Skip to main content
NCBI home page
Maternal & Child Nutrition logoLink to Maternal & Child Nutrition
. 2018 Jan 22;14(3):e12579. doi: 10.1111/mcn.12579

Measuring exclusive breastfeeding social support: Scale development and validation in Uganda

Godfred O Boateng 1, Stephanie L Martin 2, Shalean M Collins 1, Barnabas K Natamba 3, Sera L Young 1,4,✉
PMCID: PMC6025842  NIHMSID: NIHMS924594  PMID: 29356347

Abstract

The World Health Organization recommends exclusive breastfeeding (EBF) for 6 months and continued breastfeeding for at least 2 years. Social support has been widely recognized to influence breastfeeding practices. However, existing scales do not measure exclusive breastfeeding social support (EBFSS), rather they assess social support for any breastfeeding. Further, they are tailored towards high‐income settings. Therefore, our objectives were to develop and validate a tool to measure EBFSS in low‐income settings. To develop the scale, local and international breastfeeding experts were consulted on modifications to the Hughes' Breastfeeding Social Support Scale. It was then implemented in an observational cohort in Gulu, Uganda, at 1 (n = 238) and 3 (n = 237) months post‐partum (NCT02925429). We performed polychoric and polyserial correlations to remove redundant items and exploratory factor analysis at 1 month post‐partum to determine the latent factor structure of EBFSS. We further applied confirmatory factor analysis to assess dimensionality of the scale at 3 months post‐partum. We then conducted tests of predictive, convergent, and discriminant validity against EBF, self‐efficacy, general social support, and depression. The modification of the Hughes' scale resulted in 18 items, which were reduced to 16 after examining variances and factor loadings. Three dimensions of support emerged: Instrumental, Emotional, and Informational, with alpha coefficients of 0.79, 0.85, and 0.83, respectively. Predictive, convergent, and discriminant validity of the resultant EBFSS scale was supported. The EBFSS scale is valid and reliable for measuring EBFSS in northern Uganda and may be of use in other low‐income settings to assess determinants of EBF.

Keywords: exclusive breastfeeding, maternal health, scale development, social support, Uganda, validation

Key messages.

  • This is the first validated scale specific to social support for exclusive breastfeeding (vs. “any” breastfeeding).

  • The exclusive breastfeeding social support scale identifies Instrumental, Emotional, and Informational dimensions, with Cronbach's alphas of 0.78, 0.85, and 0.78, respectively.

  • Exclusive breastfeeding social support is significantly associated with exclusive breastfeeding.

  • This scale has great potential for adaptation in other settings.

1. INTRODUCTION

The World Health Organization recommends exclusive breastfeeding (EBF; i.e., providing only breast milk with no other foods or liquids, except oral rehydration therapy and medications) for the first 6 months of life and continued breastfeeding for 2 years because of the many short‐ and long‐term health benefits for mothers and children (Horta, Loret de Mola, & Victora, 2015; Rollins et al., 2016; Victora et al., 2015; WHO, 2002a). Although rates of continued breastfeeding at 12 months are high in low and lower–middle income countries, rates of EBF for the first 6 months in these same countries remain low (Victora et al., 2016).

This is the case in Uganda, for example, where despite near‐universal initiation and continuation of any breastfeeding at 6 months, only 43% of infants aged 4–5 months are exclusively breastfed (Uganda Bureau of Statistics & ICF, 2017). Qualitative research in Uganda has identified mothers' perceived breast milk insufficiency, in terms of quality and quantity, and inadequate time as key barriers to EBF (Engebretsen et al., 2007). Further analysis of Demographic and Health Survey data identified low literacy and education levels and giving birth at home as barriers to EBF (Ickes, Hurst, & Flax, 2015).

There are multiple determinants of exclusive and continued breastfeeding, which can be organized using an ecological model (Bentley, Dee, & Jensen, 2003). Social support, which can be provided at the interpersonal, community, and organization levels, is widely recognized for its influence on breastfeeding practices (Imdad, Yakoob, & Bhutta, 2011; Raj & Plichta, 1998; Rollins et al., 2016; U.S. Department of Health and Human Services, 2011). Lack of social support from family members has been identified as a barrier to EBF through qualitative research in Uganda (Ickes, Heymsfield, Wright, & Baguma, 2017) and Kenya (Kimani‐Murage et al., 2015), and social support was predictive of recommended infant feeding practices in a cross‐sectional suvey in Uganda (Ickes, Wu, Mandel, & Roberts, 2017). Peer support interventions have been shown to increase EBF rates in Uganda (Tylleskär et al., 2011). Further, in many developing countries, positive associations between lay and health worker support and EBF have been well documented (Imdad et al., 2011; Rollins et al., 2016). In the context of HIV, social support can also be quite important (Young et al., 2011).

Although the extant literature generally shows a positive association between breastfeeding and social support, there are some examples of negative associations (e.g., Meedya, Fahy, & Kable, 2010). These inconsistencies could be the result of methodological issues: social support instruments may not accurately measure the intended construct, be inadequately adapted for the context, or not assess support specific to EBF (Wills & Shinar, 2000). Hence, understanding the relationship between social support and EBF can inform the design of interventions that seek to increase EBF rates.

Existing instruments for measuring breastfeeding social support focus on social support for any breastfeeding and have typically been developed for high‐ and upper–middle income countries (Casal, Lei, Young, & Tuthill, 2017), with two exceptions. Hirani, Karmaliani, Christie, Parpio, and Rafique (2013) developed a scale to measure any breastfeeding support in Pakistan, and Zhu et al. (2013) developed a scale to measure family support of EBF in China. However, items used in the family support scale also measure support for any breastfeeding. In sum, a review of the literature did not identify any scales to measure exclusive breastfeeding social support (EBFSS) in low‐ and middle‐income countries.

However, the distinction between “breastfeeding social support” and “exclusive breastfeeding social support” is important, particularly in areas with predominant breastfeeding. For example, in East Africa, prolonged breastfeeding is widespread but so is the introduction of other foods and liquids before 6 months of age (Engebretsen et al., 2007; Wyatt, Yount, Null, Ramakrishnan, & Webb Girard, 2015). Therefore, a mother could receive high levels of social support for breastfeeding from her social network, but these same people could encourage her to give other foods and liquids early, ultimately discouraging EBF. If any breastfeeding social support was assessed in this example, it would suggest that the mother's high level of breastfeeding social support was negatively associated with EBF, even though support specifically for EBF was not measured. As such, there is a real need for a scale that measures the social support that women receive for EBF.

Therefore, we sought to adapt, construct, and validate an EBFSS scale for use in northern Uganda.

2. METHODS

2.1. Study settings and data collection

Data were collected in the context of Prenatal Nutrition and Psychosocial Health Outcomes Study (PreNAPS), ClinicalTrials.gov # NCT02922829, and Postnatal Nutrition and Psychosocial Health Outcomes Study (PostNAPS), ClinicalTrials.gov # NCT02925429, in Gulu, Uganda. Data were collected between October 10, 2012, and January 19, 2015, at Gulu Regional Referral Hospital. The goal of the parent study was to examine the relationships between food security, psychosocial health and nutrition during pregnancy and post‐partum.

