Diluted F100 v infant formula in treatment of severely malnourished infants < 6 months

By Caroline Wilkinson and Sheila Isanaka

Caroline Wilkinson was Nutrition Advisor with Action Contre la Faim - France (ACF-F), until November 2008. She spent most of 2007 in the Democratic Republic of the Congo (DRC) undertaking this study, a reflection of her particular interest in infant malnutrition. Currently she is Nutrition Advisor with UNHCR and chair of the Nutrition in Emergencies Working Group under the UN Standing Committee on Nutrition (SCN).

Sheila Isanaka is an experienced researcher in the field of nutrition in humanitarian contexts. This includes work with the SCN on the implications of the 2006 WHO Growth Standards for therapeutic feeding programmes and with Medecins sans Frontieres (MSF) in an analysis of their response to the 2007 crisis in Niger.

The authors would like to acknowledge the hard work and contribution of ACF International and Ministry of Health teams, the scientific partners involved in this project, Rebecca Brown, ACF-UK, and of course, the mothers and infants who participated in the study. The study was financed by ACF-F.

This article reports the main findings of a study which aimed to address one of the ongoing debates concerning the treatment of infants <6m - which supplemental milk is the most appropriate to use in their treatment?

Much progress has been made in the past 20 years concerning the management of severe acute malnutrition (SAM), particularly in children. However, while admission and discharge criteria and treatment protocols are clearly defined and more or less harmonised amongst actors at an international level for children over 6 months of age, this is not the case for severely malnourished infants less than six months of age (infants <6m).

Infants <6m pose particular challenges in terms of their vulnerability and immaturity. Treatment necessitates specialised care that is usually inpatient based. However the evidence base for management is weak, with no consensus on the case definition of SAM, management protocols (e.g. which supplemental milks and quantities to use, routine medical treatment, the focus on relactation/establishing exclusive breastfeeding as a primary objective) or indicators for defining success of the treatment. This means management can vary greatly by programme. It also risks that infants <6m are excluded from admission without follow-up, are admitted but are treated inappropriately (e.g. introducing supplemental feeds to exclusively breastfed infants), or only access treatment programmes once they reach 6 months of age.

A child in therapeutic programme in Sittwe, Myanmar

Infants <6m can represent a large proportion of admissions to inpatient feeding units. For example, in Action contre la Faim (ACF) programmes in South Kivu, Democratic Republic of Congo (DRC) in 2006, infants <6m represented 36% of total admissions, while constituting 21% and 12% of admissions in Myanmar and Niger, respectively. Since infants <6m are rarely measured in nutrition surveys, it is difficult to ascertain the coverage of therapeutic programmes and the extent of infant <6m SAM in the population¹. However with the majority of acutely malnourished children now managed in community-based programmes, the proportion of infants <6m amongst inpatient admissions is likely to rise. Improved programme coverage and community awareness may also increase absolute numbers admitted. In addition, changing from the National Center for Health Statistics (NCHS) references to the 2006 WHO standards significantly increases the proportion of infants <6m classified as severely malnourished².

Mortality rates amongst the youngest infants treated for SAM appear to be higher than in older children and their fragility makes their treatment more complex than for older children. Data from children admitted for treatment in ACF therapeutic feeding centres (TFCs) in Kabul in 1999³ show that the younger infants had much lower survival rates than their older counterparts (see box 1). Data from the same centres in Afghanistan between 2002 and 2004 showed improved mortality rates of infants <6m of age, at 4.7% of total admissions⁴. This improvement is undoubtedly due to a change in admission criteria, identifying infants earlier, and to improved management of these infants and better support to their mothers.

Objectives

ACF undertook a study to compare the efficacy of diluted F100 (DF100) milk currently used in the treatment of severe acute malnutrition in nonoedematous infants <6m, with a standard generic infant formula (IF) in an inpatient TFC setting.

The study was conducted in two ACF-USA managed TFCs in Baraka and Uvira, South Kivu, DRC and in two ACF-Spain managed TFCs in Mayahi and Keita, Niger, between June 2007 and April 2008. The DRC results only are presented here.

