An integrated infant and young child feeding and small‐quantity lipid‐based nutrient supplementation programme in the Democratic Republic of Congo is associated with improvements in breastfeeding and handwashing behaviours but not dietary diversity

Abstract Integrating small‐quantity lipid‐based nutrient supplements (SQ‐LNS) into infant and young child feeding (IYCF) programmes can increase consumption of essential nutrients among children in vulnerable populations; however, few studies have assessed the impact of integrated IYCF–SQ‐LNS programmes on IYCF practices. A 2‐year, enhanced IYCF intervention targeting pregnant women and infants (0–12 months) was implemented in a health zone in the Democratic Republic of Congo (DRC). The enhanced IYCF intervention included community‐ and facility‐based counselling for mothers on handwashing, SQ‐LNS, and IYCF practices, plus monthly SQ‐LNS distributions for children 6–12 months; a control zone received the national IYCF programme (facility‐based IYCF counselling with no SQ‐LNS distributions). Cross‐sectional preintervention and postintervention surveys (n = 650 and 638 in intervention and control areas at baseline; n = 654 and 653 in each area at endline, respectively) were conducted in mothers of children 6–18 months representative of both zones. Difference in differences (DiD) analyses used mixed linear regression models. There were significantly greater increases in the proportion of mothers in the intervention (vs. control) zone who reported: initiating breastfeeding within 1 hr of birth (Adj. DiD [95% CI]: +56.4% [49.3, 63.4], P < 0.001), waiting until 6 months to introduce water (+66.9% [60.6, 73.2], P < 0.001) and complementary foods (+56.4% [49.3, 63.4], P < 0.001), feeding the minimum meal frequency the previous day (+9.2% [2.7, 15.7], P = 0.005); feeding the child in a separate bowl (+9.7% [2.2, 17.2], P = 0.01); awareness of anaemia (+16.9% [10.4, 23.3], P < 0.001); owning soap (+14.9% [8.3, 21.5], P < 0.001); and washing hands after defecating and before cooking and feeding the child the previous day (+10.5% [5.8, 15.2], +12.5% [9.3, 15.6] and +15.0% [11.2, 18.8], respectively, P < 0.001 for all). The enhanced IYCF intervention in the DRC was associated with an improvement in several important IYCF practices but was not associated with a change in dietary diversity (minimum dietary diversity and minimum acceptable diet remained below 10% in both zones without significant differences between zones). The provision of fortified complementary foods, such as SQ‐LNS, may be an important source of micronutrients and macronutrients for young children in areas with high rates of poverty and limited access to diverse foods. Future research should verify the potential of integrated IYCF–SQ‐LNS to improve IYCF practices, and ultimately children's nutritional status.

(DRC) Ministry of Public Health and National Nutrition Program (PRONANUT) 21.5], P < 0.001); and washing hands after defecating and before cooking and feeding the child the previous day (+10. 5% [5.8, 15.2], +12. 5% [9.3, 15.6] and +15. 0% [11.2, 18.8], respectively, P < 0.001 for all). The enhanced IYCF intervention in the DRC was associated with an improvement in several important IYCF practices but was not associated with a change in dietary diversity (minimum dietary diversity and minimum acceptable diet remained below 10% in both zones without significant differences between zones). The provision of fortified complementary foods, such as SQ-LNS, may be an important source of micronutrients and macronutrients for young children in areas with high rates of poverty and limited access to diverse foods.
Future research should verify the potential of integrated IYCF-SQ-LNS to improve IYCF practices, and ultimately children's nutritional status.

| INTRODUCTION
Malnutrition during the critical period from conception to 2 years of age increases the risk of premature death and has lifelong developmental and health consequences (Dewey & Brown, 2003). Optimal infant and young child feeding (IYCF) practices-including initiation of breastfeeding in the first hour of birth, exclusive breastfeeding for the first 6 months of life, and the introduction of micronutrientrich, age-appropriate foods from 6 months-are particularly important for ensuring the healthy growth and development of young children (Disha, Rawat, Subandoro, & Menon, 2012;Dykes & Hall-Moran, 2009;Jones et al., 2014;Menon, Bamezai, Subandoro, Ayoya, & Aguayo, 2015;PAHO/WHO, 2003). In 2010, the United Nations Children's Emergency Fund (UNICEF) introduced the communitybased IYCF counselling package, which includes several planning, training, and counselling tools to support programming and capacity development on community-based IYCF counselling (UNICEF, 2010).
