This study is likely the first to report on demographic, economic and social correlates of a complete national SUD patient cohort and to compare a broad range of correlates across patients with different SUDs, including sedatives/hypnotics patients. Compared to the general population, the SUD patients had on average substantially lower education level, income, and wealth, they were less often in paid work and more often recipients of social security benefits. Among SUD patients, users of illicit substances scored worse on all socio-economic status indicators, compared to users of licit substances. Comparison of patients with different SUD diagnoses further revealed important differences within and across patient groups. Demographic differences, such as the relatively low mean age among cannabis patients or the high share of females among sedatives/hypnotics patients should be noted, as should the large differences in education level, labour market participation, income and wealth across patient groups, since these factors are likely to impact on treatment outcome and on the potential for future individual welfare and economic independence.
Our findings corroborate those from previous studies with regard to correlates of PWSUD [8, 9, 18, 30, 31]. Moreover, our findings accord with those in previous studies with regard to comparisons between people with AUD and DUD, finding that the former group is characterized by higher age [20, 22, 23], higher proportion in paid work [22, 23] and higher income [22]. Focussing on sedatives/hypnotics patients specifically, we found that these were older and more often women compared to other SUD patients, whereas, for most other socio-demographic correlates, these patients did not deviate from other patients. A recent review of the epidemiology of sedatives/hypnotics misuse (i.e. benzodiazepines) showed that the literature is mixed, with no clear patterning of socio-demographic correlates of this group [26]. These mixed findings may, however, reflect the fact that the primary studies included in the review were based on heterogenous samples, including patients in SUD treatment and general population samples.
What can explain the observed socio-demographic differences between PWSUD and the general population and between substance-specific groups of PWSUD? Overall, it seems likely that the underlying mechanisms are complex, and, in the following, we will briefly discuss three possible main pathways.
The first pathway pertains to socio-demographic selection mechanisms into extensive or problematic substance use. With regard to alcohol and illicit drugs, socio-demographic differences are more prominent when use has evolved to extensive use, and this occurs more often among males and in low socio-economic status groups [19]. On the other hand, extensive use of sedatives/hypnotics is equally or more often seen in women [26]. Moreover, it is well established that impulsivity, or poor inhibitory control, is a precursor for SUD vulnerability [32], and as impulsivity is also associated with low academic achievement [33], this individual correlate may, in part, explain the association between low education level and SUD. Moreover, impulsivity occurs more frequently in PWDUD compared to PWAUD [34], which may partly explain the higher share of low education level in the former group. The tendency for those with SUDs from illicit drugs to live in urban areas may, in part, reflect easier access to illegal drugs at a lower price, and a possible influx of drug users from rural areas.
The second pathway pertains to socio-demographic differences in the utilization of SUD treatment services. Studies from the US and European countries reported higher likelihood of specialized treatment utilization among those with low income and those with psychiatric co-morbidity [35, 36]. Other factors that seem to increase treatment-seeking include younger age and lower education level [36]. Moreover, while women are generally more likely than men to seek treatment in general health and mental health care services, some studies found that women are less likely to seek treatment for alcohol problems [37, 38]. We found an elevated likelihood of urban dwelling among SUD patients. This may possibly be explained by easier access to treatment services, even in a treatment system with universal coverage, such as the Norwegian one. Some evidence suggests that PWAUD are less likely than PWDUD to seek treatment [36], and as impulsivity is associated with lower academic achievement and also occurs more frequently among PWDUD than among PWAUD, differential treatment utilization by PWAUD and PWDUD may contribute to the higher proportion of low education level among patients with illicit SUDs compared to other SUD patients.
The third pathway pertains to the social drift reflecting consequences of extensive substance use. People with SUDs are at elevated risk of early retirement, unemployment, low income and need of social assistance [39, 40]. Further, SUDs affect the ability to commit in family life and they are associated with family dysfunction and child abuse and neglect [41]. The impact of SUDs on unemployment and financial difficulties seems to be larger for PWDUD compared to PWAUD [40], which corroborates our findings. Also in this regard, the more frequent occurrence of impulsivity in PWDUD compared to PWAUD is likely important; impulsivity and low self-control are found to be important predictors of unemployment [42], and may thus in part explain our observations of a lower proportion in paid work, and thereby also lower income and greater financial difficulties, among patients with illicit SUDs compared to other SUD patients. Moreover, the high financial costs of extensive substance use, and particularly so for illicit drugs [43], may also in part explain the lower participation rate in legal employment among PWSUD from illicit, as compared to licit, substances.
Although we cannot determine which of these three pathways, or what combination of them, are at work here, improved knowledge of the socio-demographic correlates of SUD patients is warranted for several purposes. More exact knowledge regarding socio-economic patient correlates may help to better design, implement and evaluate drug treatment services to improve outcome and the potential for future individual welfare and economic independence. For instance, opioid patients with a high level of marginalization and a low level of own resources are in need of far more societal efforts to reach a stable situation than cocaine and alcohol patients who are better educated, included in the work force and with a better financial situation. Cannabis patients also have a low level of education and economic means, although younger than opioid patients and thus with fewer drug consuming years. These and other observed differences between patients with respect to gender, age, education, income (including stable income like disability pension and short term income like supplementary benefits) and labour market participation may be taken into account when designing treatment programs for subgroups of PWSUD. They may also help explain differences in treatment outcomes across patient groups. Better knowledge of patient characteristics may further improve measures aimed at reducing social inequality related to SUD as it may reveal for which subgroups inequality is particularly distinctive.
Further, the complex mechanisms underlying progression into, and development of, various specific SUDs is currently not well understood. Using linked administrative datasets of national cohorts of SUD patients to describe and compare a range of demographic, economic and social correlates may contribute to this end, as suggested by the possible pathways discussed above.
Limitations
While register-based data, as employed here, are not hampered by the typical limitations found in survey studies (selection and attrition bias, response biases and few observations of PWSUD), register data come with other limitations. Routine data may vary in data quality, coding may vary between persons and institutions, and it is often difficult to gain exact information on how such data were generated [44]. In our context, we do not know precisely what underlies the making of a specific SUD diagnosis, and as co-use of several substances is typical in PWSUD [45], it is possible that a single substance diagnosis (e.g. alcohol use disorder) may be somewhat arbitrarily assigned. This type of misclassification, however, is relevant only if it is systematic and correlated with socio-demographic correlates, which cannot be precluded. Further, our findings reflect a cross-sectional ‘snap-shot’ of patients in treatment for SUDs. Thus, time dynamics, for instance with regard to changes in substance use careers and changes in social drift, could not be assessed in this study. Finally, transferability of our findings to other settings may depend on the extent to which they differ from the Norwegian setting, for instance when there are substantial differences in prevalence of substance use, the type of treatment system and treatment coverage, and the type of welfare schemes.
Adjustment for age and gender amplified, attenuated, or even reversed the unadjusted differences in socio-demographic correlates across patient groups. This complexity is noteworthy with regard to understanding likely underlying mechanisms and for external validation of findings from other patients populations.