Relapse of undernutrition and food insecurity in a nutritional program in HIV care: mixed-methods study in Tigray region, Ethiopia.

Background Food insecurity is one of the major contributors to poor attainment of nutritional recovery among people living with HIV who are enrolled in nutritional programs. Nevertheless, nutritional programs in HIV care settings implemented in many countries are not designed to address food insecurity. This study in Tigray region Ethiopia examined relapse of undernutrition, and in particular how food insecurity had an impact on effectiveness of the nutritional program, specifically relapse of undernutrition after nutritional recovery Methods This study employed mixed-methods approach involving quantitative and qualitative methods. In the quantitative part, hospital records were collected of 1757 adults and 236 children living with HIV who were enrolled in the nutritional program. Logistic and cox-regression analysis were used to analyse the data. In the qualitative study, data was collected through in-depth interviews with 20 adults, 15 caregivers of children living with HIV, and 13 health providers and program managers. Thematic framework analysis was used to analyse the qualitative data. Results Among those who graduated from the nutritional program, 18% of adults and 7% of children relapsed. Lower educational status (primary and secondary), no membership of a community HIV support group, ambulatory and bedridden functional status, longer periods on ART (more than 24 months), presence of an opportunistic infection and poor baseline nutritional status were associated with relapse. Furthermore, those from rural areas, who did not attend formal education, were employed and had bedridden functional status, anaemia and worst nutritional status were likely to have more frequent episodes of relapse than their counterparts. Findings of the qualitative study also highlighted that poverty, poor livelihood, and food insecurity were the fundamental challenges to the effectiveness of nutritional programs in HIV care including relapse. Household food insecurity contributed to the selling and sharing of the nutritional supports and negatively impacted program effectiveness by contributing to relapse of undernutrition. Conclusions nutritional programs take into consideration the underlying determinants of food insecurity in the design, implementation, and funding of nutritional programs in HIV care, the success of the nutritional programs like those implemented in Ethiopia will be undermined.

and mechanisms for their implementation vary widely across countries. Many nutritional programs in sub-Saharan Africa provide ready to use therapeutic foods (RUTF) or Ready to Use supplementary foods (RUSF) for a short period of time to tackle undernutrition amongst people living with HIV [15,16]. Alongside nutritional support, nutritional assessment and counseling are the essential components of nutritional programs in HIV care in sub-Saharan Africa [17].
Clinic-based short term food assistance programs have been proven to be beneficial for food-insecure HIV patients to improve ART adherence and temporary weight gain but do not sustainably overcome the long term nutritional needs of people living with HIV [18][19][20][21]. The absence of long term positive impact of food assistance on nutritional status among people living with HIV may be related to a narrow focus of the programs.
Various demographic and socioeconomic factors contribute to food insecurity among people living with HIV. Sociodemographic factors such as household family size, marital status (not married) and older age were associated with food insecurity among people living with HIV in Kenya [6]. An Ethiopian study reported that elementary or lower educational status, below-average monthly family income, and lower food diversity were associated with food insecurity among people living with HIV who were on ART [7,22]. Other studies from other parts of sub-Saharan Africa have also shown that lower educational status, absence of nutritional support and low socioeconomic status were associated with poorer food insecurity [23,24].
Research suggests that food insecurity is one of the major contributors to low attainment of nutritional recovery among people living with HIV enrolled in a nutritional care [25,26]. Drop out (default) from the nutritional program in HIV care settings have also been found to be related to household food insecurity in sub-Saharan Africa [25,26]. However, despite the high magnitude of food insecurity amongst people living with HIV and the role of demographic and socioeconomic issues to undernutrition, nutritional programs in HIV care settings are not designed to respond to such issues.
Nutritional programs in HIV care neither consider nor assess individual or household food insecurity when enrolling people living with HIV into nutritional programs. For this reason, maintaining nutritional status after nutritional recovery is a major challenge for both adults and children enrolled in the programs [25][26][27][28]. A study from Ethiopia showed that nutritional programs in HIV care settings were characterised by high (20%) relapse of undernutrition after nutritional recovery [25]. A small number of studies have highlighted that sharing and selling of the nutritional support are common practices among food insecure households [25,29]. However, none of these studies have examined the impact of food insecurity on the effectiveness of the nutritional program, particularly when it comes to relapse of undernutrition. In addition, in a study that examined the use, perception, and acceptability of a nutritional support in HIV care, food insecurity was common among those enrolled in the nutritional program, though the impact on effectiveness of the nutritional programs was not explored [30]. Overall, there is a lack of evidence on how food insecurity may influence the effectiveness of nutritional programs in HIV care in Ethiopia-particularly in relation to relapse of undernutrition after nutritional recovery. Hence, this study aims to address this gap by examining the extent and predictors of relapse of undernutrition for patients enrolled in the nutritional program in HIV care in Ethiopia, with a particular focus on how food insecurity contributed to the problems of effectiveness of the nutritional program in terms of relapse of undernutrition.

