Powering health: how clean energy access reduces cardiovascular mortality in children

Abstract

Background

Despite the efforts of policymakers to provide universal access to clean and affordable energy as outlined in Sustainable Development Goal 7 (SDG7), billions of people worldwide lack access to clean fuels, leading to major public health concerns. This study investigates the impact of lack of access to clean energy on cardiovascular disease (CVD) mortality among children across multiple age groups.

Methods

Using a panel dataset from 66 countries over the period from 2000 to 2021, we apply System Generalized Method of Moments (GMM) method. To explore how the effects vary across the distribution of mortality rates, we utilized quantile regression analysis, capturing heterogeneity and the influence of outliers.

Results

The empirical results reveal that lack of access to clean fuels and technology is a significant and consistent predictor of higher CVD mortality in children across all age groups. Specifically, on average, a 10-percentage point increase in the share of the population without access to clean fuels is associated with an increase in CVD infant deaths of approximately 2 per 100,000 people. The estimates from the quantile regression results reveal a heterogeneous impact where the effect is pronounced at higher quantiles of the mortality distribution.

Conclusion

Lack of access to clean energy fuels or technology and insufficient public health investment are key contributors to child CVD mortality, particularly among those already at higher risk. The findings highlight the need for targeted policy interventions that prioritize clean energy access through accelerated transition to net zero, especially for high-mortality regions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-025-24242-6.

Background

Despite the efforts of policymakers to provide universal access to clean and affordable energy as outlined in Sustainable Development Goal (SDG) seven, an estimated 2.1 billion people worldwide lack access to clean fuels technology, relying on dirty fuels. This leads to detrimental health risk, causing millions of deaths from diseases related to pollution, including cardiovascular disease (CVD) [1]. CVDs include a range of disorders of the heart and blood vessels, such as rheumatic heart disease, coronary heart disease and cerebrovascular disease. Traditionally, the literature identifies behavioral risk factors such as smoking, alcohol use and unhealthy diet as the prime risk factors of CVDs.

With the pressing issue of climate change in recent years, environmental risk factors, including air pollution associated with lack of access to clean energy and their impact on CVDs, have become top priorities to researchers and policy makers [3–6] provides compelling evidence that the use of solid fuel was significantly correlated with a range of health problems such as chest pressure while breathing, coughing and asthma. Economic challenges emerged as the foremost obstacle to the adoption of clean cooking, accompanied by other contributing factors. Similarly, [7] show that the use of dirty fuels such as charcoal and biomass are significant causes of ill-health in Ghana. Existing evidence shows that CVDs have caused significant economic burdens worldwide [8, 9].

Access to clean cooking fuels and technologies has been increasingly recognized as a critical determinant of public health [10–12]. A growing body of research links household air pollution from traditional biomass fuels to a wide range of adverse health outcomes, including respiratory infections, low birth weight, and cardiovascular diseases [2, 13]. However, the existing literature has predominantly focused on adult morbidity and mortality, with limited attention to child cardiovascular mortality, a gap this study addresses. Moreover, most prior studies rely on cross-sectional data or single-country settings, limiting the generalizability and causal inference of their findings [2, 14]. More importantly, the existing literature pays little attention to addressing the heterogeneous impacts of access to clean energy sources on CVD mortality among children. This paper contributes to the literature by providing new evidence on the impact of clean fuel access on child CVD mortality using a global panel dataset and rigorous econometric techniques, including quantile regressions to uncover distributional effects. By doing so, it advances our understanding of the heterogeneous health benefits of clean energy transitions and underscores the urgency of policy interventions targeted at vulnerable populations.

In this paper, we investigate the causal impact of access to clean fuels and technology on the incidence of child CVD mortality using panel data from a global sample of countries and employing advanced statistical methods, including system GMM and quantile regressions. The findings of the study reveal that lack of access to clean fuels increases the death rates of children from CVD significantly. More precisely, 10% increase in the share of the population without access to clean fuels is associated with an increase in CVD-related infant mortality of approximately 2 per 100,000 people.

The findings have important policy implications for health policies and energy transition. In line with sustainable development goals (SDG), particularly SDG 7 (clean and affordable energy access) and SDG 13 (climate action), it is critical that policy makers should ensure universal access and transition to clean energy sources to reduce the risk of deaths from CDCs. Various incentives such as subsidies and tax credits are essential for motivating people to adopt and sustain the use of clean energy technologies to effectively reduce risk and deaths of children due to CVDs.

