Exploring Heterogeneity in Treatment Effects: The Impact and Interaction of Asset-Based Wealth and Mass Azithromycin Distribution on Child Mortality

Exploring Heterogeneity in Treatment Effects: The Impact and Interaction of Asset-Based Wealth and Mass Azithromycin Distribution on Child Mortality | medRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-P4HH5NV'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search Exploring Heterogeneity in Treatment Effects: The Impact and Interaction of Asset-Based Wealth and Mass Azithromycin Distribution on Child Mortality View ORCID Profile Elisabeth A. Gebreegziabher , View ORCID Profile Ali Sié , View ORCID Profile Mamadou Ouattara , Mamadou Bountogo , Boubacar Coulibaly , Valentin Boudo , Thierry Ouedraogo , Elodie Lebas , View ORCID Profile Huiyu Hu , View ORCID Profile Pearl Anne Ante-Testard , Steven E Gregorich , View ORCID Profile Kieran S. O’Brien , View ORCID Profile Michelle S. Hsiang , View ORCID Profile David V. Glidden , View ORCID Profile Benjamin F. Arnold , View ORCID Profile Thomas M. Lietman , View ORCID Profile Catherine E. Oldenburg doi: https://doi.org/10.1101/2025.07.05.25329685 Elisabeth A. Gebreegziabher 1 Francis I. Proctor Foundation, University of California San Francisco , San Francisco, CA, USA 3 Department of Epidemiology and Biostatistics, University of California , San Francisco Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Elisabeth A. Gebreegziabher Ali Sié 2 Centre de Recherche en Sante de Nouna , Nouna, Burkina Faso Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Ali Sié Mamadou Ouattara 2 Centre de Recherche en Sante de Nouna , Nouna, Burkina Faso Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Mamadou Ouattara Mamadou Bountogo 2 Centre de Recherche en Sante de Nouna , Nouna, Burkina Faso Find this author on Google Scholar Find this author on PubMed Search for this author on this site Boubacar Coulibaly 2 Centre de Recherche en Sante de Nouna , Nouna, Burkina Faso Find this author on Google Scholar Find this author on PubMed Search for this author on this site Valentin Boudo 2 Centre de Recherche en Sante de Nouna , Nouna, Burkina Faso Find this author on Google Scholar Find this author on PubMed Search for this author on this site Thierry Ouedraogo 2 Centre de Recherche en Sante de Nouna , Nouna, Burkina Faso Find this author on Google Scholar Find this author on PubMed Search for this author on this site Elodie Lebas 1 Francis I. Proctor Foundation, University of California San Francisco , San Francisco, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Huiyu Hu 1 Francis I. Proctor Foundation, University of California San Francisco , San Francisco, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Huiyu Hu Pearl Anne Ante-Testard 1 Francis I. Proctor Foundation, University of California San Francisco , San Francisco, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Pearl Anne Ante-Testard Steven E Gregorich 5 Department of Medicine, University of California , San Francisco, San Francisco, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Kieran S. O’Brien 1 Francis I. Proctor Foundation, University of California San Francisco , San Francisco, CA, USA 4 Department of Ophthalmology, University of California , San Francisco, San Francisco, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Kieran S. O’Brien Michelle S. Hsiang 3 Department of Epidemiology and Biostatistics, University of California , San Francisco 6 Institute for Global Health Sciences, University of California , San Francisco, USA 7 Department of Pediatrics, Division of Pediatric Infectious Diseases , UCSF Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Michelle S. Hsiang David V. Glidden 3 Department of Epidemiology and Biostatistics, University of California , San Francisco Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for David V. Glidden Benjamin F. Arnold 1 Francis I. Proctor Foundation, University of California San Francisco , San Francisco, CA, USA 4 Department of Ophthalmology, University of California , San Francisco, San Francisco, CA, USA 6 Institute for Global Health Sciences, University of California , San Francisco, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Benjamin F. Arnold Thomas M. Lietman 1 Francis I. Proctor Foundation, University of California San Francisco , San Francisco, CA, USA 3 Department of Epidemiology and Biostatistics, University of California , San Francisco 4 Department of Ophthalmology, University of California , San Francisco, San Francisco, CA, USA 6 Institute for Global Health Sciences, University of California , San Francisco, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Thomas M. Lietman Catherine E. Oldenburg 1 Francis I. Proctor Foundation, University of California San Francisco , San Francisco, CA, USA 3 Department of Epidemiology and Biostatistics, University of California , San Francisco 4 Department of Ophthalmology, University of California , San Francisco, San Francisco, CA, USA 6 Institute for Global Health Sciences, University of California , San Francisco, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Catherine E. Oldenburg For correspondence: catherine.oldenburg{at}ucsf.edu Abstract Full Text Info/History Metrics Data/Code Preview PDF Abstract Objective To examine how child mortality among children aged 1–59 months varies by asset-based wealth status in rural Burkina Faso and to assess the interaction between mass azithromycin distribution and wealth status on child mortality at both the household and community levels. Methods We used data from a cluster-randomized trial and a population census data on household characteristics and assets. A wealth index score for each household, used to classify the population by wealth was generated using principal component analysis. We used the Relative Index of Inequality (RII), the Slope Index of Inequality (SII), and the concentration index to assess wealth-related inequalities in mortality, and the Gini Index to assess variability in child mortality across households and communities. Poisson regression models were used with person-time at risk included as an offset and robust standard error to estimate changes in mortality rates by wealth and treatment arm. We assessed interaction on both the multiplicative and additive scales. Results Mortality declined with increasing wealth at both household and community levels, with a significant gradient at the community level (RII = 1.17, 95% CI: 1.05–1.29; SII = 2.3 per 1,000 person-years, 95% CI: 0.2–4.4), reflecting higher mortality among the poorest. The effect of AZ did not vary significantly by wealth index, and the change in mortality rates across wealth levels was similar between the two treatment arms. There was no statistically significant interaction between AZ and asset-based wealth on either a multiplicative or additive scale at the household or cluster level. Conclusion Our findings show a wealth gradient in child mortality, with households and communities in the poorest quintiles experiencing the highest mortality rates. These disparities were consistent across both AZ-treated and placebo groups, suggesting that AZ’s role in health disparities may primarily address gaps in treatment access rather than broader wealth-related disparities. AZ appears to offer similar benefits across economically diverse communities, with no evidence suggesting enhanced benefits for disadvantaged communities or for prioritizing treatment based on wealth status. Further work is needed to address the wealth-related disparities in child mortality in these communities. Trial Registration ClinicalTrials.gov Identifier: NCT03676764 Introduction The Sustainable Development Goals aim to end preventable deaths of children under 5 by 2030.