■ Introduction

Alcohol dependence represents one of the leading global public health challenges, given its high rates of morbidity and mortality as well as its pronounced tendency toward recidivism [1]. In contemporary addictology, craving occupies a central role in understanding the mechanisms underlying the onset and maintenance of addiction. Craving is defined as a subjectively intense, often overwhelming, desire or urge to consume a psychoactive substance [2, 3]. Although it has been a subject of debate for decades, it is now widely accepted that craving is not binary (“present or absent”) but rather a dynamic and multidimensional, varying in intensity under the influence of biological, psychological and environmental factors [4].

Craving is understood through complementary neurobiological and cognitive-behavioural frameworks: repeated alcohol exposure sensitises meso­limbic dopaminergic pathways, so that alcohol-related cues acquire strong incentive value even as subjective pleasure from drinking declines [3, 5, 6], while cognitive models describe craving as a learned, automatic response that consumes cognitive resources and impairs inhibitory control [3, 7]. Because of its capacity to override the intention to abstain, craving is recognised as a core diagnostic criterion for alcohol use disorder in DSM-5 [8], for dependence syndrome in ICD-10 [9] and, more recently, within the construct of a “strong internal drive” in ICD-11 [10].

The evaluation of craving is of paramount importance as it represents a key parameter in both the diagnostic process and empirical research. The intensity of craving during early abstinence is considered among the most robust predictors of relapse. Empirical studies have demonstrated that patients reporting elevated levels of craving during inpatient treatment exhibit a significantly higher likelihood of relapse within the first six months post-discharge [11, 12]. Furthermore, craving substantially contributes to the patient’s subjective experience of distress, eliciting anxiety, sleep disturbances and social dysfunction thereby complicating and potentially undermining the therapeutic process [13].

Beyond its direct relationship with alcohol dependence severity, craving is increasingly understood as embedded within a broader affective context. Depressive symptomatology is highly prevalent among individuals with alcohol dependence [14], and negative affective states are thought to intensify craving through negative-reinforcement mechanisms, whereby alcohol consumption is used to alleviate distress rather than to pursue reward. Consistent with this framework, prior studies have reported a positive association between the severity of depressive symptoms and craving intensity in alcohol-dependent populations [15-18], suggesting that depressive symptomatology may represent a clinically meaningful correlate of craving, distinct from, and yet related to, the severity of alcohol use itself.

Several instruments have been developed to quantify alcohol craving, including the Obsessive Compulsive Drinking Scale (OCDS) and the Multidimensional Alcohol Craving Scale (MACS), both of which assess multiple craving dimensions but require longer administration time. The Penn Alcohol Craving Scale (PACS) developed by Flannery et al. [19] was designed as a brief, five-item alternative capturing frequency, intensity, duration and resistance to craving, and has demonstrated strong psychometric properties across multiple cross-cultural adaptations, including Korean [20] and Chinese [21] versions (Cronbach’s α = 0.91-0.95). Its brevity and established cross-cultural reliability make it particularly suitable for routine clinical use, where longer instruments are often impractical. Despite its wide international use, no validated Serbian-language version of the PACS, or of any comparable craving instrument currently exists, leaving a gap in the objective monitoring of craving among Serbian-speaking patients undergoing treatment for alcohol dependence.

The adaptation of a validated Serbian instrument would enable clinicians to more effectively monitor therapeutic response and provide timely identification of patients at high risk of relapse.

Objectives

The primary objective of this study was the translation, cross-cultural adaptation and psychometric evaluation of the Serbian version of the Penn Alcohol Craving Scale (SPACS) in patients with alcohol dependence. Specific objectives included the assessment of reliability (internal consistency and test–retest stability) as well as the examination of the scale’s construct validity through its correlation with the severity of alcohol dependence (AUDIT) and the degree of depressive symptomatology (BDI-II).

Hypotheses

H1: The SPACS will demonstrate high internal consistency and test–retest reliability.

