About the SA Blog Network

Molecules to Medicine

Molecules to Medicine

Demystifying drug development, clinical research, medicine, and the role ethics plays
Molecules to Medicine Home

Drug screens-any more than theater?

The views expressed are those of the author and are not necessarily those of Scientific American.

Email   PrintPrint

Under the influence?

I’ve been doing a lot of traveling recently, and am increasingly disturbed by the growing surveillance society and the misplaced reassurances that are used to assuage the public, coined “security theater” by Bruce Schneier. Here we’ll look at this drama in the context of screening for drugs of abuse. In a later post we’ll look more closely at the parallels between drug screening and the TSA. Both are problematic and neither produces the desired outcomes. We’ll look at why.

There are three key issues to consider: one is the accuracy of the test; the second are ethical concerns; the third, the societal and political context of drug testing.
In reading about clinical trials, one commonly misunderstood, yet basic concept is that of the accuracy of clinical test results—are you measuring what you think that you are? I’ve spoken about this a bit in posts about Orasure’s HIV test and turned again for a quick review to Marya Zilberberg’s “Between the Lines.”⁠ We’ll look at evidence and accuracy in a different context here, with our focus being on understanding false negative and false positive results. Drug screening is used as an example, as such screening is increasingly being used in work and social services settings, and raises similar concerns about accuracy and ethics as we see in clinical trials.

Background rates of drug use

According to the government’s Substance Abuse and Mental Health Services Administration⁠1 (SAMHSA) the rate of drug use in the US among full-time workers averaged 8.2%, ranging from 6.1% among health care workers (HCW) reporting use within the past month compared to 16.9% among accommodations and food workers and 13.7% for construction workers. Overall, rates were higher in those younger than 35. Among HCW, nurses had much higher rates than physicians, whose rates varied by specialty, being highest among anesthesiologists and emergency room staff.

Drug screening

Screening is performed in different settings and for different purposes. For example, there is routine screening of long distance truck drivers and airplane pilots, as they are considered a potential public safety risk. Screening can be done pre-employment or, more commonly, in response to a workplace accident. Toxicology screens are also commonly performed in Emergency Departments when evaluating a patient with an unexplained change in mental status or behavior.

Among full-time workers in the US, 42.9% reported that tests for illicit drug or alcohol use occurred as part of “pre-hire” testing—so more than 47 million adults were subjected to testing as part of the hiring process. (2004) Further, 29.6%, or 32 million full-time workers reported random drug testing at work.⁠ A 2010 study reported about 130 million drug screens⁠. Is this necessary? Does this make us safer?

Obviously, we want tests such as these to be accurate – to provide a valid measure of the state of the “real world”.  There are two ways in which tests such as these can be correct. They can detect true positives and true negatives:  That is, they can give a positive reading when the subject of the testing has used one of more of the drugs that are being tested for, and likewise they can give a negative [or “all clear”] reading when the testing subject has not used drugs within the testing window. They can also err in two ways: They can give a false negative, or a negative reading when the subject has actually used drugs, and they can give a false positive, or a positive reading when the subject of the testing has not used any drug.  The ability of the test to return true positives and avoid false negatives is called the “sensitivity” of the test.  Similarly, the ability of the test to return true negatives and not false positives is called the “specificity” of the test.

Urine testing

Most commonly, urine samples are used. Urine tests are very sensitive, but not very specific—this means that there are likely to be many falsely positive results. So a positive result will generally be followed up with a more expensive and specific urine (and occasionally blood) test being done for confirmation—but even among these tests, none are 100% accurate.

Urine tests have the advantage of showing drug use over a longer period of time; the drugs are also present in higher concentrations in urine than in blood. The advantage of testing blood is that it more often reflects the parent drug, rather than metabolites, and that it reflects more “real-time” use.

The likelihood of these false negative and false positive errors varies, and depends on both the accuracy of the test, and upon the expected [or “real-world”] percentage of true positive results in the group being tested. Due to a quirk of statistics, if the incidence of drug usage in the population under study is low, there will be higher likelihood of false positives. The smaller the number or % of true positives, the higher the rate of false positives.

