Ethnic inequalities in child stunting and feeding practices: results from surveys in thirteen countries from Latin America.
ABSTRACT: BACKGROUND:Although the prevalence of child stunting is falling in Latin America, socioeconomic inequalities persist. However, there is limited evidence on ethnic disparities. We aimed to describe ethnic inequalities of stunting and feeding practices in thirteen Latin American countries using recent nationally representative surveys. METHODS:We analyzed national surveys carried out since 2006. Based on self-reported ethnicity, skin color or language, children were classified into three categories: indigenous/ afrodescendant/reference group (European or mixed ancestry). Stunting was defined as height (length)-for-age?
Project description:BACKGROUND:Guatemala has the highest prevalence of stunting among under-five children in Latin America. We aimed to compare indigenous and non-indigenous under-five child populations in relation to stunting, as well as to explore the intersectionality of ethnicity by wealth and by place of residence. We also studied how the ethnic inequalities changed over time, using five ENSMI surveys from 1995 to 2014. METHODS:Five national health surveys carried out between 1995 and 2014 were analysed. World Health Organization (WHO) 2006 growth standards were used to calculate stunting prevalence. Self-reported ethnicity was classified as indigenous or nonindigenous. Wealth was measured through an asset-based index, and households were classified into quintiles (for analyses of the whole populations) or tertiles (for analyses of intersectionality with ethnicity). Area of residence was recorded as urban or rural, according to country definition. RESULTS:Overall stunting prevalence declined by 9.8 percentage points (95% CI -16.4 to - 3.3) from 1995 to 2014. The slope index for absolute inequalities in stunting - which corresponds to the difference in prevalence between the wealthiest and poorest households - ranged from - 52.9 to - 60.4 percentage points, with no significant change over time. Children in rural areas were consistently more stunted than those in urban areas, but rural indigenous children were significantly worse than any other group. Indigenous children in the poorest tertile of family wealth consistently presented the highest stunting prevalence, compared to all other groups. Time trends in stunting were assessed through the average annual absolute change (AAAC). The fastest decline was observed among indigenous children from the middle wealth tertile (AAAC = - 1.21 percentage points per year (pp/y); 95% CI - 1.45 to - 0.96) followed by nonindigenous children also from the middle tertile (AAAC = - 0.80 pp./y; 95% CI - 0.99 to - 0.60). Stunting prevalence in the two poorest tertiles of indigenous children in 2015 was similar to what nonindigenous children presented in 1995, 20 years earlier. In the wealthiest tertile, indigenous children were far worse off than nonindigenous children 20 years earlier. CONCLUSIONS:In terms of stunting prevalence, poor and rural indigenous children are twenty years behind nonindigenous children with similar characteristics.
Project description:BACKGROUND:Latin American and Caribbean populations include three main ethnic groups: indigenous people, people of African descent, and people of European descent. We investigated ethnic inequalities among these groups in population coverage with reproductive, maternal, newborn, and child health interventions. METHODS:We analysed 16 standardised, nationally representative surveys carried out from 2004 to 2015 in Latin America and the Caribbean that provided information on ethnicity or a proxy indicator (household language or skin colour) and on coverage of reproductive, maternal, newborn, and child health interventions. We selected four outcomes: coverage with modern contraception, antenatal care coverage (defined as four or more antenatal visits), and skilled attendants at birth for women aged 15-49 years; and coverage with three doses of diphtheria-pertussis-tetanus (DPT3) vaccine among children aged 12-23 months. We classified women and children as indigenous, of African descent, or other ancestry (reference group) on the basis of their self-reported ethnicity or language. Mediating variables included wealth quintiles (based on household asset indices), woman's education, and urban-rural residence. We calculated crude and adjusted coverage ratios using Poisson regression. FINDINGS:Ethnic gaps in coverage varied substantially from country to country. In most countries, coverage with modern contraception (median coverage ratio 0·82, IQR 0·66-0·92), antenatal care (0·86, 0·75-0·94), and skilled birth attendants (0·75, 0·68-0·92) was lower among indigenous women than in the reference group. Only three countries (Nicaragua, Panama, and Paraguay) showed significant gaps in DPT3 coverage between the indigenous and the reference groups. The differences were attenuated but persisted after adjustment for wealth, education, and residence. Women and children of African descent showed similar coverage to the reference group in most countries. INTERPRETATION:The lower coverage levels for indigenous women are pervasive, and cannot be explained solely by differences in wealth, education, or residence. Interventions delivered at community level-such as vaccines-show less inequality than those requiring access to services, such as birth attendance. Regular monitoring of ethnic inequalities is essential to evaluate existing initiatives aimed at the inclusion of minorities and to plan effective multisectoral policies and programmes. FUNDING:The Bill & Melinda Gates Foundation (through the Countdown to 2030 initiative) and the Wellcome Trust.
