Merge pull request #1880 from ihmeuw/mncnh_intrapartum_stillbirth_rates

Document IP/AP stillbirth split with data

Author: pletale

docs/source/models/concept_models/vivarium_mncnh_portfolio/concept_model.rst

@@ -341,7 +341,7 @@ Intrapartum component
 .. note::
 
   Only live births or stillbirths (NOT abortions/miscarriages/ectopic pregnancies) will proceed to the intrapartum component,
-  as described above.
+  as described above. Both antepartum and intrapartum stillbirths will proceed to the intrapartum component. However, antepartum stillbirths will only be eligible for intrapartum interventions that act on maternal health (such as misoprostol and azithromycin) and will not be eligible for intrapartum interventions intended for neonatal health (such as antenatal corticosteroids) as the fetus will have already passed prior to the onset of labor, but delivery of the fetal remains will still be necessary. Intrapartum stillbirths will remain eligible for all intrapartum interventions. 
 
 .. warning::
 

docs/source/models/intervention_models/intrapartum/acs_intervention.rst

@@ -100,7 +100,7 @@ Vivarium Modeling Strategy
 
 To be eligible to receive the ACS intervention (see the :ref:`intrapartum intervention module document <2024_vivarium_mncnh_portfolio_intrapartum_interventions_module>` 
 for how to obtain this information in the MNCNH portfolio simulation), a simulant must be expected to give birth to a early or moderate preterm infant with a believed GA
-of 26 to 33 weeks.
+of 26 to 33 weeks. Pregnancies that end in live births or intrapartum stillbirth will be eligible for ACS if believed GA is within this range, however antepartum stillbirths will not be eligible.
 
 This intervention requires adding an attribute to all simulants who expect to give birth to a preterm infant (i.e., based on believed GA if 26 to 33 weeks from 
 pregnancy module output)  to specify if a parent-child dyad receives ACS or not.  We will track this and the model will 
@@ -185,6 +185,8 @@ Assumptions and Limitations
   We do not consider anything related to ultrasound exposure for the 'accurate GA dating' criteria in the [WHO-2022]_ recommendations.
 - Despite the fact that our preterm cause model (based on the GBD cause) considers under 37 weeks of gestation, and despite the [WHO-2022]_ recommendations 
   that infants with 24-34 weeks of gestation receive ACS, we will only apply the ACS intervention to simulants with 26-33 weeks of gestation, based BMGF assumptions (see email from CT on 6/30/2025).
+- We assume that ACS would not in practice be given if antepartum stillbirth is detected prior to labor (i.e., the birth attendant could detect no fetal health rate via stethoscope or doppler upon presentation to a health facility (i.e., ANC or delivery facility)).
+  Given ACS is only provided at facilities with CPAP capability (i.e., CEmONC facilites), we assume that these facilities would have the capacity to detect antepartum stillbirth and therefore would not administer ACS in these cases.
 
 .. todo::
 

docs/source/models/intervention_models/mncnh_pregnancy/iv_iron_antenatal/iv_iron_mncnh.rst

@@ -147,6 +147,7 @@ Assumptions and limitations
     * We have a `JIRA ticket to address this limitation <https://jira.ihme.washington.edu/browse/SSCI-2377>`_ if we choose to do so.
 - We assume that there is no individual-level heterogeneity in the effect of IV iron on hemoglobin concentrations, despite having some data that could inform this. 
   We chose not to include stochastic uncertainty in order to simplify the data prep needed for this intervention model.
+- We assume that the effect of IV iron on probability of stillbirth (as mediated through hemoglobin concentration) applies equally to both types of stillbirth (intrapartum and antepartum).
 
 Validation and Verification Criteria
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

docs/source/models/intervention_models/mncnh_pregnancy/oral_iron_antenatal/oral_iron_antenatal.rst

@@ -533,8 +533,8 @@ Assumptions and limitations
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 - There is an observed association between severe anemia during pregnancy and stillbirth, as shown in [Young-et-al-2019]_. Therefore, there should hypothetically be some effect of IFA on stillbirth given that it improves hemoglobin concentration during pregnancy; however, we do not consider this effect due to lack of evidence that shows a direct effect of IFA on stillbirth outcomes.
-
 - We assume no effect modification by when in pregnancy oral iron is received. In reality, the effect on stillbirth is likely greater for those who have taken oral iron for longer.
+- We assume that the effect of MMS on probability of stillbirth applies equally to both types of stillbirth (intrapartum and antepartum).
 
