Dissertation - Sleep And Stress In Students

ABSTRACT Students deal with many changes in their first year of university, e.g. moving away from home, getting a job and trying to balance social activities with a heavy course load. All of this impacts upon anxiety levels. Dahlin, Joneborg and Runeson (2005) found first year students were more anxious than third or sixth year students. Anxiety disrupts sleep and can be serious if it becomes insomnia. Harvey (2000) found insomniacs have more pre-sleep cognitive activity than good sleepers. This can be general thoughts to worries and phobias, all of which slow sleep onset rate. Based upon previous research, this study aimed to see if there was a direct relationship between anxiety and poor sleep in first year undergraduates using the Generalised Anxiety Disorder Inventory (GADI) and sleep diary over one week. It was hypothesized that the more anxious a participant was (high GADI scorer), the poorer their sleep would be. Poor sleep was explored using three main variables – total sleep duration, time taken for sleep onset and number of wakings. There was no significance found for GADI scores and sleep total or sleep onset. There was significance found for GADI and number of wakings. Anxiety was directly related to number of wakings but not sleep total or sleep onset which does not support previous work that showed first year’s are highly stressed (Dahlin, Joneborg and Runeson, 2005) which would subsequently affect their sleep. Further research has been suggested in order to look more into the pre-sleep cognitive activity of students (Harvey, 2000), to look into the gender differences reported by Dahlin, Joneborg and Runeson, (2005) and to look at whether first year undergraduates have comparable stress levels and sleep disruption to first year postgraduates. INTRODUCTION The average person sleeps for 6 ½ to 8 hours per night and Coren (1996) has estimated that we sleep for 1 ½ hours less than we did 100 years ago as many of us are in a constant mode of sleep deprivation. Sleep is divided into four stages of slow wave sleep (or non-REM) sleep and REM (rapid eye movement) sleep which appears in a cyclic fashion and Dement and Kleitman (1957) found that different cycles and stages have different brain wave frequencies.When we are awake our brain waves are fast and desynchronized. As our body relaxes ready to sleep our body temperature and heart rate decreases, along with muscle tension. This is when alpha waves begin to appear. Sleep stage 1 is called the hypnagogic state, where dreamlike hallucinatory images resembling vivid photos occur. This is also when a person may get the feeling of falling and the body can jerk awake suddenly. The alpha waves get slower and smaller. During stage two, the brainwaves are synchronized, coming in longer and slower. Bursts of high frequency, called sleep spindles or k-complexes, occur which decrease the brains response to external stimuli to keep the person asleep. They occur once every minute and are triggered by noise. Stage 3 is when slow wave sleep (SWS) occurs. There are large and slow delta waves and sleep spindles are less common. Heart rate, breathing rate and metabolic rate continue to fall. Stage 4 SWS is where only delta waves occur and sleep spindles are eliminated. This is when metabolic rate is at its lowest. After all stages of SWS, REM sleep occurs. The brain waves are similar to when we are awake (fast and desynchronized) and this is when dreaming occurs. In this stage, a person is not easily awoken and heart rate and breathing rate increase. Skeletal muscles are completely relaxed. There are 4 to 6 cycles per night with the REM sleep period getting longer throughout the night. These sleep cycles can be affected by numerous factors including jet lag, shift work and disorders such as insomnia, sleep apnoea and depression. Jet lag is more severe when travelling west to East due to phase

