Our data corroborate the situation in many other countries (5–10) of discordance between expert harm rankings of popular drugs of abuse and their regulation by narcotic laws, as evidenced most strikingly by the assessment of alcohol—judged to be among the most harmful substances abused in our country. The relatively high prevalence of alcohol use/abuse (compared to that of less-frequently abused but perhaps more dangerous substances) likely contributes to its dimension-specific ratings, e.g., harm to others, as well as to its overall position. Similarly, the decreasing prevalence of nicotine use in Germany (as tobacco smoking has been banned from many public areas such as hospitals, educational establishments, public transport, restaurants, pubs, and discos during the last 10 years or so) may contribute to a lower-than-expected harm ranking. In addition it should be mentioned that nicotine use, despite its ability to produce considerable behavioral dependence is hardly associated with dramatic psychiatric effects, e.g., in contrast to alcohol or hallucinogen use. This study was the first to compare the harms of various NOAs with harms of well-characterized substances of abuse, and as expected identifies the harms of NOAs to be considerably lower than those of the traditional substances of abuse. The present study was also the first to include synthetic cannabinoids and propofol in an overall-harm ranking schema, which may be beneficial for the psychoeducation of users, for regulatory considerations, or for defining fields of political action for health promotion.

NPS (cathinones and synthetic cannabinoids) have been assigned to the top harm-level group here. Policy-makers and clinicians would benefit from further data about the NPS-phenomenon, e.g., associated morbidity (32, 33) and mortality which are on the rise (33).

Compared with the EU-rating from 2014 (10), cannabis, methadone and nicotine were assessed as less harmful, while crack, methamphetamine, GHB, cathinones, ecstasy, and psychotropic mushrooms were seen as more harmful (Figure 3). Cannabis and hallucinogens (i.e., ketamine, psychotropic mushrooms and LSD) were considered to be on the harm level of benzodiazepines or barbiturates. It should be mentioned that psilocybin (in Figure 1 listed as psychotropic mushrooms) and LSD have both enjoyed re-emerging therapeutic potential in psychiatric diseases and appear to show low abuse potential in that context (34).

It is interesting to note that opioid analgesics were not within the top ranks of harmful drugs. This could perhaps be related to the fact that an “opioid epidemic” (such as that in the USA, Canada and Australia), is yet not apparent in Germany or in Western Europe (35–38). The relatively low harm rankings of prescription opioids in our study stand in stark contrast to the high level of stigmatization of illicit opioids. These findings are congruent with the multi-decision analysis of nine experts (8 from the United Kingdom and 1 from the Netherlands) suggesting that the overall harms of non-medically used prescription opioids are less than half that of injected street heroin (39).

Methadone was assessed as less harmful than standard opioid analgesics, which viewpoint might be biased by addiction medicine physicians' conception of methadone primarily as a standard opioid dependence maintenance treatment, which in this context has been repeatedly shown to reduce morbidity and mortality (15). In the context of illicit use and abuse, methadone's harms (e.g., apneic and torsades-de-pointe deaths, addiction, and diversion) are obviously considerably higher than those of several other drugs ranked above it. This exposes a major limitation of drug harm-ranking studies based upon subjective assessments as they may not allow for clear differentiation between the harms of a drug with therapeutic indication in a medical context vs. illicit use/misuse outside of that context. These discrepancies in ranking of analgesics among other agents suggest that perhaps raters' experience in pain medicine should have been surveyed as well.

It cannot be excluded that our ratings may be biased toward metropolitan rather than rural perception of substance use harms; clarifying this would require further study in larger samples. Also, a possible gender influence on drug harm perceptions was not explicitly investigated here (40, 41). As we had sent out the questionnaires without tracking all recipients, requesting forwarding to other German addiction medicine experts, we are unable to provide information about the exact number of experts who finally received our questionnaires. However, such modus operandi is not unusual for studies of this kind (5). Other limitations, similar to previous studies (5–10) include the fact that the present work cannot claim to meet strict requirements for representativeness. We aimed to reduce subjectivity biases by recruiting a large and homogeneous study group (all physicians specializing in addiction medicine). However, no official statistic exists for how many specialists with more than 5 years of experience in tertiary care of SUD were working in Germany at the time of the study. We estimate that number to be somewhere between 250 and 500 physicians, thus our sample may yield a minority viewpoint. In Germany, addiction medicine experts usually are psychiatrists or general practitioners. Unlike the English (5), EU (10) and Australian (9) studies, we used no consensus–feedbacks. While this additional step may have increased the likelihood of survey participants' agreement (42), we decided against this course, because consensus-based decisions per se do not eliminate subjectivity (43) and there exists no “one-size-fits-all-method” for benefit-risk assessment (44). Furthermore, prior consensus-based studies utilized smaller samples comprising addiction experts from different professions (5, 9, 10), whose heterogeneity of experiences in the treatment of SUD more likely needed a consensus-based decision strategy than did our homogeneous group. Similar to the Netherlands (6), the Scottish (7), and the French research groups (8) we performed an “ad-hoc” assessment, using validated health and social dimensions, which have been utilized in previous (5, 10) and recent (9) empirical studies. This decision to use an “ad-hoc” format maximized the return of completed questionnaires.

Apart from the novel inclusion of NOAs, synthetic cannabinoids and propofol, there are a few strengths of the present study: (i) the utilization of one of the largest samples in this type of study; (ii) the considerable multidimensional addiction medicine experience of the participants, including that of rehabilitation clinic specialists (Figure 2), which in Germany focuses heavily upon psychosocial dimensions and outcomes; (iii) comparison with the previous EU-rating (Figure 3); and (iv) the addition of comparisons of illicit and licit drug rankings to the current literature.

The results of this cross-sectional questionnaire-study update the average overall harm (with component harms from various health and social dimensions) arising from use/misuse of various psychoactive substances (including prescription analgesics) from the perspective of German addiction medicine specialists. It should be emphasized however that these relative overall rankings apply to population-level risks, and depending on the individual and situational context as well as on the intensity of the individual misuse, nearly every psychoactive substance can be used in a very dangerous and harmful way.