Behavioural and financial incentives may improve HIV treatment outcomes

While making medications free can remove barriers to access for individuals who cannot pay for treatment, data suggest that for most people accessing care in industrialised countries, "making medications available for free or low cost will not solve problems with medication non-adherence," according to a presentation by Kevin Volpp from the University of Pennsylvania last week at IDWeek 2014 in Philadelphia, United States.

Volpp, a leading researcher on the impact of financial and organisational incentives on health behaviour and outcomes, described a range of other economic and behavioural incentives that could be used to reduce attrition along the HIV cascade of care – increasing the number of people tested and diagnosed with HIV who remain in care, take their antiretroviral medications, and have fully suppressed viral load.

Once effective treatments for conditions exist, individual behaviour is key to optimising outcomes using those treatments. Studies have shown that if the causes of premature mortality are divided into five domains – genetics, social circumstances, environmental exposures, behavioural patterns, and health care – behavioral causes account for nearly 40% of all deaths in the US. This suggests that the "single greatest opportunity to improve health and reduce premature deaths lies in personal behaviour."

But humans are not perfectly rational, and providing them with information about what to do or not do may not be enough to get them to do what is best for them – such as taking treatment that may save their lives. This is a universal problem that has been seen across many diseases.

Volpp also shared Centers for Disease Control and Prevention (CDC) data on the HIV care cascade that was described in other presentations at IDWeek, showing substantial attrition at each step of the continuum of care after diagnosis, so that of all the people who receive a diagnosis of HIV in the US, only 28% actually achieve viral suppression on antiretroviral therapy.

But it may be possible to look at and leverage the types of errors in personal health decision-making to design interventions – including behavioural and economic interventions.

Volpp described some of the types of decision errors that people can make, as well as some potential solutions:

• Status quo bias/inertia: Set up the system so that the default favours healthy behaviour;
• Present-biased preferences (myopia): Make rewards for beneficial behaviour frequent and immediate;
• Overweighting small probabilities: Provide probabilistic rewards (e.g., lottery) for self-interested behaviour;
• Regret aversion: Tell people they would have won had they been adherent;
• Loss aversion: Put rewards at risk if behaviour does not change;
• Framing and segregating rewards: $100 reward likely more effective than $100 discount on premium.

Volpp suggested that different approaches could be applied to different parts of the care cascade. For initial diagnosis and linkage to care, programmes could look at the use of defaults, while it may be advisable to consider the use of incentives to improve retention in care, medication adherence, and achieving viral load suppression.

Default bias: For instance, changing the default bias from opt-in to opt-out has been shown to have a dramatic effect on organ donation in Europe. In countries such as Denmark, UK, Germany, and the Netherlands, where individuals have to give active consent for organ donation upon death (opting-in by signing consent forms in advance), organ donation rates are quite low, ranging from 4% to 28%. But the rate is above 98% in six out of the seven European countries that have an opt-out default bias (and 86% in the seventh).

One way to increase the likelihood that people with chronic illnesses will get their prescription for long-term medications refilled would be to set up a system offering automatic refills by phone via an interactive voice response (IVR) system. If this is done, the default should be set up so that actions benefiting the long-term interests of individuals are easier to choose – the path of least-resistance.

But opt-out is not always an option in some settings.

An alternative option would be the use of "active choice" that requires participants to make a choice to increase participation in automatic refill programmes. A variation of this is "enhanced active choice", which favours one alternative by highlighting losses resulting from the non-preferred alternative. Rather than an opt-in message "Press 1 if you would like to be transferred to a customer care representative now," or "Press 2 if you are not interested," the programme could switch to an enhanced active choice default: "Press 1 if you prefer to refill your prescriptions by yourself each time" or "Press 2 if would you prefer us to do it for you automatically."

Volpp described one large study in which the opt-in enrolment via IVR into an automatic refill program was only 16% – which was increased to 32% through the use of enhanced active choice.

Some examples of how defaults and enhanced active choice could be used:

• HIV testing on an opt-out basis in emergency departments and in clinics in high-prevalence areas;
• Follow-up appointments made automatically at the time of diagnosis;
• Enhanced active choice to sign up for automatic prescription refill programmes.

Reward systems: Reward systems can help incentivise desired behaviour, but "the design elements in reward programmes are critical," said Volpp. For instance, rewards that are offered once per year ignore myopia, and may fail to change current behaviour. Setting single high-threshold targets for a reward may provide incentives only to those who are already close to achieving it, and may do little to encourage the rest.

Financial incentives are also complex and research findings may be counter-intuitive. For instance, in resource-limited settings where cost is a barrier to accessing medications, offering treatment for free may greatly increase access to care. But in other contexts, with people who are insured, decreasing co-pays has not been shown to greatly increase adherence and may not change outcomes. This is possibly because payment of the co-pay is not a barrier for people, and because the reduction in cost was small or the reward feedback too infrequent.

Technology may create new opportunities to reward adherence, however, by "automated hovering," or indirectly watching behaviour. This leverages three key developments:

• Increased focus on healthcare financing on population-based health as opposed to fee for service;
• Social media and wireless devices;
• Behavioural economics and better understanding of what makes people predictably irrational, and the implications this has for health behaviour.

Data on whether an individual takes medication or uses a scale or pedometer could be captured by the pill bottle or other device and transmitted to a central server automatically. The programme could capture behaviour, calculate an incentive, and transmit a communication to the participant. Funds could then be electronically transferred to the participant. Whether this would work in practice, or whether people would try to cheat the system remains to be seen.

The cost of a programme could be decreased by providing incentives for adherence as part of a lottery. In one study, rates of non-adherence to the blood thinner warfarin were significantly lower using daily lottery-based incentives, where participants had a 1 in 5 chance of winning $10 a day, and a 1 in 100 chance of winning $100 each day if they took warfarin the previous day. 

However, financial incentives may not be the most effective approach. Several studies in other disease areas suggest that social incentives (peer mentoring) may be a cost-effective way of improving outcomes. In one 6-month controlled trial involving people with diabetes, participants were randomised to one of three study arms: control (usual care), meeting with a peer mentor at least weekly, or financial incentives to improve glucose control ($100 to drop HbA1c by 1 point and $200 to drop 2 points or achieve HbA1c of 6.5%).

At the end of the study, the mean change in HbA1c from baseline was essentially unchanged with usual care and reduced by 0.46% via financial incentives. However, those who met weekly with peer mentors did the best, reducing their HbA1c by 1.08% in the peer mentor arm.

One question that came up after the session was how to afford the cost of these programmes – either providing financial incentives to patients or paying for peer counsellors.

"I think it is important to point out that a lot money is spent on treatment...so the question is, who is at financial risk if people do not do well on treatment, and would these programmes be willing to spend some of the money that is currently spent on treatment on services that might help those treatments achieve better outcomes?" Volpp replied. "There may be managed care organisations that might be perfectly willing to do that, because it might be in their financial interest to do so – and that may be more broadly true of other payers as well,” he concluded.