[ 1 ] Although significant progress has been made in reducing global child mortality, improving child survival remains an international priority.[ 2 ] As of 2022, the global under-five mortality rate was 37 deaths per 1000 live births, a significant decline from the 93 deaths per 1,000 live births in 1990.[ 2 ] However, there is considerable variability in mortality rates and improvements across regions,[ 3 ] with substantial inequalities observed in sub-Saharan Africa.[ 4 ] For instance, in 2019, two regions accounted for the majority of under-5 deaths: sub-Saharan Africa with 55% (53–57) of global under-5 deaths, and south Asia with 26% (26–27) of the total.[ 5 ] Considering such disparities, assessing local and regional trends can help identify areas for improvement despite overall progress, and help prioritize child survival interventions for those who need them most.[ 3 ] Understanding context-specific differences is essential for evaluating health equity, informing policy decisions, and effectively addressing local health disparities and their specific challenges. The leading causes of child mortality, particularly in sub-Saharan Africa, include infections such as malaria, diarrhea, and pneumonia.[ 6 ] Addressing these issues involve strengthening health systems, implementing nutritional interventions, improving water, sanitation, and hygiene services, and preventing infections.[ 7 ] Antibiotic-based strategies, such as mass distribution of azithromycin (AZ), have also been used to reduce child mortality in these settings. Previous randomized controlled trials, such as the MORDOR study, conducted in Malawi, Niger, and Tanzania, as well as the recent community-randomized CHAT trial in Burkina Faso, and the AVENIR trial in Niger have shown that childhood mortality was lower in communities that received mass distributions of oral AZ.[ 8 , 9 , 10 ] Previous evidence suggests that AZ is not only effective in reducing child mortality but can also help buffer against disparities in child mortality by potentially addressing gaps in treatment.[ 11 ] A secondary analysis of the MORDOR trial found that the effect of MDA with AZ was larger in communities farther from clinics.[ 11 ] Since distance from health facilities is a barrier to accessing routine and life-saving interventions,[ 12 , 13 ] hard-to-reach and rural communities, which often lack resources, [ 13 , 14 ] may benefit more from mass AZ treatment. Another key determinant of access to treatment and care is socioeconomic status (SES).[ 15 ] Socioeconomic factors are known to be important determinants of child mortality, by which disparities in health outcomes have been previously noted.[ 4 , 16 ]Individuals in poorer communities often have less access to quality care, services and information,[ 15 ] and may therefore also benefit more from mass AZ treatment. However, despite its suggested effect in addressing gaps in treatment, there is no clear evidence whether MDA of AZ interacts with wealth status to help reduce wealth-related inequalities in child mortality. Additionally, poorer communities may have greater exposure to infectious diseases due to crowding, poorer housing structures, and limited access to clean water and sanitation.[ 17 ] Consequently, the poorest communities may experience a higher infection burden and benefit more (have more harm averted) from treatment.[ 18 ] For instance, previous research shows that reducing childhood infections, particularly in families with lower SES, may more substantially reduce the burden of cardiovascular disease in adults compared to families with higher SES.[ 19 ] Therefore, examining whether the effect of AZ varies by wealth level may help determine whether AZ MDA is more beneficial for households and communities with lower SES, which typically have reduced access to resources, healthcare, and antibiotics, and aid in resource prioritization and the development of targeted approaches. Evaluating these effects at both the household and cluster/community levels may provide a comprehensive understanding of how MDA interventions could affect wealth-related inequalities. Since socioeconomic determinants operate at multiple levels such as individual, household, and community,[ 20 ] this approach may provide insights into effects within individual households, as well as broader patterns and resource gaps at the community level, enabling more targeted strategies. Using data from a cluster randomized trial of 278 villages in Burkina Faso[ 9 ] (CHAT) and data on household characteristics, we examined whether there is an asset-based wealth gradient in under-five mortality in these communities and whether there is an interaction between mass AZ distribution and wealth status on child mortality. Methods Study Design, Setting, and Population This was a secondary analysis that utilized data from the CHAT trial (ClinicalTrials.gov, NCT03676764 ) and a pre-census survey data collected prior to the first CHAT census, before the trial began. The CHAT trial was a cluster-randomized study aimed at assessing the effectiveness of mass AZ distribution for prevention of mortality in children aged 1-59 months.[ 21 ] It was conducted The trial was conducted from August 2019 to February 2023 and involved children under-five in the Nouna District of Burkina Faso, in both the Nouna Health and Demographic Surveillance Site (HDSS) and the surrounding non-HDSS area.[ 21 ] Under-five mortality in Burkina Faso declined substantially,[ 22 ] from 184 per 1,000 live births in 2003 to 48 per 1,000 live births in 2021, largely due to improvements in prevention and care.[ 1 ] Malaria, pneumonia, and diarrhea were the leading causes of child death.[ 23 ] A previous study found that the median rate of visits to government-run primary healthcare facilities for children under five in the villages was 6.7 per 100 child-months, with the majority due to pneumonia (37.5%), malaria (25.1%), and diarrhea (9.1%).[ 13 ] The median distance from healthcare facilities for a child was approximately 5 km.[ 13 ] Over 40% of the population of Burkina Faso lives below the poverty line,[ 24 ] with poverty concentrated in rural areas.[ 25 ] Data Collection In the CHAT trial, mass drug administration (MDA) of AZ or placebo was assigned at the cluster, specifically at the community level. Eligible children in treatment clusters received twice yearly doses of oral AZ, while those in placebo clusters received twice yearly doses of oral placebo over a period of 3 years (2019-2023; 6 treatment distributions in total).[ 21 ] A census was conducted every 6 months to record births, deaths, pregnancies, and migrations. The study employed an open cohort design, with varying person-time contributions from enrolled children as children could age in or out of the cohort or could move in or away or die. Vital status updates for each child, along with their household and community identifiers, were recorded during each 6-month census phase. Throughout the study, six rounds of census and treatment were performed, and a total of 1,086 deaths were observed across 119,139 person-years.[ 9 ] Detailed methods of the CHAT trial have been described previously.