As of 2015, 60 countries had incorporated elements of the package; however, there has been little research assessing the use of this package at scale (Lamstein, 2017).
In settings with high rates of poverty and limited access to diverse diets, the distribution of nutritional supplements that are rich in micronutrients and macronutrients may improve IYCF practices and ultimately children's nutritional status (Dewey & Adu-Afarwuah, 2008). Small-quantity lipid-based nutrient supplements (SQ-LNS) are sachets containing ≤20 g (≤120 kcal) of a lipid-based supplement, usually composed of peanuts, vegetable oil, milk powder, sugar, and added vitamins and minerals. In order to minimize disruption of IYCF practices, it is often recommended that SQ-LNS is mixed into local The enhanced IYCF intervention included facility-and communitybased counselling on handwashing, SQ-LNS, and IYCF, as well as the monthly distribution of SQ-LNS for children 6-12 months.
Randomized trials have indicated that daily SQ-LNS as part of complementary feeding can improve child growth (Adu-Afarwuah et al., 2007;Christian et al., 2015;Hess et al., 2015;Iannotti et al., 2014) and motor development (Adu-Afarwuah et al., 2007). An effectiveness trial in Bangladesh also showed that SQ-LNS could be provided at scale to pregnant and lactating women and to their children (from 6 to 24 months) with ultimate improvements in linear growth among children . However, there is no research on the integration of SQ-LNS into national or subnational IYCF programmes. Integrating SQ-LNS into IYCF programmes could contribute to increased political, financial, community, and human resources devoted to IYCF and may ultimately improve IYCF practices, as has been hypothesized about the integration of micronutrient powders (MNP) into IYCF programmes (Siekmans, Bégin, Situma, & Kupka, 2017). To date, a few integrated IYCF-MNP programmes have documented improved IYCF practices in diverse settings (Locks et al., 2017;Locks et al., 2018;Mirkovic et al., 2013). Only one study has assessed the impact of SQ-LNS distributions on IYCF practices, and this study focused on results from randomized trials of SQ-LNS where IYCF behaviour change was not part of the intervention (Arimond et al., 2017). Our current analysis focuses on the impact of the DRC enhanced IYCF programme (an integrated IYCF-SQ-LNS programme) on IYCF and handwashing practices.

| Programme implementation and study participants
The enhanced IYCF programme was piloted in Haut-Katanga District in Katanga Province; two health zones were selected, one as an intervention area (Kasenga) and one as a control area (Kipushi). Cross-sectional baseline and endline surveys were conducted in both areas. The zones were selected for programmatic purposes with the intention of implementing the programme in the control area if the evaluation demonstrated an impact on childhood stunting and anaemia. The two zones were selected on the basis of the following specific criteria: (a) health areas in Haut-Katanga District with large populations, (b) nonadjacency to Lubumbashi (the provincial capital), and (c) nonadjacency to each other. Distance from Lubumbashi was prioritized given the potential popularity of free SQ-LNS distributions and the programmatic costs of distributions to children who resided outside of the intervention zone. Similarly, geographic distance between the intervention area and control area was essential to prevent spillover. Kasenga and Kipushi were the only two zones that fulfilled the selection criteria; however, there were some key differences in the areas. Kasenga (the intervention area) is a rural, ethnically homogenous (Bemba) area with an agricultural economy, whereas Kipushi (the control area) is a mining area that is home to several different ethnic groups.
