The Tobacco and Alcohol Research Group blog

More about TARG

Recent Posts




Archive for September, 2015

Supervised injectable heroin for refractory heroin addiction

by Eleanor Kennedy @Nelllor_

This blog originally appeared on the Mental Elf site on 28th August 2015.

Opioid use is the number one reason for seeking substance misuse treatment across 30 European countries. Opioids are drugs derived from the opium poppy and these include the drug heroin (EMCDDA, 2015).

Heroin dependence has negative consequences for both the individual and society as persistent use of the drug is associated with poor health, criminal offences and damaged personal relationships (Ferri et al, 2011). Drug-free treatments and substitution treatments are the two interventions used to overcome heroin dependence.

Methadone is the most common substitution treatment in the EU, however, heroin prescribing is well established in Denmark, Germany and The Netherlands, an option in the UK and Spain, and currently under investigation in Belgium and Luxembourg (EMCDDA, 2015).

A recent systematic review and meta-analysis aims to compare supervised injectable heroin (SIH) as a treatment for heroin users who have not responded to more standard treatments such as methadone maintenance treatment (MMT) or residential rehabilitation (Strang et al, 2015).

NICE guidance recommends the use of methadone or buprenorphine as the first-line treatment in opioid detoxification.


Electronic databases (PubMed, Web of Science and Scopus) were searched for studies that reported on the effects of SIH treatment in participants with heroin-dependence unresponsive to standard treatments.

The studies had to have opiate use, retention in treatment, mortality and side-effects as outcome variables.

Studies were excluded if they were methodological papers, assessed unsupervised heroin treatment provision, focused on policy aspects, cost effectiveness, community perspectives or patient satisfaction.

The meta-analysis focussed on Mantel-Haenszel random effects pooled risk ratios for SIH treatment compared to the comparison groups.


There were a total of six papers included in the main review and meta-analysis. These studies were based in Switzerland, The Netherlands, Spain, Germany, Canada and England.

All studies explored SIH compared to MMT (oral methadone) in chronic heroin-dependent individuals who have repeatedly failed in orthodox treatment.

The results of rate of retention and the use of illicit heroin following treatment are shown in Table 1. The rates of retention varied across studies, with only one study reporting a lower rate of retention for the SIH group (Van den Brink et al, 2003). The statistical evidence indicated a lower rate of illicit heroin use in individuals receiving SIH treatment in all six studies.

Table 1: Retention in treatment and use of illicit heroin results

Study Retention in treatment Use of illicit heroin
Perneger et al, 1998 SIH 93% vs MMT 92% p = 0.002
Van den Brink et al, 2003 SIH 72% vs MMT 85% P = 0.002
March et al, 2006 SIH 74% vs MMT 68% P = 0.02
Haasen et al, 2007 SIH 67% vs MMT 40% P < 0.001
Oviedo-Joekes et al, 2009 SIH 88% vs MMT 54% P = 0.004
Strang et al, 2010 SIH 88% vs MMT 69% P < 0.0001


A meta-analysis was conducted to explore retention in treatment, mortality outcome and side-effects.

  • Retention in treatment was significantly better for the SIH than for the MMT treatment groups as demonstrated by four studies; RR = 1.37, 95% CI = 1.03 to 1.83
  • Mortality was lower in the SIH than in the MMT treatment groups but this was not significant; RR=0.65, 95% CI = 0.25 to 1.69
  • There was a higher risk of side effects in the SIH compared to the MMT treatment groups based on analysis of five studies; RR = 4.99, 95% CI = 1.66 to 14.99

This review provides good evidence that heroin-assisted treatment works for a small group of patients with refractory heroin dependence.

Strengths and limitations

All of the included studies were randomised controlled trials comparing traditional oral MMT to SIH in participants with chronic heroin-dependence who have not been successfully treated. The review followed PRISMA guidelines and was inclusive of all languages and publication dates, so the likelihood of important papers being excluded is minimal.