[ 9 , 21 ] The CHAT trial was reviewed and approved by the Institutional Review Boards at the University of California, San Francisco, and the Comité National d’Ethique pour la Recherche in Ouagadougou, Burkina Faso. Written informed consent was obtained from the caregiver of each enrolled child. Before the study began, the study team conducted a pre-census survey in study communities, collecting data on household-level wealth indicators in regions outside of the HDSS. These data included habitat type (e.g.simple detached house, villa, multiple dwelling building, apartment building), ownership status (e.g., owner, tenant, housed by employer/parents/friends, other), and housing structure, such as the construction material used for walls (hard, semi-hard, banco, straw), floors (tiles, cement, clay, sand), and roofs (concrete, tiles, sheets, straw/leaf, crammed earth). It also included the main source of cooking energy (electricity, gas, coal/wood, petroleum stove) and access to clean water and toilets (e.g., type of toilet such as flush toilet, latrines, unconfined latrines, no toilet) and the main source of drinking water (faucet, well, drilling, river). Data on asset ownership was also collected, including items such as radio, television, video/DVD player, telephone, freezer, gas cooker, generator, computer, car, motor, and tricycle. The pre-census data covered 15,291 households in 170 clusters involved in the CHAT trial, while 12,872 households in CHAT were not included in the pre-census. Statistical Analysis Methods We conducted three steps: 1) generated the wealth index score for each household, 2) assessed how child mortality rates change with the asset-based wealth index, and 3) evaluated the interaction between the azithromycin MDA intervention and wealth on all-cause mortality. The latter two analyses were conducted at both the household and cluster levels. Descriptive analyses were used to describe the characteristics of households from the pre-census, by treatment arm. While the overall sample size was based on childhood mortality for the AZ versus placebo comparison in CHAT trial,[ 9 ] only households present in both the pre-census and the trial were included in the household-level analyses, while clusters with at least one household in the pre-census were included in the cluster level analyses. Principal Component Analysis (PCA) The wealth index score for each household was created using PCA, which is a data reduction technique that reduces dimensionality by replacing many correlated variables with a smaller set of uncorrelated ’principal components’ that explain most of the variance.[ 26 , 27 ] To prepare variables for PCA, household characteristics were recoded as either improved or unimproved based on DHS categorization.[ 28 ] Ownership of assets was recorded as binary. We assessed household characteristics, their correlations, eigenvalues and the factors that explained majority of the variability. Of these characteristics/assets, 14 were selected for inclusion in the PCA based on higher prevalence, component loading, and their positive association with wealth status. These included improved wall, floor, roof, and toilet, as well as ownership of radio, TV, video/DVD, mobile, freezer, solar plate, solar lamp, motorcycle, bicycle and cart. We used the first component that explained the largest proportion of the total variance to create the wealth score, with higher scores representing wealthier households. We categorized the wealth index into quintiles, dividing households into five groups from the least wealthy 20% to the wealthiest 20% based on this relative measure of poverty.[ 29 ] Wealth status and Mortality The wealth index score for each household was merged with the mortality data, which included the number of deaths, person-time at risk (aggregated by household), and the corresponding treatment value (AZ vs placebo) for the cluster in which each household was located. We examined whether mortality rate in children under five changes with increase in each quintile of the wealth index. Quintiles were analyzed both as a linear ordinal predictor and as categorical variables (with the least wealthy group as the reference), following tests for linear trends across categories. We used a Poisson regression model with person-years as an offset, a log link, and robust standard errors clustered at the cluster level to account for the cluster-level treatment in the household-level analysis. The margins command in Stata was used to estimate incidence rate differences (IRDs) with 95% CIs using the delta method. In the adjusted analysis, we included distance to facility as a covariate, since it has been previously identified as strong predictor of child mortality in this setting,[ 30 ] and is often associated with wealth.[ 31 ] We also assessed wealth-related inequalities in mortality by calculating the Relative Index of Inequality (RII) and the Slope Index of Inequality (SII) which are key measures used in epidemiologic studies to quantify and compare health inequalities across populations.[ 32 ] The RII represents the ratio of predicted outcomes between the wealthiest and poorest quintiles in the wealth distribution of the population, whereas the SII reflects the absolute difference between these groups. [ 32 ] RII and SII values of 1 and 0 indicate no inequality, respectively, while higher values indicate worse outcomes for the most disadvantaged group. We used bootstrapping with 1,000 replicates, resampling clusters with replacement, to calculate the 95% confidence limits for these estimates. Additionally, we used the Gini index and the concentration index. The Gini index, commonly used to measure income inequality, is also used to measure health inequality by providing estimates that capture the distribution of health or health risks.[ 33 , 34 ] We used the Gini index to assess the overall distribution and degree of inequality in child mortality across households and communities. The Gini index ranges from 0 to 1, where a value of 0 represents perfect equality and a value of 1 indicates perfect inequality.[ 34 ] The concentration index captures the extent to which health outcomes differ across individuals or communities ranked by an indicator of socioeconomic status, and is commonly used to measure socioeconomic related health inequality.[ 35 ] The concentration index was used to measure how child mortality is concentrated among different wealth groups. It is defined as twice the area between the concentration curve and the 45° line, which represents equality, and ranges from -1 to 1. For an ill-health outcome (such as child mortality), a negative concentration index indicates that ill health is higher among the poor. A value of - 1 represents maximal pro-rich inequality (disproportionate burden on the poor), 0 indicates no inequality, and 1 represents maximal pro-poor inequality (disproportionate burden on the wealthy).