Determinants of infant nutrition status in rural farming households before and after harvest

Abstract

Infant and young child feeding (IYCF) practices determine infant growth, development and health. Despite global recommendations for exclusive breastfeeding until 6 months, adherence rates are low worldwide for different reasons, largely dependent on environment. In low‐income countries, inappropriate IYCF leads to poor nutrition status. This study examined IYCF practices and nutrition outcomes in rural farming households in Tanzania before and after harvest. Mothers and their infants were recruited from two regions in Tanzania. Demographics, health status, IYCF practices, anthropometrics and haemoglobin were measured; preharvest and postharvest. Regression analysis modelled the relationship between IYCF and nutrition outcomes. Despite high rates of breastfeeding a large proportion did not meet early initiation of breastfeeding and minimum acceptable diet standards. Undernutrition was high with 30–40% of infants classified as stunted depending on season, and the majority (81%) were anaemic. Early initiation of breastfeeding was associated with higher Length‐for‐age z‐score and weight‐for‐age z‐score and lower risk of stunting and underweight (p < 0.05). The introduction of fluids other than breast milk in the first 3 days after birth was associated with lower weight‐for‐age z‐score and increased underweight (p < 0.05). Maternal age and height were strongly and positively associated with child anthropometrics. Findings confirm the importance of early infant feeding practices for growth and development and emphasize the significance of mother's nutrition status in relation to infant health. Future interventions should focus on improving maternal nutrition status before, during and after pregnancy as well as educating and supporting mothers to adopt appropriate infant feeding including breastfeeding practices for the prevention of undernutrition.

Key messages.

• There is poor compliance with WHO IYCF guidelines in rural Tanzania including inadequate breastfeeding, early introduction of complementary foods and failure to meet minimum frequency and diet diversity standards.

• Chronic undernutrition reflected in low LAZ scores and high stunting rates remains a major nutrition problem.

• Early infant feeding practices and mother's age, height and BMI are all significant predictors of child nutrition status.

• The findings emphasize the importance of early targeted interventions for rural communities with a focus on maternal health, nutrition and nutrition knowledge before, during and after pregnancy.

1. INTRODUCTION

Infant and young child feeding (IYCF) practices determine infant growth, development and health (Nyaradi, Li, Hickling, Foster, & Oddy, 2013; Prendergast & Humphrey, 2014). Low‐income countries are most vulnerable to the causes and consequences of undernutrition. In sub‐Saharan Africa, ~36% of children <5 years are stunted, ~18% are underweight and ~8% are wasted (United Nations Children's Fund [UNICEF], 2015). The rates and the forms of undernutrition vary between countries and even within countries depending on regional conditions; for example, rural Tanzania has above average rates of stunting compared with urban regions (Tanzania Demographic and Health Survey and Malaria Indicator Survey [TDHS‐MIS], 2015–2016) Micronutrient deficiencies are also common in Tanzania, with ~81% of infants 6–11 months classified as anaemic (TDHS‐MIS, 2015–2016). Given the importance of IYCF, the World Health Organization (WHO) and the United Nations Children's Fund (UNICEF) developed and validated age appropriate indicators so that IYCF practices can be measured and monitored (Daelmans, Dewey, & Arimond, 2009). The indicators assess the WHO infant feeding recommendations including exclusive breastfeeding (EBF) to 6 months, timely introduction of appropriate complementary foods and continued breastfeeding to 2 years using standardized criteria (Kramer et al., 2004). The WHO/UNICEF IYCF indicators allow researchers and other stakeholders to assess current feeding practices and determine where improvements are needed to meet the Global Strategy for Infant and Young Child Feeding (WHO, 2002). Globally, EBF rates are low, and only 37% of infants in low‐ and middle‐income countries are EBF to 6 months (Victora et al., 2016). In Tanzania, despite relatively high rates of EBF at 1 month (84%), only 27% of infants are EBF at 5 months (TDHS‐MIS, 2015–2016). In line with the drop in EBF, significant proportions of infants are receiving complementary foods as young as 2 months, and by 4–5 months about 61% of infants have eaten solid or semi‐solid foods (TDHS‐MIS, 2015–2016).

Poor feeding practices, food insecurity and poverty are the main drivers of chronic undernutrition in Tanzania (DPG Nutrition, 2010; Kulwa, Mamiro, Kimanya, Mziray, & Kolsteren, 2015). Governments and nongovernmental organizations are now focusing on the role of agriculture in improving nutrition globally (Dewey & Adu‐Afarwuah, 2008; Ruel et al., 2013; Scaling Up Nutrition, 2012; UNICEF, 2013; Wiegers, van Dorp, & Torgerson, 2011). This initiative was emphasized with the publication of the Sustainable Development Goals, one of which aims to “end hunger, achieve food security and improved nutrition, and promote sustainable agriculture” (Lartey, 2015). Households in rural Tanzania depend on small‐scale subsistence agriculture for food and income. Subsistence‐scale farming systems are highly sensitive to extreme weather patterns, which affects the timing of crop harvests and amount and types of food stocks available for the family to consume throughout the year (International Fund for Agricultural Development, 2014). Yearly climate conditions also impact household income, price of food and subsequently, and food consumption (Kaminski, Christiaensen, & Gilbert, 2016; Omambia & Gu, 2010). Inadequate diet will impact child growth directly and indirectly through impact on maternal nutrition and health (Ramakrishnan, Young, & Martorell, 2017; Victora et al., 2008). In addition, season and food access were identified as constraints to successful IYCF interventions by the WHO and UNICEF (Daelmans, Mangasaryan, et al., 2009). Given the challenges faced when collecting data in rural settings (Pierce & Scherra, 2012), very few studies have compared IYCF practices and child nutrition status in rural regions and in different seasons. This research reports IYCF practices and nutrition status of children less than 2 years of age in rural Tanzania during preharvest and postharvest seasons and examines the factors that contribute to infant nutrition status.

