The CRAFFT Substance Use Assessment (CRAFFT 2.1+N) is a self-report screen for substance use, substance-related riding and driving risk, and risk of substance use disorder in young people aged 12 to 21 (Knight et al., 2002; CABHRe, 2021). Branching means a young person answers between 5 and 20 questions. The result places them in a Low, Medium, or High risk category, with riding and driving risk and nicotine use reported alongside the score.
Adolescence is the period when substance use most commonly begins, and early use carries particular risks. Motor vehicle crashes are a leading cause of death among young people, and riding with an impaired driver is a danger even for those who do not use substances themselves. Nicotine dependence can also develop quickly in these years, with loss of autonomy over nicotine emerging well before daily use is established. The American Academy of Pediatrics recommends routine substance use screening as part of adolescent preventive care, and brief, systematic screening gives clinicians an opportunity to identify risk early and open a supportive conversation before difficulties escalate (CABHRe, 2021). The CRAFFT was developed at Boston Children’s Hospital and is among the most widely studied adolescent substance use screens available.
The CRAFFT 2.1+N has three parts. Later questions are shown only when they are relevant, so a young person answers between 5 and 20 questions.
Part A asks on how many days in the past 12 months the young person used:
The answers to Part A determine which of the remaining questions are shown.
Part B contains the six CRAFFT questions that produce the score. The name is a mnemonic for them:
The Car question is always asked, so every young person is screened for riding and driving risk. The other five are answered only by those who report any use of alcohol, marijuana, or anything else to get high. A young person who reports only vaping or tobacco use answers the Car question and then moves to Part C.
Part C is the Hooked on Nicotine Checklist (HONC; Wheeler et al., 2004), a 10-item follow-up completed only when any vaping or tobacco use is reported in Part A. It asks about signs of diminished autonomy over nicotine, such as cravings and difficulty quitting, and is scored and reported separately from the CRAFFT total.
Routine screening. The CRAFFT 2.1+N is designed for routine, confidential screening of young people in primary care, mental health, school-based health, and emergency settings. The young person completes it in private, and the risk category orients the clinician to an appropriate level of follow-up (CABHRe, 2021).
Safety screening. The Car question flags riding and driving risk even in young people who report no substance use of their own, which gives the clinician a natural opening for a safety conversation.
Opening a conversation. The CRAFFT is designed to start a discussion as well as to classify risk. The report reproduces the official CRAFFT score interpretation chart, which the CRAFFT authors intend to be shown to the young person as part of feedback (CABHRe, 2021).
The CRAFFT 2.1+N produces a total score, a risk category, and up to two clinical flags (riding and driving risk, and nicotine). The total score is the number of Yes answers across the six CRAFFT questions in Part B. The four Part A frequency questions decide which questions each young person answers and do not contribute to any score. A young person who reports no substance use in the past 12 months answers only the Car question, so their maximum possible score is 1. One who reports any use answers all six, giving a range of 0 to 6. Higher scores indicate a greater likelihood of substance-related problems.
NovoPsych implements the self-administered CRAFFT 2.1+N and applies the official screening cut score of 2 or more (Knight et al., 2002) across the full 12 to 21 age range.
When any vaping or tobacco use is reported, the ten HONC questions are also presented. They are scored as a separate count of Yes answers from 0 to 10 and do not contribute to the CRAFFT total.
The total score reflects the likelihood of a disorder, and the risk category classifies the result using the fuller pattern of responses. Following the CRAFFT manual, each result is placed in one of three categories based on whether any substance use was reported in the past 12 months, the response to the Car question, and the total score (CABHRe, 2021).
| Risk category | How it is reached | Total score | What the result offers |
|---|---|---|---|
| Low risk | No substance use in the past 12 months and No to the Car question | 0 | An opportunity to reinforce and encourage the young person’s healthy choices. |
| Medium risk | No substance use in the past 12 months and Yes to the Car question, or any use in the past 12 months with a total score of 0 or 1 | 0 to 1 | An opportunity for brief advice about the risks of substance use and of substance-related riding and driving. |
| High risk | Any use in the past 12 months and a total score of 2 or more | 2 to 6 | A positive screen, indicating a substantially elevated likelihood of a DSM-5 substance use disorder that likely warrants further follow-up. A positive screen is not a diagnosis. |
Higher CRAFFT 2.1+N scores are associated with a greater likelihood of a DSM-5 substance use disorder. In the validation study, the percentage of adolescents meeting criteria among those scoring at or above each point rose with the score (Mitchell et al., 2014). This is why a score of 2 or more is treated as a positive screen.
| CRAFFT 2.1+N score | Adolescents meeting DSM-5 substance use disorder criteria |
|---|---|
| 1 or more | 32% |
| 2 or more | 64% |
| 3 or more | 79% |
| 4 or more | 92% |
| 5 or more | 100% |
| 6 | 100% |
These values are cumulative and describe adolescents in the validation sample with a similar score. They do not give an individual young person’s chance of having a disorder.
