Journal of Addictive Behaviors,Therapy & RehabilitationISSN: 2324-9005

Reach Us +44 190 292 8240
All submissions of the EM system will be redirected to Online Manuscript Submission System. Authors are requested to submit articles directly to Online Manuscript Submission System of respective journal.

Research Article, J Addict Behav Ther Rehabil Vol: 3 Issue: 3

Psychometric Properties of Alcohol Smoking and Substance Involvement Screening Test (Assist V3.0) Among University Students

Peter Olutunde Onifade1*, Abidemi Olubunmi Bello2, Olumide Abiodun3, John O.Sotunsa4, Oluwakemi Anike Ladipo5 and Ocheze Adesanya6
1Drug Addiction Treatment, Education and Research Unit, Neuropsychiatric Hospital, Aro, Abeokuta, Ogun State, Nigeria
2Department of Medicine, Babcock University Teaching Hospital, Ilishan, Ogun State, Nigeria
3Department of Community Medicine, Babcock University Teaching Hospital, Ilishan, Ogun State, Nigeria
4Clinical Services Division, Babcock University Teaching Hospital, Ilishan, Ogun State, Nigeria
5Senior Medical Laboratory Scientist, Babcock University Teaching Hospital, Ilishan, Ogun State, Nigeria
6Medical Social Worker, Babcock University Teaching Hospital, Ilishan, Ogun State, Nigeria
Corresponding author : Dr. Onifade P.O
Neuropsychiatric Hospital, Aro, PMB 2002, Abeokuta, Ogun State, Nigeria
Tel: +2348035061082
E-mail: [email protected]; [email protected]
Received June 19, 2014 Accepted September 05, 2014 Published September 09, 2014
Citation: Onifade PO, Bello AO, Abiodun O, Sotunsa JO, Ladipo OA et al., (2014) Psychometric Properties of Alcohol Smoking and Substance Involvement Screening Test (Assist V3.0) Among University Students. J Addict Behav Ther Rehabil 3:3. doi:10.4172/2324-9005.1000126


Psychometric Properties of Alcohol Smoking and Substance Involvement Screening Test (Assist V3.0) Among University Students

Background: Urine drug test was in use among the undergraduates in the study area for 4 years. The World Health Organization’s Alcohol Smoking and Substance Involvement Screening Test (ASSIST) was introduced recently. This study aimed to determine the reliability of the self-report and its validity against urine drug test. Methods: This study of diagnostic accuracy was conducted among students of Babcock University, Nigeria, in 2013. Each student had urine drug test and interview with the use Alcohol Smoking and Substance Involvement Screening Test (ASSIST) on the same day. The laboratory officers and the interviewers were blind to the results of each other. Results: The 2797 participants were mostly 18-20 years (61.2%) and females (65.1%), Urine of 0.1% tested positive to cannabis and Methamphetamine, 0.4% to Opiates. The three-month selfreport gave the prevalence rates of Amphetamine Type Stimulants, Opioids, Diazepam, Cannabis and cocaine at 1.2%, 2.6%, 1.4%, 1.0%, and 0.3% respectively. Against the urine drug test, ASSIST had low sensitivity and high specificity. Its diagnostic accuracy was greater than 95%. Eleven domains of ASSIST had internal correlation coefficients of greater than 0.7. Conclusions: The ASSIST version 3 has acceptable psychometric properties and is valid for use among university students.

Keywords: ASSIST, internal consistency; sensitivity; specificity; predictive value; diagnostic accuracy


ASSIST; Internal consistency; Sensitivity; Specificity; Predictive value; Diagnostic accuracy


ASSIST: Alcohol Smoking and Substance Involvement Screening Test; UDT: Urine Drug Test; GTLR: Greater than Low Risk of Harm from Use of Drugs


