Craving self‐reports as outcome measures in drug addiction trials: A systematic review of ClinicalTrials.gov

Abstract

Background and Aims

The subjective experience of drug craving is characterized by an overwhelming urge to consume substances. Due to strong validity and ease of use, self‐report measures are widely employed to assess substance‐related motivational dynamics. Multi‐item questionnaires are increasingly favored for capturing the multidimensional nature of craving, providing valuable insights for clinical care and relapse prediction. This review aimed to summarize craving self‐report measurement tools registered in clinical trials and examine their methodological parameters.

Methods

A search was conducted in November 2022 and updated in November 2024 using the same methodology on ClinicalTrials.gov for trials reporting drug craving as an outcome measure. Trials using craving measurement tools were screened and included.

Results

From 5383 initial trials, 1255 met eligibility criteria. Craving was reported as the only primary outcome measure in 14.6%, one of the primary outcomes in 21.3% and as secondary or exploratory in 64.1% of the studies. The most studied substances were nicotine (32.5% of studies) and alcohol (27.5%), followed by opioids (14.2%). Multi‐item questionnaires were the most frequently used approach (55.4%), followed by single‐item questionnaires (27.3%) to measure craving. Among 107 identified multi‐item questionnaires, only 38 were used three or more times. The most common multi‐item questionnaires were the Questionnaire on Smoking Urges (20%), Penn Alcohol Craving Scale (12.1%) and Alcohol Urge Questionnaire (9.8%). Most trials focused on evaluating phasic (now) craving (51.6%) over tonic (in a certain time‐interval) craving (38%).

Conclusion

Craving, measured through self‐reports, is increasingly targeted as a primary outcome measure in drug addiction trials registered on ClinicalTrials.gov. Craving self‐reports are highly variable, underscoring the need for standardized tools to enhance comparability across studies.

INTRODUCTION

Substance use disorders (SUDs) are chronic and complex mental health conditions characterized by drug‐seeking behaviors despite adverse psychological, physical or social consequences [1]. Craving has long been recognized as a central component of motivational drive in SUDs and a primary cause of drug‐taking behavior and relapse [2, 3, 4]. There has been substantial discussion surrounding the definition and measurement of craving over the past half‐century [5, 6]. Most researchers define craving as an overwhelming subjective experience of the urge/desire to consume substances, although it involves a plethora of cognitive processes [6, 7, 8].

Current models of the underlying mechanisms of craving broadly fall under two etiological classes [7]: (a) those with a conditioning mechanism and (b) those with a cognitive mechanism [2, 9, 10, 11]. Conditioning models propose that craving is a conditioned response elicited by stimuli associated with either positive (euphoria) or negative (withdrawal symptoms) reinforcement [2, 9]. Cognitive models [12] describe craving as the result of effortful processing that occurs after the disruption of ‘automatic’ drug‐seeking behavior or as the elaboration of transient, intrusive drug‐related thoughts triggered by external or internal cues [9, 10].

The specific operationalization of craving by different instruments reflects explicit or implicit assumptions about the nature of drug craving and its role in addiction‐related processes [6]. There is a growing body of evidence supporting the evaluation of the subjective experience of drug craving with self‐reports to assess the impact of interventions, inform clinical care and help predict the risk of relapse in individuals with SUDs [2, 6, 13, 14]. The association between craving and relapse is significantly influenced by underlying anxiety or depressive symptoms. Intense craving episodes, often driven by negative emotional states, substantially increase the risk of relapse. Individuals with depression and mood/anxiety disorders demonstrate increased reactivity to drug‐related cues, reinforcing the craving relapse cycle. These dynamics highlight the predictive value of craving intensity for subsequent substance use, as consistently observed in clinical and laboratory studies [15].

One simple tool for assessing craving is a single‐item visual analogue scale (VAS), which is designed to capture intensity of craving for the target substance at a given moment [10]. Acknowledging that craving is a multi‐dimensional experience [16] and may not be effectively measured by a single‐item questionnaire, a variety of self‐report assessment tools have been developed to measure craving across SUDs [10]. Multi‐item self‐report questionnaires remain the most commonly used method for evaluating craving, because they comprehensively address a diverse range of content, including items about the intentions to use drugs, expectations of specific outcomes from use and perceived lack of control over drug use [17, 18, 19].

Despite extensive studies on craving in the field of addiction and multiple theoretical models aimed at understanding its underlying mechanisms, there is no clear theoretical framework for selecting craving measurement tools in the literature [13, 20]. In a 2009 survey conducted by Pavlick et al., [10] it was found that almost half of United States (US) substance abuse treatment agencies addressed craving in their treatments. However, a negligible proportion (5%) used psychometrically validated self‐report measures, with the majority (86%) using open‐ended questions despite the availability of multiple published self‐report instruments.

Although this survey has potential datedness, its findings echo persistent challenges highlighted in subsequent research. Recent reviews focusing on measurement tools for specific substances (e.g. opioids) similarly reveal significant variability in structure, content and psychometric validation, underscoring the difficulties in standardizing assessments [17, 21]. Collectively, these findings demonstrate that, despite advances in understanding the psychology and neurobiology of craving, the development of numerous heterogeneous multi‐item questionnaires has resulted in a lack of harmony across assessments, further complicating their use and comparison in research and clinical settings [8, 17, 21]. The absence of standardization in craving measurement leads to inconsistent and idiosyncratic approaches, which significantly impede scientific research and hinder the accurate assessment of craving [10, 17]. While this heterogeneity is partly warranted and because of differences in study context and aims, it makes it challenging to identify, which measures effectively capture craving levels and to determine the efficacy of specific interventions. International guidelines, such as those from the European Medical Association (EMA), highlight the importance of using validated tools like the Obsessive‐Compulsive Drinking Scale, underscoring the potential for more consistent practices in future research. The present review provides an overview of existing craving measurement tools registered in the ClinicalTrials.gov database to systematically explore different types of craving assessments, identify the most commonly used ones and analyze their time frames and constructs.

METHODS

This systematic review was pre‐registered on the Open Science Framework (OSF) (https://osf.io/vmcf7/) and conducted according to the Preferred Reporting Items for Systematic Reviews and Meta‐Analysis (PRISMA) standards [22] (Figure 1). A comprehensive search for clinical trials involving assessment tools measuring craving in SUDs was conducted on 5 November 2022, and updated on 17 November 2024, using the ClinicalTrials.gov database of the US National Library of Medicine (https://www.clinicaltrials.gov/) [23]. Both searches followed the same methodology, including identical search strategy, inclusion/exclusion criteria and data extraction processes.

FIGURE 1.

PRISMA flowchart for craving assessment tools as outcome measures in drug addiction trials registered in the ClinicalTrials.Gov. The PRISMA flow chart illustrates the search strategy and screening process used to identify relevant trials entries from ClinicalTrials.gov database, specifically focusing on trials where craving assessment tools were used as an outcome measure in drug addiction trials. The flowchart displays the number of trials identified at each stage of the process: identification: total number of trials identified through database searching; screening: number of trials screened for eligibility based on predefined inclusion/exclusion criteria; eligibility: number of trials assessed for eligibility after screening; and included: number of trials ultimately included in the systematic review and analysis. In addition to standard PRISMA stages, the final included trials were further catgorized by substance. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta‐Analyses.

Search strategy and screening

ClinicalTrials.gov was searched to identify trials reporting craving as an outcome measure in SUDs from inception until 5 November 2024, using the terms ‘craving’, ‘urge’ and ‘desire’. Initially, trials were screened by title and study overview (conditions of the trials), followed by status screening (i.e. ‘completed, recruiting or terminated’). Full‐text reviews of the identified trials were conducted in parallel by two independent reviewers.

Trials were eligible for inclusion if they (a) reported ‘craving’, ‘desire’ or ‘urge’ as an outcome measure, and (b) included individuals with SUDs, such as alcohol and nicotine dependence, in either inpatient or outpatient settings. Trials were excluded if they (a) involved individuals with behavioral addictions (e.g. food, gambling, internet, video games and sex) or (b) had statuses of suspended, terminated, withdrawn or unknown.

