Assessing Cigarette Smoking Withdrawal in Unmedicated Individuals: Validating the Wisconsin Smoking Withdrawal Scales Version 3 (WSWS3)

INTRODUCTION

Nicotine withdrawal is an important component of the nicotine dependence construct. Key definitions of nicotine dependence comprise the notion that the occurrence of withdrawal, occasioned by falling blood levels of nicotine, is a core manifestation of dependence (Baker et al. 2004; Baker, Piper, et al. 2012). Consistent with this, self-report assessments of smoking withdrawal often predict success in quitting smoking (Piasecki et al. 2002; Piper et al. 2011; Schnoll et al. 2016); individuals who report the greatest increases in withdrawal symptoms during their quit attempt are relatively unlikely to quit, with increases in negative affect and craving being especially predictive in this regard (Piasecki et al. 2002; Piper et al. 2011; Robinson et al. 2019; Schnoll et al. 2016). These distressing symptoms may partially explain why smoking lapses are especially likely in the first 1–2 weeks of the quit attempt (Germovsek et al. 2021; Hughes, Keely, and Naud 2004; Kenford et al. 1994). In addition, smoking cessation pharmacotherapies appear to work, in part, by suppressing abstinence-induced increases in withdrawal symptoms, especially craving (Bolt et al. 2012; Chakraborti, Coffman, and Piper 2022; Ferguson, Shiffman, and Gwaltney 2006). Thus, withdrawal symptoms may be of clinical value in that they index those at risk for cessation failure.

The characterization and assessment of smoking withdrawal were pioneered by John Hughes and Dorothy Hatsukami in the mid-1980’s when they developed the Minnesota Nicotine Withdrawal Scale (MNWS) (Hatsukami et al. 1984; Hughes et al. 1984, 1986, 1991). Since that time, a great deal of research has led to other withdrawal scales such as the Cigarette Withdrawal Scale (CWS) (Etter 2005) and the Wisconsin Smoking Withdrawal Scale (WSWS) (Welsch et al. 1999). Withdrawal and its assessment remain topics of clinical importance and debate with regard to withdrawal’s association with tobacco dependence and personality, which symptoms to assess (e.g., craving, newer symptoms such as anhedonia), whether to measure post-quit levels and/or change from pre-quit levels, and the prospective predictive value for cessation outcomes (Baker, Breslau, et al. 2012; Baker, Piper, et al. 2012; Gilbert et al. 2000; Hughes 2007b; Piper 2015; Shiffman et al. 2004).

The original WSWS, developed in 1999, comprised 28 items with seven subscales (Welsch et al. 1999). In 2021, Smith and colleagues sought to achieve broader coverage of withdrawal, reduce the number of items, and produce a brief version. This effort, in which participants used smoking cessation pharmacotherapies, led to a 6-item Brief version (WSWS2-B) and a 19-item Long version (WSWS2-L), the latter comprising six subscales (Craving, Negative Affect, Hunger, Sleep, Restlessness, and Concentration) (Smith et al. 2021). The Brief version was developed for situations in which assessment burden and time are limited. The Long version permits reliable assessment of each subscale and so is appropriate for situations where individual symptoms are of interest. Psychometric analyses showed that the WSWS2-L subscale structure had acceptable fit indices and the whole scale and the subscales had acceptable internal consistency. The six items comprised by the WSWS2-B were ones with high factor loadings on their respective WSWS2-L subscales (i.e., Craving, Negative Affect, Hunger, Sleep, Restlessness, and Concentration). Both WSWS versions were meaningfully correlated with dependence measures such as the Wisconsin Inventory of Smoking Dependence Motives (WISDM) and significantly predicted short-term abstinence (Smith et al. 2021). The WSWS2-L and WSWS2-B were highly inter-correlated across three different analysis samples (r’s < .95). However, the WSWS2 was developed and validated from smoking cessation clinical trials where all participants used pharmacotherapies, which might have influenced scale performance.