H2: SPACS scores will correlate positively with AUDIT scores, reflecting the association between craving and alcohol dependence severity.

H3: SPACS scores will correlate positively with BDI-II scores, reflecting the association between craving and depressive symptomatology.

H4: SPACS scores will be negatively associated with duration of abstinence.

■ Material and methods

Study design and subjects

A cross-sectional observational study was conducted at the Clinic for Psychiatry of the Military Medical Academy (MMA) in Belgrade, between June and December 2025. The sample consisted of 54 patients (48 men and 6 women) enrolled consecutively in inpatient or outpatient treatment for alcohol dependence upon meeting eligibility criteria during this six-month recruitment period. Given the exploratory nature of this initial cross-cultural adaptation and the limited number of eligible patients presenting during the study period, no a priori sample size calculation was performed; recruitment concluded at the end of the pre-defined study period, yielding a final sample of 54 participants.

The inclusion criteria were age over 18 years, confirmed diagnosis of alcohol dependence according to ICD-10 classification by two independent psychiatrists [9], ability to understand the study and signed voluntary consent. The exclusion criteria were complicated withdrawal syndrome or delirium tremens, comorbid misuse of other psychoactive substances, presence of acute psychotic syndrome, high suicidal risk and cognitive impairments precluding comprehension of the questionnaires. Complicated withdrawal syndrome was excluded based on clinical assessment by the treating psychiatric team and participants with uncomplicated withdrawal symptoms remained eligible. Eligible participants were clinically assessed as cognitively intact and capable of completing the study questionnaires. No participant had a comorbid psychiatric diagnosis. Participants did not receive chronic pharmacological treatment; anxiolytic medication (diazepam) was administered “as needed” (PRN – Pro Re Nata) for management of withdrawal symptoms, and no participant was being treated with antidepressants or other centrally acting medications with known effects on craving.

Abstinence was defined as complete cessation of alcohol consumption, ascertained through patient self-reporting, collateral information obtained from family members or caregivers and blood-alcohol concentration testing at the time of assessment. To examine the association between craving intensity and length of abstinence, parti­cipants were grouped into four categories based on abstinence duration: 1) up to 3 months, 2) 4-12 months, 3) 1-2 years and 4) longer than 2 years [19, 22].

The study was approved by the Ethics Committee of the Military Medical Academy (No. 70/2025, June 4, 2025). All participants provided written informed consent in accordance with the Declaration of Helsinki and the standards of Good Clinical Practice (GCP).

Instruments

Penn Alcohol Craving Scale (PACS) is a 5-item self-assessment instrument designed to measure the frequency, intensity and duration of alcohol cravings, as well as the ability to resist it and the patient’s overall subjective feeling of the level of craving during the last seven days. Each item is scored on a 7-point Likert scale, ranging from 0 (none/not at all) to 6 (very pronounced/very often), with the total score (range 0–30) reflecting the intensity of the subjective craving experience. The original PACS was developed by Flannery et al. [19]; the present study utilised an identical version, adapted for the Serbian-speaking population through a back-translation process.

Beck Depression Inventory–Version II (BDI-II) was included as a convergent validity measure based on the theoretical rationale outlined in the Introduction. The BDI-II evaluated the presence and severity of depressive symptoms. This instrument assesses cognitive, affective, and somatic manifestations of depression experienced by participants over the preceding two-week period. The BDI-II consists of 21 items each rated on a four-point Likert scale ranging from 0 to 3 with higher scores reflecting greater intensity of depressive symptoms. Total scores are interpreted according to established thresholds as follows: mini­mal depression (0-13), mild depression (14-19), moderate depression (20-28) and severe depression (29-63). The BDI-II is extensively recognised and utilised in clinical practice having been rigorously validated by Beck et al. [23]. An adapted and validated version of the scale specifically designed for the Serbian population was applied [24].

