Results from a new study could help healthcare services develop better tools to identify people more likely to take risks with their health.
Individuals who are more prepared to take risks may be more likely to engage in behaviours that could harm their health, such as smoking and excessive drinking. Tools for measuring and understanding attitudes to risk could therefore help healthcare services develop more targeted interventions to improve public health.
To estimate attitudes to risks, people can be asked directly about risky behaviours, or ‘indirectly’, for instance via lottery-style questions. In the lottery approach, individuals are presented with hypothetical gambles, where they choose between options with different probabilities of happening. Normally, these are tied to economic outcomes, for instance, choosing between a low financial reward with a more certain outcome, or a high reward with a less certain outcome.
There are concerns, however, that using lotteries with monetary outcomes may not accurately predict attitudes to health-related risks. In a new study published today in Health Economics, researchers from HERC investigated a potential alternative: a novel lottery method using health states as outcomes.
The study was based on a representative sample of 4,000 UK adults who each completed an online survey. Respondents were asked questions about how often they engaged in several harmful health behaviours, such as smoking and excessive drinking, and whether they would adhere to taking a full course of antibiotics. They were then asked directly to rate their willingness to take risks, both in general and in health-related situations, from ‘Not at all prepared to take risks’ to ‘Fully prepared to take risks.’
In the second part of the survey, the respondents completed a series of hypothetical gambles, where the outcome was either full health or poor health (the gamble); or a certain outcome for an intermediate health state. Each scenario carried a different probability of achieving full health in the gamble.
The questions were designed to find out how low the probability of ending up in poor health would have to be for the respondent to choose the gamble. Participants were scored as being more risk-averse if the outcome of full health had to be more certain before they took the gamble.
The researchers compared the power of the answers from the direct questions and lottery situations to predict whether individuals were more likely to engage in risky health behaviours.
Key findings
- The direct questions were better able to predict whether individuals engaged in health-related risky behaviours, compared with the indirect lottery approach.
- In the direct approach, the participants’ self-reported willingness to take health-related risks more strongly predicted risky health behaviours, compared with their self-reported willingness to take risks in general.
- In the lottery approach, the placement of the gamble had a strong effect on participants’ choices. Respondents were significantly more likely to take the gamble when it was labelled “Option 1” and placed on the left hand side, compared with when it was labelled “Option 2” and placed on the right.
Various surveys, including those unrelated to health, have previously reported a left-side selection bias, suggesting that people may be more likely to choose options that are shown earlier in a list or shown on the left hand side. According to the research team, left-side bias occurs more often in cognitively demanding surveys (for instance, those based on hypothetical scenarios), compared with simpler questions (such as favourite fruit or colour).
In this study, around 1,400 respondents were screened out because they failed to recognise that the poor state of health was worse than the intermediate one. This suggests that the lottery scenario may have been a challenging concept for some respondents, causing them to pick the first option automatically to complete the question with minimal effort.
Lead author Murong Yang, a doctoral student at HERC, said: ‘Our results indicate that risk preferences elicited from direct questions are better predictors of risky health-related behaviours, compared with those elicited using a lottery-style approach with health outcomes. However, developing lottery questions in a less complex way might reduce the left-side bias effect, and improve their predictive power.’