A standardized, pilot‐tested data extraction sheet was used to extract data from the included trials. A Python script was used to import the records from ClinicalTrials.gov (the code is publicly available on the OSF page of the project (https://osf.io/vmcf7/).

Two independent reviewers, M.T. and K.G., conducted the initial selection by screening titles and study conditions. After preliminary screening, the reviewers discussed any differences in the selected trials. Disagreements were resolved through group discussion, with two additional reviewers, P.R. and N.N., serving as arbiters when needed.

Data extraction

The data extraction table comprised four main categories including (a) participant characteristics (age range/mean and gender); (b) general characteristics of the trials (start date, status, outcome‐measure type—i.e. primary, secondary, or other pre‐specified—and location); (c) design characteristics of the trials (i.e. enrollment, phases, design, type of intervention and substance of use); and (d) assessment characteristics [assessment type— i.e. single or multi‐item; assessment title—i.e. Cocaine Craving Questionnaire (CCQ); version—i.e. brief or modified; construct model—i.e. single dimension or multi‐dimension; time frame—i.e. now, past 24 hours].

It should be noted that a single trial may include several outcome measures (i.e. both primary and secondary), so the number of trials may differ from the number of outcome measures. Because general and design characteristics do not change between outcome measures, data related to these first two categories were reviewed on a per‐trial basis. Conversely, assessment characteristics were reviewed on a per‐measure basis, as a single trial might use different assessments for primary and secondary outcome measurement. Because of the heterogeneity of craving multi‐item surveys and the use of unspecified self‐reports in some trials, only multi‐item questionnaires used more than three times in the database were included in the analysis. Questionnaires identified fewer than three times are listed in the supplementary material (Table S1).

RESULTS

The initial systematic search of the ClinicalTrials.gov database yielded 5383 results. After excluding protocols based on conditions of trials (e.g. trials of behavioral addictions or urge incontinence) and status (i.e. terminated trials), 1464 trials were selected for full‐description screening. The dataset was further refined by focusing on trials that included craving, urge or desire as outcome measures, resulting in a final selection of 1255 trials (Figure 1). Thirty‐one disagreements at the data extraction stage and nine during full‐description screening were resolved with the help of a third rater (N.N. and P.R., supervised by H.E.).

General characteristics of trials

Start date, completion status and outcome measures

Since the inception of ClinicalTrials.gov in 1997, the number of registered clinical trials has increased substantially (note that 39 trials were identified before the US Food and Drug Administration Modernization Act of 1997, which established the clinical trials registry). Before 2005, trial registration on ClinicalTrials.gov was limited (1998–2004, n = 89 trials); however, beginning in 2005, registration accelerated rapidly. Between early 1987 and late 2004, 32 trials evaluated stimulant craving, the most studied substance. After 2005, 30% (n = 377) of the trials focused on nicotine craving, 25.9% (n = 326) on alcohol craving and 13.2% (n = 166) on opioid craving. More than a third of (36.7%) of the trials were registered in the last 5 years (2020–2024, n = 461) (Figure 2). According to their reported completion status, 61.7% (n = 775) of the trials were completed, but only 35.4% (n = 275) included results, while 64.5% (n = 500) were completed with no results available. In contrast, 37.5% (n = 471) of the trials were active (including recruiting, active but not recruiting and not yet recruiting), and less than 1% (n = 9) were classified as enrolling by invitation.

FIGURE 2.

Trials with craving as an outcome measure by start date and substance of use. The annotated heat map illustrates the absolute number of clinical trials registered in each year according to their actual start date and substance of use since for which craving was measured as outcome. The map spans from the first identified entry in 1987 to 5 November 2024. X‐axis (year): the map displays trial entries from 1987 through 2022, with the first column representing all trials registered between 1987 and 2004, followed by subsequent annual entries from 2005 to 2022. Y‐axis (substance indication): each row corresponds to a different substance category (e.g. nicotine, alcohol opioids, stimulants, marijuana, multi‐substance and not specified). Color intensity: the heat map uses color intensity to show the absolute number of trials registered for each substance and year. Darker colors represent a higher number of trials, while lighter colors indicate fewer trials.

Regarding outcome measures, 14.6% (n = 184) of trials reported craving, desire or urge as the only primary outcome measure, while 21.3% (n = 268) included these as a primary outcome measure among others. More than half of the trials (58.8%, n = 739) included craving as a secondary outcome measure. Trials studying craving in SUDs were conducted across five continents and in 44 countries. North America (the US and Canada) accounted for the majority of the main research sites for registered trials (74.9%, n = 940), followed by Europe (12.2%, n = 154). Asia (6.8%, n = 86), Africa and Oceania together accounted for less than 1.5% of trials over time. The origins of 5% (n = 19) of the trials were not specified or were conducted in various locations. The US accounted for 70.1% (n = 880) of all registered trials, which was expected, given that ClinicalTrials.gov is operated by the US National Library of Medicine (Figure S1).

Design characteristics of trials

Enrollment, phase and design

The median enrollment size across all trials was 60 participants [interquartile range (IQR) = 60.0), whereas the mean was 283.7 (SD = 250.0) (Figures [Link], [Link], [Link],[Link], [Link]). Nearly all the trials included adults and older adults (97.2%, n = 1220) across genders, with 2.7% (n = 35) including children and adults (13 years to 24 years). Among all trials with reported phase information (51.3%, n = 644), Phase 1 (8.5%, n = 107) and phase 2 (22.3%, n = 280) were the most frequently represented (Figure S3). In the interventional trials, randomization was used in nearly all cases (81.6%, n = 1025), and parallel assignment was used in 61.6% (n = 773).

Intervention and substance

In terms of intervention type, nearly half of the trials focused on pharmacotherapies (e.g. varenicline, buprenorphine) (46.4%, n = 583), followed by behavioral interventions (e.g. cognitive behavioral therapy, mindfulness) (25.3%, n = 318). Brain stimulation devices (e.g. transcranial magnetic stimulation, deep brain stimulation) were used in 11.9% (n = 150) of the trials, while 12.1% (n = 152) of the trials involved other interventions (e.g. biological, dietary supplements) (Figure 3).

FIGURE 3.

Trials by assessment title and types of interventions. The annotated heat map illustrates the relationship between craving assessment tools (n ≥ 3) (questionnaires) and the types of interventions used in clinical trials. Each cell in the heat map represents the use of a specific assessment title in trials using particular interventions. Abbreviations sorted by frequency in each category of substances. Multi‐item Questionnaires of Nicotine: QSU, Questionnaire on Smoking Urges; MNWS, Minnesota Nicotine Withdrawal Scale; CEQ, Craving Evaluation Questionnaire; TCQ, Tobacco Craving Questionnaire; MPSS, Mood And Physical Symptoms Scale; WSWS, Wisconsin Smoking Withdrawal Scale; FTND, Fagerström Test For Nicotine Dependence; SJWS, Shiffman Jarvik Withdrawal Scale; NCA, Nicotine Craving Assessment; QVC, Questionnaire Of Vaping Craving; WISDM, Wisconsin Inventory of Smoking Dependence Motives; HWQ, Hughes‐Hatsukami Withdrawal Questionnaire. Multi‐item Questionnaires of Alcohol: PACS, Penn Alcohol Craving Scale; AUQ, Alcohol Urge Questionnaire; OCDS, Obsessive Compulsive Drinking Scale; ACQ, Alcohol Craving Questionnaire; CrEQ, Craving Experience Questionnaire; AASS, Alcohol Abstinence Self‐Efficacy Scale; AUDIT, Alcohol Use Disorders Identification Test; SitCQ, Situational Confidence Questionnaire; DAQ, Desire for Alcohol Questionnaire; ADS, Alcohol Dependence Scale; YCS, Yale Craving Scale. Multi‐item Questionnaires of Opioid: OCS, Opioid Craving Scale, DDQ, Desire for Drug Questionnaire, SCS, Substance Craving Scale, HCQ, Heroin Craving Questionnaire, OCDUS, Obsessive Compulsive Drug Use Scale, PACSAD, Penn Alcohol Craving Scale adapted for opioids, SOWS, Subjective Opioid Withdrawal Scale. Multi‐item Questionnaires of Stimulant: CCQ, Cocaine Craving Questionnaire; SCS, Substance Craving Scale; SCQ, Stimulant Craving Questionnaire; MCCS, Minnesota Cocaine Craving Scale; CSSA, Cocaine Selective Severity Assessment; CCS, Cocaine Craving Scale; ACSA, Amphetamine Cessation Symptom Assessment. Multi‐item Questionnaires of Marijuana: MCQ, Marijuana Craving Questionnaire; MWCh, Marijuana Withdrawal Checklist. SIQ: Single‐Item Questionnaire.

