ABSTRACT:

Identifying the influence of socioeconomic, care, and feeding factors on children’s nutritional status is important for the evaluation and targeting of public policies based on nutritional interventions. We investigated the sociodemographic and biological factors associated with children aged 6 to 23 months leaving the low weight-for-age condition (weight-for-age z-score < -2) during their participation in a supplementary feeding program (SFP). This is a cohort study with 327 low-income children living in the inland of the state of São Paulo, who joined the SFP with low weight-for-age when they were six months old. The dependent variable was “maintained low weight-for-age during participation in the program” (dichotomous), and the independent variables related to: 1) maternal characteristics: marital status, schooling, age, and work situation; 2) child characteristics: being weaned, gender, birth weight, and age at weighing. We used a multiple multilevel logistic regression for the modeling. Factors positively associated with children’s weight gain were higher age at weighing (OR = 1.20; 95%CI 1.08 - 1.34; p = 0.001); higher birth weight (OR = 1.0011; 95%CI 1.0001 - 1.0019; p = 0.022), and being weaned when joining the program (OR = 0.20; 95%CI 0.08 - 0.52; p = 0.001). Actions focused on promoting appropriate birth weight and breastfeeding, and on adequate and timely introduction to healthy complementary feeding are important strategies to maximize the effects of the SFP on weight gain in the first two years of life of children from low-income families.

Keywords:
Socioeconomic factors; Supplementary feeding; Poverty; Weight by age; Multilevel analysis

INTRODUCTION

Identifying the influence of socioeconomic, care, and feeding factors on children’s nutritional status is important for the evaluation and targeting of public policies based on nutritional interventions¹1. Vitolo MR, Gama CM, Bortolini GA, Campagnolo PDB, Drachler ML. Alguns fatores associados a excesso de peso, baixa estatura e déficit de peso em menores de 5 anos. J Pediatr (Rio J) 2008; 84(3): 251-7. http://dx.doi.org/10.2223/JPED.1776
http://dx.doi.org/10.2223/JPED.1776... .

The assessment of the nutritional status of children younger than two years is most commonly conducted using three anthropometric indexes: weight-for-age (W/A), length-for-age (L/A), and weight-for-length (W/L). The difficulty in measuring the length of small children in the routine of health care services is one of the main obstacles in using the W/L and L/A indexes. Also, body weight is a measure more easily changeable than length since short-term dietary restrictions promptly reduce its value, and weight deficits can be corrected more quickly with dietary adjustments than length ones²2. World Health Organization. Physical status: the use and interpretation of anthropometry. Genebra: World Health Organization; 1995. (Technical Report Series, 854).. Therefore, the W/A index is the most appropriate to evaluate the effect of dietary interventions on the nutritional status of children in a short period of time³3. Augusto RA, Souza JMP. Efetividade de programa de suplementação alimentar no ganho ponderal de crianças. Rev Saúde Pública 2010; 44(5): 793-801. http://dx.doi.org/10.1590/S0034-89102010000500004
http://dx.doi.org/10.1590/S0034-89102010... .

In 2004, approximately 2.2 million children in the world had their deaths attributed to low W/A⁴4. World Health Organization. Global health risks: mortality and burden of disease attributable to selected major risks. Genebra: World Health Organization ; 2009.. Data from the National Demographic and Health Survey⁵5. Brasil. Pesquisa Nacional de Demografia e Saúde (PNDS: 1996 e 2006). Brasil reduz taxa de desnutrição infantil e atinge meta estabelecida pela ONU. Portal da Saúde [Internet]. Dez. 2010 [acesso em: 5 mar. 2013]. Disponível em: Disponível em: http://www.brasil.gov.br/editoria/saude/2010/12/saude-brasil-reduz-taxa-de-desnutricao-infantil-e-atinge-meta-estabelecida-pela-onu
http://www.brasil.gov.br/editoria/saude/... showed that low W/A in children younger than five years decreased from 4.2 to 1.8% between 1996 and 2006. However, socioeconomic and demographic differences persist, as younger children with lower purchasing power have worse nutritional status, and low W/A is about five times higher among them.