The paper makes several contributions to literature. First, it provides the first empirical evidence on the impact of lack of access to clean fuels and CVD mortality in children under 15 using global panel data from 66 countries. Second, the paper enhances methodological rigor by applying quantile regression to robustly address endogeneity and outlier effects in estimating the health impacts of household indoor pollution across. Third, the paper offers evidence on the heterogeneous impacts from the quantile regression that reveals higher adverse effects of dirty fuel exposure at higher CVD mortality quantiles, highlighting disproportionate impacts on the most vulnerable populations. The quantile regression results underscore that the impact of lacking access to clean fuels is not uniform; it intensifies at higher quantiles of the CVD mortality distribution. This highlights that children in countries with already high mortality rates are disproportionately affected, emphasizing the need for targeted interventions where the health burden is most severe.

The remainder of the paper is structured as follows. Section 2 describes the econometric methodology and data used in this study to empirically identify the causal relationship between access to clean energy and Child CVD mortality rates. Section 3 presents empirical results and analysis followed by discussions in Sect. 4 while Sect. 5 concludes the paper.

Methods

Study design

This study draws on a global panel dataset of 66 countries to investigate the relationship between household energy access, specifically, the lack of access to clean cooking fuel technologies, and cardiovascular disease (CVD) mortality rates of children under 15. The countries included span regions across the world, offering broad geographical and socio-economic diversity. The sample period spans from 2000 to 2021. The full list of countries in the sample is provided in Table A1 in the Appendix.

Countries were selected based on the availability and reliability of data on both household energy access (e.g., share of population without access to clean fuels and technologies for cooking) and health outcomes (e.g., age-specific CVD mortality rates per 100,000 population). Our use of panel data over such long period allows us to exploit both temporal variation (e.g., changes in fuel access over time within countries) and cross-country variations, which is a key identification strategy to estimate the causal impact of household fuel use patterns on CVD outcomes for children under 15. The multi-country, multi-year design strengthens the external validity of the findings and provides valuable insights for global health and energy policy.

Outcome variables

The outcome variable in this study is CVD mortality rate, expressed as the number of deaths per 100,000 population per year for children under 15. This measure captures the burden of cardiovascular child mortality across countries and over time, enabling cross-national and longitudinal comparisons that are not confounded by differences in population size or age distribution. CVD mortality encompasses a range of fatal conditions related to the heart and blood vessels, such as heart failure, ischemic and hypertensive heart disease and cerebrovascular disease (stroke). The data are sourced from Institute For Health Metrics and Evaluation (IHME), Global Burden of Disease.

Explanatory variable

The key independent variable in this study is the share of the population without access to clean fuels and technologies for cooking in each country-year observation. This indicator reflects the extent to which households lack access to clean fuels or technologies such as clean cookstoves, electricity and natural gas that are key to reduce exposure to indoor air pollutants. Access to clean fuels is a critical determinant of household air pollution exposure. Traditional cooking fuels such as biomass, kerosene, or coal release harmful pollutants including particulate matter and carbon monoxide that are strongly linked to cardiovascular inflammation, atherosclerosis, and increased risk of heart attacks and strokes. The data are sourced from the World Health Organization - Global Health Observatory. The indicator follows the SDG 7, which tracks progress toward universal access to clean energy. This variable serves as a proxy for long-term exposure to indoor air pollution, enabling us to estimate its causal relationship with cardiovascular mortality outcomes.

Control variables

To account for confounding factors and ensure that the estimated relationship between access to clean fuels and cardiovascular mortality is not driven by omitted variables, the model includes a range of key covariates, including domestic government health expenditure as % of GDP, real GDP per capita, unemployment rates, outdoor air pollution (PM2.5) and government effectiveness. These controls are chosen based on theoretical relevance and empirical precedent in the public health and development economics literature [15].

Health Expenditure (% of GDP): This variable captures the proportion of national income devoted to healthcare services and infrastructure. Higher public and private investment in health can directly influence mortality outcomes by improving access to preventive care, diagnosis, and treatment of cardiovascular conditions [16].

Real GDP per Capita (constant 2015 US dollar): This variable serves as a proxy for a country’s overall level of economic development. It is crucial to control national income because the level of economic development determines the quality of infrastructure, nutrition, health systems, and environmental regulations, all of which are associated with lower mortality from non-communicable diseases such as CVD [17]. For example [18], shows that the level of socioeconomic status is strongly associated with cardiovascular risk factors.