[ 35 , 36 ] Wealth-AZ Interaction We examined whether the effect of AZ on child mortality varies by asset-based wealth and whether the relationship between wealth and mortality varies by treatment arm. We used the interaction directed acyclic graph (DAG)[ 37 ] shown in Figure 1A and 1B to determine and visualize the relationships between variables relevant for the interaction analyses. We used Poisson regression models similar to those described above, including both the main effects of wealth and AZ, as well as their interaction term to assess interaction on a multiplicative scale. To evaluate interaction on an additive scale, we calculated the relative excess risk due to interaction (RERI) with bootstrap 95% confidence intervals, resampling clusters with replacement, using 1000 repetitions. Based on the interaction DAG, distance to facility was included as a covariate in adjusted models. Download figure Open in new tab Figure 1 Interaction Directed Acyclic Graphs Illustrating the Interaction Between Mass AZ Treatment and Asset-Based Wealth In addition to household-level analyses, we conducted these analyses at the cluster level by aggregating deaths and person-time by cluster and averaging household wealth scores to determine wealth quintiles for each cluster. This approach provided insights into community-level relationships and allowed the use of pre-census household wealth data to calculate the wealth index for clusters with at least one household in the pre-census. As a result, clusters with at least one household in the pre-census were included in the cluster-level analysis, enabling the inclusion of mortality data from clusters missing households in the pre-census data. Sensitivity Analyses We conducted several sensitivity analyses. First, since the first 4 principal components explained a much larger proportion of the total variance than just the first component, we combined them (using weighted sum) to create the wealth score, which was then used in analyses similar to those conducted for the wealth index constructed from the first component.[ 26 ] Second, because households not included in the pre-census were excluded from the household-level analysis, we examined whether the effect of AZ on mortality varied by selection/inclusion into the analysis ( Figure 1C ). Third, we assessed whether selection/inclusion into the household-level analyses varied by treatment arm to determine if changes in mortality by wealth could be related to selection through treatment ( Figure 1D ). The latter two sensitivity analyses were conducted to assess potential selection bias and determine whether the results from the selected population can be generalized to the full sample and target population.[ 37 ] SAS 9.4 (SAS Institute, Cary, NC) was used for data cleaning and for generating the datasets for analyses. Stata version 14.2 (StataCorp, College Station, TX) was used for all other analyses. Results There were 15,291 households in 170 clusters included in the analysis. About 53% of these households were in communities randomized to receive AZ ( Table 1 ). Over 97% of participants in both arms lived in self-owned habitats, and over 75% resided in simple detached houses or villas. Housing materials varied: while three-quarters of households had relatively improved roof materials (concrete, tiles, or sheets), less than 40% had improved floor materials (tiles or cement), and fewer than 10% had houses with improved wall materials (hard or semi-hard walls). 5.3% of households in the AZ group and 4.5% in the placebo group had access to flush toilets or latrines. In both arms, less than 1% had access to clean drinking water, and similarly low percentages used electricity or gas for cooking. The most commonly owned assets, with over 70% prevalence in both groups, included mobile phones, bicycles, and solar lamps. The largest differences between the groups were in the ownership of radios and solar plates, with 44.6% and 57.5% in the AZ group compared to 49.3% and 61.5% in the placebo group, respectively. View this table: View inline View popup Download powerpoint Table 1 Characteristics of households from the pre-census (n= 15, 291 households in 170 clusters) PCA In the PCA, the first four components had eigenvalues greater than 1, explaining 46% of the variability, with the first component accounting for the largest portion at 20.5%. The Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy was 0.78, indicating that the data was suitable for PCA.[ 38 ] Ownership of solar plates, motorbikes, radios, and TVs, as well as having an improved floor, had component loadings greater than 0.3, indicating a larger contribution to the asset score. Approximately 99% of the standardized asset score values ranged from -3.5 to 4.2, with a maximum score of 6.2, and a mean of 2.6. Based on the asset score rankings, the wealth index had five levels, with the least wealthy 20% of households in the 1st quintile and the wealthiest 20% in the 5th quintile. Wealth and Mortality The mortality rate across all households was 9.7 per 1,000 person-years, 95% CI: 8.3 to 11.2 ( Table 2 ). Mortality rates varied by wealth, with households in the first quintile experiencing the highest rate (12.3 per 1,000 person-years, 95% CI: 9.4 to 15.2) and those in the fifth quintile having the lowest rate (7.0 per 1,000 person-years, 95% CI: 5.3 to 8.6). There was a 11% decrease in mortality rate for each one-level increase in wealth quintile (IRR = 0.89, 95% CI: 0.83 to 0.95). With measures of inequality, on the relative scale (RII), children in households in the lowest wealth quintile were 1.12 times (95% CI: 1.05 to 1.19) more likely to die than those in the wealthiest households. On the absolute scale, households in the lowest socioeconomic status experienced approximately 1.5 additional child deaths per 1,000 person-years (95% CI: 0.5 to 2.5) compared to the wealthiest households ( Table 3 ). The Gini index was 0.59 (95% CI: 0.56 to 0.62) at the household level and 0.40 (95% CI: 0.37 to 0.43) at the community level, indicating high inequality in child mortality within households, with some households experiencing much higher child mortality rates than others, and moderate inequality across communities ( Table 3 ). The Concentration Index was -0.14 (95% CI: -0.25 to -0.04) at the household level and -0.15 (95% CI: - 0.24 to -0.06) at the community level, reflecting modest but statistically significant inequality in child mortality by wealth, disadvantaging the poor ( Figure 3 ). View this table: View inline View popup Download powerpoint Table 2 Mortality Rates, Rate Ratios, and Rate Differences by Asset-Based Wealth Quintiles View this table: View inline View popup Download powerpoint Table 3 Estimates of Inequality Indices: Relative Index, Slope Index, and Concentration Index at household and cluster level Mortality rates decreased with increasing wealth quintiles at both the household and cluster levels, with the greater reductions occurring from the first to the second and the fourth to the fifth quintile ( Figure 2 ). Although the trend was similar, the change in mortality rate by wealth was slightly more pronounced at cluster level, with a 15% reduction for each increase in quintile level (IRR = 0.85, 95% CI: 0.77 to 0.94). The wealth related inequalities in mortality were also slightly larger at the community level with RII of 1.17, 95%CI (1.05 to 1.29) and SII: 2.3 per 1000-person year, 95%CI (0.2 to 4.4). These disparities, though small, were statistically significant. The observed changes in mortality by wealth quintiles remained consistent when adjusting for distance to the facility ( Table 2 ). Download figure Open in new tab Figure 2 Mortality rate by a) Household wealth index quintile and b) wealth at cluster level Download figure Open in new tab Figure 3 Concentration Index of Child Mortality by Wealth Status at Household and Community Levels Wealth-AZ Interaction Mortality rates decreased with increasing wealth quintiles for both AZ and placebo clusters (IRR for AZ = 0.87, 95% CI: 0.80 to 0.95; IRR for placebo = 0.90, 95% CI: 0.82 to 0.99). Across all quintiles, the mortality rate was lower in the AZ group compared to the placebo group (IRR = 0.83, 95% CI: 0.63 to 1.1). The effect of AZ compared to placebo did not change significantly with the wealth index, and the change in mortality rate by wealth did not vary significantly between the arms ( Table 4 , Figure 4 ). View this table: View inline View popup Download powerpoint Table 4 Interaction between wealth and treatment on under-5 mortality at household and cluster level Download figure Open in new tab Figure 4 Mortality rate by treatment and wealth index quintile at a) household level and b) cluster level Although there was a slight difference in the pattern of change at the household versus cluster level ( Figure 4 ), there was no significant interaction between AZ and wealth on either the multiplicative or additive scale at both the household and cluster levels ( Table 4 ). Differences in RII and SII by arm were also minimal ( Table 5 ). View this table: View inline View popup Download powerpoint Table 5 Estimates of Inequality Indices by treatment arm at household and cluster level Sensitivity Analyses In the sensitivity analysis using the first 4 components, the results were similar (Supplementary Figure 1 and 2). We found that that the probability of households being included in the pre-census (selected) did not vary significantly by study arm (RR= 1.14, 95%CI (0.9 to 1.44). Furthermore, the effect of AZ on mortality did not vary substantially between households included in pre-census (selected) and missing households (P value for AZ*selection interaction=0.862, Supplementary Table 1). Discussion We found a wealth gradient in child mortality at both household and community levels, with more pronounced disparities at the community level. At both levels, those in the least wealthy quintile had the highest mortality rates. These wealth-related disparities were similar in both AZ and placebo groups. Although AZ-treated groups consistently had lower mortality rates, the effect of AZ compared to placebo did not vary significantly across wealth quintiles. There was no significant interaction between AZ and wealth status on either a multiplicative or additive scale at the household or cluster levels. Wealth and Mortality We found that mortality rates declined substantially with increasing wealth, consistent with previous studies that identified wealth as a key factor in under-5 mortality variability.[ 4 , 39 ] Wealth may influence child mortality in several ways. At the community level, factors such as ecological settings (e.g., local environment), political economy (e.g., economic conditions and governance), and health systems (e.g., access to healthcare services) can all influence child mortality.[ 20 ] At the household level, factors linked to a family’s income and wealth can influence access to goods and services—such as housing, food, transportation, and healthcare.[ 20 , 40 ] These factors can individually and collectively affect mortality through more proximal factors like poorer living conditions, higher susceptibility to infections, and reduced access to quality care,[ 20 , 41 , 42 ] likely contributing to the higher child mortality observed among the poorest households and communities. Additional barriers include limited access to routine services, such as timely childhood vaccination, which are often lower among the most disadvantaged populations in Sub-Saharan Africa.[ 43 ] Distance to facilities and the inability to afford care—factors commonly associated with poverty—have been noted as reasons for such disparities and as general barriers to care.[ 13 , 44 , 45 ] However, in our study, adjusting for distance to facilities did not change the disparities in mortality observed by wealth. Over the years, Burkina Faso has made notable progress in reducing childhood mortality by increasing access to free care, adopting seasonal malaria chemoprevention, expanding services, and improving the quality of care.[ 46 ] While free healthcare for children under 5 has improved healthcare utilization and helped reduce inequalities,[ 47 ] some disparities remain, with the least wealthy households and communities experiencing the highest mortality rates. The modest change in mortality rates for those in the 40 th to 80 th percentiles suggests that the wealth impacts on under 5 mortality may be most pronounced in the wealthiest and poorest 20%. A closer examination of how wealth translates to health outcomes at the extremes of the wealth spectrum could be helpful. Although the Gini index showed higher variability in child mortality across households than communities, disparities by wealth appeared to be higher at the community level, as seen with the relative and slope inequality indices, as well as the concentration index. The more pronounced community-level disparities (than the household level) may suggest that aggregate factors—such as healthcare accessibility, environmental conditions, and community resources—may play a larger role, especially in the context of residential segregation by wealth. Targeted approaches addressing the specific challenges of vulnerable populations,[ 48 ] including enhancing community resilience,[ 49 ] may further reduce differential mortality. We did not find evidence of an interaction between AZ and wealth on mortality rates. The wealth gradient in child mortality was similar in both AZ and placebo-treated clusters. Previous research indicated that AZ could help reduce health disparities by being particularly beneficial for distant communities with reduced access to healthcare and antibiotics.[ 11 ] Although the mechanism of action is not clear, AZ may improve short-term health outcomes by clearing infections,[ 50 ] which is critical given that many infectious diseases can worsen rapidly without treatment. This can make AZ particularly effective at reducing mortality and morbidity in populations that are geographically disadvantaged, where treatment access is limited. While distance to facilities is a significant barrier to accessing routine and life-saving interventions,[ 12 , 13 ] it often represents one of many obstacles to care. Thus, while mass AZ distribution may also help address some of the wealth-related disparities due to barriers to treatment and care, its effectiveness in addressing broader disparities influenced by factors such as economic inequality, nutrition, and living conditions may be more limited. Regardless of its impact on wealth-related disparities, AZ remains effective in reducing child mortality,[ 8 , 9 , 10 ] as reflected with the lower mortality rates in AZ treated communities. Therefore, using AZ with comprehensive approaches that strengthen health systems and services in poor communities, improve living conditions and nutritional status of disadvantaged children, increase awareness and coverage of interventions, and enhance the quality and equity of care may better help address wealth-related disparities.[ 48 ] The consistent effect of AZ across different wealth levels may imply that it provides similar benefits across economically diverse communities, with no evidence suggesting enhanced benefits for disadvantaged communities or for prioritizing treatment based on wealth status. Previous studies exploring the heterogeneity of AZ’s effect by underlying mortality rate[ 51 ] or by nutritional status[ 52 ] did not find strong evidence of effect modification. While concerns about resistance call for targeted approaches,[ 53 ] AZ appears beneficial for a broad range of populations. Findings from the AVENIR trial also showed that nontargeted intervention for all children aged 1 to 59 months yielded a greater benefit against mortality compared to restricting treatment to infants 1-11 months.