Introduction:

Child mortality indices are principal indicators of population health and well-being. Therefore, there are many indicators of child mortality which have been used to determine levels and trends of population health, including the neonatal and post neonatal, infant, child and under-5 mortality rates (U5MR) (1). The U5MR has been selected as one of the most important indicator of child mortality because it presents the best concept of capturing mortality risks during the susceptible years of child hood(2). Because of the importance of reducing U5MR for societies, it is one of the United Nation 2015 Millennium Development Goals aims (3).

Altogether, the statistics on child and infant mortality rates imply the countries level of socio-economic development and quality of life. They used for evaluating and observing population and health plans and policies(4). Acute respiratory infections, gender differences, vitamin A deficiency, HIV, malnutrition and etc are most reasons for U5MR. About 25-30 percent of children born to infected mothers grow to be infected with HIV and most of them died before 5 year old in developing countries. 54 percent of child hood death is because of malnutrition globally (3-7).

Given the importance of under 5 mortality, many studies have been examined to determine the determinant of it for example, Susuman and Hamisi survey a demographic scenario for U5MR in Tanzania. Based on their study maternal education can improve mothers behavior in using health serrvices which reduce under 5 years mortality(7). Ram et al compared neonatal and U5MR in 597 district of India.  According to findings 37 percent of mortality rate fell down each year between 2001 and 2012(3).  Fernandes et al examined the effects of health system strengthening on under 5, infant and neonatal mortality in Mozambique. Based on this study, improvement in the public sector health workforce, institutional birth coverage and government health financing are some reasons that decrease U5MR in Mozambique(6).

An ordinary issue on the outline of policy makers is the adjustment the concept of welfare state, particularly when talk about expediency of reducing the existence of government in significant fields like health or education. In this framework, analyzing the effects of public health expenditures on health and longevity, with the purpose of determining of the consequences of health polices is in interest. Public medical expenditures become effective in improving life expectancy(8).

Many factors could effect on child mortality in macro level.  Therefore identifying and assessing the factors affecting the child mortality could be the first step in planning to reduce the mortality and promoting the society health and life expectancy.  Also, understanding this determinants of health policy makers can help to take more cost-effective interventions and policies to reducing child mortality. Given the importance of these factors, the aim of this study was to investigate the macro determinants of under-five mortality rate (U5MR) using econometrics models in Organization of Cooperation and Development (OECD) countries in 2010 and 2013.

Material and methods

This was a cross sectional study. An econometrics model was used to analyze factors which effected under five mortality rate (U5MR) in OECD countries in 2010 and 2013. For this purpose, we used macroeconomics data on under-five mortality rate, healthcare expenditure, GNI, physician and nurses’ density, ratio of female to male primary, secondly and tertiary enrollment from World Health Organization (WHO) and world bank datasets for OECD countries in 2010 and 2013.

Pearson correlation and regression models was used to examine the relationship between the U5MR and other variables. So, the association between U5MR and the other variables were examined with Pearson correlation at first. Then the regression model were fitted. Mortality models such as under 5 mortality are in nonlinear form. Therefore, for estimating these models they must be changed into log-log form.  The primary econometrics model of under-five mortality determinants used in this study is shown below:

15U5MR= α+ HCEβ1+ GNIβ2 + PHYDβ3 + NURDβ4 +LIT1β5+LIT2β6+ LIT3β7 +U'>

Where:

HCE =  public health expenditures, (% of total health expenditures)

GNI =  gross national income per capita, PPP (current international $)

PHYD =  Physicians’ density (physician per 1000 people)

NURD =  nurses’ density (nurse per 1000 people)

LIT1 =  Ratio of female to male primary enrollment

LIT2 =  Ratio of female to male secondary enrollment

LIT3 = Ratio of female to male primary enrollment

U = residuals.

The final econometrics form of model was log-log form as following:

15LU5MR= β0 +β1LHCE+ β2LGNI+ β3LPHYD+ β4LNURD+ β5LHCE + β6LIT1+ β6LIT2+ β7LIT3 +ε '>

Where:

L = Symbol of logarithm

Β­₀ = logarithm of α and

Ɛ = the symbol of residuals.

We estimated the model at first, then Breusch-Pagan test for heteroskedasticity and Ramsey RESET test for omitted variable were done (9).  The stata.11 was used for analyze the model.

Results

The Pearson correlation between U5MR and independent variables has been shown in table1. As it appears all variable had negative correlation with U5MR in 2010. Correlation between U5MR and other variables, except LIT2, were negative in 2013. The correlation between U5MR and HCE, GNI, PHYD, NURSD were statistically significant, but LIT1, LIT2 and LIT3 had no significant correlation with U5MR in 2010. NURSD and LIT3had the strongest and the weakest correlation with U5MR, respectively in 2013.