Study procedures have been described elsewhere (Natamba et al., 2014, 2015; Widen et al., 2017). Briefly, 403 women who met the eligibility criteria: gestational age between 10 and 26 weeks, living <30 km from Gulu Regional Referral Hospital with known HIV status, were enrolled in the PreNAPS study and followed monthly during pregnancy. All PreNAPS participants who delivered after May 9, 2013, were invited to participate in PostNAPS if they had a live singleton birth, and all women accepted the invitation (n = 246). Data on EBF, breastfeeding self‐efficacy, depression, wealth, sociodemographic factors, and other health‐related outcomes were collected at 1 week, then 1, 3, 6, 9, and 12 months post‐partum.

2.2. Measuring EBF

EBF is defined 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)” (WHO, 2002b). In this study, participants were first questioned on their knowledge of and intention to EBF. They were then asked about infant feeding practices at delivery, since birth, and specifically about a range of non‐breast milk foods mostly fed to infants.

2.3. EBFSS item generation

The first step in scale construction and development is to specify the domain and to identify the set of items that make up that domain (Raykov, 2015). To measure EBFSS, we adapted the Hughes' (1984) Breastfeeding Social Support Scale, which was designed to measure perceived support for any breastfeeding among women in the south‐eastern United States. We adapted the scale to include items that reflected the support needed for EBF and the Ugandan context. The Hughes Breastfeeding Social Support Scale originally consisted of 30 items, which reflected three domains of social support: Emotional, Informational, and Instrumental. As part of assessing the content validity of the items, we consulted four experts in the fields of nutrition, public health, breastfeeding, and medicine to determine if the items were (a) appropriate or tangential indicators of EBF and (b) relevant to a low‐resource setting. These same experts also reviewed each of the items for clarity and intent.

2.4. Other survey items

Demographic data collected via survey included information on household size, parity, age, gravidity, and educational level. Wealth in tertiles was assessed using principal component analysis of participants' possession of 20 different household assets outlined in the socio‐economic module of the 2009–10 Uganda National Panel Survey (Natamba et al., 2014; Uganda Bureau of Statistics, 2010). Health information included maternal HIV status (determined at the post‐natal care clinic prior to enrollment into the PostNAPs study), maternal depression (assessed via the Center for Epidemiological Studies Depression [CES‐D] Scale), and maternal general social support (assessed through an adapted version of the Duke UNC Functional Social Support Questionnaire; Broadhead, Gehlbach, de Gruy, & Kaplan, 1988; Radloff, 1977; Tsai et al., 2012).

Other breastfeeding characteristics included assessments of participants' correct knowledge about EBF that was determined by two questions that asked how long a baby could thrive on breast milk alone and the best way to feed a baby for the first 6 months. We also measured EBF self‐efficacy using a questionnaire adapted from Dennis (2003), and how the infant was fed using a questionnaire on infant feeding practices at 1 and 3 months post‐partum.

2.5. Data analysis

Data from 1 month (n = 238) and 3 month (237) post‐partum visits were used. Data were analysed in six phases: (a) descriptive analysis, (b) item reduction, (c) extraction of factors, (d) test of dimensionality, (e) reliability, and (f) validity. Data were analysed using Mplus v. 7.40 (Los Angeles, CA, USA: Muthén & Muthén) and STATA v. 14 (College Station, TX, USA: StatCorp LP).

2.5.1. Descriptive analysis

We estimated the proportions, means, and standard deviations of the demographic, health, and breastfeeding behavior variables.

2.5.2. Item reduction

We then conducted item reduction to ensure that we only included a list of EBFSS items that were parsimonious and internally consistent (Thurstone, 1947). With a list of 18 items derived from expert content validation, we examined the proportions, mean, and variation between items for adequate variance. Because all of the 18 items were measured at the ordinal level, we estimated polychoric (inter‐item) and polyserial (item‐total) correlations to determine items that would be dropped from the tentative scale (Raykov & Marcoulides, 2011a, 2011b). The set of items obtained should correlate highly with the construct that they intend to measure (Edwards, 1957).

2.5.3. Extraction of factors

Extraction of factors is the stage in scale development in which the optimal number of factors that fit a set of items are determined. The emphasis is on the number of factors, the salience of factor loading estimates, and the relative magnitude of residual variances (Raykov, 2015; Raykov & Marcoulides, 2011a).

We used two techniques to identify the appropriate number of factors to retain from our set of items at 1 month post‐partum. Exploratory factor analysis (EFA) was used with the Guttman's (1954) eigenvalue rule of lower bound and Kaiser (1960) eigenvalue >1 rule, and Cattell's (1966) scree test to determine the number of factors to retain. Subsequently, item loadings below the threshold of 0.40 were dropped. Also, items with cross loadings greater than 0.40 were dropped (Nunnally, 1978). The extraction process used oblique rotation with weighted least squares with mean and variance adjustment (Muthén & Muthén, 2015). A number of model fit indices with satisfactory thresholds were used to determine meaningful model fitness for the data. The fit indices included chi‐square test of model fit, root mean square of error approximation (RMSEA ≤ 0.10), Tucker Lewis Index (TLI ≥ 0.95), comparative fit index (CFI ≥ 0.95), and standardized root mean square residual ≤ 0.08 (Bentler, 1990; Bentler & Bonett, 1980; Browne & Cudeck, 1993; Hu & Bentler, 1999; Kline, 2010; Tucker & Lewis, 1973). The results from this analysis provided the hypothesized factor structure to be tested at a later time point (i.e., from the 3‐month post‐partum visit).

2.5.4. Test of dimensionality

The test of dimensionality is a confirmatory test where the hypothesized factors or factor structure extracted from a previous model are tested (Brown, 2014). Our hypothesized factor structure was tested using confirmatory factor analysis (CFA) using oblique rotation with weighted least squares with mean and variance adjustment. A number of model fit indices with satisfactory thresholds were used to determine meaningful model fitness for the test of dimensionality. This included RMSEA (≤0.10), TLI (≥0.95), CFI (≥0.95), and weighted root mean square residual (WRMR < 1.0) (Bentler, 1990; Browne & Cudeck, 1993; Hu & Bentler, 1999; Muthén & Muthén, 2015; Tucker & Lewis, 1973).

2.5.5. Model modification

Model modification is used in CFA to improve model fitness using modification indices. It provides remedies for discrepancies between proposed and estimated models (Bollen, 1989; Kline, 2010). In this study, we made minor modifications to our models based on previous results, which did not show very strong support based on our model fit indices. To improve this, we used modification indices in Mplus (Bowen, 2014). We ensured that the modifications made were theoretically justifiable, few in number, and did not have a major impact on estimates of other parameters in the model (Bollen, 1989). We then examined the results for the largest modification indices and for the error terms associated with the observed indicators. The identified error terms were then correlated to improve model fitness (Bollen, 1989; Kline, 2010). With satisfactory model fit indices, we then tested the reliability and validity of the scale.

2.5.6. Reliability of EBFSS scale

Reliability is the degree of stability exhibited when a measurement is repeated under identical conditions (Porta, 2008). We assessed the reliability of the scale using coefficient alpha and coefficient of stability (Cronbach, 1951; Raykov & Marcoulides, 2011b). The coefficient alpha assesses the internal consistency of the scale, that is, the degree to which the set of items in the scale co‐vary, relative to their sum score (DeVellis, 2012; Raykov, 2015; Raykov & Marcoulides, 2011b). This was assessed for EBFSS at 1 and 3 months post‐partum using Cronbach's alpha. An alpha coefficient of 0.70 was determined as acceptable threshold for reliability; 0.80 and above is preferred for the psychometric quality of scales (Bernstein & Nunnally, 1994; Cronbach, 1951; Nunnally, 1978).

