Area, population, and study design
Guadeloupe comprises five islands, the most important of which are Grande-Terre and Basse-Terre, separated by a narrow sea channel. The banana plantations and soils polluted by chlordecone are mostly located in the southern part of Basse-Terre. This analysis was conducted in the ongoing TIMOUN cohort study. Enrolled subjects were representative of pregnant women giving birth in the local public hospitals (one in Grande-Terre and one in Basse-Terre), which account for 70% of all deliveries in Guadeloupe. Obstetricians enrolled women in this cohort during the second trimester prenatal visit, from December 2004 through December 2007. At enrollment, they provided informed consent and completed, during face-to-face interview with midwives, a standardized questionnaire covering sociodemographic characteristics, medical and obstetrical history, and lifestyle. This questionnaire also included 6 items about prepregnancy frequency of consumption of foods known to be contaminated by chlordecone (fish and root vegetables (AFSSA, 2007)). Women included in the present analysis were the first 194 for whom both food consumption data and blood chlordecone determinations were available.
2.2. Collection of data about food consumption habits
During the 3–5 days of hospitalization following delivery, a semiquantitative food frequency questionnaire (SQFFQ), asking about usual dietary intake during this pregnancy, was administered to the women by trained interviewers.
The SQFFQ was developed from a questionnaire covering 105 food items used in two previous general population studies in Martinique and Guadeloupe (Dubuisson et al., 2007). This initial questionnaire was intended to identify consumption of potentially contaminated food only and was thus completed by 109 other food and beverage items, including tap and bottled water (totaling 214) to provide an exhaustive assessment of maternal diet and nutrient intake (Willett, 1998b).
Categories for frequency of consumption were: never, less than once a month, 1 to 3 times a month, once a week, 2 or 3 times a week, 4 or 5 times a week, once a day, and more than once a day. Responses were further translated into the corresponding median number of meals a month for each item. Portion sizes were evaluated with the help of a booklet of pictures for the principal 11 food categories: raw vegetables, cured meats, meat, fish, root vegetables, other cooked vegetables, French fries, starches (pasta, rice, and semolina), bread, cheese, and desserts. Pictures and portions for these categories were extracted and adapted from a booklet used in the French SU.VI.MAX survey (Hercberg et al., 1994).
Reported energy intake was calculated by matching food consumption data and food composition data, derived from a French table (Hercberg, 2005) and completed with the Canadian (Health Canada, 2007) and the US (USDA, 2007) food composition tables. Subjects eating fewer than 800 kcal or more than 6000 kcal per day (n=3) were excluded because their SQFFQ responses were considered unreliable ( Willett, 1998b).
2.3. Dietary chlordecone intake assessment
Data about chlordecone levels in food come from a survey conducted in Guadeloupe from July 2006 to July 2007 (AFSSA, 2007). This survey was designed to assess the dietary intake of chlordecone in the general population older than 3 years, according to the WHO guidelines (FAO/WHO, 1997 and FAO/WHO, 2000). Factors considered in planning the types and numbers of samples to collect included the known contribution of foods to dietary chlordecone intake and the importance in the Caribbean diet of food items for which chlordecone contamination is not known. Local retailers and markets were divided into two strata based on their location in a polluted or an unpolluted area. The sampling places were drawn according to the food supply habits of the population, as described in the food consumption survey, with the same rate for each stratum (AFSSA, 2007). Self-production was also taken into account. European Union (directive 2002/63/EC) guidelines establishing community methods of sampling for the official control of pesticide residues in and on products of plant and animal origin were applied to collect the 744 food samples. Analytical methods and quality control have been described elsewhere (Bordet et al., 2007). Data were expressed as mean concentrations (μg/kg of fresh product) stratified according to the area where they were sampled (polluted or unpolluted areas of the island). The relations between the 59 items for which contamination data were available and items in the SQFFQ were determined to help attribute chlordecone values to the latter. When no information was available, assumptions were based on the same principles as those defined in previous dietary intake assessment studies (AFSSA, 2007).
Daily dietary intake of chlordecone was assessed for each subject according to a previously described procedure (AFSSA, 2007). The quantity of each food (or beverage) item eaten daily was multiplied by its mean estimated chlordecone level. Estimated quantities of chlordecone intake from all 214 items were then summed to obtain a total daily intake (μg/day). We designate this as the food consumption–food contamination approach (FCFC).
