Introduction

Globally, there has been a call for healthcare workers, lactating mothers, and their support systems to support initiation (within 1 hour) and maintenance of exclusive breastfeeding (EBF) for at least 6 months and thereafter introduction of appropriate complementary feeding [1]. Breastfeeding is a lowcost intervention with numerous benefits to both mother and baby. Optimal breastfeeding reduces infection-related deaths in young infants [11] saving the lives of about 800,000 children [1]. EBF up to 6 months versus EBF for 3-4 months followed by mixed breastfeeding lowers the risk of gastrointestinal infection for the baby, increases maternal weight loss after birth, and delays the return of menstrual periods [12]. Breastfeeding is also critically important for the prevention of stunting and promoting neuromotor development [6-8]. The benefits of breastfeeding go beyond the health sector since families save money otherwise used to buy breastmilk substitutes [9]. Breastfeeding aligns with the 2030 agenda for sustainable development under the goals of good health and wellbeing (SDG 3) and responsible consumption and production (SDG 12).

Despite progress by many countries in achieving the WHO target of 50% infants EBF by the year 2025 [1], there is great variation between settings due to differences in socio-demographic, cultural and religious factors [2].  Kenya ascribes to treaties that uphold EBF for six months and also regulates the marketing and distribution of breast milk substitutes [10]. Following the creation of legal and policy frameworks, breastfeeding rates have improved from 32% in 2008 to 60% in 2022 [3]. Child mortality has fallen during this period which can be partly attributed to increased breastfeeding.

Factors associated with maintenance of EBF include older maternal age, higher level of maternal education, higher family income, married mother, previous experience of breastfeeding (more children), vaginal delivery, and non-smoking [13,14]. Further, Alzaheb [15] found that in addition to factors above and particularly influencing initiation were rooming-in and prelacteal feeding. However, there is a paucity of data on the frequency of EBF and reasons for interruption in deprived communities provided with comprehensive support for new mothers regarding breastfeeding and healthy child-care practices. Gaining understanding of how this support performs is important for the development of programmes to promote breastfeeding.

We performed a secondary analysis of data from infants participating in a trial of pro/synbiotic administration (PROSYNK) [16]. The trial used peer mothers who had been proven to be effective in delivering breastfeeding messages to improve postdischarge survival and growth of low birthweight infants [17].

Methods

The protocol for the PROSYNK study has been published previously [16].

Study Setting

The study was conducted in Homa-Bay County on the eastern shore of Lake Victoria in western Kenya. The county has a population of 1,131,950 [18] and covers an area of 3,154.7km². The population density is 359 people/km² with 262,036 households and an average of 4.3 members per household. The county has 8 sub-counties with 206 health facilities [19]. The main economic activities are subsistence farming, fishing, and small-scale trading. Most of the county’s population lives in deprivation and malnutrition is prevalent [3]. With a score of 0.46 on the Human Development Index (HDI), the county ranks below average in most socio-economic indicators. The population is rapidly growing; the fertility rate of 5.2 children per woman is higher than the national average of 4.6 children per woman [18]. The rate of EBF in the county is 38.4% [4].

The recruitment site was Homa-Bay County Teaching and Referral Hospital, with an average of 300 deliveries per month. This referral facility receives clients from the 8 sub-counties and from neighboring counties.

Study Design

This is a secondary analysis of data from a 4-arm, open label, individually randomized, controlled, phase-2, study of a probiotic and two synbiotics to improve gut health and reduce systemic inflammation [16].

Study Population

Six hundred newborns were recruited to PROSYNK between October 2020 to January 2022. Eligible newborns were singletons with a birthweight (BW) or current weight (if BW not known) ³2000 g, who were well and had taken at least one breast feed well, lived within the catchment area and informed consent had been secured from the mother/carer. STATA 16.1 was used to get a random sample of participant’s files with complete follow-up data up to 12 months. Consideration to ensure equal distribution in the four arms of the PROSYNK study (aiming at 91 in each study arm).

Study procedures

Newborns had been randomly allocated, stratified by maternal HIV status, 1:1:1:1 to three intervention arms and one control arm. All mothers/carers were visited daily for the first 10 days of life and thereafter weekly until infants were 6 months old for doi.org/10.29011/2577-2228.100403

pro/synbiotic administration. During home visits, communication of healthy infant-care practices including support for maintenance of EBF was provided. Infants in the control arm received the same visits as those in the intervention arm but without administration of pro/synbiotics.

Socio-demographic (maternal age, marital status, parity, mode of delivery, residence, maternal HIV status and reported smoking) and information on feeding including reasons for EBF interruption were collected by questionnaire at enrolment and during home visits at 6 weeks, 3 months, and 12 months. Anthropometric measurements were conducted at the same time points. Interrupted EBF was defined as the introduction of any fluid or food other than breastmilk at any point during the six months of close monitoring. Interruption of EBF occurring during 0-2 months was classified as “early” and occurring between 3-5 months as “late”.

Data reliability and validity

Use of standardized questionnaires, trained personnel, randomization to facilitate equal distribution of potential confounders, verification of data using source documents and consistency checks and feedback loops to address issues raised during the data abstraction process as detailed in a data management plan maximized data quality. In addition, inter-rater reliability regarding the assigned cause of interrupted EBF was performed by the principal investigator by sampling 10% of each data abstractor’s files for re-abstraction. A Cohen’s Kappa >0.65 was maintained for all files evaluated.