Description of the nutritional program and study settings
This study was conducted in three purposively selected general hospitals (Mekelle, Lemlem Karl, and Shul) in the Tigray region, Ethiopia. In the Tigray region, there were 37 hospitals including 20 primary district hospitals, 15 general hospitals and 2 regional referral hospitals at the date of data collection (August 2016) [31]. The three hospitals were selected based on the distance from where health service was managed, coordinated and supervised in Tigray region. One hospital (Mekelle hospital) was the closest and located in Mekelle city where the health system in the region was managed and supervised while the remaining two (hospitals Shul and Lemlem Karl) were far from Mekelle city with approximately comparable distance from the centre. The selection of these hospitals was also made to minimize double-counting due to inter-hospital transfer of the study participants.
In Ethiopia, all HIV patients enrolled in chronic HIV care are regularly screened for undernutrition during their follow up visits for HIV care services. After enrolment in the nutritional program, sociodemographic, clinical, anthropometric, and nutritional outcome data is collected for monitoring and provision of RUTF or RUSF nutritional support. The enrolment and exit criteria and ration sizes are different for children and adult HIV patients enrolled in the nutritional program. While RUTF is given to individuals with severe acute undernutrition, RUSF is provided to moderate and mild undernutrition.
Depending on the severity of undernutrition, adult and child patients are provided with two sachets of RUSF daily for a maximum of three months for moderate acute malnutrition (MAM) or body mass index (BMI) of ≤ 17.99. For individuals with severe acute undernutrition, four sachets of RUTF daily for a maximum of six months for severe acute undernutrition (SAM) or BMI ≤ 16, together with counselling is provided [25,32].
An HIV patient (adult or a child) enrolled in the nutritional program is recorded as having an outcome of 'recovery', 'non-response', 'default' or 'relapse' after nutritional support for the program durations, summarised below (Table 1): Table 1 Nutritional program outcomes in the nutritional program in Ethiopia. in the research if they were interested. They were assured that their participation or otherwise in the research would be confidential and have no impact on their access to services. Health providers and program managers were identified as working in the program for more than 1 year and invited to participate in the research by email, where they could contact the researcher if they would like to be involved. Participant recruitment continued until data saturation was achieved. Interviews were conducted at a venue convenient to participants, largely the HIV service while the program managers were interviewed at their workplace (Tigray regional Health Bureau head office).

Data collection procedure
Nine data collectors (three in each hospital) familiar with the hospital data management system were recruited and trained to collect quantitative data. Two different checklists for adults and children were interview guide was field-tested and further refined based on emerging concepts and ideas. In-depth interviews were conducted by the first author (FT) in the local language Tigrigna and audio-recorded.
Comprehensive field notes were taken during the fieldwork.