Project description:BACKGROUND:With the adoption of the Sustainable Development Goals (SDGs), there is a renewed commitment of tackling the varied challenges of undernutrition, particularly stunting (SDG 2.2). Health equity is also a priority in the SDG agenda and there is an urgent need for disaggregated analyses to identify disadvantaged subgroups. We compared time trends in socioeconomic inequalities obtained through stratification by wealth quintiles and deciles for stunting prevalence. METHODS:We used 37 representative Demographic and Health Surveys and Multiple Indicator Cluster surveys from nine Latin American and Caribbean (LAC) countries conducted between 1996 and 2016. Stunting in children under-5?years was assessed according to the 2006 WHO Child Growth Standards and stratified by wealth quintiles and deciles. Within-country socioeconomic inequalities were measured through concentration index (CIX) and slope index of inequality (SII). We used variance-weighted least squares regression to estimate annual changes. RESULTS:Eight out of nine countries showed a statistical evidence of reduction in stunting prevalence over time. Differences between extreme deciles were larger than between quintiles in most of countries and at every point in time. However, when using summary measures of inequality, there were no differences in the estimates of SII with the use of deciles and quintiles. In absolute terms, there was a reduction in socioeconomic inequalities in Peru, Honduras, Dominican Republic, Belize, Suriname and Colombia. In relative terms, there was an increase in socioeconomic inequalities in Peru, Bolivia, Haiti, Honduras and Guatemala. CONCLUSIONS:LAC countries have made substantial progress in terms of reducing stunting,. Nevertheless, renewed actions are needed to improve equity. Particularly in those countries were absolute and relative inequalities did not change over time such Bolivia and Guatemala. Finer breakdowns in wealth distribution are expected to elucidate more differences between subgroups; however, this approach is relevant to cast light on those subgroups that are still lagging behind within populations and inform equity-oriented health programs and practices.
Project description:OBJECTIVE:Much is known about national trends in child undernutrition, but there is little information on how socio-economic inequalities are evolving over time. We aimed to assess socio-economic inequalities in stunting prevalence over time. DESIGN:We selected nationally representative surveys carried out since the mid-1990s for which information was available on asset indices and on child anthropometry. We identified twenty-five countries that had at least two surveys over an interval of 10 years or more, totalling eighty-seven surveys. Stunting prevalence was calculated according to wealth quintiles. Absolute and relative inequalities were calculated and time trends were obtained by regression. Setting Nationally representative household surveys from twenty-five low- and middle-income countries. SUBJECTS:Children <5 years of age. RESULTS:National prevalence declined significantly in twenty-two of the twenty-five countries. In eighteen out of twenty-five countries, relative reductions were higher among the rich than among the poor. Overall, there was no indication that inequalities improved. Striking examples are Nepal, with a 17·0 percentage points decline in stunting per decade, but where inequalities increased sharply; and Brazil, where stunting fell by 6·7 percentage points and inequalities were all but eliminated. CONCLUSIONS:Global progress in reducing stunting has not been accompanied by improved equity, but countries varied markedly in how successful they were in reducing prevalence among the poorest children. It is important to document how some countries were able to reduce inequalities, so that these lessons can be used to foster global progress, particularly in light of the increased importance of within-country inequalities in the post-2015 agenda.
Project description:<h4>Background</h4>Stunting is determined by using the World Health Organization (WHO) child growth standard which was developed using precise measurements. However, it is unlikely that large scale surveys maintain the same level of rigour and precision when measuring the height of children. The population measure of stunting in children is sensitive to over-dispersion, and the high prevalence of stunting observed in surveys in low and middle-income countries (LMIC) could partly be due to lower measurement precison.<h4>Objectives</h4>To quantify the incongruence in the dispersion of height-for-age in national surveys of <?5 y children, in relation to the standard WHO Multicenter Growth Reference Study (MGRS), and propose a measure of uncertainty in population measures of stunting.<h4>Methods</h4>An uncertainty factor was proposed and measured from the observed incongruence in dispersion of the height-for-age of <?5 y children in the MGRS against carefully matched populations from the Demographic Health Survey of 17 countries ('test datasets', based on the availability of data). This also allowed for the determination of uncertainty-corrected prevalence of stunting (height-for-age Z score?<?-?2) in <?5 y children.<h4>Results</h4>The uncertainty factor was estimated for 17 LMICs. This ranged from 0.9 to 2.1 for Peru and Egypt respectively (reference value 1). As an explicit country example, the dispersion of height-for-age in the Indian National Family Health Survey-4 test dataset was 39% higher than the MGRS study, with an uncertainty factor of 1.39. From this, the uncertainty-adjusted Indian national stunting prevalence estimate reduced to 18.7% from the unadjusted estimate of 36.2%.<h4>Conclusions</h4>This study proposes a robust statistical method to estimate uncertainty in stunting prevalence estimates due to incongruent dispersions of height measured in national surveys for children <?5?years in relation to the WHO height-for-age standard. The uncertainty is partly due to population heterogeneity, but also due to measurement precision, and calls for better quality in these measurements.
