A Canadian paper that weighs the cost of using pre-exposure prophylaxis (PrEP) to prevent HIV infections against the lifetime total cost of one HIV infection finds that, using figures available for Quebec and Ontario provinces, PrEP would be cost-saving under most scenarios, even if the overall lifetime cost of HIV care falls in the future.
The paper, published before CROI 2015, in fact almost certainly underestimates the potential cost benefits of PrEP. The researchers at the time of writing had access to estimates of the cost of PrEP and of HIV care but did not have access to the actual effectiveness of PrEP as seen in Ipergay, the study they use for their modelling. They therefore base their estimation of the cost of one HIV infection averted on the efficacy seen in the previous iPrEx study – 44%. In fact effectiveness seen in Ipergay was 86% – meaning that the cost per infection averted would be considerably less.
In addition, the researchers use a high estimate of the overall cost of the drugs used in taking an Ipergay-style regimen. In the event, the average cost per person was about half of this.
To calculate the cost of PrEP, the researchers added the cost of a year’s worth of emtricitabine/tenofovir (Truvada) to indirect costs such as work time missed for clinic visits, clinic staff wages, free condoms and the cost of tests. The cost of Truvada was CA$9505 per year and other costs were CA$2497, or CA$12,001 in total.
The cost of a lifetime’s HIV infection was based on diagnosis at age 30 (the average in Quebec) and on the current additional life expectancy (excluding people who inject drugs) of someone diagnosed at that age – 35.2 years. The researchers assume the person with HIV starts antiretroviral therapy (ART) immediately. They cost for two ART regimens, a cheap one (efavirenz/emtricitabine/tenofovir [Atripla]) and an expensive one (boosted darunavir (Prezista) plus (Truvada)). They also assume a 10% a year failure rate for these first two regimens and a second-line regimen of raltegravir or etravirine plus supporting drugs.
The annual drug costs are CA$14,093 or CA$22,040 under the least and most expensive scenario. Hospital costs were actually somewhat less than the annual cost of PrEP care: CA$2016 a year. This is because trial participants in Ipergay were seen every two months whereas the HIV clinic visits are assumed to be quarterly. This leads to a cost per patient, per year, of CA$16,109 for the cheapest drug regimen.
However, the modellers also added in something that is not always added to models of the cost of HIV infection: its cost to the economy in terms of unemployment and loss of productivity. Taking Quebecois estimates of the unemployment rate in people with HIV and the time lost to productivity due to hospitalisations and clinic visits, they calculated that the non-medical cost to the economy of each case of HIV was CA$11,550 per year. The total annual cost to the economy of each person with HIV was therefore CA$27,659 in total for the cheapest drug regimen.
The modellers note that whereas there is clearly a higher unemployment rate among people with HIV in Canada compared with the general population, this is not the case with gay men in general, so they assume typical employment rates for people using PrEP. Given the expected life expectancy remaining to the average person diagnosed with HIV, this would mean that the cost per lifetime of an HIV infection would be $1,439,984 for the cheapest regimen.
The researchers then used a figure called the “number needed to treat”. This represents the number of people that would need to be given PrEP to prevent one HIV infection. In iPrEx this number was just under 52, leading to a cost per HIV infection averted of CA$621,390. This means that PrEP would save CA$818,594 per infection averted.
However, the future costs of treating HIV are likely to be less than they are now. This is because the situation of HIV-positive people may change: for instance, fewer in the future are likely to be unemployed. In addition, some people with HIV will die early of other causes, so the cost of their HIV infection will disappear from the model. Because of this, it is standard practice in modelling the long-term effects of health interventions to apply discounts to future costs avoided.
In the case of this model and in accord with Canadian recommendations, the modellers discount the cost of an HIV infection at 3% or 5% a year per year of infection. In the case of a 3% discount, this brings the lifetime cost of HIV infection down to $662,295 and PrEP still saves $40,905 per infection averted. But with a 5% discount, the lifetime HIV cost becomes $448,901 and PrEP then costs $172,489 per infection averted – not cost-effective under any country’s guidelines. Adding in allowances for improving the quality of life, a possible “herd immunity” effect of PrEP and other benefits brings down this cost to $60,223 but it is still not cost-saving. This is as far as this model goes. However, the cost per infection averted of giving PrEP becomes far lower if one uses the number needed to treat (NNT) to prevent one HIV infection that was actually seen in Ipergay. This was 18, not 52 as in iPrEx.The annual cost per infection averted of PrEP then becomes CA$216,018. If the NNT observed in PROUD is applied (13), then it is $156,013. This is clearly cost-saving even at the 5% discount rate, saving CA$232,883 per lifetime HIV infection averted even at the 5% discount with the Ipergay NNT – or $292,888 using the PROUD NNT figure.
In addition, as noted before, the researchers made the conservative estimate that people given PrEP under the Ipergay regimen actually have so much sex they end up taking it daily. In fact, the Ipergay participants overall used 50% of the drugs that full daily use would involve. This would mean that the cost per infection averted of an annual Ipergay regimen would be $108,009 and the lifetime saving at 5% discount would be $340,892. Even if the unemployment rate were to suddenly equalise and only direct medical costs due to HIV are allowed, this means that at the 5% discount rate the lifetime cost of an HIV infection becomes about $260,000 – still less than the cost per infection averted of an Ipergay regimen, even if taken daily.
It may be objected that the $12,001 cost of each person put on PrEP is only an annual cost and this is being weighed against a lifetime’s cost of HIV treatment – whereas PrEP could also be taken for many years. The assumption is, however, that HIV incidence in the same people if they were not given PrEP would not change. In other words, for each year on PrEP, an additional year’s worth of HIV infections is avoided, balancing out the annual cost of PrEP.
If average risk behaviour in the population who were in need of PrEP but who did not actually take it were to fall, this would mean they would experience fewer HIV infections and the savings from PrEP would be lower. Equally, if risk behaviour in people taking PrEP were to rise relative to people who do not take it – a fear often expressed – then PrEP would become relatively more expensive and save less money. If the average age at diagnosis were to rise by ten years to 40, then clearly a lifetime’s worth of HIV treatment would cost less – about 40% less by the researchers’ calculation – but PrEP would still be cost-saving for the iPrEx NNT if HIV lifetime costs were un-discounted and for the Ipergay and PROUD NNTs at all discounts.
Finally, the estimate used for the lifetime cost of one HIV infection is higher in this paper than in some papers. The US CDC estimates that the direct medical costs of lifetime HIV infection are US$618,900. This would bring the cost per HIV infection down to less than $200,000 at 5% discount – but even so PrEP at the NNTs seen in Ipergay and PROUD would still be cost effective.
Clearly, PrEP has considerable start-up costs and will not start saving money for health systems over the short term. Large savings, relative to the cost of not implementing PrEP, will only accrue over decades, and will only be significant if a significant proportion of the highly at-risk population decides to take PrEP – and as long as people at low risk do not. It may be a challenge to promote PrEP to those who need it, while withholding it from people at low risk.
However, given the results of the new studies, it is becoming hard to devise scenarios in which Truvada-based PrEP does not produce cost-savings in the long run – something managers of health systems should consider when deciding whether to implement it.