-
Bland-Altman analysis revealed a satisfactory result: the actual MUI, GR, CRIS and CRQIS values in 2018 were within 95% CI of the predicted values (Supplementary Table S1, available in www.besjournal.com). In Bland-Altman analysis, the actual value and the predicted value were in good agreement for all parameters (Supplementary Figure S2, available in www.besjournal.com).
Index Bland-Altman 95% CI Outiler percentage (%) Exact percentage (%) MUI (μg/L) −30.32 39.80 9.68 9.68 GR (%) −1.32 2.03 6.45 12.90 GRIS (%) −4.44 3.90 12.90 12.90 CRQIS (%) −5.54 7.75 6.45 0 Table S1. Bland-Altman analysis result
Figure S2. Provinces with 2018 actual and predicted MUI, GR, CRIS and CRQIS Bland-Altman analysis. (A) Provinces with 2018 MUI and predicted 2018 MUI Bland-Altman Analysis; (B) provinces with 2018 GR and predicted 2018 GR Bland-Altman Analysis; (C) provinces with 2018 CRIS and predicted 2018 CRIS Bland-Altman Analysis; (D) provinces with 2018 CRQIS and predicted 2018 CRQIS Bland-Altman Analysis. (A), (C), and (D) show three lines for the upper limit of the 95% consistency limit, exact consistency levels for both, and the lower limit of the 95% consistency limit. Figure (B), the seven lines are the upper confidence interval for the upper 95% consistency limit, the upper 95% consistency limit, the lower confidence interval for the upper 95% consistency limit, exact consistency levels for both, and the lower confidence interval for the lower 95% consistency limit.
The MUI in children, as shown in Figure 1, increased significantly from 1995 to 1997, then decreased to an adequate range of 100–200 μg/L from 1997 to 2011. After the adjustment of MIS in 2012, the MUI in children also showed a downward trend; however, the overall level of iodine nutrition for children at the national and provincial levels remained adequate. The MUI in children in all provinces was divided into three categories according to their changes, and the trends are shown in Supplementary Figure S3A–C, available in www.besjournal.com.
Figure 1. Time series analysis of IDD surveillance in different years. Solid lines show actual values, dots show 2018 actual values, and dashed lines show prospected values. IDD, iodine deficiency disorders; CRIS, coverage rate of iodized salt; CRQIS, consumption rate of qualified iodized salt; MUI, median urinary iodine concentration; GR, goiter rate
Figure S3. Time series analysis of IDD surveillance in different years (MUI of provinces). (A) Provinces with MUI of children present decreased tendency; (B) provinces with MUI of children present increased tendency; (C) provinces with MUI of children present unchanged tendency. Solid lines show actual values, dots show 2018 actual values, dashed lines show projected values.
The national GR showed a decreasing trend from 1995 to 2017, and has remained stable in recent years (Figure 1). The GR was below 5% in all provinces. In 2019–2022, the GR in some provinces is expected to increase while that in the other provinces remains unchanged (Supplementary Figure S4A–B, available in www.besjournal.com).
Figure S4. Time series analysis of IDD surveillance in different years (GR of provinces). (A) Provinces with GR of children present increased tendency; (B) provinces with GR of children present unchanged tendency. Solid lines show actual values, dots show 2018 actual values, dashed lines show projected values.
The national trend of CRIS increased, then remained stable, and then decreased after 2014 (Figure 1). CRIS is predicted to continue to decline from 2019 to 2022. The trends of CRIS in different provinces were essentially consistent with the national CRIS (Supplementary Figure S5A–B, available in www.besjournal.com).
Figure S5. Time Series Analysis of IDD Surveillance in Different Years (CRIS and CRQIS). (A) Provinces with CRIS present decreased tendency; (B) provinces with CRIS present unchanged tendency; (C) provinces with CRQIS present decreased tendency; (D) provinces with CRQIS present unchanged tendency. Solid lines show actual values, dots show 2018 actual values, dashed lines show projected values.
The national CRQIS was similar to the CRIS; the CRQIS in each province is shown in Supplementary Figure S5C–D. By 2017, the CRQIS in Tianjin (59.9%), Shanghai (64.6%), Zhejiang (79.4%), Ningxia (78.6%) and several other provinces significantly decreased. In the 2019–2022 predictions, the CRQIS in Tianjin, Hebei, Shanghai, Shandong, Zhejiang, Ningxia, and several other provinces are expected to continue to decline. The CRQIS in Tianjin and Shanghai is likely to decrease significantly to below 70%. The CRQIS in other provinces is expected to remain essentially unchanged, above 85%.
-
The results of global spatial autocorrelation analysis are shown in Table 1. Data from surveys conducted in 2016, 2017, and 2018 showed a statistically significant positive special correlation (P < 0.0001). Moran’s I for the CRIS, RQIS and CRQIS increased each year, and the spatial autocorrelation was significant. The spatial autocorrelation of MIS in 2017 was larger than that in 2016, and that in 2018 was similar to that in 2017. The spatial autocorrelation of the SD of SI in 2017 was higher than that in 2016 and 2018, and low spatial autocorrelation of the CV of SI was found in 2016, 2017, and 2018.