2. METHODS

2.1. Study design, areas, population and sampling

This study was part of the AgriDiet project http://agridiet.ucc.ie/. The study was conducted in the rural areas of Morogoro (Mvomero district) and Shinyanga (Kishapu district) regions in Tanzania Mainland. The regions were selected based on accessibility for data collection and household farming practices, which were different in each region but representative of Tanzanian agriculture. The districts were selected based on the proportion of household practicing subsistence level agriculture coupled with a high rate of undernutrition. Kishapu (4,333 km²) has a population of 272,990, 7.7 people per household and practices crop and livestock farming. Mvomero (7,325 km²) has a population of 312,109, 5.3 people per household and practices mainly crop farming. The main crops in both regions include maize, rice, sorghum, sweet potatoes green gram, and groundnuts. Agriculture in both regions is dependent on season and rainfall, which follows a unimodal and in some parts bimodal rainfall pattern. Districts (within regions) are divided into divisions, and divisions are further subdivided into wards and villages. For the purpose of this study, one division was randomly selected from the district register, two wards were randomly selected within each division, and one village was randomly selected from each ward. The study villages were Lubaga and Mwakipoya in Kishapu district, and Makuyu and Milama in Mvomero district. A random sample of 110 mother–child pairs was selected from each district (proportion relative to village size) using a master list of households with children less than 2 years, which was created with help from the village and hamlet leaders. Pregnant women, critically ill children or children with evidence of chronic disease were excluded from the nutrition study. Mother–child pairs were recruited during preharvest (February–March 2014), and the same mother–child pairs were followed up postharvest (August 2014) in a longitudinal study design. Preharvest is considered a period of food shortage. Postharvest data were collected 6–8 weeks after the start of harvest in both regions. All eligible mothers were informed about the study and gave their consent before they were invited to participate. Selected mothers who refused to be interviewed were randomly replaced by others from the list. All data were collected by a team of trained researchers following standard protocols.

2.2. Ethical clearance

Ethical approval was granted by the National Institute of Medical Research, Tanzania (Reference: 1679). All procedures were conducted according to the principles described in the Declaration of Helsinki.

3. DATA COLLECTION

3.1. Demographics and health information

A pretested structured questionnaire was used to record information on maternal and infant characteristics, including child's age, sex, birth weight, vaccination status and place of delivery, mother's age, education, marital status, occupation, availability and use of health care services.

3.2. Infant and Young Child Feeding Practices (IYCF)

Infant feeding practices were recorded in accordance with the WHO guidelines (Daelmans et al., 2009). The WHO recommends EBF to 6 months followed by timely introduction of appropriate complementary foods combined with continued breastfeeding up to 2 years or beyond. The WHO defines EBF as “no other food or drink, not even water, except breast milk (including milk expressed or from a wet nurse) for 6 months of life, but allows the infant to receive ORS, drops and syrups (vitamins, minerals and medicines).” The WHO IYCF indicators are grouped into core and optional indicators, which are age specific and assessed based on information collected for the day preceding the survey, except for “early initiation of breastfeeding” and “ever breastfed”. Mothers were asked questions relating to breastfeeding practices and the introduction of complementary foods. Mothers were asked to recall and describe every item of food and drink that their child consumed over a period of 24 hr prior to the day of interview. The 24‐hr recall was used to calculate diet diversity according to WHO criteria. The questionnaire included extra questions to collect retrospective data such as the duration and frequency of feeds and the timing of the introduction of fluids and foods other than breast milk.

3.3. Anthropometrics and haemoglobin

Mother and infant weights were measured using a Seca 874 weighing scales wearing minimal clothing. Infant weight was measured using the “tared weighing” method while held by their mother. Infant recumbent length and mothers height were measured using the UNICEF Portable baby/child/adult length and height measuring system SET‐2 (S0114540) following standard procedures (Cogill, 2003). All anthropometric equipment was calibrated prior to use. Length‐for‐age z‐score (LAZ), weight‐for‐age z‐score (WAZ) and weight for‐length z‐score (WLZ) were calculated using WHO ANTHRO Software v 3.2.2 (Anthro, WHO, Geneva) and WHO Child Growth Standards (De Onis, 2006). Stunting, underweight and wasting were defined as z‐scores below minus 2 standard deviations (SD) of the median values of the reference data. Haemoglobin (Hb) was measured using the HemoCue instrument (HemoCue 201+System, Ängelholm, Sweden). A finger‐prick capillary blood sample was collected into a sterile microcuvett. Hb concentration was recorded from the portable photometer, which was calibrated daily. Hb concentrations were corrected for altitude (−0.2 g/dl) for participants living in Shinyanga, which is between 1,000 and 1,499 m above sea level, in accordance with correction factors reported by Berger et al. (1997). Children having Hb concentration values of ≥11 g/dl were classified as normal, 10–10.9 g/dl mild anaemia, 7–9.9 g/dl moderate anaemia and <7 g/dl severe anaemia (WHO, 2011).

3.4. Data processing and analysis

All data were entered into a study database, which was imported into IBM SPSS Statistics for Windows (Version 20.0. IBM Corp, 2011, Armonk, NY) for further analysis. Descriptive statistics were generated for each variable. The distribution of continuous variables was assessed visually using box plots and Q‐plots and the Shapiro–Wilk test confirmed normal distribution. Nonnormal data were log transformed. If transformation failed to achieve normality, nonparametric methods were used. Multicollinearity was checked using the tolerance and the variance inflation factor. The variance inflation factor ranged from 1 to 1.27, and the tolerance test was <1 for all models, which was within normal limits. Results are presented as frequencies and proportions for categorical variables, Mean ± SD for continuous variables or median and interquartile range for nonnormal continuous variables. One‐way analysis of variance, independent sample t tests and paired sample t tests were used to test group differences for continuous variables. Chi‐squared tests were used to test group differences for categorical variables. McNemar's test was used to test differences in paired categorical data. Multiple linear regression models were used to determine the combination of independent variables that predict continuous dependant variables preharvest and postharvest. Binary logistic regression models were used to determine the combination of independent variables that predict binary dependant variables preharvest and postharvest. The proportion of infants classified as wasted was small; hence, it was not included in the logistic regression. Following Greenland (1989) recommendations only a limited subset of candidate variables were included in the models (Greenland, 1989). Variable selection was based on bivariate correlation coefficients and support from previous research describing confounding. The preharvest and post‐harvest regression models included the following independent variables: initiation of breastfeeding within 1 hr of delivery (yes/no, categorical), infant fed extra fluid other than breast milk within 3 days after delivery (yes/no, categorical) and EBF (weeks, continuous). Frequency of breastfeeding (times per day, categorical) and breastfeeding on demand (yes/no, categorical) were included in the preharvest model only while age at which complementary foods were introduced (months, continuous) and minimum acceptable diet (yes/no, categorical) were used in the postharvest model only. Covariates were selected based on their contribution to confounding in each model and included infant age and gender, birth weight, birth order, number of children, mother's age, mother's height and mother's body mass index (BMI). Independent variables were incorporated using the Enter method. Regression coefficients and 95% confidence intervals are reported for all multivariate models. Each regression model followed the method noted below:

$$\mathrm{Y}\left(\text{anthropometric index}\right)=\text{β0}+\mathrm{\beta }1\left(\text{IYCF indicator}\right)+\mathrm{\beta }2\left(\text{mother},\text{infant characteristics}\right)+\mathrm{ϵ}$$

