Comparison of Weight-for-Height Based Indices for Assessing the Risk of Death

Summary of Published Paper

Mortality rates among children with severe malnutrition vary considerably between different treatment centres. This variation is due to differences in the severity of malnutrition amongst those enrolled at the treatment centre (risk of mortality is related to degree of malnutrition on admission) and other factors like dehydration, associated infection and differences in management of patients. Until recently little attempt has been made to account for the influence of severity of malnutrition in feeding centre admissions on different mortality rates at treatment centres. At the SFP meeting in Dublin (February 1997), reported on in the recent issue of Field Exchange, ACF described a newly developed model which would allow for a more meaningful comparison of mortality rates in different treatment centres by taking into account the nutritional status of children on admission.

This model was developed as part of a study on mortality rates in nine different therapeutic feeding centres in Rwanda, Sierra Leone and Madagascar. Of 1441 children admitted to these centres, 1047 were discharged following recovery and 147 died during treatment. Mortality was related to standard WHO nutritional indices of WHZ¹, WHP² and also to simple ratios of weight to powers of height , in particular to BMI (weight (kg) /ht (m)2), in order to examine how these relate to risk of death.

The study found that the optimal ratio of weight/height (ORWH) for assessing the risk of death for children in a TFP was wt/ht ^1.74 (95% CI of the Beta estimate 1.65-1.84). In children without oedema, ORWH and BMI were better predictors of death than WHP or WHZ. In spite of the fact that all centres had optimal management regimes and there were no epidemics recorded, there were still large variations in mortality rates between different centres. These variations were due to differences in severity of malnutrition on admission, and other factors like specific nutritional deficiencies, dehydration, hepatic dysfunction, associated infections and/or differences in management.

The development of an optimal height ratio may be good news for practitioners in the field as the ratio is easier to calculate than WHP or WHZ. Calculations do not require reference population data and the only tool that is required is a pocket calculator.

Practical Use of the Model

The ACF study was not designed to identify cut-off points for using ORWH as criteria for admission to nutrition centres so that the model cannot be used to replace currently used admission criteria like percentage of the median or presence of oedema. However, it can be used to calculate expected rate of death at treatment centres which can then be compared to actual rates in order to provide an indication of quality of management within the centre. Unexpectedly high death rates could then prompt an examination of management skills and protocols. Excel spread sheets are available from ACF which calculate expected death rates based on admission weight and height and the presence or absence of oedema , using the ORWH. These spread sheets can transform basic data entered by centre staff and provides graph plots of observed and expected mortality. This is a simple and effective way of measuring centre performance.

At this stage there are still a number of considerations to take into account when considering the ORWH model:

• children who defaulted in the study and whose data were not included in the model were significantly more malnourished on admission. This could influence the validity of the model in other situations so that the 1.74 ratio factor may not be appropriate in all circumstances.
• the model looks exclusively at death as a measurement of failure, but does not consider poor weight gains, or high default rates. These should also be considered when measuring the success of a programme
• Further validation of the 1.74 ORWH model is required.

A number of agencies at the SFP meeting expressed interest in applying the model in practice. The Excel model can be obtained from ACF.

¹WHZ, weight/height index expressed as z score which is the deviation from the median of the National Centre for Health Statistics' (NCHS) standard expressed in multiples of the standard deviation.

²Weight/height index expressed as percentage of the median of the NCHS

Reference; C. Prudhon et al Comparison of Weight- and Height-based indices for Assessing the Risk of Death in severely Malnourished Children, Am J Vol.144, No.2,1996
For further information contact Yvonne Grellety ACF, 9, Rue Dareau, 75014 Paris, France.
E-mail grellety@acf.imafinet.fr