The global public health community has undertaken numerous efforts to compare the seriousness of different health problems and the effectiveness of different health interventions across countries. Most existing health prioritization exercises have been expert assessments of the seriousness of different health problems [1] or the attractiveness of different health interventions [2]. In contrast, there have been few cross-country comparisons of how people themselves, especially in low- and middle-income countries (LMIC), rank the seriousness of the health problems they face. An exception is the Kaiser-Pew Global Health Survey, but this survey was conducted before the COVID-19 pandemic and did not explicitly include respiratory illness [3]. The available evidence suggests that people in LMIC may have very different priorities for health policy interventions than experts think they should have [4].
Health policymakers are increasingly recognizing the value of incorporating people’s preferences about healthcare priorities and new frameworks have been proposed for incorporating the public’s perception of disease risks in policymaking [5]. Understanding public preferences for disease risk is important because 1) health resource allocation decisions that are aligned with public preferences are more likely to receive support and be easier to implement; 2) health interventions that are guided by public preferences increase people’s quality of life and provide “peace of mind”; and 3) people may have better information about the risks they face in their specific local context and personal realities than experts working with aggregate data [6].
We examine how people in seven large LMIC (Colombia, India, Kenya, Nigeria, South Africa, Tanzania, and Vietnam) prioritized the seriousness of seven health problems (alcohol and drugs, HIV/AIDS, malaria, TB, other respiratory illness, water-borne diseases, and COVID-19) in January-February 2022. The choice of these seven diseases, rather than cancer, heart disease, and other chronic conditions, is because these other diseases are most common among elderly, and less likely to prompt immediate new public health interventions. Our main objective was to determine whether respondents’ rankings would be dominated by a newly emergent infectious disease (COVID-19) that was affecting populations in the countries surveyed and that was receiving extensive media attention.
During the COVID-19 pandemic, large surveys using in-person interviews have been practically impossible to implement because respondents understandably would not allow enumerators into their homes. Most survey data that has been collected has been through internet panels. The results reported in this paper come from web-surveys that YouGov implemented for us in Colombia, India, Kenya, Nigeria, South Africa, Tanzania, and Vietnam. These countries represent about 25% of the world’s population and about 29% of the total population of LMIC.
In each country, YouGov randomly selected 1200 members from their existing internet panels. Each sample was chosen to be representative of that country’s internet-connected population (including access through computers, mobile phones, and tablets). Respondents were interviewed in an online survey that included incentives for participation. The proportion of the population connected to the internet varies across countries: 70% in Colombia, 43% in India, 30% in Kenya, 36% in Nigeria, 70% in South Africa, 22% in Tanzania, and 70% in Vietnam [7]. Our results are therefore not representative of a country’s overall population. The internet-connected population tends to be younger, more educated, higher incomed, more urban and more male. Sample design and survey execution details are in the Online Materials.
The questionnaire implemented via the internet survey included a set of best-worst scaling stated preference questions [8]. From a list of seven health problems, respondents were asked to select the one that they thought was the most serious and one that was least serious. (see Online Materials.) These two health problems were then removed from the list and the choice task was repeated using the five remaining problems. The respondent was then asked a third time to indicate the most and least serious problems from the remaining three problems. The respondent’s answers to these three best-worst questions provides their complete ranking of the seven problems.
We used a rank-ordered logit model to analyze the data generated by this best-worst scaling method. The model results provide one of the most commonly used ways of averaging respondents’ rankings. For example, if malaria is ranked the second most serious by almost all respondents, who in turn are almost equally split in terms of what other disease they rank most serious, then malaria will be predicted to be the most serious disease. (See Online Materials).
Figure 1a-g displays the predicted severity ranking of the seven health problems in each of the seven countries. The most striking result is that other respiratory diseases were the top-ranked health problem in every country except Vietnam. In Vietnam, respiratory diseases were ranked the second most serious health problem, after COVID-19. Our expectation was that COVID-19 would be respondents’ top concern due to availability bias, i.e., that information about COVID-19 was widely available in the media and a topic of discussion in households almost everywhere at the time the survey was conducted [9]. However, this was not the case. Instead, COVID-19 occupied an effective three-way tie for the second-highest ranking, along with alcohol/drug problems and HIV/AIDS.