The control area received the government IYCF strategy based on the Essential Nutrition Actions (World Health Organization, 2013), whereas the intervention area received the enhanced IYCF intervention (Table 1). Site visits and key informant interviews in both areas prior to developing the intervention protocol revealed that in both areas, counselling on IYCF was provided in some health facilities, but that the coverage and training of health workers varied substantially across facilities and the use of job aids was rare. Community health workers (CHWs) were present in both the intervention and control areas (primarily to support vaccination, vitamin A supplementation, and growth monitoring programmes); however, they generally did not provide IYCF counselling.
The 2-year enhanced IYCF programme targeted all 23,000 pregnant women and infants 0-12 months in Kasenga, using an expanded and locally adapted version of the UNICEF community-based IYCF programme tools. The enhanced IYCF programme also included monthly distributions of SQ-LNS for children 6-12 months and counselling for their mothers on the appropriate use of SQ-LNS. All training and informational materials were developed on the basis of extensive formative research on the knowledge, attitudes, practices, and barriers to optimal IYCF practices. The formative research included focus groups and key informant interviews with mothers, fathers, health workers, and district officials as well as market visits to identify which nutrient-rich foods were available during different times of the year (Tripp et al., 2015). In the intervention area, 30 health workers and 286 CHWs were trained to participate in the enhanced IYCF programme. In addition to receiving IYCF-SQ-LNS training and counselling tools, CHWs in the intervention area were also given bikes to improve their ability to travel to remote areas and reinforce their community presence.
In the intervention area, mothers of children aged 6-12 months were expected to receive monthly distributions of SQ-LNS from health facilities as well as community-and facility-based counselling on its appropriate use. SQ-LNS was locally branded as "Kulabora," which translates to "eating better." Monthly distributions composed of four strips of seven sachets (totalling 28); each of the seven sachets contained images and text supporting one of seven key messages on Kulabora use, which also re-enforced key IYCF and handwashing messages: (a) one packet per child per day, (b) wash your child's hands with soap and water before feeding, (c) breastfeed your child before giving food, (d) put a small amount of food that you think your child will eat in a separate bowl, (e) mix the Kulabora into the food, (f) feed the food mixed with the Kulabora to your child, and (g) Kulabora is for children from 6 to 12 months of age. • Fortified foods may fill an important nutrient gap in areas with high rates of poverty and food insecurity; however, future research must assess whether integrated IYCF-SQ-LNS programmes or programmes providing other micronutrient-rich food supplements can improve children's nutritional status.

| Survey methodology
• Mothers in the intervention area were substantially more likely to report receiving IYCF messages from community health workers (CHWs) than mothers in the control area.  approximately equal segments, and small clusters were combined with nearby villages. A list of all children 6-17.9 months in each selected cluster was then developed, and 22 children from each cluster were randomly selected; there were no replacements for households who refused to participate. The intervention targeted pregnant women and infants 0-12 months; however, the surveys included children 6-18 months. Children <6 months were excluded because they would not have been exposed to the SQ-LNS component of the programme.
Children up to 18 months were included because it was hypothesized that the programme's nutritional impact would be retained for several months after the programme ended, and there were substantial cost implications of including only children 6-12 months, as sampling would have required visiting a greater proportion of households in both health zones. Sample sizes (650 children per health zone per survey) were determined for the primary outcomes of the evaluation (to provide 80% power to detect a relative decline in the prevalence of anaemia of 15%). A total of 660 households were targeted in each health zone during both the baseline and endline surveys.
In both surveys, trained interviewers asked caregivers about household sociodemographic characteristics and knowledge, attitudes, and behaviours relating to nutrition, IYCF, and SQ-LNS. (Across both surveys >97% of caregivers were biological mothers and are thus referred to as mothers from here on). Mothers were asked if they had heard of anaemia and if so to list the causes. They were also asked to recall breastfeeding practices in early life such as whether the child had ever been breastfed, how many hours or days after birth the mother initiated breastfeeding, when the mother introduced water for the first time, and when she introduced solid and semi-solid foods.
