Changes in sleep pattern and dream activity across and after the COVID‐19 lockdown in Italy: A longitudinal observational study

Summary A large body of evidence has documented the impact of the global COVID‐19 outbreak – and especially the lockdown period – on sleep quality and quantity. Here, we present the first Italian longitudinal study on sleep and COVID‐19 considering four different time points collected during lockdown (from 29 March 2020 to 3 May 2020) and a subsequent follow‐up period (October 2020). We used an online survey to collect socio‐demographic and COVID‐19 related information. Subjects were also asked to complete a sleep diary at each time point of the study. Our longitudinal sample included 147 participants. Statistical comparison across time intervals showed remarkable changes in sleep patterns during and after the lockdown. In particular, during lockdown we observed longer sleep latency, less ease of falling asleep, a higher total bedtime, and a lower dream frequency. The week‐by‐week evaluation described relatively stable patterns in the observed measures during the lockdown period, except for dream frequency, affected by a rapid increase in the early phase of lockdown. Our findings are in line with the current literature. Furthermore, the prospective longitudinal investigation comprising several time points offered the possibility of (a) observing the temporal dynamics and the different entities of such changes over time, and (b) reducing the typical memory bias for these studies.

the Italian Government imposed a total lockdown (March 9 to May 4, 2020), forcing its inhabitants to social distancing and home confinement.
Not surprisingly, all these substantial changes led to an increase in psychological problems among the general population, as demonstrated by the high levels of anxiety, depression, and post-traumatic stress symptoms (Torales et al., 2020).
Although the adverse effects of the COVID-19 pandemic on mental health were well documented all over the world, the consequences on the sleep pattern described a complex scenario. In line with timely warnings stated by a task force of the European Academy of Cognitive-Behavioral Treatment of Insomnia in the first months of 2020 (Altena et al., 2020), current findings showed both worsening and improving aspects of sleep following the adoption of the restrictive measures .
In general, most studies described an increase in sleep-related disturbances, defining a global phenomenon called "COVIDsomnia" (Gupta & Pandi-Perumal, 2020). Insomnia complaints during pandemic were reported with 20-45% prevalence among the general population worldwide (Becker et al., 2021). However, the results were not uniform across different populations or specific sleep domains. On the one hand, evidence showed that people went to bed and woke up later during lockdown (Cellini et al., 2020;Gupta, 2020;Wright et al., 2020), along with a longer time spent in bed (Blume et al., 2020;Li et al., 2020;Marelli et al., 2021;Wright et al., 2020). On the other hand, the time effectively spent sleeping -and consequently the sleep efficiency -underwent a parallel decrease (Blume et al., 2020;Casagrande et al., 2020;Franceschini et al., 2020;Gupta, 2020), as possibly documented by the high level of hypnotic uptake (Beck et al., 2020) to compensate the sleep debt.
As expected, people directly facing the disease -such as healthcare workers or COVID-19 patients -had experienced a higher risk of developing sleep problems and psychological distress (Jahrami et al., 2021). Moreover, sociodemographic factors such as female gender were associated with higher mental distress and poorer sleep quality (Salfi, Lauriola, et al., 2020).
Mental activity during sleep also underwent relevant quantitative and qualitative changes during the pandemic. Recent national surveys showed increased dream and nightmare recall frequency during the Italian lockdown . Moreover, waking distressing events caused by the spread of the virus affected the subsequent dream features, in line with the well-known "continuity-hypothesis" (Schredl, 2006). Namely, many studies described higher negative contents and emotional charge in dream reports compared with the pre-pandemic period Iorio et al., 2020;MacKay & DeCicco, 2020;Mota et al., 2020).
Overall, the above-mentioned findings point out that the pandemic -particularly the home confinement period -led to drastic and complex consequences on sleep and dreaming activity. However, it is worth noting that most studies on sleep and COVID-19 adopted a cross-sectional design (Alimoradi et al., 2021). Despite these studies having the benefit of rapidly collecting data from a large sample and on a broad range of variables -especially during the ongoing COVID-19 era -they do not allow the establishment of any causeand-effect relationship. A limited number of studies tried to explore more in-depth the dynamics of the observed effects using longitudinal designs Beck et al., 2021;Saraswathi et al., 2020;Scarpelli, Gorgoni, et al., 2021;Zhang et al., 2020). Notably, most of the current longitudinal studies on sleep aspects during the pandemic referred to a single time point throughout lockdown, compared with a baseline (pre-lockdown) or a follow-up (post-lockdown) period.
Here, we present the first, to our knowledge, Italian longitudinal study on sleep and COVID-19 considering multiple time points collected during and after the lockdown period. In particular, we aimed to describe the temporal dynamics of sleep and dream-related aspects collected by prospective diaries at different time points during the course of lockdown and in correspondence of a follow-up period, when the measures were gradually eased. The main advantage of this study concerns the opportunity to observe not only the difference between conditions during and after (or before) the lockdown period but also the temporal evolution of sleep patterns during the different phases of lockdown, revealing the possible influence of the passage of time (i.e., adaptation effects).
Starting from these assumptions, the aims of the present study were (a) to confirm the significant changes in several sleep domains during and after lockdown, accordingly with previous literature, and  Data reported in this study were part of a wider longitudinal research project concerning the psychological impact of home confinement in Italy. Other data with different purposes have been presented elsewhere Quaglieri et al., 2021).

