Gambling disorder (GD) is a clinical diagnosis characterized by frequent and continual engagement in gambling activities, resulting in detrimental impacts across family, social, economic, and occupational domains of living (ICD-10, 2016). Problem gambling is similarly characterized but in the absence of meeting criteria for a formal diagnosis (Potenza et al., 2019). Both the disorder and problem gambling are considered serious public health issues given their impact not just for the individual involved but also for their personal network (e.g., an estimated 5 to 15 people are affected for every one individual with GD/gambling problems; Kalischuk et al., 2006). Global reported prevalence rates vary by country (i.e., 0.1%-5.8%), although direct comparisons present challenges given methodological differences (for a comprehensive review, see Calado and Griffiths (2016)). In Norway (where the present study was carried out), the prevalence of self-reported gambling problems in 2019 was 1.4%, which increased from 0.9% in 2015 (Pallesen et al., 2020).

In the context of increasing prevalence, there is a high importance placed on the identification of temporal risk markers associated with GD. One such risk marker may be substance use disorder (SUD). Notably, a substantial body of literature has identified that there is a significant co-occurrence of SUDs (e.g., alcohol, drugs, tobacco) among individuals with GD (e.g., Lorains et al., 2011, Yakovenko and Hodgins, 2018). Estimates of co-occurrence are typically highest for nicotine dependence, reaching upwards of 60%, and lowest for drug use disorders (DUDs), leaving alcohol use disorders (AUD) in between (often around 20%-30 %) (Lorains et al., 2011, Rash et al., 2016). The literature has identified several shared risk and etiological factors of both GD and SUDs including being male, young age, deprived socio-economic background, somatic and mental illness, as well as genetics (Dowling et al., 2017, Merikangas and McClair, 2012, Potenza et al., 2019, Yau and Potenza, 2015).

While co-occurrence is recognized, there are few studies that have delineated the temporal sequencing between GD or problem gambling and SUDs using individual-level data (Yakavenko & Hodgins, 2018), and mixed findings have been reported regarding the type of SUD involved. For example, using a large cross-sectional nationally representative face-to-face household survey from the US (n = 3,435), Kessler et al., (2008) reported that GD largely preceded tobacco dependence, whereas alcohol or drug abuse preceded GD. On the other hand, in another cross-sectional US-based study also using face-to-face interviews, Cunningham-Williams et al. (2000) reported that cannabis dependence preceded GD in a combined sample of drug treatment-seekers and a high-risk drug user non-treatment sample (n = 990), although in the specific case of stimulants (i.e., cocaine or amphetamines), this type of SUD was more likely to follow GD. Moreover, in a cross-sectional study using an Australian sample of treatment-seeking problem gamblers (n = 267), Haw and Holdsworth (2016) reported that for all three categories of self-reported SUDs (alcohol, nicotine, drugs), these were most often precursors to problem gambling. Taken together, these findings – which are all based on self-reported onset of both problem gambling and SUD – seem to suggest some mixed patterns of temporality that may be dependent on particular factors, including the type of population sampled (e.g., general population, treatment-seekers) and type of SUD. Noteworthy, DUD (i.e., cannabis use) and AUD have consistently been found to precede GD across these studies.

This limited literature offers little regarding possible differences in temporality with regard to sex differences. However, findings using cross-tabulated census data from the Norwegian Patient Registry, suggest sex differences in timing of SUD for patients with GD (Leino et al., 2021). More specifically, males with GD had a higher risk of a SUD diagnosis within the same year whereas females with GD had a higher risk of diagnosis of SUD one or more years after initial onset of GD. This study also found SUD to be more prevalent among individuals with GD (i.e., 22.5%), as compared to GD among individuals with SUD (i.e., 0.7%). While the Leino et al. (2021) study use the same registry data as the current study to investigate SUD and GD, the studies differ in terms of their aim, scope, and research questions. For example, Leino et al. (2021) used aggregate level data and did not undergo any inference-based statistical analysis, whereas the present study examines individual level data using inferential analysis. The time span is also longer in the present study, and we further distinguished between type of SUD experienced.

It should also be noted that methodological factors pose major challenges for the identification of temporality and the interpretation of previous findings. As aforementioned, common in the literature is the use of retrospective cross-sectional studies asking participants to recall the onset or occurrence of symptomology concerning both GD and SUDs. Moreover, the use of subjective data (i.e., self-report data from surveys and interviews) rather than the use of objective data (e.g., medical records and registry data), is most often employed. Self-report data, in combination with cross-sectional retrospective designs, enhances the potential risk for introducing multiple forms of bias, including recall bias, social desirability, and differential recall (Coughlin, 1990). Shared method variance bias is also of concern. This makes it challenging to accurately delineate the temporal order between onset of disorders. Consequently, prospective studies using objective data are needed to enhance the rigor in this emerging area of work and to further add to the understanding of which type of disorder is more likely to precede the other.