An ecological epidemiological analysis was carried out based on secondary information using standard metrics for measuring health inequalities among provinces, the first subnational level in Ecuador. The territorial organization of the Ecuadorian State establishes decentralized governments with political, administrative and financial autonomy [23]. These territorial areas, called provinces, have been grouped, according to their location, into four regions: La Sierra (Andean highlands), El Oriente (Amazon region), La Costa (Pacific lowlands), and the Galápagos Islands. The Andean highlands region is made up of the provinces of Carchi, Imbabura, Pichincha, Cotopaxi, Tungurahua y Chimborazo, Bolívar, Cañar, Azuay and Loja. The Amazon Region is composed of Sucumbíos, Napo, Pastaza, Orellana, Morona Santiago, and Zamora Chinchipe. The Pacific lowlands contains the provinces of Esmeraldas, Manabí, Los Ríos, Guayas and El Oro. The Galapagos Islands are composed of 13 main islands [24]. These provinces have enormous differences in terms of socioeconomic and health conditions [25].
According to the rules of operation, these governments have as their main function local implementation of national programs, which are financed and monitored by the central government [23].
The purpose of the analysis was to quantify the magnitude of the relationship between poverty, access to services, and SRH outcomes. To do this, it was necessary to have disaggregated information that allowed for comparisons over time and at the population level. The largest disaggregation of information, for the period studied, was at the provincial level.
As measures of gaps, we estimated absolute (the magnitude of the difference in health between subgroups) and relative (the ratio of differences in health indicators between subgroups). For measures of gradient, the slope index of inequality (SII) and the relative index of inequality (RII). Finally, the Wagstaff adjusted Concentration Index, as described below.
All analysis used as the socioeconomic indicator (stratifier), the multidimensional poverty index (MPI) of Ecuador, that is, the official measure of poverty in the country. The MPI comprises 4 equally weighted dimensions (education; social security and employment; health, nutrition and water; and environment, habitat and dwelling) and 12 indicators. Education indicators are related to school attendance and schooling achievements. Social security measures pension contribution, while employment measures unemployment and child/adolescent work. Health, nutrition and water uses income and access to water. Finally, habitat and environment uses access to sanitation and trash collection, while dwelling conditions uses overcrowding and dwelling deficit. An individual is considered as poor if they are identified as deprived in at least one third of the indicators [26].
Analyzed indicators
In the development agenda, signed by most countries in the world in the year 2000, it was agreed to improve maternal health, which meant a reduction of 75% in the maternal mortality rate between 1990 and 2015 [8]. Reducing the maternal mortality rate necessitated increasing the quality of the obstetric health care. In underdeveloped and developing countries, the aim of increasing the quality of obstetrics care is attempted through increasing hospital coverage of obstetric emergency [27, 28]. Similarly, adolescent pregnancies are a global priority, due to the social impact and the risk they represent for adolescent health. As such, the indicators of home birth, obstetric complications, and fertility in adolescents, are elements affecting the frequency of maternal mortality. For these reasons and due to the availability of information for all provinces in the period analyzed, the indicators included in the present investigation were selected.
Obstetric complications ratio
This addresses the morbidity that occurs during pregnancy, delivery, and puerperium, and is defined as the ratio of women who received care for gynecological and obstetric events, identified through ICD-10 codes (Codes: 000–099) among total live births in the same period, multiplied by one thousand.
Post-abortion complications ratio is a subcategory of obstetric complications indicator that considers only women with abortion discharges.
Ratio of births attended at home. Number of births attended at home per 1000 live births as an indicator of access to reproductive health services. Evidence has suggested that compared to hospital deliveries, there are suboptimal outcomes for both mother and child at home deliveries [29, 30].
Percentage of births by caesarean section
Indicator of structural quality of reproductive health services, calculated as the number of births by Caesarean section over the number of deliveries attended in the same period.
Age-specific fertility rate 15–19 (adolescent pregnancy) (APFR 15–19)
This is the age-specific fertility rate for the group of women aged 15 to 19 years. The indicator was calculated as the quotient of the number of live births from women in the group of 15 to 19, in a given geographical area in a year, by every thousand women in that age group. The legal age for marriage in Ecuador is 18 as of 2015, but data from 2017 reports that 22% of Ecuadorian girls are married before that age [31]. Early childbearing negatively affects both women’s and children’s wellbeing [32, 33].
