Changes in the morbidity and mortality estimates
We analysed the number of Deaths, DALYs and YLLs attributable to air pollution in eight East African countries. The total number of deaths attributable to air pollution in all these countries combined was 183,309 in 2019. This indicates a decrease of 14.26% in mortality numbers when compared to the numbers (213,797) in 1990. Despite the reduction in the total number of deaths, the number of deaths attributable to air pollution increased for all causes, except for LRI which decreased by 43.09% in 2019 when compared to 1990. The country changes in the total number of deaths attributable to air pollution decreased in five countries (Burundi, Rwanda, Tanzania, Ethiopia, and South Sudan), and increased in three remaining countries (Kenya, Uganda, and Somalia) as presented in Fig. 1. The highest reduction in the region was observed in Ethiopia (-40.90%) and Rwanda (-34.47%), while the highest rise came from Somalia (+ 67.49%).
The Disability-Adjusted Lost Years (DALYs) summed from all countries were 7,510,511 years, a reduction of 40.19% in 2019 when compared with the year 1990 (12,557,393). Regardless of the overall drop in DALYs, all causes related to air pollution increased the DALYs except the LRI (Fig. 2). The higher decline in DALYs was observed in Ethiopia (-62.63%), while the higher rise in DALY was in Somalia (+ 45.55%) as shown in Fig. 1. The principal cause of the DALY during that period was LRI (77.89%), while the least cause was the TBLC (0.49%). Other health burdens respectively contributed by 9.81%; 5.89%; 4.48% and 2.38% for stroke, IHD, COPD, and T2DM. Our analysis showed a general decrease of YLLs in all East African countries, except for Somalia which observed an increase of 43.85% of YLLs attributable to air pollution in 2019 when compared to 1990. The highest decrease in YLL came from Ethiopia (-64.15%) followed by Rwanda (-59.69%).
Figure 2 exhibits the changes in the number of deaths, DALYs and YLLs attributable to air pollution in the region as by their source causes. The higher number of deaths increase came from TBLC (88.60%) and IHD (65.66%), while LRI is the only cause of observed decrease (-43.09%) of deaths in 2019 when compared to the year 1990 (Fig. 2). Similar to the mortality causes, DALY and YLL coming from all causes also increased in 2019 when compared to the numbers of 1990, except for the LRI. The Years of life lost (YLLs) were marked by an overall decrease from the rate of 67,076 years per 100,000 persons in 1990 to 22,173 years per 100,000 persons in 2019, which represents a decrease of 66.94% between these two reference dates. This decreases principally due to the decrease of YLL coming from LRI disease. In terms of the causes of death, the primary cause of death for the subset of the East African region was LRI and stroke, causing the respective annual average contribution of 46.1% and 21.4% for the 30 years. The TBLC is the least cause with an annual average of 1.8%. Our analysis found that LRI is also the primary cause of DALY and YLL, making up a multi-annual average contribution of 66.6% and 69.7% for DALY and YLL, respectively.
Even though the total burden of diseases was marked by an increase in most of the countries, the East African region observed a decrease in both rates of number mortality and morbidity per 100,000 persons attributable to air pollution as presented in Fig. 3. Ethiopia led other countries in terms of cutting the rate of the number of mortality and morbidity attributable to air pollution, while LRI is the most reduced cause. The reduction of the LRI burden contributed to the reduction of the overall rate of mortality and morbidity.
Countries' efforts in the reduction of Deaths, DALYs and YLLs
Figure 4 and Table 1 exhibit the DRR, DARR and YRR for eight considered in this study. Ethiopia and Rwanda are the only countries, in the region, which have decreased the Death, DALY, and YLL rates above the median of the region. Ethiopia presented the highest DRR, DARR and YRR of 34.3%; 34.2%; and 34.1%, respectively. The remaining six countries have not managed to reduce their death, DALY, and YLL attributable to air pollution to at least the extent of the East African region. The lowest change in DRR came from Burundi (-56.7%), while the lowest improvement in both DARR and YRR came from Somalia with the relative scores of -88.4% and − 94.6%, respectively. The countries with negative values of DRR, DARR and YRR are still not capable of improving their Death, DALY, and YLL rates at least on the median values of the East African countries.
