Study population
In total, 4,546 children aged 6–35 months (2,336 in 2013 and 2,210 in 2016) were included in this study. They consisted of 1,285 and 1,089 NLBC, 819 and 784 FLBC, and 232 and 337 PLBC in 2013 and 2016, respectively.
The characteristics of the children are presented in Table 1. In 2013 and 2016, NLBC and FLBC had similar median ages (19–20 months), but were 2–4 months younger than PLBC. The gender of the children and the frequency of premature birth did not differ significantly according to parental migration status.
The characteristics of the caregivers were similar in the two surveys. NLBC and FLBC were primarily cared for by their mothers. The maternal median age was 26–27 years and about 70% of mothers were educated to a middle school or higher level. More than 95% of PLBC were cared for by their grandparents (median age around 50 years), and less than one third of the caregivers were educated to a middle school or higher level (21.5% in 2013 and 31.1% in 2016). In 2013, The caregivers of PLBC had a significantly higher risk of depression than the caregivers of FLBC or NLBC (prevalence of 50.6%, 37.1%, and 40.0%, respectively). The risk of depression among the caregivers in 2016 was lower than that in 2013, and the prevalence of depression among the caregivers of NLBC, FLBC, and PLBC was 35.9%, 31.8%, and 38.0%, respectively (Table 1).
FLBC had a higher household economic status than NLBC and PLBC in 2013, but there was no significant difference in 2016. PLBC had a lower household economic status than NLBC and FLBC in both 2013 and 2016 (Table 1).
Breastfeeding and dietary intake
Child breastfeeding and dietary intake are presented in Figure 1. More than 85% of children were breastfed, and there was no significant difference according to parental migration status in 2013 and 2016. PLBC had a shorter duration of breastfeeding than FLBC and NLBC (10.82 vs. 11.89 and 12.26 months in 2013, and 9.98 vs. 12.28 and 11.31 months in 2016 respectively), and in 2016, FLBC had a longer duration of breastfeeding than NLBC. Dietary diversity of children increased from 2013 to 2016, irrespective of their parental migration status. In 2013, FLBC had slightly higher meal frequency and dietary diversity than NLBC, but in 2016, the differences were of lesser magnitude or absent. Compared with NLBC and FLBC, FLBC had higher meal frequency and dietary diversity at 6–17 months but slightly lower meal frequency and dietary diversity at 18–35 months.
Nutritional outcomes
Figure 2 shows the LAZ, WAZ, and WLZ scores of the children with different parental migration status and the results of univariate analysis. FLBC had significantly higher LAZ (-0.57 vs. -0.89, p < 0.01) and WAZ (-0.19 vs. -0.46, p < 0.01) scores in 2013 and significantly higher WAZ scores (-0.19 vs. -0.31, p < 0.05) in 2016 compared with NLBC. PLBC had significantly lower LAZ scores than FLBC in both 2013 (-1.07 vs. -0.57, p < 0.01) and 2016 (-0.83 vs. -0.61, p < 0.05), but no significant difference in LAZ scores between PLBC and NLBC was found. In addition, PLBC had significantly lower WAZ (-0.77 vs. -0.46 and -0.19, p < 0.01) and WLZ (-0.24 vs. 0.10 and 0.19, p < 0.01) scores in 2013 but similar scores in 2016 compared with NLBC and FLBC.
The prevalence of stunting among NLBC, FLBC, and PLBC was 19.0%, 11.8%, and 18.6% in 2013, and 13.4%, 8.9%, and 15.4% in 2016, respectively; results of univariate analysis showed FLBC had significantly lower risks of stunting than NLBC and PLBC in both 2013 and 2016 (p < 0.01). The prevalence of underweight among NLBC, FLBC, and PLBC was 9.2%, 6.5%, and 12.1% in 2013, and 4.3%, 4.0%, and 3.0% in 2016, respectively; results of univariate analysis showed FLBC had significantly lower risks of underweight than NLBC (p < 0.05) and PLBC (p < 0.01) in 2013, but no significant differences were found among them in 2016. the prevalence of wasting among NLBC, FLBC, and PLBC was 3.3%, 3.3%, and 5.2% in 2013, and 1.6%, 1.8%, and 0.6% in 2016, respectively. No significant differences in risk of wasting were found among children with different parental migration status (Figure 2).
Table 2 shows the results of multivariate analysis on these anthropometric indicators. After controlling for the confounding factors, FLBC had significantly higher LAZ (adjusted mean difference: 0.13, 95%CI: 0.01, 0.26) and lower risk of stunting (OR 0.73, 95%CI: 0.56, 0.96) than NLBC in 2013, but no such significant differences were found in 2016. No other significant differences in these indicators were found by multivariate analysis between children with different parental migration status.
Figure 3 shows the Hb concentration and the prevalence of anemia of children aged 6–35 months. FLBC and NLBC had similar hemoglobin concentrations at 6–35 months in 2013 and 2016. While PLBC had 0.6–0.8 g/dL significantly higher Hb concentrations than FLBC and NLBC at 6–17 months.
The prevalence of anemia among NLBC, FLBC, and PLBC aged 6–17 months was 61.5%, 63.6%, and 43.1% in 2013, and 66.2%, 60.1%, and 42.5% in 2016, respectively. PLBC had a significantly lower risk of anemia than FLBC and NLBC (p < 0.05). The prevalence of anemia among NLBC, FLBC, and PLBC aged 18–35 months was 32.7%, 36.0%, and 32.2% in 2013, and 29.8%, 35.8%, and 32.9% in 2016, respectively, and there was no significant difference according to parental migration status (Figure 3).
Table 3 shows the results of multivariate analysis on the Hb concentration and the risk of anemia. The results further supported the association found in the univariate analyses: significantly higher Hb concentration and lower risk of anemia was found in PLBC at 6–17 months (p < 0.05) but not at 18–35 months when compared with FLBC and NLBC at the same age group; no significant difference in the Hb concentration and the risk of anemia was found between NLBC and FLBC.