Unemployment Rate (% of labor force): Unemployment is included to control for economic downturn at the population level. High unemployment can lead to psychosocial stress and reduced access to healthcare, both of which are risk factors for cardiovascular disease [19]. We control for government effectiveness as a proxy for quality of governance and institutional quality using standardized data from the World Bank’s Governance Indicators database. This is in line with the evidence in the recent literature that shows the importance of institutions for citizens’ health status [20]. In addition, we also control outdoor air pollution measured in average concentration of particulate matters (PM2.5). All the control variables are sourced from the World Development Indicators database.

In addition to these covariates, we also control for country and year fixed effects in the panel data regression framework, allowing the model to control for unobserved, time-invariant country characteristics and global year-specific shocks. Together, they help strengthen the interpretation of the relationship between household energy access and CVD mortality.

Statistical analysis

To examine the impact of lack of access to clean energy on CVD-specific death rates of children under 15, this study employed a series of panel data statistical analysis, including pooled ordinary least square (OLS), system generalized moments (GMM) and quantile regression techniques to ensure robustness of the findings. The use of panel data statistical techniques allows for identification of causal effects of access to clean energy on CVD mortality by controlling for individual country heterogeneity and temporal variations [21]. The pooled OLS approach is useful to serve as a benchmark approach, but it does not address endogeneity issues that may be caused by reverse causality or omitted variable bias. While system GMM is a common approach to deal with endogeneity in the literature, it has several potential pitfalls in the implementation of GMM estimators, including concerns related to instrument proliferation and weak identification [22]. Our main method of statistical analysis is based on quantile regressions [23] which account for both outliers and unobserved heterogeneity.

The benchmark panel data model is specified as follows:

1

where denotes deaths per 100,000 people of children under age 15 in country and year, t; denotes the share of population with lack of access to clean energy fuels or technology; denotes a vector of covariates, including health expenditure as percent of GDP, real GDP per capita, unemployment rate, outdoor air pollution and government effectiveness. denotes country-specific fixed effects, absorbing time-invariant characteristics across countries; is year-fixed effect; t is country-specific linear time trend and represents the error term. The parameter, , captures the response of to changes in . The estimation of Eq. (1) using pooled OLS or system GMM yields only a snapshot of the average relationship between and .

To account for heterogeneity across the entire distribution of the , we also employ the quantile regression approach given by:

2

where is the conditional quantile of CVD given the vector of explanatory variables including and the covariates defined in Eq. (1), and ​ is a vector of coefficients that are quantile-specific, representing the change in the τ^th quantile of CVD for a one-unit change in the explanatory variables. The quantile regression estimates are obtained by solving the minimization problem given by:

3

The main advantage of the quantile regression technique is that it provides detailed view of the conditional distribution of CVD, capturing heterogeneity and varying effects across the entire distribution of the outcome variable. Most importantly, the estimated coefficients from the quantile regression are efficient and robust to outliers in the CVD death rates across countries.

Descriptive analysis

Table 1 presents summary statistics for the variables used in the analysis. The outcome variables measure age-specific deaths of children under 15 from CVD per 100,000 population, while the key explanatory variables capture the share of population with lack of access to clean energy along with control variables of socio-economic and health system characteristics across countries.

Table 1.

Summary statistics

	Obs.	Mean	Std. dev.	Min	Max
Outcome variables: Death rates per 100,000 people
Age < 1 year	1,373	9.75	22.02	0.00	566
Age 1–4 years	1,373	1.35	1.83	0.00	23
Age 5–9 years	1,373	0.74	1.18	0.00	22
Age 10–14 years	1,373	1.19	1.59	0.00	23
Explanatory variables
Lack of access to clean fuels (%)	1,373	6.59	12.57	0.00	62.40
Health expenditure (% of GDP)	1,373	4.76	2.14	0.82	12.63
Log real GDP per capita (2015 USD)	1,373	9.58	1.05	6.58	11.63
Unemployment rate (%)	1,352	7.76	5.11	0.25	37.32
Outdoor air pollution (PM2.5 µg/m³)	1,373	18.83	10.01	4.90	70.04
Government effectiveness (index)	1,305	0.67	0.88	−1.22	2.47

The sample consists of 66 countries spanning the period from 2000 to 2021. List of countries is provided in Appendix A1

Among infants (under age 1), the average CVD mortality rate is approximately 10 per 100,000, with a notably high standard deviation of 22, and a maximum value reaching about 566, indicating substantial variability across countries. Mortality rates sharply decline with age among children: for ages 1–4, the mean is 1.4, falling to 0.7 for ages 5–9, and slightly increasing to 1.2 for ages 10–14.

