Assessment of potential effects of exposure to disinfection by-products on the risk of specific birth defects is problematic because of rarity and diversity of the congenital malformations. Few previous studies [10–13] have focused on selected birth defects and only two have assessed the associations between exposure and the risk of all the most common specific birth defects [12, 13].
Our results showed no consistent association between exposure and the risk of birth defects in general. Of the 11 specific birth defects studied, the risk of ventricular septal defects, cleft palate, and anencephalus were substantially elevated in the high exposure compared to the reference category. The meta-analysis of 7 birth defects together with the Norwegian, California and British data strengthened the evidence for ventricular septal defects.
Validity of results
The present study had enough power to estimate the relations between exposure to disinfection by-products and the most common specific birth defects, which could be relevant adverse effects based on previous literature. The meta-analysis together with the Norwegian, Californian and British results improved the precision of the finding on ventral septal defects. The specific birth defects are expected to be also more homogenous as to their causal factors compared with birth defect groups. We excluded half of the births because of insufficient water quality data. This exclusion was unlikely to introduce selection bias, because it was made on municipality level and the characteristic of the excluded individuals did not differ substantially from the included, as shown in Table 4.
Our outcome assessment was based on birth registration, as in the vast majority of the previous studies of disinfection by-products and birth defects [5–8, 10, 12–14], rather than clinical examination for the purposes of the study. This is a source of misclassification, which is likely to be random, i.e. not related to the exposure of interest, and thus lead to underestimation of the effect estimates. The sources of misclassification could include variation in diagnostic criteria and errors in reporting information provided by physician or hospital. Important features in the Taiwan national health care system limit the amount of outcome misclassification. Taiwanese pregnant women are almost all covered by health insurance (>99%,) and access to prenatal care is free and good (at least 10 visits during pregnancy). The clinical surveillance of pregnancy begins at 1 month after conception and continues through 7 days after birth. Birth defects were mostly diagnosed by physician, most often by paediatrician using ultrasound. According to the law, all live births in Taiwan must be reported within 15 days after birth. In general, the birth defects might be underreported, because we did not include the birth defects that were diagnosed after 7 days of age, gestational age less than 20 weeks, and induced abortions due to birth defects. This underreporting was likely to be nondifferential, i.e. not related to exposure. In most situations non-differential misclassification of a binary outcome will produce bias toward the null, provided that the misclassification is independent of other errors .
A major challenge of this study was the imprecision of exposure assessment from using aggregate municipal measures for classifying individual exposures. We had no information on the amounts of beverage and tap water consumption by pregnant women and exposure to volatile disinfection by-products through inhalation and dermal absorption, which might introduce non-differential misclassification and decrease the accuracy of exposure assessment. Future exposure assessment should include exposure through multiple routes such as bathing, showering and swimming, as well as water consumption. Unfortunately, we do not have sufficient information on alcohol consumption, cigarette smoking, vitamin consumption, medication, and genetic factors. Adjustment for population density adjusted indirectly for municipal differences in these behavior factors, but also eliminated partly regional differences in TTHM levels between rural and urban. Residential mobility during pregnancy may be an issue in our study. Two studies conducted in the United States showed that 25%  and 37%  of women moved during pregnancy. We have no information on the change of residence during pregnancy between three exposure categories (TTHMs 20+ μg/L; TTHMs 10–19 μg/L; TTHMs 5–9 μg/L) and the reference category (TTHMs (0–4 μg/L) in Taiwan birth registration. Since the misclassification of exposure is likely to be non-differential related to the outcomes of interest, the level of such misclassification would likely bias the effect estimations toward to the null. We systematically carried out stratified analyses in different categories of exposure and other covariates to elaborate the residual confounding and potential effect modification. The stratified analyses did not indicate any major confounding or effect modification.
The interpretation of meta-analysis is more difficult when the specific effect estimates differ substantially from each other, especially if there are estimates both below and above unity. The random-effects model has become a standard approach to incorporate heterogeneity. We elaborated the heterogeneity between the specific effect estimates, but presented systematically summary estimates from both fixed and random-effects models to offer readers a possibility for their own informed judgment. We also made an attempt to explain the observed heterogeneity, although the small number of studies limited the applicability of this approach. The type of study, study populations, and outcome assessment were relatively similar between the summary studies, but there were different approaches to exposure assessment. The present and three previous studies using routine measurements of trihalomethanes [11, 13] constituted a rather homogeneous group, but different exposure contrasting. Also the Swedish study used water source and chlorination practice as the basis for exposure assessment , and the Norwegian study was based on the amount of organic content expressed in color and the presence of chlorination, it was possible to use a similar contrast between the highest exposure category and the reference group of no exposure . Additionally, the sensitivity analysis indicated that the largest studies had a substantial impact on the summary effect. .