Interviewers also asked mothers to recall all of the foods and drinks the child consumed in the previous day. Interviewers were instructed to count the number of separate meals or snacks provided to the child (to calculate meal frequency) and also to use a list of 17 food groups and mark yes or no for each food group depending on whether the food was consumed in the previous day. Because mothers did not receive SQ-LNS before the baseline survey, dietary recalls excluded SQ-LNS (for comparability). In the endline survey, mothers were specifically asked about SQ-LNS receipt and children's consumption of SQ-LNS. In the data analyses phase, the 17 food groups were collapsed to the standard seven food groups recommended in the WHO/UNICEF IYCF indicators (WHO/UNICEF/IFPRI/UCDavis/ FANTA/AED/USAID, 2008). In accordance with the WHO/UNICEF indicators, minimum dietary diversity was defined as greater than or equal to four food groups (out of seven) in the previous day, minimum meal frequency as greater than or equal to two times per day for breastfed infants aged 6-8 months, greater than or equal to three times for breastfed children aged 9-23 months, and greater than or equal to four times for nonbreastfed children 6-23 months. For breastfed children, minimum acceptable diet was defined as minimum meal frequency and minimum dietary diversity. For nonbreastfed infants, minimum acceptable diet was defined as at least two milk feeds, minimum meal frequency and at least four out of six food groups (excluding the dairy food group) in the previous day. For handwashing indicators, 2. CHWs received bikes to improve their mobility in their catchment areas.
3. CHWs received new CHW guide with images on IYCF supporting the four key messages.
4. CHWs received logbooks to track activities.
5. Standardized expectations for the role of CHWs in the enhanced IYCF programme. Guidebooks, trainings, and logbooks standardized the expected contact points between mothers and CHWs for IYCF counselling. CHWs were expected to (a) conduct monthly home visits to pregnant women and women with children 0-12 months and (b) conduct monthly group meetings with~20 pregnant women and women with children 0-12 months (CHWs were instructed to hold group meetings more frequently if they had a large catchment area).
6. Improved and standardized supervision-CHWs were expected to review their logbooks each month with the CHW supervisor (a health worker at the health facility).

Investments in mass media and IEC
• None Investments in mass media and IEC materials: 1. Creation and distribution of IEC materials addressing IYCF, handwashing, and SQ-LNS including posters (CHW and health worker guides, leaflets, and Kulabora packaging).
2. Weekly radio messages re-enforcing the four key messages.
Note. ANC: antenatal care; CHW: community health worker; IYCF: infant and young child feeding; SQ-LNS: small-quantity lipid-based nutrient supplements; UNICEF: United Nations International Children's Emergency Fund.
interviewers observed whether the household had soap and then asked the mother whether she used soap the previous day and if yes, to list the purposes for which she had used soap.
In the endline survey only, mothers were asked about exposure to specific components of the enhanced IYCF programme such as the following: whether she received information about breastfeeding during her last pregnancy, whether she received information about complementary feeding or handwashing at any point, and follow-up questions on the content and source of that information. Mothers were also asked about SQ-LNS knowledge, attitudes, and practices, as well as general indicators of programme exposure such as whether she participated in a group session on infant feeding at her last visit to the health centre for her child, whether she knows her CHW, and whether she had heard the radio messages on IYCF.

| Data analysis
We compared change in prevalence of key IYCF indicators from baseline to endline in the two health zones using difference in differences (DiD) analyses. Unadjusted and multivariable DiD (95% CI) estimates were obtained from mixed linear regression models with an interaction term between variables for health area (intervention vs. control) and time (endline vs. baseline), and cluster as a random effect (Card & Krueger, 1993). Based on a review of the literature, we determined a priori that multivariable models would adjust for child's sex and age; maternal age, education (completion of secondary school and completion of primary school vs. less than a primary education), and ethnicity (Bemba vs. other ethnicity); and household's primary source of income (agriculture, wage labour or daily work, or other), whether there was another child under 5 years of age in the household, and asset tertile.
Asset tertile was derived from a principal component analysis of all households in the baseline and endline survey's binary yes-no responses to ownership of a radio, television, mobile phone, refrigerator, stove, chair, bed, lamp, oven, hoe, sewing machine, bicycle, car, truck, and electricity (Vyas & Kumaranayake, 2006). Analyses were first conducted stratified by child's age (among children ≤12 months and children >12 months); however, when no notable differences were found in the two age groups, they were collapsed.
