Diagnostic criteria for severe acute malnutrition among infants under six months of age
Summary of research1
Location: Kenya
What we know: Diagnosis of acute malnutrition in infants under six months old (U6M) is currently based on weight-for-length z score (WHZ).
What this article adds: Data were analysed from a cohort of infants U6M admitted to Kili? County Hospital (KCH) in Kenya from 2007 to 2013 to determine the association of anthropometric indexes with risks of inpatient and post-discharge mortality. Among 2,882 admissions, 4.9% died in hospital (half within 48 hours) and 5.3% died within one year of discharge (50% follow-up). Mid upper arm circumference (MUAC) and weight-for-age z score (WAZ) predicted inpatient and post-discharge mortality better than WLZ (P < 0.0001). A single MUAC threshold of <11.0 cm performed similarly to age-adjusted MUAC thresholds (all P > 0.05) and better than WLZ <-3 for both inpatient and post-discharge mortality (both P < 0.001). Height for age (HAZ) performed similarly to WLZ. HIV exposure was associated with inpatient and post-discharge mortality. In conclusion, MUAC or WAZ identify infants U6m at highest risk of death.
Diagnosis of severe acute malnutrition (SAM) in infants under six months of age (U6M) is based on weight-for-length z score (WLZ), using the same thresholds as for older children (WHO, 2013). However, WHZ standards are not available for infants <45 cm in length and there are concerns about the reliability of anthropometric indexes in early infancy. In older children mid-upper arm circumference (MUAC) predicts mortality better than WLZ. Relations between anthropometry and mortality may be confounded by age, HIV and low birth weight. Diagnostic thresholds for WLZ, MUAC and other indexes have not been fully evaluated against mortality risk among U6M infants. The aim of this study was to determine the association of anthropometric indexes with risks of inpatient and post-discharge mortality among U6M infants recruited at the time of hospitalisation.
Methods
Data were analysed from a cohort of U6M infants (four weeks to six months of age) admitted to Kili? County Hospital (KCH), Kenya, between January 2007 and December 2013. Post-discharge analysis included the subset of infants discharged alive, resident within the Kili? Health and Demographic Surveillance System (KHDSS) and followed up from January 2007 to March 2014. Primary outcomes were inpatient death and death during follow-up over one year after discharge. Adjusted risk ratios (RRs) were calculated for inpatient mortality and hazard ratios (HRs) for post-discharge mortality for different anthropometric measures and thresholds, including weight-for-age z score (WAZ), length-for-age z score (LAZ), weight-for-length z score (WLZ) and mid-upper arm circumference (MUAC). Discriminatory value was assessed using receiver operating characteristic curves.
Results
A total of 2,882 infants were admitted and included in the initial analysis: 1,730 (60%) were male and the median age at admission was 3.0 months (inter-quartile range (IQR): 1.7–4.5 months). At admission, 642 infants (22%) were wasted (WLZ <-2); 317 (11%) of whom were severely wasted (WLZ <-3) and 962 (33%) were stunted (LAZ <-2). None of the infants had kwashiorkor (oedematous malnutrition); 191 infants (6.6%) had a positive HIV antibody test; 41 (21%) of whom had a WLZ <-3. A total of 346 infants were excluded from the main analysis because of missing data, mainly because the infant was too sick to measure. Forty-seven infants (14.7%) missing WLZ had a length <45 cm; hence WLZ could not be computed. Adjusted for age and HIV, the relative risks (RRs) of inpatient mortality associated with missing anthropometric indexes were: 3.04 (95% CI: 2.29, 4.04), 5.02 (95% CI: 3.14, 8.00), 3.11 (95% CI: 2.31, 4.19) and 3.56 (95% CI: 2.25, 5.65) for WLZ, WAZ, LAZ and MUAC respectively.
Anthropometry as a predictor of mortality
Overall, 140 of 2,882 infants (4.9%) died during admission. Sixty-nine (49%) deaths occurred within the ?rst 48 hours of hospitalisation. WLZ, WAZ, LAZ and MUAC were all associated with inpatient death even after adjusting for confounders. The current criteria for diagnosing SAM (WLZ <-3) identi?ed 317 infants (11%), of whom 40 (12.6%) died. WAZ identi?ed 630 (21.9%) severely underweight infants, of whom 77 (12.2%) died. MUAC <11.0 cm identi?ed 682 infants (23.7%), of whom 80 (11.7%) died, whereas among the 2,200 infants with MUAC ≥11.0 cm, 60 (2.7%) died. In multivariate analysis, HIV exposure and anthropometric criteria were consistently associated with inpatient mortality. Small size at birth (preterm or low birth weight) was associated with mortality in the WLZ model only.