In this review the authors focussed on supervised administration of heroin only, which contrasts with a 2011 Cochrane Review that also included studies where heroin was prescribed for take-home administration (Ferri et al, 2011). By restricting the inclusion criteria, stronger conclusions can be made about the efficacy of this type of treatment which may guide the introduction of new interventions. Additionally the authors’ address several key misgivings about SIH, which further supports the argument that SIH is an effective treatment for treatment-resistant heroin dependence. For example, the concern that SIH may undermine other existing treatments is countered by the difficulty in recruitment experienced by many of the six trials under review.

There are some limitations, e.g. the safety of injectable diamorphine requires further research as the instances of sudden-onset respiratory depression is at a rate of about 1 in 6,000 injections.

The supervision and administration of SIH makes it more expensive than oral forms of opioid maintenance treatment.


The authors concluded that:

Based on the evidence that has been accumulated through these clinical trials, heroin-prescribing, as a part of highly regulated regimen, is a feasible and effective treatment for a particularly difficult-to-treat group of heroin dependent patients.

The importance of supervision during administration is emphasised throughout the review. As mentioned above, all of the participants engaged in SIH had previously repeatedly failed in orthodox treatment, however, the evidence supports SIH as a treatment option for these individuals.

Will this systematic review and meta-analysis be sufficient for policy makers to start recommending supervised injectable heroin for heroin users who have not responded to other standard treatments?


Primary paper

Strang J, Groshkova T, Uchtenhagen A. et al. (2015) Heroin on trial: systematic review and meta-analysis of randomised trials of diamorphine-prescribing as treatment for refractory heroin addictionBr. J. Psychiatry 2015;207:5-14. doi:10.1192/bjp.bp.114.149195.

Other references

EMCDDA (2015) European Monitoring Centre for Drugs and Drug Addiction. 2015. Available at:

Ferri M, Davoli M, Perucci CA. Heroin Maintenance for chronic heroin-dependent Individuals. Cochrane Database of Systematic Reviews 2011, Issue 12. Art. No .: CD003410. DOI: 10.1002 / 14651858.CD003410.pub4.

Van den Brink W, Hendriks VM, Blanken P, Koeter MWJ, van Zwieten BJ, van Ree JM. (2003) Medical prescription of heroin to treatment resistant heroin addicts: two randomised controlled trialsBMJ 2003;327(August):310. doi:10.1136/bmj.327.7410.310.

Perneger T V, Giner F, del Rio M, Mino A. (1998) Randomised trial of heroin maintenance programme for addicts who fail in conventional drug treatmentsBMJ 1998;317(July):13-18. doi:10.1136/bmj.317.7150.13.

March JC, Oviedo-Joekes E, Perea-Milla E, Carrasco F. (2006) Controlled trial of prescribed heroin in the treatment of opioid addiction. J. Subst. Abuse Treat. 2006;31:203-211. doi:10.1016/j.jsat.2006.04.007. [PubMed abstract]

Haasen C, Verthein U, Degkwitz P, Berger J, Krausz M, Naber D. (2007) Heroin-assisted treatment for opioid dependence: Randomised controlled  trialBr. J. Psychiatry 2007;191:55-62. doi:10.1192/bjp.bp.106.026112.

Oviedo-Joekes E, Brissette S, Marsh DC, et al. (2009) Diacetylmorphine versus methadone for the treatment of opioid addiction. N. Engl. J. Med. 2009;361:777-786. doi:10.1056/NEJMoa0810635. [Abstract]

Strang J, Metrebian N, Lintzeris N, et al. (2010) Supervised injectable heroin or injectable methadone versus optimised oral methadone as treatment for chronic heroin addicts in England after persistent failure in orthodox treatment (RIOTT): a randomised trial. Lancet 2010;375(9729):1885-1895. doi:10.1016/S0140-6736(10)60349-2. [Abstract] [Watch Prof John Strang talk about the RIOTT trial]

– See more at:

SMS texting to quit smoking: a meta-analysis of text messaging interventions for smoking cessation

by Olivia Maynard @OliviaMaynard17

This blog originally appeared on the Mental Elf site on 26th August 2015.