Respondents placed lower priority on tuberculosis, malaria, and water-borne diseases. The lower priority placed on water-borne disease may be because our sample includes more urban and higher income and education respondents. Thus water-borne illness may not be as important a problem in this internet-connected population, relative to poorer, rural households.
We compare respondents’ perceptions of the seriousness of TB and malaria to actual prevalence rates across countries. Panel (a) of Figure 2 displays the relative rankings of reported per capita prevalence and the severity of malaria across the seven countries based on coefficients from the best-worst model. Panel (b) presents the same information for TB. These data show that respondents’ perceptions of the seriousness of malaria and tuberculosis are strongly correlated with country-level disease prevalence.
To examine the relationship between respondents’ perceptions of the seriousness of health problems/diseases and their socioeconomic and demographic characteristics, we also re-estimated the rank-ordered logit model adding respondent covariates [10]. Figure 2c presents relative risk parameters (1 = no difference) for four binary variables (gender, living in a large city, having been infected with COVID-19, and having been vaccinated); Figure 2d presents relative risk parameters for four continuous and categorical variables (age; difficulty of making ends meet; a scale measuring respondent’s knowledge of COVID-19; and respondent’s perceived probability of being infected with COVID in the future).
The interpretation of the relative ratio (the exponentiated coefficients from the rank order logit model) is different for the binary and continuous variables. For the binary variables, the coefficient represents the shift from the “0” level (e.g., male) to the “1” level (e.g., female). For the continuous variables an alternative notion of change is needed, and we standardize by each such variable’s standard deviation (see Online Materials).
The most striking finding in Panels (c) and (d) of Figure 2 is that these relative risk ratios are generally small, suggesting that the covariates have similar associations with respect to the seriousness of the different health problems, once overall country-specific differences are taken into account. The statistical significance is high for the large relative risk ratios, but not for the smaller ones (See Online Materials). Looking at the specific binary indicators, respondents in large cities rank respiratory diseases higher than other respondents. Women rank respiratory diseases and COVID higher than men. The indicator for having been vaccinated has almost no influence on rankings. Intriguingly, respondents’ rankings of malaria and respiratory diseases are higher if someone in the respondent’s household has had COVID.
A one standard deviation change in age or the difficulty of making ends meet had little influence on health problem rankings. However, this was not true of the two COVID variables. Those who were more knowledgeable about COVID rank both COVID and respiratory diseases more highly. However, the largest effect was for the perceived likelihood of getting infected with COVID in the next 12 months. For this COVID variable, a one standard deviation increase was associated with a sizeable increase in respondents’ ranking of the seriousness of the COVID health problem. This finding would not have been apparent by examining past COVID infection or vaccination status.
The COVID-19 pandemic has had profound direct and indirect effects on health care systems throughout the world. The demand for COVID-19 care has crowded out care for other diseases. At the same time, in many countries the supply of health care services has decreased, as health care personnel have suffered from COVID and a substantial increase in workloads. In addition to these supply and demand pressures on health care systems, there has been an enormous media focus on the COVID-19 epidemic. Thus, it would not surprise us to find respondents preoccupied with COVID-19, to the neglect of other health problems.
Results from this multi-country study of disease and health problem priorities do show respondents in seven LMIC to be quite concerned about COVID-19. In all seven countries respondents ranked COVID-19 as more serious than traditional health concerns like malaria, TB, and water-borne illnesses. However, the emergence of this new infectious disease and all the associated media attention has still not placed COVID-19 at the very top of most respondents’ rankings. That distinction belongs to other respiratory diseases associated with air pollution and smoking, clearly suggesting that the COVID-19 pandemic has not completely crowded out concern about other serious health problems. To the contrary, knowledge of COVID-19 increases concern about other respiratory diseases. Given the close correlation found between perception and prevalence of malaria and TB, healthcare policies should ensure that the attention given to COVID-19 does not crowd out other important health concerns, particularly other respiratory diseases. Future replication of the health problems ranking questions used in these surveys would provide further insights into the priority people place on addressing endemic COVID-19.