The Car question is answered by every respondent, regardless of their own substance use, and works as a safety screen in its own right. It asks about the young person’s lifetime rather than the past 12 months. A Yes response flags reported exposure to substance-impaired driving, one of the leading causes of death for young people, and moves a young person who reports no substance use from Low to Medium risk. The CRAFFT authors recommend pairing this conversation with their Contract for Life resource, available at crafft.org (CABHRe, 2021).
When any vaping or tobacco use is reported in the past 12 months, the HONC is administered and reported separately from the CRAFFT 2.1+N results. Even one Yes answer on the HONC suggests a serious problem with nicotine that warrants consideration of further assessment (CABHRe, 2021).
For respondents aged 18 or older, total scores of 2 or 3 are weaker evidence of a substance use disorder than the same scores in younger adolescents. When a respondent in this age group scores 2 or 3, the report notes that the result should be interpreted with additional care. The evidence behind this is set out under Clinical Cut Score in the Psychometrics tab.
Because the questions ask about the past 12 months, the CRAFFT 2.1+N is a periodic screen, typically administered about once a year. A CRAFFT result predicts substance use disorder most accurately about one year later, with accuracy fading over longer intervals (Shenoi et al., 2019). Each administration is interpreted as a fresh screen, and the report does not calculate change between administrations.
The report displays up to two charts. The first shows the total score as a single bar against the 0 to 6 range, with the regions below and at or above the cut score shaded and labelled. The second reproduces the official CRAFFT score interpretation chart, showing the percentage of adolescents meeting DSM-5 substance use disorder criteria at or above each score from 1 to 6, with the respondent’s score highlighted (Mitchell et al., 2014). No percentage is published for a score of 0, so the second chart is left out when the total score is 0.
Knight et al. (1999) developed the CRAFFT by selecting candidate questions from three established adolescent screening instruments (the RAFFT, the Drug and Alcohol Problem Quickscreen, and the POSIT) and testing them against the Personal Involvement with Chemicals Scale in 99 adolescent clinic patients aged 14 to 18. The six questions whose combined score most closely matched this benchmark (r = .84) were arranged into the CRAFFT mnemonic. The main validation study followed in 538 adolescent clinic patients aged 14 to 18, establishing the cut score of 2 or more that is still in use (Knight et al., 2002).
The CRAFFT 2.1 replaced the original yes or no opening questions with questions about the number of days each substance was used in the past 12 months, to reduce underreporting. The CRAFFT 2.1+N then added the vaping and tobacco frequency question and the HONC follow-up (CABHRe, 2021). The six scored questions are unchanged across these versions.
The six CRAFFT questions showed internal consistency of α = .68 in both original validation samples (Knight et al., 1999, 2002), a level consistent with a deliberately brief screen that samples varied content. Across 15 years of later studies, the developers report internal consistency from .67 to .85 (mean .74) and test-retest reliability from .77 to .98 (mean .88; CABHRe, 2021). An independent review of the 1999 to 2010 literature reported internal consistency between .65 and .86 across settings (Dhalla et al., 2011). These estimates come from earlier CRAFFT versions. Because the six scored questions are unchanged, the developers apply this evidence to the CRAFFT 2.1 and 2.1+N (CABHRe, 2021).
Criterion validity against structured diagnostic interviews is the CRAFFT’s central evidence base. In the primary validation study, a score of 2 or more identified any DSM-IV substance abuse or dependence diagnosis with a sensitivity of .80 and specificity of .86, and dependence with a sensitivity of .92 and specificity of .80. Areas under the curve ranged from .90 to .93 (Knight et al., 2002).