Nicotine was the most frequently studied substance, appearing in 32.5% (n = 408) of the trials, followed by alcohol (27.5%, n = 346). Opioids (14.2%, n = 179) and stimulants (13.1%, n = 165) were the next most common substances. Marijuana/cannabis was the least frequent (5.3%, n = 67). Fewer than 4% (n = 47) of the trials involved multiple substances, and in 3.1% (n = 39), the substance was not specified. Other substances, such as water pipe, inhalants and ketamine, appeared in less than 1% (n = 4) of the trials (Figure 2).

Characteristics of assessment tools

In terms of assessment type, 59.1% (n = 945) of the trials used multi‐item questionnaires, while 25.1% (n = 401) used single‐item questionnaires to assess craving. Of all the single‐item questionnaires, 7.4% (n = 30) used EMA to measure craving. Notably, 15.7% (n = 251) of the trial registries did not provide sufficient information about their craving assessment instrument. Among the 1597 outcome measures reviewed, 108 different multi‐item craving questionnaires were recognized. Of these 108 multi‐item self‐reports, only 38 distinct questionnaires appeared three or more times in the database and were analyzed accordingly. Half of the multi‐item questionnaires measure phasic (immediate) craving (51.6%, n = 420), while 38% (n = 315) assessed tonic (over a certain time interval) craving. Approximately 6% (n = 50) of the surveys targeted both phasic and tonic craving.

Craving multi‐item questionnaires are classified in two primary categories here: multi‐item surveys with a single dimension and multi‐item surveys with multiple dimensions in addition to, a category of multi‐item surveys addressing abstinence effects (e.g. withdrawal).

Multi‐item questionnaire with single‐dimension

Multi‐item questionnaires with single dimension are designed to assess craving as a unified construct by measuring the intensity, frequency and duration aspects of craving through a brief number of items. While these aspects are distinct, they are conceptually combined to provide a comprehensive evaluation of the craving experience. Rosenberg [24] highlighted that such tools are not typically based on explicit theoretical models of craving, but aim to measure its core features—intensity, frequency and duration—over specific time frames (e.g. currently or the past 24 hours). Similarly, Kavanagh et al. [17] emphasized that multi‐item measures are particularly useful in capturing the phenomenon more comprehensively than single‐item tools, as they reduce measurement error and enhance internal consistency. Approximately 20.5% (n = 167) of the multi‐item questionnaires in our database fall into this category.

Multi‐item questionnaire with multiple‐dimensions

Multi‐dimensional instruments usually represent craving or substance use models with multiple constructs. These models are based on the premise that craving comprises several related but independent constructs, and assessing these constructs may contribute to greater conceptual, empirical and theoretical robustness of measurement [13, 17]. These self‐report instruments fall under three categories.

Craving based on obsessive‐compulsive theory

The measurements in this category were designed to reflect the resemblance between main features of obsessive‐compulsive disorder and substance use disorder [25] and developed following Yale–Brown Obsessive‐Compulsive Scale (Y‐BOS). They are inspired by the theory that the intrusive and disruptive thoughts and images associated with drug craving are parallel to anxiety‐provoking obsessions [9, 17, 24]. Additionally, drug‐seeking behavior, excessive consumption and the resulting impairment are analogous to the repetitive and ritualized compulsions performed to alleviate anxiety. These self‐reports are structured around two dimensions: obsessions and compulsions. Of the reviewed questionnaires, 8.3% (n = 68) were based on this theoretical framework.

Craving based on cognitive theory

A range of craving measurement tools have been developed following Tiffany's cognitive craving theory in 1990 [12] [32‐item Questionnaire of Smoking Urges (QSU)]. According to this theory, craving emerges when cognitive control processes or situational factors impede the execution of an automated drug‐use action schema triggered by environmental cues [9, 11, 17, 19]. These questionnaires often conceptualize craving in four (occasionally five, known as desire subscale) domains: purposefulness, expectancy, emotionality and compulsivity [17, 24]. Approximately half (49.6%, n = 404) of the reviewed multi‐item questionnaires were developed according to this cognitive theory of craving.

Craving based on elaborated intrusion theory

The development of this instrument is based on the elaborated intrusion (EI) theory of desire [20], which views craving as an emotional‐cognitive phenomenon. It begins with a spontaneous intrusive thought that leads to cognitive elaboration and sensory imagery. These measurements assess the features of craving by evaluating its intensity, vividness of sensory imagery linked to it and its intrusive nature. Only one questionnaire is identified in this systematic review according to this model: the Craving Experience Questionnaire (1.1%, n = 9).

Abstinence effects

Craving was initially associated with substance withdrawal and considered a component of motivational processes. This view reflects the historical conception of craving—rather than an established physiological or biological link—before cognitive concepts of automatic and non‐automatic processing were applied to the study of addictive behaviours [13, 26, 27, 28]. A variety of measures have been developed to measure abstinence effects, which refer to the outcomes resulting from stopping drug use, whether because of deprivation or cessation [13, 28, 29]. These instruments mainly use either a simple adjectival checklist of symptoms or multi‐item, factor‐analyzed measures to evaluate abstinence effects. The self‐reports in this category may vary based on their primary assessment goal (i.e. withdrawal effects, physical/psychological dependence). Consequently, they primarily assess craving in conjunction with behavioral and physical withdrawal symptoms such as anxiety, depression, difficulty concentrating or sleeping, hunger, irritability, impatience and restlessness [28]. Nearly 15% (n = 121) of the multi‐item questionnaires aimed to assess craving alongside withdrawal measures.

EMA

EMA offers valuable opportunities for assessing craving by collecting real‐time data in daily life [15]. The first use of EMA to evaluate alcohol craving was recorded in France in 2014 (NCT02087358). Since then, EMA has been widely applied to assess nicotine craving between 2014 and 2024, reflecting its feasibility and addressing methodological constraints in clinical settings. As a result, there has been an increase in trials using EMA, particularly for marijuana and opioids. Of all these trials only six used EMA with multi‐item self‐reports, while 28 trials used it with single‐item VAS. Nearly all trials (n = 34) provided sufficient information on the application of EMA (Table S3).

Craving self‐report assessments by substance

Overall, the QSU [18] was the most frequently used multi‐item questionnaire, appearing in 20% (n = 163) of the trials. This was followed by the Penn Alcohol Craving Scale (PACS) [30] (12.1%, n = 99), the Alcohol Urge Questionnaire (AUQ) [31] (9.8%, n = 80), the Obsessive‐Compulsive Drinking Scale (OCDS) [25] (7.2% n = 59) and the Marijuana Craving Questionnaire (MCQ) [32] (5.6% n = 46), which were the next most common multi‐item surveys.

Nicotine

In nicotine trials (n = 408), a total of 509 outcome measures were analyzed. Multi‐item questionnaires were the most frequently used assessment type (60.7%, n = 309), with 33 different questionnaires identified across all nicotine trials, although only 12 were used three or more times and are reported here (Table 1).

TABLE 1.

An overview of the multi‐item craving questionnaires recruited as outcome measures in clinical trials.