[ 54 ] While it is important to explore whether certain populations benefit more from MDA with AZ and prioritize them accordingly, continuing to treat all children who can benefit from it remains crucial in the absence of evidence for a targeted approach or better alternatives.[ 55 ] A study weighing the risks and benefits of MDA AZ found that the short term benefits outweigh the risks, as AZ is an effective short-term strategy for reducing mortality that can be used along with larger-scale structural and systemic changes.[ 56 ] This study has some limitations. The first is reduced statistical power due to relatively low mortality rates. The need to restrict the analysis to households included in the pre-census—i.e., the exclusion of missing households—further lowered our sample size. Analyzing data at the cluster level partly mitigated this issue by averaging household wealth scores and aggregating mortality by cluster, allowing us to incorporate mortality information from households missing wealth scores. This, however, relied on the assumption that the households with non-missing scores can adequately represent those with missing scores. Additionally, our sensitivity analyses suggest that the missingness of households in the pre-census appears to be a random process and does not differ systematically by study arm. The scenario shown in Figure 1D indicates that selection can be a threat to generalizability of findings to target population if S (selection) for inclusion in the analysis is associated with changes in mortality by wealth (△Y W ).[ 37 ] Our sensitivity analysis shows that treatment (A) is not associated with selection (S), thereby breaking the link between these variables. Additionally, the effect of AZ on mortality in the selected and missing households was also similar. This suggests that the threat to validity shown in Figure 1C may not apply, since there is no association between selection into the sample (S) (selection) and the effect of AZ on mortality (△Y A ) in our data. This implies that the observed effects in the selected population can be generalized to the full sample and target population.[ 37 ] Second, there is a potential for misclassification or measurement error, as the asset-based wealth index may not fully capture the complexity of wealth within households. Although this limitation could affect the accuracy of our analyses, it would likely impact all communities similarly and would not be differential by mortality levels. The extent to which these scores adequately represent wealth can also be affected by the PCA. Although the proportion of variability explained by the first component was not large, the combination of the first 4 components, which explained a much larger proportion of variability, generated similar findings. Additionally, a summary of household characteristics by wealth index quintile shows that asset ownership increases with wealth, as expected. This trend further strengthens our confidence in using the scores generated from PCA to represent wealth status (Supplementary table 1). Lastly, since we did not have additional data on clusters and households, we could not adjust for other potential confounders of the wealth-mortality association, such as bed net access. Therefore, the objective was to examine how mortality rates vary with wealth in these communities and examine how it interacts with AZ, rather than to estimate the causal effect of asset-based wealth on child mortality. Conclusion Our results suggest that while wealth-related disparities in child mortality were observed at both the household and community levels, mass AZ treatment did not appear to mitigate these disparities, highlighting the need for other targeted approaches that could address this differential mortality. We did not find evidence of an enhanced benefit of AZ for disadvantaged communities or for prioritizing treatment based on wealth status. Therefore, it may be beneficial for MDA AZ programs to continue providing treatment to all children who could benefit. Further work is needed to address the underlying factors contributing to disparities in child mortality in these communities. Ethics Approval The randomized controlled trial from which the data was obtained was reviewed and approved by the Institutional Review Boards at the University of California, San Francisco and the Comité National d’Ethique pour la Recherche (National Ethics Committee of Burkina Faso) in Ouagadougou, Burkina Faso. Data Availability The datasets analyzed during the current study are not yet publicly available but will be made accessible through the Open Science Framework (OSF) repository upon finalization. The data will be shared in accordance with relevant open data sharing guidelines. A link to the repository will be provided once the data is available. Funding The CHAT trial was supported by the Gates Foundation (grant number OPP1187628). The conclusions and opinions expressed in this work are those of the authors alone and should not be attributed to the Foundation. Under the grant conditions of the Foundation, a Creative Commons Attribution 4.0 License has already been assigned to the Author Accepted Manuscript version that may arise from this submission. Please note that works submitted as preprints have not undergone a peer review process. Research reported in this manuscript was also supported by the National Institutes of Health Eunice Kennedy Shriver National Institute of Child Health & Human Development (NIH/NICHD) F31 Award (1F31HD114434-01A1: E.A.G.). Role of the Funder/Sponsor The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Conflict of Interest None declared Author Contributions Elisabeth A. Gebreegziabher: Conceptualization, Methodology, Software, Formal analysis, Investigation, Visualization, Writing – original draft, Writing – review & editing. Mamadou Ouattara, Mamadou Bountogo, Boubacar Coulibaly, Thierry Ouedraogo, Elodie Lebas: Project administration, Supervision, Resources, Writing – review & editing. Valentin Boudo, Huiyu Hu: Data curation, Writing – review & editing. David V. Glidden, Steven E. Gregorich, Benjamin F. Arnold: Conceptualization, Statistical analysis, Writing – review & editing. Pearl Anne Ante-Testard, Kieran S. O’Brien: Writing – review & editing. Michelle S. Hsiang: Conceptualization, Methodology, Writing – review & editing. Ali Sié, Thomas M. Lietman, Catherine E. Oldenburg: Conceptualization, Funding acquisition, Project administration, Supervision, Writing – review & editing. Supplementary Data Supplementary Table 1- Sensitivity Analyses Assessing the Role of Selection Supplementary Table 2 -Summary of household characteristic by wealth index quintiles Supplementary Figure 1- Mortality rate by a) Household wealth index quintile and b) wealth at cluster level with wealth score generated from 4 principal components Supplementary Figure 2 - Mortality rate by treatment and wealth index quintile at a) household level and b) cluster level with wealth score generated from 4 principal components References 1. ↵ How Burkina Faso cut its under-five mortality by 74%. Exemplars In Global Health . December 2022 . 