DSM-5 replaced the separate abuse and dependence diagnoses with a single substance use disorder graded as mild, moderate, or severe (American Psychiatric Association, 2013). The cut score has since been re-examined against DSM-5 criteria in two studies.
| Study | Sample | Criterion | Sensitivity | Specificity | AUC |
|---|---|---|---|---|---|
| Mitchell et al. (2014) | 525 primary care patients aged 12 to 17 | Any substance use disorder | .91 | .93 | .97 |
| Mitchell et al. (2014) | 525 primary care patients aged 12 to 17 | Moderate to severe disorder | .88 | .87 | .95 |
| Shenoi et al. (2019) | 2,143 emergency department patients aged 12 to 17 | Mild disorder | .84 | .93 | .90 |
| Shenoi et al. (2019) | 2,143 emergency department patients aged 12 to 17 | Moderate to severe disorder | 1.00 | .92 | .98 |
Shenoi et al. (2019) administered the CRAFFT on tablet computers across 16 paediatric emergency departments, the largest criterion validation to date. Baseline scores also predicted diagnoses one to three years later, with predictive accuracy strongest at one year.
In a within-person comparison, 136 primary care patients aged 12 to 17 completed the CRAFFT both on a computer and with a clinician, in random order. Sensitivity and specificity were high for both and did not differ between modes, while self-administration was faster (49 compared with 74 seconds) and elicited more disclosure on the Car question (Harris et al., 2016). Across 201,134 screenings in 314 United States practices, electronic self-administration was associated with higher disclosure of substance use than clinician interview in every age band from 12 to 20, with the largest difference among 12 to 13 year olds (Gao et al., 2025). The developers recommend the self-administered version whenever possible (CABHRe, 2021), and NovoPsych implements the CRAFFT in this format.
The cut score of 2 or more was derived in the primary validation by maximising the product of sensitivity and specificity, and it was optimal for every criterion examined (Knight et al., 2002). The same threshold performed optimally against DSM-5 criteria in both later validations (Mitchell et al., 2014; Shenoi et al., 2019).
The developers advise that a higher cut score may be more appropriate for respondents aged 18 or older (CABHRe, 2021). In 382 adults aged 18 to 25 recruited for past-month alcohol or cannabis use, specificity at the standard cut score was low (13 to 19 percent), and a score of 4 or more gave the best balance of sensitivity and specificity (Bagley et al., 2017). Because no alternative threshold has been validated for routine use in this age range, NovoPsych retains the official cut score across the 12 to 21 range and notes in the report that scores of 2 or 3 in respondents aged 18 or older are weaker evidence of a disorder.
A positive screen (a High risk result) means the young person’s pattern of responses places them in a group with a substantially elevated likelihood of a substance use disorder. A more thorough conversation or assessment is usually the next step, which may involve taking a detailed substance use history, conducting a structured diagnostic interview, or referring to a service with adolescent substance use expertise. The CRAFFT manual provides brief counselling talking points that many providers use to begin the conversation straight after reviewing the result. A positive screen is not a diagnosis, and some young people who screen positive will not meet criteria for a substance use disorder once further assessed.
Motor vehicle crashes are one of the leading causes of death for young people, and riding with an impaired driver is a risk even for adolescents who do not use substances themselves. The Car question is therefore asked of every respondent, whatever their own use, and a Yes response is flagged in the report. For a young person who reports no substance use, a Yes moves the result from Low to Medium risk. The flag creates a natural opening for a safety conversation, and the CRAFFT authors provide a Contract for Life resource designed for that discussion between young people and their families.
The CRAFFT total score (0 to 6) comes only from the six CRAFFT questions, which ask about alcohol and drug involvement. Like the other opening frequency questions, the vaping and tobacco question is not part of that score. Nicotine is screened separately: any reported vaping or tobacco use brings up the Hooked on Nicotine Checklist (HONC), a 10-item screen for signs of diminished control over nicotine, which is reported alongside the CRAFFT score. Even one Yes on the HONC is treated as meaningful, because loss of autonomy over nicotine can emerge quickly in adolescence, well before daily use is established.
The CRAFFT 2.1+N is designed for ages 12 to 21, and NovoPsych administers it across that full range. Interpretation shifts slightly at the top of the range. Research in adults aged 18 to 25 suggests that scores of 2 or 3 are common among people who use substances without having a substance use disorder, so low positive scores are weaker evidence of a disorder in this age group than in younger adolescents. The report notes this automatically for respondents aged 18 or older. Scores of 4 or more remain strong signals at any age in the range.
Young people tend to disclose more when they answer sensitive questions privately on a screen than when a clinician asks them aloud. A national study of over 200,000 screenings found higher disclosure with self-administration in every age group, with the largest difference among 12 to 13 year olds, and a direct comparison found self-administration just as accurate as clinician interview and faster. The CRAFFT developers recommend self-administration whenever privacy can be protected. Clinicians should still make sure the young person can respond without a parent or carer seeing their answers.
