Update neonatal decision tree table with RDS data inputs #1584
Conversation
docs/source/models/concept_models/vivarium_mncnh_portfolio/concept_model.rst
Outdated
Show resolved
Hide resolved
docs/source/models/concept_models/vivarium_mncnh_portfolio/concept_model.rst
Outdated
Show resolved
Hide resolved
docs/source/models/concept_models/vivarium_mncnh_portfolio/concept_model.rst
Outdated
Show resolved
Hide resolved
| * - 8 | ||
| - XX relative risk on RDS mortality of facility having CPAP or NICU capabilities vs. not | ||
| - Need to confirm this will impact mortality not incidence. Also need to determine how neonatal mortality in general will be modeled and how we will handle overlaps with preterm and LBWSG RR's on all cause mortality | ||
| - 0.64 (95% CI 0.50-0.82) relative risk on RDS mortality of facility having CPAP capabilities |
There was a problem hiding this comment.
Are you sure 0.64 is the right number? Checking the source, it looks like this was the effect size for reducing "death or use of additional ventilatory support," whereas the RR for "mortality" was 0.53 (95% CI 0.34 to 0.83). Does that look right?
Also, is this the relative risk of "the facility having CPAP capabilities," or is it the relative risk of the neonate actually getting the CPAP intervention? My reading of the Cochrane review was that it's the latter (actually getting the intervention) -- does that seem right? We could potentially model these as different things in our simulation (i.e. a neonate born in a facility with CPAP, but they didn't actually get it), so we want to make sure we know what the number is measuring.
Also, it would be good to know what population this effect size applies to. E.g., was CPAP given to all preterm babies, or just those with RDS? According to the methods section, it says the type of participants were "preterm infants (less than 37 weeks of gestation) with respiratory failure becoming evident soon after birth," so I think it's the latter. We should probably note that somewhere, maybe in the notes column.
There was a problem hiding this comment.
Are you sure 0.64 is the right number? Checking the source, it looks like this was the effect size for reducing "death or use of additional ventilatory support," whereas the RR for "mortality" was 0.53 (95% CI 0.34 to 0.83). Does that look right?
I see also that "Three out of five of these trials were conducted in the 1970s. Therefore, the applicability of these results to current practice is unclear. Further studies in resource‐poor settings should be considered and research to determine the most appropriate pressure level needs to be considered," which make me think all of this is less solid than I would like.
| in general will be modeled and how we will handle overlaps with preterm and LBWSG RR's on all cause mortality. | ||
| Source: `2017 Cochrane review <https://pubmed.ncbi.nlm.nih.gov/28321847/>`_ | ||
| * - 10 | ||
| - Percentage of preterm deaths caused by RDS: Ethiopia - 45.3%; Nigeria - 52.4%; Pakistan - 35.6% |
There was a problem hiding this comment.
I think we have to be careful with these numbers because of how GBD models things. Specifically, I think the percentages listed here tell us
Among all babies who were born preterm and died, what fraction died of RDS?
But what we need for the model is
Among all babies who were born preterm and died of preterm birth complications as defined by GBD, what fraction died of RDS?"
For example, in Table 4 in the Ethiopia paper, I divided the causes of death into two groups:
-
Causes explicitly modeled by GBD (% of deaths):
- sepsis (26.1%)
- pneumonia (3.0%, presumed to be captured in the cause "Lower respiratory infections")
- meningitis (0.8%)
- congenital anomalies (3.4%)
- asphyxia (13.6%, presumed to be captured in the cause "Neonatal encephalopathy due to birth asphyxia and trauma)
- Intraventricular haemorrhage (1.1%, presumed to be captured in GBD's intracerebral and subarachnoid hemorrhage causes, but we should double check that)
-
Causes not explicitly modeled by GBD (% of deaths):
- respiratory distress syndrome (45.3%)
- apnoea (1.4%)
- necrotising enterocolitis (0.8%)
- others (4.5%)
Assuming the causes in the second category are included in GBD's "Neonatal preterm birth" cause (which we should double-check) and the ones in the first category aren't (which we should also double-check), the fraction of RDS deaths among GBD's preterm birth deaths would be 45.3 / (45.3 + 1.4 + 0.8 + 4.5) = 87.1%. (I think previously I had included intraventricular hemorrhage in the preterm birth cause, which would give 85.3%.)
@aflaxman and @pletale does that seem right?
If so, I think we'll have to do something similar for the other sources.
There was a problem hiding this comment.
Also, would it be better to put these numbers on the preterm birth cause page instead, so that we don't have to duplicate the data?
There was a problem hiding this comment.
Also, would it be better to put these numbers on the preterm birth cause page instead, so that we don't have to duplicate the data?
I think both the numbers and the reasoning behind them (i.e. this calculation that Nathiel wrote out, once Alix has double checked it) should be on the preterm page. I think there is a place in the table for the number already.
There was a problem hiding this comment.
I'm struggling to find confirmation of what is included in GBD's neonatal preterm birth complications cause. The neonatal disorders methods appendix doesn't go into those details. Do either of you know where we might be able to confirm what is included? For now, I'll add a note that we need to confirm this.
| * - 8 | ||
| - XX relative risk on RDS mortality of facility having CPAP or NICU capabilities vs. not | ||
| - Need to confirm this will impact mortality not incidence. Also need to determine how neonatal mortality in general will be modeled and how we will handle overlaps with preterm and LBWSG RR's on all cause mortality | ||
| - 0.64 (95% CI 0.50-0.82) relative risk on RDS mortality of facility having CPAP capabilities |
There was a problem hiding this comment.
Are you sure 0.64 is the right number? Checking the source, it looks like this was the effect size for reducing "death or use of additional ventilatory support," whereas the RR for "mortality" was 0.53 (95% CI 0.34 to 0.83). Does that look right?
I see also that "Three out of five of these trials were conducted in the 1970s. Therefore, the applicability of these results to current practice is unclear. Further studies in resource‐poor settings should be considered and research to determine the most appropriate pressure level needs to be considered," which make me think all of this is less solid than I would like.
| in general will be modeled and how we will handle overlaps with preterm and LBWSG RR's on all cause mortality. | ||
| Source: `2017 Cochrane review <https://pubmed.ncbi.nlm.nih.gov/28321847/>`_ | ||
| * - 10 | ||
| - Percentage of preterm deaths caused by RDS: Ethiopia - 45.3%; Nigeria - 52.4%; Pakistan - 35.6% |
There was a problem hiding this comment.
Also, would it be better to put these numbers on the preterm birth cause page instead, so that we don't have to duplicate the data?
I think both the numbers and the reasoning behind them (i.e. this calculation that Nathiel wrote out, once Alix has double checked it) should be on the preterm page. I think there is a place in the table for the number already.
…m/ihmeuw/vivarium_research into mncnh_neonatal_decision_tree_data
There was a problem hiding this comment.
Looks good. After reviewing Abie's CPAP PR (and preterm birth PR), it seems like a lot of these numbers actually belong on the appropriate intervention or cause page, even though we had place holders for them on the concept model page. We should think about how to get these numbers onto the pages where they're needed while still having a single source of truth. But for now, here is good.
I am still looking for sources for CPAP capabilities by facility type for Nigeria and Pakistan, so using placeholder values from EmONC 2016 Final Report (Ethiopia) for now. The data here doesn't include NICU capabilities or relative risk on RDS, only CPAP, but I think we could find NICU data if it's needed!