The efficacy of different smoking cessation interventions is always a hot topic around our woodland campfire. We’ve blogged previously about the effectiveness of both pharmacological and psychological treatments for smoking cessation, as well as their effectiveness among different populations of smokers.

A recent systematic review and meta-analysis investigated the efficacy of SMS text message interventions for smoking cessation. Unlike the majority of other smoking cessation interventions, using mobile phones to deliver health information allows for direct interaction between clients and practitioners without face-to-face interaction and permits the collection of large amounts of data. It is therefore cost-effective and easily scalable to large populations.

Previous meta-analyses have looked at the effectiveness of text messaging interventions for smoking cessation, but the review recently published by Spohr and colleagues is the first to investigate which elements or moderators of text message interventions are the most effective in supporting smoking cessation.

This review


The authors searched for randomised controlled trials which investigated the efficacy of text messaging interventions for smoking cessation. Only studies which included a follow-up measure of smoking abstinence were included. 13 articles met all of these inclusion criteria.

The authors also extracted information on the use of each of the moderators described below:

  • Intervention type:
    • SMS only;
    • ‘SMS plus’ (where SMS support is combined with either face-to-face or web-based support).
  • Message frequency:
    • Fixed message schedule (a consistent number of messages throughout the intervention);
    • Decreasing schedule (most messages at quit attempts, followed by a gradual reduction);
    • Dynamic schedule (depends on the stage of cessation the client is at).
  • Message track:
    • Fixed message track (users cannot influence the course of the intervention);
    • Dynamic message track (user quit status and stage of change can influence intervention messages).
  • Message tailoring:
    • Tailored messages (customised message content to a specific individual);
    • Targeted messages (customised messages to a population subgroup).
  • On-demand messaging:
    • On-demand messaging (allow users to text a keyword in emergency situations to receive additional support. Some interventions allow users to connect with other users for support and encouragement).
  • Message direction:
    • Unidirectional messaging (by the researcher);
    • Bidirectional messaging (to obtain data from the client).
  • Message interaction:
    • Researcher-initiated (containing intervention messages and assessment questions);
    • User-initiated (containing requests for additional support).

Intervention success was assessed using seven-day point prevalence as the primary outcome measure, as 11/13 of the studies reported these results. Two other studies only reported 6 month continuous abstinence.

The researchers used an intention to treat analysis.

Perhaps surprisingly (given the ubiquitous nature of smartphones) the reviewers only found 13 trials to include in their analysis.


The 13 articles resulted in a cumulative sample size of n = 13,626. Participants were primarily adult smokers (six studies), but four studies recruited participants aged 15 or over and three targeted adolescents and young adults (ages 16-25).

Smoking quit rates for the text messaging intervention groups were 35% higher as compared to control groups (OR = 1.35, 95% CI = 1.23 to 1.49).

Overall, the analysis of the intervention moderators did not find strong evidence that any particular moderator was more effective than any other:

