The aim of this study was to assess the diagnostic capability of self-reported psychoactive substance use among patients admitted to psychiatric consultation or emergency department in Benin. To achieve this, the frequency of psychoactive substance use in the lifetime and over the last three months of these patients were determined.
The frequency of self-reported substance use in this study was 81% confirming also the high frequency rate of substance use among people with mental health disorders in our study population. Surprisingly, the frequency reported here is higher than that found elsewhere: 57% in psychiatric emergencies in Philadelphia , and 68.5% in Mwanza, Tanzania . The increase in the availability of different substances over time could be one of the major reasons for the exponential growth in substance use.
The pattern of substance use revealed that the most frequently used substance among patients were alcohol (71.3%), tobacco (45.5%) and cannabis (26.8%). Similar profiles of substance use were observed among psychiatric patients in Burkina Faso , Tanzania  and US . Licit nature of alcohol and tobacco explained their predominance. In the category of illicit substances, cannabis ranked first. The availability and relatively lower cost of cannabis explains its high consumption compared to other illicit substances.
Urine test by NarcoCheck kit was positive in 58.6% of participants, which reflected that the urine test rate was higher than the self-reported substance use (52.2%). The urine test rate found in our study is lower than that reported in a study conducted in the US . This difference could be accounted for by the fact that in our study, a systematic urine test was performed for all participants reported substance use over the past three months, in contrast to the US study which performed the urine test only for participants who reported the use of substance over the past three days. Cannabis (THC) was the most detected substance (35.7%). Several studies have also reported high frequencies of urinary detection of cannabis compared to other drugs [2, 25, 26]. However, the frequency of urine test of cannabis was higher than the frequency of self-reporting use (35.7% versus 26.8%). There could be several reasons for the difference between reporting frequency and urine testing. Patients may not want to disclose their use. Moral, socio-cultural and legal restrictions on cannabis use may deter some patients to disclose their use. The stigma associated with use may also reduce the willingness to disclose their use. In addition, the long elimination period of cannabis allows for easy detection, even in subjects whose use is three months old. Contrastively, there was poor detection of alcohol by urine test in this study (21.7 versus 36.3 for self-reporting). This is due to the fact that ethylglucuronide is rapidly eliminated from the body, making it difficult to detect in those who have consumed alcohol for more than three days [2, 22]. As a result, urine alcohol tests cannot be used as reference tests. The questionnaire is still of interest in alcohol screening, especially as it provides an opportunity for a nuanced discussion between the patient and the clinician about alcohol consumption. However, the detection of ethylglucuronide in urine can be used as a relapse control marker allowing the clinician to intervene at an early stage .
The analysis of the different tests of validity of self-reporting allows to classify the substances into three different categories. In the first category, cocaine and tobacco come first. The sensitivity and specificity observed for these two substances are high, indicating an interesting intrinsic performance for self-reporting using the ASSIST questionnaire.
Cannabis, opioids, anxiolytics and alcohol are in a second category where the sensitivity ranges from 14 to 41%. All of them have a high specificity above 95% except anxiolytics. The kappa coefficient between self-reporting and urine testing for this second category ranges from 0.21–0.40 indicating moderate agreement. Urinalysis for this category adds valuable value to self-reporting. It will help to clarify the diagnosis and to set up a therapeutic follow-up adapted to the incriminated substance.
In a third category occurs alcohol and hallucinogens, substances for which the kappa is close to 0 indicating no agreement. Here, urine tests cannot be used as a reference and other approaches for the detection of those substances should be sought. Against this backdrop, self-reporting provides more information than urine test.
The overall analysis of the results shows that all substances have high negative predictive values. This indicates that patients who declare not to use drugs are telling the truth. Urine test is more recommended for patients who report using at least one substance for further investigation. The positive predictive values found for tobacco, cannabis, cocaine and anxiolytics were all above 60% suggesting good consistency between positive self-report and positive urine test. Similar positive predictive values for tobacco, cannabis, cocaine and anxiolytics have been found in other studies [26, 28, 29]. The extrinsic performance of the self-report, in this case the ASSIST questionnaire, will depend on its intrinsic informative capacity and the context of use.
These different contexts of use explain that several researches found various degrees of concordance between self-reported substance use and urine test [11, 15, 17, 25, 30]. In line with literature, it should be noted that the accuracy of self-reports depends also upon the characteristics and nature of the population involved . In our study, the consistency between self-reported substance use and urine test was moderate. Urine tests are necessary to complete the self-report in order to clarify the diagnosis and to monitor the patient’s treatment. However, in some cases, self-reporting provides more valuable information than urinalysis indicating that the preeminence of each measure will depend on the scenario and the substance. The development of flowcharts integrating different measurement tests ranging from self-reporting to different bioassays for different substances and scenarios will be of paramount support to clinicians and toxicologists in making the appropriate decisions.
By way of conclusion, this study exhibits a moderate diagnostic capability of self-reported psychoactive substance use among patients admitted to psychiatric consultations or emergencies in Benin, suggesting that self-report may not estimate the exact prevalence of substance use. Optimal identification of psychoactive substances use in psychiatric patients requires both history and urine testing. The integration of these two approaches is an excellent method to find out the level, frequency and nature of drug used. A careful integrated interpretation of the two measures is therefore required in psychiatric patients and in general population investigation.