The main explanatory variable, the share of the population lacking access to clean fuels for cooking, has a mean of 6.6% with a standard deviation of 12.6%. This significant variation underscores the unequal distribution of clean energy access across the sample.

The values for the control variables also exhibit significant variations. Domestic government health expenditure as a percentage of GDP averages 4.76%, ranging from 0.82 to 12.63%, reflecting differing levels of investment in health systems. Log GDP per capita has a mean of 9.57, with values spanning from 6.58 to 11.63, consistent with the inclusion of both low- and high-income countries in the panel. The unemployment rate averages 7.76% with a standard deviation of 5.27, suggesting significant differences in labor market conditions across the sample countries that may influence health outcomes. There are significant variations in the values of outdoor air pollution and government effectiveness as well.

These descriptive statistics highlight the substantial cross-country heterogeneity in both health outcomes and socio-economic indicators, justifying the use of a panel data approach, controlling for fixed effects to address issue of unobserved heterogeneity in the regression analysis.

Figure 1 presents the simple cross-country scatterplots, revealing the correlation between lack of access to clean energy fuels/technology and CVD-related child deaths per 100,000 people across different age groups. The graph in all panels show that death rates from CVDs are positively correlated with the share of population without access to clean fuels.

Fig. 1.

Cross-country correlations between incidence of CVD mortality and lack of access to clean energy fuel/technology

Finally, we examined the time series properties of the variables using panel unit root test proposed by [24] that accounts for cross-sectional dependence. The results reported in Table A2 of the appendix consistently indicate that all our variables (except access to clean fuels) are integrated of order one, I (1). That is, they are non-stationary at their levels but become stationary after first differencing.

Results

Pooled OLS regression results

Table 2 presents the estimation results from pooled OLS regressions. The Table shows the statistical estimates on the association between access to clean fuels and age-specific cardiovascular disease (CVD) mortality rates per 100,000 population across four age groups. The results demonstrate a consistent and statistically significant positive relationship between the lack of access to clean fuels and CVD mortality across all age cohorts. The estimated coefficient on lack of access to clean fuels is statistically significant at 1% significant level in all regressions.

Table 2.

Benchmark pooled OLS estimates

	Outcome variable = mortality rate from cardiovascular disease
	(1)	(2)	(3)	(4)
	Age < 1 year	Age 1–4 years	Age 5–9 years	Age 10–14 years
Lack of access to clean fuels and technology	0.240***	0.026***	0.012***	0.015***
	(0.051)	(0.006)	(0.003)	(0.004)
Health expenditure (% of GDP)	−0.521	−0.040*	−0.030**	−0.072**
	(0.345)	(0.024)	(0.012)	(0.029)
Government effectiveness	−4.725***	−0.359***	−0.279***	−0.314**
	(1.297)	(0.116)	(0.077)	(0.149)
Ln(GDP per capita)	2.901*	0.257**	0.212***	0.193
	(1.677)	(0.114)	(0.078)	(0.173)
Unemployment rate	−0.036	0.025	0.007	0.020
	(0.157)	(0.017)	(0.005)	(0.016)
Outdoor air pollution (PM2.5)	−1.603	0.089	0.087	0.191
	(1.025)	(0.097)	(0.093)	(0.164)
Trend	0.582	−0.025	−0.098**	−0.048
	(0.607)	(0.086)	(0.042)	(0.070)
Number of observations	1207	1207	1207	1207
R-squared	0.203	0.315	0.286	0.216

This Table presents the Pooled OLS estimates of the effect of lack of access to clean fuels and technology on CVDs mortality of children under 15. Heteroskedasticity robust standard errors (clustered at country level) are reported in parentheses

*, **, and *** denote statistical significance at the 10%, 5% and 1% levels respectively

In the population aged less than one year (Column 1), a 10 percentage point increase in the population without access to clean fuels and technologies is associated with a 2.4 increase in CVD deaths per 100,000 people (p < 0.01). The magnitude of this effect declines with age, with estimated coefficients of 0.026, 0.012, and 0.015 for children aged 1–4, 5–9, and 10–14 years, respectively, all statistically significant at the 1% level. These results highlight the detrimental role of dirty fuels that lead to household air pollution and causing cardiovascular disease, suggesting heightened vulnerability among younger populations (see also [13]).

