Impact of Health Aid Investments on Public Opinion of the United States: Analysis of Global Attitude Surveys From 45 Countries, 2002-2016

Abstract

Objectives. To assess whether increasing health aid investments affected public opinion of the United States in recipient populations. Methods. We linked health aid data from the United States to nationally representative opinion poll surveys from 45 countries conducted between 2002 and 2016. We exploited the abrupt and substantial increase in health aid when the US President's Emergency Plan for AIDS Relief (PEPFAR) and President's Malaria Initiative (PMI) were launched to assess unique changes in opinions of the United States following program onset. We also ascertained increased exposure to health aid from the United States by systematically searching for mentions of US health aid programs in popular press. Results. Favorability ratings of the United States increased within countries in proportion to health aid and were significantly higher after implementation of PEPFAR and PMI. Higher US health aid was associated with more references to that aid in the popular press. Conclusions. Our study was the first, to our knowledge, to show that US investments in health aid improved the United States' image abroad. Public Health Implications. Sustained global health investments may offer important returns to the United States as well as to the recipient populations.

FIGURE 1—

Association Between Health Aid (in Millions) and the Probability of “Very Favorable” Opinion of the United States: 45 Low- and Middle-Income Countries, 2002–2016 Note. Line represents nonparametric kernel-weighted local polynomial regression fit (with Epanechnikov kernels and zero degree polynomials for local mean smoothing). Each marker represents health aid donations to a given country and year (x-axis) and the corresponding proportion of respondents in that country and year who had a very favorable opinion of the United States (y-axis). Labels for select countries are displayed as reference examples; legend displays full list of markers and countries.

FIGURE 2—

Association of Health Aid Disbursements With Higher Probability of “Very Favorable” Opinion of United States at the (a) Individual Level and (b) Country Level: 45 Low- and Middle-Income Countries, 2002–2016 Note. Point estimates and 95% confidence intervals obtained from multivariable regression models that included aid to all 5 sectors (in deciles), logged gross national income per capita, total population size, mobile phone ownership, and country and year fixed effects. Individual-level models were also adjusted for age and gender and were weighted by the probability of respondents being selected for the survey. Individual-level model based on responses from 266 679 individuals. Aggregate-level model based on 258 country–year observations. Standard errors were clustered at country level (n = 45).

FIGURE 3—

Association of Health Aid Disbursements With Stories Published in Popular Press That Mentioned PMI or PEPFAR by Frequency: 45 Low- and Middle-Income Countries, 2002–2016 Note. PEPFAR =  President’s Emergency Plan for AIDS Relief; PMI = President’s Malaria Initiative. Local mean smoothing regression of the frequency and sentiment of news stories as a function of US donations for health in a given country and year. Left panel shows the sum of articles published in a given country and year. Right panel shows the sum of sentiment scores of news stories published in a given country and year. Labels for select country–year observations are displayed as reference examples. Articles were scraped from online archives of 135 newspapers in 45 study countries, collapsed at country–year level, and combined with annual foreign aid disbursement data from the Foreign Aid Explorer. Local polynomial regression was fitted with Epanechnikov kernels and zero degree polynomials for local mean smoothing. Appendix B (available as a supplement to the online version of this article at http://www.ajph.org) shows association by sentiment.