Sleep Abnormalities in Schizophrenia: State of the Art and Next Steps

Rates of sleep disturbance are elevated in patients with schizophrenia spectrum disorders, as suggested by several studies. A survey of a large sample (N>1,800) of patients with nonaffective psychosis showed that about half had clinically significant levels of insomnia (26). A similar incidence was confirmed using self-reported or clinician-rated measures for chronic schizophrenia (4) as well as for first-episode psychosis (27, 28). Sleep disturbance is not only highly prevalent in patients with schizophrenia, but it also closely interacts with their psychotic experiences (29, 30). Patient accounts underscore a bilateral relationship between sleep difficulties and psychotic symptoms, such that patients’ sleep can be disrupted by psychotic experiences (31), while at the same time their sleep problems contribute to the occurrence of their psychotic symptoms (29). In patients with schizophrenia, sleep disturbance is also associated with poor clinical outcomes (32). In a first-episode psychosis sample, insomnia was found to be associated with poorer outcomes in all quality-of-life domains assessed (33). In another study comprising inpatient and outpatient participants, insomnia was associated with suicidal ideation, lifetime suicide attempts, and greater psychopathology in individuals with schizophrenia, thus suggesting that insomnia was a risk factor for suicidal ideation and suicide attempts in these patients (34). Additionally, in a study in which a time-lagged mixed-multilevel model was applied, poor sleep quality predicted elevated paranoia in patients with psychotic disorders (35). Altogether, these findings strongly support the co-occurrence of sleep disturbance and clinical symptomatology in patients with schizophrenia and related psychotic disorders.

Disturbed sleep may also precede the onset and endurance of psychotic symptoms, as suggested by recent evidence. For example, an analysis of data from the general population found that insomnia significantly increased the likelihood of reporting at least one psychotic symptom (36). In another study, the authors reported that both nightmare occurrence and its severity were associated with hallucinations and paranoid thoughts (37). When the amount of sleep is deliberately reduced in experimental manipulation studies in healthy individuals, either via total sleep deprivation (37, 38) or sleep restriction (39), it results in increases in psychotic experiences, including perceptual distortions and paranoia. Additionally, a recent review showed that sleep deprivation leads to dose-dependent aberrations that qualitatively resemble positive, negative, and cognitive symptoms of schizophrenia (40). Experimentally controlled sleep deprivation is also associated with deficits in a range of translational biomarkers for schizophrenia, including prepulse inhibition, smooth pursuit, and antisaccades, in nonclinical populations with a negative history of psychosis (40). In individuals at high risk for psychosis and schizophrenia, sleep disruption has been associated with severity of psychotic experiences, such as hallucinations and delusions (41), disrupted cognition (42), and overall worse functioning (8). A robust association between sleep problems and specific prodromal symptoms (e.g., suspiciousness and perceptual abnormalities) was recently established in a large sample of youths at clinical high risk for psychosis (43). Thus, disturbed sleep not only co-occurs with psychotic symptoms but can also induce or exacerbate these symptoms in patients with schizophrenia, as well as in at-risk individuals.

Sleep architecture abnormalities have been extensively investigated in patients with schizophrenia, with numerous studies published over the past 60 years. Pioneering work by Caldwell et al. (44) showed that individuals with schizophrenia had reduced slow-wave sleep relative to comparison subjects. Follow-up studies have confirmed and extended these findings to both chronic and early-course schizophrenia (45, 46), while also establishing that slow-wave sleep deficits are present in about 40%−50% of individuals with schizophrenia (46). REM abnormalities, including both an increase (47) and a decrease (48) in the amount of time spent in REM and a reduction in REM sleep latency (49), have been reported in individuals with schizophrenia. Of these alterations, only REM latency was found to be significantly reduced in patients with schizophrenia compared with control subjects in a meta-analysis of sleep architecture findings published in the early 1990s (10), although this reduction was not observed in antipsychotic-naive and antipsychotic-withdrawn individuals (12). Compared with control subjects, individuals with schizophrenia showed increased sleep-onset latency, decreased total sleep time, and decreased sleep efficiency, all of which suggest an overall disrupted sleep pattern. Although these alterations have been reported in both antipsychotic-naive and medicated individuals with schizophrenia and appear to be relatively stable across the course of illness (11), they are shared across psychiatric groups, including in individuals diagnosed with anxiety and mood disorders, thus suggesting an overall disrupted sleep pattern rather than alterations involved in the neurobiology of schizophrenia. In addition, abnormalities in sleep architecture parameters provide no direct information on underlying neuronal dysfunctions in individuals with schizophrenia.