Prevalence rates of psychoactive substance use vary from country to country, but it is estimated that 243 million (5.2%) of the world population aged 15-64 had used an illicit drug at least once in the previous year [1]. At least 85 million adult Europeans have used an illicit drug at some point in their lives, representing around a quarter of Europe’s adult population. An estimated 15.4 million (11.7%) of young Europeans aged 15–34 years used cannabis in the last year. Cocaine, amphetamines and ecstasy are the most commonly used illicit stimulants in Europe. [2] Among students aged 13-18 years in South American countries, lifetime prevalence of alcohol is 39.0- 78.2%, Cannabis is 3.6-16.6%, inhalants is 2.67-16.55% and Cocaine is 0.90-3.96 [3]. In Germany, more than one-quarter of all young people have had experience with cannabis and an estimated 150 000 people are dependent on heroin. The main drug of choice in China is heroin - 87.6% of drug users abuse it [4]. A recent household survey of 6752 representative sample of Nigeria estimated the lifetime prevalence rates alcohol, tobacco, sedatives, stimulants, and cannabis to be 58%, 17%, 14%, 2.4% and 3% respectively [5]. A more recent national household survey of alcohol and drug use in Nigeria reported the prevalence rate of alcohol as 39.0% lifetime, 30.3% one-year and 24.5% past 30 days. Past 30-day prevalence rates of other commonly used drugs are solvent /inhallant (3.2%), Tranquilizer (2.9%), Opiates other than Heroin (2.2%), Heroin (1.8%), Cannabis (1.8%), and Cocaine (1.4%) [6]. In 3 Nigerian Universities, stimulants other than the Amphetamine-types has the highest lifetime prevalence rate (53.4%), followed by alcohol (35.8%), tranquilizers (12%), opiates other than Heroin (11.9%) and cigarettes (11.3%). Lifetime prevalence rates of cannabis, cocaine and heroin are 2.1%, 2.1% and 7.2% respectively [7].
There are harmful consequences of drug use. An estimated 183,000 drug-related deaths were reported in 2012, corresponds to a mortality rate of 40.0 deaths per million among the population aged 15-64 [1]. Drug use is one of the major causes of mortality among young people in Europe, through overdose and drug-related diseases and accidents, violence and suicide, giving excess mortality that is 10 to 20 times greater than expected [2]. Among the population of the world aged 15-19 years, alcohol use is the 5th highest contributor of global burden of disease, causing the loss of 34 equivalent years of full health per 100,000 population. Loss of 7 equivalent years of full health per 100,000 is attributable to drug misuse among males in that same age group [8]. Among US Latina female adults, 32.3% and 15.2% reported that they were affected by alcohol and drugs during sex respectively. The use of these drugs was associated with HIV risk behaviors [9]. Similarly, some of the harms reported among undergraduates in Nigeria included sex regretted the next day (2.5%), sex without condom (2.1%), damage to objects (2.1%), and poor school performance (1.5%) (7). Use of alcohol is also significantly associated with risky sexual behavior among a community sample in Lagos Nigeria [10].
Though controversial, student urine drug testing (UDT) is a form of intervention towards substance demand reduction [11-13]. In 2009, Babcock University, Nigeria, initiated random student UDT for those, by a number of indices, were suspected to be using substances. In 2012, mandatory student UDT was introduced as an integral part of academic session registration. UDT can detect recent substance use but cannot determine level of involvement with the substance and therefore cannot specify the intensity of intervention needed by those with positive results. The World Health Organization’s Alcohol, Smoking and Substance Involvement Screening Test (ASSIST, version 3) is able to specify level of involvement with substances and the corresponding treatment needs of the individuals [14]. Therefore, it was used in addition to the mandatory UDT in the 2013 academic registration.
The psychometric propertied of ASSIST were determined in 2 phases. The first phase was a multisite study involving 9 countries and patients who were recruited from substance abuse treatment units, general medical settings and psychiatric hospitals. It evaluated the test-retest reliability of version 1. The test–retest reliability coefficients ranged between 0.58 and 0.90. The findings in this phase were used as guide to select smaller set of items that formed version 2 [15]. The phase II study, too, was multisite and evaluated the construct, concurrent and discriminative validity of the ASSIST version 2, which was compared with a battery of standardized, internationally used assessments. The internal consistency of the different domains ranged between 0.77 and 0.94. The sensitivity values of ASSIST (V2) against hair drug test for Cocaine, Amphetamine-Type stimulants, Benzodiazepine and Opioids were between 66 and 91% while specificity values were between 80 and 91%. Results of the principal components analysis in the study suggested weighting and recoding of scores on individual questions. Therefore ASSIST version 3 has established weighting to make it a more accurate screening than version 2 [16].
The ASSIST (v3.0) consists of eight items. The first 7 items cover ten substances: tobacco, alcohol, Cannabis, cocaine, amphetamine type stimulants, inhalants, sedatives, hallucinogens, opioids and ‘other drugs’. Item 1 elicits information about lifetime use of substances. The second item asks about frequency of use during the prior three months. Items 3-5 and 7 elicit information in line with ICD-10/DSMIV diagnostic criteria of substance dependence, namely, strong desire or urge to use; use leading to health, social, legal or financial problems; failure to do was normally expected because of use of substance; and loss of control over substance use, respectively. Item 6 is about friend or relative’s expression of concern about the individual’s use of substances. The last item elicits information about non-medical use of drugs by injection. The concurrent and construct validity properties of ASSIST (version 2) were reported as acceptable [14].
Some studies have reported on the properties of version 3. A study among patients in a university teaching hospital reported Internal Consistency (Cronbach’s) coefficients of 0.91 for the Global Continuum Substance Risk Score and a range of 0.74 to 0.93 for the ASSIST-specific substance scores [17]. In another study in same university teaching hospital but among the elderly, the internal consistency ranged between 0.66 and 0.89 for substance specific domain scores [18]. To the best of authors’ knowledge, there was no published report on the reliability and validity of ASSIST version 3 among undergraduates; and no such report among any population in Nigeria. Therefore, this study aimed to determine the reliability and validity of ASSIST version 3 with respect to Urine Drug Test.

