A pitfall of crowdsourcing: addictive medicines in an online study of treatment success

Last year 23andme acquired the research group CureTogether. CureTogether crowdsources patients' views on which drugs work well for particular conditions. This essay notes some benefits of these surveys and some risks associated with misuse of these results.

CureTogether has surveyed 11,613 people with depression, and here are their top-rated therapies:

  • Exercise
  • Spend time with pet
  • Play outside
  • Music with exercise
  • Music therapy
  • Daily aerobic exercise

The top 6 therapies are safe and inexpensive. Research shows them to be effective. Patients can still benefit from medical input, thus providing a cure that is patient-driven and personalised. They should still attend medical consultations to monitor for deterioration, need for antidepressants and psychotherapy, and so on, but overall, the survey could improve patient outcomes.

However, for their insomnia review, I cannot say the same. On aggregate, 8988 individuals with insomnia said that the best treatments are:

  • Xanax (Alprazolam)
  • No light/dark room
  • Ativan (Lorazepam)
  • Xyrem (Gamma-hydroxybutyric acid)
  • Imovane (Zopiclone)
  • Sexual activity
  • Stillnox (Zolpidem)
  • Valium (Diazepam)
  • Exercise
  • Rivotril (Clonazepam)

There are some good and safe therapies mixed in there, but benzodiazepines - Xanax, Ativan, Stillnox and Rivotril - round out four of the top ten. If you are not familiar with benzos, they bear some resemblance to alcohol. Benzos:

  • reduce anxiety,
  • induce intoxication, forgetfulness, difficulty coordinating movements, confusion, loss of inhibitions and emotional lability (Hindmarsh, 1997).
  • can induce tolerance - as months go by, an increased dose is required achieve the same effect)
  • can cause withdrawal effects

Benzos are frequently prescribed inappropriately:

Inappropriate chronic benzodiazepine use is more subtle but relatively common: for anxiolytics, 36% of past-year users (3% of the adult population in the US) report using these drugs for 4 consecutive months or longer. The risks of such long-term use are much better documented than the benefits. (Griffiths + Weerts, 1997)

One cannot rely on the subtlety of inappropriate benzo use to be conveyed by the survey. Ashton reports similarly:

Despite repeated recommendations to limit benzodiazepines to short-term use (2–4 weeks), doctors worldwide are still prescribing them for months or years. This over-prescribing has resulted in large populations of long-term users who have become dependent on benzodiazepines and has also led to leakage of benzodiazepines into the illicit drug market.

Benzodiazepine tolerance and withdrawal are particularly problematic where benzodiazepines are available without a prescription. Although most of the readers are likely to be located in the USA where these drugs are prescription-only, it is particularly important that readers of the survey do not import and self-medicate with these drugs. Take the example from the Bluelight drug use forum:

I am currently addicted to alprazolam, I take 2mg a day. Here is my situation I started taking alprazolam everyday since november of 2009 because I was having problems sleeping and it worked great at knocking me out I was taking .5 mg a night to sleep, so stupid i know. I continued taking that dose for about six months until I started noticing withdrawl symptoms about 18 hours after my last dose. I began taking .25 mg in the afternoon and .50 at bedtime. So now i'm up to 2mg a day, 1mg morning and 1mg at night. I am fully aware of the dangers of doing what i'm doing I'm just tired of living in this cycle, my emotions are blunted and I can tell a difference in my cognitive ability and memory. I have only been using 2mg a day for the past 4 months due to problems in my family where I was under a lot of stress with work and the death of a close family member, which is no excuse. (556m4, a member of bluelight forums)

How does the CureTogether survey represent the side-effects of Xanax? 87% reported reduced anxiety, 41% reported decreased inhibitions and 37% reported forgetfulness. Interestingly, 12% report insomnia and 6% report tremor, which can represent dependence.

CureTogether also reports that Xanax works well for anxiety. However, Ashton notes that problem of dependence there also:

Although some authors recommend long-term use of benzodiazepine anxiolytics for certain conditions, it is likely that the drugs are preventing withdrawal symptoms rather than reducing anxiety.

The problem is that we do not know how to interpret the side-effects reported in CureTogether because we do not know if they were concealed by ongoing benzodiazepine with dependence nor how long the benzos were used. One sees that what CureTogether surveys select for is very different from a cure for insomnia. It depends much more on immediacy and salience.

If you asked CureTogether for a drug to increase your happiness, you get the feeling it would offer heroin. Heroin has a large-pleasurable effect, it's true. But check out their recommendations for back pain:

  • Oxycodone
  • Hydromorphone
  • Fentanyl

This is far from the idea you want your patient to have when entering a dialogue about the management of back pain. You want to start with drugs that have fewer side-effects, like Panadol (or Tylenol in America) and anti-inflammatories, whereas CureTogether's top-three are are all opiates (morphine-like drugs), and therefore are addictive. Hydromorphone and fentanyl are extremely potent, and cannot safely be given to those who have not taken opiates before. Opiates are not given straight away because they cause dependence in 1/500 patients. (One can hardly imagine how the world would be if every patient prescribed Panadol (Tylenol) was similarly addicted!) Hydromorphone is particularly addictive, rated roughly as enjoyable as heroin by the bluelight forum.

So the CureTogether surveys ranks drugs more effective if their effects are more salient, and are for some particularly attracted to some addictive drugs. Although some therapies suggested by these surveys are good, the drugs that they recommend have some unfavourable long-term outcomes. I conclude that CureTogether can generate interesting ideas to discuss with a doctor, but is not safe to use for self-medication.