The diagnosis of substance-induced psychosis

Substance-induced psychosis (SIP) is a psychotic condition appearing in close temporal relationship to drug intake in people without primary psychotic disorders. In ICD-10, SIP is coded as a decimal specification of a substance use disorder, F1x.5, where ‘.5’ indicates a psychotic state and the x is replaced by numbers from one to nine according to which substance is believed to have caused the psychosis. In ICD-11, the system is kept, but evolved to differentiate between even more substances. The North American system of DSM 5 indicates a diagnosis of 292.1 or 292.2 for alcohol or substance/medication-induced psychosis, respectively.

There is scarce knowledge regarding how common SIP is. We recently found a relatively stable annual incidence rate of SIP in treatment in Scandinavia over the past twenty years of between 9.3 and 14.1 per 100.000 person-years, though cannabis-induced psychosis significantly increased in the second half of the period (Rognli et al., Reference Rognli, Taipale, Hjorthøj, Mittendorfer-Rutz, Bramness, Heiberg and Niemelä2022). We are not aware of similar studies outside of Scandinavia. Further, SIP represents between 6.5% (Thompson et al., Reference Thompson, Marwaha, Winsper, Everard, Jones, Fowler and Marshall2016) and 10.3% (O'Connell, Sunwoo, McGorry, & O'Donoghue, Reference O'Connell, Sunwoo, McGorry and O'Donoghue2019) of all first-episode psychoses (FEP) entering early intervention services.

In the acute phase, SIP and schizophrenia are hard to distinguish (Dawe, Geppert, Occhipinti, & Kingswell, Reference Dawe, Geppert, Occhipinti and Kingswell2011; Medhus, Mordal, Holm, Mørland, & Bramness, Reference Medhus, Mordal, Holm, Mørland and Bramness2013b). The difference is related to time, as a SIP should (Kendler, Ohlsson, Sundquist, & Sundquist, Reference Kendler, Ohlsson, Sundquist and Sundquist2019) resolve relatively rapidly with abstinence, reflected in the diagnostic criteria. According to ICD-10 (and similarly in ICD-11), symptoms should be significantly attenuated within one month and completely within six months (World Health Organization, 2004). According to the DSM-5, a SIP diagnosis should not be given if the psychosis lasts ‘a considerable period of time, for example over one month’ (American Psychiatric Association, 2013).

The neurobiological mechanisms underlying different SIPs may vary from one drug to another. Specifically, cannabis, amphetamines, and cocaine, as well as hallucinogens exhibit distinct neurobiological effects that could contribute to SIP, as described well in two recent reviews (Fiorentini et al., Reference Fiorentini, Cantù, Crisanti, Cereda, Oldani and Brambilla2021; Garson, Castle, & George, Reference Garson, Castle and George2023). We fully acknowledge the capacity of some types of substances to, at times and in some, precipitate psychotic symptoms. These associations and possible mechanisms are well described by others, such as the mentioned reviews, and will not be covered here.

In this current narrative review, we argue that even though there are properties with many drugs of abuse that could increase the risk of SIP, these characteristics do not sufficiently explain SIP. The construct validity of SIP has been questioned (Mathias, Lubman, & Hides, Reference Mathias, Lubman and Hides2008; Thirthalli, Benegal, & Gangadhar, Reference Thirthalli, Benegal, Gangadhar, Keshavan and Sachdev2010), and recent evidence has further substantiated this critique. Our aim is to present an updated understanding of SIP and to discuss implications of this understanding. Ultimately, we challenge the concept of SIP altogether.