Materials and Methods

Design and ethical consideration
This study of diagnostic accuracy was a part of a larger study to assess substance use treatment needs of students in a Nigerian university. Though the UDT was mandatory for all the students according the university’s policy, the interviewer-administered ASSIST was voluntary and student were included in the study only if they signed the written informed consent. The Babcock University Human Research Ethical Committee (BUHREC) approved the protocol.
Study population and setting
The students of Babcock University, Nigeria, constituted the study population. All of them were eligible. The university is one of the private universities located in the South-Western part of Nigeria, owned and operated by the Seventh Day Adventist Church. It has about 8,500 students.
Sampling and recruitment
The study was a census. The students were recruited in November 2013 during the year’s academic school registration. It was mandatory for all of them to take the urine drug test but recruitment for the ASSIST interview was voluntary.
Data collection
Testing days were assigned to the students based on their hostels and the UDT and ASSIST were administered same day to students who participated in both. The participants were free to decide on which test to take first. If the UDT was done first, the student had no idea of the results before the ASSIST interview. The teams of laboratory officers and the counselors were blind to the results of each other. At the ASSIST interview stations, socio-demographic questionnaire was given to each student to voluntarily complete and was later attached to the respective completed ASSIST questionnaire.
Test methods
Urine drug test: The urine drug test was taken as the reference standard in this study because it was already instituted by the university authority. DRUGCHECK® Dip Drug Test manufactured by Express Diagnostic International Incorporation, USA, was used for the UDT. It is based on the principle of competitive immunochemical reaction between a chemically labeled drug (drug protein conjugate) and the drug or drug metabolites which may be present in the urine sample for the limited antibody binding sites. The test is a one-step immunoassay for the qualitative detection of multiple drugs and drug metabolites in human urine at the following cut off concentrations (ng/ml): Amphetamine (d-Amphetamine) -1000; Barbiturate (Secobarbital) – 300; BUP (Buprenorphine) - 10; BZO (Oxazepam) – 300; COC150 (Benzoylecgonine) -150; Cocaine (Benzolecgonine ) – 300; MDMA (3,4-methylenedioxymethamphetamine) – 500; MET500 (d-Metamphetamine) -500; MET (d-Methamphetamine) – 1000; MTD (Methadone) -300; OPI1300 (Morphine) – 300; OPI (Morphine) – 2000; OXY (Oxycodone) – 100; PCP (Phencyclidine) – 25; PPX (Propoxyphene) -300; TCA (Nortriptyline) -1000; and Cannabis (Tetrahydrocannabinol) - 50. [19] The laboratory officers who administered the urine drug test had training and had been conducting the test for 4 years. At the point of urine sample collection, a chaperon handed a marked sample bottle over to each student in a private room. With measure to protect privacy, the students were expected to fill the bottle while in the line of sight of the chaperon. This was to minimize the possibility of diluting, adulterating or swapping the sample.
ASSIST: The WHO Alcohol, Smoking and Substance Involvement Screening Test (ASSIST, version 3) was used as the index test because it yields the level of treatment needs in addition to lifetime and three-month prevalence of drug use. ASSIST is an intervieweradministered, self-report substance use instrument designed by the World Health Organization for use across a range of countries and cultures at primary health care settings.
