A cross-sectional analysis was conducted with data from two public databases combined by child ID. One of the databases included information from the national survey, The Food and Nutrition Surveillance Survey for Stages of Life (FNSS), 2015 (In Spanish: Encuesta de Vigilancia Alimentaria y Nutricional por Etapas de Vida, VIANEV) [18]. The other database included administrative data from public health centers, called theHealth Benefits Report from the Integrated Health Services Systems of Comprehensive Health Insurance (ISHS), Ministry of Health 2015 (In Spanish: Reporte de Prestaciones de Salud del Sistema Integrado de Aseguramiento en Salud del Seguro Integral de Salud del Ministerio de Salud, SIASIS) [19]. The data extracted from the ISHS corresponded with the timeframe of the FNSS (2015) to ensure the indicators from both datasets were recorded during the same timeframe.
The FNSS is a nationally representative survey conducted by the National Center for Food and Nutrition, of the National Institute of Health (in Spanish: Centro Nacional de Alimentacin y Nutricin). The survey utilized a cluster sampling survey design with randomized selection to represent the national population. The survey examined nutritional outcomes and dietary intake in children under 3years.
Dietary intake was measured with a 24-h dietary recall on two non-consecutive days. The weight of each food consumed was estimated by weighting an approximately equivalent portion of the food with the survey participant. The dietary information was converted into the amount of nutrients consumed by the child during each day. To estimate the distribution of intake of nutrients, the survey used the software, PC-SIDE, developed by Iowa State University [20]. The intake of nutrients was compared to the dietary recommendations to meet estimated energy requirements (EER) by age and sex, defined by the Department of Nutrition for Health and Development of the World Health Organization [21,22,23,24]. The calculation determined if each participant met the nutrient requirements for their age of each nutrient category. The complete methodology is described in the surveys final report [18].
The FNSS survey used a portable spectrophotometer to estimate hemoglobin concentration of the participants. The hemoglobin concentrations were used to diagnosis anemia using cut-off points defined by the World Health Organization. For children aged 659months, the survey used the cut-off point defined by the World Health Organization of less than 11g per liter [25]. Due to the method of measurement, it is not possible to distinguish between the type or cause of anemia. Iron-deficiency anemia is the most common in the population [26].
The FNSS provided information regarding the water source of households surveyed. The water source was defined as safe drinking water if it same from public water source such as public water piped into the household, shared water pipe outside of the home, or a community well [18]. The information is represented in the current study by the categorical variable, access to safe drinking water, in which 0=no access to a safe drinking water source and 1=the household has access to a safe drinking water source.
The FNSS provided information that is used by the current study to control for the effect of poverty of the household. The variable is not a key predictor of interest, but it is included in the logistic regression analysis to control for its influence. The variable, basic needs met, indicators if the home meets the basic needs of the family. The home does not meet the basic needs if it is constructed with non-structurally sound material (plastic, carboard, etc.), dirt floor, overcrowded, no toilet (indoor or outdoor), a child aged 612 does not attend school, or if the head of the household did not complete primary school.
The FNSS survey database provided the following variables; child met iron recommendations, child met micronutrient recommendations (iron, zinc, vitamin A), child met micronutrient and energy requirements (iron, zinc, vitamin A, and calories), anemia diagnosis, access to safe drinking water, basic needs met, sex, age, and area of residence (metropolitan Lima, urban or rural).
The ISHS is a public database that contains administrative data reported by all public medical centers that accepts the public health insurance [19]. It provides information on the medical care provided to the population The database is available by request from the Ministry of Health of Peru. The illnesses reported in the databased, that were diagnosed and treated in the health centers, were categorized according to the International Classification of Diseases: Preparation of Short Lists for Data Tabulation [27]. The information of interest for this study from the ISHS database is the cases of intestinal infection or parasitic disease in child in 2016. The information is represented as a binary variable that indicates if the child was diagnosed with an intestinal infectious disease (1=at least 1 reported infection) or if the child has had no diagnosis of an intestinal infectious disease at a public health center during the year 20152016 (0=no reported infection). The category of intestinal infectious disease for this study includes bacterial intestinal infections, viral intestinal infections, and parasitic intestinal infections.
The FNSS and ISHS survey databases were combined by the Department of Information Technology of the Integrated Health Insurance Program, upon request by the authors. The public institution combined the FNSS and ISHS survey databases with the participants national ID number. The institution maintains confidentiality of the information and does not share any identifiable information of the participants.
Descriptive statistics were analyzed to better understand the experience of children with anemia in comparison to child without anemia. The differences between the two groups were compared with a Chi-square test to identify if the differences were statistically significant. Two logistic analyses were conducted to assess the strength of association between the key predictors and anemia in children in Peru. The first model assessed the association between anemia and intestinal infections, intake of iron, a poverty measure, and sex. The second model assessed the association between anemia and access to safe drinking water, intake of iron, a poverty measure, and sex. The model with intestinal intake and model with access to safe drinking water are analyzed separately because intestinal infection is a mediator between safe drinking water and anemia, and thus blocks the flow of association through the causal path [28].
Any cases that had omitted variables were not included in the analysis. The analysis adjusted for sampling design and clustering. The analysis was conducted with STATA/SE 16.1 [29].
