A little about insomnia from an insomniac

[bluefoxicy]

Since the sleep graph fiasco going on in another thread, a lot of users have spoken about using the Fitbit to track sleep disorders.  I've been doing that for quite a while, so figured I'd do a short write-up.

Remember that insomnia is a significant medical condition.  Minor, occasional insomnia is normal; long-term, chronic, and severe insomnia is dangerous, leading to brain damage and to dangerously-poor ability to function.  Always inform your doctor about your insomnia and any actions you've taken to remediate it; for extreme cases, find a qualified psychiatrist experienced in treating insomnia.

A little background on me.

Several months ago, I'd gone in for psychiatric care to treat ADHD.  I had a diagnosis and active medication until I was 16, and have been off drugs for about 15 years; it got really, really bad.  I knew I had some sleeping issues, but thought I'd resolved them—mostly.

At the time, my Fitbit (in Normal mode) was showing me as sleeping 5.5-6.5 hours per night.  Eventually I took the time to look at the graphs when I woke up, checking the times I'd awakened against the red spans on the graph.  It showed times I'd spend lying awake in bed as "asleep", and counted those.

Switching the tracker used to Sensitive wasn't 100% accurate, but it came close.  A few spots awake where I was asleep, and a few asleep where I was awake; and it highlighted the spans I'd spent lying awake as solid wake times.

That's where I started trying to treat insomnia:  I was getting 2-3 hours of sleep a night and, near as I could tell without a sleep journal to accurately re-model all my sleep data, it'd been going on for over a year.

Along the way, I used some drugs, did some research, and learned a lot.  I'm schizoid, so I don't have friends; I like information.  Pharmacology is fascinating, and sleep therapy is relevant.

Actigraphy Sleep Tracking

Fitbit provides wrist actigraphy sleep tracking, and some mild instructions about sleep tracking.  The instructions documented are as follows:

• The normal setting counts significant movements as being awake (such as rolling over) and is appropriate for most users.
• The sensitive setting will cause your tracker to record nearly all movements as time spent restless or awake. This setting may be helpful for users who wake up feeling tired even though their sleep history shows sufficient rest.

It's notable the directions given all suggest a predictive strategy:  if you think you'll have a sleep-disorder-laden night, you should set your Fitbit to Sensitive ahead of time.  That's not entirely helpful, and so many Fitbit users with sleep disorders have retroactively reset the setting for sleep segments after the fact.

One might wonder why it's normally okay to count rolling over or shifting around as sleep, yet completely-inaccurate if you wake up feeling tired.  The answer has something to do with wrist actigraphy in the role of sleep tracking, on which a number of papers have been published.

A review of these studies published in 2004 in the American Academy of Sleep Medicine cites some discussion of sleep actigraphy's detection of sleep and waking events.

More recently, Pollak and colleagues have suggested different ways to compute the performance of an actigraph’s ability to detect sleep and wake.21 They use “predicted value for sleep” (PVS) which is the proportion of actigraphic sleep epochs that are also classified as sleep by PSG, and use “predicted value for wakefulness” (PVW) which is the proportion of actigraphic wake epochs that are also classified as wakefulness by PSG

Essentially, sleep actigraphy is metered against Polysomnography in the detection of sleep and wakefulness.  These are opposite states, and their detection is considered separate for good reason.

If a person sleeps well, then the great majority of the measured time is accurately measured as sleep.  Failure to detect wakefulness will only mis-detect a small amount of time, thus the total measurement will be quite-accurate.

If a person sleeps poorly, then failure to detect wakefulness will measure them as sleeping much more than they actually do; while failure to detect sleep will only matter for a smaller portion of the measurement.

To measure actigraphy, the researchers score how frequently it concurs with Polysomnography on both counts.  Adjustments to actigraphy algorithms can change these both:  a more-sensitive algorithm will detect more-subtle motions as wake events, and a less-sensitive algorithm will detect sleep during higher periods of activity.  In essence, the algorithm is adjusted to suppress detection of one or the other.