The coefficient of stability is used to assess the degree to which the participant's performance is repeatable, that is, how consistent their sum scores are across time. We assessed coefficient of stability through test–retest reliability (Raykov & Marcoulides, 2011b). This was indexed by a correlation coefficient of EBFSS at 1 and 3 months post‐partum.

2.5.7. Validity of EBFSS scale

Validity of the scale is the extent to which “an instrument is indeed measuring what it purports to measure” (Raykov & Marcoulides, 2011b). The final scale items and associated dimensions were tested for their predictive, convergent, and discriminant validity using data at 1 month post‐partum (n = 238).

Predictive validity, the extent to which tests scores predict criterion measurements to be made in the future, was assessed using Pearson product moment correlation with Fisher's transformation, linear, and logistic regression (DeVellis, 2012; Raykov & Marcoulides, 2011b). Predictive validity of all the dimensions of social support was assessed against EBF behaviour and an adapted measure of EBF self‐efficacy.

Convergent validity is the extent to which a construct measured in different ways yields similar results. Evidence of convergent validity of the construct is provided by the extent to which it correlates highly with other methods/variables designed to measure the same construct (Churchill, 1979; Raykov & Marcoulides, 2011b). In this study, we estimated a Pearson moments correlation with Fisher's transformation between scores of EBFSS and general social support.

Discriminant validity is the extent to which a measure is indeed novel and not simply a reflection of some other construct (Churchill, 1979). Discriminant validity is indicated by “predictably low correlations between the measure of interest and other measures that are supposedly not measuring the same variable or concept” (Churchill, 1979). We assessed discriminant validity through a Pearson's product–moment correlation of EBFSS and maternal depression, based on Fisher's transformation, with the hypothesis that there would be low or no correlations between the two factors.

We also assessed the difference between “known” groups (Churchill, 1979), that is, groups that we deemed to experience higher scores on EBFSS. This approach examines the distribution of a newly developed scale score over known binary items. We used maternal HIV status and level of education as our known groups. We hypothesized that HIV positive and participants with more than primary education would have significantly higher EBFSS mean scores than HIV negative and participants with less than primary education. Criterion concurrent validity was not assessed, as there was no “gold standard” for measuring EBFSS.

2.6. Ethics

Cornell University Institutional Review Board and Gulu University Institutional Review Committee approved ethics for human subject participation. Permission to conduct the study in Uganda was granted by the Ugandan National Council for Science and Technology. Study participants provided a written informed consent before they were enrolled into the study.

3. RESULTS

3.1. Scale adaptation

The adaptation of breastfeeding social support scale items produced an initial set of 18 items (SA 1). The experts who reviewed the scale items suggested 15 of the 30 items from Hughes' original scale that were not common among individuals in low‐resource settings or did not reflect that EBFSS had to be removed from the list. Three additional items were suggested to be added: (a) “approved of me exclusively breastfeeding my baby,” (b) “did tasks I would normally do so that I could exclusively breastfeed,” and (c) “made me feel good about my decision to exclusively breastfeed.” Based on the recommendation of experts, response categories were reduced from four to three and included (a) no help at all, (b) less than you would like, and (c) as much as you would like.

3.2. Participant characteristics

Participants had a mean age of 25.2 ranging from 16 to 42 years (Table 1). On average, each woman had two children with a mean household size of 4.6. One quarter were primiporous (23.1%), and 36.8% were HIV positive. As for education, approximately half (55.7%) had less than primary education, and one third (34.1%) of the participants were estimated to be poorer.

Table 1.

Univariate analysis of demographic and breastfeeding characteristics of mixed HIV women in Uganda at 1 month post‐partum (n = 238)

Socio‐demographic characteristics (range) N (%) | mean (SD)
Household size (1–13) 4.6 (2.2)
Number of children (0–8) 1.6 (1.5)
Maternal age (16–42) 25.2 (5.3)
Gravidity (1–10) 2.9 (1.8)
Primigravida (%) 55 (23.1)
HIV positive (%) 88 (36.8)
Maternal education (%)
Less than primary 131 (55.7)
Wealtha (%)
Poorer 82 (34.1)
Middle 80 (33.5)
General social support (10–30) 19.1 (4.2)
CESb‐depression scale (0–53) 18.5 (10.9)
Breastfeeding characteristics N (%) | mean (SD)
Correct breastfeeding knowledgec (%)
Yes 139 (58.2)
Insufficient milk production (%) 62 (30.9)
Exclusive breast feeding (EBF; 1 month) 151 (63.9)
EBF self‐efficacy score (37–86) 66.4 (10.1)
EBF social support scores (range) mean (SD)
Instrumental (1 month; 3–9) 4.8 (1.7)
Instrumental (3 months; 3–9) 4.9 (1.8)
Emotional (1 month; 8–24) 13.2 (3.7)
Emotional (3 months; 8–24) 15.2 (3.9)
Informational (1 month; 5–15) 7.4 (2.3)
Informational (3 months; 5–15) 7.9 (2.8)
Open in a new tab

Note. N = sample size; SD = standard deviation.

a

Wealth was developed out of an asset score grouped into poorer, middle, and richest. We present only the poorer and middle wealth categories.

b

CES = Center for Epidemiological Studies.

c

Women were considered as having correct breastfeeding knowledge if they answered 6 months to a question on the length of time that a child could thrive on breast milk and chose exclusive breastfeeding as the best way to feed a baby in the first 6 months of life.

The general social support scale produced a mean score of 19.1 ± 4.2, with a Cronbach's alpha of 0.86. Mean CES‐D score was 18.5, with a Cronbach's alpha of 0.89 (Table 1).

Of the 238 participants, 58.2% were considered as having adequate EBF knowledge. At 1 month post‐partum, 63.9% reported exclusively breastfeeding their children, and 30% reported insufficient breast milk production. The mean score for EBF self‐efficacy at 1 month post‐partum was 66.4 ± 10.1, with a Cronbach's alpha of 78.6.

Generally, women received far less social support for EBF than they would have liked for each of EBFSS items (Table 2; Figure 1). Items reflecting Emotional social support (EM on Figure 1) had the greatest proportion of participants that agreed that they received as much support as they would like. This included “‘Cared Well” (24.4%), “Good Mother” (20.6%), and “Approved EBF” (17.2%; see SA 1 for complete questions).

Table 2.

Univariate proportions of response categories and polychoric/polyserial correlation coefficients for exclusive breastfeeding social support items among women at 1 month post‐partum in Uganda (n = 238)

Scale items Response categories Item Polychoric correlation coefficients Polyserial correlation coefficients
As much as you would like Less than you would like No help at all M SD
Cared well 24.4 43.3 32.4 1.92 0.75 0.28–0.69 0.69
Good mother 20.6 44.1 35.3 1.85 0.73 0.25–0.59 0.75
Approved EBF 17.2 44.1 38.7 1.78 0.72 0.24–0.60 0.62
Showed EBF 15.1 34.0 50.8 1.64 0.73 0.37–0.65 0.62
Taught care 13.9 41.6 44.5 1.69 0.70 0.43–0.52 0.68
Did task 13.0 41.2 45.8 1.67 0.69 0.27–0.64 0.67
Meals 11.8 40.3 47.9 1.68 0.68 0.28–0.83 0.68
Cared baby 9.7 45.8 44.5 1.65 0.65 0.23–0.51 0.62
Laundry 8.4 29.8 61.8 1.47 0.65 0.19–0.38 0.54
Listened 8.0 39.5 52.5 1.55 0.64 0.34–0.63 0.71
Concern phy 7.6 47.1 45.4 1.62 0.62 0.38–0.75 0.67
Advice EBF 7.6 33.2 59.2 1.48 0.63 0.34–0.72 0.62
Answered Qs 6.7 18.5 74.8 1.32 0.59 0.30–0.70 0.72
Praised EBF 6.3 33.6 60.1 1.46 0.61 0.43–0.83 0.72
Concern sad 5.5 30.3 64.3 1.41 0.59 0.37–0.67 0.79
Feel confident 5.5 49.6 45.0 1.61 0.59 0.34–0.75 0.67
Feel good EBF 4.6 41.2 54.2 1.50 0.59 0.45–0.83 0.79
Get help 2.9 20.6 76.5 1.26 0.50 0.53–0.70 0.74
Open in a new tab

Note. M = mean; SD = standard deviation; Polychoric correlations = inter‐item correlations; Polyserial correlations = item‐total correlations; EBF = exclusive breastfeeding; questions are ordered from highest to lowest endorsement of items; items struck through were dropped from the list after exploratory factor analysis.