Twenty people designated as counselor, with background in psychiatry, community medicine, psychiatric nursing, social works and psychology, were trained to use the ASSIST and administer ASSIST-based brief intervention. They administered the ASSIST over a period of 3 weeks.
Statistical methods
For the internal consistency analysis of ASSIST, its scores were aggregated into 6 domains as determined by the World Health Organization during the ASSIST Phase I reliability study [14]. The domains are:
1. Lifetime Substance Use score (sum of the number of different substances ever used according to Question 1);
2. Global Continuum of Substance Risk score or Total Substance Involvement (sum of response weights to Questions 1-8 across substance classes);
3. Specific Substance Involvement score (sum of response weights to Questions 2-7 within each of the following drug classes: tobacco, alcohol, cannabis, cocaine, amphetamine type stimulants, inhalants, sedatives/sleeping pills, hallucinogens, opioids, other). It is anticipated that this score will be used most often by clinicians and health care workers to estimate risk associated with a specific substance;
4. Current Frequency of Substance Use Score (past 3 months). The frequency of substance use in the last 3 months (*excluding tobacco and ‘other’ drugs) of each individual substance and the sum of all substances combined according to Question 2;
5. Dependence (sum of Questions 1, 2, 3, 6 & 7 across substances);
6. Abuse (sum of Questions 1, 2, 4, 5 & 6 across substances).
All Domains (with the exception of Domain 3) were split into two major groups – those that included all substances, and those that included illicit substances only (alcohol and tobacco excluded). The rationale given for the split by the designer of the instrument was to test the validity of the ASSIST for screening for illicit drugs in the absence of alcohol and tobacco, the use of which may have ‘swamped’ some of the domains. Domain 3 was split into ten, one for each of the ten substances. Supplementary Table 1 shows the details and items in the domains. The table also shows the difference on the maximum score obtainable in each domain between ASSIST version 3 and earlier versions. The difference is occasioned by the modification to the response options and scoring. In the Phases I and II of ASSIST (V1.0, V2.0, and V2.1) validation, the instrument was scored using simple Likert scoring categories which were identically weighted for similar questions. That is, Q1 (0, 1); Q2, Q3, Q4 and Q5 (0, 1, 2, 3, 4); Q6, Q7 and Q8 (0, 1, 2). The response and scoring changes made in version 3 were Q1 (0, 3); Q2 (0, 2, 3, 4, 6); Q3 (0, 3, 4, 5, 6); Q4 (0, 4, 5, 6, 7); Q5 (0, 5, 6, 7, 8); Q6 and 7 (0, 6, 3); and Q8 (0, 2, 1). The intraclass correlation coefficient (ICC) was used to determine the internal consistency in each of the domains. ICC yields same figure as Cronbach’s alpha for dichotomous variables [20]. In addition, it yields the correlations coefficient the items and the p-value.
Table 1: Socio-demographic variables.
For diagnostic accuracy of ASSIST with reference to the UDT, the following psychometric properties were determined: Sensitivity (percentage of UDT positives picked as positive by ASSIST for a specific substance), Specificity (percentage of UDT negatives picked as negative by ASSIST for a specific substance), Positive Predictive Value ( percentage of those positive on ASSIST who had positive UDT for a specific substance), Negative Predictive Value (percentage of those negative on ASSIST who had negative UDT for a specific substance), Diagnostic accuracy (percentage of correct ASSIST results for a specific substance), and Diagnostic odds ratio ( the odds of ASSIST having positive results against the odds if it having negative results for a specific substance). The formulae for calculating the properties are available [21].