Alcohol Use Disorders Identification Test (AUDIT) is a standardised instrument developed by the World Health Organization (WHO) for the screening and identification of individuals exhibiting risky or harmful patterns of alcohol consumption as well as those potentially meeting criteria for alcohol dependence. This instrument comprises 10 items encompassing three principal domains: alcohol consumption, dependence-related behaviours and alcohol-related problems. Each item is scored on a scale from 0 to 4, with higher scores indicative of more severe alcohol-related symptoms. The total score ranges from 0 to 40. According to established interpretive criteria, total scores are classified as follows: 0-7 low risk, indicating absence of significant alcohol-related problems, 8-15 hazardous alcohol use, 16-19 harmful alcohol use and 20 or higher suggests possible alcohol dependence. The AUDIT has been demonstrated to be a reliable and internationally validated screening tool, suitable for identifying individuals who may require further clinical evaluation or intervention concerning their alcohol use [25].

Socio-demographic Questionnaire. For the purposes of the study, a structured questionnaire was specifically developed to collect detailed demographic information and relevant clinical variables. Demographic data included participants’ gender, age, level of education and marital status. Clinical variables encompassed the duration of alcohol dependence, the number of previous treatment episodes and the current duration of abstinence. This questionnaire was designed to enable comprehensive characterisation of the study po­pulation and to facilitate analyses of potential associations between demographic and clinical factors and the outcomes of interest.

Procedure

The Serbian version of the PACS was developed using the standard back-translation method. Two independent board-certified psychiatrists translated the scale from English to Serbian, after which an independent translator produced a back-translation into the original language. An expert committee, comprising board-certified psychiatrists, psychiatric residents, a psychologist, a social worker and a linguist reviewed the translated version to confirm its semantic equivalence with the original. As no separate pilot testing was conducted prior to administration, instrument comprehensibility was established through this expert committee review process alone.

The finalised Serbian version of the PACS was administered together with the AUDIT, the BDI-II and the socio-demographic questionnaire indivi­dually to each participant. Test–retest reliability was assessed in a subsample of the first 31 consecu­tively enrolled participants who completed the SPACS a second time after a seven-day interval
under identical administration conditions; this interval was selected to be sufficiently long to mitigate recall bias, while remaining brief enough to minimise response variance attributable to the­rapeutic intervention during the retest period [26, 27]. As this subsample was not randomly selected, abstinence-duration subgroups may be unevenly represented within it.

Receiver operating characteristic (ROC) analy­sis was not performed as the cross-sectional design of this study did not include longitudinal follow-up; all participants were in abstinence at the time of assessment and no relapse occurred within the study period, precluding the use of a binary external criterion. Consequently, diagnostic cut-off values were not established, and trends across the four abstinence-duration groups were evaluated using variance analyses.

Statistical analysis

Data were processed using IBM SPSS Statistics version 26.0. Descriptive statistics (mean, SD, frequencies, percentages) were used to summarise socio-demographic and clinical characteristics. Normality of distributions was assessed using the Shapiro-Wilk test; SPACS and AUDIT scores significantly deviated from a normal distribution (p < 0.01), while BDI-II scores did not (p = 0.10). Accordingly, Spearman’s ρ was used for correlations involving non-normally distributed variables, including convergent validity analyses of SPACS with AUDIT and BDI-II and the association between SPACS and duration of abstinence, while Pearson’s r was used for item-total correlations and test–retest stability as standard psychometric indices of linear measurement stability computed on the continuous item- and total-score data. Internal consistency was evaluated using Cronbach’s a, including alpha-if-item-deleted values. Group differences in SPACS scores across abstinence duration categories were examined using one-way ANOVA. Given a significant Levene’s test indicating heterogeneity of variance, results were corroborated using Welch’s ANOVA and the non-parametric Kruskal-Wallis test. Effect size (η²) is reported for the ANOVA. Statistical significance was set at p < 0.05.

■ Results

Socio-demographic and clinical properties

A total of 54 participants of a mean of 50.7 years of age were enrolled in the research. The sample predominantly consisted of male participants (N = 48, 88.89%). The average duration of alcohol dependence in the examined group was 15.13 years of age while the respondents consumed alcohol for the first time at an average of 22.83.

