HIV Transmission Code including Resistance Transmission

[mmcleod89]

Currently there is some (simplified & heavily dummied) transmission code

• viral load groups for partners are just randomised currently; they should be calculated from the population statistically

• We need to include considerations for:

• [ ] Primary infection (first three months)

• [ ] Undetectable

• [ ] STIs

• Updating tests

• Add resistance tranmission: the model here should be confirmed with the HIV group based on our probability discussions earlier.

Useful info:

• risk_nippnp means "risk of new infected partner per new partner" i.e. the risk that any given new partner is infected.

[mmcleod89]

STI Code:

• Rate STI = 0.05 (this should probably go in a data file)

• Rate persistance of STI = 0.2

• Initial STI chances at the beginning of simultion = rate_sti * number of STPs (short term partners)

• This needs to initialise AFTER sex behaviour has initialised

• Place this in the sex behaviour init after new partners have been initialised

• Update for someone without an STI: BEFORE num partners update (i.e. last timestep's partners: rate_sti * num STPs

• Update for someone with an STI: probability of still having STI = rate_persist_STI * num STPs

Undetectable Viral Load:

• We have a rate of transmission for undetectable HIV which is sampled at the beginning of the simulation: %sample(tr_rate_undetec_vl, 0.0000 0.0001 0.0010, 0.7 0.2 0.1);

Probability of Viral Load Group for STP

• First check the proportion of people in that sex & age group who are in each viral load category
• If there are no infected people in the age and sex group, calculate the proportion of people in that sex (ignoring age) in each viral load group
• Map(sex, age, vl) -> Probability
• Map(sex, age) -> [Probability List]
• Update this map for the population each timestep at the beginning of HIV transmission code
• For individual transmissions, just look up the values

Transmission rate

• fold_tr: transmission factor, sampled at the beginning of the simulation %sample_uniform(fold_tr, 1/1.5 1 1.5);
• Additionally, new parter transmission factor: fold_tr_newp; %sample_uniform(fold_tr_newp, 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1/0.8 1/0.6 1/0.4);
• We have a set of 6 base tranmission rates, one for each viral load group (note that viral load groups are in increasing order until the last group which is for primary infection):

1. tr_rate_undetec_vl*fold_tr_newp
2. 0.01fold_trfold_tr_newp
3. 0.03fold_trfold_tr_newp
4. 0.06fold_trfold_tr_newp
5. 0.10fold_trfold_tr_newp
6. tr_rate_primary*fold_tr_newp
   • these values can be calculated at the setup of the HIV transmission module
   • If tranmission occurs we should note the viral load group of the person who infected them

• risk_nip is the probability of transmission from infected short term partners
• Initially set by VL group, but modified by e.g. STI, mutations/prep etc.
• So mutations need to be calculated before tranmission event

[mmcleod89]

Starting from * probability of infection from infected new partner; in SAS code:

• Calculate viral load group probabilities (in setup)
• Transmission risk based on VL group (base transmission risk + random factor which is indivually sampled)
• Presence of mutations (@andrew-phillips-1 should we mimic the SAS code or use sampling methods we discussed in previous meeting?)
• risk changes for STI / mutation / prep etc. (ignore pop_wide_tld section) Adding a small prep module is helpful here to hold / sample relevant values
• HIV tranmission: record some things for useful stats

[mmcleod89]

SAS code assumed all STPs for a given person have the same age group. For now, HIVpy allows each of these STPs to have their own age group. This will have an impact on their VL group probabilities etc.

[andrew-phillips-1]

@mmcleod89 Hi Michael I would like you to introduce any enhancements that you suggest and which we agree on. That means we might expect model results to differ from our SAS version, but I think that is a price worth paying for enhancing the model.