4. RESULTS

4.1. Socio‐demographic characteristics

The socio‐demographic characteristics of mother–child pairs (n = 220) are presented in Table 1. Twenty‐two mothers could not be contacted and were lost to follow‐up postharvest. Mothers were aged between 17 and 44 years; the majority were married and had at least primary school education. Fifty percent of mothers were earning <50,000 Tanzania Shillings ($23 USD) per month, and 10% of mothers had no income. Majority of mothers attended antenatal care services during their pregnancy and delivered their babies in a health clinic. Fifty nine percent of mothers received folic acid and iron (fefo) supplements. Almost 75% of mothers reported that they had not received any information about IYCF practices during their antenatal care visit.

Table 1.

Infants and mother characteristics

Characteristic	Category	N/median	%/IQR
Infant gender	Male	108	49.1
	Female	112	50.9
Infant age groups (months)a	0–5 months	47	21.4
	6–11 months	69	31.3
	12–24 months	104	47.3
Infant birthweight	Birth weight ≥ 2,500 g	168	95.5
	Birth weight < 2,500 g	8	4.5
Mother age group (years)	15–22 years	74	33.6
	23–30 years	77	35.0
	31–44 years	69	31.4
Mother anthropometryb	Preharvest weight (kg)	53.36	39.30–90.10
	Postharvest weight (kg)	56.55	38.00–94.40
	Height (cm)	156.66	139.55–170.66
	Preharvest BMI (kg/m2)	22.13	16.86–33.10
	Postharvest BMI (kg/m2)	22.81	16.59–33.75
Vitamin A after delivery	Yes	52	23.6
	No	164	74.5
Infant vaccination status	Completed all vaccine	165	85.9
	Not completed	12	6.7
	Infants <9 months	15	7.3
Folic acid, iron or Fefo	Folic acid	50	28.6
	Iron	21	12.0
	Fefo (combined)	104	59.4
Mothers who received information on feeding practices during RCH visit	Yes	58	26.4
	No	162	73.6
Type of information received at RCH	Breastfeeding	19	8.6
	Introduction of CF	4	1.8
	Breastfeeding and intro of CF	34	15.5
	Did not receive any information	163	74.1
Marital status	Single	12	5.5
	Married	179	81.4
	Divorced/separated	14	6.4
	Cohabit	15	6.8
Mothers employment	Farmers	203	92.3
	Employed/formal sector	1	0.5
	Others	16	7.2
Education level of the mother	No schooling	39	17.7
	Primary	167	75.9
	Secondary	14	6.4
Mothers monthly income (TZS)c	No income	22	10.0
	Cannot estimate income	62	28.2
	<100,000	131	59.5
	100,000 < 500,000	5	2.3

Note. BMI: body mass index; CF: complementary foods; TZS: Tanzania shillings; Fefo: combined folic acid and Iron; RCH: reproductive and child health.

^a Preharvest age presented, therefore postharvest age of children equals age + 6 months.

^b Mother's weight (kg), height (cm) and BMI (kg/m²) are presented as median with interquartile ranges (IQR).

^c Euro was equivalent to TZS 2,223 February 2014 (Bank of Tanzania, 2014).

4.2. WHO infant and young child feeding practices

Overall, 47% of mothers reported that they initiated breastfeeding within 1 hr of delivery, 26% reported that their infant was EBF from 0 to 5 months, and 55% were predominantly breastfed at the time of the survey (Table 2). The majority of infants were still breastfed at 12 months of age in preharvest and postharvest seasons. During the preharvest period the majority of infants aged 6–8 months received solid foods, but only 20% met minimum diet diversity standards. At postharvest, all infants received solid foods and 36% met the criteria for diversity. Further analysis of diet diversity by age group showed a higher proportion of 18–23 months only children met minimum standards compared with younger age groups preharvest, although the difference was not statistically significant (p = 0.08). There was no significant difference in the proportion of children meeting minimum diet diversity standards postharvest. Minimum meal frequency criteria were met by 75% of children during preharvest and 90% of children postharvest, resulting in an overall acceptable diet for 17% of children during preharvest and 31% during postharvest. Consumption of iron‐rich or iron‐fortified foods by infants during preharvest was very low in both seasons. Based on WHO IYCF indicator criteria and using data from the 24‐hr recall, 79% of children preharvest and 71% of children postharvest were fed appropriately for their age (Table 2). In addition to the WHO IYCF indicators, this study collected retrospective information in relation to other infant feeding practices. Results indicated that giving infants fluids other than breast milk within the first 3 days of delivery was common (38%); mostly water/with sugar (75%). EBF decreased each month from 44% at 1 month, 23% at 3 months, and 15% at 5 months, to only 8% at 6 months. Breastfeeding on demand was practiced by 81% of the mothers and 55% breastfed for more than 10 times per day. Complementary foods were introduced early (mean age 3.4 months); only 25% of infants received their first solid foods at 6 months. The first food was maize porridge (99%). More than 70% of infants only received foods produced on the household farm, and the other ~30% received some foods bought at market such as sardines.

Table 2.

