Study design
A cross-sectional design in two waves was conducted at maternal and child hospital of the Mayang in 2015 and in 2018. The Mayang was a minority nationality autonomous county, located in Midwest of China. The county consist of 18 towns or communities and the total birth numbers were 3988 in 2018.
The population of this study consisted of caregivers with children aged 6 to 23 months, involving the five towns or communities of the Mayang. The sample size was calculated by the following formula:
According to the Report of Nutrition Development for Children Aged 0-6 in China[3], the prevalence of anemia among children aged 6 to 23 months was approximately 25%. In this study, π=0.25, δ=5%, α=0.05, n≈290. The sample size increased to 300 considering the missing of objectives.
The total of 312 and 311 children aged 6 to 23 months were recruited in 2015 and in 2018 by multistage sampling technique. For first wave, five towns or communities were firstly selected at random in Mayang according to the total number of birth last year. Secondly, five villages were randomly selected in each town or community according to the total number of birth last year. The total of 25 villages was selected. Lastly, 10 to 15 children aged 6 to 23 months were randomly selected in each village lining up to children’s age. The second wave was conducted in 2018. The participants and method of sampling were consistent with the first wave. The number of birth was from maternal and child annual health report of Mayang.
Dietary data
The questionnaires asked caregivers about children’s consumption in the previous 24h. Total of 15 types of foods were recorded according to guide to infant feeding and nutrition of China[14] in this study. These foods included: breast milk, powered formulas, milk powder and fresh milk, water and soup, sugar water and drink, cereals, tubers, dark leaf vegetables and fruits, other vegetables and fruits, meat, egg, dairy products, bean and bean products, nut and multi-nutrient powders. The frequency of breast milk, powered formulas, milk powder, fresh milk and multi-nutrient powders were recorded in previous 24h, and other foods were recorded as yes or no (0=no and 1=yes).This food list was similar in both wave of survey. The questionnaires of dietary data collection were implemented through interviews by trained investigators.
Dietary patterns analysis
We used principal component analysis to identify dietary patterns. To understand the dietary patterns from 2015 to 2018, we combined both data sets before running the principal component analysis. Firstly, the frequency of 15 foods were included analysis of principal component analysis. The Kaiser-Meyer-Olkin statistic was used to analyze the compliance of variables. Then the number of dietary patterns was identified based on the eigenvalue (>1) and screen plot, factor interpretability, and variance explained. The food with factor loadings of > |0.3| were considered to significantly contribute to the identified factors. Lastly, the standardized scores which represent the sum of intakes of food weighted by their factor loadings were calculated for each child. The higher factor score in each pattern indicated that the participant was more related to this dietary pattern. Factor scores were categorised into four quartiles. Q1 indicates that the dietary pattern was weakly related to the dietary pattern and Q4 indicates that the dietary pattern was strongly related to the dietary pattern.
Dietary patterns assessment
World Health Organization infant and young child feeding indicators were used to assess dietary patterns[15]. Total 8 indicators was included in World Health Organization infant and young child feeding indicators. Definitions and parameters for child feeding associated with each indicator were used to disaggregate the data by the specified age groups. Children’s age was in group of 6 to 23 months in our study; therefore, four evaluation indicators were used, including: minimum dietary diversity, minimum meal frequency, minimum acceptable diet, consumption iron-fortified food.
Anemia assessment
The blood sample was collected by pricking children’s fingers to obtain capillary blood. Hemoglobin (Hb) was measured by microchemical reaction method. The hemoglobin concentration was detected by using hemoglobin machine which made from Hemocue AB Company of Sweden with model hemocue 301.The result was expressed in g/dL. The cut off points for anemia were: for children aged 6 to 23 months, normal Hb levels≥11.0 g/dL, anemia<11.0 g/dL.
Other related variables
Information about infant included: sex (boy or girl), age (6-11 months, 12-17 months and 18-23 months), birth weight ( normal or low birth weight), gestational age ( term or premature), fever and diarrhea in the previous 2 weeks (yes or no). Information about caregivers included: caregivers’ groups (parents or grandparents and others),educational level (illiteracy, primary, junior, and senior and above), occupations(homemakers or others), and ethnicity (Han and others or Miao).These information was collected by same questionnaire with dietary questionnaire.
Statistical analyses
The data are expressed as numbers and percentages for categorical variables.Significant differences were assessed by χ2 .The association between the related factors and anemia was assessed using the logistic regression model. Firstly, the bivariate logistic regression analyse was performed to analysize the age, sex, birth weight, gestational age, episode of diarrhea or fever in the previous 2 weeks, and dietary patterns of children as well as caregivers’ group, educational level, occupations, ethnicity. Then, factors with a value of P ≤ 0.10 in bivariate analysis were included in the multivariable logistic regression model.Odds ratio (OR) with 95% confidence interval (CI)were calculated to determine the strength of associations. A value of P <0.05 was considered indicative of statistical significance.All analyses were performed using Statistical Product and Service Solutions 13.