Table1: Pearson correlation between U5MR and independent variables in 2010 and 2013

The determinant of U5MR was investigated using the cross sectional regression models. The Breusch-Pagan test showed there were no heteroskedasticity in two regression models (p>0.01).Also, The null hypothesis in Ramsey RESET test, which indicates there are no omitted variable in the model was not rejected (p>0.01). The table 2 shows the result of the regression models. The result of 2010 showed that all variable, except Ratio of female to male primary enrollment (LLIT1) had negative effect on under-five mortality rate (U5MR). The β-coefficient of LHCE was -0.964 and was statistically significant. Also, gross national income per capita (LGNI) and Ratio of female to male secondary enrollment (LLIT2) had significant effect on U5MR. The R² of the estimated model was 0.63.

As table 2 shows, the β-coefficient forLHCE, LGNI, LPHYD, LNURD, LLIT1, LLIT2 and LLIT3 were -0.746, 0 .089, -0.721, -0.512, -6.52, -0.53 and -0.08 , respectively in 2013. Only, The effects of  LPHYD and LNURD was statistically significant in this year.

Table2: The result of regression analysis of determinant of U5MR in OECD countries

Discussion

Empirical studies on child mortality at macro levels are more useful for designing health policies  (10).  Under-five mortality rate (U5MR) is one of the most important index in health sector because implies the level of development and health status of countries. We analyzed some factors that influence on U5MR. The result of the study showed that all variable included had negative correlation with U5MR in 2010 and all variables, except LIT2, had negative correlation with U5MR in 2013.

 The regression analysis depicted that public health expenditures had negative effect on U5MR in OECD countries in 2010 and 2013, but this effect was significant in 2010. Also, per capita gross national income had negative effect on child mortality so that under the ceteris paribus 10% increase in GNI could decrease U5MR by 6.27% in 2010. The physician and nurse density, both, had decreasing effect on child mortality in 2013. Two variables of 3 variable in which included for maternal schooling had negative impact on U5MR and only Ratio of female to male secondary enrollment in school had significant effect in 2010. Also, all of these three variable had negative effect on U5MR in 2013. The R2 of regression showed that 63%  and 64% of variance of under-five mortality rate are explained by independent variables in which included in the regression models in 2010 and 2013, respectively.

Some study conducted to investigate determinant of child mortality all over the world and include different variable in their studies (10-15). Limin Wang found that access to electricity, income, vaccination in the first year of birth, and public health expenditure significantly reduce child mortality that confirms our findings about the effect of gross national income and healthcare expenditures on child mortality(10). Also, according to Howeling et al study the relation between GNI and U5MR is negative but this relation is weaker for the poor compare with the rich people. Ethnic fragmentation and public spending strongly associated with U5MR among the poor compared with the rich. The effect of female literacy, democracy and state strength on U5MR was not different in poor and rich(16). It worth noting that public and private health expenditure may have different effect on child mortality so that some evidence have been shown the increase the private health expenditures could have inverse effect on child mortality(17).

 Rezaei et al examined determinants of infant mortality in Iran. The study results showed that the number of physician per capita, mean years of schooling and GDP were inversely related to infant mortality(18). Therefore, the access to healthcare services such as physician and nurse services have been shown as important determinants of societies’ health (19). the study that conducted by Robinson for investigating relationships between infant and under-five mortality rate and the distribution of health professionals, GNP per capita, and female literacy in 155 countries, showed that higher density of general practitioners and nurses in the countries is associated with lower infant and under-five mortality rates. The  correlation coefficients for U5MR were −0·81 and−0·72 for physicians and nurses, respectively (P < 0·001). This implies that countries with higher number of physician and nurses have low child mortality  which this confirms our findings(20). Totally, better access to health workforce such as physician and nurse is related to improved service operation and health achievements, too(21). Inequality in distribution of healthcare resources especially physicians adversely effect on population health. Almost   rural and remote area of countries have fewer access to this resources than urban area. Therefore people and, especially children, usually have inappropriate health status and higher mortality in these zones. This issue is true in urban poor, too (22, 23).  

In the case of female literacy, it is important factor in child health and low literacy could  resulted in many adverse health outcome (24, 25). Literacy empowers women to create healthy households and increase their ability in use of healthcare services particularly in child bearing and child rearing periods. This could improve the family health especially for children(26). Altogether, our study depicted that female literacy could decreasing effect on under-five mortality rate. The result of many study which conducted for investigating the determinant of child mortality is consistent whit this finding (18, 27).

We used macro data of WHO data set. It is valid data set but this data may potentially have some errors. Also, some other factors may could effect on child mortality, then other study could try to investigate the effects of other variables on under-five mortality rate in OECD countries and other settings. We suggest to future study to consider micro data and other econometrics methods to provide new evidence about child mortality.

Conclusions

Our  results showed that all variables, including health expenditures, gross national income per capita, and density of physicians and nurses and female literacy, had invers effect on under-five mortality rate in OECD countries. Given the substantial focus of international public health programmes on improving child survival and importance of information about determinants of child mortality, this findings could be very useful for policy maker. Therefore, policy maker should consider the effects of this factors in planning for  improving the child survival and control of child mortality.

Conflict of interests: None

Acknowledgment

We thank all the staff and officials who helped us in this project.