Synthesis with previous knowledge
Six [5–8, 11, 22] of the eight previous studies [5–8, 10–12, 22] have provided evidence of an increase in the risk of neural tube defects related to exposure to disinfection by-products. Two large case-control studies in California focusing on ancephalus and spina bifida did not provide a clear pattern of the relation of exposure to THMs . A case-control study in New Jersey reported no significant association between exposure to THMs and the risk of spina bifida . Interestingly in the present study the risk of anencephalus was elevated in the highest exposure category. The meta-analysis of the four available studies provided an inconclusive, heterogeneous summary estimate.
The previous findings on cardiac defects as a group have been heterogeneous and inconsistent [5, 6, 8, 10, 12, 13]. In the present study, the risk of ventricular septal defects was almost two times higher in the high exposure category compared to the reference category (adjusted OR 1.81, 95% CI: 0.98, 3.35). This is consistent with the Norwegian study, which reported an exposure-related increase of ventricular septal defects (medium exposure: adjusted OR 1.63, 95% CI: 1.02, 2.58; high exposure 1.81, 1.05–3.09), and the British study (high exposure vs. reference: adjusted OR 1.43, 95% CI: 1.00, 2.04). The corresponding summary OR was 1.59 (95% CI: 1.21, 2.07). Recently the risk of ventricular septal defects has also been found to be related to the level of traffic-related air pollution .
Previous five studies [5, 6, 8, 11–13] on cleft lip and palate defects have given heterogeneous and/or inconsistent results. In the present study, the risk of cleft palate was related to the levels of TTHMs with an exposure-response pattern, yielding adjusted OR of 1.17 (95% CI: 0.74, 1.86) for medium and 1.56 (95% CI: 1.00, 2.41) for highest exposure category. Similar results were reported also from the Norwegian and California studies [9, 11], but the British study resulted in an adjusted OR of 0.95 (95% CI: 0.87, 1.10) . The four studies gave the summary odds ratio of 1.00 (95% CI: 0.87, 1.15).
Previous studies conducted in Massachusetts  and Norway [8, 12] have provided rather homogeneous and consistent evidence of an effect on urinary tract defects as a group. In the present and British  studies there were no association between exposure and urinary tract defects per se, the risks of a smaller group denoted as obstructive urinary tract defects (ICD-9: 753.2) and renal agenesis and dysgenesis (ICD-9: 753.0) were found to be slightly elevated both in the high and low exposure categories. The Norwegian study showed some evidence of an increased risk of urinary tract defects as a group and an exposure response pattern for the risk of obstructive urinary tract defects (ICD-8: 753.2) . Urinary tract defects are rare and the effect estimate imprecise, and therefore the findings of the role of exposure to disinfection by-products remain inconclusive. A negative association was found for hypospadias. This could be explained by chance while unity was within the 95% confidence interval. Unfortunately the diagnosis of hypospadias may also be compromised because it is made within the 7 days from delivery. Therefor the results of hypospadias should be interpreted cautiously.
Our study and two recent studies conducted in Norway and in England and Wales suggest that prenatal exposure to disinfection by-products increases the risk of ventricular septal defects at much lower levels than found in United States [5, 7], and Canadian  drinking water sources, probably explained by qualitative geographic differences in the levels of natural organic matter (disinfection by-products precursor) or higher concentration of other non-volatile disinfection by-products (eg. haloacetic acids). The present and the two previous studies from Norway and California suggest also an increased risk of cleft palate.
The specific mechanisms for the effects of trihalomethanes (THMs) on the risk birth defects are still unknown. Some animal studies show reproductive and developmental toxicity of some of these compounds, such as chloroform (CF) and bromodichloromethanes (BDCM), when given at high doses . There is evidence that metabolites of chloroform may accumulate in the amniotic fluid of pregnant mice . In addition, BDCM can disrupt syncytiotrophoblast formation and inhibit chorionic gonadotrophin secretion in vitro . This implies that the placenta is a likely target of BDCM toxicity in human and thus BDCM may have teratogenic effects on fetus.
An alternative explanation is that THMs may lead to birth defects via genetic damage to maternal gametes. For example, CF may be oxidatively metabolized and decomposed to electrophilic phosgene, which is more likely to react and bind to cell components including proteins, phospholipid polar heads, and reduced glutathione . This may result in chromosomal abnormalities, enzymatic malfunction, and disruption of cellular membranes, all of which could interfere with uterine development or directly influence on the conceptus.
Although THMs are the most prevalent in chlorinated water, they are also markers of a complex mixture of disinfections by-products. Some animal studies also show reproductive and developmental toxicity of haloacetic acid, non-volotile disinfection by-products, such as dichloroacetic acid (DCAA) and trichloracetic acid (TCAA) . Further detailed toxicological assessments of mixtures of chlorination by-products are also needed, as humans are most commonly exposed to complex mixtures of these compounds rather than to a single compound .