  • Intervention type:
    • There were no differences (Q= 0.56, df = 1, p = 0.46) in intervention efficacy between those which provided text-only support (= 6) as compared with text messaging plus additional support (k = 7).
    • However, text-only interventions had a slightly larger effect size than those with text messaging plus additional support.
    • There were no differences (Q= 0.89, df = 1, p = 0.35) in intervention efficacy between those which promoted the use of nicotine replacement therapy (NRT) (= 7) as compared with those which did not (k = 6).
  • Message frequency:
    • There were no differences (Q= 0.96, df = 2, p = 0.62) in intervention efficacy between those which provided decreasing schedule (= 8) as compared with fixed schedule (k = 3) or variable schedule (= 2) support.
    • However, those which had a fixed schedule had larger effect sizes than either of the other types of schedules.
  • Message track:
    • There were no differences (Q= 0.38, df = 1, p = 0.54) in intervention efficacy between those which used a fixed message track (= 5) as compared with a dynamic message track (k = 8).
  • Message tailoring:
    • There were no differences (Q= 1.54, df = 2 p = 0.46) in intervention efficacy between those which used message tailoring (= 8) as compared message targeting (k = 1) or a combination of both (k = 4).
    • All studies included some form of message content tailoring.
  • On-demand messaging:
    • There were no differences (Q= 0.15, df = 1, p = 0.70) in intervention efficacy between those which used on-demand messaging (= 11) as compared with those which did not (k = 2).
    • There were no differences (Q= 0.02, df = 1, p = 0.88) in intervention efficacy between those which provided peer-to-peer support (= 5) as compared with those which did not (k = 8).
  • Message direction:
    • All of the studies used bidirectional messaging so the effectiveness of this moderator to unidirectional messaging could not be assessed.
  • Message interaction:
    • There were no differences (Q= 0.17, df = 1, p = 0.68) in intervention efficacy between those which included assessment messages (= 7) as compared with those which did not (k = 6).

Text messaging does not compare well to many other more effective methods of smoking cessation.


This meta-analysis found that smoking cessation interventions which used text-messaging increased the odds of successfully quitting smoking by 35%.

To put this in perspective, other reviews have found that telephone quit lines increase smoking cessation success by 60%, social support increases success by 30% and practical counselling by 50%. NRT and other medications have been shown to increase cessation success by between 50% and 310% (Fiore et al., 2000).

None of the moderators investigated here were found to be more effective than any other. There was some evidence that interventions which used fixed schedules were more effective than those which used either decreasing or variable schedules. Similarly, there was some evidence that text-only support programs were more effective than those which provided a ‘text-plus’ service. However, there was no robust statistical evidence for these differences.

Overall, these results provide no evidence that text-messaging interventions, which are more complex and time-demanding (i.e. text-plus, on-demand messaging, variable schedules, social support communication), are any more effective than the simplest interventions.

However, given the cost-effectiveness, relative ease of delivery and promise of efficacy of these interventions, future research should continue to determine what moderators make an effective text-based intervention.

Limitations and future directions

These results should be treated with caution however, for a number of reasons:

  • The authors relied on data obtained from the original articles to compile this meta-analysis, rather than contacting the researchers themselves. As data regarding the actual use by users of some of the moderators such as social support and on-demand messaging was not reported in articles, we cannot be certain whether the failure of these moderators to increase quit success is because they are simply not more effective, or because users didn’t actually use these services.
  • The number of studies included in this meta-analysis was small (only 13 studies). This is even more the case for the moderator analyses. Drawing firm conclusions from the statistical evidence is therefore difficult.



Primary paper

Spohr SA, Nandy R, Gandhiraj D, Vemulapalli A, Anne S, Walters ST. (2015) Efficacy of SMS Text Message Interventions for Smoking Cessation: A Meta-Analysis. Journal of Substance Abuse Treatment, 56, 1-10. doi:

Other references

Fiore MC, Bailey WC, Cohen SJ, Dorfman SF, Goldstein MG, Gritz ER, … Lando HA. (2000) Treating tobacco use and dependence: a clinical practice guideline: Publications Clearinghouse.

– See more at:

A green man walked into a bar to the future…

By Jasmine Khouja (@Jasmine_Khouja), with contributions from Olivia Maynard, Suzi Gage, Gibran Hemani and David Troy

Alcohol and cigarettes may not seem out of place at a music festival; discussing the science behind alcohol, cigarettes and genetics may. This August, my colleagues and I from the MRC Integrative Epidemiology Unit (IEU) and Tobacco and Alcohol Research Group (TARG) at the University of Bristol took our research to the Green Man Festival 2015 ( which, though primarily a music festival, hosts the Einstein’s Garden where festival-goers can play, learn and converse about science. ‘Future’ was the theme so we took our ‘Bar to the Future’ to show where our research may lead.