| Web-based survey
Socio-demographic and COVID-19 related information: an initial questionnaire was administered to assess socio-demographic characteristics, such as gender, age, educational level, occupation, marital status. A subsequent set of questions assessed COVID-19 related information (e.g., COVID-19-infected relatives or friends, forced quarantine period).

Sleep quality measures: Sleep quality was assessed by the
Pittsburgh Sleep Quality Index (PSQI) . The PSQI is a well-known self-report questionnaire investigating sleep quality over one month. It comprises 19 items, from which partial scores in seven subscales (ranging from 0 to 3) and a global score (ranging from 0 to 21) are calculated. The subscales refer to seven different components: subjective sleep quality, sleep latency, sleep duration, sleep efficiency, sleep disturbances, use of sleep medications, daytime dysfunction. A PSQI global score >5 indicates poor sleep quality.  The normal distribution of the original data was checked and skewed data were transformed into normal distribution using square roots transformation (positive skew) and power transformation (negative skew).

| Data analysis
All data were analyzed using Statistical Package for Social

| Longitudinal changes in sleep pattern across and after the lockdown period
One-way MANOVA comparing the sleep patterns across the different lockdown periods and a subsequent FU showed a statistically significant effect of "Time" (Wilks'λ = 0.654, F 28,119 = 2.244, P = 0.001). The univariate repeated measures ANOVAs (Table 2) showed a significant main effect of "Time" for the following vari-

