Contrary to our hypothesis, the additional cue exposure component in virtual reality did not increase abstinence rates compared to nonspecific relaxation training supplemented with a validated smoking cessation program. This needs to be discussed in light of evidence of the prognostic value of craving at baseline for abstinence as the primary outcome in the VR-CET group: lower maximum craving predicted a positive therapeutic outcome. One implication of this result is that smokers experiencing high levels of cue reactivity (CR) at baseline may not have received enough cue exposure sessions. Other authors with four sessions (29, 30) did not find significant craving reductions in smokers either. While positive effects of VR-CET could be found, the corresponding studies included ten (31) sessions in smokers or eight VR-CET sessions in patients with alcohol dependency (32). Our own data indicate that four sessions of CET can be enough for smokers with low levels of baseline craving but should be increased for smokers with high baseline craving. It is important to consider that the severity of nicotine dependence (FTND) was also involved in the regression but did not predict abstinence, indicating that baseline craving is an independent measure. There is also previous evidence that craving during withdrawal and smoking severity predict relapse (33). In our data, baseline craving ratings were independent of withdrawal, as smokers in our RCT smoked as usual while smoking cessation was being prepared in group-based cognitive behavioral therapy in sessions one and two, with smoking cessation being initiated between weeks two and three. In a post hoc analysis, a therapy-adherent subsample was defined by CO values below 10 in the measurements during group therapy sessions 3–6. According to our design, every smoker assigned to the VR group took part in the CET for relapse prevention, irrespective of the therapy adherence defined by smoking cessation. However, the predictive value of maximum craving even increased in this subsample, indicating that the effect we found was independent of withdrawal. However, this points to the importance of individual interventions that should be considered even in RCTs. A possible approach is to work individually with smokers who have not yet reached abstinence before proceeding with the assigned relapse prevention interventions, along with the cost of reduced standardization.
The German treatment guidelines recommend that “behavioral treatments to support tobacco abstinence should include several components (especially psychoeducation, motivation reinforcement, measures for short-term relapse prevention, interventions to strengthen self-efficacy, everyday practical counseling with concrete behavioral instructions, and practical coping strategies (problem-solving and skills training, stress management))” 2. Cue exposure is not recommended in these guidelines and is not carried out systematically. The current trial, therefore, aimed to investigate the efficacy of a low-dose cue-exposure treatment as an adjunct to guideline-based tobacco cessation treatment and revealed the necessity of assessing the prerequisites for the intervention and allowing an individualized decision for its adaptation. This study examined the efficacy of VR-CET in comparison to PMR, both as a supplement to a cognitive behavioral smoking cessation program that has already been published and proven to be effective (22, 23), with high standards related to RCTs. The randomized trial included a representative sample of smokers with characteristics typical of such a study design: more male smokers, a mean FTND score of approximately 4.5 points, an average daily consumption of 20 cigarettes, and a high proportion of participants (84%) with at least one previous quit attempt. The primary outcome was abstinence according to the "Russell Standard" definition(24). The average achieved abstinence rates were good compared to the average long-term success rate stated in the S3 treatment guidelines, taking into account the absence of concomitant medication. Other recent studies on smoking cessation, such as studies on hypnotherapy (34) or mindfulness-based procedures (35), have reported similar abstinence rates of up to 20–25%. Adding further treatment components typically has not been proven to enhance effective CBT protocols in other mental disorders. Based on our data, a possible explanation for the insufficient efficacy of cognitive behavioral interventions in individuals with addiction could be the lack of individualization of the interventions. This affects not only the number of sessions but also the time course of application.
A future approach for the treatment of addiction is to consider motivational consumption states, tailoring the assignment to an investigated intervention more individually. This approach is promising for demonstrating an additional effect of the applied treatment component that our study could not show for our primary outcome. Likewise, for the secondary measures, there was no specific effect of the additional treatment. Nevertheless, the overall number of smoked cigarettes and the desire to smoke decreased, while an increase in self-efficacy increased. Fluctuations throughout the observed period still show improvements at the end compared to baseline measures, an effect that can essentially be explained by the successfully treated individuals. Another explanation for the lack of effect of VR-CET compared to nonspecific treatment is that the heavy smokers did not receive any medication, which is suggested in the guidelines indicating/recommending combination therapy. Without medication, heavy smokers in particular are likely to experience more withdrawal symptoms. Since the VR training started one week after smoking cessation, these withdrawal symptoms might have affected the effectiveness of the cue exposure treatment. Again, an individual approach for an RCT could overcome such limitations.