Substance	Name of questionnaire (frequency)	Theoretical model	Construct validity		Time frame	Items (version)	Authors (y)	Last used
			Factors	Subscales
Nicotine	CEQ (n = 25)	Abstinence effects	–	Smoking satisfaction Psychological reward Aversion enjoyment of respiratory tract sensations craving reduction (item 11 only)	Tonic (not specified)	12 (modified)	[33] (1992)	2024
	FTND (n = 8)	Abstinence effects	2	TFF CPD Nicotine yield inhalation Other questions	Tonic (not specified)	6 (original)	[34] (1991)	2022
	HWQ (n = 3)	Abstinence effects	–	Cigarette craving Irritability Anxiety Difficulty concentrating Restlessness Headache Drowsiness GI tract disturbances Fatigue Impatience Hunger Eating Sweating Dizziness	Tonic (past 24 h)	14 (original)	[35] (1986)	2022
	MNWS (n = 46)	Abstinence effects	–	Craving for cigarette (item 1) Irritability, frustration or anger Anxiety Difficulty concentrating Restlessness Increased appetite or weight gain Depressed or sad mood Insomnia or sleep problems	Tonic (past 24 h)	8 (original)	[36] (1986)	2024
	MPSS (n = 9)	Abstinence effects	–	Irritability Poor concentration Restlessness Depressed mood hunger Intensity and frequency of smoking urges (items 6 and 7)	Tonic (past 24 h and past week)	6 (original)	[37] (2004)	2024
	NCA (n = 6)	Not specified	–	Craving, desire and urge for a cigarette (items 1, 4 and 5)	Phasic	5 (original)	[38] (unknown)	2012
	QVC (n = 4)	Cognitive theory	3	Desire Intention Positive outcome	Phasic	6 (original)	[39] (2018)	2022
	QSU (n = 157)	Cognitive theory	2	Strong desire and intention to smoke, with smoking perceived as rewarding for active smokers An anticipation of relief from negative affect with an urgent desire to smoke	Phasic	10 (brief)	[18] (2001)	2024
	SJWS (n = 7)	Abstinence effects	5	Stimulation/sedation Physical symptoms Psychological symptoms Craving (desire to smoke)	Phasic	25 (original)	[40] (1976)	2021
	TCQ (n = 23)	Cognitive theory	4	Emotionality (smoking in anticipation of relief from withdrawal symptoms or negative mood) Expectancy (anticipation of positive outcomes from smoking) Compulsivity (an inability to control tobacco use) Purposefulness (intention and planning to smoke for positive outcomes)	Phasic	12 (brief)	[41] (2003)	2024
	WSWS (n = 8)	Abstinence effects		Anger Anxiety Concentration Sadness Hunger Sleep Urges (items 9, 11, 20 and 26)	Tonic (past 24 h and past week)	28 (original)	[42] (1999)	2022
	WISDM (n = 3)	Abstinence effects	–	Affiliative attachment Automaticity Loss of control Behavioral choice‐melioration Cognitive enhancement Craving (items 11, 29, 37, 50) Cue‐exposure or associative processes Negative reinforcement Positive reinforcement Social or environmental goads Taste and sensory processes Tolerance weight control	Tonic	68 (original)	[43] (2004)	2023
Alcohol	AASS (n = 8)	Abstinence effects	–	Negative affect Social/positive Physical and other concerns Craving/urges (items 1, 7, 10, 11 and 19)	Tonic (past week)	20 (original)	[44] (1994)	2019
	ACQ (n = 20)	Cognitive theory	4	Compulsivity (urges and desires in anticipation of loss of control over drinking) Expectancy (urges and desires to drink in anticipation of the positive benefits of drinking) Purposefulness (urges and desires coupled with intent and planning to drink) Emotionality (urges and desires to drink in anticipation of relief from withdrawal/negative affect)	Phasic	12 (brief‐revised)	[45] (1994)	2024
	ADS (n = 3)	Abstinence effects	3	Loss of behavioural control and heavy drinking Obsessive‐compulsive drinking style Psychoperceptual and psychophysical withdrawal	Tonic	29 (original)	[46] (1982)	2024
	AUQ (n = 80)	Cognitive theory	1	Desire to drink, expectation of a desired outcome from drinking Inability to avoid drinking if alcohol was available	Phasic	8 (brief)	[31] (1995)	2024
	AUDIT (n = 4)	Not specified	–	Alcohol consumptions Drinking behavior Alcohol‐related problems	Tonic	10 (original)	[47] (1993)	2023
	CrEQ (n = 9)	Elaborated intrusion theory	3	Imagery Strength Intrusion	Tonic (past week)	30 (original)	[20] (2011)	2024
	DAQ (n = 4)	Cognitive theory	3	Positive and negative reinforcement Strong desires and intentions Strong desires and intentions	Phasic	14 (original)	[48] (1998)	2024
	OCDS (n = 55)	Obsessive‐compulsive theory	2	Obsessive (measures thoughts about drinking) Compulsive (drinking behaviour)	Tonic (past week)	14 (original)	[25] (1995)	2024
	PACS (n = 89)	Intensity‐frequency‐duration	1	Frequency, intensity and duration of thoughts about drinking Ability to resist drinking if available Craving for alcohol	Tonic (past week)	5 (original)	[30] (1999)	2024
	SitCQ (n = 4)	Abstinence effects	3	Negative affect situations (Unpleasant emotions, physical discomfort and conflict with others) Positive affect situations (pleasant emotions, social pleasure to use and pleasant times with others) Urges and testing control (control over my use of alcohol and drugs temptations) (item 5)	Tonic	8 (brief)	[49] (1986)	2024
	YCS (n = 3)	Intensity‐frequency‐duration	1	Desire for a cigarette/a drink Intensity of the strongest desire to smoke/drink Intensity of the desire when tried to refrain from smoking/drinking Intensity of the desire to smoke/drink after having first cigarette/drink Intensity of the desire to smoke/drink during the stressful situation	Tonic (last week)	5 (original)	[50] (2015)	2021
Opioid	DDQ (n = 13)	Cognitive theory	3	Desire and intention Negative reinforcement control	Phasic	13 (original)	[51] (2002)	2024
	HCQ (n = 8)	Cognitive theory	4	Desire to use heroin Lack of self‐efficacy Compulsivity Relief	Phasic	14 (original)	[52] (1993)	2023
	OCDUS (n = 3)	Obsessive‐compulsive theory	3	Heroin thoughts and interference Desire and control resistance to thoughts and intention	Tonic (past week)	13 (original)	[51] (2002)	2024
	OCS (n = 29)	Intensity‐frequency‐duration	1	Current intensity Intensity during 24 h Frequency Responsiveness to drug‐related conditioned stimuli Perceived likelihood of use if in a setting with access to drugs	Phasic‐tonic	3 (original)	[53] [54] (1997) (2014)	2024
	PACSAD (n = 8)	Intensity‐frequency‐duration	1	Frequency, Intensity and duration of thoughts about drinking Ability to resist drinking if available Craving for alcohol	Tonic (past week)	5 (original)	[30] [55] (1999) (2020)	2024
	BSCS (n = 5)	Intensity‐frequency‐duration	1	Craving intensity Craving frequency Craving length Number of times think had craving during 24 h Total time spent craving during 24 h	Phasic (past 24 h)	16 (brief)	[56] (1999)	2024
	SOWS (n = 3)	Abstinence effects	–	Motoric Autonomic Gastrointestinal Musculoskeletal Psychic symptoms	Phasic	16 (original)	[57] (1987)	2024
Stimulants	ACSA (n = 3)	Abstinence effects	3	Anxiety and mood Fatigue Craving (items 15 and 16)	Tonic (past 24 h)	16 (original)	[58] (2008)	2021
	CCQ (n = 19)	Cognitive theory	4	Desire to use cocaine Intention and planning to use cocaine Anticipation of positive outcomes Anticipation of withdrawal or distressing symptoms Lack of control over use	Phasic	10 (brief)	[59] (1993)	2023
	CCS (n = 3)	Intensity‐frequency‐duration	1	Current intensity Intensity during 24 h Frequency Responsiveness to drug‐related conditioned stimuli Perceived likelihood of use if in a setting with access to drugs	Phasic‐tonic	3 (original)	[53] (1995)	1997
	CSSA (n = 4)	Abstinence effects	–	Hyperphagia Hypophagia Carbohydrate craving Cocaine craving intensity and frequency (items 4 and 5) Bradycardia Sleep problems Anxiety Energy level Activity level Tension Attention Paranoid ideation Anhedonia Depression Suicidality Irritability	Tonic (past 24 h)	18 (original)	[60] (1998)	2011
	MCCS (n = 7)	Intensity‐frequency‐duration	1	Intensity Frequency Duration	Tonic (past week)	5 (original)	[61] (1991)	2023
	SCQ (n = 14)	Cognitive theory	1	Desire to use drug Anticipation of positive outcomes from drug use Intention and planning to use drug Lack of control over drug use	Phasic	10 (original)	[59] (2015)	2024
	SCS (n = 10)	Intensity‐frequency‐duration	1	Current intensity Intensity during 24 h Frequency Responsiveness to drug‐related conditioned stimuli Perceived likelihood of use if in a setting with access to drugs	Phasic‐tonic	3 (original)	[53] (1995)	2024
Marijuana	MCQ (n = 39)	Cognitive theory	4	Emotionality (smoking in anticipation of relief from withdrawal symptoms or negative mood) Expectancy (anticipation of positive outcomes from smoking) Compulsivity (an inability to control tobacco use) Purposefulness (intention and planning to smoke for positive outcomes)	Phasic	17 (brief)	[62] (2009)	2024
	MWCh (n = 4)	Abstinence effects	–	Anger Craving for marijuana Depressed mood Decreased appetite Headaches Irritability Nervousness Restlessness Sleep difficulty Strange dreams	Tonic (past 24 h)	22 (original)	[63] (1999)	2017

