respond to feedback from team

docs/source/models/causes/neonatal/preterm_birth.rst

@@ -134,6 +134,8 @@ Note that these probabilities are not used directly in the model and are include
     * - neonate died without RDS
       - The child simulant died due to preterm birth **without** respiratory distress syndrome within the first 28 days of life
 
+.. _2021_cause_preterm_birth_mncnh_transition_probability_definitions:
+
 .. list-table:: Transition Probability Definitions
     :widths: 1 5 20
     :header-rows: 1
@@ -142,10 +144,10 @@ Note that these probabilities are not used directly in the model and are include
       - Name
       - Definition
     * - mr_w
-      - preterm mortality risk
+      - Preterm with RDS Mortality Risk
       - The probability that a simulant who was born alive dies from preterm with RDS during the neonatal period
     * - mr_wo
-      - preterm mortality risk
+      - Preterm without RDS Mortality Risk
       - The probability that a simulant who was born alive dies from preterm without RDS during the neonatal period
 
 

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

@@ -489,11 +489,9 @@ V&V Checks:
 4. Low Birthweight/Short Gestation Risk Effect on Neonatal Moratlity Model
 5. Hemoglobin Risk Exposure
 6. Hemoglobin Risk Effect on Maternal Hemorrhage
+7. :ref:`Intervention Models <neonatal_intervention_models>`
 
-.. toctree::
-    :maxdepth: 1
-
-    mncnh_interventions
+  a. :ref:`CPAP for treating Preterm with RDS <cpap_intervention>`
 
 .. _mncnh_portfolio_4.0:
 

