Physician recognition of active drug use in HIV-infected patients is lower than validity of patient's self-reported drug use.

A French survey of 325 HIV-infected subjects with a history of injecting drugs allowed us to study the recognition of patients' injection drug use (IDU) by physicians providing HIV-infection care, and to analyze the correlation between patient demographics and incorrect IDU identification. Kappa for concordance of physician's reports of their patient's IDU with patient's declaration was 0.37; concordance was lower among socially vulnerable patients. This contrasted with a nested study of validity of patient's self-report of opioid use: Kappa for patient's declaration of opioid use within the past two days against a biological assay was 0.61, and concordance was higher among socially vulnerable patients. Concordance of physicians' ratings and patients' reports of IDU was not more than fair, even though physicians were knowledgeable about their patient's IDU history. This concordance varied with social status in a way that did not correspond with variations in self-reported opioid use validity, suggesting that identification of active IDU might be partly based on incorrect interpretation of subjective cues.