Project description:Global stunting prevalence has been nearly halved between 1990 and 2016, but it remains unclear whether this decline has benefited poor and rural populations within low- and middle-income countries (LMICs).We assessed time trends in stunting among children <5 y of age (under-5) according to household wealth and place of residence in 67 LMICs.Stunting prevalence was analyzed in 217 nationally representative Demographic and Health Surveys and Multiple Indicator Cluster Surveys from 67 countries with ?2 surveys between 1993 and 2014. National estimates were stratified by wealth and area of residence, comparing the poorest 40% with the wealthiest 60%, and those residing in urban and rural areas. Time trends were calculated for LMICs by using multilevel regression models weighted by under-5 population, with stratification by wealth and by residence. Trends in absolute (slope index of inequality; SII) and relative (concentration index; CIX) inequalities were calculated.Mean prevalences in 1993 were 53.7% in low-income and 48.2% in middle-income countries, with annual average linear declines of 0.76 and 0.72 percentage points (pp), respectively. Although similar slopes of declines were observed for the poorest 40% and wealthiest 60% groups in all countries (0.78 and 0.74 pp, respectively), absolute and relative inequalities increased over time in low-income countries (SII increased from -19.3% in 1993 to -23.7% in 2014 and CIX increased from -6.2% to -10.8% in the same period). In middle-income countries, socioeconomic inequalities remained stable. Overall, stunting prevalence decreased more rapidly among rural than for urban children (0.78 and 0.55 pp, respectively).The prevalence of stunting is decreasing. Poor-rich gaps are stable in middle-income countries and slightly increasing in low-income countries. Rural-urban inequalities are decreasing over time.
Project description:<h4>Background</h4>Global progress in reducing the burden of undernutrition tends to be measured at the population level. It has been hypothesized that population-level improvements may mask widening socioeconomic inequalities, but little attempt has been made to assess whether this is true.<h4>Methods</h4>Original data from 131 demographic health surveys and 48 multiple indicator cluster surveys from 1990 to 2011 were used to examine trends in socioeconomic inequalities in stunting and underweight, as well as the relationship between changes in prevalence and changes in inequality, in 80 countries. Socioeconomic inequality is measured using the corrected concentration index.<h4>Results</h4>Countries with a higher prevalence of stunting tend to have larger socioeconomic inequalities in stunting (Spearman rank correlation = -0.27 P = 0.014). In most countries, there has been no change in inequality in stunting: in 31 out of 53, the 90% confidence intervals around the changes overlap the zero value. In the remaining 22, there was a reduction in inequality in 11 and an increase in 11. The distributional patterns underlying the summary inequality statistics vary considerably across countries, but in most there have been considerable gains to the poorest quintile.<h4>Conclusions</h4>Reductions in the prevalence of undernutrition have generally not been accompanied by widening inequalities. However, inequalities have also not been narrowing. Rather, the picture is one of a strong persistence of existing inequalities. In addition, there are different distributional patterns underlying changes in the summary indices of inequality which will need to be taken into consideration in designing programmes to reach the poor.