World Health Organization indicators for assessing infant and young child feeding

	Preharvest		Postharvest
Infant and young child feeding practice indicators	N	%	N	%	p
Core indicators
1. Early initiation of breastfeeding: Infants born in last 24 months and breastfed within 1 hr of birth (n = 218)	103	47.2	–	–	–
2. Exclusive breastfeeding under 6 months: Infants 0–5.9 months who received only breast milk on the day before (n = 47)	12	25.5	–	–	–
3. Continued breastfeeding at 1 year: Infants 12–15 months who received breast milk on the day before (n = 59; n = 38)	57	96.6	38	100	0.33
4. Introduction of solid, semi‐solid or soft foods: Infants 6–8 months who received these foods the day before (n = 26; n = 12)	25	96.1	12	100	0.14
5. Minimum dietary diversitya: Infants 6–23 months who received foods from 4 or more food groups on the day before (n = 171; n = 185)	34	19.9	66	35.6	<0.001
6. Minimum meal frequencya: Proportion of breastfed infants 6–23 months who received solid, semisolid or soft foods a minimum number of times or more on the day before (n = 171; n = 185)	128	74.8	166	89.7	<0.001
7. Minimum acceptable dieta: Breastfed infants 6–23 months who received at least the minimum diet diversity and meal frequency on the day before (n = 171; n = 185 post)	29	16.9	57	30.8	<0.001
8. Consumption of iron‐rich or iron‐fortified fooda: Infants 6–23 months who received an iron‐rich food or fortified food on the day before (n = 171; n = 185)	31	18.1	68	36.7	<0.001
Optional indicators
Infants ever breastfed: Infants born in the last 24 months who were ever breastfed (n = 220)	220	100	–	–	–
Continued breastfeeding at 2 years: Infants 20–23 months who received breast milk on the day before (n = 10; n = 60)	9	90.0	18	30.0	<0.001
Age‐appropriate breastfeedingb: Infants 0–23 months who are appropriately breastfed (n = 220; n = 193)	173	78.6	137	70.9	0.02
Predominant breastfeeding under 6 monthsc: Infants 0–5 months who received breast milk as the predominant food source on the day before (n = 47)	26	55.3	–	–	–

Note. Indicators for assessing infant and young child feeding practices, conclusions of a consensus meeting held 6–8 November 2007 in Washington DC, USA; Working Group on Infant and Young Child Feeding Indicators 2007(WHO, 2008).

^a The differences in proportions were compared preharvest and postharvest (McNemar test), and all were all significant at p < 0.001.

^b Age‐appropriate breastfeeding: This combine two groups: infants 0–5 months of age who received only breast milk during the previous day and children 6–23 months of age who received breast milk, as well as solid, semisolid or soft foods, during the previous day.

^c Predominant breastfeeding means that the infant's predominant source of nourishment has been breast milk. However, the infant may also have received liquid, ritual fluids and ORS, drops or syrups.

4.3. Nutrition status of infants and young children

Infant nutrition status indicators are presented in Table 3. LAZ was significantly lower for males compared with females during pre‐harvest (p = 0.01). LAZ was significantly different between age groups, with lower mean LAZ for older children pre‐harvest (p = 0.02) and postharvest (p = 0.05). WAZ (p = 0.03) and WLZ (p < 0.001) were significantly lower for older children preharvest, but there was no significant difference by age group postharvest. Pair‐wise comparisons for the total group showed a significant decrease in LAZ from preharvest to postharvest (p < 0.001), an increase in WAZ that did not reach significance (p = 0.06) and a significant increase in WLZ preharvest to postharvest (p < 0.001). Further analysis splitting the file by LAZ/WAZ/WLZ category (severe, moderate, mild, and normal) showed an increase in LAZ postharvest for children classified as severely stunted preharvest (p = 0.06). LAZ change was not significant for the moderately stunted group (n = 44), but WAZ (p = 0.009, n = 20) and WLZ (p = 0.049, n = 6) increased significantly postharvest. In contrast, there was a significant decrease in LAZ for the mildly stunted (p < 0.001, n = 60) and normal groups (p < 0.001, n = 76), no change in WAZ, and an increase in WLZ postharvest, but the difference was only significant for the mildly wasted group (p < 0.001, n = 36).

Table 3.

Mean LAZ, WAZ, WLZ and prevalence of stunting, underweight and wasting among infants by age, sex and seasonsa

% stunting	Preharvest (N = 215)					Postharvest (N = 198)					p
	N	Moderate/severe < −2	Mild > −2 to < −1	Normal > −1	Mean z‐score	N	Moderate/severe < −2	Mild > −2 to < −1	Normal > −1	Mean z‐score
Total	215	31b	30	39	−1.30 ± 1.19a	198	40b	32	28	−1.63 ± 1.13a	<0.001 a ; 0.01 b
Male	108	39	33	28	−1.49 ± 1.16	101	45	35	21	−1.76 ± 1.03
Female	107	22	28	50	−1.09 ± 1.19	97	36	29	35	−1.48 ± 1.21
p‐value		0.01 c			0.01 d		0.22c			0.07e
0‐5 months	45	18	31	51	−0.96 ± 1.02	—	—	—	—	—
6–11 months	66	27	31	42	−1.17 ± 1.20	38	24	34	42	−1.16 ± 1.11
12 months+	104	39		61	−1.51 ± 1.22	160	44.4	31.3	24.4	−1.65 ± 1.22
p‐value		0.03 c			0.02 e		0.01 c			0.05d
% underweight
Total	208	13b	27	60	−0.69 ± 1.18a	198	15b	23	62	−0.58 ± 1.21a	0.06a; 0.80b
Male	103	14	29	57	−0.80 ± 1.16	101	15	24	61	−0.65 ± 1.12
Female	105	12	25	63	−0.57 ± 1.18	97	16	22	63	−0.50 ± 1.27
p‐value		0.55c			0.15d		0.90c			0.36e
0‐5 months	45	0	20	80	−0.23 ± 0.88	—	—	—	—	—
6–11 months	62	10	32	58	−0.62 ± 1.22	38	5	16	79	−0.18 ± 0.99
12 months+	101	21	27	53	−0.94 ± 1.21	160	18	24	58	−0.69 ± 1.19
p‐value		0.01 c			0.03e		0.05c			0.30d
% wasting
Total	208	5b	18	77	0.03 ± 1.31a	198	2b	12	86	0.31 ± 1.20a	<0.001 a; 0.07b
Male	103	3	24	73	0.01 ± 1.37	101	1	11	88	0.27 ± 1.17
Female	105	7	12	81	0.05 ± 1.24	97	3	12	85	0.32 ± 1.24
p‐value		0.24c			0.87d		0.60c			0.81e
0‐5 months	45	0	9	91	0.70 ± 1.17	—	—	—	—	—
6–11 months	62	5	16	79	0.07 ± 1.36	38	0	5	95	0.56 ± 0.97
12 months+	101	7	24	69	−0.29 ± 1.22	160	3	13	84	0.23 ± 1.25
p‐value		0.09c			<0.001 e		0.27c			0.13d

Note. Z‐score values are presented as mean ± SDs based on World Health Organization, 2006 reference standard.