Figure 1.

Figure 1

Open in a new tab

Item response clustered bar graph showing the distribution of response categories for initial 18 exclusive breastfeeding (EBF) social support items at 1 month post‐partum (n = 238). INS = Instrumental social support; INF = Informational social support; EM = Emotional social support; items strike through were dropped from the list after exploratory factor analysis

Women felt that they received the least support for Informational social support items (INF on Figure 1). For instance, only 2.9% of the participants agreed that they received as much help as they would want on the item, “Get Help.” In fact, the majority of the participants reported receiving no help at all for “Answered Qs” and Get Help (74.8% and 76.5%). However, 13.9% to 15.1% reported receiving as much Informational support as they would like for “Showed EBF” and “Taught Care.”

Items that reflected Instrumental social support (INS on Figure 1) also had lower endorsements by participants. For example, 45.8% reported receiving no support at all for “Did Task,” 47.9% received no support with preparing meals, and 61.8% received no support with laundry. The proportion reporting receiving as much support as they would want ranged from 8.4% for support with laundry to 13.0% for support with performing tasks that women would do normally. Similar proportions were reported by participants at 3 months post‐partum (SA 2).

3.3. Item reduction

The average item‐total correlations (polyserial coefficients) for all 18 items were strong and ranged from .54 to .79 (Table 2). Inter‐item correlations (polychoric coefficients) for all the items ranged from .19 to .83. All the items had adequate variance, and the magnitudes of correlation coefficients were satisfactory; hence, none of the items were dropped at this analytical phase.

3.4. Extraction of factors

Three factors were identified from the 18 items (Table 3). The factor analysis using Geomin oblique rotation produced 3 eigenvalues (8.93, 1.66, and 1.28) > 1 (Table 4) and accounted for 66% of the variance in the data. An examination of the scree plot showed a steep curve that levelled off at factor number 3 with corresponding eigenvalue >1 (Figure 2). This also pointed to a three‐dimensional scale.

Table 3.

Factor loading results of exploratory factor analysis of 18 items indicative of exclusive breastfeeding social support in Uganda (n = 238)

Social support factors
Items 1 2 3
Instrumental Did task 0.55
Meals 0.90
Laundry 0.90
Emotional Approved EBF 0.41
Cared well 0.61
Feel confident 0.49
Listened 0.57
Good mother 0.69
Concern phy 0.82
Concern sad 0.77
Praised EBF 0.55
Informational Answered Qs 0.44
Advice EBF 0.67
Get help 0.79
Showed EBF 0.87
Taught care 0.74
Feel good EBF 0.00 0.50 0.50
Cared baby 0.31 0.22† 0.22†
Open in a new tab

Note. We acknowledge that all items in exploratory factor analysis will have loadings on all factors but show in this table only the largest factor loadings (cut‐off = 0.40) that are significant (*p ≤ .05). “Feel good EBF” was dropped because of cross loadings; “Cared baby” was dropped because factor loadings did not meet the cut‐off and were not significant (†) at (p ≤ .05).

Table 4.

Model fit indices of factor extraction at 1 month post‐partum and test of dimensionality at 3 months post‐partum

Factor extraction (n = 238)
Rotation Analytical technique χ 2 b df c RMSEA d CFI e TLI f SRMR g
Geomin oblique EFAa (18 items)
3 factors 233.87 102 0.074 0.97 0.95 0.06
Eigenvalues 8.93, 1.66, 1.28
Test of dimensionality (n = 237) χ2 df RMSEA CFI TLI WRMRh
Initial test (Geomin oblique) CFA (3 factors) 310.17 101 0.09 0.94 0.93 1.27
After specification (Geomin oblique) CFA (3 factors) 160.25 98 0.05 0.98 0.97 0.81
Open in a new tab

Note. All goodness‐of‐fit tests were statistically significant at p < .0001; Geomin oblique = type of rotation; all values in bold are used in estimating model fitness.

a

EFA = exploratory factor analysis.

b

χ2 = chi‐square goodness of fit statistic.

c

df = degrees of freedom.

d

RMSEA (<0.08) = root mean square error of approximation.

e

CFI (>0.95) = comparative fit index.

f

TLI (>0.95) = Tucker Lewis Index.

g

SRMR (<0.08) = standardized square root mean residual.

h

WRMR (<1.0) = weighted root mean square residual.

Figure 2.

Figure 2

Open in a new tab

Scree plot showing cut‐off point for retained scale factors using exploratory factor analysis (n = 238)

Upon examination of factor loadings, one item with a low factor loading (<0.40)—“Cared Baby”—and a second item “Feel Good EBF” with significant cross‐factor loading (>0.40) were dropped from the list of items (Table 3).

The remaining 16 items fell into three groups that we identified as Instrumental, Emotional, and Informational factors as assumed from the descriptive results. The Instrumental factor consisted of three items reflecting tangible support received by participants (e.g., “Did Task”). The Emotional factor consisted of seven items reflecting Emotional support (e.g., “Showed Concern”). The Informational factor consisted of six items that captured Informational support that was beneficial (e.g., “Answered Qs”). Each factor had a Cronbach's alpha value of greater than 0.70. All the model fit indices used in this study showed a very strong support for a three‐dimensional scale: RMSEA (0.07), CFI (0.97), TLI (0.95), and standardized root mean square residual (0.06; Table 4).

3.5. Test of dimensionality

Based on EFA results, we hypothesized that the 16‐item scale would represent a three‐dimensional scale. Using CFA, we tested this hypothesis with data from 3 months post‐partum (n = 237). The initial CFA results partially supported our model. The overall fit of this model for the initial test was poor on an absolute basis χ2 (101) = 310.17, p < .001; however, the descriptive model fit statistics provided partial support: RMSEA (0.09), CFI (0.94), TLI (0.93), and WRMR (1.27; Table 4).

3.6. Model modification

To improve model fit indices, we used modification indices in MPlus to determine which items needed respecification and which error terms needed to be correlated to reduce χ2 value (Bollen, 1989; Bowen, 2014). One of the items previously loading on Informational EBFSS was respecified to load on Emotional EBFSS (SA 3). Additionally, three pairs of items for each of the domains were partially correlated to increase model fitness (cf. footnote in SA 3 for full modification details).