docs/source/models/intervention_models/neonatal/cpap_intervention.rst

@@ -0,0 +1,135 @@
+.. _cpap_intervention:
+
+==================================
+CPAP for treating Preterm with RDS
+==================================
+
+.. contents::
+   :local:
+   :depth: 1
+
+.. list-table:: Abbreviations
+  :widths: 15 15 15
+  :header-rows: 1
+
+  * - Abbreviation
+    - Definition
+    - Note
+  * - CPAP
+    - Continuous positive airway pressure
+    - 
+  * - RDS
+    - respiratory distress syndrome
+    - 
+
+Intervention Overview
+-----------------------
+
+CPAP is used for the prevention and treatment of respiratory distress syndrome. A recent (2020) Cochrane Review found that CPAP was associated with lower risk of treatment failure (death or use of assisted ventilation) and lower overall mortality with moderate‐certainty evidence.
+
+This section describes how a CPAP intervention can be implemented and calibrated for the :ref:`MNCNH Portfolio model <2024_concept_model_vivarium_mncnh_portfolio>`.
+
+.. list-table:: Affected Outcomes
+  :widths: 15 15 15 15
+  :header-rows: 1
+
+  * - Outcome
+    - Effect
+    - Modeled?
+    - Note (ex: is this relationship direct or mediated?)
+  * - Preterm with RDS Mortality Probability
+    - Decrease multiplicatively
+    - Yes
+    - 
+
+Baseline Coverage Data
+++++++++++++++++++++++++
+
+39.3% of CMONC facilities and 7.5% of BMONC facilities have CPAP, according to the 2016 Ethiopia EmONC Final Report.  Please use these as a placeholder for now while we try to find reliable values for Nigeria and Pakistan. 
+
+The :ref:`MNCNH Portfolio model <2024_concept_model_vivarium_mncnh_portfolio>` includes data on current coverage levels for the specific locations we are focused on modeling.  We might be able to borrow strength from other locations and times by predicting coverage more country-years simultaneously, perhaps even in combination with other key intervention technologies, based on sources such as existing `Service Provision Assessment (SPA) <https://www.dhsprogram.com/methodology/Survey-Types/SPA.cfm>` and `Service Availability and Readiness Assessment (SARA) <https://www.who.int/data/data-collection-tools/service-availability-and-readiness-assessment-(sara)>` data.
+
+Vivarium Modeling Strategy
+--------------------------
+
+This intervention requires adding an attribute to all simulants to specify if a birth happens in a facility with access to CPAP.  Since the neonatal mortality model does not explicitly represent incidence of RDS, we will also not track who receives CPAP.  Instead the model will have different cause-specific mortality rates for RDS for individuals born with and without access to CPAP.
+
+We will use the decision tree below to find the probability of RDS mortality with and without access to CPAP that are logically consistent with the baseline delivery facility rates and baseline CPAP coverage (and population level RDS mortality probability). Since the neonatal model includes different mortality probabilities for both the early and late neonatal periods, this calculation must be performed for both time periods as well.
+
+We will then add an attribute to each simulant indicating whether the birth occurs at a facility with access to CPAP (which will be dependent on the facility choice, i.e. no access for home births and lower access for BEmONC than CEmONC facilities).
+
+We will then use the conditional probabilities for simulants with and without access to determine which simulants die from RDS.
+
+A `2020 Cochrane review <https://pmc.ncbi.nlm.nih.gov/articles/PMC8094155/>`_ found a relative risk of 0.53 (95% CI 0.34-0.83) of RDS mortality for neonates receiving CPAP.   (Note that the population that this effect size applies to is preterm infants with "respiratory failure becoming evident soon after birth".)
+
+
+Calibration Strategy
+--------------------
+
+The following decision tree shows all of the paths from delivery facility choice to CPAP availability.  Distinct paths in the tree correspond to disjoint events, which we can sum over to find the population probability of RDS mortality.  The goal here is to find internally consistent probabilities of RDS mortality for the subpopulations with and without access to CPAP, so that the baseline scenario can track who has access to CPAP and still match the baseline RDS mortality rate.
+
+.. graphviz::
+
+    digraph CPAP {
+        rankdir = LR;
+        facility [label="Facility type"]
+        home [label="p_RDS_without_CPAP"]
+        BEmONC [label="CPAP"]
+        CEmONC [label="CPAP"]
+        BEmONC_wo [label="p_RDS_without_CPAP"] 
+        BEmONC_w [label="p_RDS_with_CPAP"]
+        CEmONC_wo [label="p_RDS_without_CPAP"] 
+        CEmONC_w [label="p_RDS_with_CPAP"]
+
+        facility -> home  [label = "home birth"]
+        facility -> BEmONC  [label = "BEmONC"]
+        facility -> CEmONC  [label = "CEmONC"]
+
+        BEmONC -> BEmONC_w  [label = "available"]
+        BEmONC -> BEmONC_wo  [label = "unavailable"]
+
+        CEmONC -> CEmONC_w  [label = "available"]
+        CEmONC -> CEmONC_wo  [label = "unavailable"]
+    }
+
+.. math::
+    \begin{align*}
+        p_\text{RDS} 
+        &= \sum_{\text{paths without CPAP}} p(\text{path})\cdot p_\text{RDS_wo}\\
+        &+ \sum_{\text{paths with CPAP}} p(\text{path})\cdot p_\text{RDS_w}\\[.1in]
+        p_\text{RDS_w} &= \text{RR}_\text{CPAP} \cdot p_\text{RDS_wo}
+    \end{align*}
+
+where :math:`p_\text{RDS}` is the probability of dying from Preterm with RDS in the general population, and :math:`p_\text{RDS_w}` and :math:`p_\text{RDS_wo}` are the probability of dying from Preterm with RDS in setting with and without access to CPAP.  For each path through the decision tree, :math:`p(\text{path})` is the probability of that path; for example the path that includes the edges labeled BEmONC and unavailable occurs with probability that the birth is in a BEmONC facility times the probability that the facility has CPAP available (7.5% in Ethiopia in 2016)
+
+
+When we fill in the location-specific values for delivery facility rates, CPAP coverage, relative risk of mortality with CPAP access, and mortality probability (which is also age-specific), this becomes a system of two linear equations have two unknown (p_RDS_w and p_RDS_wo), which we can solve analytically.
+
+Scenarios
+---------
+
+.. todo::
+
+  Describe our general approach to scenarios, for example set coverage to different levels in different types of health facilities; then the specific values for specific scenarios will be specified in the :ref:`MNCNH Portfolio model <2024_concept_model_vivarium_mncnh_portfolio>`.
+
+  This is because specific scenarios might combine interventions, such as scaling up both an intervention like Antenatal corticosteroids (ACS) that lowers the prevalence of RDS together with increased coverage of CPAP.
+
+
+Assumptions and Limitations
+---------------------------
+
+- We assume that CPAP availability captures actual use, and not simply the machine being in the facility 
+- We assume that the delivery facility is the final facility in the care continum for deliveries that are transferred due to complications
+- We assume that the relative risk of RDS mortality with CPAP in practice is similar to that found in the Cochrane Review meta-analysis
+
+Validation and Verification Criteria
+------------------------------------
+
+- Population-level mortality rate should be the same as when this intervention is not included in the model
+- The ratio of RDS deaths per birth among those without CPAP access divided by those with CPAP access should equal the relative risk from the Cochrane Review
+- The baseline coverage of CPAP in each facility type should match the values in the artifact
+
+References
+------------
+
+* https://pmc.ncbi.nlm.nih.gov/articles/PMC8094155/

docs/source/models/intervention_models/neonatal/index.rst

@@ -0,0 +1,13 @@
+.. _neonatal_intervention_models:
+
+======================
+Neonatal Interventions
+======================
+
+.. toctree::
+    :maxdepth: 1
+
+    cpap_intervention
+
+
+