There is substantial evidence from both humans and animal models that nicotine withdrawal causes decreased sensitivity to rewarding events and that readministering nicotine re-establishes sensitivity (Epping-Jordan et al. 1998; Pergadia et al. 2014). Recent research suggests that decreased reward responsiveness appears to manifest as withdrawal related anhedonia: i.e., loss of subjective pleasure from life events (Cook et al. 2015, 2017; Hughes, Klemperer, and Peasley-Miklus 2020; Klemperer et al. 2021; Betts et al. 2025). Research shows that such anhedonia is alleviated by nicotine use, is correlated with other withdrawal symptoms, and predicts cessation likelihood (Betts et al. 2025; Cook et al. 2015; Leventhal et al. 2014). Such evidence has led to suggestions that anhedonia be included in withdrawal instruments; this research does just that. The current study examined two items designed to assess anhedonia to achieve greater coverage of the withdrawal construct and determine if they improve scale functioning. Thus, two items were added to the WSWS-L form and one item was added to the WSWS3-B.

This study evaluates the performance of the WSWS3 scales on measuring unmedicated withdrawal among a contemporary sample of people who smoke. Most previous studies evaluating smoking withdrawal assessments, including the ones evaluating WSWS2, involved participants who used smoking cessation pharmacotherapy (Smith et al. 2021). Such pharmacotherapies significantly suppress withdrawal symptoms and appear to suppress some symptoms (e.g., craving) more than others (Bolt et al. 2012). Therefore, almost all recent research on the validity and characteristics of withdrawal and the scales that measure it, are based on research that may not permit the full expression of the withdrawal syndrome. Most assessments of unmedicated withdrawal are decades old and involved earlier withdrawal measures (Hughes et al. 1986). Smoking patterns and populations have changed markedly since the early studies that used unmedicated individuals (Meza et al. 2023), illustrating the need for contemporaneous efforts to validate withdrawal assessments.

The current study involved the psychometric analysis of the WSWS3-L and WSWS3-B, which includes expanded content coverage from the WSWS2 to include the assessment of anhedonia. We analyzed data from a treatment-delay, one-group clinical trial in which participants tried to quit smoking without using pharmacotherapy early in their quit attempt. Participants attempting to quit smoking were instructed to abstain from smoking and not to use any smoking cessation pharmacotherapy during the first week of their quit attempt. They initiated 8 weeks of pharmacotherapy (C-NRT: nicotine patch + nicotine mini-lozenge) starting 1 week past the target quit day. This design allows for the examination of unmedicated withdrawal symptoms for the first week of the quit attempt, when withdrawal often peaks (Hughes 2007a). We examined scale structure and reliability, sensitivity to withdrawal changes during early quitting (e.g., abstinence), associations with theoretically relevant constructs and variables (e.g., dependence, smoking heaviness), and prediction of cessation outcomes. We compared correlations of the WSWS3 versions with theoretically related constructs and determined how such correlations differed from those obtained with the MNWS.

METHODS

This study design was registered and trial protocol is available at clinicaltrials.gov (NCT04969198). Figure 1 displays the study timeline relative to the target quit day “TQD”.

RESULTS

DISCUSSION

The updated Wisconsin Smoking Withdrawal Scale Version 3 (WSWS3) includes a new subscale of anhedonia, in addition to prior subscales of craving, concentration difficulties, hunger, negative affect, restlessness, and sleep problems. Recent research strongly suggests that anhedonia may be an important component of the smoking withdrawal syndrome (Betts et al. 2025; Cook et al. 2015, 2017; Hughes et al. 2020; Klemperer et al. 2021); including anhedonia items in the WSWS3 will allow researchers to further examine its status as a possible component of the withdrawal syndrome and investigate if and how it contributes to the maintenance of addiction and interferes with cessation processes. This research is unusual in that it evaluates the performance of withdrawal assessments in individuals who were attempting to quit smoking without any pharmacotherapy early in their quit attempt. The WSWS3 displayed reasonable psychometric properties and concurrent, convergent, and construct validity across several domains. The WSWS3 meets some of the thresholds for both single factor (WSWS3-B) and seven factor (WSWS3-L) solutions. However, some items cross-loaded on two factors (e.g., “feeling tired” loading on sleep and negative affect factors) and some measures of fit were subthreshold. The WSWS3 total scores displayed good concurrent, convergent validity with strong correlations with the Minnesota Tobacco Withdrawal Scale-Revised (MTWS-R). The WSWS3-L subscales of craving, negative affect, and anhedonia generally displayed significant positive correlations with related constructs (e.g., QSU, DASS-21, SHAPS), indicating convergent validity. Consistent with prior studies (Baker, Piper, et al. 2012; Smith et al. 2021), the WSWS3 brief and long total scores displayed modest, significant positive relationships with measures of tobacco dependence (e.g., WISDM, FTCD), but not smoking heaviness (e.g., cigarettes/day, exhaled carbon monoxide). The validity of the WSWS3 measures relative to the MSWS-R measure is indicated by the former’s prediction of the time-to-first cigarette dependence measure, a particularly good predictor of smoking cessation success (Baker et al. 2007). Additionally, only the WSWS3 predicted of smoking abstinence early in the quit attempt (at 1 week post quit day), which may be largely driven by craving.