An assumption of cause

In earlier times, diagnoses often included information about the assumed cause of the mental disorder, such as reactive and endogenous depression and psychosis. Modern diagnostics place more emphasis on describing signs and symptoms and avoid pointing to possible underlying causes. In the introduction to DSM III in 1980, Robert L. Spitzer wrote: ‘For most […] disorders, however, the etiology is unknown. A variety of theories have been advanced, buttressed by evidence – not always convincing – to explain how these disorders come about. The approach taken in DSM-III is atheoretical with regard to etiology or pathophysiological process except for those disorders for which this is well established and therefore included in the definition of the disorder’ (Spitzer, JB, & Skodol, Reference Spitzer, JB and Skodol1980). With the introduction of DSM III, it was thus recognized that for most mental disorders the cause is largely unknown and at least very complex and varies from individual to individual. Diagnoses hence need to be descriptive. A diagnose should be based on what we can observe and describe, not be causal or explanatory, because we rarely know or fully understand the cause of the disorder. Some diagnoses deviate from this principle. In Post-Traumatic Stress Disorder (PTSD), the assumption of cause may be seen as more immanent, and treatment focuses on both the cause (reprocessing the traumatic event) and its consequences (dampening anxiety and avoidance) (Kirkpatrick & Heller, Reference Kirkpatrick and Heller2014). However, even in PTSD vulnerability will play a role as the majority of those exposed to potential traumatic events do not develop PTSD. The term ‘substance-induced psychosis’ is also an explanatory diagnosis. But SIP differs from PTSD in terms of how the causality assumptions impact treatment. In SIP, removing the assumed cause (drug use) is the predominant intervention, while treatment of the consequence (psychosis) to receives less attention. As we shall see, substance use is not a sufficient cause of SIP, and several other risk factors are largely overlooked, partly due to this nomenclature. We argue that in SIP, the assumption of cause disproportionately guides our attention to substance use, possibly resulting in suboptimal treatment.

Substance use is not the cause of SIP

The fact that an SIP diagnosis follows the use of substances is not an argument for causality, but rather a tautology, driven by the diagnostic criteria themselves. It does not bring us very far in understanding the nature of SIP. Even if substance use forgoes the psychosis and some ways precipitates it, substance use only becomes a necessary risk factor because we choose to have a distinct diagnostic category for it. Contrary to this we believe it should be viewed as only one of many risk factors for SIP. It is necessary, also given the diagnostic criteria, but it is not sufficient. The majority of those who use any drug do not become psychotic. This applies even to those with extensive drug use. For instance, a prospective cohort study on methamphetamine users showed a strong dose–response association between days of use and probability of having psychotic symptoms the same month, but even among the most frequent users, some had no psychotic symptoms (McKetin, Lubman, Baker, Dawe, & Ali, Reference McKetin, Lubman, Baker, Dawe and Ali2013). In experimental studies where subjects were administered increasing amounts of amphetamines or cannabis to induce psychosis, not everyone became psychotic, even at high doses (Angrist & Gershon, Reference Angrist and Gershon1970; van der Steur, Batalla, & Bossong, Reference van der Steur, Batalla and Bossong2020). We have previously published a study comparing blood concentrations of amphetamines in people admitted to an acute psychiatric ward with blood concentration in apprehended drivers suspected for driving under the influence (Medhus, Holm, Mørland, & Bramness, Reference Medhus, Holm, Mørland and Bramness2013a). Interestingly, among those positive for amphetamines, blood concentrations were in fact lower among the psychiatric patients than among the apprehended drivers, indicating that other factors than amphetamine blood concentration are essential for the development of psychosis.

A relatively recent study from one of the co-authors showed that infections, which have been identified as possible risk factor for schizophrenia, are also associated with an increased risk of developing SIP (Hjorthøj, Starzer, Benros, & Nordentoft, Reference Hjorthøj, Starzer, Benros and Nordentoft2020). Another study showed that family vulnerability to psychosis is significantly more common in those who developed SIP compared to the general population (Kendler et al., Reference Kendler, Ohlsson, Sundquist and Sundquist2019; McKetin et al., Reference McKetin, Clare, Castle, Turner, Kelly, Lubman and Berk2023). More studies are needed to identify other risk factors for SIP besides substance use, but these examples demonstrate that vulnerability is also a key factor for SIP, similarly as for the development of primary psychosis (Løberg et al., Reference Løberg, Helle, Nygård, Berle, Kroken and Johnsen2014). In other words, just like for other mental disorders, the etiology of SIP is multifaceted, and the idea of substance use as the cause stands corrected in the view of scientific evidence.