Domestic government health expenditure, measured as a share of GDP, is negatively and significantly associated with CVD mortality in all age groups. For example, among children in age group 10 to 14 years, a one-percentage-point increase in health spending corresponds to a 0.07 decrease in CVD mortality per 100,000 population (p < 0.05). In the younger cohorts, the magnitude of the protective effect is smaller: −0.04 for ages 1–4, and − 0.03 for ages 5–9. These findings underscore the crucial role of public investment in health infrastructure and services in reducing premature cardiovascular deaths. Good governance, proxied by government effectiveness, has a statistically significant and negative association CVDs across all age groups.

Unemployment rate does not exhibit a statistically significant association with CVD mortality in any of the age-specific regressions while GDP seems to have unexpected signs that may be due to nonlinearity or multicollinearity with unemployment rate. These results point to the broader socio-economic determinants of health.

We plot the residuals from a regression of CVD-related mortality on country and year fixed effects against lack of access to clean energy illustrate the relationship after accounting for time-invariant country characteristics and global time trends. The results reported in Appendix A1 support the main conclusions.

System GMM estimation results

An important issue in the estimation of the impact of a lack of access to clean cooking and heating fuels on cardiovascular disease (CVD) mortality a potential bias due to endogeneity. This issue arises as the use of unclean fuels can be correlated with the error term in the model. This correlation can lead to biased and inconsistent estimates, making it difficult to establish a clear causal relationship between access to clean fuels and CVD mortality. The primary sources of this endogeneity are omitted variable bias, reverse causality and measurement errors.

Omitted variable bias is arguably the most pervasive source of endogeneity in studies of household air pollution and health outcomes. It arises when factors such as socioeconomic status that influences both the health outcome variable and lack of access to clean fuels are excluded from the model. Reverse causality or simultaneity can also be a key source of endogeneity as poor cardiovascular health could influence a household’s choice of fuel. This is because household members suffering from a pre-existing cardiovascular condition tend to be too ill to work, leading to a decrease in household income that will force them to switch from cleaner, more expensive fuels to cheaper and more polluting options. As such, the poor health outcome precedes and influences the fuel choice, reversing the expected causal direction. In addition, measurement errors can significantly distort the findings of observational studies without appropriate approach to address endogeneity.

Given potential endogeneity concerns in modeling the determinants of child cardiovascular disease (CVD) mortality, this study employs a two-step system GMM estimator with a lag length of one and Windmeijer-corrected standard errors to ensure consistent and efficient estimates [25].

The key identifying assumptions that underpin our System GMM results rests on two core assumptions about the internal instruments: instrument relevance and instrument validity (exclusion restriction). To be relevant, the instruments must be sufficiently correlated with the endogenous variables they are instrumenting. This assumption is generally considered to hold, as past values of a variable are typically strong predictors of its current value. Regarding the validity, the key assumption for the exclusion criteria is that the instruments must be uncorrelated with the error term. This means that past values of our variables should only affect the current CVD mortality rate through their effect on the current values of the variables in the model, not through any other unobserved channel.

The system GMM approach partially addresses the empirical challenges of endogeneity bias by exploiting the panel structure of the data and using lagged values of endogenous variables as internal instruments although it is not a complete solution due to the issue of several pitfalls such as instrument proliferation [22].

The results, reported in Table 3, offer evidence that lack of access to clean fuels and technology is a consistent and statistically significant determinant of child mortality from CVD across all age groups. Specifically, the coefficient for this variable is positive and significant at the 5% level in all specifications except for Column (4) (10% significance level), indicating that higher rates of reliance on polluting fuels are associated with increased CVD mortality. The estimated effect is particularly pronounced the younger age cohorts of children under one year (column 1), where a 10-percentage point increase in the share of population lacking access to clean fuels is associated with a 2.7 increase in the mortality rates per 100,000 people. Although the magnitude of this effect diminishes in the higher age bands (columns 2 to 4), it remains statistically significant, underscoring the persistent health burden imposed by household air pollution.

Table 3.