LAZ: length for age z‐score; WAZ: weight for age z‐score; WLZ: weight for length z‐score; infants and young children < −2SD were classified as stunted, underweight and wasting; <−1 group of infants who were mild malnourished.

A p‐value of <0.05 was considered statistically significant.

Seven infants showed signs of oedema pre‐harvest and 3 showed signs post‐harvest.

^a Paired sample t test were used to compare differences in mean z‐scores preharvest to postharvest.

^b McNemar's test was used to test the significance differences of proportions of stunted, underweight and wasted infants from preharvest to postharvest.

^c Chi squared test was used and that this was used to compare differences for categorical variables between groups.

^d Independent sample t test was used to compare mean differences for continuous variables between two groups.

^e One‐way analysis of variance was used to compare mean differences for continuous variables between three age groups.

In line with z‐scores, the prevalence of stunting increased from preharvest (31%) to postharvest (40%; p = 0.01; Table 3). Stunting rate was significantly higher in males (39%) compared with females (22%) in preharvest (p = 0.01); however, there was no significant difference in postharvest nor were there any gender differences for underweight or wasting at either time‐point. Stunting was significantly higher in older age groups preharvest (p = 0.03) and postharvest (p = 0.01). The proportion of underweight children was higher in older age groups preharvest (p = 0.01) and postharvest (p = 0.05); however, there was no change in prevalence rates preharvest to postharvest (p = 0.80). Finally, there was a decrease in the prevalence of wasting, but it did not reach statistical significance (p = 0.07).

Mean infant Hb concentration was 9.38 ± 1.31 g/dl preharvest and 9.31 ± 1.41 g/dl postharvest. There was no significant difference between males and females or age groups at preharvest or postharvest time‐points nor was there a significant change in Hb concentrations preharvest to postharvest. This translated to high rates of anaemia preharvest (91%) and postharvest (85%). There was no difference in the proportion of males or females classified as anaemic, but there was a significant difference between age groups (p = 0.04). During preharvest period, 98% of children 6–11 month were classified as anaemic compared with 84% of children 0–5 months old and 88% of 12–24 months old (p = 0.04). This trend was observed during the postharvest period but did not reach significance (p = 0.10).

4.4. Relationship between infant feeding practices and nutrition status

In the preharvest period, mother's age and height were positively associated with LAZ and child age and the number of children in the household were negatively associated with LAZ (F = 4.042, p < 0.001, R ² = 0.20). Birth weight, mother's age and height were positively associated with WAZ, and child age, number of children, feeding fluids within 3 days of delivery and more weeks of EBF were negatively associated with WAZ (F = 4.173, p < 0.001, R ² = 0.21). There was a positive association between birth weight and WLZ, but child age was negatively associated with WLZ (F = 2.732, p = 0.002, R ² = 0.13; Table 4). At postharvest, mother's height and initiation of breastfeeding within 1 hr were positively associated with LAZ and child age was negatively associated with LAZ (F = 2.877, p = 0.001, R ² = 0.13). Mother's height, mother's BMI and initiation of breastfeeding within 1 hr were positively associated with WAZ, and child age and feeding fluids within 3 days of delivery were negatively associated with WAZ (F = 2.755, p = 0.002, R ² = 0.12). Mother's BMI was positively associated with WLZ, and feeding fluids within 3 days of delivery was negatively associated with WLZ (F = 1.904, p = 0.034, R ² = 0.07). Regression analysis failed to generate a model that could predict Hb during preharvest or postharvest periods.

Table 4.

Associations between LAZ, WAZ, WLZ and Mean Hb conc. with selected independent variables (linear regression results)

Dependent variables	Pre‐harvest								Post‐harvest
	LAZ		WAZ		WLZ		Hb		LAZ		WAZ		WLZ		Hb
Independent variables	β	p	β	p	β	p	β	p	β	p	β	p	β	p	β	p
N	159	—	156	—	156	—	159	—	164	—	164	—	164	—	162	—
R2	0.2	—	0.21	—	0.13	—	−0.02	—	0.13	—	0.12	—	0.07	—	0.04	—
Child sexa (males)	−0.08	0.28	−0.02	0.84	0.05	0.55	−0.13	0.16	0.04	0.57	−0.02	0.84	−0.03	0.68	0.01	0.88
Child age (months)	−0.31	<0.001	−0.28	<0.001	−0.25	<0.01	−0.03	0.62	−0.18	0.01	−0.15	0.04	−0.09	0.27	0.15	0.05
Child birthweight (kg)	0.01	0.87	0.16	0.04	0.2	0.01	0.10	0.30	0.04	0.60	0.05	0.52	0.05	0.48	−0.03	0.73
Maternal age (years)	0.36	0.01	0.38	0.01	0.23	0.11	−0.03	0.84	0.09	0.49	0.21	0.14	0.17	0.23	0.02	0.89
Maternal height (cm)	0.39	<0.001	0.28	<0.001	0.06	0.43	−0.05	0.77	0.36	<0.001	0.24	<0.001	0.08	0.31	0.04	0.63
Maternal BMI (kg/m2)	0.02	0.78	0.1	0.18	0.09	0.23	−0.01	0.9	0.02	0.77	0.18	0.03	0.21	0.01	0.09	0.22
Number of children	−0.34	0.01	−0.23	0.09	−0.07	0.62	0.11	0.43	0.05	0.71	−0.07	0.63	−0.10	0.48	0.03	0.84
First baby	0.11	0.23	0.15	0.1	0.09	0.34	0.13	0.18	−0.07	0.38	−0.02	0.85	0.03	0.77	−0.03	0.75
BF (within 1 hr)a	0.11	0.13	0.12	0.1	0.08	0.32	−0.05	0.57	0.16	0.02	0.15	0.04	0.11	0.17	−0.11	0.17
Fluids (within 3 days)	−0.08	0.28	−0.15	0.04	−0.13	0.11	−0.06	0.52	−0.08	0.28	−0.18	0.02	−0.19	0.01	0.03	0.57
EBF (weeks)	−0.11	0.18	−0.18	0.03	−0.15	0.08	0.13	0.21	−0.04	0.69	−0.09	0.31	−0.12	0.2	−0.05	0.49
BF on demanda	−0.05	0.58	0.02	0.84	0.04	0.71	0.03	0.77	—	—	—	—	—	—	—	—
BF frequency (10×)a	0.02	0.81	0.07	0.46	0.07	0.49	−0.05	0.52	—	—	—	—	—	—	—	—
Solids (months)b	—	—	—	—	—	—	—	—	0.04	0.68	0.02	0.79	0.03	0.73	0.06	0.50
Min acceptable dieta , b	—	—	—	—	—	—	—	—	−0.02	0.84	0.02	0.75	0.05	0.49	0.21	0.01

