To our knowledge, this is the first study conducted on BS candidates that simultaneously describes the self-reporting of alcohol and substance use with objective urine toxicological tests. The overall prevalence of risky drinking in BS candidates was quite similar to that of the general population (7% vs. 5%, respectively) according to the AUDIT-C questionnaire. However, two additional relevant conclusions were derived from this study. First, we found a much higher proportion of male BS candidates with a risky drinking pattern compared to women BS candidates (18% vs. 2%). Second, male BS candidates showed a more than two-fold greater prevalence of risky drinking than men in the general Spanish population (18% vs. 7.6%, respectively)⁽¹⁷⁾.

In contrast, no differences according to sex were found with the ASSIST questionnaire, with a general prevalence of risky drinking of 5%. Although the AUDIT and ASSIST are screening tests, there are some differences in relation to item content and assessment time period that may account for these disparities.

Previous studies on risky drinking in BS candidates have yielded conflicting results. On one hand, the Longitudinal Assessment of Bariatric Surgery-2 [10] and Assessment of Bariatric Surgery studies [9] described a higher prevalence of risky drinking among BS candidates (19.6% and 16.6%, respectively), although data by sex were not provided. Nevertheless, these higher prevalences are likely overestimated due to methodological issues, since risky drinking was classified not only by the AUDIT score but also when the participant gave any positive response to additional or modified items suggestive of binge drinking, alcohol abuse, or dependency. Therefore, our data are not completely comparable to the results of these studies. On the other hand, the Swedish Obese Subject study [18] found no differences in risky drinking between BS candidates and matched controls. Nonetheless, this study had an important bias since participants who reported alcohol problems or with consumption ≥ 2.5 standard drinks were excluded. In summary, differences in assessment tools or criteria may account for the discrepant results found. It should be taken into account that BS candidates tend to underreport psychopathology or substance use in the pre-surgical assessment [19]. Indeed, most risky drinker individuals tend to underestimate their consumption and risk [20].

With regard to the alcohol biomarker, 15% of our patients presented EtG values ≥ 500 ng/ml, demonstrating that alcohol had been consumed within the previous 72 h. Surprisingly, alcohol biomarkers have seldom been used in BS candidates. Two studies using Peth, with a determination ≥ 0.05 µmols/L, reported similar proportions (12–18%) [13,14] as was also described in a retrospective study [15] in which 15.3% of individuals showed values of heavy alcohol use (≥ 200 ng/mL). Although an EtG ≥ 500 ng/ml is not diagnostic per se, in our study, this value increased the probability of being a risky drinker by 5 to eightfold. However, as expected, EtG and alcohol questionnaires were not concordant, similar to the previous studies with Peth, since different aspects were assessed with subjective and objective methods.

Nonetheless, the most significant predictive factor of risky drinking was sex in our study. Sex has previously been identified as an important risk factor for alcohol misuse [21]. Males seeking BS had a tenfold greater risk of risky drinking than females. Unexpectedly, being single, divorced, or widowed was a protective factor in our analysis, although the magnitude of its effect makes it negligible and the relationship between alcohol consumption and social support is complex [22]. Age and having any psychiatric diagnosis (apart from AUD) were not significant predictors of risky drinking in our study, in contrast to what some studies after BS have suggested [10].

The overall prevalence of AUD in the present study was 6%, quite similar to that obtained in other preoperative studies [10] using procedures not as robust as our structured criteria.

Assessment of alcohol use is relevant since an increased risk of developing any alcohol problem has been described, especially 2 years after BS. The risk is somehow greater in patients who had undergone Roux-en-Y-gastric bypass (RYGB)¹⁶, but also, it has been reported also in patients who had undergone sleeve gastrectomy [10,23,24], even in patients without a previous history of unhealthy alcohol use [25]. Anatomical and metabolic changes secondary to BS can modify the pharmacokinetics of alcohol, producing higher and longer blood concentrations for equivalent intakes prior to surgery, and thus with greater alcohol exposure in the brain [26]. There may also be some shared underlying physiological features between addiction and obesity in brain reward circuitry or its disruptions [27,28] and common genetic influences in both disorders [2] as well as a switch in coping strategies from eating to alcohol intake [29]. The Labs-2 study [21] reported a significant increase in regular drinking and in the prevalence of AUD after BS (especially following RYGB, increasing from 7% presurgery to 16% 7 years post-surgery), as well as a greater risk of illicit drug use.

All these findings have important clinical implications. BS patients, mainly males but also females, must be advised that alcohol use can increase or generate more negative health consequences after BS, even with amounts not considered harmful. This warning should be given to all patients, with or without any preoperative alcohol use, and should be maintained in the long term, since some studies [21] show significant figures up to 7 years after BS. This might be due to a certain relaxation of the patients over time with respect to their care. Therefore, regular monitoring is warranted, since besides the heightened risk of any type of harmful alcohol intake including AUD, alcohol use could potentially lead to poorer long-term BS outcomes, such as recurrent weight gain due to increased caloric intake from alcohol, malnutrition because of the possible compromise of adherence to nutritional guidelines and supplementation, higher rates of alcohol-related health complications including the risk of alcohol-associated liver disease [23] and mental health issues. Moreover, family, social, work, and economic consequences may also arise. The multidisciplinary team should be aware of any sign suggestive of alcohol-related difficulties to work together with the patient to adequately address the problem.

Apart from alcohol, tobacco was the drug most frequently self-reported at a risky level in 15% of the sample, similar to other studies (10 to 19%)^4,30.

Regarding other recreational substances, the global prevalence of risky use was 9%, being somewhat higher than the 4.5% reported by previous research [10,30]. Cannabis use was objectively verified in 4% of the males in our sample. Urine tests detected slightly higher figures than self-assessment for the use of benzodiazepines without prescription, opioids for non-medical use, and cocaine, confirming the under-reporting in BS candidates. Since few studies in BS patients have focused on other substances beyond alcohol and tobacco, comparison with our results is difficult. A recent study performed toxicology screening [31] and found similar proportions for benzodiazepines (5.6 vs. 6.6% in our study) and a slightly higher proportion for amphetamines (1.2 vs. 0%). Nonetheless, our study detected somewhat higher positive results for cocaine (1.5 vs. 0%) and cannabinoid markers (3.6 vs. 1.8%). Finally, a very small proportion of participants met the criteria for drug dependence or abuse (1% and 1%, respectively).

Reliability and validity-related aspects are among the strengths of our study: a semi-structured psychiatric interview characterized the sample, the use of validated questionnaires and urine toxicology screening, and participants were unaware of the performed drug examination before the study assessment visit. This study also has some limitations, such as differences in the temporal dimension of the questionnaires and the fact that an EtG ≥ 500 ng/ml does not necessarily indicate a harmful or risky pattern of drinking. The voluntary nature of the study might have implied some selection bias in the sample, although participants and non-participants were equivalent in age and sex. However, we cannot say with certainty that patients who did not volunteer were not at increased risk for alcohol and other substance use.