CMAM in Cambodia – indicators of acute malnutrition for screening

By Jennifer Carter and Joel Conkle

Jennifer Carter is a second year MPH student at the Tulane University School of Public Health and Tropical Medicine in the Department of International Health and Development. She lived and worked in Cambodia for three years before returning to the country in May 2010 as a food security and nutrition intern for UNICEF Cambodia.

Joel Conkle is the Nutrition Specialist at UNICEF Cambodia. He has lived and worked in Camodia for three years and previously worked in Southern Africa as a nutrition data consultant for UNICEF.

The authors acknowledge the contribution of Magna Children at Risk and Samaritan's Purse and their staff for their continued dedication in everyday field work and for sharing the data and experiences reflected in this article. Particular thanks to Denisa Augustinova and Paul McKnight.

This article from Cambodia shares observed differences in acute malnutrition prevalence between WH and MUAC in national survey and some programming data that run counter to the pattern observed in other regions. The authors go on to discuss programming implications for the interim CMAM guidelines in Cambodia.

According to the WHO/UNICEF Joint Statement on WHO child growth standards and the identification of severe malnutrition in infants and children (2009), "the prevalence of severe acute malnutrition.based on weight-for-height below -3 SD of the WHO standards and those based on a mid upper arm circumference (MUAC) cut-off of 115 mm are very similar¹." Recent analysis of programme and survey data has shown that this is not the case in Cambodia. This issue has been noted in several studies in sub-Saharan Africa, particularly among pastoralist populations. There has, however, been little research on the discrepancy between weight-for-height (WH) and MUAC derived prevalences of acute malnutrition in Asia, where undernutrition and malnutrition follow very different patterns to other regions of the world.

In Cambodia, UNICEF has supported in-patient treatment of severe acute malnutrition for a number of years. In 2010, along with development partners, UNICEF is supporting the government to develop national guidelines for the management of acute malnutrition and to begin implementation of community screening and health centre based outpatient treatment with Ready to Use Therapeutic Food (RUTF).

Community based screening for acute malnutrition in the communities of Phnom Penh

Cambodia programme data

Data from screening by the non-governmental organisations (NGOs), Samaritan's Purse² and Magna Children at Risk³ indicates that far more children are identified as both moderately and severely acutely malnourished by WH z-scores than by MUAC criteria. Magna screening data (both WH and MUAC were used) was collected at a referral hospital in Kandal province, where the NGO operates a large, comprehensive programme for treatment of moderate and severe acute malnutrition in Cambodia. While the facility is a 24-hour paediatric ward where any sick child will be treated, many people in the surrounding community are aware of Magna's inpatient and outpatient programme for the treatment of malnutrition. Thus the children who comprise the selfselected population, who were screened prior to admission to the facility, are far more likely to present with acute malnutrition than children in a community setting. Recent analysis of data from screening at the Magna health facility shows that the estimated prevalence of moderate and severe wasting among patients (6 to 59 months of age) according to WH (< -2 SD) is 83.1% compared with 65.8% according to MUAC (<125 mm). Differences were also found to be greatest among older children (> 24 months), whereas prevalence estimates derived from WH and MUAC were found to be similar among younger children.

Further new anthropometry data from both Samaritan's Purse (collected in slum communities in Phnom Penh where the NGO is operating) and Magna are being collated and analysed and will be presented in a future issue of Field Exchange. This article focuses on the findings of a reanalysis of the Cambodia Anthropometric Survey (CAS) 2008⁴ prompted by the discrepancy noted between WH and MUAC derived prevalences of wasting in both community and facility-based programmes.

Re-analysis of CAS 2008

The CAS 2008 is a nationally representative sample of 7,495 households with children ages 0 to 59 months, making it the largest national sample of child measurements ever collected in Cambodia. The survey was conducted in order to ascertain the effects of the 2008 food price crisis on the health and nutrition of Cambodians. MUAC was included as an anthropometric measure due to the current debate over the use of WH versus MUAC as measures of acute malnutrition.

