Acute malnutrition treatment docs, with inpatient SAM tx#1931
Conversation
alibow
requested review from
SylLutze,
aflaxman,
patricktnast,
rmudambi,
tylerdy and
zmbc
| - Did not exist in prior models, but has exactly the same structure as the MAM treatment effects in the row above | ||
| * - tx_rem_rate_sam | ||
| - Uncomplicated SAM treatment | ||
| - 0 (= 0 * tx_rem_rate_sam / e_sam * tx_rem_rate_sam) |
There was a problem hiding this comment.
The (= is weird, what is this meant to be?
There was a problem hiding this comment.
It's meant to be a proof for how we got zero from a larger equation that simplifies to zero. I agree it is sort of awkward here though. Would you prefer it with that deleted?
…ng_treatment_with_inpatient_sam/index.rst Co-authored-by: SylLutze <108029462+SylLutze@users.noreply.github.com>
…uw/vivarium_research into inpatient_sam_interventions
|
|
||
| `Tekeste et al. 2012 <https://link-springer-com.offcampus.lib.washington.edu/article/10.1186/1478-7547-10-4>`__ found that an program with higher coverage of uncomplicated SAM treatment (outpatient/community treatment) had ~3 times lower complication rates than a program with lower coverage of uncomplicated SAM treatemnt (facility-based treatment). | ||
|
|
||
| * To evaluate how our model validates to this finding, we can compare the ratio of complicated SAM prevalence to total SAM prevalence in a scenario without treatment for uncomplicated SAM to a scenario with treatment for uncomplicated SAM. We would expect the difference to be >3 times as our modeled coverage differences are larger than what is expected in this publication. |
There was a problem hiding this comment.
I'm not following how uncomplicated SAM treatment would change this ratio?
There was a problem hiding this comment.
I've been considering just removing this validation target because it is informed from really indirect data anyway -- but basically by increasing the rate by which simulants transition out of the uncomplicated SAM state to mild child wasting, which will decrease the number of simulants who degrade to complicated SAM.
There was a problem hiding this comment.
But uncomplicated -> mild child wasting would take the simulant out of both the numerator and denominator of this ratio, no?
There was a problem hiding this comment.
Yeah, but only after they've contributed person time --
In a scenario with uncomplicated treatment: someone transitions into uncomplicated SAM and contributes person time to the uncomplicated state, then they transition to mild child wasting and stop contributing person time to SAM states
In a scenario without uncomplicated treatment: someone transitions into uncomplicated SAM and contributes person time to the uncomplicated state, then they transition to complicated SAM and also contribute person time to the complicated SAM state.
The measure I want to compare between scenarios is (person time in complicated SAM) / (person time in complicated SAM + person time in uncomplicated SAM)
There was a problem hiding this comment.
But there could also be a simulant who, in a scenario without uncomplicated treatment, stays for 10 timesteps in the uncomplicated SAM state and then remits without treatment, while that same simulant gets treated back to mild after 1 timestep when they do have uncomplicated treatment.
Basically, it's not obvious to me that this ratio would go in the direction you expect it to. Also, I'm not sure whether it matches what was recorded by the study.
There was a problem hiding this comment.
Yeah, I see your point, thanks. Maybe what I want is a ratio of incident counts and not person time, but I still don't think that's quite right.
Basically I have a hypothesis that it is possible to study the impact of an uncomplicated treatment program on the frequency of complicated admissions and I'd like to find a study like that and use it as a validation target. This reference is not quite right and neither is my specified sim quantity -- I think the ideal measure to compare is the impact of a an uncomplicated SAM treatment program on population-level complicated SAM incidence. I'll update with a note to see if we can easily find something to use here.
There was a problem hiding this comment.
I think the ideal measure to compare is the impact of a an uncomplicated SAM treatment program on population-level complicated SAM incidence
This sounds right to me, and has the advantage of being a lot simpler to think about 😃
…ng_treatment_with_inpatient_sam/index.rst Co-authored-by: Zeb Burke-Conte <zmbc@users.noreply.github.com>
…ng_treatment_with_inpatient_sam/index.rst Co-authored-by: Zeb Burke-Conte <zmbc@users.noreply.github.com>
|
|
||
| .. math:: | ||
|
|
||
| \overline{RR_\text{r}} = C * RR_\text{r, treated} + (1 - C) * RR_\text{r, untreated} |
…ng_treatment_with_inpatient_sam/index.rst Co-authored-by: Abraham Flaxman <abie@alum.mit.edu>
…ng_treatment_with_inpatient_sam/index.rst Co-authored-by: Abraham Flaxman <abie@alum.mit.edu>
4 participants
No description provided.