Methodology

Full details of the methodology are not included here (available upon request). Current ACF treatment protocols were applied in this study. The aim of the treatment was to support breastfeeding and to stimulate breastmilk production. The main admission criterion was failure of effective breastfeeding and the main discharge criterion was appropriate weight gain on breastmilk alone (exclusive breastfeeding) (see Box 2). Infants with nutritional oedema, major physical or mental disability affecting their ability to breastfeed or those without a mother or caregiver willing to breastfeed, were excluded from the study and were treated following protocols appropriate to their situation.

On admission to the centre, eligible infants <6m were randomised into one of the two study groups; either DF100 (control) or IF (intervention). Apart from receiving different milks, treatment of the infants and mothers between the two groups was identical.

Treatment involved infants <6m receiving routine medications according to their age and weight categories (Vitamin A, folic acid, and antibiotics). Breastmilk production was stimulated wherever possible using the Supplementary Suckling Technique (SST) in association with the supplemental milks⁵. Mothers were encouraged to breastfeed as often as possible. The milk supplement was decreased to half when the infant was gaining at least 20g/day for 3 consecutive days and was completely stopped when the infant maintained 10g/day weight gain. The infant was discharged when weight gain continued at 10g/day for at least 3 but up to 5 consecutive days. Five days were the preferred observation period to evaluate weight gain on breastmilk alone, however many mothers found it difficult to remain in the centre once the infant was no longer receiving supplemental milk.

Results

A total of 146 infants were included in analysis, 74 in the DF100 control group and 72 in the IF intervention group (Box 3). This was appreciably less than the sample size of 300 infants calculated at the outset (150 in each group) and was due to lower admission numbers in 2007 than the projections made on 2006 data. A total of 98 infants recovered and contributed to the analysis of weight gain and length of stay data.

Standard treatment outcomes of weight gain (g/kg/day) and length of stay from admission to discharge (days) as well as type of discharge (recovered, death, defaulter or transfer) were compared between the two study groups. The primary study outcomes were weight gain (g/kg/day) and death rate.

Infant and maternal characteristics

Overall, 65% of infants were less than 2 months of age on admission, 56% were girls and 44% were boys. Mean weight on admission was 2.3 kg (SD 0.6) and mean length was 47.8 cm (SD 5.6). One third (32%) of the infants were multiple births. Data on the weight-for-length index is not presented here, as 64% of the infants admitted were shorter than 49 cm in length, so weight for length index could not be calculated using the NCHS standards (32% were shorter than 45cm, also excluding them from calculation using the 2006 WHO standards). Mean weight-for-age and length-for-age upon admission (NCHS standards) were respectively -3.48 z scores (SD 0.7) and -3.48 z scores (SD 1.2).

Forty-one percent of infants <6m were reported as exclusively breastfed on admission, and 48% of mothers reported that the infant was born with weight <2.5kg. The most frequently reported reasons for admission were the infant not gaining weight at home (59%) and maternal-reported breastfeeding failure (50%). At time of admission, 64% of the mothers reported that their infant was ill; the most common complaints were weight loss, followed by diarrhoea, fever and cough. The majority received treatment other than the standard antibiotic treatment, suggesting that they were clinically unwell. Mean maternal age was 25.0 ± 7.4 years, 52% of the mothers who responded had attended school for some period in their lives and approximately one-quarter of mothers were first time mothers. Of the 146 infants studied, only 4 mothers complained of malaria upon admission. Maternal HIV status was not available. Maternal admission characteristics were much the same across the two treatment groups overall and within centre (Table 1).

Effect of intervention on weight gain (g/kg/day)

Intervention effect was estimated as the difference in mean weight gain (g/kg/day), from admission to discharge as recovered, between subjects in the intervention IF arm vs the control DF100 arm.

