GBD 2019 Static Child Wasting Model and Protein Energy Malnutrition

Overview

This page contains information pertaining to the static joint risk-cause wasting model. An alternative dynamic transition model of wasting exposure (specific to GBD 2020) is described elsewhere. Given that wasting is generally considered an acute rather than chronic condition, it is more appropriate to model it as a dynamic transition model. However, a dynamic exposure model of child wasting is not modeled by GBD and is complex and data intensive. As a simplication to the dynamic transition wasting model used for the acute malnutrition simulation, this document will describe a propensity-based static child wasting exposure modeling strategy for 2019 similar to the modeling strategy for GBD 2020 child stunting exposure.

GBD stratifies wasting into four categories: TMREL, mild, moderate, and severe wasting. All PEM cases are attributed to moderate and severe wasting, making PEM a PAF-of-1 model. Under the GBD framework, wasting is additionally a risk for measles, diarrheal diseases, and lower respiratory infections. These relationships are detailed under the risk effects page for wasting.

Abbreviations

Abbreviation	Definition	Note
WHZ	Weight for height z-score
PEM	Protein energy malnutrition

Wasting background

See description on the dynamic transition wasting exposure model document.

Wasting Exposure in GBD 2019

Child wasting REI ID = 240

Wasting categories in GBD 2019

Category	Description	Range	Note
cat1	Severe child wasting	Less than -3 WHZ
cat2	Moderate child wasting	Between -2 and -3 WHZ
cat3	Mild child wasting	Between -1 and -2 WHZ
cat4	No child wasting	Greater than -1 WHZ	TMREL

Wasting Restrictions 2019

Restriction type	Value	Notes
Male only	False
Female only	False
Risk exposure age group start	Early Neonatal	age_group_id = 2. This is the earliest age group for which the wasting risk exposure estimates nonzero prevalence.
Risk effects age group start	Post neonatal	age_group_id = 4. This is the earliest age group for which there exist wasting RRs.
Age group end (risk exposure and effects)	1 to 4	age_group_id = 5

Protein Energy Malnutrition in GBD 2019

PEM is a PAF-of-1 cause with child wasting in GBD 2019 among the unrestricted ages for child wasting. There are fatal and non-fatal components.

PEM Restrictions 2019

Restriction type	Value	Notes
Male only	False
Female only	False
YLL only	False
YLD only	False
YLL age group start	Post neonatal	age_group_id = 4
YLL age group end	95 plus	age_group_id = 235
YLD age group start	Early Neonatal	age_group_id = 2
YLD age group end	95 Plus	age_group_id = 235

Vivarium Modeling Strategy

The wasting exposure model should be implemented as a propensity risk exposure model such that a simulant’s child wasting exposure state may change as they age into the next age group, but their child wasting percentile within the population will remain constant.

If a simulant is in wasting risk exposure cat1 or cat2, they should be considered “infected” with severe and moderate PEM (respectively) and accrue YLDs according to the disability weights as well as experience the associated excess mortality rates defined in the table below.

State data

State	Measure	Value	Note
cat1 (severe wasting, severe PEM)	disability weight	\(\frac{\text{dw_s199} * \text{prevalence_s199} + \text{dw_s2036} * \text{prevalence_s2036}}{\text{prevalence_s199} + \text{prevalence_s2036}}\)
cat1 (severe wasting, severe PEM)	excess mortality rate	\(\frac{\text{deaths_c387}}{\text{population} * \text{prevalence_c387}}\)	Assumed same excess mortality rate as cat2 moderate PEM
cat2 (moderate wasting, moderate PEM)	disability weight	\(\frac{\text{dw_s198} * \text{prevalence_s198} + \text{dw_s2033} * \text{prevalence_s2033}}{\text{prevalence_s198} + \text{prevalence_s2033}}\)
cat2 (moderate wasting, moderate PEM)	excess mortality rate	\(\frac{\text{deaths_c387}}{\text{population} * \text{prevalence_c387}}\)	Assumed same excess mortality rate as cat1 severe PEM
cat3 (mild wasting)	disability weight	0
cat3 (mild wasting)	excess mortality rate	0
cat4 (tmrel)	disability weight	0
cat4 (tmrel)	excess mortality rate	0

Data values

Parameter	Source	Note
Wasting risk exposure, rei_id=240	source=’exposure’, decomp_step=’step4’, status=’best’, gbd_round_id=6, year_id=2019, gbd_id_type=’rei’, gbd_id=240
deaths_c387	source=’codcorrect’, decomp_step=’step5’, status=’best’, gbd_round_id=6, year_id=2019, gbd_id_type=’cause’, gbd_id=387
prevalence_c387	source=’como’, decomp_step=’step5’, status=’best’, gbd_round_id=6, year_id=2019, gbd_id_type=’cause’, gbd_id=387, measure_id=5
prevalence_s{198,199,2033,2036}	source=’como’, decomp_step=’step5’, status=’best’, gbd_round_id=6, year_id=2019, gbd_id_type=’sequela’, gbd_id=[198,199,2033,2036], measure_id=5
dw_s{198,199,2033,2036}	Pull from GBD 2019	Listed in the note below for easy reference

Note

PEM sequelae disability weights from GBD

Disability weight	Value	Note
dw_s198	0.051 (0.031–0.079)	Moderate wasting with edema (from the GBD 2019 risk appendix table S13)
dw_s2033	0	Moderate wasting without edema (from the GBD 2019 risk appendix table S13)
dw_s199	0.128 (0.082–0.183)	Severe wasting without edema (from the GBD 2019 risk appendix table S13)
dw_s2036	0.172 (0.115-0.238)	Severe wasting with edema (from the GBD 2019 risk appendix table S13)

NOTE: It looks like these sequelae descriptions are mislabeled if logical numbering patterns were followed, but I have confirmed they are correct despite this suspicion.

Validation

• Wasting exposure state person time should validate to GBD wasting risk exposure

• PEM exposure state person time, CSMR, EMR, YLDs, and YLLs should validate to GBD prevalence

• Moderate PEM state person time should occur among those in wasting exposure cat2 only

• Severe PEM state person time should occur among those in wasting exposure cat1 only

Assumptions and limitations

We are modeling wasting as a chronic condition rather than the acute condition that it is. This may cause us to overestimate wasting burden among the population afflicted with correlated factors and vise versa. Additionally, we may underestimate wasting exposures in the older age groups as simulants with propensities for wasting exposure will die at a higher rate than those without. We also assume that moderate and severe PEM have equal excess mortality rates when it is likely higher for severe PEM.

References