A meta-analysis finds a non-significant trend in decreased antibody response with poor sleep. What is the significance?
Steven Novella on March 15, 2023.
Optimizing the antibody response to vaccines is essential for their efficacy, both individually and as a public health measure. Great care is taken in the formulation of vaccines to maximize this response. That is the essence of vaccine research – which proteins to target, with how much of a stimulus, and with which adjuvants (vaccine additives that increase immune response). Research also focuses on the number and timing of vaccines, and what antibody titers are necessary in order to maintain adequate protection from infection, reduce severity of infection, and/or reduce the probability of spreading the disease.
There hasn’t been as much focus, generally, on the person receiving the vaccine, other than demographic factors. We cannot alter a person’s demographics (sex, age, race), but are their factors we can change that influence the response to vaccine? A recent study looks at a potentially modifiable factor and Covid vaccine response – sleep.
The study, published in Current Biology, is a meta-analysis of studies looking at the correlation of sleep around the time of vaccination and antibody response. Data on sleep is mostly self-report, with subjects divided into <6 hours of sleep vs sufficient sleep from two days prior to two days after receiving a vaccine for Covid. The idea is sound, as sleep deprivation can have negative physiological effect, including impaired immune function. Insomnia can both impair immune response, making individuals more susceptible to infection, and impair immune regulation, leading to chronic inflammation.
However, overall the data from this study was negative:
“The association between self-reported short sleep (<6 h/night) and reduced vaccine response did not reach our pre-defined statistical significant criteria.”
There was a trend toward decreased immune response, but it did not reach statistical significance. So, like many researchers, they did a subgroup analysis and found that the effect did reach statistical significance in men but not women. In their summary they conclude:
Large-scale well-controlled studies are urgently needed to define (1) the window of time around inoculation when optimizing sleep duration is most beneficial, (2) the causes of the sex disparity in the impact of sleep on the response, and (3) the amount of sleep needed to protect the response.
I would add that such studies are needed to confirm that this effect is even real. Again, a causal association here is plausible but that does not mean it is real or clinically significant. When results like this are statistically marginal, especially if you have to breakout subgroups to reach significance at all, that decreases the chance that the effect is real or that there is a significant effect size. This is where replication is essential. If these results are a statistical fluke then looking at a fresh group of data is likely to regress to the mean or show different results. If it is real, then it should reliably replicate.
This study is a meta-analysis not new epidemiological research so they may not have been setup to do this, but ideally it is best for such studies to do internal replications. In other words, once you find a subset of the data that does reach statistical significance, then check the findings against fresh data as part of the same study, before publishing. This is a good way to minimize the flood of false positives that find their way into the published research. In my opinion, this should be standard practice and high-end journals should make it a near requirement for publication.
This study was also presented as positive, and that is how the media is reporting it, without the proper caveats. In fact, I think this study should be presented as negative – the data did not reach pre-determined statistical significance, and the subgroup analysis gets buried in the results and discussion as indicating a possible direction for future research – but not as the main result.
In this particular case the negative effect from reporting a potential false positive is likely minimal. It’s already a good idea to get a good night’s sleep, and making a special effort to do so around the time of vaccination is a reasonable thing to do even based solely on biological plausibility. There is no real downside to getting good sleep. There could be a downside, however, if the results are overhyped and, for example, some people or their doctors delay vaccination until they improve their sleep.
Future research should focus on several questions. Is the correlation between sleep and vaccine response real, and is there a sex difference? If so, then what parameters optimize the response? Also, if the correlation is real, what is the cause? Is it the lack of sleep itself, or are other factors both impairing sleep and immune response? Is the impairment of immune response clinically significant – does it actually increase the risk of contracting Covid? And, what is the net effect of any clinical intervention, such as treating the insomnia, especially if this treatment delays vaccination. Clinical questions can get complicated once you go beyond a simple correlation.
In the meantime, good sleep hygiene, or even medically treating insomnia, is a good idea if you are fighting off an infection, have been exposed or will be in a high risk environment, or are getting vaccinated. Ideally, chronically poor sleep needs to be treated like the serious medical problem that it is.