Project description:BACKGROUND:The rise in contraceptive use has largely been driven by short-acting methods of contraception, despite the high effectiveness of long-acting reversible contraceptives. Several countries in Latin America and the Caribbean have made important progress increasing the use of modern contraceptives, but important inequalities remain. We assessed the prevalence and demand for modern contraceptive use in Latin America and the Caribbean with data from national health surveys. METHODS:Our data sources included demographic and health surveys, multiple indicator cluster surveys, and reproductive health surveys carried out since 2004 in 23 countries of Latin America and the Caribbean. Analyses were based on sexually active women aged 15-49 years irrespective of marital status, except in Argentina and Brazil, where analyses were restricted to women who were married or in a union. We calculated contraceptive prevalence and demand for family planning satisfied. Contraceptive prevalence was defined as the percentage of sexually active women aged 15-49 years who (or whose partners) were using a contraceptive method at the time of the survey. Demand for family planning satisfied was defined as the proportion of women in need of contraception who were using a contraceptive method at the time of the survey. We separated survey data for modern contraceptive use by type of contraception used (long-acting, short-acting, or permanent). We also stratified survey data by wealth, area of residence, education, ethnicity, age, and a combination of wealth and area of residence. Wealth-related absolute and relative inequalities were estimated both for contraceptive prevalence and demand for family planning satisfied. FINDINGS:We report on surveys from 23 countries in Latin America and the Caribbean, analysing a sample of 212?573 women. The lowest modern contraceptive prevalence was observed in Haiti (31·3%) and Bolivia (34·6%); inequalities were wide in Bolivia, but almost non-existent in Haiti. Brazil, Colombia, Costa Rica, Cuba, and Paraguay had over 70% of modern contraceptive prevalence with low absolute inequalities. Use of long-acting reversible contraceptives was below 10% in 17 of the 23 countries. Only Cuba, Colombia, Mexico, Ecuador, Paraguay, and Trinidad and Tobago had more than 10% of women adopting long-acting contraceptive methods. Mexico was the only country in which long-acting contraceptive methods were more frequently used than short-acting methods. Young women aged 15-17 years, indigenous women, those in lower wealth quintiles, those living in rural areas, and those without education showed particularly low use of long-acting reversible contraceptives. INTERPRETATION:Long-acting reversible contraceptives are seldom used in Latin America and the Caribbean. Because of their high effectiveness, convenience, and ease of continuation, availability of long-acting reversible contraceptives should be expanded and their use promoted, including among young and nulliparous women. In addition to suitable family planning services, information and counselling should be provided to women on a personal basis. FUNDING:Wellcome Trust, Pan American Health Organization.
Project description:BACKGROUND:The current focus on monitoring health inequalities and the complexity around ethnicity requires careful consideration of how ethnic disparities are measured and presented. This paper aims to determine how inequalities in maternal healthcare by ethnicity change according to different criteria used to classify indigenous populations. METHODS:Nationally representative demographic surveys from Bolivia, Guatemala, Mexico, and Peru (2008-2016) were used to explore coverage gaps across maternal health care by ethnicity using different criteria. Women were classified as indigenous through self-identification (SI), spoken indigenous language (SIL), or indigenous household (IH). We compared the gaps through measuring coverage ratios (CR) with adjusted Poisson regression models. RESULTS:Proportions of indigenous women changed significantly according to the identification criterion (Bolivia:SI-63.1%/SIL-37.7%; Guatemala:SI-49.7%/SIL-28.2%; Peru:SI-34%/SIL-6.3% & Mexico:SI-29.7%/SIL-6.9%). Indigenous in all countries, regardless of their identification, had less coverage. Gaps in care between indigenous and non-indigenous populations changed, for all indicators and countries, depending on the criterion used (e.g., Bolivia CR for contraceptive-use SI?=?0.70, SIL?=?0.89; Guatemala CR for skilled-birth-attendant SI?=?0.77, SIL?=?0.59). The heterogeneity persists when the reference groups are modified and compare just to non-indigenous (e.g., Bolivia CR for contraceptive-use under SI?=?0.64, SIL?=?0.70; Guatemala CR for Skilled-birth-attendant under SI?=?0.77, SIL?=?0.57). CONCLUSIONS:The indigenous identification criteria could have an impact on the measurement of inequalities in the coverage of maternal health care. Given the complexity and diversity observed, it is not possible to provide a definitive direction on the best way to define indigenous populations to measure inequalities. In practice, the categorization will depend on the information available. Our results call for greater care in the analysis of ethnicity-based inequalities. A greater understanding on how the indigenous are classified when assessing inequalities by ethnicity can help stakeholders to deliver interventions responsive to the needs of these groups.
Project description:Stunting in children less than five years of age is widespread in Sub-Saharan Africa. We aimed to: (i) evaluate how the prevalence of stunting has changed by socio-economic status and rural/urban residence, and (ii) assess inequalities in children's diet quality and access to maternal and child health care. We used data from nationally representative demographic and health- and multiple indicator cluster-surveys (DHS and MICS) to disaggregate the stunting prevalence by wealth quintile and rural/urban residence. The composite coverage index (CCI) reflecting weighed coverage of eight preventive and curative Reproductive, Maternal, Neonatal, and Child Health (RMNCH) interventions was used as a proxy for access to health care, and Minimum Dietary Diversity Score (MDDS) was used as a proxy for child diet quality. Stunting significantly decreased over the past decade, and reductions were faster for the most disadvantaged groups (rural and poorest wealth quintile), but in only 50% of the countries studied. Progress in reducing stunting has not been accompanied by improved equity as inequalities in MDDS (p < 0.01) and CCI (p < 0.001) persist by wealth quintile and rural-urban residence. Aligning food- and health-systems' interventions is needed to accelerate stunting reduction more equitably.