Neonatal sepsis and other neonatal infections
Disease Overview
Neonatal sepsis and other neonatal infections are infections during the neonatal period that advance to a systemic bloodstream infection (sepsis) and infections that occur during the neonatal period that are not already modelled separately in the GBD. For the MNCNH portfolio, we will include only mortality and not morbidity from this cause.
Todo
decide if for this model it would be appropriate to also include LRI and diarrheal diseases in this cause; decide based on evidence for antibiotics as a treatment for these conditions in neonates.
GBD 2021 Modeling Strategy
We focus on only the mortality due to this cause, which is estimated together with all other cause-specific mortality through the CODEm process.
The nonfatal part is described here: GBD 2021 Non-fatal Neonatal Infections Modeling Strategy (pages 32-39).
Cause Hierarchy
All causes (c_294) [level 0]
Communicable, maternal, neonatal, and nutritional diseases (c_295)
Maternal and neonatal disorders (c_962)
Neonatal disorders (c_380)
Neonatal sepsis and other neonatal infections (c_383) [level 4]
Restrictions
The following table describes restrictions on the effects of this cause (such as being only fatal), as well as restrictions on the age and sex of simulants to which different aspects of the cause model apply.
Restriction Type |
Value |
Notes |
|---|---|---|
Male only |
False |
|
Female only |
False |
|
YLL only |
True |
|
YLD only |
False |
|
YLL age group start |
Early neonatal |
GBD age group id 2 |
YLL age group end |
Post neonatal |
GBD age group id 4 |
Vivarium Modeling Strategy
Scope
The Level 4 neonatal conditions included in the MNCNH Portfolio model are closely linked to the Overall Neonatal Disorders Model. Since the LBWSG Risk Factor has an effect on overall neonatal mortality, this subcause will define a birth-weight- and gestational-age-specific cause specific mortality rate that the Overall Neonatal Disorders Model can use. This risk-specific CSMR might then be further individualized based on treatment coverage and efficacy, to ensure that individuals who have access to a treatment like antibiotics have a lower risk of mortality from neonatal sepsis than those who do not.
Assumptions and Limitations
Focusing solely on YLLs will likely underestimate the total burden of sepsis and, consequently, the DALYs averted by interventions that reduce sepsis risk. However, we anticipate this underestimation to be less than 10%.
We assume that the relationship between LBWSG and sepsis CSMR follow the relationship between LBWSG and all-cause mortality during the neonatal period, and could be refined with additional data.
Cause Model Decision Graph
We are not modeling Neonatal Sepsis dynamically as a finite state machine, but we can draw an directed graph to represent the collapsed decision tree representing this cause. Unlike a state machine representation, the values on the transition arrows represent decision probabilities rather than rates per unit time. Note that these probabilities are not used directly in the model and are included here only for clarity. If they end up being more confusing than enlightening, we should delete them.
State |
Definition |
|---|---|
live birth |
The parent simulant has given birth to a live child simulant (which is determined in the intrapartum step of the pregnancy model) |
neonate did not die of sepsis |
The child simulant did not die of sepsis within the first 28 days of life |
neonate died of sepsis |
The child simulant died of sepsis within the first 28 days of life |
Symbol |
Name |
Definition |
|---|---|---|
csmrisk |
sepsis mortality risk |
The probability that a simulant who was born alive dies from this cause during the neonatal period |
Modeling Strategy
The Neonatal Sepsis submodel requires only the birth-weight- and gestation-age-stratified cause specific mortality risks for sepsis during the early and late neonatal periods.