The majority of participants were married or in a stable relationship (62.96%) while the remainder were single or divorced; there were no widowed participants.

Secondary education was the most prevalent level of attainment among the respondents (N = 30, 55.56%). Regarding employment status, the largest proportion of participants reported being employed (N = 38, 70.37%). All respondents were currently in abstinence. Detailed socio-demographic and clinical characteristics of the study participants are shown in Table I.

Psychometric properties of the SPACS

Internal consistency of the SPACS was high, with a Cronbach’s a of 0.964, which was consistent with, and somewhat higher than, the values reported for the original PACS (a = 0.92) and its Korean (a = 0.91) and Chinese (a = 0.95) adaptations. Corrected item-total correlations ranged from r = 0.812 (Item 4, Resistance) to r = 0.947 (Item 5, Overall assessment), and alpha-if-item-deleted values ranged from 0.948 to 0.971, indicating that no single item substantially compromised internal consistency (Table II). Test–retest reliability assessed in a subsample of 31 participants over a seven-day interval was very high (Pearson’s r = 0.980, p < 0.001). Taken together, these findings confirm H1.

Convergent validity was supported by significant positive correlations between SPACS scores and both AUDIT scores (Spearman’s ρ = 0.773, p < 0.001) and BDI-II scores (ρ = 0.543, p < 0.001) (Table III), confirming H2 and H3 respectively. A significant positive correlation was also observed between BDI-II and AUDIT scores (ρ = 0.437, p < 0.001), reflecting the well-documented overlap between depressive symptomatology and alcohol-related problems.

Correlation of craving with clinical and demographic parameters

PACS scores were significantly negatively correlated with duration of abstinence (Spear-man’s ρ = −0.649, p < 0.001), confirming H4 and indicating that craving intensity diminishes progressively over the course of sustained abstinence. A one-way ANOVA confirmed a significant diffe­rence in SPACS scores across the four abstinence-duration groups, F(3, 50) = 17.07, p < 0.001, η² = 0.51, indicating a large effect. As Levene’s test indicated heterogeneity of variance across groups (p = 0.003), this finding was corroborated using Welch’s ANOVA, F(3, 18.8) = 28.66, p < 0.001 and the non-parametric Kruskal-Wallis test, H(3) = 27.71, p < 0.001; all three approaches yielded convergent results.

No significant association was found between marital status and SPACS scores, F(2,51) = 0.44, p = 0.647, and this variable was therefore not retained in the final analysis (Table IV).

Predictors of craving: partial correlation and multiple-regression analysis

To examine whether the association between SPACS and BDI-II scores was independent of alcohol dependence severity, a partial correlation controlling for AUDIT scores was computed. The association remained statistically significant, though attenuated relative to the zero-order correlation (r = 0.360, p = 0.008), indicating that depressive symptomatology contributes to craving intensity beyond its shared variance with alcohol dependence severity.

A multiple linear regression was conducted to identify independent predictors of SPACS scores, with AUDIT, BDI-II, age, gender and abstinence-duration category entered as predictors. The overall model was significant with F(5, 48) = 22.27, p < 0.001 accounting for 66.7% of the va­riance in SPACS scores (R² = 0.699, adjusted R² = 0.667). AUDIT scores emerged as the strongest independent predictor (β = 0.555, p < 0.001), followed by BDI-II scores (β = 0.232, p = 0.029). Gender was coded 0 = male, 1 = female; female gender was associated with lower SPACS scores (β = −0.173, p = 0.043), although this finding should be interpreted with caution given the small number of women participants (n = 6). Age and abstinence-duration category were not significant independent predictors in the multivariate model (p = 0.477 and p = 0.131 respectively), suggesting that their bivariate asso­ciations with craving are substantially accounted for by shared variance with alcohol dependence severity and depressive symptomatology (Table V).