The crucial question is whether substance use can cause psychosis in individuals who, if they had not used drugs, would have remained healthy. Large, randomized experiments might bring us closer to an answer, but would of course be unethical and impossible to conduct. At the individual level, it will be impossible to know whether the person would have developed psychosis even without drug intake.

One in four with SIP later receive a schizophrenia diagnosis

The knowledge that around one in four with SIP after some years are diagnosed with a schizophrenia-spectrum disorder, is relatively new. This was first described in 2007, when a 1-year follow-up study of 319 cases of SIP found a transition rate of 25% (Caton et al., Reference Caton, Hasin, Shrout, Drake, Dominguez, First and Schanzer2007). Some years later, in 2013, a large register-based study including more than 18.000 SIP cases found highest transition rate to schizophrenia-spectrum disorder for cannabis-induced psychosis (46%) and lowest for alcohol (5%), but without reporting a pooled transition rate for all SIPs together. Additional large, register-based studies from different countries followed from 2017 and onwards (Alderson et al., Reference Alderson, Semple, Blayney, Queirazza, Chekuri and Lawrie2017; Kendler et al., Reference Kendler, Ohlsson, Sundquist and Sundquist2019; Rognli, Heiberg, Jacobsen, Høye, & Bramness, Reference Rognli, Heiberg, Jacobsen, Høye and Bramness2023; Starzer, Nordentoft, & Hjorthoj, Reference Starzer, Nordentoft and Hjorthoj2018), and a meta-analysis in 2020 described a pooled transition rate of 25% for all SIPs together, and the highest for cannabis-induced psychosis (34%) (Murrie, Lappin, Large, & Sara, Reference Murrie, Lappin, Large and Sara2020). No other risk factors for schizophrenia have an effect size of comparable magnitude. People with SIP, particularly those associated with cannabis use, have comparable transition rates to schizophrenia as what we see for those with brief psychotic episode (ICD-10 F23) and atypical psychoses (ICD-10 F29) (Murrie et al., Reference Murrie, Lappin, Large and Sara2020). The understanding of the possible implications of these findings have shifted over time. While the authors of the first paper stated that ‘the distinction between a SIP and a primary psychotic disorder is important because these two disorders require fundamentally different approaches to treatment’ (Caton et al., Reference Caton, Hasin, Shrout, Drake, Dominguez, First and Schanzer2007), an editorial following the most recent transition paper (Rognli et al., Reference Rognli, Heiberg, Jacobsen, Høye and Bramness2023) emphasized the need for early intervention efforts for individuals experiencing SIP (Vassos, Reference Vassos2023).

Repeated episodes of SIP are associated with an increased risk of a diagnostic transition to schizophrenia (Crebbin, Mitford, Paxton, & Turkington, Reference Crebbin, Mitford, Paxton and Turkington2009; Kendler et al., Reference Kendler, Ohlsson, Sundquist and Sundquist2019; Rognli et al., Reference Rognli, Heiberg, Jacobsen, Høye and Bramness2023). The risk of transition is also increased with longer first admission for SIP, which probably is an indicator of the severity of the episode (Alderson et al., Reference Alderson, Semple, Blayney, Queirazza, Chekuri and Lawrie2017).

