The first two years of life are critical to reducing problems related to malnutrition . The current study scaled-up this strategy by implementing a community-based nutrition education intervention entirely delivered through HEWs. The results of the study showed that mothers' KAP regarding pulse-incorporated complementary food was low at baseline in both the intervention and control groups. After four months of education (i.e., at midline), mothers in the intervention group improved KAP, which continued through the 9-month period. Almost all mothers in the intervention group had good knowledge about the benefits of pulses, household food processing techniques, and methods of preparation; they also started preparing complementary food using pulse crops. This current study findings are similar to other large intervention studies in other counties such as China [29, 30] and Kenya , which showed improved mothers' knowledge about feeding practices when health service providers were used to provide nutrition education.
Improvements in the intervention group were seen in many measures. Frequency of pulse consumption significantly increased in the intervention group as most mothers started using pulses as a complementary food more than once per day. Frequency of pulse consumption increased in the intervention group after midline. Although, we lacked data on how much pulse crop the mothers' harvested, the provision of seed and fertilizer may have motivated those in the intervention group to prepare pulses for consumption more frequently. Furthermore, there was a significant difference in mean dietary diversity between the intervention and control groups. At the end of the intervention, consumption of pulses and nuts group improved significantly in the intervention group. Although the mean diet diversity in both groups improved at the midline and endline, the intervention group had a slightly higher mean diet diversity score than the control group. However, at endline, both groups were preparing foods from fewer food groups per day than recommended (i.e., less than four) . Meal frequency in young children was significantly higher in the intervention group at both midline and endline. This study results were similar to those of a Chinese study, which was designed to improve the feeding practices of young children and found improved diet diversity as well as meal frequency in the intervention group .
Stunting increased with age in both groups, but by endline, less children were stunted in the intervention group than in the control group, which could reflect the positive impact of the intervention. The study findings were consistent with those of similar studies in Peru and China. A Peruvian study found a significant difference in stunting between the intervention and control groups and a reduced rate of stunting in the intervention group at the end of the study . A Chinese study found a marginal significant difference in stunting between treatment and comparison groups . Other similar studies have also observed a decreased rate of stunting and improvement in linear growth at the end of intervention, although these studies found no statistically significant difference between the treatment and comparison groups [33–36]. However, other studies have found neither improvement nor significant differences in stunting between the treatment and comparison group [15, 31]. The range of the nutrition education period in most studies ranged from six to 18 months, a short time to find a significant change in stunting. In the current study, the prevalence of stunting increased with age, consistent with previous studies [33–36]. This increase in stunting with age could occur because of factors such as infectious disease, poor maternal malnutrition during pregnancy, poor hygiene, and poverty , all of which limit the impact of an intervention over a relatively short period. In the same way, in the current study, underweight also increased with age in both intervention and control groups. However, the prevalence of wasting decreased over time in the intervention group, whereas the number of children in the control group who were wasted increased at the endline.
In our study, we found household processing practices such as germination to be difficult tasks for mothers to adopt. The number of mothers who were using soaked and germinated pulse crops for complementary food was low. Only 43% and 69.7% of mothers in the intervention group reported soaking and germination pulses at midline and endline, respectively. However, the results showed that more mothers performed household processing techniques in the intervention group than in the control group. The number of mothers who adopted practices for soaking and germination of pulses in this study was higher than that in a similar study in Malawi, where only 25% of participants adopted the practices from the message . This difference could be due to the length of intervention, which was shorter than this current intervention. In general, the effect of nutrition education on mothers' knowledge and practice was high at both the midline and endline of this study, but, attitude did not change after the midline, possibly indicating that an intervention of approximately five months could be sufficient to change mothers' attitude.
The strength of this study was the involvement of trained local HEWs who can speak the local language in delivering the nutrition messages. The intervention contributed to women's empowerment by using HEWs and training mothers in their own communities. A peer mentoring approach can have positive impacts on the knowledge, confidence, and attitudes of participants . We have described the effects on HEWs elsewhere . In addition, multiple educational theories, namely Theory of Planned Behavior (TPB) and Health Belief Model (HBM) were used in designing the messages. TPB helped to identify factors that may influence pulse consumption in feeding practices through a combination of attitudes towards consumption of pulses, household processing techniques, cultural influences, and behavioral control that results in the formation of an intention . HBM helped in addressing the problem behaviour, in this case the risk of inappropriate feeding practice and low consumption of pulses leading to undernutrition . The limitations of the study included the potential for information contamination between intervention and control sites. The distribution of seed after midline was an unexpected addition to the study; however, harvesting of these planted seeds had not occurred before endline, and thus this additional variable did not impact the results at nine months.