Micronutrient powder use in Arequipa, Peru: Barriers and enablers across multiple levels.

Abstract In Peru, nearly half of children aged 6–36 months were diagnosed with anaemia in 2017. To address this disease, the Peruvian Ministry of Health implemented a national programme in 2014, distributing free micronutrient powders (MNPs) to all children of this age. However, rates of childhood anaemia remain high. The aim of this study was to explore factors at all levels of the Social‐Ecological Model that affect MNP use and adherence in Arequipa, an Andean city with childhood anaemia rates higher than the national average. We conducted in‐depth interviews with 20 health personnel and 24 caregivers and 12 focus group discussions with 105 caregivers. We identified numerous barriers, including negative side effects (constipation, vomiting, and diarrhoea), poor taste of MNP, lack of familial and peer support for its use, insufficient informational resources provided by the health system, and limited human resources that constricted health personnel abilities to implement MNP programming successfully. Facilitators identified included concern about the long‐term effects of anaemia, support from organizations external to the health system, well‐coordinated care within the health system, and provision of resources by the Ministry of Health. We found that community or organizational and societal factors were key to limited MNP use and adherence, specifically the limited time health personnel have to address caregivers' doubts during appointments and the lack of informational resources outside of these appointments. Potential policy implications could be to increase informational resources available outside of individualized counselling by strengthening existing collaborations with community organizations, increasing media coverage, and providing group counselling.

According to most recent studies, 42.6% of the world's children aged 6-59 months have anaemia, defined as low-haemoglobin concentration (WHO, 2015). Most estimates suggest that half of all childhood anaemia cases are due to iron deficiency, with other causes including haemoglobinopathies and infectious disease (Kassebaum, 2016;Miller, 2013;Rosado et al., 2010;WHO, 2016). Globally, anaemia is more prevalent in lower and middle-income countries due to higher rates of nutritional deficiencies and infectious diseases and limited access to health care resources (Shaw & Friedman, 2011;Tolentino & Friedman, 2007). In children, anaemia can impair cognitive and motor development, causing fatigue and poor school performance (Kassebaum, 2016;Powers & Buchanan, 2014). To address childhood anaemia, countries around the world have employed various multimicronutrient supplements (De-Regil, Suchdev, Vist, Walleser, & Peña-Rosas, 2011;Rosado et al., 2010;Shankar et al., 2009;Sivakumar et al., 2006;Vinod Kumar & Rajagopalan, 2006). Micronutrient powders (MNPs), with brand names including Sprinkles or "Chispitas", are one form of supplementation. MNPs are designed to be odourless, tasteless powders, typically formulated with iron, zinc, folic acid, Vitamin A, and Vitamin C (De-Regil et al., 2011;Zlotkin et al., 2005). These powders come in single-dose sachets and are consumed by mixing with semi-solid food (De-Regil et al., 2011;Zlotkin et al., 2005). MNPs are proven to be an efficacious intervention for preventing and treating anaemia in young children (De-Regil et al., 2011), and World Health Organization recommends their use in populations where the prevalence of anaemia in young children is 20% or higher (WHO, 2016).
Peru had seen high rates of early childhood anaemia, defined as a haemoglobin value of less than 11.0 g/dl, for the past 18 years (Instituto Nacional de Estadística e Informática, 2018). Between 2000 and 2011, prevalence of anaemia in children aged 6-36 months decreased significantly from 60.9 to 41.6%, corresponding with a similar decrease in childhood stunting and the implementation of government strategies for poverty and child malnutrition reduction in this time period (CARE Perú, 2011;Instituto Nacional de Estadística e Informática, 2018;Marini, Rokx, & Gallagher, 2017). However, the national prevalence of childhood anaemia has changed little since 2011, with 43.6% of children diagnosed with anaemia in 2017 (Instituto Nacional de Estadística e Informática, 2018). Across the country, prevalence is higher in rural than urban zones (53.3 and 40.0%, respectively). Moreover, anaemia is higher in the jungle and highland regions (53.6 and 52.0%) compared with coastal areas them daily for 12 months, a departure from pilot tests, which recommended consumption every other day for 6 months (Creed-Kanashiro et al., 2016;Ministerio de Salud, 2014). MNPs are distributed during well-child check-ups and promoted through free home visits and informational sessions in public health establishments (Ministerio de Salud, 2014). However, an early study of the nationwide programme showed low adherence to MNP, with only 24.4% of children consuming more Research findings on MNP use in Peru suggest that misconceptions around preparation of MNP, nausea or dislike of taste upon consumption, and negative comments from family and peers were key barriers to adherence (Creed-Kanashiro et al., 2016;Huamán-Espino et al., 2012;Munares-García & Gómez-Guizado, 2016). This research identifies the key facilitators as familial support, culturally appropriate counselling techniques, and recognition of improved child health (Creed-Kanashiro et al., 2016;Huamán-Espino et al., 2012;Munares-García & Gómez-Guizado, 2016). Analysis of MNP interventions in other countries shows that these barriers and facilitators are not unique to Peru and have had similar influence on programme effectiveness worldwide (Akoto Osei et al., 2014;Sarma, Uddin, Harbour, & Ahmed, 2016;Jefferds et al., 2010;Kodish, Rah, Kraemer, de Pee, & Gittelsohn, 2011;Sutrisna, Vossenaar, Izwardy, & Tumilowicz, 2017;Tripp et al., 2011). Numerous studies on the use and acceptability of MNP have contributed to a greater understanding of factors that can affect the success of these interventions. However, within the Peruvian context, there has been limited qualitative exploration into Given the continually high rates of childhood anaemia, the aim of our research is to explore factors at various levels that affect MNP use and adherence in Arequipa, Peru. To do this, we triangulated data gathered through focus group discussions (FGDs) and in-depth interviews (IDIs)