System GMM estimation results

	Outcome variable = mortality rate from cardiovascular disease
	(1)	(2)	(3)	(4)
	Age > 1 year	Age 1–4 years	Age 5–9 years	Age 10–14 years
Lack of access to clean fuels and technology	0.269**	0.024**	0.015**	0.018*
	(0.105)	(0.011)	(0.006)	(0.010)
Health expenditure (% of GDP)	−0.745	−0.078	−0.023	−0.146
	(0.766)	(0.050)	(0.031)	(0.115)
Government effectiveness	2.050	0.068	−0.086	−0.685
	(1.759)	(0.259)	(0.159)	(0.603)
Ln(GDP per capita)	−1.553	0.023	0.118	0.555
	(2.304)	(0.362)	(0.160)	(0.486)
Unemployment rate	−0.114	0.006	−0.007	0.017
	(0.143)	(0.022)	(0.015)	(0.032)
Outdoor air pollution (PM2.5)	0.835	0.477*	0.616***	0.746***
	(2.467)	(0.268)	(0.205)	(0.207)
Number of observations	1207	1207	1207	1207
Number of instruments	281	281	281	281
AR (2)	0.707	0.400	0.408	0.519
Hansen test over id. (p-value)	1.000	1.000	1.000	1.000

This Table presents the two-step System GMM estimates of the effect of lack of access to clean fuels and technology on CVDs mortality of children under 15. Robust Windmeijer-corrected standard errors in parentheses

*, **, *** Denotes statistical significance at the 10%, 5% and 1% levels respectively

Outdoor air pollution is also found to be a significant factor in increasing CVD mortality among children, though the significance varies by age group. The coefficient is positive and statistically significant for age cohorts above 1 year. While government spending is negatively associated with CVDs, the estimated coefficient is statistically insignificant. However, the coefficient is not significant in the broader category of children aged over one year, potentially due to greater heterogeneity within this group diluting the effect.

To assess if multicollinearity among the macroeconomic variables (GDP and unemployment rate) is an issue that drives insignificant estimates for most of the control variables, we examined the pairwise correlation matrix for all independent variables. The correlation coefficient between real GDP per capita and unemployment was found to be −0.174. This indicates only a modest negative correlation, as expected, that does not flag a problematic level of collinearity. The insignificant results for most of the control variables may suggest that more proximal factors like outdoor air pollution have a more direct and measurable impact on CVD mortality than broad macroeconomic indicators.

The test for second-order autocorrelation (AR(2)) indicates no evidence of serial correlation in the residuals across all models, with p-values comfortably exceeding conventional thresholds, a necessary condition for the lagged instruments to be valid. The Hansen test for overidentifying restrictions passes in all our models, indicating that we cannot reject the null hypothesis that our instruments are valid (i.e., uncorrelated with the residuals). However, it is worth noting that the high p-values may suggest the issue of instrument proliferation. Therefore, the GMM estimates should be interpreted with caution, and we rely on quantile regression approach as our main empirical strategy.

Accounting for heterogeneity: quantile regression

To further investigate the distributional impacts of cardiovascular disease (CVD) mortality among children and to complement the System GMM estimates, quantile regression was employed. This technique offers significant advantages over traditional mean-based approaches such as OLS or even GMM, particularly when the data exhibit heterogeneity or outliers, which can distort average effects. Quantile regression estimates the conditional median and other quantiles of the dependent variable, allowing for a more nuanced understanding of how covariates influence the outcome across different points of the mortality distribution.

The results, reported in Table 4, reveal a clear pattern of increasing marginal effects across the distribution of the outcome variable. For children under one year of age, the coefficient on the CVD death rate rises markedly from 0.095 at the 25th percentile to 0.468 at the 90th percentile, all statistically significant at the 1% level. This suggests that the relationship between CVD mortality and its determinants becomes more pronounced at the higher end of the distribution, where mortality is already elevated. In other words, the burden of CVD mortality is not evenly distributed, and the adverse effects are disproportionately concentrated among those in the upper quantiles, regions or groups experiencing the worst health outcomes.

Table 4.

Quantile regression results

Outcome variable	(1)	(2)	(3)	(4)
	t = 0.25	t = 0.5	t = 0.75	t = 0.9
Cardiovascular disease death rate, age < 1yr	0.095***	0.202***	0.320***	0.468***
	(0.028)	(0.025)	(0.033)	(0.062)
Cardiovascular disease death rate, age = 1-4yrs	0.031***	0.036***	0.055***	0.076***
	(0.004)	(0.003)	(0.004)	(0.007)
Cardiovascular disease death rate, age = 5-9yrs	0.013***	0.013***	0.016***	0.027***
	(0.002)	(0.002)	(0.002)	(0.005)
Cardiovascular disease death rate, age = 10-14yrs	0.014***	0.016***	0.026***	0.034***
	(0.003)	(0.002)	(0.003)	(0.008)
Control variables included	Yes	Yes	Yes	Yes
Number of observations	1,285	1,285	1,285	1,285

