered significantly (high-high) in some districts of Orissa.

Page 24: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 24  

Spatial dependence for delivery care has captured the better acceptability at some extent to

describe through several tests of spatial diagnosis over dependents (outcome), independents

(covariates) and error term which has come up with the spatially lagged dependent variable term

in the model which estimate are adjusted by spatial autocorrelation. All intermediate and direct

variavles have shown combined effect on the institutional delivery. Some OLS estimates has

shown significant association with the institutional delivery except for average population

covered by PHC while some covariates disappears its influence on independents once spatial-lag

(spatial dependence) parameter incorporated in the model like availability of doctors at PHC,

proportion of SC/ST population and percentage of urban population (p>0.05). Low infrastructure

adequacy at PHC, distant health facility providing ANC or delivery care and proportion of

lowest quintile have significantly (p<0.01) reduced the probability of having institutional

delivery and hypothsesis are rejected. Whereas Receipt of three or more anti-natal visits (ANC),

all weather road connectivity of village to the health center and women literacy have

singnificanlty increase the likelihood of instititonal births. Programamitc efforts of states

government to encourage the institutional birth by providing the incentives to women had

delivered baby in public health facility have singificanlty increase the utilization (p<0.001).

This study improved understanding how women's health-care-seeking behaviour is shaped by the

availability of health services and inform the development of strategies to improve the provision

and use of maternal healthcare at district/county level. As if the barriers to the accessibility of

service are to be effectively reduced any attempt and improve the adequate facility for maternal

care at PHC, to increase maternal care-seeking behavior in rural India will require resources to

be targeted at the most impoverished areas and development of strategies for reaching those not

yet reached.

Page 25: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 25  

References (to be updated):

Andersen RM, Davidson PL: Measuring Access and Trends. Changing the US Health Care System San

Francisco, CA: Jossey-BassAndersen RM, Rice TH, Kominski GF 1996.

Anseline L, Bera A (1988). Spatial Dependence in linear regression models with an introduction to

spatial econometrics. New York: Marcel Dekkar.

Anseline, L. (2002). Exploring Spatial Data with DynESDA2. CSISS and Spatial Analysis. Laboratory

University of Illionois.

Anseline, L. (1996). The Moron Scatterplot as an ESDA Tool to Access Local Instability in Spatial

Association. Spatial Analytical Perpectives on GIS , 111-125.

Arcury, T A; Gesler, W M; Preisser, J S; Sherman, J; Spencer, J; Perin, J (2005). The Effects of

Geography and Spatial Behavior on Health Care Utilization among the Residents of a Rural Region.

Health Services Research , 40(1):135-156.

Bonu S, Bhushan I, Rani M, Anderson I (2009). Incidence and correlates correlates of ‘catastrophic’

maternal health care expenditure in India. Health Policy and Planning , 24: 445–56.

De Maio FG: Ecological Analysis of the Health Effects of Income Inequality in Argentina. Public Health

2008, 122(5):487-496.

Elbers C, Lanjouw J, Lanjouw P (2003). Micro-level estimation of poverty and inequality. Econometrica ,

71 (1): 355–364.

Heard Nathan J, L. U. (2004). Investigating Access to Reproductive Health Services Using GIS:

Proximity to Services and the Use of Modern Contraceptives in Malawi. African Journal of Reproductive

Health Reproductive , 8(2):164-1.

International Institute for Population Sciences (IIPS). (2010). District Level Household Survey (DLHS-3),

India 2007–08. Mumbai: IIPS.

Kesterton AJ, C. J. (2010). Institutional delivery in rural India: the relative importance of accessibility and

economic status. BMC Pregnancy and Childbirth , 10:30.

Kumar RM, S. M. (1997,). Impact of health centre availability on utilisation of maternity care and

pregnancy outcome in a rural area of Haryana. J Indian Med Assoc , 95(8):448-50.

Larrea C, Kawachi I: Does economic inequality affect child malnutrition? The case of Ecuador. Social

Science & Medicine 2005, 60(1):165-178.

Page 26: IUSSP Conference 2013, Busan, S Korea Rural Health ...

Rachana Patel, PhD (IUSSP 2013)   Page 26  

López-Cevallos and Chi. 2010. Assessing the context of health care utilization in Ecuador: A spatial and

multilevel analysis, BMC Health Services Research 2010, 10:64

Luo W: Using a GIS-based floating catchment method to assess areas with shortage of physicians. Health

and Place 2004, 10(1):1-11.

Manju Rani, Sekhar Bonu and Steve Harvey 2008. International Journal for Quality in Health Care.

Number 1: pp. 62 –71

McLafferty, S. 1988. Predicting the effect of hospital closure on hospital utilization patterns. Social

Science and Medicine, Vol.27(3), pp.255-262.

Ministry of Health and Family Welfare, Government of India. (2009, november).

http://mohfw.nic.in/NRHM.htm. Retrieved from http://mohfw.nic.in.

National Health Policy 1983, National Population Policy 2002

Phillips, K., Morrison, K., Anderson, R., & Aday, L. (1998). Understanding the Context of Healthcare

Utilization: Assessing Environmental and Provider-Related Variables in the Behavioral Model of

Utilization. Health Services Research , 33(3): 571-596.

Rosero-Bixby, L. (2004). Spatial access to health care in Costa Rica and its equity: a GIS-based study.

Social Science & Medicine , 58(7):1271-1284.

Singh A, Pathak PK, Chauhan RK, Pan Willliam. 2011. Infant and child mortality in india in last two

decades: a geospatial analysis. PLos one,6 (11):1-19 

Subramanian SV, Delgado I, Jadue L, Vega J, Kawachi I: Income inequality and health: multilevel

analysis of Chilean communities. J Epidemiol Community Health 2003, 57(11):844-848. 

WHO, country health profile health system, India

[http://www.searo.who.int/en/Section313/Section1519_10857.htm]