Inorganic Arsenic in Drinking Water and Bladder Cancer: A Meta-Analysis for Dose-Response Assessment

In our previous paper, “Inorganic Arsenic in Drinking Water and Bladder Cancer: A Meta-Analysis for Dose-Response Assessment”, 2006, 3(4), 316–322, there were several errors in the table of data used in the analysis. In particular:

• The paper of Bates et al. [1] incorrectly listed units of concentrations. They reported in units of milligrams rather than micrograms (see the last entries in Table 3 of their paper).
• In the paper by Chiou et al. [2] we introduced an error ourselves. We listed the arsenic exposure level as ≤ 50; 50–70; 71+. These should be ≤ 50; 50–700; 710+.

With these corrections, the pooled estimate of slopes from the seven studies using the fixed effects model becomes was 0.001 (95% CI: 0.001, 0.002), with the unit of lnRR per unit increase of exposure (exposure is in μg/L as in our original paper). The chi-square statistic was quite large (i.e. Q = 497.752 on 6 degrees of freedom, p= 0.00), which rejects the null hypothesis of homogeneity and means there was evidence of heterogeneity. Using the random-effect model, and including only the five studies identified in the original paper as most relevant (excluding Bates et al [1] and Kurttio et al [3]), the pooled estimate of the slopes from the five studies was found to be 0.002 (exposure also in units of per μg/L) (95% CI: −0.001, 0.006).

The new result of the meta-analysis still supports the claim that there is a positive dose-response relationship between exposure to arsenic in drinking water and bladder cancer.

Table 1:. Risk of bladder cancer at different MCLs

MCL (ppb)	AR (u_95)	AR (Mean)	AR (L_95)
0	0	0	0
1	−1.80E-07	1.08E-06	−1.80E-07
3	−5.39E-07	3.27E-06	−5.39E-07
5	−8.98E-07	5.48E-06	−8.98E-07
10	−1.79E-06	1.11E-05	−1.79E-06
20	−3.56E-06	2.29E-05	−3.56E-06
50	−8.78E-06	6.30E-05	−8.78E-06

summarizes the revised results of the absolute risk (AR) calculation for bladder cancer associated with a variety of proposed MCLs (maximum contaminant levels) using different estimates from the meta-analysis: the best estimate, the upper-bound and lower-bound estimates of the slope factor. The best (revised) estimate of the slope factor from the meta-analysis is 1.64 × 10⁻⁵ (with unit of probability per μg/kg/day), with the upper bound of 5.38 × 10⁻⁵. These slope factors from the meta-analysis are lower than the ones from the EPA (1.5 × 10⁻³) and NRC (8.85 × 10⁻⁴).

If readers would like the revised figures and tables from the paper, please contact the corresponding author, at the above-referenced address.