2. ↵ Under-five mortality . UNICEF Datauniceforg/topic/child-survival/under-five-mortality/ . March 2024 . 3. ↵ Golding N , Burstein R , Longbottom J , Browne AJ , Fullman N , Osgood-Zimmerman A , et al. Mapping under-5 and neonatal mortality in Africa, 2000-15: a baseline analysis for the Sustainable Development Goals . Lancet . 2017 ; 390 ( 10108 ): 2171 – 82 . OpenUrl CrossRef PubMed 4. ↵ Van Malderen C , Amouzou A , Barros AJD , Masquelier B , Van Oyen H , Speybroeck N . Socioeconomic factors contributing to under-five mortality in sub-Saharan Africa: a decomposition analysis . BMC Public Health . 2019 ; 19 ( 1 ): 760 . OpenUrl PubMed 5. ↵ Sharrow D , Hug L , You D , Alkema L , Black R , Cousens S , et al. Global, regional, and national trends in under-5 mortality between 1990 and 2019 with scenario-based projections until 2030: a systematic analysis by the UN Inter-agency Group for Child Mortality Estimation . Lancet Glob Health . 2022 ; 10 ( 2 ): e195 – e206 . OpenUrl 6. ↵ Liu L , Oza S , Hogan D , Chu Y , Perin J , Zhu J , et al. Global, regional, and national causes of under-5 mortality in 2000-15: an updated systematic analysis with implications for the Sustainable Development Goals . Lancet . 2016 ; 388 ( 10063 ): 3027 – 35 . OpenUrl CrossRef PubMed 7. ↵ Amegah AK . Improving Child Survival in Sub-Saharan Africa: Key Environmental and Nutritional Interventions . Ann Glob Health . 2020 ; 86 ( 1 ): 73 . OpenUrl PubMed 8. ↵ Keenan JD , Bailey RL , West SK , Arzika AM , Hart J , Weaver J , et al. Azithromycin to Reduce Childhood Mortality in Sub-Saharan Africa . N Engl J Med . 2018 ; 378 ( 17 ): 1583 – 92 . OpenUrl CrossRef PubMed 9. ↵ Oldenburg CE , Ouattara M , Bountogo M , Boudo V , Ouedraogo T , Compaoré G , et al. Mass Azithromycin Distribution to Prevent Child Mortality in Burkina Faso: The CHAT Randomized Clinical Trial . JAMA . 2024 ; 331 ( 6 ): 482 – 90 . OpenUrl CrossRef PubMed 10. ↵ O’Brien Kieran S , Arzika Ahmed M , Amza A , Maliki R , Aichatou B , Bello Ismael M , et al. Azithromycin to Reduce Mortality — An Adaptive Cluster-Randomized Trial . New England Journal of Medicine . 2024 ; 391 ( 8 ): 699 – 709 . OpenUrl PubMed 11. ↵ Chao DL , Arzika AM , Abdou A , Maliki R , Karamba A , Galo N , et al. Distance to Health Centers and Effectiveness of Azithromycin Mass Administration for Children in Niger: A Secondary Analysis of the MORDOR Cluster Randomized Trial . JAMA Network Open . 2023 ; 6 ( 12 ): e2346840 -e. OpenUrl 12. ↵ Karra M , Fink G , Canning D . Facility distance and child mortality: a multi-country study of health facility access, service utilization, and child health outcomes . Int J Epidemiol . 2017 ; 46 ( 3 ): 817 – 26 . OpenUrl CrossRef PubMed 13. ↵ Oldenburg CE , Sié A , Ouattara M , Bountogo M , Boudo V , Kouanda I , et al. Distance to primary care facilities and healthcare utilization for preschool children in rural northwestern Burkina Faso: results from a surveillance cohort . BMC Health Serv Res . 2021 ; 21 ( 1 ): 212 . OpenUrl CrossRef PubMed 14. ↵ Ogunkola IO , Adebisi YA , Imo UF , Odey GO , Esu E , Lucero-Prisno DE , 3rd . . Rural communities in Africa should not be forgotten in responses to COVID-19 . Int J Health Plann Manage . 2020 ; 35 ( 6 ): 1302 – 5 . OpenUrl CrossRef PubMed 15. ↵ Peters DH , Garg A , Bloom G , Walker DG , Brieger WR , Rahman MH . Poverty and access to health care in developing countries . Ann N Y Acad Sci . 2008 ; 1136 : 161 – 71 . OpenUrl CrossRef PubMed Web of Science 16. ↵ Bado AR , Appunni SS . Decomposing Wealth-Based Inequalities in Under-Five Mortality in West Africa . Iran J Public Health . 2015 ; 44 ( 7 ): 920 – 30 . OpenUrl PubMed 17. ↵ Sahasranaman A , Jensen HJ . Poverty in the time of epidemic: A modelling perspective . PLoS One . 2020 ; 15 ( 11 ): e0242042 . OpenUrl PubMed 18. ↵ Fallah MP , Skrip LA , Gertler S , Yamin D , Galvani AP . Quantifying Poverty as a Driver of Ebola Transmission . PLoS Negl Trop Dis . 2015 ; 9 ( 12 ): e0004260 . OpenUrl CrossRef PubMed 19. ↵ Liu RS , Burgner DP , Sabin MA , Magnussen CG , Cheung M , Hutri-Kähönen N , et al. Childhood Infections, Socioeconomic Status, and Adult Cardiometabolic Risk . Pediatrics . 2016 ; 137 ( 6 ). 20. ↵ Mosley WH , Chen LC . An analytical framework for the study of child survival in developing countries. 1984 . Bull World Health Organ . 2003 ; 81 ( 2 ): 140 – 5 . OpenUrl PubMed Web of Science 21. ↵ Sié A , Ouattara M , Bountogo M , Bagagnan C , Coulibaly B , Boudo V , et al. A double-masked placebo-controlled trial of azithromycin to prevent child mortality in Burkina Faso, West Africa: Community Health with Azithromycin Trial (CHAT) study protocol . Trials . 2019 ; 20 ( 1 ): 675 . OpenUrl CrossRef PubMed 22. ↵ Becher H , Müller O , Dambach P , Gabrysch S , Niamba L , Sankoh O , et al. Decreasing child mortality, spatial clustering and decreasing disparity in North-Western Burkina Faso . Trop Med Int Health . 2016 ; 21 ( 4 ): 546 – 55 . OpenUrl CrossRef PubMed 23. ↵ Liu L , Johnson HL , Cousens S , Perin J , Scott S , Lawn JE , et al. Global, regional, and national causes of child mortality: an updated systematic analysis for 2010 with time trends since 2000 . The Lancet . 2012 ; 379 ( 9832 ): 2151 – 61 . OpenUrl CrossRef 24. ↵ The World Bank in Burkina Faso . The World Bank Sep 2023 . 25. ↵ General information on Burkina Faso . SOS CHILDREN’S VILLAGES . 26. ↵ Lisa Hjelm AM , Darryl Miller , Amit Wadhwa . Creation of a Wealth Index . World Food Programme . 2017 . 27. ↵ Greenacre M , Groenen PJF , Hastie T , D’Enza AI , Markos A , Tuzhilina E . Principal component analysis . Nature Reviews Methods Primers . 2022 ; 2 ( 1 ): 100 . OpenUrl 28. ↵ Analyzing DHS Data . Guide to DHS Statistics DHS-8 dhsprogramcom/data/Guide-to-DHS-Statistics/indexhtm#t=Analyzing_DHS_Datahtm 29. ↵ Bellows N , Weinberger M , Reidy M . Using the Demographic Health Survey wealth index to create family planning market segments based on absolute income levels . BMJ Glob Health . 2020 ; 5 ( 9 ). 30. ↵ Schoeps A , Gabrysch S , Niamba L , Sié A , Becher H . The effect of distance to health-care facilities on childhood mortality in rural Burkina Faso . Am J Epidemiol . 2011 ; 173 ( 5 ): 492 – 8 . OpenUrl CrossRef PubMed Web of Science 31. ↵ Dotse-Gborgbortsi W , Nilsen K , Ofosu A , Matthews Z , Tejedor-Garavito N , Wright J , et al. Distance is “a big problem”: a geographic analysis of reported and modelled proximity to maternal health services in Ghana . BMC Pregnancy Childbirth . 2022 ; 22 ( 1 ): 672 . OpenUrl PubMed 32. ↵ Moreno-Betancur M , Latouche A , Menvielle G , Kunst AE , Rey G . Relative index of inequality and slope index of inequality: a structured regression framework for estimation . Epidemiology . 2015 ; 26 ( 4 ): 518 – 27 . OpenUrl CrossRef PubMed 33. ↵ Truman B , Smith K , Roy K , Chen Z , Moonesinghe R , Zhu J , Crawford C , Zaza S. Rationale for Regular Reporting on Health Disparities and Inequalities --- United States. Morbidity and Mortality Weekly Report (MMWR) . January 14, 2011 . 34. ↵ Steinbeis F , Gotham D , von Philipsborn P , Stratil JM . Quantifying changes in global health inequality: the Gini and Slope Inequality Indices applied to the Global Burden of Disease data, 1990-2017 . BMJ Glob Health . 2019 ; 4 ( 5 ): e001500 . OpenUrl Abstract / FREE Full Text 35. ↵ O’Donnell O , O’Neill S , Van Ourti T , Walsh B. conindex: Estimation of concentration indices . Stata J . 