Examining the Relationship between Access to Health Care and Socio-Economic Characteristics

Abstract

The link between good health and the ability to work effectively to meet livelihood needs is established but the economic implications of the reverse have often not been estimated; also how this plays out for different gender and socio-economic groups is often not estimated. The chapter examines the health care access that rural households have and examine how it relates to their education and employment in various sectors. The study used Nigeria as a case study as such the General Household Survey Data for wave 4 was used. The data were analyzed using descriptive, and Tobit regression model. The results showed that labour hours worked (in agricultural, non-agricultural and non-household activities) has a negative relationship with health care access. Age and literacy (ability to read) is important in health care access and have positive relationships with it. The policy implication of the study is that educational infrastructure must be developed along-side health policy initiatives.

1. Introduction

A major target of the sustainable development goal three on good health and well-being includes achieving universal health coverage, which includes access to health services. This is relevant for developing countries and rural areas in particular where suffering and preventable diseases often lead to untimely death. Low insurance coverage, poverty and shortage of health care staff have been cited as deterrents to healthy lives in developing countries [1]; these can lead to increased out of pocket expenses which the vulnerable groups of the society may be unable to meet conveniently [2]. Hence, achieving the sustainable development goal three requires that the issue of access to health care and its socio-economic determinants like income, education, and gender be prioritized in policy and intervention strategy designs [1].

According to [3] access to health care means having “the timely use of personal health services to achieve the best health outcomes”. Implicit in this definition is the thought that health facilities should be within the reach of all as at when needed; that it is utilized and positioned to render appropriate services from which users can achieve expected outcome. In reality many people lack access to adequate health care as a result of public policy failure which creates barriers such as poverty and other forms of inequality. On the other hand, there is the issue of self-selection as people show preferences for other ‘medical’ alternatives. However, health care system is a public policy issue and as such resources should be allocated to it to make it effective and efficient. Based on this, there is a need for empirical evidence that will support government planning and resource allocation with respect to the provider and the client.

To achieve this, appropriate definition of variables that define health care is important. In the literature [4, 5], some indicate that the ability to pay for services is a major determinant, in some case preferences indicated by clients’ behavior in terms of perception of the illness, the options available are cited as playing important roles in health care access. There is also the gender dimension: poverty, religion, cultural values and norms are considered barriers to women’s access [6]. Finally, there is the school of thought that lays emphasis on demographic and socio-economic factors as the determinants of health care access. The varied positions suggest the need to approach the question from a different perspective. This study defined an index for access to health care and used alternative set of variables which are particularly relevant to the clients’ circumstance. The study focused on the rural areas of Nigeria and used hours worked, level of literacy as major covariates of health care access.

A focus on the rural area is apt because the poor are generally known to have less access to health care services. The continued disparity between the poor and the rich within the country, is the basis for research based evidence which will set the direction for approaches and interventions that will narrow the gap. Identifying the dimensions of healthcare deprivation in rural areas, the vulnerable and the hot spots for the disadvantaged can be a spring board for closing the gap. Based on this, the study examined the relationship between access to health care and socio-economic variables with the aim of suggesting evidence based policy options that will lead to improved personal health care system in rural Nigeria.

2. Methodology

2.1 Conceptual framework

Several organizations [3, 7] have given different definitions of health care access to include timeliness, coverage, a regular source of care and capable personnel. The word ‘access’ is also seen to have both quantitative and qualitative aspects which means that it may not be fully quantified while being evaluated. Some components used to assess and evaluate it in literature include ‘being available’, ‘financial access’, ‘utilization’, and ‘barriers’. Part of the discourse on its measurement includes the capacity, demand [7] and geography or spatial dimension. But a common thread that runs through the literature is the need to have equity in health care access especially for planning and resource allocation at the macro level.

Figure 1 shows the conceptual framework which underlies this study. The definition of access to health care is taken from Penchansky and Thomas [8]. Access is grouped into five As: Affordability, Availability, Accessibility, Accommodation and Acceptability. These characteristics of health care access are defined to reflect provider and client’s expectations and characteristics; and the fact that these need to fit. The framework shows that there are aspects of healthcare access which are easily influenced by the socio-economic status of the respondent. For example, the income generated by the client could influence not only the choice of the healthcare facility but also the ability to pay; also the resources of the facility especially in terms of personnel and technology may interact with the income level or socio economic status to determine health care access [10]. On the other hand, the level of access could impinge on the individual’s continued ability to earn an income, improve the food security and livelihood status [11, 12]. Hence the relationships could be recursive in some cases. Also, the influence of external factors such as policies and institutions on elements of the framework can hamper or support clients’ access to health care. The framework suggests the need for a holistic approach to improving rural people’s access to healthcare so that it cuts across all socio-economic levels while not making any particular group worse off.

