Intoxication is a common clinical presentation and is associated with increased risk of injury, self-harm, and premature mortality. It also complicates clinical care because patients who are intoxicated may provide unreliable histories, display altered pain perception, and exhibit behaviours that increase risk to themselves or others. Despite these challenges, research consistently shows that clinicians are often inaccurate in identifying intoxication or estimating its severity. This narrative review synthesises the literature on clinician recognition of alcohol and other drug intoxication to identify challenges that contribute to misclassification, and potential solutions to identifying intoxication in healthcare settings. Misclassification is most common at mild and moderate levels, where behavioural signs are subtle and easily masked by individual variability, tolerance, or co-occurring conditions. Severe intoxication is generally recognised more reliably, but this provides limited support for early risk detection or safe decision-making across the full clinical spectrum. Structured assessment tools have been developed to improve consistency, yet most show weak relationships with biological measures, have limited applicability across substance types, or have poor feasibility in routine practice. No widely adopted clinical tool exists, and accuracy does not consistently improve with professional experience or seniority. Evidence on the detection of non-alcohol substance intoxication in healthcare settings remains limited. Most available data comes from law enforcement contexts, where structured procedures and extensive training improve performance. These approaches do not translate easily to healthcare environments characterised by rapid patient turnover, competing demands, and limited capacity for lengthy assessments. Clinicians therefore continue to rely primarily on observational judgement, even though behavioural cues vary substantially and are prone to bias.

This review integrates findings across alcohol and other drug classes, identifies common sources of perceptual and decision error, and outlines the implications of misclassification for patient safety, informed consent, diagnostic accuracy, and service utilisation. Improving recognition of intoxication will require investment in clinician training, development and validation of practical assessment tools, and targeted research examining accuracy across diverse clinical contexts. These steps are essential for safer and more consistent responses to intoxication in healthcare.