 References
 ------------

docs/source/models/other_models/pregnancy/gbd_2021_mncnh/index.rst

@@ -237,21 +237,51 @@ Assign birth outcome (live births and stillbirths only)
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 For pregnancies that were determined to result in either a live birth or a stillbirth,
-we choose which of these occurs.
+we choose whether the birth outcome is a live birth, antepartum stillbirth or intrapartum stillbirth.
 
-Among such pregnancies, the base probability of a live birth is ASFR / (ASFR + ASFR * SBR),
+Among pregnancies that result in births, the base probability of a live birth is ASFR / (ASFR + ASFR * SBR),
 with the remainder being stillbirths.
 This probability is modified by the :ref:`antenatal supplementation intervention <maternal_supplementation_intervention>`.
 
+For pregnancies that result in stillbirths, we assign the timing of the stillbirth as either antepartum (fetal deaths occurring before the onset of labor) or intrapartum (fetal death occurring during labor and before birth).
+Both antepartum and intrapartum stillbirths will proceed to the intrapartum component. The timing of stillbirth will affect the eligibility of the simulant dyad for certain intrapartum interventions.
+Specifically, antepartum stillbirths will only be eligible for intrapartum interventions that act on maternal health (such as misoprostol and azithromycin) and will not be eligible for intrapartum interventions intended for neonatal health (such as antenatal corticosteroids) as the fetus will have already passed prior to the onset of labor, but delivery of the fetal remains will still be necessary.
+Intrapartum stillbirths will remain eligible for all intrapartum interventions. 
+
+The probability of intrapartum stillbirth will be affected by the intrapartum sensors -> C-section -> obstructed labor -> intrapartum stillbirth pathway once implemented, but baseline values for rates of antepartum stillbirth rates are outlined in the table below. 
+Antepartum stillbirths still require delivery of the deceased fetus, which may occur via dilation and evacuation without labor (for smaller fetuses), spontaneous or induced vaginal delivery, or c-section - we will need to consider this possibility of our c-section modeling strategy once developed. 
+Location-specific probability of intrapartum stillbirth is extracted from a 2018 Lancet Global Health report. [Ahmed_et_al_2018]_
+
+.. todo:: 
+
+  When we implement the intrapartum sensors -> C-section -> obstructed labor -> intrapartum stillbirth pathway, we'll need to delay determination of intrapartum stillbirth. 
+
 .. note::
 
-  When we implement the intrapartum sensors -> C-section -> obstructed labor -> intrapartum stillbirth pathway, we'll need to assign only
-  antepartum stillbirths here.
+  The antepartum stillbirth rates extracted from [Ahmed_et_al_2018]_ below are specific to Sub-Saharan Africa and South Asia, and therefore apply to our modeled locations.
 
 .. note::
 
   We assume that live births and stillbirths have the same gestational age. There is ongoing work at IHME to estimate gestational age at birth distributions among stillbirths. 
+  Additionally, we assume that intrapartum and antepartum stillbirths have the same gestational age distribution. [Madhi_et_al_2019]_ indicates this is likely a reasonable assumption (see Table 2).
+
+.. note::
 
+  We will have all antepartum stillbirths still proceed to the intrapartum component, even though it is possible that a stillbirth could be detected and addressed through D&E without labor. 
+
+.. list-table:: Data values for antepartum stillbirth rates
+  :header-rows: 1
+
+  * - Location
+    - Data value (proportion)
+    - Uncertainty distribution
+  * - Ethiopia & Nigeria
+    - 0.37 (95% CI 0.27-0.47)
+    - Truncated normal distribution truncated at 0-1.
+  * - Pakistan
+    - 0.48 (95% CI 0.39-0.57)
+    - Truncated normal distribution truncated at 0-1.
+    
 .. _other_models_pregnancy_closed_cohort_mncnh_sex_of_infant:
 
 Assign sex of infant (live births and stillbirths only)
@@ -342,3 +372,12 @@ The following should validate:
 References
 -----------
 
+.. [Ahmed_et_al_2018]
+
+  Ahmed, Imran et al. 2018. Population-based rates, timing, and causes of maternal deaths, stillbirths, and neonatal deaths in south Asia and sub-Saharan Africa: a multi-country prospective cohort study.
+  The Lancet Global Health, Volume 6, Issue 12, e1297 - e1308
+
+.. [Madhi_et_al_2019]
+
+  Madhi, Shabir A., et al. 2019. An Observational Pilot Study Evaluating the Utility of Minimally Invasive Tissue Sampling to Determine the Cause of Stillbirths in South African Women.
+  Clin Infect Dis. 2019 Oct 9;69(Suppl 4):S342–S350. doi: 10.1093/cid/ciz573