Importantly, PVS measures if a sleep epoch measured by actigraphy is also a sleep epoch in polysomnography.  If actigraphy detects wakefulness but polysomnography detects sleep, that doesn't lower the PVS; it only counts against actigraphy if the actigraph claims sleep and the polysomnograph claims wakefulness.  The same is true of PVW:  if the actigraph says you're awake and the polysomnograph says your asleep, it counts against PVW.

That tuning has an important impact:

In these comparisons, low threshold actigraph algorithms (e.g., defining wake as occurring even when a small number of activity counts accumulated during the epoch) yielded the best accuracy rate and PVS; however, as sleep efficiency diminished, accuracy rate diminished (evidence level 1A). Actigraph PVW was best with high threshold algorithms (e.g., defining wake as occurring when a large number of activity counts, such as 100, accumulated during the epoch) compared to low threshold algorithms but at a cost of lower accuracy and PVS.

The above essentially says:

• If the actigraph is sensitive and detects you as awake more-often, then it will more-accurately detect sleep;
• If the actigraph is less-sensitive and detects you as asleep more-often, then it will less-accurately detect sleep

This sounds strange because "accuracy" is contextual:  it's either inaccurately detecting sleep (sleep when awake) or it's inaccurately detecting wake (wake when asleep).  If, of 100 sleep epochs, it detects 10 as sleep and 90 as awake, it's got 100% PVS accuracy; the same analysis would detect 0% PVW accuracy.

This bit is rather important:

To summarize, when compared to PSG, actigraphy was found to be valid and reliable for detecting sleep in normal, healthy adult populations but less reliable for detecting sleep as sleep became more disturbed (evidence level 2B).

As I said in the beginning:  I notated the times I had woken up and validated against both graphs.  Neither is perfectly-accurate; likewise, blindly selecting graphs based on a feeling of how well you've slept is not necessarily accurate.  I've had nights where I slept poorly but got a more-accurate measure out of Normal graphs—especially as I recovered and started increasing the restfulness of my sleep.

If you're not keeping a dedicated sleep journal, at least check the time when you wake up and don't fall right back asleep.  If you wake up and feel restless several minutes later, glance at your Fitbit.  It's on your arm.  When you wake up and remember being awake for a short (twenty minutes?) time at 1:25am and an extensive (hour or so) from 3:30am to nearly 5, look at the graphs and decide which best tells you your sleep based on those WASO events.

Sleep Restriction

I'd found also that my sleep, even once treated with drugs, was terrible.  I slept enough, but not well.  I'd been trying to go to bed at 10pm, and I woke up around 8am; that's a good 10 hours in bed, with 7 hours of sleep.  No fun there.

At some point I noticed I'd sleep better if I stayed up later.  That lead me to sleep restriction therapy, a type of cognitive behavioral therapy (CBT).

Sleep restriction is dead-simple:

• Identify your average total sleep time for the week;
• Identify the time you want to wake up;
• Stay up until you've got just enough time to sleep (maybe +15 minutes);
• When your average sleep efficiency exceeds 90% over a week, move your bed time back 15 minutes;
• If your sleep efficiency falls below 80% over a week, stay up 15 minutes later;
• Do this consistently, every day.

I handle sleep latency separately.  The Fitbit has historically calculated total sleep time by ignoring the time in bed spent falling asleep—it cuts off the sleep latency and starts from there.  I use that measurement.  If my sleep latency increases, I keep trying; if it's stubborn or large, I delay my bedtime so I'm not lying in bed awake for an hour.  Sleep latency might take a week to come down after daylight savings time, for example; if it takes longer, I move my bedtime up an hour and start again from there.

Pharmacological Intervention

Drugs can help you sleep; drug therapy also brings risks and trade-offs.  Those trade-offs can be unacceptable.

I've gone through a lot of drugs.  You'll have to ask a psychiatrist for some help here; my experiences won't necessarily be yours.

Melatonin receptor agonists include Melatonin, as well as Ramelteon (e.g. Rozarem).  Melatonin is OTC; Ramelteon requires a prescription.

Dosing melatonin is hard; higher doses disrupt sleep.  At my worst, I needed literally 20mg 8-hour release to sleep; currently I do best with 2mg of 8-hour bi-phasic release.  Melatonin dosages around 0.3mg (300mcg) instant-release can be far more-effective than doses around 1mg; some research shows 2mg of 8-hour continuous-release is effective with no tolerance or rebound insomnia in the elderly.