Data management and analysis
Quantitative data analysis involved description of the demographic, socioeconomic, clinical, immunological, nutritional and anthropometric characterises of adult and child records. In addition, description of the outcome variables such as relapses of undernutrition and frequency of relapse of undernutrition were done. For relapses of undernutrition after nutritional recovery, Kaplan Meier curve was used to estimate the average time to relapses.
To identify the baseline determinants of relapse of undernutrition, bivariate Cox regression analysis was used to estimate the crude hazard ratio. Independent variables which were statistically significant at p < 0.3 were taken to multivariate cox regression to estimate the adjusted hazard ratio, with 95% CI. In multivariate Cox regression, statistical significance was declared at p < 0.05 and the hazard ratio was used to interpret the results.
The second outcome variable was the frequency of relapse of undernutrition. To identify the determinants of frequency of relapse of undernutrition, bivariate logistic regression was conducted to estimate the crude odds ratio and statistical significance was considered at p-value < 0.3.
Assumptions of logistic regression such as multicollinearity and homogeneity of variance were checked and no collinearity was identified. Factors that were found to be statistically significant at p < 0.3 were taken to multivariate logistic regression model to determine the independent predictors of the frequency of relapse. Adjusted odds ratio with 95% CI were used to determine the magnitude and direction of relationship and statistical significance was considered at p < 0.05.
Qualitative interviews were translated and transcribed into English by the first author (FT). Framework thematic analysis [36] was used to analyse the qualitative data. A coding framework was developed and discussed in the team meetings until a consensus was reached among the authors. Three interviews were double coded by the authors and differences were discussed until consensus was reached. Data was analysed using QSR NVivo and themes and categories emerged after thorough reading and understanding of the interview data, field notes, and memos. Illustrative quotes are used to describe themes and categories.

Results Of Quantitative Study
Demographic and socioeconomic characteristics of adults and children living with HIV The most common age group for adults was the 26-35 age group (42.6%), 63% were female and twothirds lived in urban areas ( Table 2). More than 40% were married, and over three quarters (76.6%) had children, with the majority of people living in a household of less than 5 people. Nearly 62% of adults attended at least primary and secondary education and a large proportion of adults and parents of children living with HIV were unemployed. The vast majority of people reported their religion as Orthodox or other Christian religion. Over three quarters had disclosed their positive HIV status to at least someone, and 81% of adults were not members of a community HIV support group.
The most common age group for children was 5-10 years (43.2%) and 47.5% were female. The vast majority of children (82.6%) of children lived with their parents (of whom more than half were living together), and three quarters lived in urban areas. There were similar proportions of the children who were the first, second, third and fourth-child in their family. Most children attended school. Table 2 Demographic and socioeconomic of adults and children enrolled in the nutritional program. Clinical and immunological characteristics of adults and children living with HIV The functional status of the majority of adult participants was 'working' (83.8%), while 50.5% of adults and 56% of children were in WHO clinical stage I. Furthermore, 94% of adults and 88% of children were on ART at baseline or at the time of enrolment in the nutritional program. In addition, 45.7% of adults were anaemic at enrolment. 64% of adults and 72% of children had been on ART for 24 months at enrolment to the nutritional program. A quarter (25%) of adults and 22% of children had opportunistic infections, with TB being the major opportunistic infection in both groups (Table 3). Table 3 Clinical and immunological characteristics of adults and children enrolled in the nutritional program. Almost 78% of adults received less than three sachets per day during their engagement with the program. Having good appetite was one of the criteria for enrolment into the nutritional program and almost all of the adults reported having good appetite. Those with poor appetite were treated for the cause before enrolment into the nutritional program. Usually, the causes of poor appetite were opportunistic infections (Table 4).
After enrolment in the nutritional program, 55% of adults and 71% of children recovered or achieved the graduation criteria of a target BMI. Non-response (not meeting target BMI) was identified in 21% of adults and 14% of children. Amongst the study cohort, 18% of adults and 14% of children failed to complete the program. Among those who achieved nutritional recovery, 18% of adults and 7% of children later relapsed (Table 4).

Determinants of relapses of undernutrition
The total time to relapse after nutritional recovery in the nutritional program was 68.5 months (95% CI 67.0-69.9). We conducted bivariate and multivariate Cox regression to identify the independent predictors of relapses of undernutrition after nutritional recovery.
In multivariate Cox regression, statistical significance was declared at P < 0.05. Accordingly, educational status, membership of an HIV community support group, duration on ART, ambulatory and bedridden functional status, presence of opportunistic infections, moderate and severe acute 13 undernutrition were significantly associated with relapse of undernutrition after nutritional recovery.
Adults who attended primary and secondary education were 2.8 and 3.7 times more likely to relapse than those who attended tertiary and above, but there was no difference for those who had no education. Non-membership of HIV community support group was associated with a 1.7 times higher chance of relapse than those who were members of community HIV support group. In addition, those who had been on ART for longer periods were more likely to relapse than those who had been less than 24 months on ART. Adults who had opportunistic infections at baseline were 1.7 times the risk of relapse of undernutrition after nutritional recovery (Table 5). relapsed three and four times respectively after nutritional recovery.
In multivariate logistic regression, statistical significance was declared at P < 0.05 and place of residence, education, employment status, functional status, presence of anaemia, and baseline nutritional status were statistically significantly associated with frequency of relapse of undernutrition. Individuals who were from rural areas were three times more likely to relapse more than once than those from urban areas. Those who had not attended formal education were more likely to relapse more than once, compared to those who attended secondary education. There was no difference between those attending primary, secondary education and above. Regarding employment, those who were employed were more likely to relapse more than once than those who were unemployed. Individuals who were bedridden were five times more likely to relapse more than once than those who were healthy. Individuals were anaemic at enrolment were nine times more likely to relapse more than once than those who were not. Those who were severely and moderately undernourished at baseline were four and nine times respectively more likely to relapse more than once than those who had mild undernutrition (Table 6).