Maternal mortality ratio (MMR)
This is the main outcome indicator of sexual and reproductive health care, and a central indicator in the Sustainable Development Goals. It is calculated as the quotient of the number of maternal deaths in women from 15 to 49 selecting ICD-10 codes O00 to O99, excluding codes O96 and O97 (late maternal deaths) according to the standard procedure in Ecuador for identifying maternal deaths, over the total number of births, multiplied by 100,000.
Information sources
Public secondary information was used from the National Institute of Statistics and Censuses (Instituto Nacional de Estadística y Censos, INEC), the National Secretariat for Planning and Development (Secretaria Nacional de Planificación y Desarrollo, SENPLADES), the Coordinating Ministry for Social Development (Ministerio Coordinador de Desarrollo Social, MCDS) and the Ministry of Public Health of Ecuador (Ministerio de Salud Pública del Ecuador, MSP). In particular, information was extracted from the following sources that are collected by the National Institute of Statistics and Censuses (INEC) of Ecuador:
Analysis of trends in indicators
To explore the behavior of the indicators over time, we first did a descriptive analysis of their values between 2008 and 2015 for Ecuador as a whole, calculating the value for each year according to the previous definitions. This analysis made it possible to identify indicators that are overall improving, worsening or remaining unchanged during the period.
Analysis of health inequalities
To generate an intra-country analyses of health inequalities in Ecuador, we compared provinces in the country by the different indicators and using the different metrics of inequalities.
For the gap analysis (absolute and relative), we first ordered provinces by the percentage of households living in poverty (using the multidimensional measure of poverty) and then divided them into 5 equal size groups (quintiles) [34]. For each quintile, we estimated the different indicators as the weighted averages of the indicator value in the provinces in that quintile:
$$ {SRHI}_q=\sum \limits_i^n{SRHI}_i\ast \frac{Pop_i}{POP_q} $$
where SRHIq is the indicator for the quintile (1 to 5), SRHIi is the value of the indicator in province i and \( \frac{Pop_i}{POP} \) measures the relative size of the population of each province related to the population of all province in that quintile.
For the gap analysis, the absolute gap is the difference in SRHIq between quintiles 5 and 1, while the relative gap is the ratio between the same quintiles:
$$ AbsoluteGap={SRHI}_{q5}-{SRHI}_{q1} $$
$$ RelativeGap=\frac{SRHI_{q5}}{SRHI_{q1}} $$
The absolute gap will result in a positive value if the indicator is pro-rich (higher value for the quintile 5) and negative if pro-poor. The relative gap will be > 1 for pro-rich indicators and < 1 for pro-poor indicators.
We estimated the slope index of inequality (SII) using linear regression models where the dependent variable is the value of the indicator at the province level and the independent variable is the ridit for the percentage of households living in poverty by province [34,35,36]:
$$ {SRHI}_i=\alpha +\beta {Ridit}_i+\varepsilon $$
where Riditi is the ridit score of the share of poverty in each province, α is the intercept or the level of SRHI in absence of poverty, β is the SII and ε is the term of error. The SII measures the difference in the outcome between the provinces in the extremes of the actual distribution. For this case, negative values of the SII indicates that the SRHI is higher at the poorest province compared with the richest province. The SII is an absolute measure that is sensitive to changes in the level of the health indicator in the population even if differences among provinces remain constant.
Thus, we also estimated the relative index of inequality that is the ratio of the coefficients in the estimated regression, equivalent to the ratio on the SRHI at the extremes of the poverty distribution, that is the difference of zero poverty versus 100% poverty [35, 37]:
$$ RII=\frac{\widehat{SRHI_0}}{\ \widehat{SRHI_1}}=\frac{\alpha }{\ \alpha +\beta } $$
We finally estimated the Concentration Index, which is derived from a concentration curve of the health indicator and the population ordered by a socioeconomic indicator,in this case, provinces ordered by share of poverty. The CI is generally defined as twice the area between the concentration curve and the line of equality (the 45-degree line), and ranges from − 1 to 1. For CI, the zero value is an indication of absence of inequality, while for this analysis, positive values indicates pro-poor indicators, that is, indicators that present higher values for provinces with a larger share of poor households and, in the opposite, negative CI values identify pro-rich indicators [34, 35, 38].
We did all the analyses in Stata 15.