| DISCUSS ION
In agreement with our hypotheses, we found significant changes in sleep-related aspects during and after lockdown. In particular, we observed longer sleep latency and reduced ease of falling asleep during lockdown than post-lockdown. In parallel, time spent in bed underwent a significant decrease during the follow-up period compared with most of the intervals within the lockdown period.  Scarpelli, Gorgoni, et al., 2021).
Several hypotheses have been proposed to explain these changes, as a function of specific lockdown-related aspects. For example, the increased electronic device usage near bedtime during the isolation period (Cellini et al., 2020; could exacerbate the subjective difficulty in falling asleep. Indeed, it is well established that the evening exposure to screen light may have detrimental effects on human sleep and circadian rhythms due to the lengthening of sleep latency and the increasing of intrasleep wakefulness (Green et al., 2017). Also, it is conceivable that the higher levels of stress and anxiety experienced during the social restriction period (for a review, see (Xiong et al., 2020)) may have caused an increase in physiological arousal. Consequently, longer sleep onset timing could reflect the known relationship between mental distress experienced during the day and the quality of subsequent sleep (Kalmbach et al., 2015). Furthermore, our sample consists mostly of women, who seem to be particularly vulnerable to the psychological consequences of the pandemic (Salfi, Lauriola, et al., 2020). In addition to mirroring the data on sleep latency, the perceived difficulty of falling asleep during lockdown can also be associated with the possible reduction of the homeostatic sleep pressure in that period. Indeed, the lack of traditional social, school, and work schedules caused remarkable changes in sleep timing and circadian rhythms (Blume et al., 2020).
Even the increase in TBT during lockdown could be explained within this theoretical framework. It should be hypothesized that people have adapted daily activities to their individual chronotype, and this allowed the re-adjustment of the natural sleep-wake schedule in the absence of social time cues (Leone et al., 2020). In general, studies worldwide described a shift to later bedtime and rise time during the quarantine period (Cellini et al., 2020;Gupta, 2020;Wright et al., 2020). Also, the reduced sunlight exposure may have concurred to the reduction of the externally imposed sleep-wake rhythms (Smit et al., 2021). However, we should consider that the increase in time spent in bed was not paralleled by the same rise of the effective sleeping time.
The pattern of our results is consistent with previous literature (for a review, see Jahrami et al., 2021). Moreover, the divergent findings as a function of the specific sleep domain suggested the low ability of composite measures (e.g., PSQI) to provide an accurate picture of the multiple consequences of this unprecedented situation .
Along with the above-mentioned results concerning sleep, we also found significant changes across time points regarding the oneiric activity. In particular, we described a rapid increase in the number of dreams during the early phases of lockdown, reaching the highest peak in the second time interval and then gradually decreasing until the last week of lockdown. During the follow-up period, we observed a drastic fall in the dream frequency compared with the confinement period. Consistently with previous literature, the increased dream rate during lockdown Iorio et al., 2020;Schredl & Bulkeley, 2020) could be explained by a sort of "coping strategy" to metabolize stressful events and negative affect experienced during daytime life (Scarpelli et al., 2019). These results are also in keeping with previous studies showing significant changes in dream occurrence after experiencing traumatic events (e.g., Tempesta et al., 2013). Concerning the increased dream frequency in the early phase of lockdown, a possible explanation could be related to the "alarm reaction" phase occurring at the beginning of lockdown (higher levels of anxiety), followed by an adaptation to circumstances (Fancourt et al., 2021;Quaglieri et al., 2021). In line with the previously stated role of dreams during traumatic or stressful events, such a rapid increase may reflect the urgent need to adopt coping mechanisms at the very beginning of the overwhelming experience.
Furthermore, in previous research, higher dream frequency was also related to lower sleep quality -specifically to sleep fragmentation and intra-sleep wakefulness (Schredl & Reinhard, 2008;Wyk et al., 2019). In particular, we could speculate that the changes in sleep timing during the lockdown (e.g., the later bedtime and rise time, the longer TBT) may have increased both the probability of REM awakenings and the number of REM-wake transitions and -consequently -the recall of dreams upon morning awakenings (Koulack & Goodenough, 1976;Scarpelli et al., 2020). However, we could not directly establish this relationship given the lack of corresponding measures in the current study.
To the best of our knowledge, this is the first longitudinal study on sleep and dreams, collecting prospective data through repeated measures during an extended period (from March to October 2020) of the ongoing COVID-19 pandemic. The within-subjects design allowed us to observe the subjective trajectories of sleeping and dream activity during the different phases of lockdown and postlockdown. This aspect represents a great advantage compared with cross-sectional studies, which provide a single snapshot of a given moment in time, without any information about what occurs immediately before and after that specific period.
Despite our study focused on the first lockdown period ("first contagion wave" -spring 2020), it is worth mentioning other national studies also considering the second lockdown period ("second contagion wave" -autumn 2020), characterized by lower sleep restrictions (Conte et al., 2021;Salfi et al., 2021). Interestingly, their results showed the high and permanent sensibility of sleep and dreaming to lifestyle modification, as reflected by the changes in both sleep quality (Salfi et al., 2021) and dream activity (Conte et al., 2021) closely related to the course of the pandemic. Overall, this evidence confirms the current results and allows us to read them from a wider time perspective.
Some methodological constraints of the present study should be taken into account. Firstly, the absence of any pre-pandemic infor- Anyway, we should consider that the longitudinal collection of prospective data potentially represents the best solution to describe the actual scenario referred to the selected variables, although it could further imply issues of self-selection. Indeed, the retrospective assessment used in most studies allowed a rapid data collection from a large sample but affected the reliability of results due to the underlying memory bias (under-or overestimate).
Regarding the investigation about dream activity, we did not collect dream reports. Consequently, we only described the quantitative aspect of dreams without any reference to the nature of the contents and their possible relation with waking experiences.
Finally, our sample is not fully representative of the Italian population due to the unbalanced dataset (e.g., gender, age, occupation, marital status), and so caution is needed in interpreting and generalizing the results.

| CON CLUS IONS
Our findings were consistent with previous literature showing lockdown as a key factor responsible for the reported changes in sleep and dream activity during the pandemic. Furthermore, the longitudinal investigation comprising several time points offered the possibility to observe the temporal dynamics and the different entities of these changes over time.
To sum up, our pattern of results pointed out two main conclusions (a) sleep-related aspects underwent critical changes during lockdown: reduced ability to fall asleep (timing and ease), increased amount of time spent in bed, and higher dream recall frequency; (b) sleep pattern observed during lockdown remained quite stable over time, except for dream activity: an initial rapid increase was observed, followed by a partial stabilization.
The understanding of the effects of the pandemic on sleeping and dreaming is still under investigation. Further longitudinal studies should be conducted to shed more light on the evolution of such effects over a period of time and in different phases of the pandemic.

ACK N OWLED G EM ENTS
Open Access Funding provided by Universita degli Studi di Roma La Sapienza within the CRUI-CARE Agreement. [Correction added on 4 June 2022, after first online publication: CRUI funding statement has been added.]

CO N FLI C T O F I NTE R E S T
None of the authors have potential conflicts of interest to be disclosed.

AUTH O R CO NTR I B UTI O N S
VA, MG, FF, AMG and LDG conceived and designed the experiment; SSc, PZ, SSd, EM, AQ collected and pre-processed the data; VA and MG analyzed the data; VA, MG and LDG wrote the paper.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.