Ambient Air Pollution and Chronic kidney disease risk in Deltan communities: A Policy Brief, 2023

Evidence of high ambient air pollutant levels in Warri

No air monitoring data existed in the State at the time this study was conducted. Ambient air pollutants were measured using portable air sensors, in collaboration with two environmental scientists and eight citizen scientists from four communities at varying distances from the petrochemical refinery: A (3 km/semi-urban), B (3.5 km/urban), C (10 km/urban), and D (13 km/rural). The air sensors were calibrated, and the citizen scientists were trained on how to use them and record their findings. For each community, two people took repeated measurements of six air pollutants over a period of 4 weeks. The geographical coordinates of each observer’s location, temperature, and relative humidity were also recorded.

The average levels of PM_2.5, PM₁₀, and volatile organic compounds (VOCs) were higher than the WHO acceptable limits in all four communities. However, CO₂ levels were only acceptable in the communities that were the farthest away from the refinery ( Figure 1). Ozone (O₃) was within the acceptable limits in all communities. The mean VOC ranged between 0.280 and 0.320 ppm (acceptable limit = 0.220 ppm). The mean PM₁₀ concentration was highest in the two communities closest to the refinery (A = 55.54 and B = 55.43 μg/m³), while PM_2.5 was highest in the urban community closest to the refinery (B = 28.01 μg/m³) (Okoye, 2024, pp. 250-254). Higher than acceptable levels of NO₂ (>0.1-0.2 ppm) were recorded on certain days in all communities, whereas for most other days, it was negligible (0.0 or 0.1 ppm). The PM_2.5 concentrations for three of the communities are five times higher than the WHO acceptable limit and higher than IQAir (2023) report of 23.9 μg/m³ and 14.8 μg/m³ for Nigeria and Warri respectively, based on estimated satellite data (IQAir, 2023). Furthermore, the concentrations are much higher than 7.8 μg/m³ achieved in Angola, the least polluted African country, which ranked 114th out of 134 countries, while Nigeria ranked the 35th most polluted.

Figure 1. Mean air pollutant concentrations compared with acceptable limits.

The calculated individual exposure (IE = mean air pollutant concentration x duration of exposure) of all air pollutants was statistically significantly higher in participants who had CKD than those who did not. However, there was a weak negative correlation between estimated glomerular filtration rate (eGFR) and IEPM_2.5, IEPM₁₀, IECO₂, and IEVOC, respectively ( Figure 2).

Figure 2. Correlation between calculated individual exposure (IE) to air pollutants and eGFR.

A. Correlation between IE_PM2.5, IE_PM10 and eGFR. B. Correlation between IE_CO2 and eGFR; and IE_VOC and eGFR.

The HQ was estimated by dividing the mean concentration of the individual air pollutants by their respective WHO minimum acceptable limits. An HQ ≤1 is considered a negligible hazard, while >1 indicates exposure concentrations exceeding the reference limit, but not necessarily a statistical probability of harm occurring. The calculated HQ for PM _2.5, PM ₁₀, VOC, and CO ₂ based on the WHO minimum allowable limits, were elevated in all four communities. The total HQ for all pollutants were 11.27, 11.63, 9.63, and 10.63 for communities A-D. However, these aggregate figures do not necessarily represent magnitudes or synergies of health effects.

High prevalence of CKD near the refinery

A cross-sectional study was conducted over a period of six months to assess the prevalence and risk factors of CKD among 1460 community members selected by multi-stage sampling from the four communities. Adults aged 18-64 years who had resided in their respective communities for at least 5 years were recruited. The four study communities are depicted as follows: ‘A’ - nearest to refinery/semiurban, B - near/urban, C - far/urban and D - farthest/rural, to ensure participants’ privacy and anonymity. The majority of participants were female (71%) and there was no significant difference across the four communities. The mean age was 44±13 years; it was highest in far/urban and lowest in the nearest/semi-urban community (P ≤ .001). More than half of participants in the far/urban (66.6%) and far/rural communities were above 50 years (56%), compared with 50% respectively for the near communities.