Construct validity is supported by the fact that all of the WSWS3 total scale scores and subscale scores showed the characteristic inverted U waveform that Hughes (Hughes 2007a) listed as a criterion of withdrawal. However, the late decrease seen after 1-week post-TQD may likely reflect participants’ initiation of pharmacotherapy. Craving stands out from the other WSWS3 subscales with respect to its sensitivity to the quit attempt as well as its association with early abstinence. This is consistent with other research that shows that craving is typically the strongest predictor of abstinence and best accounts for the mediation of smoking pharmacotherapy effects on abstinence (Bolt et al. 2012; Piper et al. 2011). It may be that the presence of craving accounts for the fact that both the WSWS3-L and WSWS3-B predicted abstinence at 1 week post-quit while the MTWS did not. Therefore, for clinical or research contexts where prediction of abstinence is a key goal, it may be critical to use withdrawal scales that capture craving (such as the WSWS3). Additional research is needed to clarify when and for whom non-craving withdrawal symptoms predict clinically important outcomes (e.g., Schnoll et al. 2016; Robinson et al. 2019; Tonkin et al. 2020).

The fairly similar findings obtained with the long and brief forms of the WSWS3 in terms of their responses to smoking deprivation and their relationships with dependence assessments, suggest that the brief WSWS3 should be used in situations in which assessment burden is a concern. The long form should be used primarily when analyses of particular subscales (e.g., craving) are of interest.

Recent research suggests that anhedonia may be an important component of the smoking withdrawal syndrome (Betts et al. 2025; Cook et al. 2015, 2017; Hughes et al. 2020; Klemperer et al. 2021). In the present study, anhedonia scale scores increased significantly with the occurrence of the quit attempt, a critical criterion for a withdrawal symptom. Anhedonia scale scores were significantly associated at most (pre-quit and 1 day post-TQD), but not all (1-week post-TQD) times, with a valid anhedonia assessment that has been shown to strongly relate to the inability to quit smoking (Snaith-Hamilton Pleasure Scale: SHAPS) (Betts et al. 2025; Snaith et al. 1995). It remains an open question, whether, and to what degree, the anhedonia items capture relative overlap or independence of reward sensitivity dysregulation, negative affect, and general distress. For instance, anhedonia scale was significantly associated with depressive symptomology (e.g., DASS-21 depression subscale) and anhedonia items appear to contribute to construct coverage that is related and unrelated to the Negative Affect subscale (e.g., CFA fit improved when these factors were allowed to cross-load). It is unlikely that the anhedonia item meaningfully increased the association of the WSWS3 measures with relevant criteria. However, it may be important to include it in withdrawal scales given the wealth of data that it does, in fact, constitute a withdrawal symptom (Betts et al. 2025; Cook et al. 2015, 2017; Hughes et al. 2020; Klemperer et al. 2021).