This situation was strongly evident in the past three decades, with successive changes in the determinants of health and disease conditions. Among the reasons for the progress reached are socioeconomic changes, interventions external to the healthcare sector (such as income transfer programs), vertical health programs (e.g., promotion of breastfeeding and immunization), creation of the Public Health System (Sistema Único de Saúde - SUS) and the Family Health Program, and implementation of several national and state programs to improve health, child nutrition, and food security⁶6. Victora CG, Aquino EML, do Carmo Leal M, Monteiro CA, Barros FC, Szwarcwald CL. Maternal and child health in Brazil: progress and challenges. Lancet 2011; 377: 1863-76. https://doi.org/10.1016/S0140-6736(11)60138-4
https://doi.org/10.1016/S0140-6736(11)60... . In this context, we have the State Milk Program “Vivaleite,” established in December 1999 to improve the nutritional status of low-income children.

A previous study³3. Augusto RA, Souza JMP. Efetividade de programa de suplementação alimentar no ganho ponderal de crianças. Rev Saúde Pública 2010; 44(5): 793-801. http://dx.doi.org/10.1590/S0034-89102010000500004
http://dx.doi.org/10.1590/S0034-89102010... measured the effectiveness of Vivaleite on the growth of participants (children exposed to the program), using newcomers (children not exposed to any action of the project and who did not consume food supplement) as a control group, and revealed that participants showed higher mean W/A z-score than newcomers, with this effect being as high as their nutritional deficiency when joining the program. From then on, there was an interest in investigating whether sociodemographic and biological factors interfere in the weight gain of children participating in the Supplementary Feeding Program (SFP), in order to find elements that promote the discussion of intervention actions to maximize the results obtained. Thus, we aimed to study the factors associated with children aged 6 to 23 months leaving the low W/A condition during their participation in the governmental program for fortified milk distribution “Vivaleite.” The study concerns children living in the inland of the state of São Paulo, who were six months old and had low W/A when they joined the program.

METHODS

DATA DESIGN AND SOURCE

Previously³3. Augusto RA, Souza JMP. Efetividade de programa de suplementação alimentar no ganho ponderal de crianças. Rev Saúde Pública 2010; 44(5): 793-801. http://dx.doi.org/10.1590/S0034-89102010000500004
http://dx.doi.org/10.1590/S0034-89102010... , a cohort study was conducted with data from 25,433 low-income children living in 311 cities in the inland of the state of São Paulo, who entered the Vivaleite SFP from January 2003 to September 2008, with age ranging from 6 to 24 months when they joined the program, complete data record, no report of health issues, no diarrhea in the prior 15 days, no hospitalization in the previous 3 months, no twins, and at least one weight measurement, besides the one held when they started in the program. Out of all children assessed, 88.7% (22,556 subjects) were active in the system, and the others had already left the SFP when the database was obtained.

Data were collected from the Children Record of the Vivaleite Project and the Quarterly Follow-up Worksheet on Anthropometric Data, both forms implemented in January 2003 by the Vivaleite coordination. The forms included information about household composition, characteristics of the mother/guardian and the child, and weight and height measures. For data collection, all managers in the cities periodically received training and printed information compiled by the Vivaleite nutrition team about the content of each item of the form, and the care needed to measure weight and height.

STUDY SAMPLE CHARACTERIZATION

The present study worked with a dynamic cohort sample of all 327 children who joined the SFP at six months of age and had low W/A (z-score < -2) on their first weighing, based on the growth standards of the World Health Organization¹1. Vitolo MR, Gama CM, Bortolini GA, Campagnolo PDB, Drachler ML. Alguns fatores associados a excesso de peso, baixa estatura e déficit de peso em menores de 5 anos. J Pediatr (Rio J) 2008; 84(3): 251-7. http://dx.doi.org/10.2223/JPED.1776
http://dx.doi.org/10.2223/JPED.1776... . The children were weighed every four months, and, for this research, we decided to monitor them until they reached 23 months of age. The schedule was not always followed over time, with children being weighed in the unscheduled ages of 9, 11, 13, 15, 17, 19, 21, and 23 months, as well as the scheduled 10, 14, 18, and 22 months, so we opted to work with the child’s actual ages at weighing. Since we worked with a dynamic cohort, some natural losses occurred over time. Each child had at least two and no more than five weighings, with weighing “1” corresponding to their measure when they joined the program.

