Table 2

Pooled risk estimates for non-Hodgkin’s lymphoma by exposure assessment method, study design, publication year period and geographic region, based on meta-analysis of articles on occupational pesticide exposure published between 1987 and 2017.

	Number of risk estimates	sRR	95% CI	Heterogeneity measures
	Number of risk estimates	sRR	95% CI	I² (%)	P value	Q	P value
Exposure assessment method						6.23	0.07
Group-level	17	1.21	1.05 to 1.40	63	<0.01
Self-reported exposure	13	1.49	1.16 to 1.91	76	<0.01
Expert-level	10	1.74	1.39 to 2.19	0	0.68
Study design						22.1	<0.01
Cohort (prospective)	8	1.04	0.96 to 1.13	23	0.24
Cohort (retrospective)	4	1.11	0.89 to 1.39	11	0.34
Case–control	28	1.66	1.39 to 1.98	57	<0.01
Publication year period
1987–2005	19	1.15	1.00 to 1.32	21	0.2	8.5	<0.01
2006–2017	21	1.59	1.34 to 1.87	78	<0.01
Geographic region						3.89	0.14
Europe	18	1.42	1.13 to 1.77	55	<0.01
North America	18	1.27	1.10 to 1.47	70	<0.01
Other	4	1.77	1.31 to 2.39	38	0.18
Case–control studies only
Exposure assessment method						0.1	0.95
Group-level	9	1.63	1.20 to 2.21	61	<0.01
Self-reported exposure	11	1.67	1.21 to 2.31	71	<0.01
Expert-level	8	1.73	1.33 to 2.27	0	0.47
Exposure assessment method during publication year periods
1987–2005
Exposure assessment method						2.30	0.32
Group-level	8	1.04	0.83 to 1.28	31	0.18
Self-reported exposure	8	1.26	1.05 to 1.51	0	0.51
Expert-level	3	1.37	0.86 to 2.17	0	0.51
2006–2017
Exposure assessment method						4.68	0.1
Group-level	9	1.35	1.11 to 1.64	74	<0.01
Self-reported exposure	5	1.94	1.14 to 3.30	91	<0.01
Expert-level	7	1.88	1.45 to 2.24	0	0.7

I²=percentage of variation across studies due to heterogeneity. Q=Cochran’s Q.
sRR, summary risk ratio.