Note: These questionnaires are used in the trials registered on ClinicalTrials.gov with equal and more than three times repetition in the result from January 1997 to November 2024. Table sorts according to substance frequency.

Abbreviations: AASS, Alcohol Abstinence Self‐Efficacy Scale; ACQ, Alcohol Craving Questionnaire Revised; ADS, Alcohol Dependence Scale; ACSA, Amphetamine Cessation Symptom Assessment; AUDIT, Alcohol Use Disorders Identification Test; AUQ, Alcohol Urge Questionnaire; BSCS, Brief Substance Craving Scale; CCQ, Cocaine Craving Questionnaire; CCS, Cocaine Craving Scale; CEQ, Cigarette Evaluation Questionnaire; CrEQ, Craving Experience Questionnaire; CSSA, Cocaine Selective Severity Assessment; DAQ, Desires for Alcohol Questionnaire; DDQ, Desire for Drug Questionnaire; FTND, Fagerström Test for Nicotine Dependence; HCQ, Heroin Craving Questionnaire; HWQ, Hughes‐Hatsukami Withdrawal Questionnaire; MCCS, Minnesota Cocaine Craving Scale; MCQ, Marijuana Craving Questionnaire; MNWS, Minnesota Nicotine Withdrawal Scale; MPSS, Mood and Physical Symptoms Scale; MWCh, Marijuana Withdrawal Checklist; NCA, Nicotine Craving Assessment; OCDS, Obsessive Compulsive Drinking Scale; OCS, Opioid Craving Scale; PACS, Penn Alcohol Craving Scale; PACSAD, Penn Alcohol Craving Scale adapted for opioids; QSU, Questionnaire on Smoking Urges; QVC, Questionnaire of Vaping Craving; SCS, Substance Craving Scale; SCQ, Stimulant Craving Questionnaire; SitCQ, Situational Confidence Questionnaire; SJWS, Shiffman Jarvik Withdrawal Scale; SOWS, Subjective Opioid Withdrawal Scale; TCQ, Tobacco Craving Questionnaire; WISDM, Wisconsin Inventory of Smoking Dependence Motives; WSWS, Wisconsin Smoking Withdrawal Scale; YCS, Yale Craving Scale.

Nearly half, 50.8%, (n = 157) of the nicotine trials used the QSU [18]. The Minnesota Nicotine Withdrawal Scale (MNWS) [35, 36, 64] (14.8%, n = 46) was the next most frequent multi‐item survey followed by the Cigarette Evaluation Questionnaire (CEQ) [33] (8%, n = 25), and the Tobacco Craving Questionnaire (TCQ) [65] (7.4%, n = 23) (Table 2). Approximately 3.2% (n = 10) of the nicotine trials also used untitled multi‐item questionnaires (not psychometrically evaluated for validity or reliability) (Table S2). A majority (63.7%) of the nicotine multi‐item questionnaires were designed to measure phasic craving (i.e. right now) (e.g. QSU and TCQ), while 29.4% measured tonic craving (i.e. past 24 hours or past week) (e.g. MNWS and CEQ) (Figure 4).

TABLE 2.

A brief overview of the nicotine multi‐item craving questionnaires recruited as outcome measures in clinical trials.

Theoretical model	Name of questionnaire (frequency)	Time frame	Items (version)
Cognitive theory	QSU [18] (n = 157)	Phasic	10 (brief)
	TCQ [41] (n = 23)	Phasic	12 (brief)
	QVC [39] (n = 4)	Phasic	6 (original)
Abstinence effects	MNWS [66] (n = 43)	Tonic	9 (original)
	CEQ [33] (n = 19)	Tonic	12 (modified)
	FTND [34] (n = 8)	Not specified	6 (original)
	MPSS [37] (n = 7)	Tonic	6 (original)
	WSWS [42] (n = 7)	Tonic	28 (original)
	SJWS [40] (n = 7)	Phasic	25 (original)
	WISDM [43] (n = 3)	Tonic	68 (original)
	HWS [35] (n = 3)	Tonic	9 (original)
Not specified	NCA [38] (n = 6)	Phasic	5 (original)

Note: Table sorts according to the frequency of the assessments.

Abbreviations: CEQ, Cigarette Evaluation Questionnaire; FTND, Fagerström Test for Nicotine Dependence; HWS, Hughes‐Hatsukami Withdrawal Questionnaire; MNWS, Minnesota Nicotine Withdrawal Scale; MPSS, Mood and Physical Symptoms Scale; NCA, Nicotine Craving Assessment; QSU, Questionnaire on Smoking Urges; QVC, Questionnaire of Vaping Craving; SJWS, Shiffman Jarvik Withdrawal Scale; TCQ, Tobacco Craving Questionnaire; WISDM, Wisconsin Inventory of Smoking Dependence Motives; WSWS, Wisconsin Smoking Withdrawal Scale.

FIGURE 4.

Parameter space of multi‐item craving self‐reports used as outcome measures in trials registered in the ClinicalTrials.Gov. This sunburst chart illustrates the breakdown of multi‐item self‐report questionnaires used to assess craving, categorized by substance, theoretical framework, questionnaire title and version. Each layer provides progressively detailed information (the numbers in parentheses indicate the frequency of use in the reviewed studies). Substance category (innermost layer): represents the substances targeted by the questionnaires. Categories include by frequency, nicotine, alcohol, opioids, stimulants, marijuana. Underlying theoretical model (second layer): indicates the theoretical basis of the questionnaires, grouped into single‐dimensional and multi‐dimensional frameworks. Single dimension: IFD, intensity‐frequency‐duration. Multiple dimensions: CT, cognitive theory; OC, obsessive‐compulsive; AE, abstinence effects (craving and withdrawal symptoms); and NS, not specified. Questionnaire title (third layer): lists the specific titles of craving questionnaires categorized by substance. Nicotine: QSU, Questionnaire Of Smoking Urges; MNWS, Minnesota Nicotine Withdrawal Scale; CEQ, Cigarette Evaluation Questionnaire; TCQ, Tobacco Craving Questionnaire; MPSS, Mood and Physical Symptoms Scale; FTND, Fagerström test for Nicotine Dependence; WSWS, Wisconsin Smoking Withdrawal Scale; NCA, Nicotine Craving Assessment. Alcohol: PACS, Penn Alcohol Craving Scale; AUQ, Alcohol Urge Questionnaire; OCDS, Obsessive Compulsive Drinking Scale; ACQ, Alcohol Craving Questionnaire. Opioids: OCS, Opioid Craving Scale; PACSAD, Penn Alcohol Craving Scale adapted for opioids; DDQ, Desire for Drug Questionnaire; HCQ, Heroin Craving Questionnaire; SCS, Substance Craving Scale. Stimulants: CCQ, cocaine craving questionnaire; SCQ, Stimulant Craving Questionnaire; SCS, Substance Craving Scale; MCCS, Minnesota Cocaine Craving Scale. Marijuana: MCQ, Marijuana Craving Questionnaire. Questionnaire version (outermost layer): specifies the format or item count for each questionnaire. B, brief; O, original; AD, adapted; M, modified; RB, revised brief (using only few items of the questionnaires' brief version). Small circles (linked to outermost layer): represent the number of times each specific version was identified in the results. Background color of circles: reflects the craving time frame assessed: phasic, immediate craving; tonic, craving in a given moment; phasic‐tonic, combination of both; not specified, no clear indication.