Key Messages
• Caregivers experienced frustrations with MNP use, especially difficult administration due to taste, which were compounded by doubtful comments from peers on the effectiveness of MNP and insufficient time in well-child check-ups to talk through concerns.
• Awareness of the long-term effects of anaemia prompted caregivers to use MNP, and informational sessions at public day cares were an effective method of disseminating information about MNP use.
• Policy implications from this study could include strengthening community engagement strategies and mass media campaigns in order to address doubts about MNP use and decrease the strain on limited informational resources within the health system. with caregivers of children aged 6-36 months, and IDIs with health personnel. We applied a modified version of the Social-Ecological Model (SEM) to structure the IDIs and FGDs, as well as the analysis and presentation of results (Centers for Disease Control and Prevention, 2013).
The SEM derives from the concept that individual's behaviours are influenced by factors at various levels of their environment, in this version: the individual, interpersonal, community or organizational, and policy levels (Centers for Disease Control and Prevention, 2013). Notably, our modified version utilizes the community or organizational level for factors within health establishments and other organizations that could vary between sites and the policy level for government-led resource allocation and intervention design. However, it should be noted that many themes are cross-cutting and do not easily fit into only one level.
Other versions of the SEM have been successfully employed to evaluate factors affecting child nutrition outcomes and interventions (Kelly et al., 2017;Stang & Bonilla, 2017).

| Study setting
This research project took place in the province of Arequipa, the second largest city in Peru, with~970,000 inhabitants (Instituto Nacional de Estadística e Informática, 2009). The city is surrounded by the Andean mountains and is located 2,328 m above sea level ("Are-quipa|Peru,", n.d.). Local health authorities, the "Red de Salud Arequipa Caylloma" ("Arequipa Red" from now on), requested that we conduct this study in Arequipa, given the continually high anaemia prevalence rates among children aged 6-36 months in the area-44.5% in 2016 (Ministerio de Salud, 2017). We conducted IDIs and FGDs across eight urban and peri-urban districts out of the 29 districts in Arequipa.
According to "Arequipa Red" unpublished sources, these districts comprised more than half of the cases of early childhood anaemia in this province.

| Study design
Qualitative methods, comprised of IDIs with caregivers and health personnel and FGDs with caregivers, were employed. These participant types are the key stakeholders in the MNP intervention, involving both allowed for data triangulation between the two groups (Joint United Nations Programme on HIV/AIDS, n.d.

| Caregivers
Caregivers were recruited for both IDIs and FGDs. The inclusion criteria for caregivers were any primary caregivers-whether biological parent or not-over the age of 18 who self-reported spending at least 5 days a week providing care for a child aged 6-36 months who lived within the study site. The primary caregiver is usually responsible for the feeding practices of the child and therefore the administration, or lack thereof, of MNP.