Figure 1 illustrates that the risk factors for psychosis in the landscape of drug use v. SIP v. transition to schizophrenia may influence either the precipitation of SIP (step 1) or the transition to schizophrenia (step 2). If these risk factors are more important in the first step, this indicates that SIP resembles more a schizophrenia spectrum disorder, while a greater influence on the transition step would indicate SIP not being so closely related to schizophrenia spectrum disorders. Only a few studies have investigated this. The study by Kendler and coworkers (Kendler et al., Reference Kendler, Ohlsson, Sundquist and Sundquist2019) and the study by Hjorthøj and coworkers (Hjorthøj et al., Reference Hjorthøj, Starzer, Benros and Nordentoft2020) investigated if known risk factors for schizophrenia also increase the risk for SIP (step 1) and the transition from SIP to schizophrenia (step 2). Hjorthøj found that serious infections are a risk factor SIP, but not for the transition from SIP to schizophrenia, providing evidence for step 1 (Hjorthøj et al., Reference Hjorthøj, Starzer, Benros and Nordentoft2020). This speaks for SIP belonging more to the schizophrenia spectrum. Kendler on the other hand found that familial vulnerability predicts both SIP, and the transition to schizophrenia (Kendler et al., Reference Kendler, Ohlsson, Sundquist and Sundquist2019), giving support to both steps. The family-based vulnerability was a greater risk factor for the transition, than for SIP, at a level we would find for people receiving a diagnosis of schizophrenia, without first having a SIP diagnosis.

Figure 1. The influence of vulnerability on the relationship between drug use, SIP, and schizophrenia. If the vulnerability influences the precipitation of SIP (step 1) rather than the transition to schizophrenia (step 2) this speaks for moving SIP closer to the primary psychosis realm.

How does this affect our understanding of SIP? For those who convert, the initial diagnostic assessment of SIP could of course have been wrong, and the correct understanding of the repeated episodes was that they were signs of an underlying chronic disorder (schizophrenia), with periodical (drug-induced) symptom exacerbations. Alternatively, something fundamental changed over time and with repeated admissions, causing a different disorder to occur. However, when asking if the SIP was erroneously diagnosed in the first place (if the ‘true’ diagnosis in fact would have been schizophrenia) or if the diagnosis changed from SIP to schizophrenia, it can be useful to remind ourselves that diagnostic cut-offs are constructed. The psychosis continuum perspective advocated by among others Jim van Os et al. describes how psychosis proneness over time may make psychotic symptoms become abnormally persistent and subsequently clinically relevant, depending among other things on the degree of environmental risk the person is additionally exposed to (van Os, Linscott, Myin-Germeys, Delespaul, & Krabbendam, Reference van Os, Linscott, Myin-Germeys, Delespaul and Krabbendam2009). This is in our view a third and more viable understanding, and one which is compatible with placing SIP in the psychosis continuum.

Clinical implications

There may be good reasons for being restrictive in diagnosing people with schizophrenia-spectrum disorder (Mitchell, Reference Mitchell2007). Clinicians may not want to give the patient too poor a prognosis and ‘label’ people with something that could be a chronic illness. However, for fear of falling into one ditch, there is a risk of falling into the opposite, namely that by holding back, sick people do not get the help they need. This applies to people with repeated SIPs without healthcare personnel being able to properly assess and treat the psychotic symptoms.

There are comprehensive clinical guidelines for the treatment of first-episode psychosis (FEP), and they recommend rapid and high-quality interventions in many areas, and long-term follow-up (Early Psychosis Guidelines Writing Group, 2016; International Early Psychosis Association Writing Group, 2005; NICE, Reference NICE2013). The brief interventions aim to reduce the duration of untreated psychosis, as this is considered one of the few potentially modifiable predictors of the outcome of schizophrenia (Malla, Reference Malla2022; Penttilä, Jääskeläinen, Hirvonen, Isohanni, & Miettunen, Reference Penttilä, Jääskeläinen, Hirvonen, Isohanni and Miettunen2014). However, SIP is not considered a FEP, and efforts to reduce duration of untreated psychosis seemingly do not apply to these patients. Most clinical guidelines on comorbid psychosis and substance use disorder prerequisite that the patient has an F2x-category diagnosis (NICE, Reference NICE2011, Reference NICE2013, 2016, 2020). To the extent that guidelines for SIP exist, the recommendations are mostly limited to stopping drug use. The message is also that differential diagnosis in relation to schizophrenia-spectrum psychosis is difficult, but that it is important to arrive at the correct diagnostic conclusion since the implications for further follow-up are great. This indicates that SIP is to be treated differently. In this way, a diagnosis of F1x.5 v. F2 seems to have great consequences. Our diagnostic practice may thus contribute to a poorer prognosis. This is also reflected in the recent review that states the striking paucity of information on outcomes, treatments, and best practices of SIP (Fiorentini et al., Reference Fiorentini, Cantù, Crisanti, Cereda, Oldani and Brambilla2021).