To gain better insight into the neurobiology of schizophrenia, a growing number of EEG studies have recently focused on investigating spontaneous brain activity during sleep, including sleep-specific EEG rhythms such as sleep spindles and slow waves, in schizophrenia patients compared with control subjects. Traditionally, spindles have been identified by visual inspection, a process that is subjective, inherently reliant on individual expertise, and particularly time-consuming (50). However, the development of automated spindle detection algorithms (15, 51, 52) has allowed us to compute several parameters reflecting different aspects of spindle activity objectively and efficiently, including amplitude, duration, and density. In addition, the recent availability of high-density EEG systems (≥64 channels) has provided enhanced spatial resolution relative to standard recordings, which can be exploited to identify local differences in brain activity during sleep. Deficits in sleep-spindle parameters, including amplitude, duration, and density, have been found in patients with chronic schizophrenia compared with control subjects in numerous sleep-EEG studies (14, 15, 19, 24, 53, 54). A decrease in spindle density has been the most frequently reported impairment (13). Moreover, spindle deficits in patients with chronic schizophrenia were primarily localized in frontal-parietal and prefrontal regions (14, 15, 53). Although most of these spindle deficits were observed in patients with schizophrenia who were taking antipsychotic medications, no significant correlation between altered spindle parameters and antipsychotic medication doses was established in any of these sleep studies. Additionally, a sleep study that included a nonschizophrenia psychiatric group taking antipsychotic medications found no reduced spindle activity in these individuals (15). In contrast to sleep architecture findings, in two studies conducted by our research group, patients with schizophrenia showed reduced spindle activity even when compared with patients diagnosed with mood disorders (14, 15), thus suggesting that spindles may be more specifically altered in psychotic disorders. Along with spindles, slow waves are the main oscillatory rhythms occurring during NREM sleep (55). As for sleep spindles, several algorithms have recently been developed to detect several slow-wave parameters, including slope, amplitude, and density, in both psychiatric and control populations (56, 57). In patients with chronic schizophrenia, findings of slow-wave parameters have been less consistent when compared with reports on spindles. Reductions in slow-wave activity (1–4 Hz NREM sleep-EEG power) and slow-wave density have been reported in several sleep studies, which included hospitalized, acutely ill patients (45, 46, 58 –60), whereas other EEG studies found no slow-wave impairments in chronically medicated patients with nonacute illness compared with control subjects (15, 61 –63). Although several factors, including medication status and slow-wave selection criteria, could have contributed to the discrepancies in these findings, one intriguing possibility is that slow-wave alterations reflect an increase in symptoms in schizophrenia.

Sleep studies conducted by our research group and others have investigated spindle and slow-wave abnormalities in patients with early-course schizophrenia and in patients with first-episode psychosis. Two initial studies that utilized a limited EEG montage and identified spindles through visual inspection found no difference in spindle activity in 11 (64) and eight (65) patients with early-course schizophrenia, respectively, compared with control subjects. In contrast, by performing sleep high-density EEG recordings and employing an automatic spindle-detection algorithm, our research group established that antipsychotic-naive patients and minimally treated first-episode psychosis patients had a significant reduction in spindle duration and density, which was localized in a frontal region, compared with individuals without psychosis (17). A reduction in sleep spindle density was also reported recently in patients with early-onset schizophrenia compared with both a major depressive disorder group and a nonclinical control group (66). Another sleep study, which used a spindle-detection algorithm in antipsychotic-naive patients with early-course schizophrenia and other psychotic disorders, nonpsychotic first-degree relatives of schizophrenia patients, and two control groups, reported that patients with early-course schizophrenia had significant reductions in spindle amplitude and density compared with the control group (18). Relatives of schizophrenia patients also showed reduced spindle activity when compared with their respective control group. Notably, in that study, the authors reported a reduction in slow-wave activity, but this finding was not further investigated. In a recent sleep high-density EEG study, our research group examined several slow-wave parameters, including density, negative peak amplitude, and upslopes and downslopes in antipsychotic-naive or minimally treated patients with first-episode psychosis compared with a control group. Patients with first-episode psychosis had a significant reduction in slow-wave density in a large frontal-central area, which included the prefrontal cortex, whereas the other slow-wave parameters did not differ between groups (16). Altogether, these findings indicate that spindle and slow-wave abnormalities are present during the early stages of schizophrenia.

Increasing evidence suggests that spindle and slow-wave abnormalities are associated with the clinical symptoms and cognitive impairments commonly reported in patients with schizophrenia. Our research group found that reduced spindle amplitude and density were negatively correlated with the intensity of positive symptoms in patients with chronic schizophrenia (15, 20), as well as with the severity of negative symptoms in patients with first-episode psychosis (17). We also found that reduced slow-wave density predicted higher positive symptom scores among patients with first-episode psychosis (16), whereas other sleep studies have reported that higher slow-wave activity in the frontal and temporal cortex correlated with greater negative symptom severity in patients with chronic schizophrenia (67, 68). Antipsychotic medication exposure is a potential confounder that may affect the relationship between sleep-oscillatory abnormalities and clinical symptoms. However, these abnormalities are unlikely to be related to antipsychotic medications, given the absence of correlation between spindle deficits and antipsychotic medication doses in patients with chronic or early-course schizophrenia who showed significant relationships between spindle alterations and clinical symptoms (15 –17) and the presence of spindle deficits that were correlated with clinical symptoms in antipsychotic-naive schizophrenia patients (17, 18). Nonetheless, longitudinal studies of medication-naive and minimally treated patients with first-episode or early-course schizophrenia are needed to fully establish the impact of antipsychotic medications on the relationships between spindle (and slow-wave) abnormalities and clinical symptoms, including how these relationships may vary in the same cohort of individuals with schizophrenia in the acute and the chronic stages of illness.

Cognitive dysfunction is a core feature of schizophrenia and represents one of the best predictors of poor functional outcome in patients (69). We have reported that decreased prefrontal spindle density was associated with worse working memory performance in patients with schizophrenia compared with control subjects (53). Individuals with psychosis, which included patients with schizophrenia and other psychotic disorders, showed reduced spindle amplitude that was found to be associated with poorer cognitive scores, including lower estimated verbal IQ, compared with control subjects (18). Additionally, several sleep studies have shown that in patients with chronic schizophrenia, reduced spindle density was associated with impaired memory consolidation, involving both procedural and declarative memory (19, 61, 63, 70). Evidence suggests that memory consolidation may rely on the precise temporal coordination of sleep spindles with slow waves. Specifically, it was reported that spindle density and slow-wave and spindle coordination, assessed with phase and amplitude coupling, predicted memory consolidation better than either parameter alone in both schizophrenia patients and control subjects (71). Altogether, these findings indicate that sleep spindle and slow-wave abnormalities are associated with the clinical symptoms and cognitive dysfunctions of schizophrenia.

To further explore the link between sleep-oscillatory deficits and clinical and cognitive symptoms, a novel approach involves computational modeling. For example, simulations with artificial neural (attractor) networks have shown that a reduction in GABA activity produces random jumps from spontaneous activity into an attractor state, which facilitates the occurrence of psychotic symptoms, such as delusions and hallucinations (72), and biophysically realistic models of working memory have shown that alterations in GABA-regulated gamma oscillations are implicated in the working memory impairments commonly reported in individuals with schizophrenia and psychosis (73). Because sleep oscillations are regulated by GABA-ergic neurotransmission and sleep-oscillatory abnormalities are associated with clinical symptoms and cognitive deficits in schizophrenia, in silico experiments could be leveraged to further characterize these relationships.

Sleep-oscillatory impairments point to dysfunctions in underlying neuronal circuits and molecular neurotransmission. Sleep spindles are generated by the interplay of the thalamic reticular nucleus with the other thalamic nuclei, especially in the dorsal thalamus, and are then relayed through thalamocortical projections to the cerebral cortex. Pioneering electrophysiological work has demonstrated that thalamic reticular nucleus neurons can generate sleep spindles in isolation, whereas spindles disappear within the thalamocortical network after disconnection from the thalamic reticular nucleus (74). Subsequent in vivo, in vitro, and in silico studies have demonstrated that the interactions of chemical synapses and electrical coupling among inhibitory neurons of the thalamic reticular nucleus lead to generation and synchronization of spindle sequences within this nucleus (75). In addition, by combining optogenetics and multielectrode recordings, it has been shown that the optogenetic drive of the thalamic reticular nucleus can switch the thalamocortical firing mode from tonic to bursting, which is necessary to generate the spindle oscillation (76). Thalamothalamic and corticothalamic loops are involved in the coordination and maintenance of sleep spindles, as revealed by in vitro recordings (77) and optogenetic activation (78). The thalamic reticular nucleus is strategically located between the thalamus and the cortex, and it is heavily interconnected with the dorsal thalamus, including the mediodorsal nucleus (79). In individuals with schizophrenia, we previously established that reduced mediodorsal thalamic volumes correlated with decreased spindle activity in the prefrontal cortex (PFC) (53). Neuroimaging studies have reported abnormal resting-state functional MRI (MRI), including reduced mediodorsal PFC and increased sensory-motor thalamocortical connectivity in both chronic and early-course schizophrenia (80, 81). Additionally, by performing sleep-EEG recordings and resting-state fMRI, it was shown that patients with schizophrenia had increased motor thalamocortical connectivity, along with reduced spindle density, and the spindle density reduction was inversely correlated with thalamocortical connectivity (24). Combined, these converging findings strongly indicate that spindle deficits reflect abnormalities in underlying thalamocortical circuits in schizophrenia. Slow waves are positive-negative large-amplitude deflections that reflect synchronized periods of neuronal depolarization (up state) and hyperpolarization (down state). Slow waves emerge primarily from the recurrent, synchronized interaction between cortical neurons (82). Thus, a reduction in slow waves results primarily from impaired cortical neuronal activity, which is more prominent in frontal-prefrontal areas, where slow-wave activity usually peaks. A reduction in prefrontal cortex activity has been consistently reported in schizophrenia, including during cognitive tasks (83) and after transcranial magnetic stimulation (TMS) activation (84). Also, a recent study established reduced prefrontal cerebral blood flow in individuals at clinical high risk, in patients with first-episode psychosis, and in patients with chronic schizophrenia compared with control subjects (85).

Regarding the molecular mechanisms implicated in sleep-oscillatory deficits, neurons of the thalamic reticular nucleus express two T-channel subtypes, the CaV3.2 and the CaV3.3 (86). Electrophysiological recordings in mice have shown that CaV3.3 calcium channels regulate spindle occurrence in the thalamus, especially the thalamic reticular nucleus, which is considered the spindle pacemaker (87), and that deletion of CaV3.2 and CaV3.3 in knockout mice leads to suppression of sleep-spindle rhythmogenesis (88). The CaV3.3 is encoded by the CACNA1I gene, and this gene was validated as a candidate schizophrenia risk gene in a landmark genetic study (conducted by the Schizophrenia Working Group of the Psychiatric Genomics Consortium) and established as one of 108 independent genomic loci that exceeded genome-wide significance. De novo genetic variations in CACNAA1I were confirmed in schizophrenia probands (89), and in silico studies have established that these variations have the capacity to disrupt CaV3.3 channel-dependent functions, including rebound bursting in thalamic reticular nucleus neurons occurring during spindle oscillations (90).

The thalamic reticular nucleus is the only thalamic nucleus that contains entirely GABA neurons (79). Electrophysiological experiments have demonstrated that GABA_A receptor-mediated depolarization in reticular neurons activates T-type Ca⁺ channels, which are responsible for the burst spiking associated with spindle oscillation, whereas both GABA_A and GABA_B receptors are involved in synchronizing the spindle oscillatory activity within the thalamus (91). In rodents, thalamic reticular nucleus GABA activity controls sensory auditory gating, which is disrupted by psychotomimetic compounds, including amphetamine, and it is reversed by antipsychotic medications, such as haloperidol (92). These findings from animal studies are consistent with electrophysiological evidence of sensory processing dysfunction in individuals with schizophrenia, as previously reviewed (93). Postmortem studies have demonstrated GABA and glutamatergic abnormalities in schizophrenia, including in the PFC (94 –96) and the thalamus (97, 98). In addition, treatment studies have shown that clozapine, one of the most effective antipsychotic compounds, is associated with enhanced thalamocortical GABA activity in patients with schizophrenia, and the beneficial effects of ECT and TMS are related to increased GABA-mediated inhibitory neurotransmission on excitatory cortical neurons (99 –101). Evidence also shows that knockout mice for the type-5 metabotropic glutamate receptor, which has emerged as a potential therapeutic target for schizophrenia (102), have reduced slow-wave activity during NREM sleep, as well as decreased spindle range power and sleep spindle density, compared with wild-type mice (103, 104). Altogether, these findings indicate that aberrant thalamocortical circuits, dysregulated Ca⁺ channel activity, and altered GABA and glutamate neurotransmission underlie spindle and slow-wave impairments in schizophrenia.

A biomarker is “a defined characteristic that is measured as an indicator of normal biological processes, pathogenic processes, or responses to an exposure or intervention” (105). While clinical symptoms can vary over the course of the illness and are assessed subjectively, biomarkers are objective measures that tend to be stable over time (106, 107). Also, neurophysiological biomarkers, such as sleep-EEG oscillations, offer precise temporal measures of neural circuit dysfunctions underlying the pathophysiology and clinical manifestations of schizophrenia (108). Several subtypes of biomarkers have been defined, according to their putative applications, by the U.S. Food and Drug Administration–National Institutes of Health Biomarker Working Group. Among those subtypes, sleep abnormalities may serve as diagnostic, risk/susceptibility, and predictive biomarkers for schizophrenia. Our research team has reported sleep-spindle deficits in several groups of patients with chronic and first-episode schizophrenia compared with control subjects (14, 15, 17, 53). Other research groups have confirmed and extended these findings to early-course and early-onset schizophrenia (18, 19, 109). Of note, some of these sleep studies reported that spindle deficits were not observed in patients diagnosed with other psychotic disorders or in non-schizophrenia patients receiving antipsychotic medications (15, 18). Combined, these findings suggest that spindle impairments are unlikely to be related to medication exposure and may be potentially specific for schizophrenia. Studies utilizing larger cohorts of control groups and psychiatric patients, including patients with schizophrenia and other major psychiatric disorders, are therefore needed to establish whether spindle abnormalities are putative diagnostic biomarkers for schizophrenia. In this regard, it is important to point out that most studies have focused on sleep abnormalities in individuals with schizophrenia who were diagnosed according to DSM clinical criteria. However, it is critical to recognize that future work should further investigate sleep abnormalities in the entire psychosis spectrum for identifying relevant biomarkers for these psychotic disorders. A sleep-based biomarker of schizophrenia should also be easily measurable, including in routine clinical settings. Toward this goal, work from our research group, as well as other research groups, has shown that sleep-spindle abnormalities can be assessed even with a few EEG electrodes (18, 20), thus not requiring technologically intensive (i.e., high-density EEG systems) diagnostics that are available mostly at academic centers. However, more work needs to be done to make these assessments even more accessible to individuals with schizophrenia, who are usually treated in publicly funded mental health clinics. Additionally, environmental (e.g., history of trauma) and psychological (e.g., comorbid anxiety or depression) factors are known to affect sleep parameters and tend to co-occur with psychosis. Thus, future studies should investigate the contribution of these factors to sleep-oscillatory abnormalities in schizophrenia and related psychotic disorders.

A risk/susceptibility biomarker indicates the potential for developing a disease or medical condition in individuals who do not currently have clinically apparent disease (i.e., schizophrenia). As reported above, our group and others have reported deficits in both spindle and slow-wave abnormalities in patients with early-course schizophrenia. However, we still do not know when these sleep impairments first occur and how they may affect the development and manifestation of psychopathology in patients with schizophrenia spectrum disorders. Clinically high-risk (or prodromal) individuals represent a unique group enriched for precursors of major psychiatric disorders, especially schizophrenia (110). Future work characterizing spindles and slow waves in clinically high-risk individuals could help establish whether these sleep-oscillatory abnormalities may serve as risk/susceptibility biomarkers for schizophrenia. Specifically, if spindle (or slow-wave) impairments are present in clinically high-risk youths and are most prominent in those who eventually convert to first-episode psychosis, these sleep-oscillatory deficits may contribute to the development of a biologically informed risk assessment for schizophrenia. Also, prodromal individuals tend to have different longitudinal courses, with one subgroup showing remission of psychotic symptoms, a second subset with persistence and progression of subsyndromal symptoms, and a third group that will eventually transition to a diagnosable psychotic disorder (111). By assessing the ability of sleep abnormalities to forecast the ultimate clinical outcome of these individuals, future studies may contribute to sleep-based putative predictive biomarkers for schizophrenia.

Endophenotypes are molecular, neuropsychological, imaging, or electrophysiological parameters that reflect the genetic liability for a given disorder. An ideal endophenotype for schizophrenia should be associated with the illness in the population, be inheritable, co-segregate with the illness within families, and be detected in nonaffected family members of psychiatric patients at higher rates than in the general population (106). Our group and others have reported reduced sleep spindles and slow waves in first-degree relatives of patients with schizophrenia compared with age- and gender-matched control subjects with no family history of schizophrenia (18, 112). Building on these findings, future sleep studies utilizing larger groups of patients and their first-degree relatives will be able to assess the viability of altered sleep oscillations as candidate endophenotypes for schizophrenia.

Circadian rhythms regulate the ability of organisms to adapt to environmental challenges and directly affect the health and longevity of many species (113). Circadian disturbances have been reported in individuals with schizophrenia and seem to interfere with their daily functioning. For example, schizophrenia patients with disturbed circadian rhythms tend to perform worse on cognitive tests compared with those with unaltered circadian regulation (114). In a study measuring both actigraphy and melatonin rhythms, about half of individuals with schizophrenia had severe disruptions of the circadian sleep-wake timing, including advanced and delayed or free-running (non-24-hour period) sleep-wake cycles compared with the control group (115). In addition, in a recent actigraphy-based meta-analysis that assessed several circadian variables, including motor activity, relative amplitude, interdaily stability, intradaily variability, and acrophase, it was found that individuals with schizophrenia had reduced motor activity compared with control subjects (116). Molecular studies have reported a loss of rhythmic mRNA expression of the clock genes CRY1 and PER2 in skin fibroblasts in individuals with chronic schizophrenia, as well as decreased expression of CRY1, PER2, and clock genes in individuals with first-episode psychosis compared with control subjects (117), and postmortem work using time-of-death data showed that genes with daily rhythms in individuals with schizophrenia were very distinct from those identified in the dorsolateral PFC in comparison subjects and had a different rhythmic pattern (118). Despite this evidence, which is presented in greater detail in a recent review (119), the co-occurrence and interplay between sleep and circadian abnormalities in individuals with schizophrenia has yet to be fully established. Thus, future studies should assess the relationship between sleep and circadian disturbances in schizophrenia and related psychotic disorders.

In patients with schizophrenia, sleep disturbances are pervasive and have been increasingly implicated in clinical and cognitive impairments. It follows that ameliorating sleep alterations has a clear therapeutic potential for these patients. One of the most effective interventions for sleep disturbances in both psychiatric patients and control populations is cognitive-behavioral therapy for insomnia (CBT-I). CBT-I, which involves a combination of behavioral, cognitive, and educational components, is an efficacious, cost-effective intervention (120), and it is considered the gold standard for the treatment of insomnia (121, 122). Of note, DSM-5 clinical guidelines are that sleep disturbances meeting the threshold of a sleep disorder should be diagnosed and treated as such, even in the presence of another disorder such as schizophrenia. This highlights the importance of treating altered sleep, rather than dismissing it as secondary to the presence of clinical symptoms in schizophrenia and psychosis, and it is supported by evidence that CBT-I was effective in treating insomnia in individuals diagnosed with psychotic disorders, although that study was not adequately powered to detect clinical symptom changes in these patients (123). In addition, in a randomized controlled trial in which university students with insomnia were assigned to receive digital CBT-I or usual care, the sleep intervention at 10 weeks reduced insomnia, paranoia, and hallucinations, and insomnia was a mediator of changes in these psychotic symptoms (124). Another recent study investigating the effects of CBT-I in individuals with schizophrenia and related psychoses reported a significant improvement in the severity of psychotic symptoms and psychological distress following treatment (125). Combined, these findings suggest that CBT-I can ameliorate sleep disturbances and other clinical symptoms in individuals with schizophrenia, although some variability exists in treatment outcomes. One way to reduce such variability involves profiling and selecting patients according to their sleep issues. In support of this approach, a recent study found that individuals with psychosis with classic severe insomnia were more likely to show sleep improvements after CBT-I compared with patients with other sleep subtypes, including insomnia with normal sleep duration and insomnia with hypersomnia (125). Another promising, novel approach involves utilizing treatment response to establish which features contribute to an effective intervention. Identifying these features (i.e., age, gender, severity of sleep and clinical symptoms, and cognitive dysfunctions) will enable clinicians to predict which individuals are most likely to benefit from CBT-I, in line with a personalized, precision-medicine approach (126). At the same time, employing creative adaptations of standard CBT-I protocols can be used to fit a larger proportion of the clinical population, as recently shown in psychiatric wards (127). It is also important to account for other factors that affect outcome, including therapist training, therapeutic rapport, and confidence in intervention. Thus, future studies employing a combination of these approaches, as well as taking into consideration these factors along with the treatment goals of the participants, will help in conducting more effective CBT-I-based treatments in schizophrenia. Future work should also assess the impact of CBT-I interventions on changes in sleep spindles, slow waves, and cognitive functions, such as memory consolidation, in individuals with schizophrenia and related psychotic disorders, which would further reveal the link between abnormal sleep and these disorders. Another challenge to be met is the training of qualified personnel to deliver this intervention in the community, given the shortage of CBT-I therapists outside of university settings. Nonetheless, the clear and current consensus is that CBT-I interventions are acceptable, effective, and cost-effective in ameliorating sleep disturbances in individuals with schizophrenia and other psychotic disorders across at-risk, chronic, and acute stages (128) and that these interventions can and should be applied more widely (129).

Besides CBT-I, other approaches are available to ameliorate sleep disturbances in patients with schizophrenia. For example, antipsychotic medications tend to have beneficial effects on sleep patterns in individuals diagnosed with schizophrenia, although the impact of these compounds on sleep parameters is variable, inconsistent, or, in some cases, still largely unknown (130). Additionally, even in those individuals who are treated with more than one antipsychotic medication, thus raising the issue of polypharmacy, sleep complaints remain significant (131). Among other pharmacological options, hypnotics and benzodiazepines are the compounds most commonly prescribed in clinical settings. These medications facilitate the falling asleep process, as reflected by decreased sleep latency, and can also increase stage N2 NREM, as well as sleep-spindle activity. It would therefore be logical to investigate whether these compounds can mitigate spindle impairments in patients with schizophrenia. A recent double-blind crossover study investigating the effects of eszopiclone, a nonbenzodiazepine hypnotic, on sleep spindle and memory in schizophrenia patients and control subjects found that this medication increased the number and density of spindles in individuals with schizophrenia over baseline levels, although it failed to significantly enhance memory consolidation (132). Thus, future work should further investigate the ability of benzodiazepines and other hypnotic compounds to ameliorate sleep-spindle deficits and related cognitive impairments in patients with schizophrenia.

In addition to pharmacological interventions, noninvasive brain stimulation is emerging as a promising approach to enhance sleep and ameliorate sleep disturbances. The aim of brain stimulation is primarily to increase slow-wave sleep, the deepest NREM sleep stage, which is characterized by slow waves, and it is thought to play a critical role in the restorative and memory aspects of sleep (133, 134). Multiple brain-stimulation methods to enhance slow-wave sleep were recently proposed, including electric, magnetic, and sensory stimulation (135 –137). In addition, studies have reported the development of experimental procedures that allow monitoring and selectively enhancing, in real time, sleep-EEG oscillatory activity, including spindles and slow waves, using auditory stimuli or electric-transcranial input delivered through closed-loop systems (138, 139). Future studies should therefore assess whether these interventions can enhance sleep depth and related sleep-oscillatory deficits in patients with schizophrenia. To successfully develop these noninvasive brain-stimulation paradigms, important challenges will need to be addressed, including establishing whether brain-stimulation enhancement in sleep leads to improvement in sleep quality and functional impairments in individuals diagnosed with schizophrenia and psychosis. Other issues involve acceptability, efficacy, potential side effects, and financial costs and accessibility. Understanding the mechanism of action of brain-stimulation paradigms will also be critical, thus requiring rational designs that consider functional network dynamics and neuroanatomical information. To achieve these goals, more research in animal models sharing anatomical and functional characteristics with humans (i.e., nonhuman primates) will help to characterize in greater detail the neural circuits affected by neuromodulation. Moreover, experimental protocols combining different interventions (for example, electrical and auditory stimulation) will establish whether synergistic effects can be achieved above and beyond administering each stimulation modality separately. Along these lines, studies are needed to compare the sleep-enhancement effects of psychological (i.e., CBT-I), pharmacological, and brain-stimulation interventions, as well as to explore the cumulative effects of combining some of these interventions, and to establish the impact of improved sleep on the clinical and cognitive dysfunctions of psychotic patients. Eventually, this work will contribute to the development of novel sleep-informed treatments for individuals with schizophrenia and related psychotic disorders.