For screening tests, false positives are estimated to occur in 5 to 10% of tests, and false negatives in 10-15%. For confirmatory tests, data has been more difficult to find. Estimates of accuracy of the drug tests are < 1% for false positive confirmatory tests, and perhaps 5% false negatives, due to where the thresholds are commonly set.* Using this as an example, and an 8% drug usage rate, we would get, in a 1000 tested workers:

Drug present Drug absent total
Test pos 76 9 85
Test neg 4 911 915
total 80 920 1000


About 4 people will be reported “clean,” a false negative, though they use drugs.
Of 920 who did not use drugs, 1%, or ~9 people will show up as pos.
Of 85 people who test positive, 9/85, or 10.6% will erroneously be identified as drug users. So while 1% error rate (false positive) sounds good, 10% of the positives will be false.

What drugs are being tested for?

Testing conducted according to SAMHSA’s guidelines checks for five illicit drugs plus, in some cases, alcohol (ethanol, ethyl alcohol, booze).  These five illicit drugs are:⁠
•    Amphetamines (meth, speed, crank, “Ecstasy” (MDMA))
•    THC (cannabinoids, marijuana, hash)
•    Cocaine (coke, crack)
•    Opiates (heroin, opium, codeine, morphine)
•    Phencyclidine (PCP, angel dust)”

The typical 10-Panel Test also includes:
•    Barbiturates (phenobarbital, butalbital, secobarbital, downers)
•    Benzodiazepines (tranquilizers like Valium, Librium, Xanax)
•    Methaqualone (Quaaludes)
•    Methadone (often used to treat heroin addiction)
•    Propoxyphene (Darvon compounds)

Testing is not typically done for alcohol, hallucinogens, inhalants, steroids, hydrocodone, suboxone, or “bath salts” (an amphetamine like drug).

Drugs firms 'creating ills for every pill'

Sites also tell you the length of time that drugs can be detected in blood or urine.⁠
• Alcohol – 1 oz. for 1.5 hours
• Amphetamines – 48 hours
• Barbiturates – 2-10 days
• Benzodiazepines – 2-3 weeks
• Cocaine – 2-10 days
• Heroin Metabolite – less than 1 day
• Morphine – 2-3 days
• LSD – 8 hours
• Marijuana – casual use, 3-4 days; chronic use, several weeks
• Methamphetamine – 2-3 days
• Methadone – 2-3 days
• Phencyclidine (PCP) – 1 week
(Hair testing has extended detection time.)

While testing might pick up some drug use, a whole cottage industry has developed to sell products to defeat the testing, including urine substitutes, or giving advice as to how to beat the tests.⁠ Thus, an individual with a small amount of knowledge could circumvent scheduled screening. This is why drug screens are random, not scheduled, except for pre-employment.

Approaches to testing

SAMSHA ⁠has two different approaches to testing. One is a “zero tolerance” test, which reduces false negatives. Any level of drug detected is considered a positive (or failed) test. The other is the “cut-off” type of test—a drug may be detected in small amounts, but the test will be reported as negative (passed) if below a certain threshold.

False positives
A major concern for those who don’t use illicit drugs and for whom the repercussions of a positive test are grave, is the issue of false positives. Unfortunately, many drugs cross-react with the test, giving potentially erroneous results, especially on screening tests. For example, the USC’s Keck Hospital⁠ notes “more than 300 over-the-counter drugs and foods that can affect the test (see modified table).

If you take or eat: You could test positive for:
Ibuprofen (common pain reliever) Marijuana, barbiturates, or benzodiazepines
Cold remedies Amphetamine
Hay fever remedies Amphetamine
Nasal decongestants Amphetamine
Diet pills Amphetamine
Sleep aids Barbiturates
DHEA (dehydroepiandrosterone ) Anabolic steroids
Novocaine (used in dentistry) Cocaine
Dietary supplements containing ephedrine Amphetamine
Poppy seeds** Opiates/morphine
Hemp food products Marijuana
Quinolone antibiotics (Levaquin, etc.) Opiates


**Large amounts of poppy seeds—for example, a pastry filled with poppy seeds, not a poppy seed bagel—would need to be consumed to cause a positive test. However, sophisticated testing can discern poppy seeds from opiates.”
A suspected false positive on a screening assay—designed to be highly sensitive but not very specific—is then followed up with testing by another technique. Unfortunately, no test is completely accurate, and even the “gold standard” confirmatory test, gas chromatography-mass spectometry, has false positives and is subject to performer error.