Based on principal learning theories, smokers who experienced intense craving were expected to profit from VR-CET, which was not confirmed by our data. We already discussed the low number of repetitions of VR-CET sessions and the lack of withdrawal medication to be related to this lack of expected effects. Compared to other fields where exposure treatment is applied, e.g., patients with posttraumatic stress disorder (PTSD), there are additional instructions to apply emotion regulation skills during exposure (36). Therefore, a possible shortcoming of our VR-CET could be the lack of available skills to handle the CR-evoking scenario. In future studies, inhibitory control during CET could be supported by noninvasive brain stimulation, an approach that has already been tested during exposure in patients with PTSD (37). Furthermore, another enhancement of the VR-CET could be the application of emotion regulation strategies for craving during sessions. Another approach is the use of biofeedback to regulate the intensity of VR-CET, as the physiological response is another important level of CR that is not completely captured by craving (38).
Furthermore, the strong control group needs to be discussed as a possible contributor to the unexpected effects. In the case of PMR, we used a control intervention without known specific effectiveness in smoking cessation, yet a subgroup of smokers may have benefited from this intervention as an aid in coping with relapse-prone situations. Involving more specific characteristics than the FTND, e.g., impulsivity, depression, and sensation seeking(39), could provide more features for the decision to apply the respective additional therapy.
Last but not least, the study was influenced by the COVID-19 pandemic (2020/2021): Compared to previous smoking cessation studies by our working group (23, 39), the follow-up response rates were significantly lower. In the course of the conservative ITT, these participants were considered relapsing smokers. From March 2020 onward, both the smoking behavior of the participants and the technical processes of the study, such as recruitment and contacting the participants, were affected by the pandemic situation. International data on smoking during the pandemic were mixed(40), while German data (DEBRA study) show a surprising increase in smoking behavior and a decrease in smoking cessation success during the pandemic, despite the known effects of smoking on lung damage (https://www.debra-study.info/). There is evidence that relapse rates are particularly high among smokers with high stress levels34. In our sample, we found elevated anxiety levels among participants who had quit smoking, which may have been related to increased perceptions of stress.
Furthermore, the rapid development of VR technology must be considered. Realistic scenarios are the goal for increasing the efficacy of VR-CETs; therefore, it needs to be considered whether technical improvement in quality could increase their efficacy. Subjective ratings of the four scenarios are reported elsewhere (Schröder et al., submitted), and increases in subjective presence could be another future direction for improving CET efficacy.
Despite these limitations, the high methodological standards of our trial were comparable to those of pharmacological studies in terms of external monitoring and data management, a safety assessment, predefined analysis plans and blinded follow-ups. Four VR scenarios were established and tested in this large study sample, where acceptance of the VR-CET among smokers was shown.
In conclusion, we were able to demonstrate the relevance of CR to smoking cessation, but the specific parameters of our CET – such as duration, character and intensity – may not have been adequate to sufficiently change CR, e.g., an insufficient number of training repetitions. Future directions in VR incorporate further sensory modalities such as olfactory or haptic information that need to be investigated regarding a potential increase in the efficacy of cue exposure. In our scenarios, smelling or even drinking coffee or alcohol could intensify the scenarios. Furthermore, the presentation of additional high-risk situations (e.g., an emotional conflict to experimentally induce rumination) in VR could be used to broaden the scope of processes and situations strongly related to relapse that are addressed in VR-CETs.
There are other important parameters that need to be considered in future studies, including the number and frequency of cue exposure sessions and longer or more adaptive protocols that include personal high-risk smoking situations, internal or emotional cues or craving-adapted session protocols. Furthermore, an individualized approach based on baseline CR, self-efficacy, and coping strategies might significantly improve VR-CET.