EVIDENCE LIBRARY
Alcohol–Cocaine Coexposure: Why Two Excluded Patients Changed a Severity Comparison
An 81-patient study found an alcohol-group severity comparison was nonsignificant before two observations were excluded. The result does not establish a DUI impairment threshold.
In an 81-patient emergency-care study, the difference in clinical severity between cocaine patients with and without detected blood ethanol was not statistically significant when all patients were included. After excluding two observations from the ethanol-negative group, the investigators reported a significant difference. Both analyses matter when evaluating what the paper supports about combined alcohol and cocaine exposure.
The study examined clinical intoxication severity, not driving performance. It does not establish a blood-alcohol or cocaethylene concentration that predicts a particular driver’s impairment.
How the alcohol groups were defined
Zucoloto and colleagues studied adults evaluated for suspected cocaine or crack intoxication at a São Paulo poison-control service between December 2015 and March 2017. Inclusion required a positive urine cocaine screening result. Three of 84 initially assessed patients were excluded for exposure to other stimulants, leaving 81.
Group A contained 62 patients without a positive blood ethanol determination; Group B contained 19 with detected blood ethanol. These were observed clinical groups, rather than patients randomly assigned to receive a controlled alcohol dose.
The classification did not establish that Group A had never consumed alcohol with cocaine. Thirty of its 62 patients reported associated alcohol use. Cocaethylene was also detected in one Group A patient. The group labels therefore need to retain their sampling-time meaning: ethanol detection at assessment is different from a complete history of coexposure.
What the severity score measured
The Stimulant Intoxication Scale, or SIS, combined heart rate, blood pressure, orientation, mental state and temperature. It was applied during the initial assessment alongside specimen collection.
The mean SIS was 3.2 in Group A and 4.2 in Group B. Across the whole sample, 74 patients were classified as mildly intoxicated and seven as moderately intoxicated. The highest observed score was 13; no patient entered the scale’s severe category above 15.
These clinical measurements address a different outcome from lane control, divided attention, reaction time or safe driving. A higher SIS is not a quantified increase in crash risk.
The significant comparison depended on excluding two observations
The comparison using the full groups gave p = 0.103, above the study’s 0.05 significance criterion. This did not establish a statistically significant difference. It also did not demonstrate that the groups were equivalent or that ethanol had no possible effect.
The researchers identified two Group A observations as outliers and repeated the analysis without them. That comparison gave p = 0.008, supporting higher clinical scores in Group B under the revised analysis.
The paper describes the second analysis after discussing the outliers. Its significant finding should therefore be reported together with the initial nonsignificant result and the exclusion decision. The two analyses are not independent replications, and omitting the first would conceal the result’s sensitivity to how observations were handled.
Nor was this a randomized alcohol challenge. Group differences could reflect timing, amounts consumed, patient characteristics or other uncontrolled factors. A significant revised comparison does not by itself isolate a causal effect of alcohol.
Detection and concentration produced different information
Blood collected in EDTA tubes was analyzed for cocaine and cocaethylene by liquid chromatography with diode-array detection. Cocaine was detected in 20 patients and cocaethylene in five. The urine screening used for enrollment measured benzoylecgonine, a cocaine metabolite; it was a different specimen and analytical question.
Cocaethylene detection was more frequent in the blood-ethanol-positive group: four of 19 patients, versus one of 62. The authors reported that this detection-frequency difference was significant. However, the between-group differences in measured cocaine and cocaethylene concentrations were not statistically significant.
A greater proportion of positive results is not the same finding as a higher concentration among positive samples. Only five cocaethylene-positive patients contributed to the concentration observations, which sharply limits individualized interpretation.
Timing limits an earlier-driving inference
The interval between last exposure and hospital arrival could be estimated in only 23 patients. In those patients it ranged from one to 24 hours. Most patients therefore lacked the detailed exposure timeline needed to relate a later specimen to an earlier event.
The paper does not provide quantitative ethanol concentrations for a BAC dose-response analysis. It also does not validate the clinical score or the drug concentrations against driving behavior. A positive cocaethylene result cannot be converted through this dataset into an exact prior BAC, consumption time or degree of driving impairment.
The site’s cocaethylene interpretation guide explains the broader distinction between overlapping cocaine–ethanol exposure and person-specific conclusions. This study adds a narrower evidentiary point: a claim of greater clinical severity must disclose the group definition, small sample and outlier-sensitive analysis that produced it.
Primary source
Zucoloto AD, Eller S, de Oliveira TF, et al. Relationship between cocaine and cocaethylene blood concentration with the severity of clinical manifestations. American Journal of Emergency Medicine. 2021;50:404–408. https://doi.org/10.1016/j.ajem.2021.08.057.