Figure 1.

Framework for health care access. Source: Author’s Concept based on Penchansky and Thomas, [8] cited in McLaughlin and Wyszewianski [9].

2.2 Analytical framework

2.2.1 The relationship between health care access and socio-economic status

The relationship of access to health care to socio-economic indices (Education, Income and Employment) was determined by first establishing an indicator of health care access. Rural people often have several health care options which may be formal or informal; also some combine different options in a bid to maintain a healthy life or restore themselves to health. This behavior reflects the different aspects of health care access, and may also be a reflection of what they feel comfortable with.

2.2.1.1 Measuring health care access

Health care access has been measured in different ways. The IOM [3] identified two quantifiable areas: utilization and outcomes. Indicators for both were then identified which could indicate problem areas as well as show when problems occur; these permitted a level of measurement but access was treated as an intervening variable to health care utilization and outcomes. A conditional logit model was proposed by Jang [5]; the model combined a choice model with a Floating Catchment Area in such a way that the peculiarity of a client to use a hospital was captured. In this study, indicators were identified for each of the fives ‘A’s defined above. These were then used to generate an index using the Principal Component Analysis. The definition of the each A is relatively encompassing as such client and organizational peculiarities are included. The index generated was then used as the dependent variable in the regression model. The terms and the indicators are listed in Table 1. To examine the relationship between health care access and indicators of socio-economic variables [10], the people were classified as: (I) Able to read vs. Unable to read, (II) Male vs. Female (III) Hours worked in different jobs per week.

S/N	Item	Indicator variables	Unit
1	Availability	The existence of a formal health care facility/Actually Visited; The facility has some qualified personnel; The facility is equipped	Yes/No; Yes/No; Yes/No
2	Affordability	No charges; Client paid the charges and or bought medication	None; Amount Spent in Naira; Amount spent in Naira
3	Accessibility	Transportation cost to the nearest facility	Transportation cost in Naira
4	Accommodation	Wait time for consultation; Walk-in possible	Minutes; Yes/No
5	Acceptability	Actual Consultation made	Yes/No

Table 1.

Indicators for the measurement of access to health care.

The PCA is a technique for reducing the dimensionality of large datasets, increasing interpretability but at the same time minimizing information loss [13]. It does so by creating new uncorrelated variables that successively maximize variance. Although for inferential purposes a multivariate normal (Gaussian) distribution of the dataset is usually assumed, PCA as a descriptive tool needs no distributional assumptions and, as such, is very much an adaptive exploratory method which can be used on numerical data of various types. Olajide, stated ‘In mathematical terms, from an initial set of n correlated variables, PCA creates uncorrelated indices or components, where each component is a linear weighted combination of the initial variables. Mathematically, the transformation is defined by a set of p-dimensional vectors of weights or loadings (1) that map each row vector X_(i) of X to a new vector of principal component scores(2) given by (3).

$$W_k=\left({\omega }_1,\dots ,{\omega }_p\right)\left(k\right)$$E1

$$t_{\left(i\right)}=\left(t_i,\dots t_p\right)\left(i\right)$$E2

$$t_{k\left(i\right)}=X_{\left(i\right)}\bullet W_{\left(k\right)}$$E3

in such a way that the individual variables of t considered over the data set successively inherit the maximum possible variance from x, with each loading vector w constrained to be a unit vector [14].

For example, from a set of variables X₁ through to X_n.

$${\mathrm{PC}}_1={\mathrm{a}}_{11}{\mathrm{X}}_1+{\mathrm{a}}_{12}{\mathrm{X}}_2+\dots +{\mathrm{a}}_{\text{1n}}{\mathrm{X}}_{\mathrm{n}}$$E4

$${\mathrm{PC}}_{\mathrm{m}}={\mathrm{a}}_{\text{m1}}{\mathrm{X}}_1+{\mathrm{a}}_{\text{m2}}{\mathrm{X}}_2+\dots +{\mathrm{a}}_{\mathrm{mn}}{\mathrm{X}}_{\mathrm{n}}$$

Where:

a_mn represents the weight for the mth principal component and the nth variable [14].

