Table 3

Pooled effect estimates for cough where n represents the number of studies and p values derive from the Q combinality test for homogeneity between study effect estimates

Pollutant	Method of pooling or subgroup	Multiplicative change in symptom odds per unit (μg.m⁻³) increase in pollutant			Expected impact on symptom odds of a 50 μg.m⁻³ rise in pollutant	Q test p value	n
Pollutant	Method of pooling or subgroup	Pooled effect estimate	95% confidence interval			Q test p value	n
*Number greater than the total number of studies as some studies present results separately for symptomatic and non-symptomatic/asthmatic children.
PM_2.5	Overall pooled effect estimate
	Fixed effects model	1.005	1.003	1.007	1.28	<0.025	6
	Random effects model	1.010	1.005	1.016	1.64
PM₁₀	Overall pooled effect estimate
	Fixed effects model	1.000	0.999	1.001	1.00	<0.001	12
	Random effects model	1.004	1.002	1.006	1.22		12
	Subgroups based on outcome measures (fixed effects models)
	Symptom prevalence only	1.000	0.999	1.001	1.00	<0.001	8
	Subgroups based on the population studied (fixed effects models)
	Asthmatic or symptomatic children	1.000	0.999	1.001	1.00	<0.001	9*
	Non-symptomatic children	1.002	1.001	1.004	1.11	<0.025	5*
	Subgroups based on environmental conditions (fixed effects models)
	High ozone levels	1.010	1.006	1.014	1.64	<0.001	3
	High SO₂ levels	1.001	1.000	1.002	1.05	<0.025	2
	High PM₁₀ levels only	1.000	1.000	1.000	1.00	<0.001	8
	Subgroups based on pollutant lag (fixed effects models)
	Same day or prior two days	1.000	0.999	1.001	1.00	<0.001	8
	Subgroups based on analytical approach (fixed effects models)
	Use of a two-stage approach	1.005	1.003	1.007	1.28	<0.001	3
	Use of a population daily average outcome	1.000	0.999	1.001	1.00	<0.001	9
	Autocorrelation modelled using GEE	1.005	1.004	1.007	1.28	<0.001	6
	Autocorrelation not modelled using GEE	1.000	0.999	1.001	1.00	<0.001	6