The CRAFFT 2.1+N is a periodic screen, so each administration stands alone as a fresh snapshot of risk. The strongest evidence supports re-screening about once a year: a large prospective study found that a CRAFFT result predicted substance use disorder most accurately about a year later, with accuracy fading over two and three years. Re-screening sooner makes sense whenever circumstances change, for example after a significant life event or if new concerns emerge. Differences between administrations reflect changes in screening status and do not measure an amount of improvement or deterioration.
Knight, J. R., Sherritt, L., Shrier, L. A., Harris, S. K., & Chang, G. (2002). Validity of the CRAFFT substance abuse screening test among adolescent clinic patients. Archives of Pediatrics & Adolescent Medicine, 156(6), 607–614. https://doi.org/10.1001/archpedi.156.6.607
© John R. Knight, MD, Boston Children’s Hospital, 2022.
Reproduced with permission from the Center for Adolescent Behavioral Health Research (CABHRe), Boston Children’s Hospital.
[email protected]
For more information and versions in other languages, see http://www.crafft.org.
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). https://doi.org/10.1176/appi.books.9780890425596
Bagley, S. M., Anderson, B. J., & Stein, M. D. (2017). Usefulness of the CRAFFT to diagnose alcohol or cannabis use disorders in a sample of emerging adults with past-month alcohol or cannabis use. Journal of Child & Adolescent Substance Abuse, 26(1), 18–23. https://doi.org/10.1080/1067828X.2016.1175986
Center for Adolescent Behavioral Health Research. (2021). The CRAFFT 2.1 manual (Version 28 October 2021). Boston Children’s Hospital. https://crafft.org
Dhalla, S., Zumbo, B. D., & Poole, G. (2011). A review of the psychometric properties of the CRAFFT instrument: 1999-2010. Current Drug Abuse Reviews, 4(1), 57–64. https://doi.org/10.2174/1874473711104010057
Gao, C., O’Connell, M. M., Howard, B. J., Sturner, R., Shrier, L. A., & Harris, S. K. (2025). Does screening mode matter? A repeated cross-sectional study of computer self-administered vs. clinician-administered screening of youth substance use in pediatric primary care. Frontiers in Adolescent Medicine, 3, 1694040. https://doi.org/10.3389/fradm.2025.1694040
Harris, S. K., Knight, J. R., Van Hook, S., Sherritt, L., Brooks, T. L., Kulig, J. W., Nordt, C. A., & Saitz, R. (2016). Adolescent substance use screening in primary care: Validity of computer self-administered versus clinician-administered screening. Substance Abuse, 37(1), 197–203. https://doi.org/10.1080/08897077.2015.1014615
Knight, J. R., Sherritt, L., Shrier, L. A., Harris, S. K., & Chang, G. (2002). Validity of the CRAFFT substance abuse screening test among adolescent clinic patients. Archives of Pediatrics & Adolescent Medicine, 156(6), 607–614. https://doi.org/10.1001/archpedi.156.6.607
Knight, J. R., Shrier, L. A., Bravender, T. D., Farrell, M., Vander Bilt, J., & Shaffer, H. J. (1999). A new brief screen for adolescent substance abuse. Archives of Pediatrics & Adolescent Medicine, 153(6), 591–596. https://doi.org/10.1001/archpedi.153.6.591
Mitchell, S. G., Kelly, S. M., Gryczynski, J., Myers, C. P., O’Grady, K. E., Kirk, A. S., & Schwartz, R. P. (2014). The CRAFFT cut-points and DSM-5 criteria for alcohol and other drugs: A reevaluation and reexamination. Substance Abuse, 35(4), 376–380. https://doi.org/10.1080/08897077.2014.936992
Shenoi, R. P., Linakis, J. G., Bromberg, J. R., Casper, T. C., Richards, R., Mello, M. J., Chun, T. H., & Spirito, A. (2019). Predictive validity of the CRAFFT for substance use disorder. Pediatrics, 144(2), e20183415. https://doi.org/10.1542/peds.2018-3415
Wheeler, K. C., Fletcher, K. E., Wellman, R. J., & DiFranza, J. R. (2004). Screening adolescents for nicotine dependence: The Hooked on Nicotine Checklist. Journal of Adolescent Health, 35(3), 225–230. https://doi.org/10.1016/S1054-139X(03)00531-7