This Table presents quantile regression estimates of the effect of lack of access to clean fuels and technology on CVDs mortality of children under 15. Robust standard errors in parentheses

*** Denotes statistical significance at the 1% level

A similar though more modest gradient is observed for other age categories. For children aged 1–4 years, the coefficient increases steadily from 0.031 at the 25th percentile to 0.076 at the 90th percentile. Likewise, for ages 5–9 and 10–14 years, the estimated effects are consistently positive and increase with the quantile, indicating that the impact of underlying risk factors intensifies in higher-risk populations. This upward trend in coefficients across quantiles underscores the presence of distributional heterogeneity, where traditional mean regression methods would potentially understate the severity of determinants in the most affected cohorts.

Importantly, the consistent significance of the coefficients across all quantiles and age groups highlights the pervasiveness of CVD mortality risk, while the rising magnitude across higher quantiles points to nonlinear vulnerabilities, possibly due to compounding disadvantages such as poor access to healthcare, environmental exposures, or socioeconomic deprivation. These findings reinforce the argument that policy responses should not only target average outcomes but also be tailored to high-risk subgroups experiencing the worst health conditions.

From a methodological standpoint, the quantile regression approach strengthens the empirical analysis by revealing effects that might be masked in mean-based estimators. The strong statistical significance and the monotonic increase in coefficients across quantiles provide robust evidence of the heterogeneous effects of cardiovascular disease determinants. These results align with and deepen the insights from the System GMM analysis, particularly by illustrating that endogeneity-adjusted mean effects may underestimate the challenges faced by the most vulnerable children.

For a better visualization of the quantile estimation results, Figs. 2, 3, 4 and 5 present the plots for the quantile estimates along with the OLS benchmark estimates for the four age groups. The blue solid lines represent the quantile estimates along with the 95% confidence interval represented by the shaded region while the OLS benchmark average estimates are represented by the horizontal black solid line.

Fig. 2.

Quantile regression estimates for age cohort of under 1 year

Notes: The solid black line represents the OLS benchmark while the solid blue lines represent the quantile estimates along with the 95% confidence interval given by the shaded area

Fig. 3.

Quantile regression estimates for age cohort of under 1 to 4 years

Fig. 4.

Quantile regression estimates for age cohort of under to 5 to 9 years

Fig. 5.

Quantile regression estimates for age cohort of under to 10 to 14 years

The results clearly indicate the heterogeneous impacts of lack of access to clean fuels on CVD mortality rates across all age groups. The quantile estimates are lower than the OLS estimates at the lower-end of the CVD mortality distribution and continuously rising to above the OLS benchmark estimates at the higher-end of the CVD mortality distribution.

The consistent association between lack of clean fuel access and elevated mortality underscores the urgent need for energy transition policies to accelerate net zero transition and universal access to clean energy by 2050, a target in which countries around the world are aiming to achieve by 2050. Similarly, the insulating effect of health spending points to the importance of sustained public investment in health systems to combat non-communicable diseases in early life.

While GDP per capita does not consistently exhibit a significant association with CVD mortality, government health expenditure and the quality of governance (government effectiveness) show a robust and significant negative relationship with CVD mortality across specifications. This suggests that the allocation of resources, particularly investment in the health sector and institutional quality, play a more critical role in improving child health outcomes than income levels alone.

Discussion

This study investigates the determinants of cardiovascular disease (CVD) mortality among children using a comprehensive panel dataset and advanced panel data estimation methods. These methods were selected to address key econometric challenges often overlooked in health-related cross-country studies, including endogeneity, unobserved heterogeneity, and outcome heterogeneity across the distribution. For example, based on a systematic review of related studies that model access to clean fuels and health outcomes [18], underscore that it is critical to account of observed and unobserved contexts and heterogeneities across units and over time.

Our results lend support to the findings of earlier studies that document exposure to pollutants from household combustion of solid fuels can trigger cardiovascular events through several biological pathways. For example [7], finds that the adoption of solid biofuels for cooking significantly increases the probability and frequency of reporting ill-health and frequently reporting ill-health by 25%. With a focus on pathophysiological mechanisms several studies such as [26] and [27], find that short-term exposure to such pollutants has been associated with CVDs such as endothelial dysfunction, increased blood pressure, and altered autonomic balance. Similarly [28] and [29], find that long-term exposure leads to systemic inflammation, oxidative stress, and atherosclerosis progression [30]. shows PM2.5 and other pollutants from use of dirty fuels penetrate deep into the lungs and enter the bloodstream, contributing to myocardial infarction, stroke, and heart failure.