Table 1
Comparison of the ratio of Death, DALY, and YLL rates attributable to air pollution in 1990 and 2019 for East African countries
Countries
|
DRR
|
DARR
|
YRR
|
1990
|
2019
|
Changes
|
1990
|
2019
|
Changes
|
1990
|
2019
|
Changes
|
Burundi
|
0.75
|
1.17
|
-56.7%
|
0.94
|
1.22
|
-29.6%
|
0.93
|
1.23
|
-31.4%
|
Ethiopia
|
1.31
|
0.86
|
34.3%
|
1.29
|
0.85
|
34.2%
|
1.30
|
0.85
|
34.1%
|
Kenya
|
0.54
|
0.73
|
-35.3%
|
0.50
|
0.68
|
-36.7%
|
0.49
|
0.67
|
-35.3%
|
Rwanda
|
1.15
|
0.99
|
13.8%
|
1.04
|
0.92
|
11.4%
|
1.04
|
0.91
|
12.2%
|
Somalia
|
1.28
|
1.76
|
-36.9%
|
1.32
|
2.48
|
-88.4%
|
1.32
|
2.57
|
-94.6%
|
South Soudan
|
1.04
|
1.20
|
-15.3%
|
1.06
|
1.64
|
-54.8%
|
1.06
|
1.68
|
-58.2%
|
Tanzania
|
0.96
|
1.01
|
-5.4%
|
0.96
|
1.08
|
-12.3%
|
0.96
|
1.09
|
-13.1%
|
Uganda
|
0.69
|
0.78
|
-12.9%
|
0.59
|
0.80
|
-34.6%
|
0.59
|
0.79
|
-35.4%
|
Temporal variability of O3, HAP, and PM2.5
We analysed the temporal change in air pollutant parameters (O3, PM2.5 and HAP) for the period of 1990 to 2019. The regional average annual air pollutant concentrations were 36.66 ppb (95% Confidence Interval: 33.68–39.64) for O3; 32.90 µg/m− 3 (95% CI: 29.17– 36.63) for PM2.5; and 0.972 (95% CI: 0.948–0.997) for HAP. Figure 5 presents the air pollutant concentration time series in the concerned countries. O3 concentration increased in all countries, while PM2.5 did not change a lot during that period. Even though the level of O3 increased in all countries, East African countries may be clustered into three groups. The first group includes countries such as Somalia, Tanzania, and Kenya. These countries have the lowest, but increasing, levels of ground-level O3 mixing ratio. The second group comprises Ethiopia, Rwanda, Burundi, and Uganda. This group is characterized by medium ground-level O3 concentrations increasing at the same rhythm from 1990. The third group contains only South Sudan which started the study period with an elevated level of O3 concentration, and it did not show any significant change in its level during the study period, and it was finally overtaken by the countries in the second group.
Furthermore, the analysis done on the ambient PM2.5 revealed that all countries have not managed to reduce the level of PM25 during the study period. The ambient particulate matter remains above 35 µg/m3 in the territories of Rwanda, South Sudan, and Uganda, and it was above 25 µg/m3 during 90.2% of the time for the whole region. In addition to the PM2.5 and O3, countries observed elevated HAP values during the whole study period. The annual regional HAP was 0.983 (95% CI: 0.964–1.001) in 1990 and this value declined to 0.967 (95% CI: 0.940–0.997) in 2019 with the drop in HAP observed in all countries, except for Tanzania, as witnessed by the Mann-Kandel trend test results presented in Table 2. Defiantly, our analysis identified the increase of the ground level O3 in all countries except for South Sudan. Furthermore, we did not detect any progress in the reduction of ambient PM2.5 in the region.
Table 2
Mann-Kandel trend test for O3, PM2.5 and HAP for each participating country
|
O3
|
PM2.5
|
HAP
|
τ
|
s
|
p*
|
τ
|
s
|
p*
|
τ
|
s
|
p*
|
Burundi
|
0.846
|
77.0
|
0.000
|
-0.121
|
-11.0
|
0.584*
|
-0.901
|
-82.0
|
0.000
|
Ethiopia
|
0.791
|
72.0
|
0.000
|
0.308
|
28.0
|
0.139*
|
-0.890
|
-81.0
|
0.000
|
Kenya
|
0.813
|
74.0
|
0.000
|
0.033
|
3.00
|
0.913*
|
-1.000
|
-91.0
|
0.000
|
Rwanda
|
0.868
|
79.0
|
0.000
|
-0.011
|
-1.00
|
0.100*
|
-0.879
|
-80.0
|
0.000
|
Somalia
|
0.912
|
83.0
|
0.000
|
0.286
|
26.0
|
0.171*
|
-1.000
|
-91.0
|
0.000
|
South Sudan
|
0.022
|
2.00
|
0.956*
|
-0.121
|
-11.0
|
0.584*
|
-0.769
|
-70.0
|
0.000
|
Tanzania
|
0.835
|
76.0
|
0.000
|
0.242
|
22.0
|
0.250*
|
0.022
|
2.00
|
0.956*
|
Uganda
|
0.670
|
61.0
|
0.001
|
0.143
|
13.0
|
0.511*
|
-0.670
|
-61.0
|
0.001
|
In Table 2, Kendell’s tau (τ) represents the measure of the monotony of the slope; Sen slope (s) reflects the overall slope of the series; and p (or p-value) implies the statistical significance of the model. Values with p-value < 0.05 indicate the rejection of the null hypothesis of no trend and thus, reveal the existence of the trend. The trend is increasing for a positive value of Kendell’s tau (τ), and vice versa. The p-values with an asterisk (p*) indicate the lack of statistical significance of the trend (i.e., \(p\ge 0.05\)).
