*3.2. D-Optimal*

Analysis of variance is widely used to predict the suitability of a model with experiment results. The obtained results (Table 2) indicated that the predicted values of the model were not conflict with the experiments. The coefficient of determination of *R*<sup>2</sup> was 0.932 and the coefficient of determination adjustment *R*2adj was 0.910. The suitability of the model was also shown in *P* values and Fisher test. *<sup>P</sup>*regression value was 0.000 (<0.05), and *P*Lack of fit was 0.221 (>0.005), which showed that the obtained model was consistent with the experiment.


**Table 2.** Analysis of variance (ANOVA).

Note: degrees of freedom, DF; sum of squares, SS; mean square, MS; Fisher, F; probability value, P; and standard deviation, SD.

The three-dimensional response surface shows the effect and interaction of the two factors on the target function. Figure 3a shows the combined effect of the Vsolvent/Vsample ratio and NaCl concentration. Figure 3b shows the image effect of NaCl concentration and time of extraction. Interaction between the Vsolvent/Vsample ratio and time of extraction is shown in Figure 3c. In general, when the value of the variables increases, the efficiency of the phthalate extraction rises and eventually reaches equilibrium.

The contribution of these factors on the extraction efficiency is shown in Figure 3d. The solvent/ sample ratio was the biggest influence (54.8%), followed by extraction time (35.8%), and final NaCl concentration (9.4%).

The optimal tool of MODDE 12.1 software was used for the optimization. The results are shown in Table 3. Experimental result was obtained at optimum conditions, yield was 92.51 (95% confidence). This proves that the model was highly meaningful, allowing good experimental results.


**Table 3.** Optimization of phthalate extraction process.

**Figure 3.** Response surface plots for the central composite design (CCD). Note: (**a**) Vsolvent/Vsample ratio vs. NaCl concentration; (**b**) NaCl concentration vs. time of extraction; (**c**) Vsolvent/Vsample ratio vs. time of extraction and (**d**) The phthalate extraction rises vs. eventually reaches equilibrium.

#### *3.3. Method Performance*

To evaluate the efficiency of the proposed method, a number of parameters of the method were investigated and manifested in Tables 4–6. The triple quadrupole detector provided a high degree of selectivity. The linear ranges of these phthalate compounds were built up from 1 to 200 μg/L. Additionally, the weights 1/x<sup>2</sup> shown through the correlation coefficient (*r*2) were greater than 0.996, a non-significant lack of fit and individual residuals deviation of <13% proved the quality of the method. The lowest LOD of these phthalate substances was 0.5 ng/L and the highest was 3.0 ng/L. The maximum of retention time was ±0.06 min, which was below the maximum tolerance deviation stated in SANTE guidelines (±0.1 min). The repeatability (RSDr) and within-laboratory reproducibility (RSDwr), which expressed percent relative standard deviation (%RSD), ranged from 1.0 to 9.1% and from 2.7 to 12.3%, respectively. All detected RSD values were smaller than 15% that meet the SANTE guideline of RSD ≤ 20%. The trueness of this method was appraised through the recovery values by adding standard solution to carbonated beverage and fat beverage samples at three different concentrations. The average recoveries of 10 phthalate compounds are demonstrated in Table 3 and are within the range required by the SANTE guidelines (between 70% and 120%).

**Table 4.** Linear dynamic range (μg/L), determination coefficients (*r*2), residuals, retention times, limit of detection (LOD) and limit quantitation (LOQ).



**Table 5.** Repeatability (RSDr) and within-laboratory reproducibility (RSDwr) for peak areas evaluated at three concentration levels.

data is presented as % RSD.

**Table 6.** Trueness results for 10 phthalate compounds in non-alcoholic beverages matrices.


Note: M: Mean (% recovery); R: Relative standard deviation (%).

#### *3.4. Levels of Phthalates in Samples*

Non-alcoholic drink samples were analyzed based on the above sample preparation method. The results are shown in Table 7 and Figure 4. As described in Table 4, DBP and DEHP were also detected in all of the 148 collected beverage samples which were analyzed, while DnOP were found in 33% of samples. The appearance of phthalate compounds ranged from 1% to 100% so that almost all of the samples were contaminated by phthalates. As can be seen in Figure 4, DEHP was the phthalate which primarily presents in the samples (>35%), followed by DBP and DEP in mineral water, fruit juice, tea, fermented milk and functional drink. Conversely, in carbonated drink samples, DnOP was the most abundant phthalate substance (>50%). In relation to fermented milk, DMP and DEHP were comparatively in the same proportion (45.5 and 47.6%). It was recognizable that there was the extensive appearance of DMP, DnOP and DBP. DnHP virtually did not appear in these kinds of beverage drinks. DnOP was chiefly found in carbonated drink samples (54.3%). Additionally, DMP was mainly detected in fermented milk samples (45.5%).

