The MNP sachet distribution and behaviour change strategies were added to an existing IYCF programme that included education and counselling about recommended breastfeeding and complementary feeding practices. The pilot was designed to distribute 60 MNP sachets to all children 6–23 months of age every 6 months (a total of 180 sachets per child aged 6–23 months). Caregivers (usually mothers, and referred to as mothers in this paper) were instructed to provide one sachet daily for 60 days and, if they stopped for any reason, to start again until the sachets were finished. In 6 months, they should return to get the next batch. The MNP was locally branded as Baal Vita and included 15 vitamins and minerals.

Two delivery models for MNP distribution were piloted in rural areas of selected districts. Both models involved FCHV (who were generally unpaid local volunteers from the community who provided various health-related services, including IYCF support). When designing the delivery models for the pilot, there was a concern that FCHVs were already overworked and that it would be too burdensome for them to also distribute MNP. As a result, FCHV delivered the behaviour change support in both models, but routinely distributed MNP sachets in only one model. MNP were distributed routinely either through FCHV in the community (FCHV model) or through the health facilities (HF model), where families had to travel to the nearest health facility to get MNP. Health facility staff also supported behaviour change in both models but, because of MNP distribution in facilities, families in the HF model were expected to have more interaction with health facility staff regarding MNP compared with families in the FCHV model. All other aspects of the two intervention delivery models, strategies and content were designed to be identical.

In both models, health facility staff and FCHV were trained prior to the start of the pilot on the intervention and their primary roles in delivering the behaviour change strategies. The aims of the behaviour change component focused on creating an enabling environment and increasing caregivers (primarily mothers) knowledge and skills which would influence their experiences, acceptability and motivation to improve key IYCF practices, collect the MNP every 6 months, practise appropriate MNP preparation and use, and have the children consume all of the MNP from each batch obtained (Stokols 1996; Fisher et al. 2008). The behaviour change strategy included distributing an MNP brochure and a reminder card for each MNP distribution in both models. Both the brochure and the reminder card included key messages describing MNP and the correct MNP preparation and use, which were meant to support maternal knowledge, skills and motivation for MNP coverage, and sustained intake adherence. The reminder card also included dates showing MNP receipt and the date for families to get the next batch of 60 MNP sachets, as well as places to mark after feeding the child food mixed with MNP each day; this tool was meant to support mothers as memory aids for when to get the next batch of MNP, as well as to help establish an intake routine. The intervention was also advertised on the radio and billboards/banners and through other branded promotional strategies (e. g. on clocks and stickers given to FCHV). The radio advertisement included a mother talking about how and why to use MNP during a usual day in her life, and the radio advertisement and billboards/banners included messages about MNP, appropriate preparation and use, and where to get more information. Each FCHV received an IYCF/MNP flip chart to use as a teaching aid, which included information on anaemia, IYCF, hygiene and sanitation, early childhood development and described MNP and appropriate MNP preparation and use. FCHVs are expected to hold mothers’ group meetings every month on various topics, and in the behaviour change strategy, one meeting was used to launch the IYCF/MNP intervention in each community.

Eighty-three per cent of the population of Nepal lives in rural areas. The country has a diverse geography including three main ecozones: plains, hills and mountains (Himalayas). The elevations range from 90 to over 8800 m, with 50% of the population living in the plains, 43% in the hills and 7% in the mountains. Households are densely spaced in the plains and parts of the hills, whereas in other parts of the hills and the mountains above 2000 m, the distances between houses are much greater (Ministry of Health and Population Nepal, New ERA & ICF International Inc. 2012). Recognising the importance of variations in the geography of Nepal, both distribution models were piloted in each ecozone. In the plains ecozone, Rupandehi (FCHV model) and Parsa (HF model) districts participated and, in the hills, Palpa (FCHV model) and Makwanpur (HF model) districts participated; surveys were not collected in the two pilot districts in the mountains ecozone.

Cross-sectional household surveys representative of children 6–23 months of age were collected in each of four pilot districts during October and November 2011 as part of programme monitoring of this public health intervention; UNICEF Nepal funded an external agency to conduct the surveys. Because there was a staggered start to the programme, these data were collected 3 months after programme roll-out in Rupandehi and Parsa districts (plains) and 15 months after programme roll-out in Makwanpur and Palpa districts (hills).