2.2.1.2 Estimating the relationship between access to health care and selected indicators of socio-economic status

Since an index was generated to summarize the individual’s access to health care, it means that it can be subject to a censoring effect. As such in order to examine the relationships between the variables, the Tobit model is used to avoid having a biased coefficient estimates [15]. The model is designed to estimate linear relationships between variables when there is either left- or right-censoring in the dependent variable. The Tobit Model presents a simple relation:

$${{\mathrm{y}}_{\mathrm{i}}}^{\ast }={\mathrm{\beta }}_0+{\mathrm{\beta }}_1{\mathrm{x}}_{\text{1i}}+{\mathrm{\epsilon }}_{\mathrm{i}}$$E5

$${\mathrm{y}}_{\mathrm{i}}=0,\text{if}{{\mathrm{y}}_{\mathrm{i}}}^{\ast }={{\mathrm{x}}_{\mathrm{i}}}^{‘}\mathrm{\beta }+{\mathrm{\epsilon }}_{\mathrm{i}}\le 0$$E6

$$={{\mathrm{y}}_{\mathrm{i}}}^{\ast }={{\mathrm{x}}_{\mathrm{i}}}^{‘}\mathrm{\beta }+{\mathrm{\epsilon }}_{\mathrm{i}}\text{if}{{\mathrm{y}}_{\mathrm{i}}}^{\ast }={{\mathrm{x}}_{\mathrm{i}}}^{‘}\mathrm{\beta }+{\mathrm{\epsilon }}_{\mathrm{i}}>0$$E7

The effect of the Xs on the probability that an observation is censored and the effect on the conditional mean of the non-censored observations are the same: β.

y_i * = HCA index (may be censored right or left).

β = Vector of parameters to be estimated.

x = Explanatory and control variables categorized into household head and household variables (sex, age, education, farm and non-farm income, employment type), household food security status. The model aimed at determining the partial effects of the x variables on the latent variable. The parameters of Eq. (5) are estimated by the maximum likelihood method. To examine the relationship between health care access and indicators of socio-economic variables, the people were classified as: (I) Able to read vs. Unable to read, (II) Male vs. female (III) Hours worked in different jobs per week.

2.3 Data

The paper used the wave 4 of Nigeria General Household Survey (GHS) data collected by the National Bureau of Statistics and the World Bank. The survey panel is implemented in collaboration with the World Bank Living Standards Measurement Study (LSMS) team as part of the Integrated Surveys on Agriculture (ISA) program. The data is nationally representative involving about 5000 households, and contains comprehensive data on socio-economic characteristics and welfare indicators. The households were selected through a random sampling procedure which ensured the distribution of EAs across the 6 geo-political zones (and urban and rural areas within) in the nation. The GHS consists of three panel questionnaires: Household, Agricultural and Community, which were administered using Computer Assisted Personal Interview (CAPI) in post planting and post-harvest periods. This study focused on the rural sector as a case study not only because it is the agricultural base of the country but also because additional empirical evidence is necessary for resource planning and policy implementation of improved health care delivery. It will also serve as a means of evaluating the current status of previously implemented rural programs in the nation. Individual level analysis was carried out using data for household heads in rural communities. All household heads were selected irrespective of sex leading to a total sample of 3433 individuals based in the rural areas but complete data for the variables of interest were found for 3217 people, so these were used in the analyses.

The Federal Republic of Nigeria is located in the south east of West Africa, with a coast at the Bight of Benin and the Gulf of Guinea. It lies between latitudes 4° and 14°N, and longitudes 2° and 15°E. It has a land area of 923,768 km² and a population of 192 million people. It has a tropical climate with variable raining and dry season periods. Agriculture, which is the main means of livelihood in the rural sector, contributes about 23.4 percent of the GDP. Health care delivery is the joint responsibility of the three tiers of the federal, state and local governments in the country. Since the Bamako Initiative of 1987, the country has made significant improvement in health care delivery and access using the community based approach. However, the sector is witnessing increasing emigration of skilled workers to the west. The rural population is about 47% the World Bank World Development Report, over 35 percent work in agriculture, in an environment that generally has low social infrastructure [16].

3. Results and discussion

3.1 Index generation

The Principal Component Analysis (PCA) indicated that the most important components in the index generated were those associated with the first 4 As: Availability, Affordability, Accessibility and Accommodation. Component 1 was responsible for about 32 percent of the variance as such it was used in the regression analysis. Tables 2 and 3 indicate the bases for the acceptance of the model and confirms that a PCA could be carried out on it.

PCA Pattern Matrixa
	Component
	1	2	3
Actually consulted or visited	0.767
Days Lost to illness			−0.938
Consultation fees (N)
Transportation cost (N)			−0.908
Travel Time (minutes)	0.744
Wait time (minutes)	0.600
Cost of medicine purchased (N)	0.478
Number of Nights Hospitalized		0.942
Cost of Hospitalization (N)		0.936

Table 2.

The component of the access to health care index.

^a

Rotation converged in 7 iterations.

Extraction Method: Principal Component Analysis.

Rotation Method: Oblimin with Kaiser Normalization.

KMO and Bartlett’s test
Kaiser-Meyer-Olkin measure of sampling adequacy.		0.647
Bartlett’s Test of Sphericity	Approx. Chi-Square	8707.521
	df	36
	Sig.	0.000

Table 3.