False negatives

In the balancing act, trying to avoid the costly additional testing and potential legalities of a false positive, the threshold set by some labs for reporting a test is high enough that even though drug may be present in low levels, it will not be reported. The Feds are pushing for more stringent reporting.

Other testing nightmares

There are other workplace testing procedures that erroneously punish non-users.⁠ For example, those with “shy bladder”, or those who only void small amounts, have been fired. If urine appears too dilute, employees have been subject to repeat testing and surveillance. Forensic toxicologist Dr. Vina Spiehler ⁠has been particularly critical of “arbitrary cutoffs for creatinine, saying they discriminate against women, vegetarians, the elderly, people who drink lots of water, and people of small body size. Spiehler also argued for greater safeguards for validity testing, since ‘for employees subject to urine drug screening sometimes the penalties for abnormal urine characteristics are more severe than those for the presence of drugs in their urine.’”


This example of drug testing is but one example to illustrate the difficulty in interpreting laboratory tests and studies. For each, you have to try and tease out not only the accuracy of the test in measuring the outcome you wish to examine, but also whether you are measuring the right thing or asking the right question.

In an upcoming post, we’ll look at pre-employment screening, its value (or lack thereof), and some of the broader ethical and societal implications of the increasing culture of surveillance.

(*Note: est of false +/- for GC-MS is based on information from toxicologists)

Special thanks to: Dr. David Parish
Ed Smith, Drug Detection Lab
Steve Codd, Beckman Coulter
and Dr. George Lundberg
for additional background info and for helping me understand the complexities of this topic, and to my family, for their thoughtful comments.

“Molecules to Medicine” banner © Michelle Banks
Tried that – Jovike/Flickr
Urine testing – publik15/Flickr
Pills for every ill – publik15/Flickr

Judy Stone About the Author: Judy Stone, MD is an infectious disease specialist, experienced in conducting clinical research. She is the author of Conducting Clinical Research, the essential guide to the topic. She survived 25 years in solo practice in rural Cumberland, Maryland, and is now broadening her horizons. She particularly loves writing about ethical issues, and tilting at windmills in her advocacy for social justice. As part of her overall desire to save the world when she grows up, she has become especially interested in neglected tropical diseases. When not slaving over hot patients, she can be found playing with photography, friends’ dogs, or in her garden. Follow on Twitter @drjudystone or on her website. Follow on Twitter @drjudystone.

The views expressed are those of the author and are not necessarily those of Scientific American.

Rights & Permissions

Comments 14 Comments

Add Comment
  1. 1. paulus 11:01 am 02/19/2013

    How very interesting that alcohol isn’t screened, although it causes more harm than all the others combined. We have more deaths from alcohol induced violence, not to mention the increase in cancer and heart related deaths due to alcohol.
    As a former prison worker, I can attest to the fact that alcohol figures prominently in many murders. During my 14 year stint as a prison worker in California, I had 12 inmates on my crew (out of 24) who had been convicted of murder. Every murder was committed under the influence of alcohol.
    Why can we not see that the reason for the anti-drug crusade was a job-saving measure for G-men in the aftermath of prohibition’s repeal.
    As we see by the article, it is not only alcohol which gets the kid glove treatment, but hydrocodone (commonly known as Vicodin[sp?]). Could the failure to test for it have any relationship to Big Pharma’s contributions to political action committees?
    Why can’t we recognize the need for a harm reduction strategy in the treatment of addiction? We spend billions arresting, trying, and incarcerating people who suffer from a medical condition called addiction.
    We don’t lock up diabetics for their medical condition.
    Let’s stop acting stupidly, get smart on crime, and temper justice with mercy. Addicts need treatment, not jail. This includes alcoholics as well–and the failure to address the elephant in the room only makes the drug testing regimen a display of hypocrisy which fails to address the underlying problem of addiction–whether the horror is generated by a ‘drug’ addiction or an alcoholic one.