| Recruitment
IDI participants were selected through convenience sampling from within or around health establishments after a screener questionnaire established that they met the inclusion criteria. Health establishments were selected by choosing one health centre and one health post per district, favouring those establishments with the highest percentage of children with anaemia out of children evaluated according to most recent data from "Arequipa Red" unpublished sources. The selection of one health centre and one health post per district was made to account for variation among establishments as health centres are typically larger, have more resources, and are located in more urbanized areas than health posts.
For FGDs, a team of field researchers started from outside the selected health establishments and entered the surrounding neighbourhoods from different directions, recruiting caregivers doorto-door until reaching a radius of five blocks. Each of five field researchers was responsible for recruiting five caregivers and would stop recruiting when they reached this number or the five-block limit, whichever came first; the team oversampled knowing some individuals might not make it to the FGDs. Caregiver eligibility was determined using a screener questionnaire and confirming that they had not previously participated in an IDI. The FGDs were held the day following recruitment in private spaces in the selected health establishments.

| Data management and analysis
IDIs and FGDs were audio-recorded (verbal consent permitting); also, a member of the field team took detailed notes and reviewed these with the interviewer afterwards. In the few cases that the participant did not give consent for audio-recording but expressed interest in participating in the research, notes were used for analysis in lieu of a transcription. This occurred in five IDIs in four districts-two with caregivers and three with health personnel. For analysis, the research team developed a codebook based on the IDI and FGD guides, in a process similar to structural coding (Saldana, 2015), distinguishing codes as either themes related to MNP use or anaemia and other methods of anaemia prevention, each cross-coded with a corresponding level of the SEM. Blind double-coding was used on every fifth transcript to ensure consistency and reduce bias. The research team reviewed these double-coded transcripts and made changes to the coding where discrepancies were identified before coding further transcripts. The data were coded and analysed for the most frequently co-occurring codes using Dedoose, a web-based application that facilitates research data management and analysis ("Home|Dedoose,", n.d.).
Findings are presented thematically using the SEM framework and are stratified by caregiver and health personnel perspectives (Tables 2 and 3).

| Ethics
This research project was reviewed and approved by the Institutional Review Boards (IRBs) of Tulane University School of Public Health and Tropical Medicine in New Orleans, Louisiana, USA (#1069759), the Universidad Peruana Cayetano Heredia in Lima, Peru (#101351), and the nonprofit organization Asociaciòn Benéfica PRISMA in Lima, Peru (#1446.17). All study participants gave their verbal consent before participating in the in-depth interviews or focus group discussions, as approved by all IRBs, and were given a written copy of the consent script.

| RESULTS
Data were collected in June-July 2017. In each of the eight study districts, IDIs were conducted with three caregivers and one to three health personnel, and one to two FGDs were conducted.

| Health personnel
We conducted 20 IDIs with health personnel: 14 nurses (who give out the MNP), four doctors, and two nutritionists (Table 4).

| Caregivers
We conducted 24 IDIs with caregivers: 23 parents and one grandparent (Table 4). We collected the exact child age from 21 of these 24 caregivers (some could only give an approximation). The average child age was 19 months, with a reasonably even distribution across the age range of the inclusion criteria (6-36 months). FGD size ranged from four to 13 caregivers, with a total of 105 participants across the 12 FGDs; again, most were parents (93) or grandparents (eight; Table 4).

of 12
The exact ages of the children of these caregivers were not documented during FGDs; however, the field team confirmed that the child's age met the inclusion criteria during recruitment (6-36-months old).

| Barriers
Health personnel emphasized side effects reported by caregiversmost frequently constipation, vomiting, and diarrhoea-as the primary reasons caregivers did not use or discontinued use of MNP.

| Facilitators
All health personnel and the majority of caregivers demonstrated a good level of knowledge when asked about the causes and consequences of anaemia, as well as the use and administration of MNP.
Additionally, caregivers and health personnel mentioned that awareness of the long-term effects of anaemia on the child's brain devel-

| Facilitators
Though some caregivers mentioned that family members and peers were not supportive, others shared receiving support from family and peers for their MNP use. Some of the support was only verbal, while other times people in their immediate circle of family and friends took a more active role in helping the caregiver with MNP administration, such as by reminding them to give it to the child or sharing ideas on how to make administration easier. One mother shared, "[The

child's] Dad is the only one who gives it to her. Sometimes I forget. I'm cooking and I already forget, but he says, 'Don't forget to give her
Chispitas.'" 3.6 | Community or organizational level

| Barriers
The key barriers that emerged at the community or organizational level were restricted access to MNP and limited informational resources about MNP use. One caregiver described an experience of being charged for MNP: They told me they would give me a prescription so I