Prevalence ratios comparing programme exposure at endline in the intervention and control areas were obtained from linear mixed models with a log link, binomial distribution, and cluster as a random effect (Spiegelman & Hertzmark, 2005). Within the intervention area at endline, we also compared prevalence ratios for IYCF knowledge and practices among mothers with high versus low programme exposure (defined as 2-3 vs. 0-1 of the three exposures assessed in the endline survey: attendance at a health centre group IYCF session; receipt of IYCF, SQ-LNS, or handwashing information from a CHW; and feeding the child SQ-LNS). The initial analysis plan was to compare mothers with any exposure to the enhanced IYCF programme (1-3 of the exposures above) to an "unexposed group"; however, only 53 mothers in the intervention area at endline answered no to all three exposures. We thus collapsed mothers with 0-1 exposures into a single reference group defined as the "low exposure" group, compared with the "high exposure" (2-3 exposures) group. When the log-binomial models did not converge, a log-Poisson link function, which provides a consistent but less efficient empirical estimate of the prevalence ratio (Zou, 2004), was used.

| Ethical approval
The National Statistics Office in Lubumbashi in DRC and the U.S.
Centers for Disease Control (CDC) approved the protocol for the impact evaluation; the CDC determined the evaluation as public health practice. Interviewers explained the survey protocol to mothers and also informed mothers of the option to refuse to participate in the survey or to stop participation at any point. Given low rates of literacy in the area, all mothers provided verbal, informed consent to participate in the baseline and endline surveys, and interviewers indicated on the questionnaire when informed consent had been obtained.

| RESULTS
The baseline and endline surveys included 1,288 and 1,307 mothers, respectively, reflecting response rates of 97.7% and 99.2%, respectively. Intervention and control areas were similar on several key demographic characteristics such as child's age and sex, maternal age, the number of households with multiple children under 5 years, and the proportion of household who receive primary health care at health posts versus dispensaries (Table 2). However, there were several differences between the two areas; notably, over 80% of households in Kasenga (intervention area) but less than 40% of households in Kipushi (control area) cited agriculture as their primary source of income. Kipushi also had a more urban population with greater ethnic diversity, more education, and more household assets.
Compared with the control area, the intervention area saw substantial increases in the proportion of mothers who reported optimal breastfeeding practices in the first 6 months and handwashing behaviours during food and defecation-related moments in the previous day; these differences remained significant in multivariable analyses adjusting for sociodemographic confounders (Table 3) Notably, the significant DiD for minimum meal frequency was largely driven by a decrease in feeding frequency in the control area as opposed to an increase in the intervention area. We also did not find Analyses from the endline survey confirmed that mothers in the intervention area were substantially more likely to report exposure to programme components compared with mothers in the control area (   In accordance with the UNICEF IYCF indicators, minimum dietary diversity is defined as four or more food groups (out of seven) in the previous 24 hr. Minimum meal frequency defined as greater than or equal to two times per day for breastfed infants aged 6-8 months, greater than or equal to three times for breastfed children aged 9-23 months, and greater than or equal to four times for nonbreastfed children 6-23 mos. Minimum acceptable diet is defined as minimum meal frequency and minimum dietary diversity in the previous 24 hr for breastfed children; for nonbreastfed children, minimum acceptable diet is defined as greater than or equal to two milk feeds, the minimum meal frequency and greater than or equal to four food groups (from a total of six food groups that excludes dairy) in the previous day.  Only among children who received 28 LNS sachets in their last distribution and whose mothers reported a quantity consumed (n = 418 in the intervention area).
receiving information from their CHW on feeding their child after 6 months and on handwashing, respectively. The comparable rates of information receipt from the CHW were respectively 0.5%, 0.3%, and 0.3% in the control area. Mothers in the intervention area were also more likely to report receiving specific messages on key IYCF behaviours including "initiate breastfeeding within one hour after birth"  ). Notably, less than half of mothers in the intervention area reported receiving the key message on dietary diversity to "add eggs, insects, and fish to child's food," and in multivariable models, there was no difference between the proportion of mothers who received this message in the intervention and control areas.