Orexin Antagonists currently only include Suvorexant (e.g. Belsomra).  This is a new drug.  It worked fantastically for me, cutting my sleep latency to an easy 20 minutes; for most people, it doesn't do anything until you take enough to make you groggy.  Suvorexant has a 9-hour half-life and will stay with you all day; maybe in the future they'll make one with a 1-2.5-hour half life that can knock you out and then leave your body.  It's C-IV, no rebound, no withdrawal; you'll need a prescription, and it's expensive so your insurance company won't want to pay for it.

GABA Drugs are generally alpha-GABA-receptor positive allosteric modulators.  They bind to the receptor and cause it to react more-strongly to GABA.  There are also GABA reuptake inhibitors such as Tiagabine, which increase the amount of GABA in your brain.  Prescription required, for obvious reasons.

GABA PAMs include benzodiapezines such as Diazepam (e.g. Valium) and Lorazepam, as well as non-benzodiapezines like Zolpedim (e.g. Ambien) or Eszopiclone (e.g. Lunesta).  These drugs can work; they can also have horrible side-effects, including memory loss, sleep driving, and hallucinations.  Eszopiclone made me incredibly high 24/7, and I almost drove into another car in a hypnotic state 20 hours after my last dose; I stopped, and went through one day of withdrawal—tachycardia, tremors, sweating, anxiety, depression, and the sensation of sunburn without sensitivity to touch and heat.  Your mileage may vary; don't abuse these things, never mix with alcohol.

Magnesium is also a GABA PAM, apparently, hence why ZMA is marketed as a sleep-promoting drug.  Use ZMA long enough and you'll have trouble sleeping when you stop taking it; it is, fortunately, not Valium.

Atomoxetine is an ADHD drug.  I'm on that now.  It's a norepinephrin and serotonin reuptake inhibitor (SNRI), weaker on serotonin.  It seems to have sharply-diminished my insomnia, so I'm now managing that with melatonin and sleep restriction.  It's been prescribed off-label for insomnia, but it's not necessarily an insomnia medication; I'm uncertain by what mechanism it's helping me.

Whether you're using prescription medications or over-the-counter drugs, talk to your doctor.  Insomnia can be crippling; in extreme cases, you may want to employ a psychiatrist—potentially, a sleep specialist.

f you have insomnia and ADHD, try to get away from the stimulants; amphetamine will seriously destroy your sleep.  Modafinil actually works quite-well for ADHD and frequently doesn't upset people's sleep, despite being a eugerogic; do not use Modafinil to stay up for days straight.  I used Modafinil for my ADHD before I started treating the insomnia, and it drove me to suicidal depression suddenly and with no warning (I don't respond to that kind of thing, so I observed it as an interesting and unpleasant data point and returned the drugs to my psychiatrist); if you get it in your head to stay up for a week straight on your shiny new ADHD drugs, you are screwing up badly.

Cconclusions

Managing insomnia is a combination effort between information tracking (e.g. sleep actigraphy, sleep journals), cognitive behavioral therapies (e.g. sleep restriction, sleep environment), and drug therapy.

You should definitely report your insomnia to your doctor; for significant cases, you should find a psychiatrist to help treat it.  Drug therapies carry risks, and can be used safely and effectively for a large subset of patients; you and your psychiatrist need to determine if the drugs are necessary, effective, and safe for you.

[CarrieM81]

Hi Bluefoxicy,

Thank you for your post, I found it extremely interesting to read.  I've been suffering with Insomnia for 2 months so not long as all but I have done a great deal of research in that time about ways that I can try and overcome the issue as I'm starting to really struggle with it.  I think you said that you have recovered and if so I would really appreciate some guidance/help on the issue?

Thanks 

[Michael23]

It was hard for me to fall asleep for a long time. And when I had a success I saw terrible nightmares which made me scream and sweat.  To say nothing about the rest quality I had. And One thing helped me was a guided hypnosis. Despite all the scepticism I had, it helped a lot. Here's the link for those who suffer as I did https://sleephypnopill.com/have_sweet_dreams/