Note. All results adjusted for age, sex and birth weight of child; first baby, number of children, maternal BMI, maternal height (cm) and maternal age.

β: adjusted regression coefficient for the predictor values and p‐values; Hb: haemoglobin; LAZ: length‐for‐age z‐score (data not presented WLZ: weight‐for‐length z‐score; BF (10×): breastfeeding 10 or more times per day; EBF: exclusive breastfeeding.

^a Categorical variables are Yes/No answers with No answers being reference category, except for child sex where female is the reference.

^b Not included in the preharvest analysis.

The model for preharvest stunting (χ² = 33.111, p = 0.002) suggested that older children and children with more siblings were more likely to be stunted. However, children from older and taller mothers were less likely to be stunted. The model for preharvest underweight (χ² = 28.783, p = 0.004) suggested that older children and children who received fluids within 3 days of delivery were more likely to be underweight, and higher birth weight and initiation of breastfeeding within 1 hr of delivery were associated with a reduced risk of underweight. Postharvest, the logistic regression model predicting stunting (χ² = 28.5, p = 0.008) showed that mother's height continued to be a strong predictor of reduced risk of stunting, as was initiation of breastfeeding within 1 hr of delivery. The postharvest model predicting underweight (χ² = 23.881, p = 0.02) showed that mother's height was associated with reduced risk and feeding fluids within 3 days of delivery and late introduction of solid foods were associated with an increased risk of underweight (Table 5).

Table 5.

The association between infant feeding practices, maternal and child characteristics with stunting and underweight and anaemia

Dependent variables	Stunted	p	Preharvest	p	Anaemia	p	Stunted	p	Postharvest	p	Anaemia	p
			Underweight						Underweight
Independent variables	OR (95% CI)		OR (95% CI)		OR (95% CI)		OR (95% CI)		OR (95% CI)		OR (95% CI)
Child sexa (males)	1.49 (0.63–3.50)	0.35	0.59 (0.17–2.00)	0.39	1.34 (0.66–2.88)	0.36	0.75 (0.35–1.51)	0.39	1.75 (0.61–5.00)	0.29	1.11 (0.52–1.28)	0.82
Child age (months)	1.15 (1.02–1.25)	<0.001	1.19 (1.05–1.34)	0.01	1.03 (0.96–1.10)	0.38	1.07 (0.98–1.15)	0.10	1.02 (0.92–1.12)	0.68	0.91 (0.81–1.00	0.06
Child birthweight (kg)	1.04 (0.48–2.24)	0.92	0.29 (0.09–0.94)	0.03	0.89 (0.44–1.77)	0.74	1.02 (0.90–1.14)	0.78	0.86 (0.52–1.42)	0.55	1.00 (0.81–1.25)	0.93
Maternal age (years)	0.85 (0.76–0.95)	<0.01	0.95 (0.83–1.09)	0.48	0.99 (0.93–1.04)	0.77	1.02 (0.92–1.12)	0.71	0.89 (0.77–1.04)	0.14	0.89 (0.77–1.04)	0.93
Maternal height (cm)	0.90 (0.84–0.96)	<0.01	0.93 (0.86–1.02)	0.13	0.93 (0.86–1.02)	0.68	0.89 (0.84–0.95)	<0.001	0.92 (0.84–0.99)	0.02	0.96 (0.81–1.08)	0.54
Maternal BMI (kg/m2)	0.98 (0.87–1.11)	0.79	0.96 (0.82–1.14)	0.70	1.04 (0.94–1.15)	0.43	0.99 (0.89–1.11)	0.92	0.94 (0.79–1.10)	0.07	0.97 (0.85–1.05)	0.48
Number of children	1.56 (1.11–2.22)	0.01	1.04 (0.65–1.65)	0.86	0.93 (0.69–1.24)	0.59	0.89 (0.64–1.25)	0.50	1.22 (0.72–2.06)	0.44	0.94 (0.64–1.45)	0.88
BF (within 1 hr)	0.68 (0.29–1.58)	0.38	0.23 (0.06–0.81)	0.02	0.93 (0.46–1.87)	0.83	0.40 (0.18–0.86)	0.02	0.42 (0.14–1.25)	0.08	3.21 (1.02–9.50)	0.05
Fluidsa (within 3 days)	1.45 (0.62–3.38)	0.38	4.01 (1.21–13.34)	0.02	0.72 (0.34–1.49)	0.38	1.64 (0.76–3.58)	0.21	3.00 (1.03–8.89)	0.04	0.47 (0.13–1.17)	0.09
EBF (weeks)	1.06 (1.00–1.12)	0.03	1.04 (0.96–1.12)	0.32	0.95 (0.91–0.99)	0.04	0.99 (0.93–1.04)	0.74	0.99 (0.98–1.00)	0.26	1.10 (0.99–1.15)	0.09
BF on demanda,b	1.76 (0.53–5.97)	0.36	1.48 (0.31–7.02)	0.62	0.59 (0.20–1.70)	0.33	—	—	—	—	—	—
BF frequency (10x)b	0.76 (0.27–2.11)	0.61	0.42 (0.09–1.83)	0.24	1.00 (0.42–2.38)	0.98	—	—	—	—	—	—
Solids (months) c	—	—	—	—	—		0.99 (0.77–1.27)	0.93	1.58 (1.07–2.31)	0.02	0.83 (0.59–1.15)	0.27
Min acceptable dieta , c	—	—	—	—	—		0.58 (0.26–1.24)	0.16	1.44 (0.46–4.43)	0.52	0.52 (0.24–1.12)	0.11

Note. All results adjusted for age, sex and birth weight of child; first baby, number of children, maternal body mass index (BMI), maternal height (cm) and maternal age.