    Link to this
  2. 2. sonoran 11:31 am 02/19/2013

    “Most commonly, urine samples are used. Urine tests are very sensitive, but not very specific—this means that there are likely to be many falsely positive results.”
    “For screening tests, false positives are estimated to occur in 5 to 10% of tests, and false negatives in 10-15%.”

    These two statements appear to be somewhat contradictory.

    Link to this
  3. 3. Judy Stone in reply to Judy Stone 1:27 pm 02/19/2013

    Data are conflicting. Also depends on which drug, which testing technique, and where the thresholds for reporting as positive are being set. If the cut-off is set high, then you get more false-negatives.

    Link to this
  4. 4. gooner 2:51 pm 02/19/2013

    Well said paulus

    Link to this
  5. 5. Marya Zilberberg 3:27 pm 02/19/2013

    Judy, great article! And thanks for mentioning Between the Lines.

    I did my calculation of what a positive test means slightly differently. I started with the urine test assuming 90% sensitivity and 90% specificity, as mentioned in your article. From this I got the following 2×2 table if testing 1,000 people from the general population (8% drug use prevalence):

    Drug use + Drug use - Total

    Urine test + 72 92 164
    Urine test - 8 828 836

    Total 80 920 1,000

    Now, the positive predictive value (or what percentage of all positive tests are true positives) is 72/(72+92) = 44%. So, for the screening urine test, the rate of false positives is worse than a coin toss: Out of 100 people testing positive for drug use, 66 will be falsely accused.

    If we go on to a confirmatory blood test in the population of people with a positive urine result (from above we know that their prevalence of drug use should be around 44%; and you gave 99% specificity and 95% sensitivity), we get the following 2×2:

    Drug use + Drug use - Total

    Urine test + 418 7 425
    Urine test - 22 653 675

    Total 440 660 1,000

    The positive predictive value here is 418/(418+7) = 90%, meaning that out of 100 people testing positive in this confirmatory test, 10 will be falsely accused. Put differently, of those 66 people that tested positive above but who are not drug users, fully 7 will be “confirmed” as drug users by this “confirmatory” test. And the consequences can be dire.

    This is a nice illustration of a couple of things. First, shotgunning screening with a crappy test in a population with a low rate of exposure is fraught with error. Second, even in a population enriched for the exposure, even with a test that looks pretty sensitive and specific, there is still a lot of room for error.

    Link to this
  6. 6. Marya Zilberberg 3:28 pm 02/19/2013

    Sorry, it looks like my 2×2 tables got misaligned.

    Link to this
  7. 7. Judy Stone in reply to Judy Stone 3:48 pm 02/19/2013

    I very much appreciate your comments and additional calculations.
    Sorry I can’t figure out how to fix the formatting…

    Your key point is even more shocking, looked at this way: “10 will be falsely accused. Put differently, of those 66 people that tested positive above but who are not drug users, fully 7 will be “confirmed” as drug users by this “confirmatory” test. And the consequences can be dire.”

    And the test doesn’t give any indication about job performance or competence, which is, in theory, the point of it. More on that in upcoming post.

    Link to this
  8. 8. vapur 5:04 pm 02/19/2013

    I find this personally relevant. When going to a homeless shelter to volunteer, they would not accept people who admitted to smoking marijuana because they stated that it had destroyed people’s lives. On the contrary, I would say that it was the government that had destroyed it: taking away potential job opportunities, incarceration, maligned reputation, etc. Employers may claim that they are Equal Opportunity Employers; however, it is evident that some people are more equal than others, regardless of their functional capacity. The biggest complaint that I have about the whole problem is that people are being arrested and pretty much turned into slave labor for the State. I think it is a indictment of our surveillance society when commercials show marijuana leading to amotivational syndrome and stupidity when I know several successful programmers with degrees. Hypocrites don’t deserve all that glory and worship, and will lose faith through each successive generation if not addressed.