Results Of The Qualitative Findings
In the qualitative analysis, food insecurity emerged as one of the key factors that directly and indirectly influenced undernutrition, nutrition program participation and utilisation, and program outcomes -through its impact on weight loss, motivation to enrol in the program, driver of selling and sharing the nutritional support, as a disincentive to graduate from the program and final impact on relapse.
Food insecurity, poor livelihood, poverty and undernutrition

Food insecurity as a motivation to participate in the nutritional program
Lack of access to adequate food was one of the key motivations for adults and caregivers to enroll in the nutritional program. Those who lived in urban areas and did not have a reliable source of income or were unemployed reported a lack of access to enough food in their household and that lack of adequate food was a motivation to enroll in the nutritional program. These motivations were shared by both male and female adult participants, reporting circumstances of food shortage in the household.

Food insecurity as a contributor to relapse
Despite the high magnitude of food insecurity among those enrolled in the nutritional program, health providers identified that there was no remedial strategy as part of the program to prevent relapses of undernutrition among those graduated. However, the above claim may be a perception as no other participant reported such an experience.

Discussion
This study highlights that household food insecurity coupled with underlying factors such as poverty and poor livelihood undermines the effectiveness of the nutritional programs in HIV care settings, particularly in relation to relapse of undernutrition. Food insecurity was explored in qualitative study and variables which may indirectly relate with food insecurity were also significantly associated with relapse in the quantitative findings. In addition, the link between food insecurity and the nutritional program emerged during the qualitative data analysis. Hence, this discussion will combine the qualitative and quantitative findings on how they relate with food insecurity and undermine the effectiveness of the nutritional program mainly relapse of undernutrition.
Studies in sub-Saharan Africa [37][38][39] have indicated that people living with HIV are more likely to be unemployed and less skilled. Consistent with the earlier studies, the current study demonstrated high rates of unemployment including from the quantitative sample, and many participants in the qualitative interviews also reporting no or unreliable sources of income related to lack of employment.
The lack of employment seemed to place significant stress on household income and contributed to food insecurity, and were reported as key motivators for enrolment in the program in the qualitative study.
While both adults and children in the study regained weight and achieved nutritional recovery after enrolment in the nutritional program, sustaining nutritional wellbeing in the presence of household food insecurity was a challenge. In the current study, 18% of adults and 7% of children relapsed after nutritional recovery, and 27% of adults relapsed more than once. These findings are significant as they are an indication of the contribution of demographic, socioeconomic, clinical and nutritional factors to the relapse of undernutrition after nutritional recovery.
The quantitative analysis found that those with lower levels of education were more likely to relapse and that those who did not attend formal education were more likely to relapse more frequently.
These are novel findings as we were unable to find previous studies that have examined the relationship between educational status and relapse of undernutrition. It is well acknowledged that a better education is related with better health more generally [40][41][42], and greater education could be related to enhanced understanding of the program, including the nutritional counseling, and strategies to prevent relapse. Those who were not employed were more likely to replace. This may relate to poor clinical status making work difficult or could also reflect the impact of lower income on food security. Surprisingly, those who were employed were more likely to relapse more than once compared to those who were not employed.
Another study from Ethiopia and Uganda found that people from rural areas have limited access to basic services such as health, education and other social services than their urban counterparts [43].
In the current study being a rural dweller was associated with being more likely to relapse multiple times than their urban counterparts. This suggests the need to consider contextual factors in supporting positive program outcomes.
Another interesting finding is that adults involved in community HIV support groups were less likely to relapse, community HIV support group was not related with the frequency of relapse of undernutrition. This could be related to community HIV support groups' role in reducing food insecurity among people living with HIV and also through improved access to services, they could offer greater support in maintaining weight and reducing HIV related stigma [44,45].