Adjusted odds ratio from binary logistic regression model with confidence interval in the parenthesis; BF: breastfeeding; CI: confidence interval; EBF: exclusive breastfeeding; OR: odds ratio.

^a Categorical variables with exception of child sex (female as reference); categorical variables are Yes/No answers with No answers being reference category.

^b Not included in the postharvest analysis.

^c Not included in the preharvest analysis.

5. DISCUSSION

This study examined IYCF practices and nutrition status of children below 2 years of age in rural Tanzania during preharvest and postharvest periods and examined the factors that contributed to infant nutrition status. Analysis highlighted the important associations between early breastfeeding practices, mother nutrition status and child nutrition status in different seasons. Overall, there was poor compliance with WHO IYCF guidelines including inadequate breastfeeding and complementary feeding practices in both seasons. A large proportion of infants received fluids other than breast milk within 3 days of delivery. Stunting was the major nutrition problem for children in both seasons increasing from preharvest to postharvest seasons. Noncompliance with early infant feeding guidelines was associated with poor nutrition status irrespective of season. Mother's age, height and BMI were all significant predictors of child anthropometrics in preharvest and postharvest.

5.1. Infants feeding practices

Although some of the findings from this cohort relate to recent national statistics, such as the proportion initiating breastfeeding within 1 hr of birth (National Nutrition Survey, 2014; UNICEF, 2015), others are quite different. For example, in this sample 26% of infants were EBF from 0 to 5 months compared with 59% reported at a national level (NBS and ICF Macro, 2016). However, other studies have reported similar findings; for example, a study conducted in Morogoro reported 22% of children were EBF at 4 months (Safari, Kimambo, & Lwelamira, 2013), and in Rukwa even lower rates of EBF at 6 months were reported (Nordang, Shoo, Holmboe‐Ottesen, Kinabo, & Wandel, 2015). Complementary feeding practices were also poor; however, in this case the findings reflect national data (Victor, Baines, Agho, & Dibley, 2014, TDHS‐MIS, 2015–2016). Although almost all infants aged 6–8 months received semisolid or soft foods only 20% met targets for minimum dietary diversity and less than 20% received any iron‐rich foods in the preharvest season. Although these figures improved postharvest, still less than 40% were meeting targets. Acknowledging that children were older postharvest, the increase in the proportion of children meeting complementary feeding indicators could potentially be due to seasonal changes in food availability and consumption. After harvest, there is more food available and access to food improves, particularly in rural regions that are dependent on agriculture for food and income (Arsenault et al., 2014; Wijesinha‐Bettoni, Kennedy, Dirorimwe, & Muehlhoff, 2013). The contrasting findings for breastfeeding practices and the poor adherence to complementary feeding recommendations combined with the seasonal differences in nutrition status indicators highlight the need for regional and/or rural focus when collecting similar data on IYCF practices and related health outcomes. Although there is a global focus on IYCF practices the data collected here and in other studies suggest that education and promotion are not reaching all communities, particularly rural communities (Nordang et al., 2015). This is supported by the fact that about 75% of the mothers in this study did not receive information about breastfeeding during antenatal visits. It is also clear that food intake and IYCF practices are sensitive to seasonal changes in food availability, which supports the existing literature calling for nutrition sensitive agricultural interventions to improve nutrition outcomes particularly in agricultural dependent households (Haddad, 2013; Masset, Haddad, Cornelius, & Isaza‐Castro, 2011).

5.2. Nutrition status of infants and young children

Nutrition status measurements highlighted a high prevalence of stunting and anaemia regardless of season, similar to recent national data, whereas wasting and underweight were less prevalent (TDHS‐MIS, 2015–2016). The decrease in LAZ and increase in stunting postharvest is particularly concerning when we expect more food is available in the postharvest. However, the overall change in LAZ is likely due to the age‐related decrease in LAZ reported previously as infants are 6 months older post‐harvest (Ali, Saaka, Adams, Kamwininaang, & Abizari, 2017; Megabiaw & Rahman, 2013), compounded by the timing of the post‐harvest survey, just 6–8 weeks after the start of harvest, so not much time for catch‐up growth (Maleta, Virtanen, Espo, Kulmala, & Ashorn, 2003). Interestingly, comparing changes depending on the child's nutrition status in the preharvest season highlights that those classified as severely or moderately malnourished show improvements in related z‐scores, whereas those classified as mildly malnourished only show improvements in WLZ, with no change in WAZ, and a significant decrease in LAZ. As the mildly malnourished and normal groups are the majority, we see these trends reflected when we compare z‐scores for the total group, a worsening of LAZ and higher prevalence of stunting, but improvements in WLZ and a slight decrease in the prevalence of wasting (p = 0.07). The difference in the change for length measurements compared with weight measurements suggests that catch‐up weight is more sensitive to changes in food availability than catch‐up linear growth. According to Maleta et al. (2003), linear growth increments typically occur 3 months after those for weight (Maleta et al., 2003).

Additionally, higher rates of stunting were reported for males compared with females. Similar findings in relation to age and gender have been reported previously (Wamani, Åstrøm, Peterson, Tumwine, & Tylleskär, 2006). The aetiology of stunting is multifactorial and includes a range of food, health and care practices classified as immediate, intermediate and underlying, which interact to influence child growth (Goudet, Griffiths, Bogin, & Madise, 2015). Therefore, stunting typically reflects the cumulative effects of poor nutrition, exposure to pathogens, underlying disease, altered gut microbiota and their interaction with the child's genotype (Pekmez, Dragsted, & Brahe, 2018; Prendergast & Humphrey, 2014). The combination of higher energy demands as children grow and inadequate food and nutrient intakes likely explains lower LAZ scores and increases in the prevalence of stunting as children get older (Dewey, 2016; Maleta et al., 2003). Similarly, Butte suggests that higher energy requirements might predispose males to increased risk of growth faltering compared with females (Butte, 2005). The alarmingly high rates of anaemia in this cohort preharvest (89%) and postharvest (83%) are another major cause for concern. The slight improvement postharvest likely reflects the improvements reported for the complementary feeding indicators and again reinforces diet sensitivity to seasonal changes.