The overall prevalence of CKD, defined as dipstick proteinuria and/or an eGFR <60 ml/min was 12.3%. The prevalence was highest in the nearest/semi-urban community (17.9%) compared with 13.1%, 10.5%, and 8.0% in the near/urban, far/urban and farthest/rural communities respectively (P≤.001). Proteinuria alone was detected in 6.8% of all participants, while 6.6% had a reduced eGFR of <60 ml/min. The prevalence of CKD reported across Nigeria and Sub-Saharan Africa varies greatly depending on the population studied, CKD definition and methodology; from as low as 2% to >20% (Abd ElHafeez et al., 2018; Chukwuonye et al., 2018). However, our findings demonstrate a higher CKD prevalence in the nearest/semiurban community, compared with a pooled prevalence of 10% and 13.7% reported for Africa and Nigeria, respectively (Abd ElHafeez et al., 2018; Raji et al., 2024).

Two-fifths of the participants with CKD were in stage 3A (i.e., eGFR 45-59 mls/min) which represents a mild to moderate decrease in kidney function requiring monitoring. The nearest/semi-urban and near/urban communities had a higher proportion of participants in stage 1 and 2 CKD (proteinuria with eGFR>60 ml/min) while the far/urban and farthest/rural communities had more participants in Stage 3A ( Figure 3). Higher occurrence of proteinuria among participants closer to the refinery suggests a glomerular mechanism of kidney injury which has been previously reported (Blum et al., 2020; Li et al., 2021; Xu et al., 2016; Yan et al., 2014). However, further experimental studies are needed to establish this. In contrast, the lower prevalence of proteinuria among the farther two communities suggests that the mechanism of kidney damage may be different. Considering that the latter two communities had an older population, aging and related comorbidities may have contributed significantly to CKD.

Figure 3. CKD staging across the four communities.

*A=nearest/semi-urban| B=near/urban| C=far/urban| D=farthest/rural.

Air pollution-CKD Association

The risk factors significantly associated with CKD were proximity to the refinery, diabetes, hypertension, increasing age, low level of education, residence in urban/semi-urban areas compared to rural areas, use of hair dyes and spending more time outdoors. However, after adjusting for confounding factors such as gender, age, LOE, diabetes and hypertension, the statistically significant independent risk factors CKD were proximity to the refinery [OR=2.00 (1.43–2.81)], increasing age [OR=1.02(1.005–1.04)], hypertension [OR=1.61(1.12-2.31)], and level of education [OR=0.63(0.44-0.91)] (Okoye, 2024, pp. 275). In a further logistic mixed model using ‘R’, which accounted for clustering effects at household level, only increasing age was an independent risk factor for CKD [OR=1.26 (1.09-1.45), P=.002,]. This observation suggests that intrahousehold homogeneity significantly accounted for the variance observed. While proximity to the refinery did not sufficiently predict CKD risk, it probably acts synergistically with other prevalent risk factors and exposures to increase the risk for CKD as explained in the multicausation theory, which is applicable to most non-communicable diseases. On the other hand, adjusting for intrahousehold homogeneity led to reduced intra-group sample sizes which can cause non-observance of statistical effects even where it exists (Type II error).

There are emerging empirical evidence on air pollution-associated kidney disease, though a majority are based on studies conducted in the global north (Chan et al., 2018; Liu et al., 2020; Okoye et al., 2022; Wu et al., 2020; Ye et al., 2021). These studies consistently report that PM and NO₂ exposure increases the risk of CKD by 4-70%. Our findings strongly support that this impact of air-pollution on the kidneys also applies to disadvantaged areas, like the Niger Delta. There is a scarcity of similar studies from Nigeria and Africa as was reported in our systematic review of 14 studies, in which no study from Nigeria or Africa investigated CKD as an outcome (Okoye et al., 2022).

Lastly, the overall prevalence of hypertension, obesity, and diabetes was 33%, 28.5%, and 6.0%, respectively.

Social determinants of CKD risk

One-third (31.5%) of the population had less than secondary-level education, and 50.5% earned less than the minimum wage. Although 86% of the population was employed, 68% were self-employed, and only 3.8% were employed by the government. Of the 68% self-employed, the majority were petty traders. Several social risk factors and toxic environmental exposures associated with CKD and NCDs were prevalent among residents of the four communities. Behavioural factors included unhealthy dietary habits (70–90%), low physical activity (47.2%), and habitual exposure to potential nephrotoxins (37–44%). Four-fifths of the population was regularly exposed to petrochemical products as part of their daily lives, 72% used household chemicals regularly, 53.2% were regularly exposed to pesticides, and 49% were exposed to toxic chemicals or dust in their jobs. Other risk factors that were relatively less prevalent included hair dye use (19%), excess salt intake (15%), use of mothballs (14.4%), use of skin lighteners (12.7%), and current smoking (3.8%).