Data Analysis

Categorical variables were summarized according to absolute frequencies and percentages of subjects.

Association of potential categorical predictors of interrupted EBF (e.g. baseline socio-demographic factors) for both mother and infant were evaluated by Chi-square or Fisher’s exact tests. Binary logistic regression was used to generate unadjusted odds ratios for factors with significant differential distribution at α<0.05. Additionally, multivariable logistic regression was used to generate adjusted odds ratios and 95% confidence intervals for variables with significant differential distribution and longer duration of breastfeeding. Lastly, confounders in the relationship between duration of breastfeeding and predictors of breastfeeding interruption were assessed. STATA 16.1 was used for data analysis.

Sample Size Determination

Using Cochran’s sample size formula and assuming that the proportion EBF for the first six months of life was 60% [3] and with P=0.05 and a 5% margin of error, 366 infants were required.

Ethical Considerations

Trial participants had consented for additional research on data collected during the initial, approved study (KEMRI/SERU/ CGHR/320/3917; PPB/ECCT/20/04/02/2020(085); LSTMsponsor, 19-048). Permission to use the study data was granted by the sponsor in a letter dated 16th February 2023. Permission to conduct research was granted by Jaramogi Oginga Odinga University of Science and Technology Board of Post Graduate Studies and Ethical Review Board (ERC 35/12/23-15/07) in letters dated 2nd November 2022 and 10th January 2023 respectively. A research permit was obtained from the National Commission of Science and Technology (586326).

Analysis was limited to anonymized data with subjects identified only by their unique study number. Tablets and laptops were password protected and study documents kept in lockable cabinets. The findings of this study together with the finding of the main trial will be disseminated to stakeholders and the scientific community.

Results

Participant flow, baseline characteristics and level of EBF interruption

A total of 534/600 (89.0%) participants recruited in the PROSYNK study had complete follow up data to 12 months from whom 366 (61.0% of the total study cohort) were randomly selected to be included in this study (Figure 1).

Overall, most mothers were aged between 15-35 years, were married, resided in rural areas, and had a vaginal delivery (Table 1). About half of mothers had 3 or more children. There were 199 (54.4%) female and 167 male (45.6%) infants. The proportion of infants exposed to HIV (63; 17.2%) was similar to that of the whole PROSYNK study cohort (data not shown).

EBF interruption occurred in 45.1% (95% CI: 0.40-0.50) infants. Among those infants who had interrupted breastfeeding, majority of them 70.91% (95% CI: 0.63-0.78) were breastfed for longer (between 3-6 months), while less than one third of infants 29.09%, (95% CI: 0.22-0.37) interruption occurred early (within 3 months). Proportion of male and female infants in the interrupted EBF categories appeared not to be different from the total proportion. Mothers who interrupted EBF were younger and those who interrupted EBF early (within 3 months) had only one child (Table 1).

Table 1: Baseline characteristics/predictors of EBF of mother and infants according to EBF interruption category.

Predictors of EBF interruption

Maternal age at enrolment was not normally distributed, the overall median age was 25 years, the youngest mother was 12 years and the oldest was 43 years. Factors/baseline characteristics that were associated with EBF interruption in a similar fashion were also associated with duration of EBF (early vs late interruption). They included maternal age (p=0.01), marital status (p<0.01), and number of children (p=0.01), as shown in Table 1 above.

The infant baseline and socio-demographic characteristics evaluated; none was significantly associated with breastfeeding interruption (Table 1). Using the enter method, factors associated with EBF interruption by Chi square cut off 0.05 were included in a multivariable logistic regression model assessing for factors associated with longer breastfeeding duration. Based on the likelihood ratio test, the model containing all predictors was significant relative to null model [LR X2 (4) = 16.6, P=0.002]. In the univariate analyses, advancing maternal age (OR=1.1), higher number of children (OR=2.4) and being in a union (OR=4.1) were significantly associated with longer breastfeeding durations. Holding other factors constant, the odds of breastfeeding for longer was 4.7 (95% CI:1.1, 14) times for women in a union as compared to women in non-unions (Table 2).

Table 2: Odds ratios (crude and adjusted) for predictors of longer breastfeeding duration.

Reasons for EBF interruption

The reasons for EBF interruption were missing in 4/165 (2.4%) of participant’s files. Overall, the most common reason reported for EBF interruption was the feeling that breast milk was not enough to meet child’s nutritional needs (50.3% of lactating mothers; Table 3). School resumption was the second most common reason (26.1%), while resumption of work was reported as the third common reason (11.5%). Maternal illness and perceived child’s sudden refusal to breastfeed were less common reasons for EBF interruption.

Reasons for EBF interruption differed according to the timing of EBF interruption. Early interruption was mostly due to school resumption (47.9%) and this was the only reason that was significantly different between early and late interruption (P<0.001). Late interruption was mostly due to perception of breastmilk was not enough to meet growing baby’s feeding needs (52.9%; Table 3).

Table 3: Reasons for EBF interruption