After respecification of the model, Instrumental EBFSS consisted of three items, Emotional EBFSS consisted of eight items, and Informational EBFSS consisted of five items (Figure 3). Our model fit indices improved greatly without having a major impact on the estimates of other parameters in the model (Table 4). In general, the results indicated satisfactory model fitness for a three‐dimensional scale: χ2 (98) = 160.25, p < .001; RMSEA = 0.05; CFI = 0.98; TLI = 0.97; and WRMR = 0.81. Based on the results, a three‐dimensional scale was determined as an appropriate fit for our data giving us a finalized scale of 16 questions reflecting EBFSS (Table 5).

Figure 3.

Figure 3

Open in a new tab

Confirmatory factor analysis results showing standardized estimates with errors for a tri‐dimensional exclusive breastfeeding social support scale at 3 months post‐partum (n = 237)

Table 5.

Final 16‐item exclusive breastfeeding social support scale validated for use among women in northern Uganda

Item label Items
Domain Please describe the amount of help you received in each of the following areas during the past month:
Instrumental Did task Did tasks I would normally do so that I could exclusively breastfeed
Meals Prepared meals
Laundry Did laundry
Emotional Approved EBF Approved of me exclusively breastfeeding my baby
Cared well Told me I was doing well caring for my baby
Feel confident Made me feel confident even when I made mistakes
Listened Listened to me talk about the new baby
Good mother Believed that I am a good mother
Concern phy Showed concern about my own physical condition and health
Concern sad Showed concern when I felt sad or depressed
Praised EBF Praised me for my efforts to exclusively breastfeed
Informational Answered Qs Answered my questions about breastfeeding
Advice EBF Gave me advice and suggestions about how to exclusively breastfeed
Get help Told me where I could get help if I had questions about breastfeeding or caring for my baby
Showed EBF Showed me how to breastfeed
Taught care Taught me how to take care of myself
Open in a new tab

Note. Responses to these questions were categorized in three options: (a) no help at all or much less than you would like; (b) less than you would like; and (c) as much as you would like. EBF = exclusive breastfeeding.

We then added the scores for each of the three dimensions at 1 and 3 months post‐partum. At 1 month, Instrumental EBFSS had a mean of 4.78 and a range of 3–9; Emotional EBFSS had a mean of 13.21 and a range of 8–24; and Informational EBFSS had a mean 7.40 and a range of 5–15. The range for each of the subscales remained the same at 3 months post‐partum, but there were marginal increases for each of the scores (Table 1). Higher scores in each of subscales suggest greater social support received by the mother whereas lower scores indicate the receipt of lower social support.

3.7. Reliability

Reliability for the three subscales was measured via Cronbach's coefficient alpha and test–retest correlation at 1 and 3 months post‐partum.

The reliability test for the three revised subscales, Instrumental, Emotional, and Informational EBFSS scales produced respective Cronbach's alpha of 0.78, 0.85, and 0.78 at Month 1 and Cronbach's alpha of 0.78, 0.85, and 0.83 at Month 3 (SA 4). These values were consistently above the published satisfactory (0.70) and preferred (0.80) thresholds for scale reliability.

We further assessed test–retest reliability by correlating scores of each subscale at Months 1 and 3 to give us the coefficient of stability. Our estimation of each of the subscales produced a significant correlation coefficient .16 for Informational EBFSS, .31 for Emotional EBFSS, and .37 for Instrumental EBFSS (SA 4). All three coefficients were below the expected threshold of .70.

3.8. Validity

3.8.1. Predictive validity

For predictive validity, we regressed EBF self‐efficacy and EBF behaviour on each of the subscales (SA 5). Instrumental (β = 1.79; 95% confidence interval, CI, [1.05–2.52], p < .001), Informational (β = 1.29; 95% CI [0.98–1.59], p < .001), and Emotional (β = 1.33; 95% CI [0.80–1.85], p < .001) EBFSS were all predictive of EBF self‐efficacy. However, only Emotional breastfeeding social support was significantly associated with EBF behaviour (odds ratio = 1.12; 95% CI [1.03–1.21], p < .01). Based on these results, the predictive validity of our new scale was supported.

3.8.2. Convergent validity

We correlated the composite scores for each subscale with the composite score of general social support to assess the convergent validity of EBFSS (SA 5). The results showed statistical significant correlations between general social support and Instrumental (r = .41, 95% CI [0.30–0.51], p ≤ .001), Informational (r = .15, 95% CI [0.03–0.28], p ≤ .05), and Emotional (r = .25, 95% CI [0.13–0.37], p ≤ .001) EBFSS. Thus, convergent validity was supported.

3.8.3. Discriminant validity

We correlated the composite scores for each subscale with CES‐D scores to assess discriminant validity (SA 5). The correlational results between depression and Instrumental (r = −.24, 95% CI [−0.36, −0.12], p ≤ .001), Informational (r = −.05, 95% CI [−0.18, 0.08], p > .05), and Emotional (r = −.08, 95% CI [−0.21, −0.21], p ≤ .05) EBFSS suggested very low correlations existed between EBFSS scale and depression. These results supported discriminant validity, that is, EBFSS measured a construct, which was significantly different from depression.

3.8.4. Differences between known groups

Finally, we differentiated the position of “known groups,” that is, maternal HIV status and level of education, against the composites scores of all three subscales (SA 5). There were no significant differences between HIV positive and negative women on any of the EBFSS subscales. However, we found that participants with less than primary education had lower mean scores on Instrumental EBFSS (4.58 vs. 4.96; t = −1.75, p > .05) and significantly lower mean scores on Informational (7.09 vs. 7.78; t = −2.37, p ≤ .05) and Emotional (12.63 vs. 13.97; t = −2.79, p ≤ .01) breastfeeding social support than participants who had more than primary education. Thus, one out of the two outcome variables used showed significant differentiation between known groups, which is also an indicator of the scale validity.

4. DISCUSSION

The 16‐item EBFSS scale described here can be considered a valid and reliable metric for measuring EBFSS in northern Uganda. It is unique for its specificity to exclusive breastfeeding, as opposed to any breastfeeding. The EBFSS scale is the first to focus specifically on support for EBF in low‐resource settings.

EFAs and CFAs revealed three factors: Instrumental, Emotional, and Informational dimensions of social support (Tables 3 and 5). These parallel the three factors in Hughes' (1984) Breastfeeding Social Support Scale and social support construct more broadly. One difference is that we include far fewer items than the Hughes scale (16 vs. 30). This brevity represents a significant advantage, in that it requires fewer resources and less time to administer.

The EBFSS scale was reliable at 1 and 3 months post‐partum (SA 4). The Cronbach's (1951) alphas for all three subscales were above the published thresholds of 0.70 and 0.80 for the psychometric quality of scales (Bernstein & Nunnally, 1994). Compared to Hughes' original scale, the magnitude of the alpha coefficient calculated for Emotional breastfeeding social support was the same (0.85); however, the alpha coefficients for Informational and Instrumental breastfeeding social support were lower in magnitude, as they consisted of fewer items. The EBFSS alpha coefficients were comparable with existing scales measuring any breastfeeding social support, whose alpha coefficients ranging from 0.77 to 0.85 (Bai, Peng, & Fly, 2008; Grassley, Spencer, & Bryson, 2013; Hirani et al., 2013). Nonetheless, the rigour imposed in assessing the reliability of this scale surpasses that for any other breastfeeding social support scale, as none examines scale reliability longitudinally.

