The utility of visual appearance in predicting the composition of street opioids.

Background: With the emergence of unregulated fentanyl, people who use unregulated opioids are increasingly relying on appearance in an effort to ascertain the presence of fentanyl and level of drug potency. However, the utility of visual inspection to identify drug composition in the fentanyl era has not been assessed. Methods: We assessed client expectation, appearance, and composition of street drug samples being presented for drug checking. Results of a visual screening test were compared to fentanyl immunoassay strip testing. We calculated sensitivity, specificity and likelihood ratios (LR) to assess the accuracy of the common assumption that samples with a "pebbles" appearance contain fentanyl. Results: In total, of the 2502 unregulated opioid samples tested, 1820 (73.5%) appeared as "pebbles", of which 1729 (95.0%) tested positive for fentanyl for a sensitivity of 75.9% (95% Confidence Interval [CI]: 74.2-77.6) and specificity of 59.4% (95%CI: 57.5-61.3). Although, the odds of samples containing fentanyl was 4.60 (95%CI: 3.47-6.11) times higher among pebbles samples compared to non-pebble samples, the positive LR for pebbles to contain fentanyl was only 1.87 (CI: 1.59-2.19). The negative LR was more useful at 0.41 (95% CI: 0.36-0.46). Conclusions: A positive screening test for pebbles is not strongly enough associated to be used as a proxy for detecting fentanyl. While the absence of the appearance of pebbles does somewhat reduce the likelihood of fentanyl being present in a given sample, the high prevalence of fentanyl and fentanyl analogues in the drug supply and the risks of consumption are such that public health providers should routinely advise people who use unregulated opioids against solely relying on visual characteristics of drugs as a harm reduction strategy.