Single‐item measures (VAS) comprised only 21.2% (n = 108) of reported outcomes, with 12.9% (n = 14) of these specifically associated with EMA (Table S3).

Approximately 18% (n = 92) of the trials lacked sufficient information. These trials provided such limited detail on the assessment tools that specific information about these outcomes was difficult to determine (e.g. only stating ‘craving self‐report questionnaires’ as the assessment tool).

Alcohol

Across 346 alcohol trials, 414 outcome measures were analyzed for assessing alcohol craving. Multi‐item questionnaires were used in the majority of trials (68.3%, n = 283), with 24 different questionnaires identified, but only 11 were used three or more times and are reported here (Table 1).

PACS [30] (31.4%, n = 89), followed by the AUQ [31] (28.2%, n = 80), and the OCDS [25] (19.4%, n = 55), were the most frequent alcohol craving multi‐item questionnaires (Table 3). Additionally, 1.4% (n = 4) of the trials used untitled multi‐item questionnaires to assess craving (i.e. not psychometrically evaluated for validity or reliability) (Table S2). Sixty percent of the alcohol craving questionnaires were designed to measure tonic craving (e.g. PACS and OCDS), while 37.7% evaluated phasic craving (e.g. AUQ) (Figure 4).

TABLE 3.

A brief overview of the alcohol multi‐item craving questionnaires recruited as outcome measures in clinical trials.

Theoretical model	Name of questionnaire (frequency)	Time frame	Items (version)
Intensity‐frequency‐duration	PACS [30] (n = 89)	Tonic	5 (original)
	YCS [50] (n = 3)	Tonic	5 (original)
Cognitive theory	AUQ [31] (n = 80)	Phasic	8 (original)
	ACQ [45] (n = 20)	Phasic	12 (brief‐revised)
	DAQ [67] (n = 4)	Phasic	13 (original)
Obsessive‐compulsive theory	OCDS [25] (n = 55)	Tonic	14 (original)
Elaborated intrusion theory	CrEQ [20] (n = 9)	Tonic	30 (original)
Abstinence effects	AASS [44] (n = 8)	Tonic	20 (original)
	SitCQ [49] (n = 4)	Tonic	8 (brief)
	ADS [46] (n = 3)	Tonic	29 (original)
	AUDIT [47] (n = 4)	Tonic	10 (original)

Note: Table sorts according to the frequency of the assessments.

Abbreviations: AASS, Alcohol Abstinence Self‐Efficacy Scale; ACQ, Alcohol Craving Questionnaire; ADS, Alcohol Dependence Scale; AUDIT, Alcohol Use Disorders Identification Test; AUQ, Alcohol Urge Questionnaire; CrEQ, Craving Experience Questionnaire; DAQ, Desires for Alcohol Questionnaire; OCDS, Obsessive Compulsive Drinking Scale; PACS, Penn Alcohol Craving Scale; YCS, Yale Craving Scale; SitQ, Situational Confidence Questionnaire.

Single‐item questionnaires were used in 21.5% (n = 89) of alcohol craving assessments, with EMA applied in only 6.7% (n = 6) of them (Table S3).

Approximately 10.1% (n = 42) of trials did not provide adequate information regarding craving assessment tools.

Opioids

In 179 opioid trials, 200 outcome measures were analyzed. Unlike alcohol and nicotine, multi‐item questionnaires were less frequently used to measure opioid craving (36%, n = 72), with 22 different questionnaires identified, but only seven were used three or more times and reported here (Table 1).

The Opioid Craving Scale (OCS) [54, 68] (40.2%, n = 29) was the most frequently used multi‐item questionnaire followed by the Desire for Drug Questionnaire (DDQ) [51] (18%, n = 13). The Heroin Craving Questionnaire (HCQ) [69] and PACS adapted for opioids [55], were both frequent as the next most common multi‐item self‐reports with (11.1%, n = 8). Nearly 30% of the opioid multi‐item questionnaires were designed to assess phasic craving (e.g. DDQ and HCQ), while 19.4% assessed tonic craving (PACS adapted for opioids) (Table 4). Approximately half (47.2%) of the assessments were designed to evaluate both phasic and tonic craving (e.g. OCS) (Figure 4).

TABLE 4.

A brief overview of the opioids multi‐item craving questionnaires recruited as outcome measures in clinical trials.

Theoretical model	Name of questionnaire (frequency)	Time frame	Items (version)
Intensity‐frequency‐duration	OCS [54] (n = 24)	Phasic‐tonic	3 (original)
	PACSAD [30] (n = 89)	Tonic	5 (adapted)
	SCS [70] (n = 5)	Phasic‐tonic	16 (original)
Cognitive theory	DDQ [51] (n = 13)	Phasic	13 (original)
	HCQ [45] (n = 28)	Phasic	14 (original)
Obsessive‐compulsive theory	OCDUS [51] (n = 3)	Tonic	14 (original)
Abstinence effects	SOWS [57] (n = 3)	Tonic	16 (original)

Note: Table sorts according to the frequency of the assessments.

Abbreviations: DDQ, Desire for Drug Questionnaire; HCQ, Heroin Craving Questionnaire; PACS, Penn Alcohol Craving Scale; OCDUS, Obsessive Compulsive Drug Use Scale; OCS, Opioid Craving Scale; SCS, Substance Craving Scale; SOWS, Subjective Opioid Withdrawal Scale.

Single‐item questionnaires were the most frequently used assessment tool for opioids (47.5%, n = 95), with EMA used in 5.2% (n = 5) of these trials (Table S3).

Less than 20% of the trials (16.5%, n = 33) provided such limited information about their assessments that significant details remained unknown (Table S2).

Stimulants

In 165 stimulant trials, 186 craving outcome measures were analyzed. Single‐item questionnaires (36%, n = 67) and multi‐item (35.4%, n = 66) questionnaires were used almost equally in assessing craving in stimulants. Among 20 different multi‐item questionnaires identified, only nine were used three or more times and are reported here (Table 1).

The CCQ [59] was the most frequently used multi‐item questionnaire (28.7%, n = 19), followed by the Stimulant Craving Questionnaire (SCQ) [71] (21.2%, n = 14) and the Substance Craving Scale (SCS) [72] (15.1%, n = 10). More than half of the multi‐item questionnaires for stimulants were designed to measure phasic craving (e.g. CCQ and SCQ), while 25.7% assessed tonic craving (e.g. Minnesota Cocaine Craving Scale) (Table 5). The SCS was designed to measure both phasic and tonic craving (Figure 4).

TABLE 5.

A brief overview of the stimulants multi‐item craving questionnaires recruited as outcome measures in clinical trials.

Theoretical model	Name of questionnaire (frequency)	Time frame	Items (version)
Cognitive theory	CCQ [59] (n = 19)	Phasic	10 (brief)
	SCQ [59] (n = 14)	Phasic	10 (original)
	DDQ [51] (n = 3)	Phasic	13 (original)
Intensity‐frequency‐duration	BSCS [70] (n = 10)	Phasic‐tonic	16 (brief)
	MCCS [61] (n = 7)	Tonic	5 (original)
	CCS [73] (n = 3)	Phasic‐tonic	3 (original)
Abstinence effects	CSSA [60] (n = 4)	Tonic	18 (original)
	ACSA [58] (n = 3)	Tonic	16 (original)

Note: Table sorts according to the frequency of the assessments.

Abbreviations: ACSA, Amphetamine Cessation Symptom Assessment; BSCS, Brief Substance Craving Scale; CCQ, Cocaine Craving Questionnaire; CCS, Cocaine Craving Scale; CSSA, Cocaine Selective Severity Assessment; DDQ, Desire for Drug Questionnaire; MCCS, Minnesota Cocaine Craving Scale; SCQ, Stimulant Craving Questionnaire.

No EMA accompanied stimulants single‐ or multi‐item assessments.

Almost 30% (n = 53) of these outcome measures provided inadequate information regarding the craving assessment tool (Table S2).

