A new US model of the cost-effectiveness of pre-exposure prophylaxis (PrEP) for men who have sex with men (MSM), prepared by researchers at Emory University in Atlanta, finds that taking even one year of PrEP is more cost-effective than measures like kidney dialysis, if it is used by people belonging to populations where HIV prevalence is at least 10%.
However, for PrEP to actually save money, relative to the lifetime cost of treating the HIV infections that would otherwise happen, either average adherence to PrEP needs to be high enough for it to be more than 92% effective, or the population prevalence of HIV has to be over 20%, or there has to be a reduction in the price of PrEP of at least 80%.
The model finds that the two most significant factors in whether PrEP saves money or not are population HIV prevalence, and adherence. (PrEP saves money when background HIV prevalence is high because it prevents more of the infections that would otherwise have occurred.)
As a result, the authors say, PrEP could be cost-effective with modest price reductions in states like Georgia but would need to be discounted substantially, or only taken by a targeted few, to be cost-effective in low-prevalence states like Montana.
The model’s ‘base case scenario’ – in which PrEP effectiveness is 44% (as in the iPrEx study), HIV prevalence in the PrEP-taking population is 10%, and the price of PrEP is the full current list price for brand-name Truvada (emtricitabine/tenofovir disoproxil fumarate – $10,711 a year) finds that the net cost of PrEP is $415,000 per case of HIV prevented, or a cost-effectiveness of $64,000 per quality-adjusted life-year (QALY) gained, compared to no PrEP.
It is important to note that this is achieved by each person in the model only taking PrEP for one year. Coincidentally, recent data from the Kaiser Permanent PrEP distribution programme in northern California found that the average time people stayed on PrEP in the programme was about a year. The model does not calculate what would happen if people took PrEP for longer.
“Cost-effective” medical interventions that nonetheless still cost money are allowed because it is accepted that medical interventions that extend life have a value over and above the merely financial, and it is unethical not to provide them if they are affordable. Some interventions are allowed even if they cost more than the highest US threshold usually quoted for cost-effectiveness, namely $100,000 per QALY gained; for instance kidney dialysis may cost as much as $150,000 per QALY gained.
In practice, however, moral and political factors as well as financial ones always influence cost-effectiveness decisions, and the authors acknowledge that a better benchmark in the case of PrEP might be the $19,530 per QALY gained cost-effectiveness of cervical cancer screening.
The three factors that would best bring PrEP cost-effectiveness in under this latter figure would be improved adherence and therefore effectiveness, or a drop in drug prices, or a rise in the background HIV prevalence of the population who took PrEP.
If PrEP effectiveness was 92%, as it was in the US sites of iPrEx, or if there was an 80% reduction in the price of generic emtricitabine/tenofovir disoproxil (about the average price drop seen when generic drugs come in), then PrEP would not merely be cost-effective, but cost-saving. In other words, because the prevented HIV infections would not have to be treated for a lifetime, its use would eventually give more money back to the US healthcare system than its initial cost.
If HIV prevalence in the population taking PrEP was 20% rather than 10%, then the reduction in the cost of PrEP would only have to be 40% to save money, even with 44% effectiveness.
Even if effectiveness is only 32%, then PrEP is still cost-saving with an 80% price reduction in populations with at least 10% prevalence.
The model also finds that behaviour change is a fourth factor with a surprisingly strong influence on the cost-effectiveness of PrEP. In the base-case scenario of no behavioural disinhibition, 10% prevalence and moderate (44%) effectiveness, then PrEP becomes cost-saving if an 80% drop in its price is achieved. But if there is a 25% increase in risk behaviour, then even an 80% price reduction does not achieve cost savings except in populations where prevalence is 20% or more.
If there is behavioural disinhibition, PrEP at 32% effectiveness does not save money at any price or at any prevalence – it actually loses more money the more it is implemented, because behavioural disinhibition at that level of effectiveness implies that introducing PrEP might actually cause a rise in infections.
Conversely, with 92% effectiveness – which was that seen in the US sites of the iPrEx Study – then PrEP would be cost-saving under nearly all scenarios, regardless of behavioural disinhibition, and even at today’s full list price for Truvada in populations with more than 20% prevalence. But there would have to be a drop in price of at least 40% in populations with 10-20% prevalence for cost-savings to be achieved, implying a price of approximately $7900 a year without behavioural disinhibition, or $7200 with it.
The model assumes that only 36% of all MSM with HIV (diagnosed are not) are on antiretroviral therapy (ART) and ART efficacy is 91%, implying that only 33% of MSM with HIV in a population are virally suppressed.
This model does not explore what happens if more men with HIV are virally suppressed. One would normally expect that as people with HIV are virally suppressed, HIV prevalence effectively falls as they drop out of the infectious population, and so PrEP becomes less cost-effective as there are fewer infections to prevent.
However, there are two other factors that the authors of the study also acknowledge they have not included in their model, and that could mean that PrEP preserves its cost-effectiveness even if more people with HIV are virally suppressed. One is that it does not take into account secondary infections – the infections that are prevented when people who take PrEP don’t pass on HIV to others. One cost-effectiveness study of PrEP in France estimated that for every person protected directly from HIV by PrEP, between 1.6 and 2.3 other people would be protected from secondary infections.
The other factor is that, as the authors say “We did not account for dense, associative and heterogeneous sexual networks that can result in pockets of very high transmission density.” In other words, it does not take into account PrEP’s ability to break chains of infection in closely-connected sexual networks before new infections happen, rather than after some have.
Taken together, these are the factors that may account for the fact that in some settings, PrEP has had a stronger effect in reducing HIV infections than simple models can account for, because it does not just reduce HIV infections in individuals, but HIV infections in the entire sexual network they belong to. As the authors of this model say, “the simplification of sexual transmission in our model may…lead to a more conservative estimate of the cost-effectiveness of PrEP.”
In addition, even a 90% reduction in the list price of Truvada might not be the maximum achievable. A 90% reduction on the list price used in this model implies a drug cost for PrEP of about $90 a month. But a 30-day supply of generic PrEP is available from the website Dynamix International to US buyers for as little as $48 and a scheme is about to be launched in Germany whereby selected pharmacies will be offering PrEP at €50 a month.
These factors, plus the fact that PrEP effectiveness has generally been observed in ‘real world’ studies to be higher than 44%, suggest that even in the US, the home of high drug costs, PrEP programmes could save money in the long term.