We included only a single exploratory pain item in the WSWS given emerging evidence that suggests that pain should be considered as a withdrawal symptom (Ditre et al. 2018; Ditre, Zale, and LaRowe 2019; LaRowe and Ditre 2020). We did not find sufficient preliminary evidence to support its inclusion in the revised WSWS3, but future research with multi-item assessments may permit a more comprehensive psychometric evaluation of a possible pain factor. It falls to future research to determine whether pain should be considered a withdrawal symptom that warrants inclusion in a tobacco withdrawal scale, and if so, what items should be included to capture a valid and reliable assessment of pain (Bagdas et al. 2026).

One challenge to validating smoking withdrawal assessments is using an externally valid participant sample. This research evaluates the performance of withdrawal assessments in a contemporary sample of adults who smoke. Today’s smoking population differs from earlier populations in multiple ways. For instance, those smoking today tend to smoke less heavily, may face greater stigma and environmental restrictions on smoking, are more likely to experiences greater socioeconomic disadvantages, and are more likely to use other nicotine delivery products (U.S. Department of Health and Human Services 2010; Filion et al. 2012; U.S. Department of Health and Human Services 2014). Any of those factors might affect both nicotine dependence and withdrawal symptoms.

It is also vital to study withdrawal in a clinically relevant context: a quit attempt in which no cessation medication is used during the critical early withdrawal period. This context has public health relevance since the great majority of quit attempts are made without medication. This context should reduce distortion and suppression of withdrawal symptoms due to medication effects.

We included the full participant sample in our analyses examining changes in withdrawal symptoms during early quitting, which includes people who were completely abstinent as well as those who smoked, albeit they smoked at significantly reduced levels relative to baseline (see Figure S2). This decision reflects previous observations that those who lapse early post-quit often report the highest withdrawal symptoms (Piasecki et al. 2002). Indeed, individuals who were not able to achieve initial abstinence tended to have higher pre-TQD baseline withdrawal levels and showed increases in symptoms that were similar to those reported by participants who were completely abstinent. Thus, including only abstinent individuals would likely eliminate individuals who are highly dependent and who would show strong withdrawal reactions to decreased smoking (See Figures S2–4 and Table S7). Thus, the data on the WSWS3 scales presented in this paper reflect ability to detect the effects of decreased smoking, including both reductions as well as total abstinence. Importantly, we observe the same general patten of effects in both of those conditions.

This study has limitations. First, the duration of the medication-free quitting period was relatively brief (1 week) which truncates the time in which unmedicated withdrawal can be studied. The one-week medication delay was selected to balance competing considerations related to participant recruitment, attrition, and the desire to deliver medication treatment before participants had become too discouraged in their quit attempt. This study design decision allowed us to examine unmedicated withdrawal; however, it likely reduced abstinence rates which may limit the precision of subscale estimates of predictive validity for 1-week abstinence (e.g., due to the modest number of abstinent cases). Second, there was a trend for withdrawal symptoms to decrease during the pre-quit period. Gilbert and colleagues have noted a tendency for affective measures to decrease with repeated assessments (Gilbert et al. 1998; Sharpe and Gilbert 1998). The first withdrawal assessment occurred in the screening session and symptoms were especially elevated at this occasion (see Table S8). Symptoms might have been elevated at this time because of the novel clinical research context or the uncertainty and stress of the screening, assessment, and consent process. However, we observed significant within-subject variability during the pre-TQD period, suggesting that baseline withdrawal assessments capture both time-varying fluctuations in symptoms as well as stable characteristic traits. Third, we did not assess long-term abstinence (e.g., at 6-months).

In sum, this research provides evidence of the reliability and validity of the new WSWS3 long and brief smoking withdrawal assessment scales. The comparability of the two forms suggests that the brief WSWS3 could be used unless particular subscales are of interest. Our findings also support the use of the MTWS as a brief measure of withdrawal, consistent with a large body of prior research (Etter and Hughes 2006; Toll et al. 2007; West et al. 2006). One might use the WSWS3 if stronger prediction of dependence or post-quit abstinence is valued due to its assessment of craving; one might use the MSWS-R or WSWS2 if one does not view craving or anhedonia, respectively, as a withdrawal symptom. More research is needed to establish the theoretical status of the different types of withdrawal symptoms and their clinical utility.

Supplementary Material