2016 ; 16 ( 1 ): 112 – 38 . OpenUrl CrossRef PubMed 36. ↵ Analyzing Health Equity: The Concentration Index . Worldbank Retrieved February 5 , 2025 37. ↵ Nilsson A , Bonander C , Strömberg U , Björk J . A directed acyclic graph for interactions . International Journal of Epidemiology . 2021 ; 50 ( 2 ): 613 – 9 . OpenUrl CrossRef PubMed 38. ↵ Carillo MF , Largo FF , Ceballos RF . Principal component analysis on the Philippine health data . arXiv preprint arXiv:190207905. 2019 . 39. ↵ Tekeba B , Tamir TT , Workneh BS , Zegeye AF , Gonete AT , Alemu TG , et al. Early neonatal mortality and determinants in Ethiopia: multilevel analysis of Ethiopian demographic and health survey, 2019 . BMC Pediatr . 2024 ; 24 ( 1 ):p. 40. ↵ Lartey ST , Khanam R , Takahashi S . The impact of household wealth on child survival in Ghana. Journal of Health , Population and Nutrition . 2016 ; 35 ( 1 ): 38 . OpenUrl 41. ↵ Reducing Child Mortality – The Challenges in Africa . The United Nations Chronicle wwwunorg/en/chronicle/article/reducing-child-mortality-challenges-africa . 42. ↵ Asif MF , Pervaiz Z , Afridi JR , Safdar R , Abid G , Lassi ZS . Socio-economic determinants of child mortality in Pakistan and the moderating role of household’s wealth index . BMC Pediatr . 2022 ; 22 ( 1 ): 3 . OpenUrl PubMed 43. ↵ Mutua MK , Mohamed SF , Porth JM , Faye CM . Inequities in On-Time Childhood Vaccination: Evidence From Sub-Saharan Africa . American Journal of Preventive Medicine . 2021 ; 60 ( 1, Supplement 1 ): S11 – S23 . OpenUrl CrossRef PubMed 44. ↵ McLaren ZM , Ardington C , Leibbrandt M . Distance decay and persistent health care disparities in South Africa . BMC Health Services Research . 2014 ; 14 ( 1 ): 541 . OpenUrl PubMed 45. ↵ Chuma J , Okungu V , Molyneux C . Barriers to prompt and effective malaria treatment among the poorest population in Kenya . Malaria Journal . 2010 ; 9 ( 1 ): 144 . OpenUrl PubMed 46. ↵ How Burkina Faso cut its under-five mortality by 74% . Exemplars In Global Health wwwexemplarshealth/stories/how-burkina-faso-cut-its-under-five-mortality December 8 , 2022 . 47. ↵ Druetz T , Fregonese F , Bado A , Millogo T , Kouanda S , Diabaté S , et al. Abolishing Fees at Health Centers in the Context of Community Case Management of Malaria: What Effects on Treatment-Seeking Practices for Febrile Children in Rural Burkina Faso? PLOS ONE . 2015 ; 10 ( 10 ): e0141306 . OpenUrl PubMed 48. ↵ Lassi ZS , Mallick D , Das JK , Mal L , Salam RA , Bhutta ZA . Essential interventions for child health . Reprod Health . 2014 ; 11 Suppl 1( Suppl 1 ): S4 . OpenUrl 49. ↵ Popay J , Kaloudis H , Heaton L , Barr B , Halliday E , Holt V , et al. System resilience and neighbourhood action on social determinants of health inequalities: an English Case Study . Perspect Public Health . 2022 ; 142 ( 4 ): 213 – 23 . OpenUrl CrossRef PubMed 50. ↵ Keenan JD , Arzika AM , Maliki R , Elh Adamou S , Ibrahim F , Kiemago M , et al. Cause-specific mortality of children younger than 5 years in communities receiving biannual mass azithromycin treatment in Niger: verbal autopsy results from a cluster-randomised controlled trial . Lancet Glob Health . 2020 ; 8 ( 2 ): e288 – e95 . OpenUrl PubMed 51. ↵ Porco TC , Oldenburg CE , Arzika AM , Kalua K , Mrango Z , Cook C , et al. Efficacy of Mass Azithromycin Distribution for Reducing Childhood Mortality Across Geographic Regions . Am J Trop Med Hyg . 2020 ; 103 ( 3 ): 1291 – 4 . OpenUrl PubMed 52. ↵ O’Brien KS , Arzika AM , Maliki R , Manzo F , Mamkara AK , Lebas E , et al. Biannual azithromycin distribution and child mortality among malnourished children: A subgroup analysis of the MORDOR cluster-randomized trial in Niger . PLoS Med . 2020 ; 17 ( 9 ): e1003285 . OpenUrl PubMed 53. ↵ Mabey D , Okomo U , Greenwood B . Priorities in reducing child mortality: Azithromycin and other interventions . PLOS Medicine . 2020 ; 17 ( 9 ): e1003364 . OpenUrl 54. ↵ O’Brien KS , Arzika AM , Amza A , Maliki R , Aichatou B , Bello IM , et al. Azithromycin to Reduce Mortality — An Adaptive Cluster-Randomized Trial . New England Journal of Medicine . 2024 ; 391 ( 8 ): 699 – 709 . OpenUrl PubMed 55. ↵ Tam CC , Offeddu V , Lim JM , Voo TC . One drug to treat them all: ethical implications of the MORDOR trial of mass antibiotic administration to reduce child mortality . J Glob Health . 2019 ; 9 ( 1 ): 010305 . OpenUrl PubMed 56. ↵ Kahn R , Eyal N , Sow SO , Lipsitch M . Mass drug administration of azithromycin: an analysis . Clinical Microbiology and Infection . 2023 ; 29 ( 3 ): 326 – 31 . OpenUrl PubMed View the discussion thread. Back to top Previous Next Posted July 06, 2025. Download PDF Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Exploring Heterogeneity in Treatment Effects: The Impact and Interaction of Asset-Based Wealth and Mass Azithromycin Distribution on Child Mortality Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Exploring Heterogeneity in Treatment Effects: The Impact and Interaction of Asset-Based Wealth and Mass Azithromycin Distribution on Child Mortality Elisabeth A. Gebreegziabher , Ali Sié , Mamadou Ouattara , Mamadou Bountogo , Boubacar Coulibaly , Valentin Boudo , Thierry Ouedraogo , Elodie Lebas , Huiyu Hu , Pearl Anne Ante-Testard , Steven E Gregorich , Kieran S. O’Brien , Michelle S. Hsiang , David V. Glidden , Benjamin F. Arnold , Thomas M. Lietman , Catherine E. Oldenburg medRxiv 2025.07.05.25329685; doi: https://doi.org/10.1101/2025.07.05.25329685 Share This Article: Copy Citation Tools Exploring Heterogeneity in Treatment Effects: The Impact and Interaction of Asset-Based Wealth and Mass Azithromycin Distribution on Child Mortality Elisabeth A. Gebreegziabher , Ali Sié , Mamadou Ouattara , Mamadou Bountogo , Boubacar Coulibaly , Valentin Boudo , Thierry Ouedraogo , Elodie Lebas , Huiyu Hu , Pearl Anne Ante-Testard , Steven E Gregorich , Kieran S. O’Brien , Michelle S. Hsiang , David V. Glidden , Benjamin F. Arnold , Thomas M. Lietman , Catherine E. Oldenburg medRxiv 2025.07.05.25329685; doi: https://doi.org/10.1101/2025.07.05.25329685 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Epidemiology Subject Areas All Articles Addiction Medicine (569) Allergy and Immunology (863) Anesthesia (300) Cardiovascular Medicine (4442) Dentistry and Oral Medicine (444) Dermatology (383) Emergency Medicine (609) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1511) Epidemiology (15230) Forensic Medicine (30) Gastroenterology (1126) Genetic and Genomic Medicine (6610) Geriatric Medicine (668) Health Economics (998) Health Informatics (4542) Health Policy (1370) Health Systems and Quality Improvement (1613) Hematology (543) HIV/AIDS (1266) Infectious Diseases (except HIV/AIDS) (15923) Intensive Care and Critical Care Medicine (1103) Medical Education (623) Medical Ethics (147) Nephrology (668) Neurology (6607) Nursing (346) Nutrition (999) Obstetrics and Gynecology (1146) Occupational and Environmental Health (957) Oncology (3338) Ophthalmology (974) Orthopedics (369) Otolaryngology (420) Pain Medicine (436) Palliative Medicine (130) Pathology (665) Pediatrics (1693) Pharmacology and Therapeutics (692) Primary Care Research (712) Psychiatry and Clinical Psychology (5448) Public and Global Health (9239) Radiology and Imaging (2202) Rehabilitation Medicine and Physical Therapy (1370) Respiratory Medicine (1196) Rheumatology (596) Sexual and Reproductive Health (714) Sports Medicine (530) Surgery (712) Toxicology (99) Transplantation (289) Urology (265) (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a01f44426916aa64',t:'MTc3OTgyNDU4NA=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();