Comparing full cascade to IUD-only intervention

[robynstuart]

Comparing full cascade to IUD-only intervention

Problem

Component analysis allows us to run individual treatments (NSAID, TXA, Pill, hIUD) standalone and compare their impacts. However, comparing the full cascade to an IUD-only scenario may not be appropriate if we use the same uptake parameters.

In the full cascade:

• Women progress through multiple treatment options (NSAID → TXA → Pill → hIUD)
• Only a subset reaches hIUD after trying earlier treatments
• hIUD uptake depends on cascade position and prior treatment failures

In an IUD-only scenario:

• hIUD is offered as first-line treatment
• No prerequisite treatments required
• Potentially different care-seeking patterns
• Different acceptance rates when offered as first vs last option

Questions

1. Uptake rates: Should IUD-only intervention use the same acceptance probability as when hIUD is offered in the cascade, or should we model different acceptance for first-line vs last-resort treatment?

2. Care-seeking: Do women seek care differently when hIUD is the only option vs when multiple treatment options exist?

3. Fertility intent filtering: In the cascade, women with fertility intent never reach hIUD (blocked by Pill requirement). In IUD-only, should we apply the same fertility intent filter, or would different counseling/messaging affect this?

4. Population coverage: The cascade naturally selects for women who failed earlier treatments. IUD-only would reach a different population. How do we account for this in impact comparisons?

Discussion

Need to determine whether component analysis should adjust uptake parameters for realistic scenario comparisons or maintain identical parameters to isolate treatment efficacy effects.

[nnoori-IDM]

Thanks Robyn. Here are some thoughts, and I’d love to hear you and Michelle’s thoughts as well:

In terms of uptake and care-seeking, in the previous plots we varied the % offered and the % accepted for each intervention. We could take a similar approach to compare the cascade under the H-IUD–only scenarios versus the package scenarios. Followings are the assumptions we had for the previous plots:

• H-IUD-only scenarios: % offered = 50% for all three; % accepted = 2%, 5%, or 8%.
• Package scenarios: % offered (applied across all interventions) = 25% or 50%.
  % accepted for H-IUD = 5%
  % accepted for the other interventions = 25% or 50%

For IUD-only intervention, I think we should apply the same fertility_intent filter.

To compare the IUD-only intervention and the package cascade, we could define outcomes per woman reached, such as number of HMB cases per woman reached or number of anemia cases or number of school disruptions averted per woman reached.