Using data from an epidemiological survey of a general population sample of Hebrew-speaking Jews living in Israel, a Western country, the prevalence of problematic substance use was reported and shorter ASSIST versions were compared to develop recommendations for a brief screening test for efficient use in primary care.

In terms of prevalence, there are two main considerations. First, for substances reported on in the previous survey from Israel (alcohol, cannabis, and sedatives) [12], prevalence of ASSIST 3.1 problematic use either descriptively increased or remained similar; for additional substances, prevalence was high for tobacco and non-negligible for prescription stimulants and painkillers, suggesting that problematic use is a pressing and possibly increasing public health concern in Israel, as worldwide [1–3]. Second, although the ASSIST 3.1 queries non-medical use of amphetamines, opioids, and other drugs, which should cover prescription stimulants and opiate painkillers, very few respondents screened into those sections. More respondents screened in when asked specifically about prescription stimulants and painkillers in a later module. This suggests that when the ASSIST is self-administered, specific categories of prescription medications should be included in the list of substances, as in the modified NIDA version [24].

Study results support several recommendations for a brief screening test. Lifetime use should be asked regarding general substance categories, specifically listing relevant prescription medications, with very rare substances in the “other” category. This may differ by country but would save time by limiting irrelevant questions. The ASSIST-FC, which queries frequency of use and other’s concerns about use, performed well at identifying those with ASSIST 3.1 problematic use, with few differences by gender or age. These two items show face validity for identifying those who may require clinical interventions. The ASSIST-FC can be used in count form (with higher values indicating more severe problems) or in binary form, with a threshold of 3+ (alcohol, 5+) generally providing prevalence similar to ASSIST 3.1, high specificity/PPV, acceptable sensitivity, and good agreement.

Those identified as having problematic use fall into two categories. The majority currently use the substance; these individuals can receive brief intervention and/or referral to specialist care for further testing and treatment, based on the health care provider’s discretion. Others had past concerns about use but did not use currently; these individuals can receive short validation of their success in abstaining and recommendation to return to the medical practitioner if abstinence becomes difficult. More generally, whether those without current use should be considered to have problematic use warrants further study. Paying extra attention to such individuals may mitigate their increased risk for relapse, but including them may lead to possibly misleading prevalence rates of problematic use, since they are not actually using.

There are additional points to consider. The ASSIST-FC did not perform well for alcohol, similar to the original ASSIST-FC study, which suggests replacing frequency of any use with frequency of binge drinking [10]. In exploratory analysis, the ASSIST-FC score with binge drinking (AUC = 0.84; 95% CI = 0.82, 0.87) did not perform better than the score with any use. Instead, perhaps craving should be included for alcohol, since the ASSIST-FCCr score performed better than ASSIST-FC for alcohol and the 9 + ASSIST-FCCr binary version showed acceptable test characteristics. The ASSIST-FCCr measures also showed better performance for other substances. Although the original ASSIST-FC did not recommend including craving since there were only small increases in test performance with craving [10], here significant improvements in test characteristics were observed. While further studies are warranted to validate the ASSIST-FCCr across substances, since many clinicians consider craving to be central to addiction and treating craving may lead to successful reduction in use, health care providers can consider including craving in the test. Additionally, the ASSIST-FC threshold could be increased to 4 + for cannabis as use becomes more normative. Last, the study aim was to recommend a shorter version of the ASSIST 3.1, so ASSIST 3.1 was the “gold standard”. But the ultimate goal is to develop efficient screening tests with strong ability to predict clinical diagnoses and treatment-related outcomes. Additional studies should assess the validity of the shorter versions (ASSIST-FC, ASSIST-FCCr) and determine their predictive ability, cross-culturally.

Study limitations are noted. First, there may be selection bias since respondents were limited to those willing and able to participate in the online survey, but using quotas enabled collection of a sample that can be considered relatively representative of the adult, Jewish, Hebrew-speaking population of Israel, in terms of key sociodemographic factors (age, gender, geographic area, religiosity, and education) [12, 16, 17]. Nevertheless, those of lower socioeconomic status may be less likely to participate in internet survey panels and may be more likely to use substances and have higher rates of problematic use; thus, prevalence may be underestimated. Second, the analyzed sample was missing key sectors of the Israeli population that would need methodological adaptations, e.g., those less likely to complete online surveys (older than 70, Ultra-Orthodox) or with cultural differences (Israeli-Arabs). Pilot studies within those groups have been conducted to determine how to collect more representative samples of the entire population in future studies. Third, non-Hebrew speakers were excluded, but > 90% of Jews in Israel have mastery of Hebrew [31]. Fourth, other biases may exist. Responses were subject to recall bias, which may be limited by the short timeframe for current use (three months). Responses were dependent on participant’s understanding of the questions, but standard, validated screening instruments were used. There may be hesitancy on reporting possibly stigmatized or illegal behaviors; this should be mitigated by the confidential online platform [22]. Fifth, no analyses were done for substances less prevalent in Israel. Sixth, the ASSIST-Lite could not be assessed for tobacco but a proxy measure (any current use) was used instead [27]. Seventh, the gold standard was the ASSIST 3.1; how well the ASSIST-FC performs as compared to clinical diagnoses should be determined in future studies. Yet, this study evaluated the shorter ASSIST versions in the general population, which may be more informative for primary care settings than the original studies, which included clinical samples [9, 10].