Though 5938 students had ASSIST interview, only 2797 could be matched with the UDT. Table 1 shows the socio-demographic variables of the participants with matched results. Most of them were aged 18-20 years (61.2%), females (65.1%), Yoruba (63.4%) and Christians (91.9%). The rates of positive drug urine test to Cocaine, Codeine, Amphetamine, and Benzodiazepine were 0% respectively. Cannabis was 0.1%; Methamphetamine, 0.1%; and Opiates, 0.4%. ASSIST based lifetime prevalence rates of Amphetamine Type Stimulants, Opioids, Sedatives (Diazepam), Cannabis and cocaine were 4.7%, 4.3%, 3.2% 3.0% and 0.5%, respectively. The 3-month prevalence rates were 1.2%, 2.6%, 1.4%, 1.0%, and 0.3% respectively.
Table 2 shows the psychometric properties of ASSIST against UDT. Neither the lifetime nor the 3-month item of ASSIST picked gave a positive result for any of the students with positive UDT for Cannabis Amphetamine Type Stimulant or Benzodiazepam, giving sensitivity of 0%. Lifetime and 3-month items for opiates use were able to detect 11% and 5% of students with UDT positive for opiates. On the other hand the specificity of ASSIST’s lifetime and 3-month items on the other hand was greater 90% for all the respective substances. Similarly, while the Positive predictive value was very low, Negative predictive value was very high. The diagnostic accuracy of lifetime and 3-month items was greater than 95% for respective substances. The odds of ASSIST’s 3-month item detecting a student as using opiates was twice that of UDT. The odds ratio was zero for other drugs.
Table 2: ASSIST Versus UDT.
Table 3 shows the psychometric properties of UDT at determining students who are with greater than low risk of harm (GTLR) from use of drugs. Except with respect to Opiates, none of the students classified as having GTLR by ASSIST had positive UDT; thus the sensitivity of UDT to detect GTLR was 3.7% for opiates and 0% for other drugs. The specificity however ranged between 99.3% and 100%. The Negative Predictive Values and Diagnostic accuracy of UDT with respect to GTLR were also greater than 90%. The diagnostic odds ratio for opiates was 5.7.
Table 3: Properties of Urine Drug Test at detecting students with Risky level of drug use.
Table 4 depicts the internal consistency properties of the various domains of ASSIST. The correlation of the items in each domain was significant (p<0.001). Eleven domains had ICC of greater than 0.7; 4 had between 0.4 and below 0.7. The domains of Lifetime illicit drug use (0.398), Hallucinogens involvement score (0.391), Total Current Frequency of Substance (0.224), and Total Current Frequency of Illicit Drug Use (0.179) had ICC of less than 0.4. Some of the items in some of the domain were excluded from the internal consistency analysis because that had zero variance (see legend of Table 4). These items were Q4d (three-months health, social, legal or financial problems due to cocaine), Q4f (three-months health, social, legal or financial problems due to inhalants), Q4h (three-months health, social, legal or financial problems due to hallucinogens), Q5d (three-months failure to do what was normally expected due to use of cocaine), Q5h (three-months failure to do what was normally expected due to use of hallucinogens), Q6f (anyone expressing concern about repondent’s use of inhalants), Q6h (anyone expressing concern about repondent’s use of hallucinogens) and Q7h (loss of control over the use of hallucinogens).
Table 4: Internal consistency of each ASSIST domain.