2.4. Other covariates
The following information was obtained from a questionnaire administered at enrollment: maternal age, place of birth (Guadeloupe, Martinique, Haiti, Dominica, continental France), town of current residence, marital status, educational level (none, primary, secondary school, high school diploma, higher education), employment status before pregnancy, weight and height before pregnancy, obstetrical history, previous breast-feeding, and smoking status. Body mass index was calculated as weight (kg)/(height (m))2 and classified as underweight (<18.5 kg/m2), normal (18.5–<25 kg/m2), overweight (25–<30 kg/m2), or obese (≥30 kg/m2). Risk of residential contamination was determined by matching women’s town of current residence with the most recent map of soil pollution risk in Guadeloupe (Cabidoche et al., 2006), drawn by matching land plots to past and present maps of banana plants. The risks thus defined were further validated by soil analyzes.
2.5. Chlordecone blood determination
At delivery, maternal blood samples were collected in EDTA tubes. After centrifugation, they were transferred to polypropylene Nunc® tubes and stored at −30 °C. Blood samples were transferred in dry ice to the Laboratory of Animal Ecology and Ecotoxicology (Liège, Belgium) which is a member of the Center for Trace Residue Analysis (CART) for determination of chlordecone concentrations.
Blood samples were prepared by adaptation of a previously described method (Debier et al., 2003). Briefly, blood was first treated by adding 100 μL of triethylamine and 10 mL of formic acid to a precisely known volume of sample (from 2 to 3 mL, depending on the amount of sample initially available). Solid phase extraction used Supelco Supelclean™ Envi-18 SPE Tubes (Supelco, Bellefonte, PA, USA). An elution device (Visiprep DL™ SPE Supelco) was used to condition the column with 10 mL methanol, followed by 10 mL of distilled water. The deproteinized sample was added to the column and rinsed with distilled water and dried under vacuum. After a first elution by 5 mL n-hexane, a second elution was performed, with the addition of 5 mL diethylether/n-hexane (85%/15%, v- v) and 50 μL of PCB 112 (100 pg/μL), used as a surrogate marker. Then the sample was evaporated and purified by concentrated sulfuric acid. The purified extracts were evaporated under a gentle nitrogen stream to a final volume of 100 μL. The extracts were analyzed by high-resolution gas chromatography, with a Thermo Quest Trace 2000 gas chromatograph equipped with a Ni63 ECD detector (Thermo Quest, Milan, Italy). Chlordecone was separated on a 30 m×0.25 mm (0.25 μm film) DB-XLB capillary column (J&W Scientific, Santa Clara, CA, USA). The linear calibration curve, with concentrations ranging from 5 to 120 pg/μL, was established with a certified chlordecone solution (Riedel-de Haën). A blank consisting of 2 mL of distilled water was run with each series of 10 samples to control the cleanup procedure. A quality control sample of human blood enriched with a defined concentration of chlordecone was run and analyzed simultaneously. Recovery was calculated on the basis of the concentration of surrogate marker PCB 112 (50 pg/μL). The results of the analyzes were accepted only for recoveries between 70% and 130%. Due to the high remanence of chlordecone molecules on the analytical column, a calculation strategy had to be used. With each series analyzed, n-hexane samples were injected before and after the blank sample and another hexane was injected after the injection of 5 samples. In those hexanes a peak corresponding to the chlordecone released by the column was found and processed as a sample. The average value found in those hexane samples was considered as the average remanence of chlordecone in the series of concern. This value was then subtracted from the chlordecone concentration value of each sample in order to better avoid the chlordecone remanence. In these conditions, the limit of detection (LOD) was fixed at triple the background noise of the chromatogram, which corresponds to the hexane samples average value. The limit of quantification (LOQ) was determined by means of spiked blood serum samples and was the lowest concentration that could be quantified. In our analytical conditions, the LOD was 0.25 and the LOQ 0.75 ng/mL. Mean and standard deviation of chlordecone blood level were estimated using the maximum likelihood estimation (MLE) approach, which is a relevant procedure when dealing with left-censored data.
2.6. Statistical analysis
2.6.1. Predictive model of BCC building
Previous studies have underlined some limitations of the FCFC appro
การแปล กรุณารอสักครู่..