Population and study design
The data were retrieved from the results of Shanghai’s annual IDD surveillance, from March to June 2019. The target population was pregnant women with at least twelve continuous months of residency in Shanghai. A multi-stage, stratified random sampling method was conducted in 16 districts of Shanghai. The formula for calculating stratified random sampling sample size which was n = z2*S2*deff/d2 was used to calculate the sample size required for analysis. According to the results of pregnant women iodine nutrition survey in 2009 in Shanghai, S = 15 µg/L, d = 15 µg/L *0.05 = 0.75 µg/L. We defined the two-sided significance level α = 0.05, 1-β = 0.8, zα/2=1.96. The deff value of stratified random sampling was 1. At least 1537 pregnant women were needed for analysis. Each district was divided into five sections, a street was randomly selected from each section, and 30 pregnant women were selected from each of the five selected streets. The different gestational weeks of the women were evenly distributed.
Questionnaire Survey
All participants were required to complete a standardized questionnaire, which included information on their demographics, salt consumption, and iodine-related knowledge through face-to-face interviews with trained interviewers. Participants’ iodine-related knowledge, attitudes and behaviours was measured using a questionnaire with a total possible score of ten points, based on respondents’ ratings of the questions. The questionnaire was adopted after five experts reviewed its clarity and comprehension and agreed on all the items used in the questionnaire. The reliability coefficient of the questionnaire was 0.87, which was considered acceptable. Iodine-related knowledge was expressed as each participant’s total score on the questionnaire. All data were reviewed by the local district CDC project team, and the Shanghai CDC project team reviewed at least 5% of the data.
Data collection and analyses of household cooking salt and urine samples
More than 50 g of household cooking salt from each subject were collected during the household assessment. And 5 ml of fasting urine All urine samples were temporarily stored in a refrigerator at 4 ℃, then stored in a freezer at -20 ℃ for 12 hours, and finally transported to the testing unit within two weeks. All cooking salt samples were stored at room temperature and transported to the laboratory within one week.
The household cooking salt iodine concentration (SIC) and urinary iodine concentration (UIC) were measured through titration and acid digestion [17, 18], respectively, at the Central Laboratory of Shanghai’s Municipal CDC and the 16 district CDCs in Shanghai. The internal quality control of the samples for the analyses of the SIC and UIC were provided by China’s National Iodine Deficiency Disorders Reference Laboratory of the CDC.
Definitions And Classifications Of Relevant Indicators
Cooking SIC was classified into two types based on cooking SIC standards in Shanghai: non-iodized salt (SIC < 5.0 mg/kg) and iodized salt (SIC ≥ 5.0 mg/kg) which included low-iodized salt (5.0 mg/kg ≤ SIC < 21.0 mg/kg) and qualified-iodized salt (SIC ≥ 21.0 mg/kg).
The usage rate of iodized salt was equal to the percentage of salt with an iodine level ≥ 5.0 mg/kg in all samples.
The usage rate of qualified-iodized salt was equal to the percentage of salt with an iodine content ≥ 21.0 mg/kg in all samples.
The ‘abundant knowledge’ group consisted of participants who scored 8–10 points on the iodine-related knowledge questionnaire. The ‘general knowledge’ group were those who scored 6–7 points on the iodine-related knowledge questionnaire. The ‘lack of knowledge’ group were those who scored ≤ 5 points on the iodine-related knowledge questionnaire.
Former smokers were participants who smoked cigarettes in the past, excluding those who took a few tentative puffs. Former drinkers were those who usually drank alcoholic beverages during non-gestational periods, excluding those who sipped some wine.
The nutritional iodine status of the pregnant women was determined using the recommended criteria of the WHO/United Nations Children's Fund (UNICEF)/International Council for the Control of Iodine Deficiency Disorders (ICCIDD). Insufficient iodine intake was defined as MUIC < 150 µg/L; adequate iodine intake as MUIC 150–249 µg/L; iodine intake above the requirement as MUIC 250–499 µg/L; and excessive iodine intake as MUIC ≥ 500 µg/L [12].
Because of the high within-person variability of a single spot urine, the WHO programme guide limited the use and interpretation based on single spot urine per participant to population median of a sufficiently large group (in general, > 30) [12]. In our study, the sampling error (95% confidence interval (CI) of the MUIC) was considered and calculated using bootstrapping. Pregnant women are divided into 48 units according to the district and trimester. When the upper cut-off level of the 95% CI the MUIC in a unit was higher than 150 µg/L, the iodine status was considered optimal for all pregnant women in this unit, these participants were categorized as high UIC and the others were low UIC.
The definition of iodine-rich foods, which included kelp, laver, seaweed, and shrimp, was based on the eating habits of Shanghai residents and the food composition tables published in the China Health and Nutrition Survey (19).
Statistical analysis
Statistical analyses were conducted with Excel (2010 Edition, Microsoft, China) and SPSS (version 21.0, China). Frequency count data are expressed as number and percentage (%), normally distributed data are expressed as mean ± standard deviation (SD), and non-parametric data are expressed as the median (25th percentile, 75th percentile). One-way analysis of variance (ANOVA) was used to compare multiple groups. In pairwise comparisons, homogeneity of variance was tested using the least significant difference (LSD) test, and heterogeneity of variance was assessed using Tamhane's T2 test. The Kruskal-Wallis one-way ANOVA was used to compare the non-parametric data of multiple groups. Binary logistic regression analyses were used to explore factors with the potential to predict low UIC. A P-value below 0.05 was considered to be statistically significant.