The concentrations of phthalate compounds presenting in non-alcoholic beverages are also illustrated in Table 8. In 6 groups of experimental samples, DEHP was the phthalate substance containing the highest mean as well as medium value among all of the samples. The mean and medium were 91.6 and 64.5 μg/L. The variation of concentrations of 148 samples ranged from 0.092 to 466.6 μg/L, which were much higher than those of DBP (22.1 and 18.8 μg/L, the variation varies from 0.093 to 73.5 μg/L). The DMP, BzBP, DPP. DiBP, DnOP and DCHP contents were nd–131.9 μg/L, 0.30–21.5 μg/L, nd –0.52 μg/L, nd–1.9 μg/L, nd–200.4 μg/L and nd–0.60 μg/L, respectively.


**Table 7.** Detection of frequency [n (%)] of phthalates in non-alcoholic beverages in Hanoi.

**Figure 4.** The distribution of 10 phthalate compounds in different kinds of samples. Note: Dimethyl phthalate, DMP; Diethyl phthalate, DEP; Dipropyl phthalate, DPP; Diisobutyl phthalate, DiDP; Benzyl butyl phthalate, BzBP; di-*n*-hexyl phthalate, DnHP; di (2-ethylhexyl) phthalate, DEHP; di-*n*-octyl phthalate, DnOP; Dicyclohexyl phthalate, DCHP and di-*n*-butyl phthalate, DPBP.

Table 8 also shows that the concentrations of phthalate compounds studied on different targets were considerably different. The DEHP in fruit juice samples had the highest mean (230.8 μg/L) and median (222.7 μg/L) among other types of beverage drinks in this experiment. Moreover, the mean and median of DEP (17.9 and 17.3 μg/L) in fruit juice drink were also far higher than other beverages. In terms of fermented milk, DMP content was detected in a range of 12.3 to 131.9 μg/L, and the average and median were 68.0 and 65.7 μg/L, respectively.


**Table 8.** Phthalate concentrations in different types of non-alcoholic beverages (μg/L).

The distribution of the total phthalate concentration in non-alcoholic drinks was different among the sample matrices. Juice drinks had the highest phthalate concentration, followed by fermented milk and tea. As shown in Figure 5, DEHP was a major contributor leading to the phthalate contamination in non-alcoholic beverage, similar to the previous study [21]. The contamination of phthalates depended on the characteristics of the samples. The sample containing preservatives (potassium benzoate) had higher phthalate concentration than that which did not use preservatives [22]. Furthermore, the sample carrying high fat content was easier to contaminate by phthalate [23]. When comparing the data above, the identification of the sources of phthalate contamination was ambiguous because of other factors such as temperature, pH, light, turbidity and storage time [5,24,25].

Among all phthalates, DEHP is the most popular substance appearing in non-alcoholic beverages in similar studies. Figure 6 illustrates the degree of DEHP contamination in recent studies and the container of these products is not necessarily made from plastic. According to the research of Ustun et al., DEHP concentrations of soda, lemonade, mineral water and high-taste water in Turkey ranged from 73 to 2312 ng/g and the highest DEHP concentration was found in Cola soft drink [21]. In contrast, based on the study of Sireli et al., DEHP concentration in fruit juice drink varies from 1.1 to 44.3 ng/g, which is much lower than Ustun' research [26]. Wu et al. reported that the DEHP content in energy drink and tea ranged from 15 to 83 ng/g [27]. DEHP concentration was remarkably high in the study of Truong et al. of chocolate and high-fat drinks (111–1753 ng/g) [23]. In our research, DEHP concentration varied from 0.1 to 466.6 ng/g, remained within the range of the above studies and predominantly concentrated in milk-containing fruit juice sample.

**Figure 5.** The distribution of the total phthalate concentration in non-alcoholic drinks.

**Figure 6.** Concentrations of DEHP in similar studies.

#### *3.5. Exposure to Phthalates*

Assessing phthalate concentration in non-alcoholic beverages has been investigated by many researchers around the world. However, in Vietnam, there are no specific statistics on phthalate content in daily beverage drinks. Identification of the existence as well as frequency of the occurrence of phthalate compounds in the matrices totally depends on instrument detection limit (IDL) and method detection limit (MDL) of the study, but comparison of phthalate contamination in non-alcoholic beverages still has scientific meaning.

Relying on the studies of Guo [28] and Sireli [26], we calculated the daily intake of DEP, DBP, BzBP and DEHP in Vietnam following the formula below:

$$\text{EDI} = \frac{\text{CQ}}{\text{bw}} \text{r}\_{\text{uptake}} \tag{2}$$

where EDI (μg/kg × day) is the estimated daily intake from drinking beverages, C (ng/g) is the phthalate concentration in beverages, r is the gastrointestinal uptake factor and bw (kg) is the body weight. In this study, average beverages intake was 150 g/day, ruptake was 1 and an average bw of 50 kg was used for Vietnam population. The result is shown in the Table 9.


**Table 9.** Characteristics of the investigated phthalates.

The daily intake of DEHP when investigating phthalates in beverages in Vietnam was higher than TDI (U.S. EPA), but the contamination of DEP, DBP and BzBP was significantly lower than the threshold of regulation. The phthalate concentration in non-alcoholic beverages did not give rise to serious consequences for adult health. However, the beverages such as fruit juice and fermented milk, which were analyzed, are consumed daily by pregnan<sup>t</sup> women. Because of this, there is likely to be a mother-to-child exposure through the placenta [29] leading to the phenomenon of hormonal disturbance in children [30].