Dual diagnosis was first described in the 1980s, referring to individuals with coexisting severe mental illness and substance use disorders (Buckley, Reference Buckley2006; Drake, Mercer-McFadden, Mueser, McHugo, & Bond, Reference Drake, Mercer-McFadden, Mueser, McHugo and Bond1998). Is SIP in fact automatically a dual diagnosis as F1x.5 by necessity will include harmful use of a substance (F1x.1 refers to use that gives negative health consequences), and as psychosis is the negative health consequence? Or does one need something from both the F1 and F2 chapters to be considered having a dual diagnosis? And in that sense, does F1x.5 represent a move to avoid dual diagnosis? Some have called for better definition of dual disorders (Hryb, Kirkhart, & Talbert, Reference Hryb, Kirkhart and Talbert2007).

SIP is in most cases treated in the acute phase in mental health care. There are many indications that it would also be sensible to follow these patients in mental health care, as is the case for other types of psychotic disorders. In the event of an FEP, the patient is usually offered long-term follow-up, including pharmacotherapy, various forms of psychotherapy, family follow-up, as well as a focus on physical health, social life, housing, work, and collaboration with the municipality (Norwegian Directorate of Health, 2013). For SIP, our impression is that the typical treatment is a brief hospitalization at an acute psychiatric ward, and discharge with a request to refrain from substance use, perhaps accompanied by a referral to an outpatient clinic for substance use disorder treatment. Here we believe that the diagnosis may hamper proper follow-up. Research from the last decade could encourage us to, rather than viewing SIP as a specification of a substance use disorder, consider it as a type of psychotic disorder. Or, in other words, a SIP should be considered a FEP in an individual who uses substances. We believe that the evidence base indicates that SIP, at least those with recurring episodes, should be offered comparable treatment as other FEPs. It is noteworthy that in a recently published paper SIP is viewed as a sub-category of FEP, also because the disorders are undisguisable in a clinical setting (Inchausti, Gorostiza, Gonzalez Torres, & Oraa, Reference Inchausti, Gorostiza, Gonzalez Torres and Oraa2023).

Ideally, a sound clinical assessment should guide treatment, based on the patient's condition and needs, regardless of diagnostic codes. However, in a fragmented and specialized healthcare system, with scarce resources and a push towards rapid discharge, we know that diagnostic codes matter regarding what department is responsible for which patients. We are aware that in clinical practice some use a diagnosis of unspecified psychosis (ICD-10 F29) together with a diagnosis of harmful use of the substance (F1x.1) instead of a diagnosis of SIP (F1x.5), because this ensures better follow-up. This illustrates how diagnostic categories may serve as gatekeepers for treatment.

There are also important judicial aspects concerning the placing of SIP in the ICD-10 F1 or F2 chapters. In many jurisdictions people who are intoxicated are viewed as criminally sane, meaning that even if they are psychotic following drug use, they stand trial and serve time, different from people with schizophrenia spectrum disorder, who often are referred to treatment, not incarcerated. This has been problematized also by other research groups (Dominique, Reference Dominique2013; Liu et al., Reference Liu, Hao, Shi, Xue, Wang, Chen and Zhao2017).