In terms of validity, we validated the EBFSS scale using predictive, convergent, and discriminant validity, as well as using differences between known groups (SA 5). Each form of validity supported the validity of our scale as a measure of EBFSS. These procedures were also more rigorous than had been used in prior breastfeeding support scales. At best, previous breastfeeding social support scales used two out of the four measures that we used for testing validity. For example, Grassley et al. (2013) and Hirani et al. (2013) estimated construct and predictive validity, with no discussion of convergent or discriminant validity. Further, Hughes (1984) discussed construct validity, but not predictive validity. No prior studies examined differences between known groups, and there were a few breastfeeding social support scales that did not discuss validity at all (Bai et al., 2008; Renee Matich & Sims, 1992).

The results of this study have programmatic and policy implications for increasing rates of EBF in northern Uganda. For one, our descriptive results of EBFSS scale items showed very low perceived Informational social support. Interventions to increase facility and family support for EBF have been effective at increasing EBF rates; and those that combine facility and community‐based approaches are most effective (Sinha et al., 2015). Our data suggest that if facility‐ and community‐based health workers and volunteers provide more information and support to mothers and other family members during the antenatal and post‐natal periods, support for EBF may increase. Our data also showed that mothers were not receiving the desired Emotional or Instrumental support that suggests the need for interventions to increase support for recommended infant feeding practices (Alive & Thrive, 2017; Aubel, 2012; Flax et al., 2014; Mukuria, Martin, Egondi, Bingham, & Thuita, 2016). Programmatically, when administered before delivery, the scale could help to identify mothers who may have challenges with EBF. Finally, this scale can be used to facilitate the evaluation of programs and enhance the development of policy‐relevant initiatives aimed at improving support for EBF.

There are a number of limitations to this study that are important to note. First, the correlational effect that we find between the different subscales at Months 1 and 3 may be due to common method variance (i.e., “variance that is attributable to the measurement method rather than the constructs the measures represent,” p. 879; Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Hence, it will be appropriate to test the hypothetical factor structure on a new sample to ascertain if they have the same meanings, latent factors, and factor loadings. Second, although we did find a significant association between EBF self‐efficacy scale and EBFSS scale, our measure of self‐efficacy was calculated using principal component analysis. A validated measure of EBF self‐efficacy would be more appropriate for exploring this relationship. Third, the lower scores of the coefficients of stability, which were below the threshold of reliability, reinforce the need to retest this scale in a new population.

Indeed, there is plenty of opportunity for further work on this scale. First, it will enhance our ability to use each of the subscales by themselves if each scale is tested for unidimensionality in another population within northern Uganda (Raykov, 2015).

Second, to increase the reliability, validity, and predictive effect of the scale, the scale should be tested in new populations in its entirety, starting with pregnant and post‐partum women in rural and urban settings in Uganda. For post‐partum women, this should be done within the first 6 months of delivery, as the scale seeks to promote EBF. The scale could, and should, be adapted for use in any low‐, middle‐, and high‐income country where knowledge about the importance of breastfeeding is common, but EBF is uncommon. With the Sustainable Development Goal 3 emphasizing ending preventable deaths of newborns by 2030, it will be interesting to explore if increasing EBFSS can better protect lives.

Other research opportunities include using this scale to establish causality between EBFSS and increased EBF. Furthermore, studies should proceed to establish a validated measure of EBF self‐efficacy, which we found to have a significant association with EBFSS, similar to McCarter‐Spaulding and Gore (2012). Lastly, it would be interesting to compare how EBFSS and BFSS scales work in other settings, with a goal of developing a cross‐culturally validated scale that has the same measurement, function, and meaning in different groups, across cultures, or over time.

In summary, this study has demonstrated the validity and reliability of an EBFSS that is relevant in a low‐resource setting where any breastfeeding is common. Its widespread adaptation should enhance our understanding of the correlates of EBFSS, and EBF itself, throughout the world.

CONFLICTS OF INTEREST

The authors declare that they have no conflicts of interest.

CONTRIBUTIONS

The authors' responsibilities were as follows: GOB and SLM wrote the introduction and key messages; GOB designed and performed the data analysis, wrote the methods, results, and discussion parts of the manuscript, and had primary responsibility for the final content of the manuscript; SLM prepared the EBF social support scale questions. SMC conducted data cleaning and management; BKN and SLY conceived and designed the PostNAPs study; BKN and SMC supervised the data collection; GOB, SLM, and BKN conceptualized the research question; GOB, SLM, SMC, BKN, and SLY provided critical review of the manuscript; all authors reviewed and contributed to the intellectual content of the manuscript, read and approved the final manuscript.

Supporting information

Data S1. Supporting info item

Click here for additional data file. (128.5KB, docx)

ACKNOWLEDGMENTS

We thank Angela Arbarch, Elizabeth Widen, the NAPS study team (Winnifred Achoko, Harriet Achola, Geoffrey Abwola, Daniel Achidri, Daniel Onen, Joe Cord, and Claire Biribiwa) and most of all the NAPS study participants.

Boateng GO, Martin SL, Collins SM, Natamba BK, Young SL. Measuring exclusive breastfeeding social support: Scale development and validation in Uganda. Matern Child Nutr. 2018;14:e12579 10.1111/mcn.12579