VARIABLES

The dependent variable was “maintained low W/A” (W/A z-score < -2) during the weighing after starting in the SFP. This variable adopted a value of zero (0) for “yes” and one (1) for “no.” The comparisons with each independent variable used ratios of the result “no” (did not maintain low W/A), according to age at weighing.

The independent variables related to maternal and child characteristics were:

STATISTICAL ANALYSIS

The dependent variable is dichotomous with the repeated observation of the same child, leading to modeling with multiple multilevel logistic regression, directly producing individual level odds ratios (OR). Estimates of average probabilities allowed us to calculate marginal (crude and Mantel-Haenszel) OR, considering the explanatory variables separately.

The statistical model is:

In which:

πij=eβ0+∑βijxij1+eβ0+∑βijxij is the probability (low weight = 1|x _ij ) = probability (“not having low weight”|x _ij );

• β₀ = fixed intercept coefficient;
• Ϛ_0j = random factor of child j connected to the intercept coefficient;
• β₁ = fixed slope coefficient (effect) of the variable x ₁;
• Ϛ_idadej = random factor of child j connected to the slope coefficient of the variable x ₁;
• β_interaçãoj = interaction coefficient of the variable idade _x x ₁ of child j;
• idade_j = age at weighing of child j;
• x_1j = value of the variable idade _x x ₁ observed in child j; and
• interação_j = value of interaction x for child j;

The modeling sequence was:

The software Stata 10.1 (StataCorp LP, 2007) processed the data.

The commands, using the variable mother’s working situation as an example, were:

In which:

• i(codunico)  = controls the repetition of observation of a same child over time;
• margtrab  = estimates global marginal probabilities, according to age⁷7. Rabe-Hesketh S. Multilevel and Longitudinal Modeling using Stata. 2ª ed. Lakeway Drive, College Station, Texas: Stata Press; 2008..

ETHICAL ASPECTS

The Committee for Ethics in Research (Brazil Platform) approved this research, under a Certificate of Presentation for Ethical Consideration (Certificado de Apresentação para Apreciação Ética - CAAE) number 01383612.3.0000.5421. The research was conducted with data collected from the database of the State Milk Program - Vivaleite Interior, formally granted by the Secretariat of Agriculture and Food Supply of the state of São Paulo, and it was not possible to use the Informed Consent Form.

RESULTS

Table 1 presents maternal and child characteristics at the time the 6-month-old infant with low W/A joined the SFP, according to categories of sociodemographic and biological variables included in this study. Regarding maternal characteristics, 76% were in a stable conjugal situation, 46% had between 5 and 8 years of schooling, 77% were aged 20 years or older, and 85% did not have a paid work. Among the children assessed, 61% were male, 53% were weaned when they joined the program, and 66% had a birth weight greater than or equal to 2.5 kg.

Table 2 shows that the interactions between the child’s age and other variables studied in isolation were not statistically significant and were removed from subsequent modelings. Modeling 2 revealed that the results of the variables mother’s marital status, maternal schooling, mother’s age, mother’s working situation, and gender were not statistically significant, and were removed from the next modeling. Based on the statistical significance decision and considering a confidence interval of 95%, the selected variables for the final modeling were being weaned when joining the program (95%CI 0.09 - 0.58) and birth weight (95%CI 1.00 - 5.96), from which we found that:

Figure 1 exemplifies the modeled average probabilities related to modeling 3. The dotted black curve represents the situation of a child born with 3.25 kg, who was weaned when joining the program, used to illustrate the more favorable situation to leave the low W/A condition. The solid gray curve represents a child breastfeeding at the time he or she entered the SFP, with a birth weight of 2.5 kg, exemplifying the less favorable situation to leave the low W/A condition.

Table 3 shows the crude and adjusted values of children’s ages at weighing for each variable of the study, in which the crude OR values (average of all ages at weighing) and Mantel-Haenszel OR values (adjusted for ages at weighing) were close or equal to 1. The exception was the breastfeeding variable.