From the point of epidemiological evidence, a strand of literature has demonstrated a strong association between household air pollution and CVD mortality. For example [31], find that individuals exposed to solid fuels had a significantly higher risk of major cardiovascular events and death compared to those using clean fuels. In a systematic review [12], also document that solid fuel use was associated with a significant increased risk of CVD mortality, especially ischemic heart disease and stroke.

Well-placed with evidence in existing literature, our results provide compelling evidence that lack of access to clean fuels and technology is a significant and consistent driver of CVD mortality in children across all age groups. This underscores the health consequences of household air pollution, particularly in settings where clean energy remains inaccessible [6, 32].

Our study differs from the existing studies that heavily rely on correlational analysis based on average relationships. While mean-based estimators provide important insights, they may fail to capture the distributional complexities inherent in health outcomes. The quantile regression results illuminate this by showing that the effect of CVD determinants intensifies at higher quantiles of the mortality distribution. Children in the upper tail, those experiencing the most severe health outcomes, are disproportionately affected by environmental vulnerabilities. For instance, the effect of early-age CVD mortality on later age groups becomes markedly stronger in the 75th and 90th percentiles, highlighting the need for policies that explicitly consider high-risk populations. This finding lends support the conclusion of [33] which documents compelling evidence that the highest risk group to faces existential threat from air pollutants tend to be affected 19 times compared to the lowest-risk group.

Public health expenditure and quality of governance (government effectiveness) are also found to be key determinants of child health outcomes, suggesting that greater investment in national health systems can yield tangible improvements in child health outcomes.

Taken together, these findings reinforce the multifaceted nature of child health vulnerability. Environmental deprivation, especially poor access to clean energy, and insufficient public health investment emerge as critical risk factors. Equally important is the methodological contribution of the study: by applying Quantile Regression, it offers both distribution-sensitive insights, allowing for more effective policy targeting.

Conclusion

Worldwide, cardiovascular diseases (CVDs) are among the leading cause of death with an estimated 17.9 million deaths annually [34]. Lack of access to clean fuels by households is a challenge to policymakers given its contribution to indoor pollution that could contribute to CVDs. In this study, we investigate the impact of lack of access to clean energy on cardiovascular disease (CVD) mortality among children across multiple age groups using long historical panel data from 66 countries across the world spanning the period from 2000 to 2021. Recognizing the potential for endogeneity and unobserved heterogeneity, the analysis employs advanced econometric techniques to uncover both average and distributional effects.

Based on our preferred quantile regression approach, we find that lack of access to clean fuels and technology is a significant and consistent predictor of higher CVD mortality in children across all age groups. The impacts are more severe for infants (under age 1). Health expenditure as a share of GDP and government effectiveness are inversely associated with child mortality. The estimates from the quantile regression results reveal that these effects are more pronounced at higher quantiles of the mortality distribution, indicating that the most vulnerable populations bear a disproportionate burden.

Our results suggest that policymakers should consider expanding access to clean household energy technologies in line with the net zero transition targets and scaling up health sector investments, particularly in communities facing elevated child mortality risks. Moreover, interventions must be differentiated across the distribution of outcomes to ensure that the most vulnerable subgroups are not left behind. Through such integrated approaches, grounded in sound empirical evidence, meaningful progress can be made toward improving child health and equity in developing settings.

Abbreviations

CVD

Cardiovascular Disease

GMM

Generalized Method of Moments

IHME

Institute For Health Metrics and Evaluation

OLS

Ordinary Least Square

SDG

Sustainable Development Goal

WHO

World Health Organization

Authors’ contributions

Authorship contribution statement: Abebe Hailemariam: Conceptualization, formal analysis, software, modelling, analysis, writing manuscript, review & editing, and visualization. Mandefrogn Mammo Lemma: Conceptualization, data analysis, writing manuscript, reviewing. Estifanos Esubalew Lewogineh: data curation, formal analysis, writing – original draft.

Funding

None.

Data availability

The data are sourced from publicly available sources including The World Development Indicators (WDI) and Global Burden of Disease. No proprietary data used in this study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.