Overall, there was no difference in weight gain (g/kg/day) between the IF and the DF100 arms. IF resulted in a non significant (P=0.95) 0.1 g/kg/day (95% CI: -3.3, 3.5) greater weight gain than DF100 (Table 2). Analyses by centre showed directional differences, with IF resulting in greater weight gain than DF100 in Baraka and lower weight gain than DF100 in Uvira. However, these differences by centre were not statistically significant. There was also no statistically significant interaction of weight gain by infant age (< or ? 2 months old) overall or within centre.

Effect of intervention on length of stay (admission to discharge as recovered)

An infant <6m being weighed

The intervention showed no significant effect on length of stay (admission to discharge as recovered) overall, but analyses did show significant differences by centre (P = 0.04). Infants who received IF stayed under treatment for 4.7 days less (95% CI: -10.7, 1.3) than those under DF100 in Baraka and 3.5 days more (95% CI:-1.5, 8.5) than the DF100 group in Uvira (Table 2).

The intervention effect on total duration of treatment was not significantly modified by infant age, overall or within centre. Trends in the intervention effect by infant age showed similar patterns by centre and overall, with infants under IF staying less time than those under DF100 in Baraka and more time in Uvira (data not shown).

Effect of intervention on type of discharge

Incidence rates were used to describe the frequency of discharge types (death, recovery and defaulting) among all infants. Compared to DF100, IF had no significant effect on the rate of death (IRR⁶: 0.93 (95% CI: 0.41, 2.11)) or the rate of recovery (IRR: 1.06 (95% CI: 0.71, 1.58)). There was no evidence of significant differences by centre or by infant age for either of these types of discharge (see Table 3).

The rate of defaulting was nearly seven times greater in the IF group than in the DF100 arm in Baraka (95% CI: 0.84, 57.79), whilst there was a significant (70%) reduction in defaulting in the IF arm compared to the DF100 arm in Uvira (95% CI: 0.10, 0.93) (P for interaction = 0.01). No evidence of interaction by infant age was observed for defaulting.

All infants were exclusively breastfeeding on discharge for between 3 and 5 days. Although interpretation is problematic⁷, the infant exit characteristics were measured against the Sphere Standards to provide a comparison with the older children (see Table 4). The infant exits did not meet the Sphere Standards of < 10% deaths (16% overall) or > 75% recovered (67% overall) and showed worse outcomes than older infants and children admitted for severe acute malnutrition (in and outpatient) in the same centres between June and December 2007 (5% deaths, 72% recovered, and 7% defaulter).

* to a centre where the infant could not be followed up in the study

Interpretation

The results presented above tend to suggest that the treatment outcomes of weight gain (g/kg/day) and length of stay (days) between the group of infants that received DF100 and those that received IF are similar. There are some directional differences of note for weight gain and length of stay, as well as for the defaulter outcome, between the two centres of Baraka and Uvira. Results for type of discharge also tend to suggest similar results between the two treatment arms. The study, however, is not able to provide conclusive evidence as to the nature of these differences or similarities between the two types of milks due to the smaller than required sample size, and the directional differences observed for some of the outcomes (weight gain, length of stay and defaulting).

Discussion

This study once again highlighted the difficulties in the current case definitions of SAM in infants <6m and raises the question of how best to target those infants most at need of therapeutic interventions. The majority of the infants <6m included in the study were either very young and/or small at admission; 65% were under 2 months old, their mean weight < 2.5 kg and their mean length at < 49 cm. Nearly half had a history of low birth weight (<2.5kg), contributed to by a high multiple birth rate. Under these circumstances, it is not possible to use strict anthropometric criteria for admission alone, as this would exclude all of the infants under < 49 cm or < 45 cm, depending on whether NCHS or WHO based norms are used.

The majority of mothers reported that the infant was not gaining weight at home. However more often than not, birth weight data or growth monitoring data were unavailable. This made it difficult to determine whether the weight of the infant at admission was linked to prematurity, full term low birth weight or failure to thrive. Lack of serial weight measures at presentation also makes it impossible to differentiate an infant that is gaining weight steadily from an infant that is not gaining or is losing weight.