REFERENCES

  1. Alive & Thrive . (2017). Mass communication | Alive and Thrive. Available at: http://aliveandthrive.org/program-components/mass-communication/ (Accessed 14 June 2017).
  2. Aubel, J. (2012). The role and influence of grandmothers on child nutrition: Culturally designated advisors and caregivers. Maternal & Child Nutrition, 8(1), 19–35. 10.1111/j.1740-8709.2011.00333.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bai, Y. , Peng, C.‐Y. J. , & Fly, A. D. (2008). Validation of a short questionnaire to assess mothers' perception of workplace breastfeeding support. Journal of the American Dietetic Association, 108(7), 1221–1225. 10.1016/j.jada.2008.04.018. [DOI] [PubMed] [Google Scholar]
  4. Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107(2), 238–246. 10.1037/0033-2909.107.2.238. [DOI] [PubMed] [Google Scholar]
  5. Bentler, P. M. , & Bonett, D. G. (1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88(3), 588–606. 10.1037/0033-2909.88.3.588. [DOI] [Google Scholar]
  6. Bentley, M. E. , Dee, D. L. , & Jensen, J. L. (2003). Breastfeeding among low income, African–American women: Power, beliefs and decision making. The Journal of Nutrition, 133(1), 305S–309S. [DOI] [PubMed] [Google Scholar]
  7. Bernstein, I. , & Nunnally, J. C. (1994). Pyschometric theory. New York, NY: McGraw‐Hill. [Google Scholar]
  8. Bollen, K. A. (1989). Structural equations with latent variables. New York, NY: John Wiley & Sons. [Google Scholar]
  9. Bowen, N . (2014). Requesting and using modification indices in Mplus. Available at: https://www.google.com/#q=Requesting+and+Using+Modification+Indices+in+Mplus (Accessed 11 May 2017).
  10. Broadhead, W. E. , Gehlbach, S. H. , de Gruy, F. V. , & Kaplan, B. H. (1988). The Duke‐UNC Functional Social Support Questionnaire: Measurement of social support in family medicine patients. Medical Care, 26(7), 709–723. [DOI] [PubMed] [Google Scholar]
  11. Brown, T. (2014). Confirmatory factor analysis for applied research. New York, NY: Guildford Press. [Google Scholar]
  12. Browne, M. W. , & Cudeck, R. (1993). Alternative ways of assessing model fit In Bollen K. A., & Long J. S. (Eds.), Testing structural equation models (pp. 136–162). Newbury Park, CA: Sage publications. [Google Scholar]
  13. Casal, C. S. , Lei, A. , Young, S. L. , & Tuthill, E. L. (2017). A critical review of instruments measuring breastfeeding attitudes, knowledge, and social support. Journal of Human Lactation, 33(1), 21–47. 10.1177/0890334416677029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Cattell, R. B. (1966). The scree test for the number of factors. Multivariate Behavioral Research, 1(2), 245–276. [DOI] [PubMed] [Google Scholar]
  15. Churchill, G. A. (1979). A paradigm for developing better measures of marketing constructs. Journal of Marketing Research, 16(1), 64–73. 10.2307/3150876. [DOI] [Google Scholar]
  16. Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16(3), 297–334. 10.1007/BF02310555. [DOI] [Google Scholar]
  17. Dennis, C. (2003). The breastfeeding self‐efficacy scale: Psychometric assessment of the short form. Journal of Obstetric, Gynecologic & Neonatal Nursing, 32(6), 734–744. 10.1177/0884217503258459. [DOI] [PubMed] [Google Scholar]
  18. DeVellis, R. F. (2012). Scale development: Theory and application (3rd ed.). Los Angeles, CA: Sage Publications. [Google Scholar]
  19. Edwards, A. (1957). Technique of attitude scale construction. New York, NY: Appleton Century‐Crofts. [Google Scholar]
  20. Engebretsen, I. M. S. , Wamani, H. , Karamagi, C. , Semiyaga, N. , Tumwine, J. , & Tylleskär, T. (2007). Low adherence to exclusive breastfeeding in Eastern Uganda: A community‐based cross‐sectional study comparing dietary recall since birth with 24‐hour recall. BMC Pediatrics, 7(1), 10 10.1186/1471-2431-7-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Flax, V. L. , Negerie, M. , Ibrahim, A. U. , Leatherman, S. , Daza, E. J. , & Bentley, M. E. (2014). Integrating group counseling, cell phone messaging, and participant‐generated songs and dramas into a microcredit program increases Nigerian women's adherence to international breastfeeding recommendations. Journal of Nutrition, 144(7), 1120–1124. 10.3945/jn.113.190124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Grassley, J. S. , Spencer, B. S. , & Bryson, D. (2013). The development and psychometric testing of the Supportive Needs of Adolescents Breastfeeding Scale. Journal of Advanced Nursing, 69(3), 708–716. 10.1111/j.1365-2648.2012.06119.x. [DOI] [PubMed] [Google Scholar]
  23. Guttman, L. (1954). Some necessary conditions for common‐factor analysis. Psychometrika, 19(2), 149–161. 10.1007/BF02289162. [DOI] [Google Scholar]
  24. Hirani, S. A. A. , Karmaliani, R. , Christie, T. , Parpio, Y. , & Rafique, G. (2013). Perceived Breastfeeding Support Assessment Tool (PBSAT): Development and testing of psychometric properties with Pakistani urban working mothers. Midwifery, 29(6), 599–607. 10.1016/j.midw.2012.05.003. [DOI] [PubMed] [Google Scholar]
  25. Horta, B. L. , Loret de Mola, C. , & Victora, C. G. (2015). Breastfeeding and intelligence: A systematic review and meta‐analysis. Acta Paediatrica, 104, 14–19. 10.1111/apa.13139. [DOI] [PubMed] [Google Scholar]
  26. Hu, L. , & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. 10.1080/10705519909540118. [DOI] [Google Scholar]
  27. Hughes, R. B. (1984). The Development of an Instrument to Measure Perceived Emotional, Instrumental, and Informational Support in Breastfeeding Mothers. Issues in Comprehensive Pediatric Nursing, 7(6), 357–362. 10.3109/01460868409009772. [DOI] [PubMed] [Google Scholar]
  28. Ickes, S. B. , Heymsfield, G. A. , Wright, T. W. , & Baguma, C. (2017). “Generally the young mom suffers much:” Socio‐cultural influences of maternal capabilities and nutrition care in Uganda. Maternal & Child Nutrition, 13(3). 10.1111/mcn.12365. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Ickes, S. B. , Hurst, T. E. , & Flax, V. L. (2015). Maternal literacy, facility birth, and education are positively associated with better infant and young child feeding practices and nutritional status among Ugandan children. The Journal of Nutrition, 145(11), 2578–2586. 10.3945/jn.115.214346. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Ickes, S. B. , Wu, M. , Mandel, M. P. , & Roberts, A. C. (2017). Associations between social support, psychological well‐being, decision making, empowerment, infant and young child feeding, and nutritional status in Ugandan children ages 0 to 24 months. Maternal & Child Nutrition. 10.1111/mcn.12483. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Imdad, A. , Yakoob, M. Y. , & Bhutta, Z. A. (2011). Effect of breastfeeding promotion interventions on breastfeeding rates, with special focus on developing countries. BMC Public Health, 11(3), S24 10.1186/1471-2458-11-S3-S24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Kaiser, H. F. (1960). The application of electronic computers to factor analysis. Educational and Psychological Measurement, 20, 141–151. 10.1177/001316446002000116. [DOI] [Google Scholar]
  33. Kimani‐Murage, E. W. , Wekesah, F. , Wanjohi, M. , Kyobutungi, C. , Ezeh, A. C. , Musoke, R. N. , … Griffiths, P. (2015). Factors affecting actualisation of the WHO breastfeeding recommendations in urban poor settings in Kenya. Maternal & Child Nutrition, 11(3), 314–332. 10.1111/mcn.12161. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Kline, R. B. (2010). Principles and practice of structural equation modeling (3rd ed.). New York: The Guildford Press. [Google Scholar]
  35. McCarter‐Spaulding, D. , & Gore, R. (2012). Social support improves breastfeeding self‐efficacy in a sample of Black women. Clinical Lactation, 3(3), 112–115. 10.1891/215805312807022923. [DOI] [Google Scholar]
  36. Meedya, S. , Fahy, K. , & Kable, A. (2010). Factors that positively influence breastfeeding duration to 6 months: A literature review. Women and Birth, 23(4), 135–145. 10.1016/j.wombi.2010.02.002. [DOI] [PubMed] [Google Scholar]
