Individually calibrate HIV+ multipliers on HPV clearance and lesion regression

[carajbro]

• Multiple multipliers (by CD4 count) that modify clearance/regression in HIV+ persons are set to be the same. Ie. multiple parameters are set equal to one set of values.
• These values were calibrated by Nick and Monisha, but based on this source: L. Ahdieh et al., Prevalence, incidence, and type-specific persistence of human papillomavirus in human immunodeficiency virus (HIV)-positive and HIV-negative women. J Infect Dis 184, 682-690 (2001).
• This may be reasonable if it correctly captures trends, but it seems unlikely to me that all these parameters are actually the same.

Values by decreasing CD4 count: [0.6, 0.55, 0.45, 0.3] (original parameter name used to set others: hpv_hivClear)

Parameters set to the same value:

• HIV+ multipliers modifying infection clearance from infected to immune in females and infected to normal in males (parameter name: rHivHpv_Clear)
• HIV+ multipliers modifying regression from CIN3 to CIN2 (parameter name: c2c3Mult)
• HIV+ multipliers modifying regression from CIN2 to CIN1 are set to this set of values squared (parameter names: c1c2Mult, rHivHpvMult)

[darcyrao]

Nicole Campos has a cervical cancer model fitted to South Africa (Campos et al JAIDS 2018) with a lot of good sources for parameters, including HPV clearance and CIN regression. We could look at their sources and estimates to compare them to ours. I've attached their supplemental appendix. QAI_2018_06_15_CAMPOS_1_SDC1.docx

[carajbro]

In (Campos et al., JAIDS, 2018) Appendix, look at sources:

• HPV clearance: 3, 16-19, 22, 23, 26-28
• No sources for lesion regression

I think our rates are an order of magnitude different:

• Campos gives baseline probabilities (~0.01-0.08, varying by age,type) * factor (~0.5-0.86)
• p = 1-exp(-rt)
• if one of our rates is 0.436, this equates to a probability of 0.3534 ?