When considering only the child’s age at weighing, the OR of not having low W/A for each age was 1.29 times the OR of not having low W/A in the immediately preceding age (95%CI 1.13 - 1.47). Figure 2 demonstrates that most children who left the condition of low W/A did so after participating in the SFP for 3 to 4 months (66.2%), regardless of sociodemographic and biological variables. After this period, the reduction was constant and happened in lower intensity.

DISCUSSION

The results of other studies that evaluated milk distribution SFPs³3. Augusto RA, Souza JMP. Efetividade de programa de suplementação alimentar no ganho ponderal de crianças. Rev Saúde Pública 2010; 44(5): 793-801. http://dx.doi.org/10.1590/S0034-89102010000500004
http://dx.doi.org/10.1590/S0034-89102010... ^,88. Castro INR, Monteiro CA. Avaliação do impacto do programa “Leite é Saúde” na recuperação de crianças desnutridas no município do Rio de Janeiro. Rev Bras Epidemiol 2002; 5(1): 52-62. http://dx.doi.org/10.1590/S1415-790X2002000100007
http://dx.doi.org/10.1590/S1415-790X2002... ^,99. Goulart RMM, Franca Junior I, Souza MFM. Recuperação nutricional de crianças desnutridas e em risco nutricional em programa de suplementação alimentar no Município de Mogi das Cruzes, São Paulo, Brasil. Cad Saúde Pública 2007; 23(1): 147-56. http://dx.doi.org/10.1590/S0102-311X2007000100016
http://dx.doi.org/10.1590/S0102-311X2007... indicated a positive impact on the growth of children who benefited from them. Despite these programs having different characteristics, and adopting distinct analytical methods, anthropometric indexes, and growth reference standards, their results point to the same direction: the SFP on its own leads to weight gain for participating children. The present study confirmed this result during the program.

The progression of weight gain in participating children did not depend on socioeconomic characteristics. On the other hand, higher birth weight and early weaning were favorable conditions to weight gain in an observation period of fewer than 18 months, with a sharper effect during the first 4 months. These results corroborate other studies¹⁰10. Griffiths LJ, Smeeth L, Hawkings SS, Cole TJ, Dezateux C. Effects of infant feeding practice on weight gain from birth to 3 years. Arch Dis Child 2009; 94: 577-82. http://dx.doi.org/10.1136/adc.2008.137554
http://dx.doi.org/10.1136/adc.2008.13755... ^,1111. Sloan S, Gildea A, Steward M, Sneddon H, Iwaniec D. Early weaning is related to weight and rate of weight gain in infancy. Child Care Health Developm 2007; 34(1): 59-64. http://dx.doi.org/10.1111/j.1365-2214.2007.00771.x
http://dx.doi.org/10.1111/j.1365-2214.20... , which revealed a higher weight gain among weaned children born with greater weight, highlighting the increased risk of overweight in the long term. Therefore, the associations found in the present study should be interpreted with caution since even though they show a positive short-term effect in the recovery of underweight children, these conditions could favor excessive weight gain over time, and should not be promoted for children’s health.

It is noteworthy the lack of interaction between the variables studied and the increase in children’s age at weighing. For each variable, the marginal effect, expressed by crude or Mantel-Haenszel OR, was the same in all ages. In other words, after the initial increase observed in the second weighing (after three or four months of participating in the program), the ratio of children who did not have low W/A remained stable in all ages until the end of monitoring.

The case observed in the present study in which a “lower proportion of children breastfeeding at the time they joined the program left the condition of low W/A” should be interpreted with care, as it does not allow the questioning of the positive effects of breastfeeding, but suggests the hypothesis that early weaning might be associated with higher weight gain in childhood, a situation being investigated in other studies¹⁰10. Griffiths LJ, Smeeth L, Hawkings SS, Cole TJ, Dezateux C. Effects of infant feeding practice on weight gain from birth to 3 years. Arch Dis Child 2009; 94: 577-82. http://dx.doi.org/10.1136/adc.2008.137554
http://dx.doi.org/10.1136/adc.2008.13755... ^,1111. Sloan S, Gildea A, Steward M, Sneddon H, Iwaniec D. Early weaning is related to weight and rate of weight gain in infancy. Child Care Health Developm 2007; 34(1): 59-64. http://dx.doi.org/10.1111/j.1365-2214.2007.00771.x
http://dx.doi.org/10.1111/j.1365-2214.20... . According to Victora et al.¹²12. Victora CG, de Onis M, Hallal PC, Blössner M, Shrimpton R. Worldwide timing of growth faltering: revisiting implications for interventions. Pediatrics 2010; 125: e473-80. https://doi.org/10.1542/peds.2009-1519
https://doi.org/10.1542/peds.2009-1519... , the first two years of life are a window of opportunity to promote the healthy growth of children through appropriate eating habits, with exclusive breastfeeding until the infant reaches six months of age, and, after that, the timely and adequate introduction to healthy complementary feeding.