■ Discussion

This study represents a preliminary psychome­tric evaluation of the SPACS in patients with alcohol dependence. Overall, the findings support the reliability and convergent validity of the SPACS, which was in line with our initial expectations; however, given the sample size constraints discussed below, these results should be interpreted as an initial step toward validation rather than a complete psychometric characterisation of the instrument.

Alcohol craving is recognised in contemporary literature as a pivotal diagnostic criterion within the DSM-5, ICD-10, and ICD-11 classifications, and has been identified as a clinically important correlate of relapse risk following inpatient treatment [8-12]. The results of the present study confirmed the high internal consistency of the SPACS with a Cronbach’s a of 0.964. These findings are congruent with the original version by Flannery et al. [19] (Cronbach’s a = 0.92), as well as recent cross-cultural validations, such as the Korean (α = 0.91) [20] and Chinese versions (α = 0.95) [21]. The high corrected item-total correlations (ranging from r = 0.812 to 0.947, p < 0.001) across all five items indicate that each individual item contributes significantly to the measurement of a unidimensional craving construct thereby justifying the use of the total score in clinical practice; item-deleted alpha values further confirmed that no single item disproportionately influenced overall reliability.

The high temporal stability of the instrument (r = 0.980, p < 0.001) indicates that the SPACS is a stable measure for assessing the subjective experience of craving over a short time interval. A high level of stability of this kind is consistent with the findings of the original authors and suggests the potential applicability of the SPACS as a reliable tool for monitoring therapeutic effects during the rehabilitation process [19].

Convergent validity was supported by a strong positive correlation between SPACS and AUDIT scores (ρ = 0.773), suggesting that craving intensity is closely related to the severity of alcohol-related problems, which is consistent with theoretical models emphasising the central role of craving in the maintenance of addiction [2, 3, 8, 9].

A moderate positive correlation was also observ­ed between SPACS and BDI-II scores (ρ = 0.543), which is consistent with theories highlighting the interplay between affective disorders and addiction [14, 28, 29]. Notably, this association remained statistically significant after controlling for AUDIT scores (partial r = 0.360, p = 0.008), indicating that depressive symptomatology contributes to craving intensity independently of alcohol dependence severity. This finding provides empirical support for the inclusion of the BDI-II as a convergent validity measure in this study.

To further characterise the relative contribution of these variables, a multiple regression model including AUDIT, BDI-II, age, gender, and abstinence-duration category was conducted, explaining 66.7% of the variance in SPACS scores. AUDIT scores emerged as the strongest independent predictor, followed by BDI-II scores, reinforcing the pattern observed in the correlational and partial correlation analyses. Age was not a significant predictor. Female gender was associated with lower craving scores in this model; however, given the small number of women participants (n = 6), this finding should be regarded as exploratory rather than conclusive.

SPACS scores were significantly negatively correlated with duration of abstinence (ρ = −0.649, p < 0.001), and this was corroborated by a large, robust group difference across abstinence-duration categories (confirmed using Welch’s ANOVA and the Kruskal-Wallis test given violated homogeneity-of-variance assumptions), consistent with previous longitudinal studies suggesting that craving declines relatively rapidly and then more gradually over the course of sustained abstinence [30]. Interestingly, abstinence duration did not retain independent predictive value in the multivariate regression model, suggesting that its strong bivariate association with craving may be substantially accounted for by its overlap with alcohol dependence severity and depressive symptomatology rather than reflecting an independent temporal effect. This finding should be interpreted as hypothesis-generating and warrants replication in larger, longitudinal samples.

However, in line with theoretical and empirical models describing craving as a relatively stable and clinically significant feature of alcohol dependence, our results indicate that recovery does not progress uniformly across patients. Specifically, elevated craving (SPACS total score ≥ 6) persisted in 20% (2 of 10) of participants who had maintained abstinence for two years or longer, consistent with prior findings indicating that a subset of patients may continue to experience craving despite prolonged abstinence, which is probably due to long-lasting neuroadaptive changes and conditioned responses to alcohol-related cues. These findings further underscore the chronic nature of alcohol dependence and highlight the potential value of long-term, continuous therapeutic strategies targeting craving management [13]. However, given the modest subgroup size (n = 10), this observation should be considered descriptive rather than confirmatory.

