When you try to answer questions such as how effective a treatment is for a symptom like insomnia it is important to find a randomized controlled study. A non randomized study would be when people are offered let’s say CBTi or Ambien and you study how well they do. In this scenario you will not know if their response to treatment is because of treatment itself or another factor such as for example how much they believed in the treatment. To be sure you are studying cause and effect of a treatment you also need to treat some people with nothing, aka placebo, aka a sugar pill. This is to make sure that the effect you see is not simply because you are doing something, whatever that something may be.
This study by Gregg and coworkers is a randomized placebo controlled study. They randimized 63 people with chronic insomnia to one of four groups.
The first group received four 30 minute CBTi sessions in person and one 15 minute treatment session by phone. The second group received Ambien 10 mg for 1 month, then 5 mg for 1 week follow by 5 mg every other day for 1 week. The third group was a combination for 1 and 2 and the fourth group received placebe in exactly the same manner as those who received Ambien.
The treatment period was 8 weeks and how well the subjects slept was checked before treatment, at 1 month (“Midtreatment”), at 2 months (“After treatment”) and then at 1, 3, 6 and 12 months follow up points after treatment.
At the 1 month follow up points the subjects that were in the Ambien only and Placebo only group were told what they had been taking and had the option of trying something else. As not enough people stayed in the Ambien only and Placebo only group the investigators did not re-assess these two groups at the 3, 6 and 12 month follow up point.
So who did the best? In which group did people see the most improvement?
Please take a look at this table below. It’s busy but pretty straight forward. What do you find?
Let’s go over the main findings. By the way n is the number of people in the group. For example (n=15) means that there were 15 people in that group.
Starting from the top looking at Sleep latency we can see that people in the CBT, Ambien (Pharmacotherapy) group and the Combitnation therapy group fell a sleep quite a bit faster at the midtreatment point compared to before treatment. However in the Ambien group the time to fall asleep started rising after people stopped taking Ambien and at the 1 month follow up point it took them about 1 hour to fall asleep. In the CBT group things were very stable and even a year after treatment ended time to fall asleep remained much lower than before treatment. 34 minutes is a pretty long sleep latency so there was certainly room for improvement but benefits were sustained.
Sleep efficiency showed a similar pattern. Improvement in all but the placebo group that were sutained in the CBT and Combination group. At the after treatment mark sleep efficiency was 83% in CBT group versus 67% in the Ambien group.
All groups seemed to have improvement in total sleep time but as opposed to sleep differences in sleep latency (time it takes to fall asleep) and sleep efficiency these differences were not statistically significant.
The numbers favor CBT when you look at the after treatment mark and beyond. In other words what you see is that with CBT you learn techniques that help you sleep better for months (even years as other studies have shown) but that with Ambien when treatment is discontinued you no longer benefit.
Maybe the most interesting finding however was what people in the Ambien and Placebo only groups reported and what they did when they had the option of switching ot another study arm. 6/14 (43%) in the placebo only group felt that treatment was not successful and wanted to try another treatment option, 5 chose CBT and 1 chose Ambien. 5/13 (38%) in the Ambien only group reported that treatment was not successful and then tried CBT. Almost the same proportion in the Ambien and Placebo groups felt that treatment was not successful.
Another interesting finding in this and other studies is that the combination of CBT + a drug was not better than CBT by itself.
There is a lot to learn from studies like this one. I would say the key things I have learned from studies on CBT versus drugs for insomnia are:
1. Neither therapy produces dramatic differences in total sleep time. To me this signals that we have a certain sleep need that is genetically determined and neither CBT or drugs changes total sleep time much.
2. You can see pretty dramatic improvements in how fast people fall asleep or their sleep efficiency both with drugs and CBTi. However, with drugs these effects are not sustained. Either because you cannot stay on a drug for ever of, more commonly because you develop tolerance to the drug.
For more than 30 years studies have shown that CBTi is an effective treatment for insomnia but to this day very few insomniacs have access to non-pharmacological therapy.
If you know of someone who does not sleep well please tell them about CBTi and also, please share your thoughts on this article or anything else insomnia related below.