A highly significant finding from the survey was that between 2005 and 2008, all improvements in the prevalence of acute malnutrition had effectively halted. According to analysis of the Cambodia Demographic and Health Surveys (CDHS), using the 2006 WHO growth standards for all, between the years 2000 and 2005 Cambodia experienced a 1.7% yearly average decrease in wasting, with the prevalence falling from 16.8% in 2000 to 8.4% in 2005^5,6. The CAS 2008 determined the prevalence of wasting to be 8.9% and not statistically significantly different from the 2005 estimate⁷.

While prevalences of moderate and severe acute malnutrition derived from WH z-scores (< -2 SD) and MUAC-for-age (MUAC/A < -2 SD) were found to be similar in the Cambodia Anthropometric Survey (CAS) 2008, at 8.9% and 8.7% respectively, MUAC (< 125 mm) unadjusted for age produced a wasting prevalence of only 3.8% (UNICEF analysis, see Figure 1). This confirms that in Cambodia, differences in prevalences derived from MUAC and WH occur at the national level, as well as in community and facilitybased nutrition programmes. With regard to severe wasting, the prevalence among children aged 6 to 59 months according to MUAC was only one third of the prevalence according to WH (See Figure 2). The greatest correspondence between both indicators is for the prevalence of moderate wasting, where MUAC prevalence is around three quarters that of WH (see Figure 3).

Reasons for WH v MUAC differences in prevalence

Part of the discrepancy between MUAC and W/H can be attributed to measurement error. The height of the youngest children is more likely to be over estimated, which leads to low levels of stunting and high levels of wasting. However, over estimation occurs mostly in the 6-11 month age group and seems to disappear by 23 months of age. However, as reflected in Figures 1-3, the majority of the difference in wasting prevalence observed by MUAC and WH in CAS 2008 analysis was in older children (> 24 months) and attributable to MUAC not selecting older children. WH remains consistent across age groups in its identification of children as wasted, while MUAC preferentially selects younger children as wasted.

Community based screening for acute malnutrition in the communities of Phnom Penh

An additional issue when considering MUAC and WH is that the two indicators select different children. Of the 9% of children identified as wasted by either measure, only 2% were selected by both MUAC and WH (see Figure 4) . For severely wasted children, only 5 out of 145 children were selected by both (see Figure 5). While previous studies have shown a similar mortality rate in children selected with either indicator among hospitalised children, there is still some uncertainty about which indicator is more appropriate for community based screening of children for therapeutic feeding.

Programming implications

Findings from the CAS 2008, along with growing international support for the development of programmes targeting acute malnutrition in non-emergency settings, has led to the development of interim community based management of acute malnutrition (CMAM) guidelines for Cambodia. The guidelines will remain in draft form until sufficient evidence is gathered from the implementation of pilot programmes.

Recommending appropriate indicators of acute malnutrition is integral to ensuring that CMAM guidelines will allow for children at increased risk of mortality due to acute malnutrition to be identified as such in community and health facilities. As the Cambodia CMAM guidelines are being developed, it is important that data is used to inform choices of anthropometric measures. In this respect, it is significant that the difference in MUAC-derived and WHderived prevalences of wasting increases with age. The indicators produce more similar estimates of acute malnutrition for children under the age of two years than for older children. This is important given the fact that the predictive powers of MUAC and WH increase with age. It makes logical sense that the mortality prediction power of MUAC is strongest among older children (> 23 months)⁸. Having a small arm circumference relative to a set cut-off point at a young age is less likely to be indicative of increased risk of mortality than at an older age. While arm circumference increases slowly between birth and 4 years of age, it does indeed increase among healthy children. Similarly, severe deficits in WH produce an only moderately increased risk of mortality among young children (< 23 months) but a marked increase in risk after 2 years of age⁹. So during the years when there is the most discrepancy between MUAC and WH, both of these indices are likely to be at their highest mortality predictive power.

Community based screening for acute malnutrition in the communities of Phnom Penh

The fact that discrepancy between MUAC and WH increases with age has significant implications with regard to food security. Wasting among young children is usually indicative of recent disease often coupled with improper feeding practices, while wasting among older children is more indicative of food insecurity. In a 2009 report for the Integrated Food Security Phase Classification (IPC) Global Partners¹⁰, WH is recommended as a better indicator for monitoring changes in food security because it does not preferentially identify younger children as malnourished as MUAC has been shown to do. During periods of food insecurity, as the prevalence of wasting increases, older children are likely to experience a relatively greater increase in acute malnutrition than younger children. Thus it is possible that using MUAC alone will mask problems among older children and thus provide an inaccurate picture of food insecurity in a country or region¹¹.