Proportion-to-size population sampling was used to select 30 clusters from rural areas of each district. In each cluster, a census was carried out and 12 children 6–23 months of age were randomly selected; their mothers were invited to participate in an interview (360 interviews per district/survey). There was no replacement of clusters, for refusal to participate in interviews, or for fewer than 12 eligible children in a cluster. In addition to the household surveys, in each selected cluster, every FCHV (usually one per cluster but sometimes more) was invited to participate in an FCHV interview. The response rate among mothers in the 3-month surveys collected in the plains was 100% in the FCHV model and 93% in the HF model, and for the 15-month survey in the hills, the response rates were 96% and 86%, respectively. The response rate for the FCHV invited to participate in the FCHV surveys was 100% in all surveys.

Participation in the interview was voluntary and women gave verbal informed consent that was witnessed and recorded. The Nepal Ministry of Health and Population approved the protocol and de-identified data were used for this secondary analysis.

In the 3-month surveys, the outcome of interest was initial MNP coverage, defined as mother’s report of ever obtaining a batch of 60 MNP sachets for her child. In the 15-month surveys, the outcomes were initial MNP coverage and MNP repeat coverage, defined as mother’s report of receiving two or more batches of 60 MNP sachets for the eligible child (eligible children were ≥12 months of age who had received at least one previous batch of MNP sachets).

Each mother reported her child’s age and sex, her education level and household assets (used to develop a wealth quintile index). As a proxy indicator of knowledge and exposure to the programme, which might also influence motivation to give her child MNP, each mother was asked whether she knew of any consequences of anaemia (open-ended responses dichotomised into no vs. knows of at least one consequence). Mothers were expected to have participated in meetings or received programme materials and messages, which would influence their knowledge, skills and motivation to get and use MNP. They were asked if they had ever heard of MNP and, if yes, were questioned about whether they had experienced specific programme components (yes vs. no) that were used as proxies of exposure to the intervention package. Proxies included whether they had ever received information about MNP at a FCHV led mothers’ group meeting, had ever heard the MNP radio advertisement, or had received an intervention brochure or reminder card. Mothers who responded that they had never heard of MNP were categorised as ‘No’ for attending a FCHV led mothers’ group meeting about MNP or hearing the MNP radio advertisement. Among mothers who had given MNP to their children, additional questions were asked about direct experiences or perceptions of using MNP that might further influence motivation and demand for MNP. These included open-ended questions about any negative or positive effects of MNP in their children after use, which we categorised as perceived no positive effects vs. one or more positive effects, and perceived no negative effects vs. one or more negative effects. Mothers were asked if their children liked consuming food mixed with MNP (yes vs. no), how accessible were MNP (easily accessible vs. not), and reported round-trip time to pick up the last batch of 60 MNP sachets, which was dichotomised as <1 h round-trip vs. ≥ 1 h round-trip.

Data were analysed using SAS 9. 3 (SAS Institute Inc. , Cary, NC, USA) accounting for complex survey design. Chi-square tests were used to examine differences in coverage and other survey population characteristics by distribution model. A logistic regression model was developed for the surveys collected 3 months after the intervention start for the outcome of receiving a batch of 60 MNP sachets. There were two categories of predictor variables included in the model: child socio-demographic factors (age, sex) and proxies of intervention exposure that should have occurred prior to getting the first batch of 60 MNP sachets (attended a mothers’ group meeting where FCHV discussed MNP, hearing the MNP radio advertisement and knowing consequences of anaemia). For the 15-month surveys, logistic regression was used to examine the odds of repeat coverage (limited to mothers of children ≥12 months of age who reported obtaining at least one batch of MNP sachets). The same socio-demographic factors and proxies of intervention exposure variables were included as above, as well as additional variables related to MNP coverage and intake, including receiving an intervention brochure or a reminder card, perception of negative or positive effects in the child after MNP use, report of whether the child liked food mixed with MNP, and round-trip time to get the batches of MNP. Models were adjusted for maternal education and household wealth quintile. Chi-square tests were also used to examine differences in FCHV characteristics and work experiences in the FCHV surveys.

At the 3-month surveys, 49. 2% of mothers in the FCHV model were in the highest wealth quintile, and 36. 1% had a secondary education or higher, compared with 23. 5% and 12. 2%, respectively, in the HF model (Table 1).