A mother breastfeeding her malnourished infants

In this study, nine infants were excluded after randomisation because they were submitted to a period of observation within the centre, prior to any treatment. These infants were gaining weight at a rate of at least 20g/day on exclusive breastfeeding. Thus we judged that this weight gain did not warrant the risk of admitting the infant for therapeutic care, exposing the infant to a higher risk of infection and interfering with exclusive breastfeeding by introducing supplemental milks. This approach would also be relevant in contexts of high HIV prevalence or where the HIV status of the mother is known to be positive; promoting exclusive breast-feeding and avoiding unnecessary mixed feeding if the infant is doing well on breastfeeding alone. This experience suggests a promising approach for determining which infants should be treated therapeutically and which should be referred for breastfeeding counselling and support. Breastfeeding status on admission relied on maternal report that likely overestimated the number of exclusively breastfed infants and early introduction of complementary foods was extremely common in both centre populations. It is clear that objective predictive indicators would help in judging which infants are at most need for therapeutic intervention.

Over half of the infants were clinically unwell on admission; there are clear problems with clinical diagnosis of illnesses amongst infants <6m. The overall quality of care did appear to differ by centre, with mothers and infants in one of the centres receiving more personalised and intensive care (medical and supportive) than in the other. This highlights the challenges to provision of appropriate care for infants and their mothers in the inpatient units. It requires more time, support and qualified care than for the older infants and children.

To be able to conclude with confidence that there is no difference between the outcomes of the infants treated on either the control DF100 or intervention IF, the study would have to be continued, increasing the sample size and hence the power of the study.

Given the increasing decentralisation of treatment for SAM in most contexts and the higher risk of death in the youngest age group of infants <6m, appropriate services for these infants who require inpatient care need to be available close to the homes of those mothers and the infants that require treatment. There is a need to rethink the way that we are currently identifying infants and mothers that require treatment and alternatives are needed for infants that do not require therapeutic intervention but whose mothers do require effective breastfeeding support. For those requiring therapeutic intervention using relactation techniques and supplemental milks, proven alternatives to DF100 would facilitate access to services at delocalised points.

Perspectives

This study provides the evidence that infants respond to both types of treatment with satisfactory weight gains in g/kg/day and an average length of stay of around 3 weeks. However, the evidence base for the treatment of acute malnutrition in infants needs strengthening.

ACF is committed to working towards the improvement of the management of acute malnutrition in infants and is working on this from both the nutritional as well as psychological perspectives. Programmes treating infants will continue to be implemented, evaluated, documented and lessons learned applied in the field for the improvement of the management of these young infants.

For further information or to obtain a copy of the full study report, please contact: Rebecca Brown, rbrown@actioncontrelafaim.org

¹To address this gap, analysis of the burden of acute malnutrition in infants <6m was undertaken as part of the Management of Acute Malnutrition (MAMI) Project, implemented by the ENN, UCL CIHD and ACF and funded by the Global Nutrition Cluster. Secondary data analysis of 22 DHS country datasets found the burden of acute malnutrition was significant, and considerably higher when using 2006 WHO Growth Standards. See summary report and Chapter 3 of the final report at www.ennonline.net/research. Findings will be summarised in the next issue of Field Exchange.

²See footnote 2.

³Golden M. Comment in including infants in Nutrition Surveys: Experience of ACF in Kabul City. Field Exchange 2000; 9: 16 - 17

⁴ACF 2007. Infants in Therapeutic Care: A descriptive study of infants admitted in therapeutic centres in Kabul, Afghanistan: 2002 - 2004 (available upon request).

⁵For supplementary suckling technique, see IFE Module 2, v1.1. Chapter 8, p86-90.

⁶Incidence Rate Ratio

⁷Comparison of disaggregated infant exit characteristics with Sphere Standards is probably not appropriate as the Sphere Standards were not established for use in this particular age group of infants <6 months old. Specific indicators would need to be determined for infants as the proportion of deaths would arguably be expected to be naturally higher than that of older infants and children.