Since the smoking ban, bars have been smoke-free but the growing popularity of e-cigarettes has sparked debate about their use indoors. Using an attention-grabbing demonstration (pictured right), we asked the public’s opinions on smoking and vaping. Our demonstration, in which cigarettes and e-cigarettes were smoked/vaped into the two separate glass tubes by means of a battery powered bed pump, shows a brown residue (tar) in the conventional cigarette tube where the e-cigarette tube is clear. Tar is not present in e-cigarettes making them, in one way, less harmful than traditional cigarettes.


bartothefutureMembers of the public shared their mixed opinions with us (pictured right) ranging from extremely positive to extremely negative.  Many seemed concerned about the long term health effects of using e-cigarettes. Scientists are unsure of the long term effects of e-cigarette use because they have not existed long enough for these effects to be assessed and the rapidly changing devices make it hard for scientists to keep up. The wide spectrum of beliefs could reflect the inconsistency of information the public receive from the media; news sources reported Public Health England’s recent suggestion that e-cigarettes are 95% safer than conventional cigarettes and may eventually be prescribed on the NHS ( which received backlash in a Lancet Journal editorial which also received media coverage ( Opinion on conventional cigarettes however, was less divided. The public are clearly aware of the dangers despite many choosing to continue smoking. Opinions on vaping indoors showed a similar pattern to the views on e-cigarettes harm; worries over normalising smoking again and second hand vape were among the concerns despite no evidence suggesting inhaling second hand vape is dangerous. The exhaled vapour usually contains traces of nicotine and flavouring as well as an FDA approved substance used in most fog machines (propylene glycol). Though positive about the use of e-cigarettes as a smoking cessation aid as they contain fewer and lower doses of toxicants, they do still contain toxicants so we would not describe them as safe for non-smokers but they are safer than conventional smoking.


The children of Green Man got involved by designing cigarette packaging warning labels. Here are a few of the designs placed on our giant cigarette packet which could show the future of cigarette health warnings.


Calorie information is now placed on food and soft drinks bought at supermarkets but rarely placed on the packaging of alcoholic drinks. Calls from public health officials and policy makers could see calorie content on labelling made mandatory. By asking the public to guess how many calories were in lager, whiskey, alcopops and wine it became apparent that the public have limited knowledge when it comes to the calorific content of alcohol. This is unsurprising as calorie content varies across brands, strength and size of alcoholic drink. Without the information being provided, it is difficult to know how many calories are in your drink (some information can be found on the drink aware website,

Straight glasses may be used more in bars rather than curved glasses in the future. To demonstrate the effect of this we asked festival-goers to half fill a curved and a straight glass with water. The majority of festival-goers struggled to find the half-way point on the curved glass yet found it relatively easily on the straight glass. This finding has been shown in the lab too; people perceive the half-way point to be lower than it is on a curved glass and consume a half pint 4 minutes faster in a curved glass than from a straight glass and drink 1 minute slower from glasses marked with volume information displaying where the ½, ¼ and ¾ points are. By simply adding volume information to glasses and using straight instead of curved glasses the public may reduce their drinking speed meaning they drink less over the course of their drinking session.


snakesWe also shared some future research which involves the Rotating snakes illusion (pictured right). When viewing this illusion the motion perception areas in the brain are activated meaning the viewer perceives the snakes as rotating. We hypothesised that if festival-goers had consumed alcohol they would see less or no rotation due to their motion perception being impaired. Though we did not observe this, there was a lot of variation in quickly the public saw the snakes rotate even if they hadn’t consumed any alcohol. This information will be useful when designing our future lab studies so that each participant is tested with and without having consumed alcohol rather than comparing alcohol consumers to non-alcohol consumers.


