Latent traits of impulsivity and compulsivity: toward dimensional psychiatry

Abstract

Background: The concepts of impulsivity and compulsivity are commonly used in psychiatry. Little is known about whether different manifest measures of impulsivity and compulsivity (behavior, personality, and cognition) map onto underlying latent traits; and if so, their inter-relationship.

Methods: A total of 576 adults were recruited using media advertisements. Psychopathological, personality, and cognitive measures of impulsivity and compulsivity were completed. Confirmatory factor analysis was used to identify the optimal model.

Results: The data were best explained by a two-factor model, corresponding to latent traits of impulsivity and compulsivity, respectively, which were positively correlated with each other. This model was statistically superior to the alternative models of their being one underlying factor ('disinhibition') or two anticorrelated factors. Higher scores on the impulsive and compulsive latent factors were each significantly associated with worse quality of life (both p < 0.0001).

Conclusions: This study supports the existence of latent functionally impairing dimensional forms of impulsivity and compulsivity, which are positively correlated. Future work should examine the neurobiological and neurochemical underpinnings of these latent traits; and explore whether they can be used as candidate treatment targets. The findings have implications for diagnostic classification systems, suggesting that combining categorical and dimensional approaches may be valuable and clinically relevant.

Keywords: Addiction; cognition; compulsive; habit; impulsive.

Fig. 1.

Heat map showing correlations between variables of interest. Left: positive correlations are shown in blue, and negative in red; larger dots are indicative of stronger correlations. Right: correlation coefficients (rounded to one decimal place in the interests of clarity). gamfreq, gambling frequency per week; alcfreq, alcohol frequency per week; marfreq, marihuana frequency per week; AUD, alcohol use disorder; SUD, substance use disorder; EIQI, Eysenck Inventory Impulsiveness; EIQV, Eysenck Inventory Venturesomeness; EIQE, Eysenck Inventory Extraversion; BISAI, Barratt Attentional Impulsiveness; BISMI, Barratt Motor Impulsiveness; BISNI, Barratt Non-Planning Impulsiveness; antisoc, antisocial personality disorder; suic, suicidality on the Mini International Neuropsychiatric Inventory; PADUACW, Padua Inventory Contamination and Washing subscale; PADUADG, Padua Inventory Dressing/Grooming subscale; PADUAC, Padua Inventory Checking subscale; PADUAT, Padua Inventory Thoughts of Harm to Self-others subscale; PADUAI, Padua Inventory Impulses to Harm Self/Others subscale; nOCPD, number of obsessive-compulsive personality disorder criteria met; ADHD, total score on attention-deficit hyperactivity disorder screen; Internet, total score on Young's Internet Addiction Test; MIDIPG, gambling disorder; SSRT, stop-signal reaction time on Stop-Signal Test; qualdec, quality of decision-making on the Cambridge Gamble Test; riskadj, risk adjustment on the Cambrige Gamble Test; eat_dis, eating disorder; pgybocs, Pathological Gambling Yale Brown Obsessive–Compulsive Disorder Scale; ederr, extra-dimensional errors on the set-shifting task.

Fig. 2.

Two-factor correlated model, showing factor loadings. Variables with non-statistically significant loading (p > 0.05) are shown in strike-through. Abbreviations – see footer to Fig. 1.