The shape of gambling risk‐curves for frequency, expenditure and proportion of income in Australia.

Background and aim: Examining risk‐curves is important to understanding the degree to which indices of gambling consumption are associated with gambling‐related harm. Risk‐curves have largely been described as J‐shaped, suggesting that at low levels of consumption harm remain constant but then increase sharply at a certain threshold. Alternative methods in recent work, however, have described risk‐curves as linear and R‐shaped, indicating that risk of harm increases as consumption increases at all levels of consumption. The aim of the current study is to estimate the shape of gambling risk‐curves using competing methods. Design Systematic comparison of gambling risk‐curves using categorical (via plots) and continuous (via bootstrapped regression analyses) operationalizations of gambling consumption. Setting and participants Data were 2873 gamblers (1417 women) from the fourth Social and Economic Impact Study of Gambling in Tasmania. Measurements Gambling‐related harm was assessed using the Problem Gambling Severity Index (PGSI) and Short Gambling Harm Screen (SGHS). Gambling consumption was assessed as annual frequency, expenditure and expenditure as a proportion of income. Findings Categorical gambling consumption data plotted with equal magnitudes evidenced J‐shaped risk‐curves. When the magnitude of gambling categories was re‐scaled to the mid‐points, risk‐curves no longer appeared J‐shaped. Additionally, bootstrapped regression analyses using the continuous gambling consumption data did not provide evidence for J‐shaped risk‐curves. Conclusions: Gambling risk‐curves in Tasmania appear not to be J‐shaped, but rather suggest that risk of gambling‐related harm increases with even small increases in gambling consumption. [ABSTRACT FROM AUTHOR]