Index 2016 2017 2018 Moran’s I z score P value Moran’s I z score P value Moran’s I z score P value Coverage rate of iodized-salt 0.066 19.133 < 0.0001 0.302 71.409 < 0.0001 0.369 102.561 < 0.0001 Qualified Rate of iodized salt 0.056 16.302 < 0.0001 0.093 22.113 < 0.0001 0.152 42.077 < 0.0001 Consumption rate of qualified iodized salt 0.080 23.028 < 0.0001 0.301 70.805 < 0.0001 0.331 91.704 < 0.0001 Mean of Iodized salt 0.150 43.005 < 0.0001 0.259 60.879 < 0.0001 0.242 66.841 < 0.0001 Standard deviation of iodized salt 0.076 21.998 < 0.0001 0.131 32.045 < 0.0001 0.043 12.126 < 0.0001 Coefficient Variation of iodized salt 0.061 17.596 < 0.0001 0.031 7.735 < 0.0001 0.050 13.876 < 0.0001 Table 1. Global spatial autocorrelation analysis of salt related indexes
-
Getis-Ord Gi* was used for local spatial autocorrelation analysis. The hotspot map for the CRIS, RQIS, CRQIS, the MIS, SD and CV of SI in 2016, 2017, and 2018 is shown in Figures 2–4.
Figure 2. Local autocorrelation maps of the coverage rate of iodized salt (CRIS) and qualified rate of iodized salt (RQIS) in 2016, 2017, and 2018. (A) Local autocorrelation map of CRIS in 2016; (B) local autocorrelation map of CRIS in 2017; (C) local autocorrelation map of CRIS in 2018; (D) local autocorrelation map of RQIS in 2016; (E) local autocorrelation map of RQIS in 2017, (F) local autocorrelation map of RQIS in 2018.
Figure 3. Local autocorrelation maps of the consumption rate of qualified iodized salt (CRQIS) and median of iodized salt (MIS) in 2016, 2017, and 2018. (A) Local autocorrelation map of CRQIS in 2016; (B) local autocorrelation map of CRQIS in 2017; (C) local autocorrelation map of CRQIS in 2018; (D) local autocorrelation map of MIS in 2016; (E) local autocorrelation map of MIS in 2017; (F) local autocorrelation map of MIS in 2018.
Figure 4. Local autocorrelation maps of the standard deviation (SD) and coefficient of variation (CV) in 2016, 2017, and 2018. (A) Local autocorrelation map of SD in 2016; (B) local autocorrelation map of SD in 2017; (C) local autocorrelation map of SD in 2018; (D) local autocorrelation map of CV in 2016; (E) local autocorrelation map of CV in 2017; (F) local autocorrelation map of CV in 2018.
CRIS (Figure 2A–2C) In 2016, there was one hot spot and four cold spots; in 2017, there were two hot spots and two cold spots; in 2018, the distribution of hot spots and cold spots was similar to that in 2017. From 2016 to 2018, the four major cold spots—C1, C2, C3, and C4—disappeared, and the new cold spots C5 and C6 appeared in 2017 and 2018. C5 was converted from a hot spot to a cold spot (areas around iodine-excess areas). The hot spot area spread from the original H1 to the southwest, forming H3, with the addition of H2, indicating a wider range of hot spots and higher iodized salt coverage.
RQIS (Figure 2D–2F) In 2016, the RQIS had one hot spot, H1, and three cold spots. In 2017, the RQIS had two hot spots and two cold spots. In 2018, a new cold spot was added, concentrated in C5 (including Ningxia and Gansu). From 2016 to 2018, hot spot H1 disappeared and eventually became a cold spot, and the RQIS decreased overall. The C1 cold spot area narrowed with the simultaneous development of two new hot spots, H2 and H3, whereas the RQIS increased.
CRQIS (Figure 3A–3C) Because the CRQIS is influenced by the CRIS, the distribution of CRQIS hot and cold spots was consistent with that of CRIS. From 2016 to 2018, the scope of the H1 hot spot narrowed; part of it became the C6 cold spot, and the CRQIS decreased. The C1, C2 and C3 cold spots disappeared or became hot spots. Hot spots spread to the southwest, forming H3 and adding H2, and the CRQIS increased. In 2018 the CRQIS increased significantly over that in 2017.
MIS (Figure 3D–3F) In 2016, there were five MIS hot spots and three cold spots; the distribution of MIS hot spots in 2017 and 2018 was similar, with two hot spots and two cold spots. Between 2016 and 2018, the hot spots H1, H2, H3 and H4 all disappeared or became cold spots, new hot spots H6 and H7 formed, and the MIS increased. The cold spot evolved from the original central C2 and C3 to the eastern coast, forming a wide area of cold spots C4 and C5, with decreasing MIS.
SD of SI (Figure 4A–4C) In 2016, there were four hot spots of the SD of SI, and one cold spot; in 2017 and 2018 there were two new hot spots, and the cold spots changed from C1 to C2 in the original central region. Between 2016 and 2018, all three hot spots—H2, H3 and H4—disappeared or even became cold spots, and the SD of SI became smaller.
CV of SI (Figure 4D–4F) The CV of SI was similar to the distribution of the cold spots and hot spots of the SD of SI. Between 2016 and 2018, hot spots H2, H3 and H4 disappeared or even became cold spots, while new hot spots H5 and H6 developed. The C1 cold spot gradually disappeared, and two new cold spots C2 and C3 emerged.
HTML
Time Series Trends and Predictions
Spatial Autocorrelation Analysis
Global Spatial Autocorrelation Analysis
Local Spatial Autocorrelation Analysis
21348Supplementary Materials.pdf |