Marijuana

In 67 marijuana trials, 66 outcome measures were analyzed, with 65.1% (n = 43) using multi‐item questionnaires to assess craving. Of the 16 different multi‐item questionnaires identified, only two were used three or more times and are reported here (Table 1).

The MCQ [32] was the most frequently used questionnaire (90.7%, n = 39), followed by the Marijuana Craving Checklist (MCC) [63] (9.3%, n = 4) (Table 6).

TABLE 6.

A brief overview of the marijuana multi‐item craving questionnaires recruited as outcome measures in clinical trials.

Theoretical model	Name of questionnaire (frequency)	Time frame	Items (version)
Cognitive theory	MCQ [32] (n = 39)	Phasic	22 (brief)
Abstinence effects	MWCh [63] (n = 4)	Tonic	17 (original)

Note: Table sorts according to the frequency of the assessments.

Abbreviations: MCQ, Marijuana Craving Questionnaire; MWCh, Marijuana Withdrawal Checklist.

The MCQ was designed to assess phasic craving, whereas the MCC was intended to measure tonic craving (Figure 4).

Single‐item questionnaires were used in 19.7% (n = 13) of the trials, with only two of these using EMA (Table S3).

Additionally, 15.1% (n = 10) of the trials did not provide sufficient data regarding the craving assessment tools (Table S2).

DISCUSSION

This systematic review provides an overview of the craving self‐report assessments used as outcome measures in drug addiction trials registered on ClinicalTrials.gov. A total of 1255 trials and 1597 outcome measures were identified, with a gradual increase in registered trials since the inception of the ClinicalTrials.gov database in 1997, which accelerated significantly after 2005. At the time of our search, most trials were completed, with over a third having been translated into published results. Craving was less frequently reported as the only primary outcome measure (15.5%) compared to when it was included as one of multiple primary outcomes (21.7%). Multi‐item questionnaires were the most commonly used assessment tools for evaluating cravings (59.1%), likely because of their ability to capture the multi‐dimensional nature of craving more effectively than single‐item questionnaires (25.1%) In total, 107 different multi‐item questionnaires were identified across various substances, but only 38 were used three or more times in the trials.

Accurate measurement is essential to avoid impeding knowledge synthesis and to prevent the emotional and time burden of repeated assessments on communities [74]. Studying the evaluation of craving and its associated processes is increasingly important, given the growing interest in academic research. Such studies can significantly aid clinicians during and after intervention by helping them assess the severity of a patient's drug use tendencies, select appropriate treatment approaches, monitor progress throughout the treatment trajectory and estimate the potential risk of relapse in individuals maintaining abstinence [10, 11, 75, 76]. These practical benefits highlight the key role of craving measurement as both a research focus and clinical tool.

Various methods, such as behavioral, psychophysiological and neuroimaging techniques have been proposed for measuring craving, despite the considerable debate among researchers about the concept of craving and its measurements [13]. Nevertheless, the idea that self‐report surveys are the most effective method for assessing craving is almost ubiquitous and justified, given that craving is an inherently subjective phenomenon [6]. Most conceptualizations of craving presume that drug‐related motivational processes can be measured through self‐reported accounts of subjective experiences [13]. Over the years, a variety of self‐report questionnaires have been developed to measure cravings for both licit and illicit drugs, leading to a wide variety of assessment tools [10, 17]. Some multi‐item questionnaires, such as the QSU‐brief [18] or the OCDS [25], have emerged as de facto standards because of their frequent use [6] by researchers and clinicians. However, several others are seldom used and often become obsolete after being applied in only one or two specific studies.

Craving assessment tools are varied in terms of theoretical models, items and time frame in which they attempt to capture. Various theoretical models have tried to conceptualize craving and explain the complexity of this phenomenon [77]. Craving is believed to originate from learning mechanisms driven by positive reinforcement [78]—the desire for drug‐induced positive outcomes—or negative reinforcement [79])—the need to avoid withdrawal effects—or both [80]. In addition to classical conditioning models, contemporary ones highlight the role of cognitive processes [9, 10], such as the intention and desire to use, perceived lack of control and preoccupation with drug‐related thoughts and behaviors. Accordingly, each assessment tool is designed to measure specific aspects of craving based on the theoretical model it is developed from. Multi‐item self‐reports developed based on the cognitive theory of craving, such as the questionnaire of smoking urges [17], are among the most frequently used tools (e.g. AUQ, CCQ and QSU). This popularity reflects the idea that craving is a complex, multi‐dimensional experience involving various cognitive processes [8].

Some of the trials applied withdrawal measurements as proxy measures of craving. However, findings regarding the association between craving and factors such as withdrawal symptoms, the number of drug‐related problems and the severity of dependence remain mixed [24]. Classical views conceptualize withdrawal as a set of symptoms that emerge during abstinence and diminish over time or with resumed drug use. These perspectives may not fully align with modern theories of craving and withdrawal. Ideally, a withdrawal scale should reliably demonstrate consistent increases with abstinence across various contexts and populations [28, 81, 82]. Although no single nicotine withdrawal scale is universally endorsed, several tools—such as the MNWS, the Mood and Physical Symptoms Scale (MPSS) and the Wisconsin Smoking Withdrawal Scale (WSWS) have shown reliability. These scales exhibit psychometric validity by consistently detecting increases in withdrawal symptoms during abstinence [28].

Generally, single‐item Likert‐type or VAS are simple assessment tools where respondents circle a numeral or mark a point to indicate their craving intensity at a given moment. While these instruments facilitate the craving assessment process, they are often considered less robust for capturing the multifaceted nature of craving [13, 17]. Multi‐item questionnaires, however, can depict more intricate models of craving. In contrast, multi‐item questionnaires provide more comprehensive models of craving, often conceptualizing it as a composite of several subsidiary constructs. Tools such as Obsessive‐Compulsive Drug Use Scale [51] and the Alcohol Craving Questionnaire [45] exemplify this approach, offering enhanced conceptual, empirical and theoretical robustness [17]. Moreover, increasing the number of items in these questionnaires significantly improves their internal consistency, making them more reliable for assessing craving [13, 17, 83].

EMAs are designed to capture phasic craving through brief prompts delivered via mobile devices in real‐world settings. These tools are particularly effective at measuring immediate fluctuations in craving, which can be influenced by dynamic environmental factors. According to Serre et al., [15] a substantial body of research has established a positive association between craving and subsequent drug use, particularly for tobacco during quit attempts. This strong predictive link underscores the value of EMA in addiction research. The rising prevalence of EMA‐based trials in recent years [24] reflects the efficacy of this approach for capturing the real‐time, situational influences on craving, offering valuable insights into its dynamic nature [15].

Self‐report measures of craving assess either: tonic craving, referring to retrospective subjective estimations of craving over a specific period of time in the past (e.g. last week), or phasic craving, referring to momentary state reports of craving (e.g. craving levels at a given moment) [6, 84]. The former reflects a generalized subjective experience of craving that persists over a long period of time and is associated with an individual's dependence, intensifies during withdrawal period and is linked to physiological tolerance [85]. The latter, however, captures dynamic changes in craving levels that can be elicited by various factors, such as exposure to drug‐related cues or ingestion of drug itself [84, 86]. Cue‐elicited cravings, although typically short‐lived, can be experienced quite intensely [85].

It could be speculated that tonic and phasic craving are conceptually related. Findings from an experiment on excessive drinkers suggest that the functions of tonic and phasic craving are different [84, 85]. This aligns with the tonic/phasic model of brain dopamine regulation, where tonic dopamine response refers to a steady‐state, low level of extrasynaptic dopamine release and phasic dopamine response refers to the acute, spike‐dependent release of dopamine, modulated by substance use, which produces the rewarding effects [83, 87, 88]. The imbalance between tonic and phasic dopamine release has been shown to have a neurobiological significant impact on the brain's reward system, leading to fluctuations in craving levels (i.e. tonic and phasic craving) and, ultimately, drug‐seeking behaviours [88, 89].