Data sampling adequacy.

3.2 Relationship between HCA index and socio-economic variables

The HCA is the response variable predicted by the model. The tobit model was used because this response variable is censored. The tobit regression coefficients are interpreted in the same way as the OLS regression coefficients except that the linear effect is not on the observed outcome but on the latent variable. The expected HCA score or index will change for each unit increase in the corresponding predictors. As such an increase in hours worked is likely to reduce the probability of accessing health care while an increase in age could lead to an increase in health care access. The literacy status as a measure of educational level suggests that it could be a barrier to accessing health care. At p < = 0.005, labor hour in a week is the most significant variable, although negative, while Age and literacy status are significant at higher levels of the test statistic. Policy instruments that will encourage high literacy levels and create health care opportunities for the aged could increase personal access to health care in rural communities. Also, advocacy on the importance of balancing work with health could be necessary to encourage more visits even with a tight schedule (Table 4).

Tobit regression; Log likelihood = −2684.85		Number of obs = 3217; LR chi2(4) = 73.62; Prob > chi^2 = 0; Pseudo R^2 = 0.0135
HCA	Coef.	Std. Err.	t	P > |t|	[95% Conf.]
Client is literate	0.234	0.12	1.93	0.054	−0.00374	0.470747
Hours worked/week	−0.14	0.02	−7.54	0	−0.17712	−0.104
Sex	0.22	0.15	1.43	0.15	−0.07955	0.510316
Age	0.01	0.004	1.85	0.065	−0.00043	0.014691
_cons	−1.78	0.28	−6.32	0	−2.3364	−1.23056
/sigma	2.53	0.0781			2.37461	2.680823

Table 4.

Tobit regression results.

Obs. summary: 2490 left-censored observations at HCA<=0

727 uncensored observations

0 right-censored observations

4. Conclusion

This study examined the relationship between health care access and some socio-economic variables. To achieve this, an index was generated based on several variables. The index showed that availability, affordability, accessibility and accommodation are really important as such policies that would enhance these should be pursued in order to improve rural health care system in terms of utilization and outcome. The policies should combine rural healthcare infrastructure development with rural health insurance schemes. The results also showed that labour hours worked (in agricultural, non-agricultural and non-household activities) has a negative relationship with health care access. Age and literacy (ability to read) is important in health care access and have positive relationships with it. The policy implication of the study is that educational infrastructure must be developed along-side health policy initiatives.

A.1 Communalities

	Initial	Extraction
Consulted	1.000	0.555
Days Lost	1.000	0.805
Pay consult	1.000	0.340
Trans cost	1.000	0.845
Time Mins	1.000	0.599
Wait time minutes	1.000	0.334
Cost drug purchased	1.000	0.545
Number of nights admitted	1.000	0.865
Amount paid	1.000	0.843

Extraction method: Principal component analysis.

A.2 Total variance explained

Component	Initial eigenvalues			Extraction sums of squared loadings			Rotation sums of squared loadings^a
	Total	% of variance	Cumulative %	Total	% of variance	Cumulative %	Total
1	2.877	31.966	31.966	2.877	31.966	31.966	2.193
2	1.687	18.741	50.707	1.687	18.741	50.707	2.057
3	1.168	12.974	63.681	1.168	12.974	63.681	2.206
4	0.927	10.303	73.984
5	0.810	8.996	82.980
6	0.553	6.149	89.129
7	0.496	5.507	94.635
8	0.277	3.082	97.717
9	0.205	2.283	100.000

Extraction method: Principal component analysis.

A.3 Component Matrixa

	Component
	1	2	3
Consulted	0.523		0.499
Days Lost	0.595		−0.608
Pay consult	0.583
Trans cost	0.706		−0.488
Time Mins	0.606
Wait time minutes
Cost Drug Purch	0.722
Number of Nights admitted	0.466	0.804
Amount paid	0.417	0.817

^a

3 components extracted.

Extraction method: Principal component analysis.

A.4 Structure Matrix

	Component
	1	2	3
Consulted	0.740
Days Lost			−0.888
Pay consult	0.447		−0.442
Trans cost			−0.916
Time Mins	0.767
Wait time minutes	0.567
Cost Drug Purch	0.606	0.457	−0.451
Number of Nights admitted		0.928
Amount paid		0.912

Extraction method: Principal component analysis.

Rotation method: Oblimin with Kaiser normalization.

A.5 Component correlation matrix

Component	1	2	3
1	1.000	0.158	−0.329
2	0.158	1.000	−0.199
3	−0.329	−0.199	1.000

Extraction method: Principal component analysis.

Rotation method: Oblimin with Kaiser normalization.