  37. Mukuria, A. G. , Martin, S. L. , Egondi, T. , Bingham, A. , & Thuita, F. M. (2016). Role of social support in improving infant feeding practices in Western Kenya: A quasi‐experimental study. Global Health: Science and Practice, 4(1), 55–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Muthén, L. , & Muthén, B. (2015). Mplus user's guide (seventh). Los Angeles, CA: Muthén & Muthén. [Google Scholar]
  39. Natamba, B. K. , Achan, J. , Arbach, A. , Oyok, T. O. , Ghosh, S. , Mehta, S. , … Young, S. L. (2014). Reliability and validity of the center for epidemiologic studies‐depression scale in screening for depression among HIV‐infected and ‐uninfected pregnant women attending antenatal services in northern Uganda: A cross‐sectional study. BMC Psychiatry, 14(1), 303 10.1186/s12888-014-0303-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Natamba, B. K. , Kilama, H. , Arbach, A. , Achan, J. , Griffiths, J. K. , & Young, S. L. (2015). Reliability and validity of an individually focused food insecurity access scale for assessing inadequate access to food among pregnant Ugandan women of mixed HIV status. Public Health Nutrition, 18(16), 2895–2905. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Nunnally, J. C. (1978). Pyschometric theory. New York, NY: McGraw‐Hill. [Google Scholar]
  42. Podsakoff, P. M. , MacKenzie, S. B. , Lee, J.‐Y. , & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879–903. 10.1037/0021-9010.88.5.879. [DOI] [PubMed] [Google Scholar]
  43. Porta, M. (2008). A dictionary of epidemiology. New York, NY: Oxford University Press. [Google Scholar]
  44. Radloff, L. S. (1977). The CES‐D scale: A self‐report depression scale for research in the general population. Applied Psychological Measurement, 1(3), 385–401. 10.1177/014662167700100306. [DOI] [Google Scholar]
  45. Raj, V. K. , & Plichta, S. B. (1998). The role of social support in breastfeeding promotion: A literature review. Journal of Human Lactation, 14(1), 41–45. 10.1177/089033449801400114. [DOI] [PubMed] [Google Scholar]
  46. Raykov, T. (2015). Scale construction and development (lecture notes). East Lansing, MI: Michigan State University. [Google Scholar]
  47. Raykov, T. , & Marcoulides, G. A. (2011a). Classical item analysis using latent variable modeling: A note on a direct evaluation procedure. Structural Equation Modeling: A Multidisciplinary Journal, 18(2), 315–324. 10.1080/10705511.2011.557347. [DOI] [Google Scholar]
  48. Raykov, T. , & Marcoulides, G. A. (2011b). Introduction to psychometric theory. New York, NY: Routledge, Taylor & Francis Group. [Google Scholar]
  49. Renee Matich, J. , & Sims, L. S. (1992). A comparison of social support variables between women who intend to breast or bottle feed. Social Science & Medicine, 34(8), 919–927. 10.1016/0277-9536(92)90260-W. [DOI] [PubMed] [Google Scholar]
  50. Rollins, N. C. , Bhandari, N. , Hajeebhoy, N. , Horton, S. , Lutter, C. K. , Martines, J. C. , … Group, T. L. B. S. (2016). Why invest, and what it will take to improve breastfeeding practices? The Lancet, 387(10017), 491–504. 10.1016/S0140-6736(15)01044-2. [DOI] [PubMed] [Google Scholar]
  51. Sinha, B. , Chowdhury, R. , Sankar, M. J. , Martines, J. , Taneja, S. , Mazumder, S. , … Bhandari, N. (2015). Interventions to improve breastfeeding outcomes: A systematic review and meta‐analysis. Acta Paediatrica, 104, 114–134. 10.1111/apa.13127. [DOI] [PubMed] [Google Scholar]
  52. Thurstone, L. (1947). Multiple‐factor analysis. Chicago: University of Chicago Press. [Google Scholar]
  53. Tsai, A. C. , Bangsberg, D. R. , Frongillo, E. A. , Hunt, P. W. , Muzoora, C. , Martin, J. N. , & Weiser, S. D. (2012). Food insecurity, depression and the modifying role of social support among people living with HIV/AIDS in rural Uganda. Social Science & Medicine, 74(12), 2012–2019. 10.1016/j.socscimed.2012.02.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Tucker, L. R. , & Lewis, C. (1973). A reliability coefficient for maximum likelihood factor analysis. Psychometrika, 38(1), 1–10. 10.1007/BF02291170. [DOI] [Google Scholar]
  55. Tylleskär, T. , Jackson, D. , Meda, N. , Engebretsen, I. M. S. , Chopra, M. , Diallo, A. H. , … Kankasa, C. (2011). Exclusive breastfeeding promotion by peer counsellors in sub‐Saharan Africa (PROMISE‐EBF): A cluster‐randomised trial. The Lancet, 378(9789), 420–427. [DOI] [PubMed] [Google Scholar]
  56. U.S. Department of Health and Human Services (2011). Executive summary the Surgeon General's call to action to support breastfeeding. Breastfeeding Medicine, 6(1), 3–5. 10.1089/bfm.2011.9996. [DOI] [PubMed] [Google Scholar]
  57. Uganda Bureau of Statistics . (2010). The Uganda National Panel Survey 2009/10: Household Questionnaire. Kampala. Accessed at: https://www.google.com/#q=The+Uganda+National+Panel+Survey+2009/10:+Household+Questionnaire (Accessed 10 May 2017).
  58. Uganda Bureau of Statistics & ICF . (2017). Demographic and Health Survey 2016: Key Indicators Report. Kampala, Uganda.
  59. Victora, C. G. , Bahl, R. , Barros, A. J. D. , França, G. V. A. , Horton, S. , Krasevec, J. , … Group, T. L. B. S. (2016). Breastfeeding in the 21st century: Epidemiology, mechanisms, and lifelong effect. The Lancet, 387(10017), 475–490. 10.1016/S0140-6736(15)01024-7. [DOI] [PubMed] [Google Scholar]
  60. Victora, C. G. , Horta, B. L. , de Mola, C. L. , Quevedo, L. , Pinheiro, R. T. , Gigante, D. P. , … Barros, F. C. (2015). Association between breastfeeding and intelligence, educational attainment, and income at 30 years of age: A prospective birth cohort study from Brazil. The Lancet Global Health, 3(4), e199–e205. 10.1016/S2214-109X(15)70002-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. WHO . (2002a). WHO | Nutrient adequacy of exclusive breastfeeding for the term infant during the first six months of life. Available at: http://www.who.int/nutrition/publications/infantfeeding/9241562110/en/ (Accessed 25 May 2017).
  62. WHO . (2002b). The World Health Organization's infant feeding recommendation. World Health Organization. Available at: http://www.who.int/nutrition/topics/infantfeeding_recommendation/en/ (Accessed 25 May 2017).
  63. Widen, E. M. , Collins, S. M. , Khan, H. , Biribawa, C. , Acidri, D. , Achoko, W. , … Young, S. L. (2017). Food insecurity, but not HIV‐infection status, is associated with adverse changes in body composition during lactation in Ugandan women of mixed HIV status. The American Journal of Clinical Nutrition, ajcn142513. 10.3945/ajcn.116.142513. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Wills, T. A. , & Shinar, O. (2000). Measuring Perceived and Received Social Support In Cohen S, Underwood L. G., & Gottlieb B. H. (Eds.), Social support measurement and intervention. A guide for health and social scientists (pp. 86–135). New York, NY: Oxford University Press. [Google Scholar]
  65. Wyatt, A. J. , Yount, K. M. , Null, C. , Ramakrishnan, U. , & Webb Girard, A. (2015). Dairy intensification, mothers and children: An exploration of infant and young child feeding practices among rural dairy farmers in Kenya. Maternal & Child Nutrition, 11(1), 88–103. 10.1111/mcn.12074. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Young, S. L. , Mbuya, M. N. N. , Chantry, C. J. , Geubbels, E. P. , Israel‐Ballard, K. , Cohan, D. , … Latham, M. C. (2011). Current knowledge and future research on infant feeding in the context of HIV: Basic, clinical, behavioral, and programmatic perspectives. Advances in Nutrition: An International Review Journal, 2(3), 225–243. 10.3945/an.110.000224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Zhu, X. , Li, Y. , & Lu, H. (2013). Reliability and validity of breastfeeding family support questionnaire. Chinese Journal of Modern Nursing, 19(7), 775–778. 10.3760/cma.j.issn.1674-2907.2013.07.010 [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data S1. Supporting info item

Click here for additional data file. (128.5KB, docx)

Articles from Maternal & Child Nutrition are provided here courtesy of Wiley

RESOURCES