Coverage of MNP at 3 months after programme implementation did not differ by delivery model (65. 0% in the FCHV model vs. 57. 4% in the HF model); nearly everyone who got a batch of MNP reported their child consumed at least one sachet (Table 1). Ever having heard about the MNP did not vary by delivery model, although having attended a mothers’ group meeting where MNP was discussed was higher in the HF model. Among those who had ever heard of MNP, coverage was higher in the FCHV model (86. 7% vs. 68. 9%, P < 0. 001). Similarly, more mothers who had heard of MNP reported that MNP were easily accessible in the FCHV model. Few mothers reported MNP stock being unavailable as a barrier to coverage (n = 4, FCHV model; n = 15 HF model; data not shown).

In both delivery models, older children (12–17 months and 18–23 months of age) had greater odds of obtaining a batch of 60 MNP sachets compared with children 6–11 months, at 3 months after programme implementation (Table 2). In the HF model, mothers of girls had approximately half the odds of getting MNP for their child compared with mothers of boys. In both delivery models, mothers who attended a mothers’ group meeting where MNP was discussed had higher odds of obtaining MNP compared with those who did not attend (Table 2). In the FCHV model, hearing the MNP radio advertisement was associated with higher odds for MNP coverage compared with those not having heard it, and mothers who knew of one or more consequences of anaemia had greater odds of MNP coverage compared with mothers who did not know of any consequences.

Among mothers in the surveys collected 15 months after implementation start, 52. 1% of those in the FCHV model and 22. 5% of those in the HF model had a secondary level of education or higher (Table 3). More than half of the mothers in the 15-month FCHV model and two-thirds of those in the HF model surveys were in the two lowest wealth quintiles.

Coverage of any MNP was higher in the FCHV model compared with the HF model (82. 5% vs. 51. 6%) in the 15-month surveys, as was ever hearing about the MNP (91. 3% vs. 74. 1%) (Table 3). Among those who had heard of MNP, coverage was higher in the FCHV model (91. 1% vs. 69. 6%), and more mothers reported that MNP were easily accessible (95. 4% vs. 76. 3%). Few reported MNP stock being unavailable as a barrier to coverage (n = 3, FCHV model; n = 3, HF model; data not shown).

In a subsample of participants with children ≥12 months of age who reported ever getting at least one batch of MNP, repeat coverage (obtaining ≥2 batches of MNP) was higher in the FCHV model (51. 8% vs. 35. 3%). Among this subsample, > 92% of mothers reported MNP were easily accessible from both models.

Among those who obtained ≥2 MNP batches, 90. 0% of mothers in the FCHV model reported they had been reminded to get the next batch of MNP. Among those reminded, 93. 3% were reminded by FCHV, 6. 7% by a family member, and 3. 3% by health facility staff (multiple answers possible, data not shown). In the HF model, 68. 8% had been reminded to get the next batch of MNP. Among those reminded, 72. 7% were reminded by health facility staff, 45. 8% by FCHV, and 6. 1% by a family member (multiple answers possible, data not shown).

Among children ≥12 months of age whose mothers had previously obtained at least one batch of MNP, being 18–24 months of age was associated with increased odds of repeat MNP coverage compared with children 12–17 months of age in both models (Table 4). Likewise, in both models, mothers who perceived one or more positive effects of MNP after giving it to their children had increased odds of repeat coverage compared with mothers who reported there were no positive effects after MNP use. In the HF model, mothers who attended a mothers’ group meeting had more than six times the odds of repeat coverage compared with mothers who did not attend.

In the FCHV surveys collected at 3 months, 72. 4% of volunteers in the HF model had no education compared with 47. 8% in the FCHV model (P = 0. 05) (Table 5). Volunteers in the HF model worked more days per week than those in the FCHV model; however, those in the FCHV model were more likely to work ≥5 hours per day. More volunteers in the HF model reported having discussed MNP with one or more mothers in the previous week than those in the FCHV model. While not statistically different (P = 0. 13), 23. 9% of volunteers in the FCHV model report needing more support to carry out work or not liking the added work of the IYCF/MNP programme compared with 10. 3% of volunteers in the HF model.

In the FCHV surveys collected after 15 months, 35. 3% of volunteers in the HF model had no education compared with 13. 8% in the FCHV model (P = 0. 05) (Table 5). There were no differences between the two distribution models in the average number of days worked as a volunteer the previous week, or in the proportion who worked an average of ≥5 h per day on FCHV duties. Although not statistically different (P = 0. 20), 56. 5% of volunteers in the FCHV model reported needing more support to carry out work or not liking the added work of the IYCF/MNP program compared with 43. 5% in the HF model.