This study determined the internal reliability of ASSIST (v3) and its specificity, sensitivity and diagnostic accuracy of ASSIST with respect to UDT.
Rate of positive UDT
The rate of positive UDT in this study was lower than the self report rate of 3-month drug use for each substance. The lower rate of UDT was reported in a similar study with ASSIST [22] and another study on concordance of a different drug use self-report and UDT [23]. Keeping in mind that urine drug screening had been taking place among the study population for about four years and that the students knew ahead of time that the UDT would be part of the academic registration, the lower rate is not unexpected because a negative UDT result may mean that the client had never used the drug detectable by the test; or had used but not recently enough or not frequently enough or not at sufficient dose to be detected by the test; or had diluted the urine prior to collection by drinking large volume of water; or had diluted it at the time of collection by adding water; or tampered with it by adding chemicals like table salt; or had switched with a sample by person not using the drug; or the test might just not be sufficiently sensitive to detect the drug in urine though present in the urine at quantity detectable by a different test. [24] Some people believe that negative results could be produced by ingesting Niacin [25] or vinegar [26]. In addition, there are commercial products and instructions to help drug users obtain negative UDT [27]. We implemented measures to ensure the integrity of the urine sample but one could not be 100% certain that the samples were not diluted adulterated or swapped.
Self- report of drug use
The lifetime self-report in this study had lower rates than that of a recent study among undergraduates in the UK [28]. The 3-month self-report had lower rates than those reported in some studies of substance use among university and non-university youths [22,29]. But the rates were higher, except in case of cannabis and cocaine, than the rates of current drug use observed in 2009 in three universities, one of which is the same as this study population. The other two universities were also based in the South-Western region of Nigeria [7].
Diagnostic accuracy
Though the sensitivity of ASSIST to determine those who were UDT positive ranged between 0 and 10% the specificity and Negative predictive values were greater than 90% for all the substances. The diagnostic accuracy (concordance) rates for the substances were greater than the 79% to 87% reported in a recent similar study of ASSIST versus urine toxicology among youths in US [22].
When classification of ASSIST into the levels of risk is taken as the reference, UDT had very low sensitivity to detect those with problematic drug use who therefore need secondary treatment. When a positive UDT is valid, it might be due to a one time use, an intermittent use, a chronic use, an abuse, drug dependence, or use by prescription. [24] In other words, a valid positive UDT result is not synonymous with substance use disorder or the need for secondary treatment.
Internal consistency
All the ASSIST domains had statistically significant correlation of items and most of them have Intra-class correlation coefficients comparable to the Cronbach’s alpha reported in the WHO validation study of ASSIST (version 2) [14] and the French version of version 3 [17,18]. The Intra-class correlation analysis in this study suggested the exclusion from the instrument of question 4 on cocaine, inhalants and hallucinogens, question 5 on cocaine and hallucinogens, question 6 on inhalants, and hallucinogens and question 7 on hallucinogens. From the face value, these items are more applicable in a population with significant proportion of people with high level of involvement with the substances. In a way, their zero variance in this population point to construct validity of ASSIST: students with higher level of involvement with substances, tending towards dependence, were more likely to be absent from school. Future studies in similar population covering wider geographical and cultural setting may or may not confirm the need to exclude the items from ASSIST meant for university students.


The study was limited to only one study location. This limits the generalizability of the findings to the entire population of university students in Nigeria and beyond.


The large sample size and the co-administration of biological screening test and ASSIT are strengths of this study.


The ASSIST version 3 has acceptable psychometric properties and is valid for use among university students. Because it is able to detect students who had not used drug recently enough to be picked by UDT and because it is able to determine level of risk and treatment needs of the students, it is recommended as an essential part of drug use screen programme in the university. It is also recommended that UDT, or better still, hair drug analysis be done alongside ASSIST administration.


Track Your Manuscript

Share This Page