|
Abstract
|
Associations between problem gambling and different forms of self-harm—including non-suicidal self-injury and suicidal ideation and attempts—have been documented in the literature.1, 2 However, the temporal direction and causal nature of these correlations are largely unresolved. Only a few studies have examined the longitudinal and temporal relationship between problem gambling and self-harm, partly due to difficulties reaching the population of interest.2 Moreover, these studies have rarely focused on how changes in the levels of these constructs over time might affect each other. In The Lancet Public Health, Heather Wardle and colleagues3 contributed substantially to existing knowledge by showing that increased problem gambling is associated with subsequent suicide attempts.
Although Wardle and colleagues’ study adequately addresses the change in the severity of problem gambling while controlling for the effect of confounding variables,3 their study design also highlights at least three methodological gaps that future research should tackle to understand the complex relationship between problem gambling and self-harm. First, future research could explore the bidirectional relationship between the two constructs (eg, whether problem gambling has a predictive effect on self-harm or vice versa, or whether they have a more complex causal relationship), considering within-person changes in both problem gambling and self-harm. Second, it is important to note that non-linear change trajectories over time along these constructs are plausible (eg, U-shaped, inverted U-shaped, or plateau-shaped curves for quadratic function); therefore, requiring a more precise description of the relationship. Finally, in addition to different subgroups of gamblers4, 5 showing different patterns of self-harm, longitudinal and causal relationships might also differ substantially due to the different motivations, personalities, and comorbid disorders in different subgroups.
These methodological concerns suggest that future research might benefit from adopting a more extensive longitudinal design to explore the longitudinal link between problem gambling and self-harm. For example, when using a panel study, such as was done by Wardle and colleagues,3 it is desirable to collect data at a minimum of three timepoints from a representative sample of the given population by systematically measuring problem gambling, different forms of self-harm, and potential confounders at each data collection wave. Using such a design would allow researchers to determine the bidirectional relationships between problem gambling and self-harm more precisely, the relationships between linear and non-linear change trajectories, and the presence of different forms of self-harm, specifically among different problem gambling subgroups. In addition, with a sufficiently large sample size, a more extensive longitudinal design could facilitate the use of statistical analysis methods that can provide a more accurate view of the complex longitudinal relationship between problem gambling and self-harm. For example, random intercept cross-lagged panel modelling would allow analysis of the bidirectional relationships between problem gambling and self-harm and could distinguish stable between-person differences from state-type within-person differences.6 Additionally, latent growth modelling can allow researchers to describe linear and higher-order growth trajectories (eg, quadratic and cubic functions) for problem gambling and self-harm to assess the relationships between the baseline levels and the different change trajectories of these constructs.7 Furthermore, latent class growth analysis could be used to distinguish and identify different problem gambling classes to investigate similarities and differences between different gambler subtypes in various forms of self-harm and motivations.8 It is noteworthy that for each of the above-mentioned analytical approaches, it is necessary to consider and control for the effect of confounding variables (eg, their initial levels and changes in them over time) that might influence the association between problem gambling and self-harm (eg, psychiatric symptoms, and emotion regulation) ...
Language: en |