Introduction

The association between low intelligence quotient (IQ) scores in children and high level of fluoride in drinking water has been reported [1]. However, how strong the association is is unknown. In China, endemic fluoride poisoning is serious, and there are thousands of people and children live in endemic fluorosis areas. The objective of this paper is to present a meta-analysis to answer the following question: Is there truly an association between low IQ and fluoride in total 20 years? How strong is the association?

To assess the strength of the association between the level of fluoride in water and IQ, weighted mean difference (WMD) is used. The magnitude of WMD determines the strength of an association between fluoride levels and low IQ. In a meta-analysis, a systematic review of the literature is conducted, and studies are included or excluded from the analysis depending on whether they meet the eligibility criteria for inclusion set by the reviewers. The data are combined statistically to provide a quantitative estimate of the size of effect of exposure and risk of disease [2].

Materials and Methods

Literature Search

MEDLINE, SCI, and CNKI search were organized for all studies published, in English and Chinese, between 1988 and 2008. The following keywords were used to search for all documents: fluorosis, fluoride, intelligence, or IQ. A further search was undertaken in the website www.fluorideresearch.org because this is a professional website concerning research on fluoride. The country was restricted to China. As a result, 18 case–control studies were located with enough data to allow for further quantitative analysis (Table 1).

Table 1 Risk of Low IQ in Fluorosis Area: Case–control Studies
Full size table

A total of 26 studies were excluded. Some excluded studies did not present enough data to allow inclusion in the meta-analysis. Another reason for exclusion was some of the papers had been included, which means that they had been published more than one time or had been published in two journals in Chinese or English language.

Meta-analysis

The analysis included 16 case–control studies. The summary WMD and 95% confidence intervals (CI) were obtained using the following two statistical methods (Table 2):

  1. 1)

    Mantel–Haenszel method: estimates the summary WMD and its standard error

  2. 2)

    DerSimonian–Laird method: estimates the summary WMD and its standard error. No test of homogeneity is available for this method

Table 2 Summary WMD
Full size table

The Mantel–Haenszel should be used in a fixed-effect model, while the DerSimonian–Laird methods should be used in a random-effect model. Tests of homogeneity of variances for the Mantel–Haenszel were performed as described by Petitti [2].

Sensitivity Analysis

Sensitivity analysis was used to assess how robust the results are to uncertain decisions or assumption about the data and the methods that were used [3] (Table 3).

Table 3 Sensitivity Analysis of This Present Study
Full size table

Results

Case–Control Studies

These studies are listed in chronological order in Table 1. Children expose to fluoride had a significant increase in the risk of developing low IQ. Twelve studies found statistically positive associations between the exposure to fluoride and low IQ (Table 1). The risk of developing low IQ ranged from a low of 0.2 to a high of 10.78 [1, 4–18].

Meta-analysis

The Mantel–Haenszel method estimated that the summary WMD for the association between exposure to fluoride and low IQ is about −4.97 (95% CI = −5.58 to −4.36; p < 0.01). The DerSimonian–Laird gave a summary WMD of −5.03 (−6.51 to −3.55; p < 0.01; Table 2). Because of the lack of homogeneity of variances among the studies, the results of the Mantel–Haenszel method may be invalid. The random-effect method that is the basis for the DerSimonian–Laird algorithm is preferred in this case. Overall, the meta-analyses found that the summary WMD on average is about 5, indicating that children who live in a fluorosis area have five times higher odds of developing low IQ than those who live in a nonfluorosis area or a slight fluorosis area.

Sensitivity Analysis

To analyze the sensitivity of our study, two studies were excluded because the samples of the two studies were large and thus may increase the heterogeneity: Li’s paper, the samples of which were 720 for the fluorosis area and 236 for the control, and Chen’s paper, the samples of which were both 320 for the fluorosis area and the control (Table 3). The results of the sensitivity analysis show that with either Li’s or Chen’s study excluded, the summary WMD and 95% CI were still near to the results before they were excluded. Further, when the two studies were excluded at the same time, the results were also stable. Hence, the results of the sensitivity analysis indicate that the results of our study are reliable and believable.

Funnel Plot for Bias Analysis

Figure 1 is the funnel plot of our study, which shows the bias of our study. From the figure, we can see that there are biases in our study. Publication bias and language bias may be the main causes.

Fig. 1
figure 1

Funnel plot of the present study, which shows that there are some biases (obtained by Review Manager 4.2.2 Software)

Full size image

Discussion

The mechanism of the action of fluoride in reducing IQ is not clear. Guan et al. [19] demonstrated that the contents of phospholipids and ubiquinone are altered in the brain of rats affected by chronic fluorosis, and therefore changes in membrane lipids could be a cause of this disorder. Furthermore, thyroid hormones play an important role in development of brain and thus might also affect IQ level [20]. However, the strength between fluoride and low IQ is still unknown.

We investigated this study to summarize the 20-year studies concerning the association between fluoride and low IQ in China and to try to find if there is truly an association between them in total 20 years and how strong the association is. Our findings show that children who live in a fluorosis area have five times higher odds of developing low IQ than those who live in a nonfluorosis area or a slight fluorosis area. Hence, further action should be undertaken to reduce the fluoride level in the water of a fluorosis area, which is also urgent and necessary.