The adjusted area under the curve (AUCs) for WLZ, WAZ, LAZ, and MUAC were 0.71 (95% Confidence Interval (CI):0.66,0.76), 0.76 (95% CI:0.72,0.81), 0.73 (95% CI: 0.68, 0.78) and 0.77 (95% CI: 0.73, 0.81) respectively. Compared with WLZ, MUAC and WAZ were better predictors of mortality (both P < 0.0001) and LAZ was similar to WLZ (P = 0.43).
Effect of age on MUAC criteria
AUCs for MUAC, WAZ and WLZ were plotted by month of age. The point estimates of AUC for MUAC and WAZ were consistently above those for WLZ. Statistically, the optimal MUAC cut-off was 11.2 cm (rounded down to 11.0 cm). However, MUAC thresholds differed with age, indicating that age-varying cut-offs should be investigated. To further understand whether single or varied MUAC cut-offs would best suit this age group, 11.0 cm was chosen as a reference threshold and the diagnostic performance of single and varied MUAC cut-offs were tested against this reference. For the single criterion between <10.0 and <11.5 cm, no difference in the association with the risk of mortality was found. From <11.0 cm; however, differences were found in case load and sensitivity for death. The performance of WAZ criteria of <-2 and <-3 was also no better than the use of MUAC <11.0 cm.
Post-discharge mortality
Of the 2,742 infants discharged alive, 1,455 (50%) lived within the area and were followed up for 12 months after discharge. Fifty infants (3.4%) had unknown outcomes; hence 1,405 were included in the ?nal analysis. During 1,318 child-years of observation, 75 infants (5.3%) died and there was a mortality rate of 57 (95% CI: 45–71) per 1,000 child-years of observation. The median time to death was 91 days (IQR: 40, 165 days). A total of 33 of 75 (44%) and 53 of 75 (71%) deaths occurred during the ?rst three and six months of follow-up. In multivariate analysis, HIV exposure and anthropometric criteria were associated with post-discharge mortality but not size at birth. Compared with WLZ, MUAC and WAZ were better predictors of mortality (both P < 0.001), but LAZ was similar (P = 0.93).
AUCs for MUAC, WAZ, and WLZ were plotted by month of age. Within each age group, the point estimate AUCs for MUAC and WAZ were consistently above those for WLZ. The statistically optimal MUAC threshold for all infants was <11.5 cm but varied between age groups. The performance of the MUAC cut-off <11.0 cm was tested against other single and varied MUAC cut-offs. For the single criterion between <10.0 and <11.5 cm, no differences in association with the risk of mortality were found from the reference, except for the cut-off <10.0 cm, which had a signi?cantly lower AUC (P = 0.003) than the reference. In all groups, differences in case load and sensitivity for death were noted. WAZ thresholds were not superior to MUAC <11.0 cm.
Discussion and conclusions
Among hospitalised infants U6m, malnutrition is common and is associated with inpatient and post-discharge mortality. MUAC and WAZ are better predictors of mortality than WLZ. A single MUAC threshold of <11.0 cm performed similarly to age-adjusted MUAC thresholds that varied with age (all P > 0.05) and performed better than WLZ <-3 for both inpatient and post-discharge mortality (both P < 0.001). Reported small size at birth did not reduce the risk of death associated with anthropometric indexes.
To treat acute malnutrition in U6M infants, WHO currently recommends re-establishing exclusive breastfeeding; this is not a malnutrition-specific intervention and most admitted sick infants would benefit. Successfully re-establishing exclusive breastfeeding in an inpatient setting is a labour intensive and time- and resource-consuming activity. Furthermore, for those who are too sick or unable to breastfeed for another reason, dilute F-100 or formula milk may be needed. The proposed cut-offs can be used to identify infants for whom potentially scarce resources can be prioritised. Further research into the effectiveness of potential interventions is required.
Footnotes
1Mwangome M, Ngari M, Fegan G, Mturi N, Shebe M, Bauni E, Berkley JA. Diagnostic criteria for severe acute malnutrition among infants aged under 6 mo. Am J Clin Nutr doi: 10.3945/ajcn.116.149815.
References
WHO (2013). Guideline: Updates on the management of severe acute malnutrition in infants and children [Internet]. WHO; 2013. [cited 2015 Apr 16]. Available from: http://apps.who.int/iris/bitstream/10665/ 95584/1/9789241506328_eng.pdf