Discussion
CHA use continues to plague our society,5 with surgical patients having higher rates of alcohol misuse and dependence than the general public.29 In our EGS patients, 16 of the 91 (17.6%) were positive on either the AUDIT or CDT, indicating some level of hazardous or harmful drinking. Our findings are lower than that reported by Moore et al,6 who found self-reported alcohol abuse rates to be 23% in their general surgery population, but within reported ranges in other studies.30 Only 47% of eligible patients approached for inclusion in this study agreed to participate. This high rate of rejection may have led to a biased sample, selecting out those that have a history of CHA or other hazardous use behavior. Self-reporting surveys regularly underestimate the true burden of this disease by up to 40% to 50%, as pointed out by Livingston and Callinan,2 and our estimates of those truly affected by heavy alcohol use could be low since half of the eligible population opted out of the study after being informed that it was evaluating alcohol use. It is this very reluctance to report alcohol misuse that motivated us to compare %dCDT and the AUDIT, as %dCDT would be a more objective screening tool for CHA use. It could then be implied that the actual burden of disease is greater than previously anticipated, further highlighting the need for this investigation.
The AUDIT screening tool was chosen as it has a reported sensitivity and specificity of identifying alcohol use disorder in general medical patients of 90% and 80% to 90%, respectively.18 The AUDIT screening tool has also been tested and validated in multiple settings with a diverse population including surgical patients,17 19–21 has proven to be superior to most biochemical markers for detecting alcohol misuse,31 and excelled over other screening tools such as the CAGE (Cut down, Annoyed, Guiltly, Eye-Opener) Questionnaire.18 32 Its limitations arise from the need for patients to cooperate and divulge the information required in the questionnaire, hence a lack of objectivity. Also, it assesses alcohol-related problems during a longer prehospitalization period and thus identifies those potentially providing false-positive indications of recent heavy drinking. The inability of some patients to participate in self-assessment questionnaires due to the severity and acuity of their disease process at admission also adds to its limitations.
Numerous biochemical markers have been used in the clinical setting to help clinicians identify patients with varying degrees of alcohol use disorders. Gamma glutamyl transferase (GGT), aspartate aminotransferase, alanine aminotransferase, and mean corpuscular volume have all been proposed as markers of liver injury that increase in the serum due to chronic alcohol use, but have suffered from poor accuracy due to low sensitivities or specificities, whereas serum %dCDT levels have reported improved reliability.24 33 In a systematic review, %CDT’s performance (including older, less specific assays) in the clinical setting elucidated its higher specificity and in some cases sensitivity at detecting alcohol misuse compared with GGT.34 %dCDT’s levels are not impacted by diet, common drugs, or comorbid diseases such as hypertension, diabetes, lipid metabolism disorders, or disorders of the gastrointestinal tract; however, older assays can be less sensitive in women than in men.35 36
When examining the relationship between %dCDT and AUDIT, the correlation was low (ρ=0.14) but also limited to low numbers, and is analogous to a study by Hermansson et al,37 who saw the correlation between serum %dCDT levels and AUDIT scores to be between 0.15 and 0.2. Of the four patients with positive %dCDT levels, three had positive AUDIT scores, but most patients with positive AUDIT scores did not have elevated %dCDT levels. This makes sense, since %dCDT specifically measures heavy alcohol use in the recent past, whereas AUDIT is designed to capture a wider range of hazardous or harmful alcohol behaviors during a longer period.
Although well documented in the literature, our study failed to show any statistically significant association between drinking behaviors and complication rates, readmission rates, or length of stay. This may be attributed to the rarity of those complications and the relatively small cohort of only 91 participants, of whom only 5 had a positive %dCDT. Patients who do not have CHA use but other risky alcohol-related behaviors (such as occasional binge drinking, or drinking and driving) may be captured by the AUDIT but will not be detected by elevated %dCDT values. These behaviors, although increasing the risk of injury, are less likely to cause the pathophysiologic changes associated with CHA use. As Stibler indicates, %dCDT may be a marker of drinking habits associated with CHA use-related pathologic findings.23 36 Examining for the assignment of any diagnosis codes included under Elixhauser et al’s27 comorbidity grouping for alcohol abuse among the participants, we found that only one patient had been assigned such a diagnosis. That patient was positive on the AUDIT, but not %dCDT, and none of the four participants with positive %dCDT had diagnoses codes indicating alcohol abuse, suggesting that %dCDT can identify people missed during clinical assessment.
Aside from the potential selection bias discussed earlier, the 47% patient participation rate underpowered our study, limiting our ability to make inferential conclusions about alcohol use and this patient population.
There is a large body of literature supporting the view that CHA use leads to increased morbidity and mortality in patients admitted to the hospital for trauma or elective surgery.7–13 38 39 Identifying at risk patients early on in their hospital course may allow clinicians to institute treatments to mitigate and/or circumvent complications in such patients. This pilot study determined that 17.6% of our participating EGS patients had either a positive %CDT or AUDIT score, indicating the presence of some level of alcohol misuse. The AUDIT, however, is neither objective nor has the sensitivity to identify patients with increased risk of the physiologic impacts of CHA use. While we were unable to show a difference in outcomes, this could be due to the relatively small number of patients we were able to recruit, as well as the smaller subset that was specifically positive for CHA use. Additional research is needed to examine the impact of CHA use in larger cohorts of EGS patients, as well as the utility of routine %dCDT testing.