A popular part of our stall was the genetics of table football. This was no ordinary game of table football, each team was made up of a genetics and an environmental player to demonstrate how our genes and environment affect the person we are in the future. The genetics players rolled a dice to decide their genetic predispositions (e.g. to alcohol dependence) which represented the element of chance in who we are. This related to an advantage or disadvantage in the game (e.g. holding a cup while playing). The environment player then picked a card (representing the element of choice in our environment) giving the player an advantage or disadvantage. The players then battled it out. The team who scored the first goal got to pick a controlled environment card; they could either lose one of their disadvantages (e.g. put the cup down) or choose to disadvantage the other team before continuing play. The game sparked discussions on how neither your genes or environment solely determine who you are, it is a team effort.

Festival-goers also tried PTC (Phenylthiocarbamide), which has a rare property; 70% of people experience a bitter taste but 30% of people taste virtually nothing when they try this chemical due to their genetics. After asking festival-goers to lick a PTC strip they were asked if they liked four bitter foods. We expected to find that those who can taste PTC are more sensitive to bitter tastes and therefore like bitter tasting foods less than those who cannot taste PTC. Information like this could be used in the future to help people decide what alcoholic drinks may taste better to them. We found that PTC tasters were as likely to like bitter tasting foods as those who can’t taste PTC.

stallThe punch line

After three days of games and discussions with the public, our gazebo proved no match for the Welsh weather and was left broken beyond repair after heavy rainfall. We were forced to shut down the stall a day early but for those who missed out, we hope to see you next year!


Does tobacco cause psychosis?

by Marcus Munafo @MarcusMunafo

This blog originally appeared on the Mental Elf site on 30th July 2015.

Hot on the heels of a recent study suggesting a dose-response relationship between tobacco smoking and subsequent risk of psychosis, a systematic review and meta-analysis (including the data from that prospective study) has now been published, again suggesting that we should be considering the possibility that smoking is a causal risk factor for schizophrenia.

As I outlined in my earlier post, smoking and psychotic illness (e.g., schizophrenia) are highly comorbid, and smoking accounts for much of the reduced life expectancy of people with a diagnosis of schizophrenia. For the most part, it has been assumed that smoking is a form of self-medication, to either alleviate symptoms or help with the side effects of antipsychotic medication.

It's widely thought that people with psychosis or schizophrenia use smoking as a way to self-medicate and relieve their symptoms.


This new study reports the results of a systematic review and meta-analysis of prospective, case-control and cross-sectional studies. The authors hoped to test four hypotheses:

  1. That an excess of tobacco use is already present in people presenting with their first episode of psychosis
  1. That daily tobacco use is associated with an increased risk of subsequent psychotic disorder
  1. That daily tobacco use is associated with an earlier age at onset of psychotic illness
  1. That an earlier age at initiation of smoking is associated with an increased risk of psychotic disorder

The authors followed MOOSE and PRISMA guidelines for the conduct and reporting of systematic reviews and meta-analyses, and searched Embase, Medline and PsycINFO for relevant studies. They included studies that used ICD or DSM criteria for psychotic disorders (including schizophrenia, schizophreniform disorder, schizoaffective disorder, delusional disorder, non-affective psychotic disorder, atypical psychosis, psychotic depression, and bipolar mania with psychotic features).

To test the first hypothesis, studies with a control group were used to calculate an odds ratio. To test the second, prospective studies in which rates of smoking were reported for patients who developed psychotic disorders compared to controls were included, so risk ratios could be calculated. To test the third and fourth, prospective and case-control studies were included, and for the onset of psychosis, cross-sectional studies were also included.

Effect size estimates (weighted mean difference for continuous data, and odds ratios for cross-sectional data or relative risks for prospective data) were combined in a random-effects meta-analysis.


A total of 61 studies comprising 72 independent samples were analysed. The overall sample included 14,555 tobacco users and 273,162 non-users.