| Facilitators
Two key facilitators at the community or organizational level were support from organizations external to the health system and well- Instead of filling things out like a maniac and not counselling her. We hate to fill out forms. Because we provide care-dedicating yourself to the patient, not dedicating yourself to filling out paperwork.
Finally, health personnel also showed dissatisfaction with training opportunities. Trainings are specific to health professions (i.e., doctors attend different trainings than nurses), and each health establishment is only required to have one representative present at the training, who is then responsible for sharing the information with others.
Health personnel shared that fragmented training caused lack of coordination among their coworkers, sometimes leading to inconsistent counselling practices. One health personnel explained, "We don't all work in the same language, we don't all have the same interest. If you could visualize the care that each one gives, how they work … it's different. Therefore, the information that they give [the patient] is different." Additionally, some caregivers expressed doubt about the quality of MNP linked to its distribution by the government. One caregiver shared, "My grandparents say that it kills neurons. That the government is giving this to us so that we are more foolish and don't know the truth." Health personnel reported hearing similar doubts from caregivers.

| Facilitators
Despite  Jefferds et al., 2010;Kodish et al., 2011;Munares-García & Gómez-Guizado, 2016;Osei et al., 2014;Sarma et al., 2016), and other studies in Peru support the idea that poor taste can cause complications in MNP administration (Creed-Kanashiro et al., 2016;Huamán-Espino et al., 2012). However, other studies show that the extent to which side effects influence MNP adherence can vary based on the counselling provided and that caregivers are less likely to forego administration if they are warned about these issues (Pelletier & DePee, 2019;Tumilowicz, Schnefke, Neufeld, & Pelto, 2017). It is possible that this theory also applies to the taste of MNP. Given that we found that caregivers frequently complained about taste while health personnel said MNP were tasteless when prepared correctly, it seems unlikely that health personnel are trained to preemptively address complications with MNP. Other studies also show that interactions with the people who distribute MNP can affect caregivers' sentiments towards and use of the product (Creed-Kanashiro et al., 2016;Kodish et al., 2011;Munares-García & Gómez-Guizado, 2016;Sarma et al., 2016). Similarly, we found that caregivers were dissuaded from using MNP if they felt they had received insufficient information from health personnel or had other negative experiences at health establishments. Issues in these interactions may be rooted in limited and inconsistent training opportunities among those who distribute MNP, which was found to be a barrier in our study as well as others (Creed-Kanashiro et al., 2016;Stephen Kodish et al., 2011 These findings suggest potential applications to strengthen the MNP intervention in this setting. One possible application is to strengthen collaborations with community organizations and leaders.
Both caregivers and health personnel recognized Cuna Más as an important resource and desired additional community engagement.
Other institutions could be involved, such as community kitchens, which are common throughout peri-urban Peru. Another community engagement strategy could be to implement educational campaigns with community health promoters, which has been successful in promoting multimicronutrient supplements in India, Indonesia, and Chiclayo, Peru (Dongre, Deshmukh, & Garg, 2011;Gross, Diaz, & Valle, 2006;Shankar et al., 2009 Finally, it is important to note that though current national policies to address childhood anaemia include provision and promotion of MNP, these components are part of a larger holistic approach led by the government that includes strategies such as anaemia treatment, diagnosis and treatment of respiratory and digestive infections, promotion of hygiene and handwashing, among others (Ministerio de Desarrollo e Inclusión Social, 2018). To get a full picture of how national policies to address anaemia are being implemented and received, as well as the effect they are having on the target population, these interventions also warrant future research.
By utilizing the SEM and triangulating our data between caregivers and health personnel, we found that community or organizational and policy barriers were at the crux of programme success, specifically the limited time health personnel have to share information and resolve caregiver doubts about MNP during appointments and the lack of informational resources external to these appointments.
Thus, interventions that increase informational resources and reduce caregiver reliance on individualized time with health personnel, such as increased community engagement, mass media campaigns, and group counselling, could potentially lead to improved MNP uptake and adherence. To get a more complete picture of the factors that affect MNP use in Peru, future research should incorporate caregivers who have limited contact with the public health system and should elaborate on findings using observational and quantitative methodologies.

ACKNOWLEDGMENTS
We would like to express our gratitude to Dr. Ricardo Castillo-Neyra,

CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest.

CONTRIBUTIONS
KR, ARP, and VPS were responsible for the conception of the study and study design. RAO and VPS acquired the funding for the study.
KR, VPS, and RAO provided project administration, and VPS and RAO provided research guidance as well. JDB, MPS, and KR collected the data. JDB and MPS analysed the data. JDB wrote the original draft of the manuscript. All co-authors edited and reflected on following drafts and gave final approval of the manuscript.