Previous epidemiological studies in Massachusetts have reported a risk of adverse health outcomes in relation to disinfection by-product (DBP) exposures. Measurement error due to the use of indirect exposure surrogates can lead to misclassification bias in epidemiological studies; therefore, it is important to characterize exposure variability in these populations to assess the potential for exposure misclassification. We used 19,944 trihalomethane (THM) samples and 9291 haloacetic acid (HAA) samples collected in 201 public water systems (PWSs) in Massachusetts to examine temporal variability under different drinking water sources and disinfection types. Annual and seasonal variability was also examined in 46 PWSs with complete quarterly THM4 (i.e., the sum of 4 individual THMs) data from 1995 to 2004 and 19 PWSs with complete HAA5 (i.e., the sum of 5 individual HAAs) data from 2001 to 2004. The quarterly ratio of THM4 and HAA5 and correlations between THM4, HAA5 and individual DBP species were examined to determine the adequacy of using different exposure surrogates in epidemiological studies. Individual PWSs were used to examine monthly variability in relation to quarterly averages. Based on all available matched samples (n=9003) from 1995 to 2004 data, we found a correlation of 0.52 for THM4 and HAA5. The correlation was stronger among the 62 ground water systems (r(s)=0.62) compared to the 81 surface water (r(s)=0.45) and 40 mixed water (r(s)=0.39) systems. Mean THM4 levels were fairly stable over the 10-year study period for 46 PWSs including 39 PWSs that did not change disinfection. Large reductions (∼40 μg/L) in mean THM4 data were found among seven systems that switched from chlorination to alternative disinfectants. As expected, the highest mean THM4 values were detected for Quarter 3, while the lowest values were found in Quarter 1. The highest HAA5 values were detected in Quarters 2 and 3 and the lowest was found in Quarter 4. Data from four systems showed mean differences up to 66 μg/L (67% change) in successive months and by 46 μg/L compared to quarterly mean concentrations. Although longer-term disinfection by-product temporality may be minimal in this study population, the use of monthly average concentrations for exposure assessment may be needed for some PWSs to minimize misclassification of narrow critical periods of exposure in epidemiological studies.
Published by Elsevier Inc.