Applicability to actively drinking and acutely detoxifying patients. All participants in this study were in abstinence at the time of assessment; consequently, the present findings cannot be directly generalised to actively drinking patients or those undergoing acute detoxification. Given that craving may manifest and fluctuate differently in the context of active use or acute withdrawal, future studies should examine the psychometric proper­ties and clinical utility of the SPACS specifically within these populations before its use is extended to such clinical contexts.

Caution regarding relapse-related interpretations. Although craving is theoretically and empirically linked to relapse risk in the broader literature, relapse was not directly assessed in the present study. Accordingly, statements regarding the SPACS’s utility for relapse prediction should be regarded as a theoretically grounded hypothesis rather than an empirically demonstrated property of the Serbian version, pending prospective validation.

Limitations. Several limitations should be considered when interpreting these findings. First, the relatively small sample size (N = 54) constrains the scope of this psychometric evaluation; in particular, it precluded exploratory or confirmatory factor analysis, meaning that the factorial structure of the SPACS – specifically, whether it reproduces the unidimensional structure of the original PACS – remains unconfirmed. Second, the single-centre design and cross-sectional nature of the study limit the generalisability of the findings and preclude conclusions about longitudinal change or causal relationships. Third, complicated withdrawal was excluded based on clinical judgment rather than a standardised instrument, and no formal pilot testing of the translated instrument was conducted prior to its administration. Fourth, the test–retest subsample was not randomly selected, which may limit the representativeness of the reliability estimate across abstinence-duration subgroups. Fifth, convergent validity was assessed using the AUDIT and BDI-II; a visual analogue scale or another established craving instrument was not administered, which limits the ability to benchmark the SPACS directly against existing craving-specific measures. Finally, the predominance of male participants (88.89%) is consistent with the epidemiological profile of alcohol dependence in the Republic of Serbia; however, this limits the generalisability of the findings, particularly regarding gender-related differences, to the female population [31]. Larger, multi-centre, longitudinal studies incorporating factor-analytic techniques are needed to confirm the factorial structure and broader clinical applicability of the SPACS.

■ Conclusions

The SPACS demonstrated strong reliability and convergent validity in this preliminary evaluation, supporting its potential value as a brief, easy-to-administer tool for assessing alcohol craving in Serbian-speaking patients with alcohol dependence. The SPACS maintains the same item count and structure as the original PACS, potentially addressing limitations of longer craving instruments like extended completion time and respondent burden. However, given the modest sample size and the absence of factor-analytic evidence, these findings should be regarded as an initial step toward validation rather than a definitive confirmation of the scale’s psychometric properties. Before the SPACS can be recommended for routine clinical or research use, further studies with larger, multi-centre samples are needed to confirm its factorial structure and to establish its applicability across broader patient populations, including actively drinking and acutely detoxifying individuals.

Conflict of interest/Konflikt interesów

None declared./Nie występuje.

Financial support/Finansowanie

None declared./Nie zadeklarowano.

Ethics/Etyka

The research was conducted after the approval of the Ethics Committee of the Military Medical Academy (No. 70/2025, June 4, 2025). All participants signed an informed consent, in accordance with the principles of the Declaration of Helsinki and the standards of Good Clinical Practice (GCP).

The work described in this article has been carried out in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki) on medical research involving human subjects, Uniform Requirements for manuscripts submitted to biomedical journals and the ethical principles defined in the Farmington Consensus of 1997.

Treści przedstawione w pracy są zgodne z zasadami Deklaracji Helsińskiej odnoszącymi się do badań z udziałem ludzi, ujednoliconymi wymaganiami dla czasopism biomedycznych oraz z zasadami etycznymi określonymi w Porozumieniu z Farmington w 1997 roku.

References/Piśmiennictwo