The concentration of multiple social and environmental risk factors in the studied population may explain the high prevalence of CKD and other NCDs. These findings support the multi-causation theory, drawing attention to the need for a multi-faceted approach to CKD prevention.

Low air pollution health risk literacy in Warri

Two-fifths of the 1460 survey participants perceived that their outdoor air was polluted, and the proportion was significantly higher (65%) among those residing near the refinery. Heightened perception of air pollution was significantly more common among young people, those who lived near refineries and urban areas, those who spent more time outdoors, and those who cooked with propane gas. Refinery activities were cited as the most popular source of air pollution. A higher proportion of those residing near the refinery attributed air pollution to the refinery/gas plant: 40.6% and 18.0% for communities A and B, respectively, compared to 7.2% and 6.1% for the farther communities C and D, respectively. Other perceived sources of air pollution include poor environmental sanitation, traffic emissions, generator fumes, open waste burning, and illegal oil refining ( Figure 4).

Figure 4. Participants’ perception of sources of air pollution (N=628).

Others = Bakery, other industries, dust, overcrowding, sawmills, septic pits, swamps.

Most participants (70.1%) perceived that air pollution is associated with health risks, 13.4% responded negatively, and 16.4% did not know. The majority of study participants (60.1%) were unaware of any medical conditions associated with air pollution ( Figure 5). This low literacy level suggests that the necessary preventive measures, such as individual behavioural changes, are lacking, and this may contribute to the high burden of air pollution associated NCDs in the community.

Figure 5. Medical conditions the participants associated with air pollution (N=1460).

Respiratory disease = cough, catarrh, difficulty with breathing, chest pain, asthma, COPD, lung cancer, lung disease | Others= infections (not specified), measles, small pox, air borne disease, diarrhoea, nausea, liver disease, stomach ache, heart burn.

Only 12.3% of the participants agreed that the ambient air environment was well controlled and up to 60% placed responsibility solely on the government. Among those who agreed that they had a role to play, the responses included maintaining environmental sanitation (53%), complaining to the government and advocacy (32%), and using personal protective measures (3.7%).

Implications for policy

Lastly, our findings are generalisable to similar vulnerable populations across the globe who reside near point emission sources. Therefore, the following recommendations may be useful for future public health interventions in these settings.

Actionable recommendations

Table 1 below details recommendations based on our findings.

Limitations of the study

Air monitoring conducted for a period of 4 weeks served as a surrogate for annual exposure; this was due to the high financial implications and tenure of the research. Urine protein was tested using dipsticks rather than the than the albumin-creatinine ratio, which is more reliable due to the high cost of the test. However, the dipsticks test is highly specific though less sensitive in detecting low levels of proteinuria. Lastly, the diagnosis of CKD was based on a spot-assessment of urine protein excretion and eGFR and may have led to an over- or underdiagnosis of CKD. The participants’ who had abnormalities were unwilling to repeat their tests due to the fear of confirming a new disease, despite repeated attempts at inviting them through phone calls and physical visits to the community. Although this was a large sample study, the cluster size variability led to significant design effects in the prediction model.

Despite the limitations, the strength of the underlying study lies in the innovative combination of multiple research methods, interdisciplinarity and involving citizen scientists in addressing a public health problem. The extensive consideration of potential clinical, social and environmental NCD risk factors and adjusting for confounding factors strengthens the study quality.

More research is required from underserved areas to explore: this exposure-effect relationship, the mechanism of air pollution associated CKD and potential interventions to reverse it, and to develop educational interventions that would effectively improve public awareness of environmental health risks. Based on our experience with intrahousehold sample selection and the resultant design effects, we suggest that subsequent studies should consider systematic selection of households (the more, the better) and intra-household selection to ensure a constant cluster size and acceptable design effects.