In addition, clinical and nutritional factors such as duration on ART, presence of opportunistic infections and anaemia were associated with relapse of undernutrition. For example, anaemia among people living with HIV is common due to the HIV illness and some ART medications [46,47] and in the current study, those with anaemia and severe and moderate acute undernutrition at enrolment were more likely to relapse. Those with the worst clinical conditions (bedridden functional status) were also more likely to relapse more than once than those apparently healthy. These clinical and nutritional conditions may contribute to undernutrition directly because worst HIV infection contributes to undernutrition [48], and also indirectly by limiting the individual from contributing economically leading to poverty and food insecurity. This is consistent with literature which states that prolonged illnesses are one of the essential contributors of poverty and food insecurity among people living with HIV [11,49]. However, there are no other studies demonstrating the role of these clinical and nutritional characteristics in contributing to relapse of undernutrition after nutritional recovery among those enrolled in the nutritional programs.
While undernutrition in people living with HIV is underpinned by underlying determinants of food insecurity such as poverty, poor livelihood, it seems that the nutritional program primarily focuses on short term treatment of undernutrition, dealing with "tip of the iceberg". As a result, the effectiveness of the nutritional program in improving nutritional status is hampered, including the reluctance of people to graduate from the program.
Similarly, sharing and selling of the nutritional support was a common practice and appeared to be driven by food insecurity and its underlying determinants. In the current study, adults and caregivers shared the nutritional support to fulfill dietary needs of their family due to lack of adequate food in the households. These findings support a study conducted in Ethiopia which reported RUTF sharing was more common in poor households indicating the negative impact of food insecurity on the nutritional program [29]. Although no program participants reported selling the nutritional support, health providers stated that this happened and did so for families to fulfill family nutritional needs.
Selling and sharing undermine the effectiveness of the program through reducing and changing ration size.
While the study has provided important qualitative and quantitative evidence regarding the predictors of relapse and the role of food insecurity, there were a number of limitations. While the relationship between food insecurity and relapse emerged during the qualitative analysis, it was not directly measured in the quantitative study where it was not possible to directly measure the relationship between food insecurity and relapse of undernutrition. The other limitations include that it was not possible to validate the accuracy and precision of measurement of the demographic, socioeconomic, clinical and nutritional characteristics as only secondary data was used. It is also likely that the originality of information in the qualitative study may have been lost during the translation and transcription, even though efforts were made to maintain the original conversation as narrated by the participants.

Conclusions
In conclusion, a significant proportion of adults and children relapsed after nutritional recovery and qualitative findings highlighted that food insecurity, poverty, and poor livelihood were common among people living with HIV enrolled in the nutritional program. Food insecurity contributed to the problems of effectiveness of the nutritional program in three ways including:(i) many adults and caregivers were motivated to join the nutritional program because of the food insecurity, (ii) Individuals from food-insecure households shared and sold the nutritional support to fulfill family dietary needs, and (iii) a significant number of adults and children relapsed because after graduation they were food insecure, leading to the loss of weight that was gained from the program. Various

Declarations
Ethical issues: There were no direct ethical concerns regarding the quantitative data, but issues of anonymity and confidentiality were sources of concern in the qualitative study. To address these concerns for the qualitative study participants, strategies such as reassurance, maintenance of confidentiality and anonymity was employed. In addition, qualitative study participants were provided with compensation for the time they spent during the interview. Ethical clearance was also secured from the Flinders University Social and Behavioural Research Ethics Committee (SBREC) (7118) in Australia and Mekelle University Ethical review committee (ERC 06211/2016), Ethiopia.

Consent to participant:
Written informed consent was obtained from all study participants.

Consent for publication: Not applicable.
Availability of data and material: The datasets used and/or analysed during the current study are not possible to share with a third party because of the agreement between the first author and the study participants to only use it for the purpose of current study. However, the authors can share deidentified aggregate data.

Competing interests:
The authors declare that they have no competing interests.