With respect to birth weight, although studies¹³13. Romani SAM, Lira PIC. Fatores determinantes do crescimento infantil. Rev Bras Saúde Matern Infant 2004 Jan/Mar; 4(1): 15-23. http://dx.doi.org/10.1590/S1519-38292004000100002
http://dx.doi.org/10.1590/S1519-38292004... ^,1414. Silveira KBR, Alves JFR, Ferreira HS, Sawaya AL, Florêncio TMMT. Associação entre desnutrição em crianças moradoras de favelas, estado nutricional materno e fatores socioambientais. J Pediatr 2010; 86(3): 215-20. http://dx.doi.org/10.1590/S0021-75572010000300009
http://dx.doi.org/10.1590/S0021-75572010... indicate its direct relationship with the future nutritional status of the child, publications that evaluate its influence in the nutritional recovery of children participating in supplementary feeding programs are rare.

Puccini et al.¹⁵15. Puccini RF, Goihman S, de Nóbrega FJ. Avaliação do Programa de Recuperação de Desnutridos do Município do Embu, na Região Metropolitana de São Paulo. J Pediatr 1996 Mar/Abr; 72(2): 71-9. http://dx.doi.org/0021-7557/96/72-02/71
http://dx.doi.org/0021-7557/96/72-02/71... , after assessing the Nutrition Recovery Program of the city of Embu (metropolitan area of São Paulo), found that children with lower birth weight had worse anthropometric evolution regarding W/A, including those monitored by the program. Spyrides et al.¹⁶16. Spyrides MHC, Struchiner CJ, Barbosa MTS, Kac G. Amamentação e crescimento infantil: um estudo longitudinal em crianças do Rio de Janeiro, Brasil, 1999/2001. Cad Saúde Pública 2005 May/Jun; 21(3): 756-66. http://dx.doi.org/10.1590/S0102-311X2005000300009
http://dx.doi.org/10.1590/S0102-311X2005... evaluated childhood growth in the first months of life of infants living in the city of Rio de Janeiro and revealed that those with low birth weight had lower weight development than others. The present study found the same results: lower birth weight positively associated with a smaller proportion of children who left the condition of low W/A.

The lack of statistical significance between the categories of the variables mother’s marital status, maternal schooling, mother’s age, mother’s working situation, and gender, and the proportion of children who left the condition of low W/A during their involvement in the SFP indicated that the maternal characteristics assessed and the child’s gender do not influence the weight gain process while the children participate in the program. The homogeneity of socioeconomic conditions of the population who benefited from Vivaleite was an important factor when considering the non-significance of the variables analyzed, as, for instance, among the mothers registered in the program, adolescents did not differ from non-adolescents.

We worked with existing data from SFP forms, which did not bring information about the eating habits of the children. The lack of data about eating habits could be a possible limitation in the evaluation of factors associated with the nutritional improvement of children. Thus, we suggest that the SFP includes this type of information in the routine of nutritional follow-up of participants. In addition, it could be argued that the option of working only with children who were 6 months old when joining the program and monitoring them until they reached 23 months of age is restrictive. Nonetheless, using other ages simultaneously could lead to a confusing epidemiological situation, which would require a much more complex statistical analysis, with results potentially less precise. By adopting this age range, the possible confounding effect of other ages was controlled. We expect similar studies with different starting ages, e.g., 12 months old, to compare the results. After comparing the final model (Modeling 3) with the results from Table 2 related to the variables breastfeeding and birth weight of modelings 2 and 3, we found that the values (OR, p, 95%CI) remained similar, that is, each of these variables was not a confounding factor for the other.