Although 95.9% and 70.2% of mothers in the intervention area had heard of Kulabora (LNS) and tried feeding it to their child, respectively, only 1.4% of mothers in the control area had heard of Kulabora, and only a single mother had fed it to her child. In the intervention area, 91.3% of mothers who received SQ-LNS reported receiving 28 sachets in their last distribution, three quarters of whom reported feeding their child all 28 sachets.
In analyses from the endline survey in the intervention area, we also found that high programme exposure (2-3 vs. 0-1 of the following: attendance at a health centre group IYCF session, receipt of IYCF or handwashing information from a CHW, and feeding the child SQ-LNS) was associated with select IYCF and handwashing knowledge and practices (Table 5). Specifically, we found that mothers with high programme exposure were significantly more likely to report waiting until 6 months to introduce water (APR [95% CI]:

| DISCUSSION
In this analysis of cross-sectional, preintervention, and postintervention surveys conducted in samples that were representative of all households with children aged 6-18 months in two health zones in Katanga Province, DRC, we found that the health zone that received the enhanced IYCF intervention experienced significantly greater increases in the proportion of mothers reporting several key breastfeeding and handwashing practices between baseline and endline compared with mothers in the control area. We did not, however, find significant DiD in the proportion of mothers who fed their children the minimum dietary diversity or minimum acceptable diet.
Notably, this analysis revealed that immediate and exclusive breastfeeding indicators in the first 6 months of life increased significantly more in the intervention area than the control area, and that in the intervention area, mothers with high programme exposure (including exposure to CHWs) were more likely to wait until 6 months to introduce water or complementary foods compared with mothers with low programme exposure. A separate study from South Kivu, DRC, also found that a nutrition programme using CHWs resulted in substantially higher rates of exclusive breastfeeding among children under 6 months in the intervention area compared with a control area (Balaluka et al., 2012). Taken  Enhancing access to diverse, micronutrient foods is also essential for optimal IYCF practices. Although key messages of the enhanced IYCF intervention emphasized the importance of adding "special foods such as eggs, insects, and fish to the child's food," less than half of mothers in the intervention area reported receiving this key message from their health worker. Furthermore, we found that in the baseline and endline surveys from both health zones, mothers reported feeding their children an average of only two food groups per day (below the WHO recommended four or more food groups per day). Few studies have analysed drivers of IYCF practices in the DRC; however, one qualitative study from South Kivu revealed that mothers cited the cost of meat, milk, eggs, fruits, and vegetables, as well as their lack of empowerment to make purchasing decisions, as the primary barriers to mothers providing their children a diverse diet (Burns et al., 2016). In this setting of high poverty, low maternal education and limited access to micronutrient-rich foods (DRC DHS, 2013;Ekesa, Blomme, & Garming, 2011), health workers may not have emphasized this message if they did not think that it was feasible for mothers to improve dietary diversity, and among mothers who did receive this message, lack of access to diverse foods was likely an important barrier to behaviour change. These findings thus highlight the particular relevance of fortified foods such as SQ-LNS to fill the micronutrient and macronutrient gap between biological needs and actual consumption of infants and young children in areas with high rates of poverty (Arimond et al., 2015).