  1. The overall prevalence of smoking in people presenting with their first episode of psychosis was higher than controls (12 case-control samples, odds ratio 3.22, 95% CI 1.63 to 6.33, P = 0.001). This supports hypothesis 1.
  2. Compared with non-smokers, the incidence of new psychotic disorders was higher overall (6 longitudinal prospective samples, risk ratio 2.18, 95% CI 1.23 to 3.85, P = 0.007). This supports hypothesis 2.
  3. Daily smokers developed psychotic illness at an earlier age compared with non-smokers (26 samples, weighted mean difference -1.04 years, 95% CI -1.82 to -0.26, P = 0.009). This supports hypothesis 3.
  4. Age at initiation of smoking cigarettes did not differ between patients with psychosis and controls (15 samples, weighted mean difference -0.44 years, 95% CI 1-.21 to 0.34, P = 0.270). This does not support hypothesis 4.

Daily tobacco use is associated with an increased risk of psychosis and an earlier age at onset of psychotic illness.


The authors conclude that the results of their systematic review and meta-analysis show that daily tobacco use is associated with an increased risk of psychotic disorder and an earlier age at onset of psychotic illness, although the magnitude of the association is relatively small.

Interestingly, the authors interpret their results in the context of the Bradford Hill criteria for inferring causality (which consider the strength, consistency, specificity, temporality, biological gradient, plausibility, coherence, experiment, and analogy of an association). They argue that, where these criteria can be applied (the specificity criterion cannot be applied because smoking affects so many disease processes, while the experiment criterion is not met because animal models of psychotic illness that capture important features such as delusions are impossible), they do appear to be met by the evidence available.


There are a number of important limitations to this study, which the authors themselves acknowledge:

  • The first is that all analyses relied on observational data, which makes strong causal inference impossible. Longitudinal prospective studies help somewhat in this respect, but only a small number were available for inclusion in the analysis of risk of developing psychosis between smokers and non-smokers. Moreover, even these studies cannot exclude the possibility that symptoms present before a first full episode of psychosis may have led to smoking initiation (i.e., self medication).
  • Another important limitation is that very few studies measured or adjusted for use of other substances (most importantly, perhaps, cannabis, which has been widely discussed as a potential risk factor for schizophrenia). This is a potentially very important source of bias.

Nevertheless, this is a well-conducted systematic review and meta-analysis that brings together a reasonably large literature. The results appear robust, although given the observational nature of the data, and the fact that only data that were comparable across studies could be meta-analysed, any conclusions regarding causality need to be very tentative.

Very few studies in this review, measured or adjusted for use of other substances such as cannabis.


It seems that we should seriously consider the possibility that smoking is a causal risk factor for schizophrenia. Of course, the data available to date aren’t definitive, and we need to be very cautious about inferring causality from observational data, but this does feel like an area where there is growing, converging evidence from multiple studies using multiple methods.

It’s also worth bearing in mind that even if smoking is a causal risk factor, this does not preclude the possibility that smoking is also used as a form of self-medication. There are several thousand constituents of tobacco smoke; it is possible that some of these alleviate symptoms, while others exacerbate them. For this reason, we shouldn’t assume that nicotine is necessarily the culprit if smoking is indeed a causal risk factor; it may be (and Gurillo and colleagues discuss the biological plausibility of nicotine in this context), but that will need to be tested.

This last point is particularly important in the content of ongoing debate regarding the potential harms and benefits of electronic cigarettes. If smoking does turn out to be a causal risk factor for schizophrenia, then whether nicotine or something else in tobacco smoke is identified as the culprit will have an important bearing on this debate, and attitudes towards these products.

There are several thousands constituents of tobacco smoke; it is possible that some of these alleviate symptoms, while others exacerbate them


Primary paper

Gurillo P, Jauhar S, Murray RM, MacCabe J. (2015) Does tobacco use cause psychosis? Systematic review and meta-analysis. Lancet Psychiatry 2015. doi: 10.1016/S2215-0366(15)00152-2 (Open access paper: features audio interview with authors)

Munafo M. Smoking and risk of schizophrenia: new study finds a dose-response relationship. The Mental Elf, 1 Jul 2015.

– See more at: