Limitations in Current Knowledge of the Sleep Reduction Model

As mentioned in section 1.4, there are a number of limitations in the body of literature that has been put forward to support the sleep reduction model of mania. These limitations include, small sample sizes (which are therefore of limited representativeness), heterogeneity in the use and dosages of medication, and a lack of information on other clinical features, such as bipolar subtypes.

Additional limitations of studies utilising therapeutic sleep deprivation include: the fact that the majority of these studies were not designed to detect mania, therefore these studies do not consistently report on the characteristics of BD participants, and sometimes used antidepressants as an additional treatment. Authors of these studies often comment that participants were informed that sleep deprivation was a treatment for depression, and this, in combination with the fact the studies could not be designed so that participants were blind to whether they had received sleep deprivation therapy, increased the likelihood of a placebo effect.

Additional limitations of prospective studies include variation in the frequency and length of the monitoring period, and variation in methods of measuring mood and sleep. It is also more difficult to ascertain causation from these studies, as observed associations between sleep and high mood could be the result of other triggering factors such as alcohol misuse,

medications, stressors and other factors that may lead to circadian rhythm disruption, such as light exposure.

In summary, therefore, there is limited evidence on whether sleep deprivation can, in fact, trigger mania in individuals with BD. Existing studies, while suggesting that there might be an

association, are limited by methodological issues, thus making it difficult to determine causation.

Assuming that sleep deprivation can, in some cases, act as an early-warning sign or trigger of impending mood episodes, there are further limitations in our knowledge on this subject.

These can be summarised as follows:

1. Some evidence finds an association between reduced sleep and depression, Therefore, the nature of the association between sleep disruption and episode polarity is inconsistent (Gruber et al. 2011). There is a need, therefore, to understand what factors may cause individuals with BD to become depressed rather than manic following sleep disruption.

2. Not all individuals experience mood-elevating effects of sleep deprivation protocols (Benedetti and Smeraldi 2009), suggesting that some individuals have a lower threshold for the antidepressant effects of sleep loss compared to others.

3. The duration of sleep deprivation that is required to trigger mania is still unknown, with varying estimates provided in different studies.

These findings may be the result of (i) the type of sleep loss experienced (chronic vs. acute) and (ii) individual differences in the response to sleep disruption.

1.6.1 Sleep Disruption and Depressive Symptoms

Some studies find that reduced sleep is associated with increased depressive rather than manic symptoms. For example, Talbot et al. (2012) examined 49 patients with bipolar disorder over one week and found that decreased sleep predicted an increase in depressive, but not manic, symptoms the following day. Furthermore, Perlman et al. (2006), in a sample of 54 BD-I participants, found that reduced sleep duration predicted an increase in

depressive, but not manic, symptoms over a six-month follow-up period. This has also been noted in sleep deprivation therapy, where there are accounts of some patients becoming more depressed, rather than euthymic or (hypo)manic, following sleep deprivation (Wirz-Justice and Van den Hoofdakker 1999). Alternatively, sleep reduction might cause a manic or depressive polarity depending on characteristics of the sleep loss, such as whether it is chronic or acute (Boland and Alloy 2013).

1.6.2 Non-Manic Responses to Sleep Deprivation Protocols

Not all studies have found associations between sleep and subsequent mood. For example, (Klein et al. 1992) compared the sleep (from actigraphy) of patients who did and did not relapse following lithium discontinuation, and did not find significant differences between the groups (although this study may have been underpowered to detect differences, due to the small sample size of 10 participants). Studies of patients undergoing TSD for depression have also reported that some individuals do not experience antidepressant effects of sleep

deprivation (Pflug 1976; Bhanji et al. 1978; Wirz-Justice and Van den Hoofdakker 1999) and it has been estimated that the treatment is effective for 50-60% of patients (Wu and Bunney 1990; Kasper and Wehr 1992). This further suggests that individuals vary in how they respond to sleep loss.

The fact that not all individuals respond to TSD in the same way might explain the varying estimates of the proportion of individuals with BD who will become (hypo)manic following sleep deprivation. However, most earlier studies of TSD were not designed to investigate what proportion of BD individuals would switch into mania, therefore included patients with a variety of diagnoses (e.g. major depressive disorder, BD-I, BD-II) and did not always report whether any patients had become hypomanic (Plante and Winkelman 2008). This, in addition

to issues with small sample sizes and varying methods, might explain the wide range in the proportion of individuals who become (hypo)manic following TSD.

1.6.3 How Much Sleep Loss Triggers Episodes?

The studies that report positive associations between sleep deprivation and mood also differ in their estimates of how much sleep deprivation will cause shifts in mood, with estimates ranging from three (Bauer et al. 2006) to 48 hours (Wehr et al. 1982). As discussed previously, this might be due to methodology and sample characteristics. It is also possible that the type of sleep loss (acute vs. chronic) may affect the polarity of the episode that is triggered, with chronic sleep loss being more likely to trigger depressive episodes, and acute sleep loss (i.e.

total sleep deprivation) increasing the likelihood of manic episodes (Wehr 1992). In addition, the time that the sleep deprivation occurs may also be important; partial sleep deprivation therapy, where patients are restricted to 4 hours of sleep per night (Kasper and Wehr 1992), is more effective when patients are sleep-deprived in the second half of the night (i.e. woken after 1am) than in the first half of the night (Goetze and Tolle 1981; Sack et al. 1985; Sack et al. 1988).

1.6.4 Individual Differences In Response To Sleep Disruption

A possible explanation for why these studies show inconsistent results is that there is individual variation in the response to sleep loss within individuals with bipolar disorder.

Vulnerability to the negative cognitive and emotional effects of sleep loss has been studied more extensively in the general population. Research in this area suggests that the differential vulnerability to sleep loss is trait-like (Van Dongen et al. 2004; Rupp et al. 2012; Taniyama et al. 2015) with tentative evidence suggesting that this is heritable (Kuna et al. 2012),

associated with chronotype (Selvi et al. 2007), and associated with polymorphisms in

circadian rhythm genes (Groeger et al. 2008; Drake et al. 2015). This, in addition to evidence that BD is a heterogeneous condition, with a multi-factorial aetiology and clinical

presentation, suggests that individuals within bipolar populations will show variation in their vulnerability to sleep disruption.

1.6.5 Summary

Therefore, despite being widely acknowledged as an important factor in the presentation, management and pathophysiology of BD, there is inconsistent evidence that changes in sleep reliably predict manic symptoms and/or episodes. Furthermore, the literature review

undertaken in this thesis was non-systematic, therefore did not examine grey literature and was limited to studies published (or translated into) English. This means that the reported literature was predominantly conducted in Western populations and could be subject to publication bias.

The literature described in this section suggests that the role of individual differences in moderating the response to sleep disruption (e.g. polarity of response) requires further investigation, in addition to the characteristics of sleep loss that may influence whether it acts as a pathway to mania (e.g. duration, time in circadian night that sleep disruption occurs).