Our study is one of few to assess the impact of a community-and facility-based integrated IYCF-SQ-LNS programme on IYCF practices (Arimond et al., 2017;Owino, Bahwere, Bisimwa, Mwangi, & Collins, 2011 Low programme exposure defined as 0 or 1 of the following, and high programme exposure is defined as at least two of the following: (a) mother attended a group session on IYCF at her last health facility visit for her child; (b) mother receiving information from her community health worker on breastfeeding, complementary feeding and/or handwashing; or (c) mother has fed her child LNS. b Univariate and multivariate prevalence ratios (PR), 95% confidence intervals, and P values obtained from linear regression models with the log link and binomial distribution. When the log-binomial model did not converge, a Poisson distribution was used.
c Multivariate models adjust for child's sex and age; maternal age, education, and ethnicity; urban versus rural; households primary source of income and household asset tertile; and whether there was another child under 5 years in the household. d In accordance with the UNICEF IYCF indicators, minimum dietary diversity is defined as four or more food groups (out of seven) in the previous 24 hr. Minimum meal frequency defined as greater than or equal to two times per day for breastfed infants aged 6-8 months, greater than or equal to three times for breastfed children aged 9-23 months, and greater than or equal to four times for nonbreastfed children 6-23 months.
Minimum acceptable diet is defined as minimum meal frequency and minimum dietary diversity in the previous 24 hr for breastfed children; for nonbreastfed children, minimum acceptable diet is defined as greater than or equal to two milk feeds, the minimum meal frequency and greater than or equal to four food groups (from a total of six food groups that excludes dairy) in the previous day.
second study combined data from four coordinated randomized trials in children without malnutrition in Africa and Asia and showed there were no differences in key infant feeding practices such as continuing breastfeeding and diet diversity, and that in two sites, children randomized to receive SQ-LNS were more likely to be fed the minimum meal frequency than children who were not randomized to LNS (Arimond et al., 2017 however, we cannot rule out residual confounding. In addition, our study design includes serial cross-sectional data from representative samples of the two areas at baseline and endline; however, we did not collect longitudinal data on individual mother-infant pairs, nor did we collect programme exposure data (such as frequency of contact with health workers and CHWs) during the baseline survey. Thus, we cannot conduct longitudinal analyses to show changes in behaviour among mothers due to programme exposure. In order to support the findings that the enhanced IYCF programme played a causal role in the observed improvements in IYCF practices, we showed that mothers in the intervention area with high levels of programme exposure were more likely to engage in optimal IYCF practices compared with mothers with low programme exposure, even after adjustment for key demographic characteristics. However, we cannot rule out the possibility that there were unmeasured confounding factors that drove mothers to pursue both programme exposure and optimal IYCF practices.
In addition, our sample included children aged 6-18 months for cost reasons, whereas the programme focused on children under 12 months and pregnant women. Our findings on indicators relating to immediate and exclusive breastfeeding are based on maternal recall of at least 6 months; this recall may ultimately reflect reporting bias as opposed to actual changes in behaviour. Notably, reporting bias in this study would likely still be a reflection of changes in knowledge if not actual changes in behaviour (i.e., mothers in the intervention area would need to know that they "should" report immediate and exclusive breastfeeding in order to document the substantial increases that were recorded). Despite documented changes in maternal report of some key IYCF practices, we did not document changes in dietary diversity; and our lack of data on food prices and availability prevents us from fully understanding this lack of change. Finally, our evaluation did not collect qualitative information from mothers, CHWs, or health facility staff, and thus, we are unable to fully elucidate barriers and motivations of mothers or frontline workers.
Our study has several strengths. We collected baseline and endline data in both the programme and control areas, thus allowing us to separate the extent to which changes in IYCF practices are due to the intervention as opposed to secular trends. Our baseline and endline surveys are representative of the study areas, they collect comprehensive data on key IYCF and sociodemographic characteristics, and they included large sample sizes, all of which allowed for rigourous, multivariate analyses adjusting for key confounders. In this study, we found that there were (a) significant DiD comparing some IYCF practices at baseline and endline in the two health zones; (b) substantial differences in the proportion of mothers exposed to IYCF programme components (counselling and key messages) in the two health zones at endline; and (c) correlations between programme exposure and optimal IYCF practices within the intervention area at endline. Taken together, these findings demonstrate a consistent association between the enhanced IYCF programme and improved IYCF and handwashing practices in Kasenga, DRC.Overall, we found that the enhanced IYCF intervention was associated with an increase in the prevalence of several important handwashing and breastfeeding practices, and that mothers in the intervention area were substantially more likely to receive IYCF mes-