Therefore, the relationship between tonic and phasic craving can be explored in clinical setting to predict maladaptive behaviours such as binge drinking [67]. Studies have shown that targeting tonic craving through interventions may help reduce phasic responses to alcohol or associated cues, emphasizing the need for multi‐dimensional craving assessments. This finding highlights the complex interaction between these two types of craving and suggests that treatment approaches should address both to be more effective [67, 90]. The self‐reports developed based on the cognitive craving theory, are typically designed to capture the instantaneous form of craving. The prevalence and frequent use of these questionnaires in our results indicate that phasic craving was more frequently addressed than tonic craving.

Our results align with previous findings in the literature [6, 78], where QSU‐brief [18] was identified as the most commonly used multi‐item survey for assessing nicotine craving. Kavanagh et al. [17] described human desire as ‘an affectively charged cognitive event in which an object or activity associated with pleasure or relief of discomfort is in focal attention’ and emphasized that desire by definition is conscious, compared to its precursors and consequences, which often occur outside of conscious awareness. Accordingly, they believed that PACS [30], the CEQ [20] and the OCDS [25] (obsession subscale) were the measures that fit to this definition. The PACS and the OCDS were indeed among the most frequently used multi‐item questionnaires for measuring alcohol craving [17]. Kleykamp et al. [21] reported that the single‐item VAS was the most commonly used instrument for assessing opioid craving, used in half of the studies they reviewed, consistent with our results. In terms of multi‐item questionnaires, the OCS [54] was more frequently used compared to the DDQ [51], the HCQ [69] and the OCDUS [51], which Kleykamp et al. [21] identified as the most common craving self‐reports. This trend reflects a growing preference for brief measures with fewer items in clinical trials targeting opioid craving.

Ultimately, the selection of craving measurement tool should be guided by the specific research objectives and context. Multi‐item self‐report assessments are suitable for comprehensive evaluations, because they provide a nuanced understanding of craving through their multiple subsidiary constructs [17]. In contrast, single‐item VAS are highly sensitive to rapid fluctuations in craving levels and are well‐suited for repeated measurements because of their simplicity in both administration and scoring [6, 24, 86]. This makes single‐item VAS particularly effective in controlled laboratory settings [91], following manipulations as drug cue exposure [13] or for real‐time craving measurements using technological devices. Moreover, the theoretical framework underlying each measurement tool influences its suitability for various research designs, making some tools more suitable for certain studies than others. By carefully considering these factors, and clearly defining the structures being assessed [92], researchers can select the most appropriate tools to accurately capture the complexity of craving.

This review provides a comprehensive survey of the variety of craving assessment tools in addiction studies. However, there is no consensus on which craving assessment measures can optimally capture the precise craving levels and determine the efficacy of specific interventions—an essential aspect of addiction studies. The call for explicit guidelines [93] is needed, as it would provide researchers with clear criteria for selecting the most appropriate and psychometrically validated tools. Establishing such guidelines could not only help reduce inconsistencies across studies, but also improve the overall reliability and validity of craving measurements, and thereby strengthening addiction research and also ensuring fidelity of implementation.

Limitation of this review

The limitations of this review are important to consider. Initially, the reliance on ClinicalTrials.gov as the primary source of data means the findings are primarily US‐centric, as trials conducted in the US are required to register on this platform [94], ClinicalTrials.gov, which is managed by the National Institutes of Health (NIH), represents the largest clinical trials database globally. While the registry includes over 330 000 registrations from 209 countries, the predominance of US‐centric entries may limit the generalizability of the results to other countries or regions where different craving assessment tools may be in use. Furthermore, the absence of a systematic psychometric and content analysis of the instruments is another limitation of this review. A future content analysis investigating the structural elements and conceptual foundations of craving assessment tools could enhance our understanding of their utility and limitations. These considerations underscore areas for future research to further refine and expand the findings of this review. A formal risk‐of‐bias assessment was not conducted because of the reliance on clinical trial registrations, which lack some key methodological details. As such, potential biases are acknowledged as a limitation of this review (e.g. incomplete reporting, industry sponsorship).

Recommendation for future studies

Nearly 20% of the reviewed outcome measures failed to provide adequate information about the tools used to assess craving. These trials only mentioned ‘assessing craving with self‐reports’ without specifying key details such as the exact title, version of the self‐report questionnaires, or citing reliable references. The use of unpublished self‐report questionnaires (Table S2) that are not psychometrically validated, coupled with inaccuracies in the references provided, can confound researchers and contribute to heterogeneity in selecting appropriate craving self‐report assessments. It is, therefore, essential to establish a minimum reporting standard for craving assessment tools. We also recommend that an optimal set of details must be included in the registration of a trial in ClinicalTrials.gov. This should include accurate references or detailed information on the title and version of multi‐item self‐report questionnaires, as well as precise data on the setting, reaction time and scale for single‐item questionnaires or EMA.

CONCLUSIONS

This article provides an overview of the current status of craving self‐reports as an outcome measure in trials registered on ClinicalTrials.gov. The results demonstrate that craving is increasingly targeted as a primary outcome measure in drug addiction trials. Given the growing global prevalence of both licit and illicit drug use, focusing on craving is important for analyzing the underlying aspects of substance use disorders and also refining treatment strategies. Multi‐item surveys for craving are more frequently used than single‐item measures, likely because they capture various the multi‐faceted nature of drug craving. While the heterogeneity of self‐report craving assessments summarized in this review reflects the varied aims and dimensions of craving targeted by different trials, the lack of standardization presents significant challenges for cross‐study comparisons. To enhance the consistency and applicability of the future research, it is helpful to establish a consensus on standardized guidelines for selection among available craving assessment tools or proposals for unified assessment tools. Such efforts can facilitate better aggregation of results, improve the quality of trials and also provide a clearer path forward in the addiction treatment research.

AUTHOR CONTRIBUTIONS

Matin Toulami: Conceptualization (equal); data curation (equal); formal analysis (equal); investigation (equal); methodology (equal); writing—original draft (equal). Keyvan Ghasemi: Data curation (equal); formal analysis (equal); software (equal); visualization (equal); writing—original draft (equal). Parnian Rafei: Formal analysis (supporting); writing—original draft (supporting); writing—review and editing (equal). Arshiya Sangchooli: Conceptualization (lead); methodology (supporting); supervision (supporting); writing—review and editing (equal). Nastaran Nafissi: Data curation (equal); formal analysis (lead); writing—review and editing (equal). Arash Khojasteh Zonoozi: Methodology (supporting); writing—review and editing (lead). Hossein Mohaddes Ardabili: Methodology (supporting); supervision (supporting). Mohsen Ebrahimi: Software (supporting); visualization (supporting). Tara Rezapour: Conceptualization (lead); supervision (equal); writing—review and editing (equal). Hamed Ekhtiari: Conceptualization (equal); methodology (equal); project administration (equal); supervision (equal); writing—review and editing (equal).

DECLARATION OF INTERESTS

None.

Supporting information

Table S1. An overview of the multi‐item craving questionnaires recruited as outcome measures in clinical trials. These questionnaires are employed in the trials registered on ClinicalTrials.gov in the result from January 1997 to November 2024. (Table sorts according to substance frequency).

Table S2. An overview of the multi‐item craving questionnaires lacking references or public validation recruited as outcome measures in clinical trials. These questionnaires are employed in the trials registered on ClinicalTrials.gov in the result from January 1997 to November 2024. (Table sorts according to substance frequency).

Table S3. Overview of the Ecological Momentary Assessments applied for assessing craving in trials registered in the ClinicalTrials.gov.

Figure S1: Origins of the trials and the top six countries with the highest number of trials.

Figure S2 A: Trends in patient enrollment, categorized by trial start dates.

Figure S2 B: Trends in substances and patient enrollment.

Figure S2 C: Trends in interventions and patient enrollment.

Figure S3 A: Enrollment data alongside the number of trials.

Figure S3 B: Trial phases with corresponding trial counts.

Toulami M, Ghasemi K, Rafei P, Sangchooli A, Nafissi N, Khojasteh Zonoozi A, et al. Craving self‐reports as outcome measures in drug addiction trials: A systematic review of ClinicalTrials.gov. Addiction. 2025;120(10):1931–1950. 10.1111/add.70064

DATA AVAILABILITY STATEMENT

The data supporting this study's findings are available from the corresponding author upon reasonable request.