Recommendations

MUAC and WH identifying different children as malnourished means that using only one indicator is likely to leave out a group of children with a similar risk of mortality. For this reason, the interim CMAM guidelines for Cambodia state that either a low MUAC score or a low WH score is grounds for inpatient or outpatient treatment of acute malnutrition, depending on the severity of the deficit and the presence of other clinical signs. This is distinct from the two-stage screening process and thus avoids the problem of 'rejected referrals,' where children referred to the health facility due to low MUAC are turned away from treatment because they do not meet the WH criterion. In a non-emergency setting such as Cambodia, the use of both MUAC and WH for community screening seems a fair compromise until additional evidence from CMAM pilot programmes in rural and urban settings can be obtained.

Community based screening for acute malnutrition in the communities of Phnom Penh

Findings from Samaritan's Purse, Magna, and the CAS 2008 regarding discrepancies between MUAC and WH warrant further investigation as to which is the better indicator of acute malnutrition. In particular, a facilitybased study is needed in order to determine whether MUAC or WH is more associated with clinical signs of malnutrition and mortality in Cambodia. Given the complex relationship that anthropometric measures have with age, there is evidence of the need to disaggregate estimates of risk (for mortality and morbidity) by age in order to better assess the indicators. There may be justification for using different indicators among different age groups, although this would add complexity and thus require sufficient evidence of differences within a population.

A separate but related issue in Cambodia is the need to revise the Integrated Management of Childhood Illness (IMCI) to be in line with the WHO/UNICEF Joint Statement on WHO child growth standards and the identification of severe malnutrition in infants and children. At present, IMCI is used at the health centre to diagnose and guide treatment of illness among children. The IMCI algorithm includes weightfor- age (WA) z-scores as the only measure of malnutrition among young children. Research has shown that this may be acceptable for children less than 2 years of age, when low weight is more likely due to wasting than stunting, but not for older children. Low WA in older children is more likely to be caused by stunting rather than wasting (a problem that will not respond to therapeutic feeding). Now that there is evidence that the prevalence given by MUAC and WH is not similar in Cambodia, more research is needed to properly inform the revision of IMCI protocol with respect to anthropometric indicators.

For more information, contact: Jennifer Carter, email: jenncart@gmail.com and Joel Conkle, email: jconkle@unicef.org

¹WHO child growth standards and the identification of severe acute malnutrition in infants and children. A joint statement by the World Health Organisation and the United Nations Children's Fund. May. 2009

²http://www.samaritanspurse.org/

³www.magnachildrenatrisk.org

⁴National Institute of Statistics (NIS), Ministry of Planning. 2008. English supplement to the Cambodia Anthropometric Survey 2008. Phnom Penh, Cambodia: National Institute of Statistics, Ministry of Planning prepared by UNICEF.

⁵National Institute of Statistics (NIS), Directorate General for Health [Cambodia], and ORC Macro. 2001. Cambodia Demographic and Health Survey 2000. Phnom Penh, Cambodia, and Calverton, Maryland USA: National Institute of Statistics, Directorate General for Health, and ORC Macro.

⁶National Institute of Statistics (NIS), Directorate General for Health [Cambodia], and ORC Macro. 2006. Cambodia Demographic and Health Survey 2005. Phnom Penh, Cambodia, and Calverton, Maryland USA: National Institute of Statistics, Directorate General for Health, and ORC Macro.

⁷See footnote 4.

⁸Pelletier, D.L. (1994). The relationship between child anthropometry and mortality in developing countries: implications for policy, programs and future research. Journal of Nutrition 124, 2047S-2081S.

⁹See footnote 8.

¹⁰Young, H. & Jaspers, S. (2009). Review of nutrition and mortality indicators for the integrated food security phase classification (IPC). SCN Task Force on Assessment, Monitoring and Evaluation, and The Integrated Food Security Phase Classification (IPC) Global Partners.

¹¹Bern, C. & Nathanail, L. (1995). Is mid-upper arm circumference a useful tool for screening in emergency settings? Lancet 345, 631-33.