5.3. Determinants of child nutrition outcome

In addition to reporting adherence to WHO IYCF recommendations and current nutrition status, this research explored the potential role of infant feeding practices in determining anthropometrics and iron status in both seasons. The results support the current infant feeding recommendations as well as previous research findings that early initiation of breastfeeding is associated with higher LAZ and WAZ and lower risk of stunting and underweight (Bentley et al., 2015; Kumar, Goel, Mittal, & Misra, 2006; Roche et al., 2016; Siddarth, 2007). Similarly, fluids other than breast milk given within 3 days after delivery was associated with lower WAZ and increased underweight, again supporting previous reports (Kumar et al., 2006). Delayed breastfeeding and/or introduction of water or other fluids in the early days after birth can impact overall energy intake for the infant and therefore impact growth (Kramer et al., 2004). In addition, the introduction of fluids other than breast milk put infants at risk of infection through the combined impact of contaminated water sources and depriving infants of the immune‐protective components available in breast milk (Hamosh, 2001; Kramer et al., 2004; Lawrence & Pane, 2007). Interestingly, early initiation of breastfeeding was more strongly associated with anthropometrics and nutrition status indicators postharvest compared with preharvest. This could be due to the fact that we see LAZ and higher prevalence of stunting as children get older or could also relate to the strong relationship between maternal characteristics and outcomes in pre‐harvest. In contrast, EBF duration was associated with an increased risk of stunting preharvest. However, this result should be interpreted with caution given the weaker odds ratio and the fact that this relationship did not hold true post‐harvest nor when included as an independent variable in a regression model with LAZ.

Regression analysis also highlighted the impact of mother and infant characteristics on anthropometric outcomes. Reflecting what we saw in the tables comparing age and gender groups, the regression analysis showed that males had a higher risk of stunting. As described previously the increased prevalence of stunting in males is partly explained by the higher energy demands but is mostly likely the result of a combination of interlinked factors. Several other studies, particularly in Sub‐Saharan Africa have reported similar higher rates of stunting in males (Wamani et al., 2006,Wamani, Åstrøm, Peterson, Tumwine, & Tylleskär, 2007,Keino, Plasqui, Ettyang, & van den Borne, 2014). Although some behaviour related explanations have been proposed, the underlying causes of why males appear more at risk of stunting compared with females is largely unknown.

Taller and older mothers and those with higher BMI were more likely to have children with higher LAZ and WAZ scores and consequently less likely to be stunted or underweight. Short maternal height is associated with increased risk of intrauterine growth restriction and low birth weight, which in turn leads to impaired child growth and accelerates the intergenerational cycles of malnutrition (Felisbino‐Mendes, Villamor, & Velasquez‐Melendez, 2014). Other research suggests that the link between mother's age and child nutrition status is likely due to the fact that younger mothers are at risk of undernutrition themselves. Younger mothers have higher nutrient requirements as they are still growing and may not have developed the coping skills required during times of food shortages, compared with older mothers (Fall et al., 2015). Indeed, several other studies have reported similar relationships between mother's nutrition status and child nutrition outcomes in developing countries (Arimond & Ruel, 2004; Menon, Bamezai, Subandoro, Ayoya, & Aguayo, 2015). In this study, the strength of relationship between maternal characteristics and child nutrition status was different depending on whether the measurement was taken preharvest or postharvest. For example, maternal age, maternal height and maternal BMI are significant predictors of LAZ in preharvest (Table 4), whereas only maternal height is significant postharvest. This suggests that depending on the child's age and on other aspects of the child's environment, different factors will have a stronger or weaker relationship with their nutrition status. Regardless, these findings again reinforce the need for early intervention to improve women's nutrition and health before, during and after pregnancy given the combined impact for child health outcomes (Walsh, 2013). Interventions that include promoting and supporting improved nutrition for adolescent girls and preventing early marriages and childbearing in rural households need to be addressed as part of long‐term solutions for child nutrition and health. Although these recommendations are not new (Bhutta et al., 2013; Daelmans et al., 2015), it is important that policy amendments and public health interventions target deliverable actions that will improve nutrition and health for young women, particularly in vulnerable subgroups such as those from rural farming backgrounds.

5.4. Strengths and limitations

This study reported findings from 220 households before and after the harvest season. The repeated measures design allowed for comparisons on several variables at critical periods in the year that relate to child health outcomes. However, it must be noted that the recall approach used in relation to certain survey questions was subject to respondent bias particularly when capturing retrospective data such as antenatal care and early infant feeding practices. In addition, the regression analysis was limited to include only those with complete data, which could potentially influence the observed results. Given the way in which WHO IYCF indicators are determined, it was not possible to include all the indicators as part of the regression analysis without significantly reducing the sample size. For this reason, only certain indicators were used in the regression models.

6. CONCLUSION

This study highlighted the extent of undernutrition and the related poor infant feeding practices in rural Tanzania. In particular, early infant feeding practices including early initiation and the introduction of fluids other than breast milk within 3 days of delivery were among the determinants of anthropometric z‐scores, stunting and undernutrition indices. Infant and maternal characteristics such as age, gender, birthweight, height and BMI were important contributors to regression models. In addition, infant feeding practices and nutrition outcomes were sensitive to seasonal related changes in food availability. Together, the findings from this research emphasize the importance of early targeted interventions with a focus on maternal health and nutrition status prior to and during pregnancy as well as targeting rural, vulnerable communities. Multicomponent interventions that aim to sustainably improve infant feeding and care practices, and food availability and diversity for all through education and incentives are needed to impact mother and infant health outcomes and the health and prosperity of entire households in the long‐term.

CONFLICTS OF INTEREST

The authors declare that they have no conflicts of interest.

CONTRIBUTIONS

HM, JK and AOS contributed to experimental design, data collection analysis and interpretation and manuscript writing.

Muhimbula H, Kinabo J, O'Sullivan A. Determinants of infant nutrition status in rural farming households before and after harvest. Matern Child Nutr. 2019;15:e12811 10.1111/mcn.12811