    Link to this
  9. 9. paulus 12:12 pm 02/20/2013

    An Associated Press article appeared in today’s papers on overdose deaths in the USA. Of the nearly 40,000 who overdosed last year, 30,000 died from overdoses of Vicodin or Oxycontin. Is this article correct in stating that patients aren’t being screened for these two painkillers, even though the rate of morbidity (death) is so high among those who ingest them?
    Again, however, this is lower than the deaths from alcohol. In 2001, the CDC reported that nearly 35,000 people died from cirrhosis of the liver due to alcohol, and almost 41,000 died from automobile accidents due to alcohol use. I notice these numbers don’t include killings, although alcohol plays a prominent role in assaults that often end in death.
    Neither can we rule out the role of alcohol’s often deadly interaction with other medications. Recent celebrity deaths of Whitney Houston and Heath Ledger show the lethality that combining painkillers, street drugs, and alcohol can play in the grim reaper’s awful tally.
    In all of this, we need to understand the addict is a victim in need of services to change his addiction.
    The best evidence and practices should be used to help him change his self-destructive behavior. Those practices suggest harm reduction strategies work best in addressing the scourge of addiction.

    Link to this
  10. 10. Judy Stone in reply to Judy Stone 12:50 pm 02/20/2013

    Yes, that is correct.
    Screening tests are available for Vicodin and Oxycontin, but the most commonly used panels do not test for these drugs.
    see Drugs-of-Abuse Testing: Understanding the Limitations for a good overview, or workplace drug testing info from the Dept of Labor. Vicodin or oxycontin (oxycodone) can be detected, but are not screened for routinely on workplace drug testing.
    Thank you for your thoughtful comments.

    Link to this
  11. 11. spike91nz 1:48 am 02/21/2013

    It seems to me that the use of drug screens is related to the avoidance of the company (or agency) for economic responsibility for accidents. If an accident occurs in the workplace a positive on a drug screen result absolves the company of injury, disability, and worker’s comp claims. The screen is only incidentally employed for public safety in most cases and is largely utilized to the benefit of the employer. The false positives are beneficial to those who stand to gain and detrimental to the employee. The tests for entry to employment are frequently less stringent then are the tests employed in the evaluation of drug use by the employee following an incident. I am interested in an estimate of how much cost is avoided by the employer (and state) in the use of drug screens for workers comp cases.

    Link to this
  12. 12. American Muse 6:25 pm 02/22/2013

    Non-medically required drug screening should be constitutionally banned. It is an invasion of privacy and amounts to illegal search and seizure. There has to be “probable cause” before such searches can be carried out. Drug screening has become a huge and lucrative business for its purveyors now and is hard to stop — like international wars or the war on drugs.

    Link to this
  13. 13. LRM12 10:04 pm 08/5/2013

    I really appreciate this article, especially this line: “Forensic toxicologist Dr. Vina Spiehler ⁠has been particularly critical of “arbitrary cutoffs for creatinine, saying they discriminate against women, vegetarians, the elderly, people who drink lots of water, and people of small body size.”

    I found out today I failed a pre-employment drug screen for what they called a “negative dilute.” I’m an active, health-conscious person. I drink the recommended 6-8 cups of green tea daily for its anti-oxidative qualities (my mother passed away from early-onset Alzheimer’s a few years ago, so antioxidants are something vitally important to me.) After 12 or 1PM I drink just water throughout the rest of the day–enough to keep me hydrated, but a larger amount during and after working out. I also try to eat a healthy diet full of colorful fruits/veggies, and some protein from fish/foul and carbs. I avoid pig products for the most part, and infrequently eat grass-fed, organic beef. I actually didn’t even drink half the amount of water/tea before my drug screen simply because it was a busy morning spent on the phone. In fact, I felt dehydrated when I went in for the test, and was told by the lab technician to drink water in the waiting room until I felt I needed to empty myself. I can’t understand that a person could potentially be denied a job for living a healthier lifestyle than the average person, especially when there are other tests that could be more accurate and effective of weeding out people with a serious drug dependency.

    “Discriminated against” was exactly how I felt after I received the results.

    Link to this
  14. 14. swimmer 11:18 am 08/12/2013

    Thanks for the article.
    I am a MRO( medical review officer).
    I call DOT and non DOT donors and review their results with them.
    We use only DOT and DHHS certified labs.
    That being said, many companies decide not to use our services for their drug screen results and evaluate the results themselves. This leads to situations like LRM12. Having a “dilute negative” being called a “positive” is a mistake.

    Link to this

Add a Comment
You must sign in or register as a member to submit a comment.

More from Scientific American

Email this Article