Since this research included only children from low-income families, participating in a governmental program, with a social focus, it is not possible to extrapolate these results to the general population. However, similar results could be expected in realities close to that of participants of Vivaleite, regarding socioeconomic, biological, and environmental characteristics.

From a public health viewpoint, the follow-up on nutritional status and the identification of factors that lead to changes in it are essential, so interventions can be carried out aimed at reducing risks and improving the quality of life of the child population. For this reason, public policies should invest in: improving prenatal care to decrease rates of low birth weight, intensifying their efforts in measures that stimulate proper intrauterine weight gain and postnatal nutrition¹⁷17. Brasil. Secretaria de Desenvolvimento Social. Diário Oficial do Estado de São Paulo. 2012; 122(35): 40.; promoting actions directed at food security, including the availability of food in natura or minimally processed fortified and access to adequate and healthy foods to prevent malnutrition and obesity in mothers and children; and developing educational actions to advocate the importance of breastfeeding and adequate eating habits, particularly in the first thousand days of life (from pregnancy until the child reaches two years of age), a crucial period that establishes eating habits and growth, and will last throughout life¹⁸18. Black RE, Victora CG, Walker SP, Bhutta ZA, Christian P, de Onis M, et al. Maternal and child undernutrition and overweight in low-income and middle-income countries. Lancet 2013 Aug; 382(9890): 427-51. https://doi.org/10.1016/S0140-6736(13)60937-X
https://doi.org/10.1016/S0140-6736(13)60... .

CONCLUSION

Although eradicating nutritional deficiencies has always been a priority in the agenda of international organizations, this issue persists. In the past decades, several countries adopted numerous strategies to prevent and control nutritional deficiencies and promote healthy growth. Among them, it is noteworthy the fortification of foods, which had positive results in some regions¹⁹19. World Health Organization. Use of multiple micronutrient powders for home fortification of foods consumed by infants and children 6-23 months of age. Genebra: World Health Organization ; 2011.^,2020. Backstrand JR. The history and future of food fortification in the United Sates: a public health perspective. Nutr Rev 2002; 60(1): 15-26. https://doi.org/10.1301/002966402760240390
https://doi.org/10.1301/0029664027602403... and limited ones in others, often due to the kind of supplement used, the amount of the food vehicle effectively consumed²¹21. Assunção MCF, Santos IS, Barros AJD, Gigante DP, Victora CG. Efeito da fortificação de farinhas com ferro sobre anemia em pré-escolares, Pelotas, RS. Rev Saúde Pública 2007; 41(4): 539-48. http://dx.doi.org/10.1590/S0034-89102006005000031
http://dx.doi.org/10.1590/S0034-89102006... , and the lack of integrated health and nutrition care actions¹⁴14. Silveira KBR, Alves JFR, Ferreira HS, Sawaya AL, Florêncio TMMT. Associação entre desnutrição em crianças moradoras de favelas, estado nutricional materno e fatores socioambientais. J Pediatr 2010; 86(3): 215-20. http://dx.doi.org/10.1590/S0021-75572010000300009
http://dx.doi.org/10.1590/S0021-75572010... .

We underline that, in low- and middle-income countries, a considerable prevalence of children with growth deficits and micronutrient deficiencies, mainly iron and vitamin A, persists, and is among the ten main preventable risk factors related to child morbimortality⁴4. World Health Organization. Global health risks: mortality and burden of disease attributable to selected major risks. Genebra: World Health Organization ; 2009.. In this context, and based on recent evidence, the World Health Organization¹⁴14. Silveira KBR, Alves JFR, Ferreira HS, Sawaya AL, Florêncio TMMT. Associação entre desnutrição em crianças moradoras de favelas, estado nutricional materno e fatores socioambientais. J Pediatr 2010; 86(3): 215-20. http://dx.doi.org/10.1590/S0021-75572010000300009
http://dx.doi.org/10.1590/S0021-75572010... recommends food fortification as a public policy action to prevent nutritional deficiencies in children younger than two years, by adding multiple micronutrients to complementary feeding, in addition to basic healthcare actions.

Supplementary feeding programs focusing on the nutritional recovery of children should associate maternal-child healthcare strategies with adequate eating habits.

References

Publication Dates

History