The way these CMRs are used is the same for all subcauses, and therefore is included in the Overall Neonatal Disorders Model page. This page describes the birth-weight- and gestational-age-specific cause specific mortality risks that are used for this cause on that page, \(\text{CSMRisk}^{\text{sepsis}}_{\text{BW},\text{GA}}\). The formula is:
where \(\text{CSMRisk}\) is the cause-specific mortality risk for sepsis, \(\text{RR}_{\text{BW},\text{GA}}\) is the relative risk of all-cause mortality for a birth weight of \(\text{BW}\) and gestational age of \(\text{GA}\), and \(Z\) is a normalizing constant selected so that \(\int_{\text{BW}} \int_{\text{GA}} \text{RR}_{\text{BW},\text{GA}} \cdot Z = 1\).
Note
the choice to use \(\text{RR}_{\text{BW},\text{GA}}\) in this equation is essentially arbitrary, and it could be replaced by any other nonnegative “weight function” \(w(\text{BW},\text{GA})\) as long it doesn’t lead to a negative “other causes” mortality risk. But with this choice, \(Z\) is equal to the \(1-\text{PAF}\) of LBWSG on all-cause mortality.
Each individual simulant \(i\) has their own \(\text{CSMRisk}_i\) that might be different from \(\text{CSMRisk}_{\text{BW}_i,\text{GA}_i}\) (meaning the average birth-weight- and gestational-age-specific CSMRisk for simulants with the birth weight and gestational age matching simulant \(i\). We recommend implementing this as a pipeline eventually because it will be modified by interventions (or access to interventions) relevant to this subcause. (Until we implement those, we will have \(\text{CSMRisk}_{i} = \text{CSMRisk}_{\text{BW}_i,\text{GA}_i}\), though.)
The following table shows the data needed for these calculations.
Data Tables
Note
All quantities pulled from GBD in the following table are for a specific year, sex, age group, and location.
Variable |
Definition |
Value or source |
Note |
|---|---|---|---|
enn_all_cause_death_count |
Count of deaths due to all causes in the early neonatal age group |
GBD: source=’codcorrect’, metric_id=1, cause_id=294 |
|
enn_death_count |
Count of deaths due to cause neonatal sepsis in the early neonatal age group |
GBD: source=’codcorrect’, metric_id=1, cause_id=383 |
|
lnn_death_count |
Count of deaths due to cause neonatal sepsis in the late neonatal age group |
GBD: source=’codcorrect’, metric_id=1, cause_id=383 |
|
live_birth_count |
Count of live births |
GBD: covariate_id = 1106 |
|
csmrisk_enn |
neonatal sepsis mortality risk in the early neonatal age group |
enn_death_count / live_birth_count |
|
csmrisk_lnn |
neonatal sepsis mortality risk in the late neonatal age group |
lnn_death_count / (live_birth_count - enn_all_cause_death_count) |
|
\(\text{CSMRisk}\) |
neonatal sepsis mortality risk |
either csmrisk_enn or csmrisk_lnn depending on the simulant’s age group |
|
\(\text{RR}_{\text{BW},\text{GA}}\) |
Relative Risk of all-cause mortality for a birth weight of BW and gestational age of GA |
interpolated from GBD data |
See Low Birth Weight and Short Gestation (LBWSG) page for details. |
\(Z\) |
Normalizing constant |
calculated from \(\text{RR}_{\text{BW},\text{GA}}\) and LBWSG exposure distribution. |
see above for details. |
Calculating Burden
Years of life lost
The years of life lost (YLLs) due to Neonatal Sepsis are calculated assuming age \(a=14 \text{ days}\), and equals \(\operatorname{TMRLE}(a) - a\), where \(\operatorname{TMRLE}(a)\) is the theoretical minimum risk life expectancy for a person of age \(a\).
Years lived with disability
For simplicity, we will not include YLDs in this model.
Validation Criteria
Neonatal Sepsis mortality risk in simulation should match GBD estimates.
Relative Risk of Neonatal Sepsis death due to LBWSG should match overall neonatal mortality RR.
References
GBD 2021 Non-fatal Neonatal Infections Modeling Strategy (pages 32-39).
GBD 2021 Neonatal Sepsis and Other Neonatal Infections Factsheet.