Assessing the effects of mining projects on child health in sub-Saharan Africa: a multi-country analysis

Table 5 Estimates of association between child health outcomes, anthropometrics, and mining exposure using the main specifications

Interaction (proximity*active)$	(1)	(2)	(3)	(4)	(5)
Interaction (proximity*active)$	Diarrhoeal episodes	Cough episodes	Height-for-Age z-scores	Weight-for-Age z-scores	Weight-for-Height z-scores
Crude model^†	0.74	1.01	−0.35*	−0.07	− 0.03
Crude model^†	(0.54–1.01)	(0.73–1.42)	(−0.62 - -0.08)	(− 0.64–0.50)	(−0.50–0.43)
Observations	59,868	58,593	35,027	35,609	34,594
Adjusted model^‡	0.68**	1.01	−0.16	0.10	0.10
Adjusted model^‡	(0.51–0.90)	(0.77–1.31)	(−0.40–0.08)	(−0.28–0.48)	(−0.28–0.48)
Observations	59,078	57,799	35,027	35,609	34,594

* p < 0.05, ** p < 0.01
$ - interaction term between clusters’ proximity (0–10 km) and mine activity status at survey year; † − model including interaction term only; ‡ − adjusted for gender, child age, twin births, maternal age, maternal education, residence, wealth index, birth order, number of children ever born to mother
The treatment group corresponds to children located within a distance radius of 10 km from active mines at the DHS survey year. The reference group (control) are children located within a distance radius of 10 km before mine activation and those born 10–50 km away regardless of mines’ activity status at the DHS survey year
Logistic regression models are used for estimating the odds ratio for diarrhoeal, and cough episodes (columns (1) and (2)) and linear regression models are used for anthropometric indicators (columns (3), (4), and (5)). The reported estimates for morbidities (i.e., diarrhoea and cough) are crude and adjusted odds ratios (OR), and the child’s anthropometrics are crude and adjusted beta coefficients. The 95% confidence intervals (CIs) are shown in parentheses and are clustered at the survey-cluster level

ISSN: 1744-8603