The relative contribution of air pollutants to morbidity and mortality
We used a two-stage approach in estimating the association between air pollution and health burden. The first stage concerned with the analysis of pollution effects at the country level, while during the second stage, we made our analysis on the regional level. The correlation analysis classified countries into three groups. The first group is formed by countries in which at least two of three pollutants significantly predicted the health effects of the observed pollution. The second includes countries in which at least two of three pollutants projected the health outcome of observed air pollution levels but without a strong statistical significance. The last group comprises countries in which only one of the air pollutants is linearly linked with health effects. The composition of those groups accompanied by the multilinear regression (MLR) model results is presented in Table 3.
Table 3
Multilinear regression model analysis results
No.
|
Country
|
MLR Model for YLLs
|
R2
|
F-statistics
(p-value)
|
I
|
Kenya
|
\(log\left(YLL\right)=5.66+0.38log\left(P{M}_{2.5}\right)+4.87log\left(HAP\right)\)
|
0.77
|
21.1 (0.000)
|
Somalia
|
\(log\left(YLL\right)=4.74+0.48log\left(P{M}_{2.5}\right)+0.41log\left({O}_{3}\right)\)
|
0.61
|
18.4 (0.006)
|
South Sudan
|
\(log\left(YLL\right)=3.81+0.14log\left({O}_{3}\right)+63.96log\left(HAP\right)\)
|
0.74
|
15.4 (0.001)
|
Burundi
|
\(log\left(YLL\right)=4.08+1.08log\left({O}_{3}\right)+80.24log\left(HAP\right)\)
|
0.68
|
11.5 (0.002)
|
II
|
Uganda
|
\(log\left(YLL\right)=3.80+0.19log\left(P{M}_{2.5}\right)+21.40log\left(HAP\right)\)
|
0.55
|
2.39 (0.138)
|
III
|
Ethiopia
|
\(log \left(YLL\right) = 7.21 + 57.80log \left(HAP\right)\)
|
0.81
|
34.6 (0.000)
|
Rwanda
|
\(log \left(YLL\right)= 6.07+ 171.57log \left(HAP\right)\)
|
0.74
|
51.0 (0.000)
|
Tanzania
|
\(log \left(YLL\right) = 6.56 + 14.53log \left(HAP\right)\)
|
0.43
|
9.10 (0.011)
|
A multilinear analysis was conducted on the countries in groups one and two to reveal the predictability of YLLs by air pollutants of the most concern in each country. The model results are presented in Table 3. If we take Kenya as a practical example, the model output indicates that a 10% change in PM2.5, results in a 3.8% change in YLL, while a 10% increase in HAP leads to a 48.7% increase in YLL. Our model revealed that HAP was the pollutant of the most concern in seven out of eight countries, with Somalia being the only country where YLL is not strongly correlated with HAP. Nevertheless, we failed to predict the YLL in the region simultaneously for three independent variables (PM2.5, O3, and HAP). The failure followed the lack of a simultaneous strong positive linear relationship between all these air pollutants and the regional total YLL. However, HAP strongly predicted regional YLL for the study period with the single linear regression equation given by Eq. (4),
$$\text{log}\left(YLL\right)=7.41+31.06\times \text{log}\left(HAP\right); \left({R}^{2}=0.93;F=169.3;p=1.95\times {10}^{-8}\right) \left(4\right)$$
This model implies that an increase of 1% in regional HAP will cause a 31.06% increase in YLL. The strong predictivity of YLL by HAP confirms the fact that HAP is a pollutant of strong concern in the region.