1

advance where a person loses hours. When travelling East to West, phase delay occurs as a person is ahead of local time. Sleeping patterns adjust after a few days; however temperature and hormone cycles take longer. When a person is suffering jetlag, their physical and mental performance can be affected. Czeisler et al (1982) showed that shift work not only disrupts sleeping but also eating and social life zeitgeibers. The problem is not the shift work; it is the changing shifts that constantly reset the biological clock resulting in fatigue, serious sleep disorders, and increased risk of heart attack, ulcers and a higher accident rate. When they introduced a phase delay system in a Utah chemical plant, it was found that output increased. Anxiety and stress are common sleep disrupters. Sleep can be disrupted on a variety of levels by anxiety from mild to extreme depending on the fixation and extremity of the stressor, for instance a trip away from home, a new job etc, but it can be disrupted more severely by anxiety disorders such as phobias, obsessive compulsive disorder and panic disorder (http://sleepdisorders.about.com/cs/sleepdestroyers/a/anxiety.htm, 2006, as accessed on 30th April 2008) Anxiety in students can be induced by numerous factors, Andrews and Wilding (2004) found, such as exam or coursework pressure, homesickness, work and future career prospects, travel concerns and fear of not fitting in or being able to cope with the workload, etc. Andrews and Wilding (2004) found that the most anxiety-inducing factors within students were financial worries and other outside pressures. In their study they looked into whether student anxiety increased after beginning their university courses. The effect of adverse life events and how these factors affected their exam performance in 351 undergraduates using the Hospital Anxiety and Depression Scale and a list of threatening experiences. They found that 20% of symptom free students had become significantly anxious halfway through their course. Of these 36% had recovered and relationship difficulties independently predicted anxiety. All affected predicted a decrease in exam performance. This was the first study to highlight the fact that British students’ anxiety levels can affect academic performance. Dahlin, Joneborg and Runeson (2005) looked into stress and depression in medical students using the Higher Education Stress Inventory and Meehan’s suicidal ideation questions in first, third and sixth year students. They found that first year students gave higher ratings to workload and feedback stressors, showing that they felt more under pressure and were dealing with it less effectively than the third and sixth year students who gave lower ratings. Third year students rated worries are about the future highest and sixth year students rated highest the third year worries as well as non-supportive climate. All rated the lack of feedback. They concluded that first year students had the highest degree of pressure and subsequent anxiety depression than any of the other two years looked at. There was also to be found a gender difference with women reporting higher levels of stress than men. Insomnia is one of the most prevalent psychological disorders which causes severe distress in the patient and can affect many aspects of their lives including social life, personal relationships, physical health and work life. It affects 33% of the population in the United States (National Sleep Foundation, 1991) and Ancoli-Israel and Roth (1999) found that it is especially prevalent in those suffering stressful life events. NHS Direct (http://www.nhsdirect.nhs.uk/articles/article.aspx?articleId=216§ionId=1, accessed April 30th 2008) defines insomnia as the disturbance of a normal sleep pattern and can last for days, months and even years. Symptoms include difficulty getting to sleep, waking during the night, not feeling refreshed by sleep and inability to concentrate during the day and have physical, psychological, physiological and pharmacological causes. Harvey (2002) modelled the maintenance of insomnia. Insomniacs tended to be more anxious about

2

getting sleep and the consequences for the next day if none or very little is experienced. This negative cognition triggers autonomic arousal and emotional distress. Harvey (2002) proposed that this state of intense anxiety triggers selective attention towards the internal and external sleep treat cues and continues to monitor them. This anxious state and attention paid to internal and external cues makes the individual overestimate the sleep deficit and its subsequent effects on daytime performance. As time goes by, these anxiety levels rise further and disrupt sleep in a more drastic way. Harvey (2000) looked at pre-sleep cognitive activity in insomniacs and good sleepers. Pre-sleep Cognitive activity such as being focused on worries, problems and noises within the sleeping environment or being less focussed on nothing in particular keep participants awake and therefore sleep onset if disrupted. These factors are more pronounced in insomniacs and is a key attribute in their disrupted or lack of sleep. Insomniacs tend to focus on not being able to sleep or getting very little or about the day’s events. For insomniacs, this pre-sleep cognitive activity tends to be less intentional, more occupying and lasts much longer than in good sleepers and therefore causes more problems for sleep onset in insomniacs than in good sleepers. Pre-sleep imagery tended to be stronger, more distressing and associated with strong physical sensations in insomniacs than in good sleepers. The previous research has shown that a variety of things can affect sleep, from shift work (Czeisler et al, 1982) to depression. Students have been found to be some of the most under pressure and anxious resulting from a variety of factors such as finances, (Andrews and Wilding, 2004). Dahlin, Joneborg and Runeson (2005) found that first year students dealt more poorly and felt more anxious and under pressure than third or sixth year students. Anxiety is a known sleep disrupter and in extreme cases can lead to insomnia (Harvey, 2002). Drawing on from previous research, it can be seen that a study looking into a possible direct link between sleep disruption and the level of anxiety experienced by students has not been done specifically. Since students appear to be subject to greater stress, this may lead to greater anxiety, which may have a negative impact on sleep. This study attempted look at how anxiety affects the sleep of first year undergraduates over a week using questionnaires. First year students tend to have higher anxiety levels due to the changes and pressures they face. Sleep is easily affected by anxiety, it was hypothesized that students with high anxiety levels, as measured by the GADI, would sleep less in total, wake more during the night and take longer to get to sleep, as there maybe a direct relationship between anxiety and poor sleep. METHOD Design This study used a between groups design. It utilised the Generalised Anxiety Disorder Inventory (GADI – see Appendix 3) (2000, Psychopharmacology Unit, University of Bristol) and a sleep diary – see Appendix 4 (Southampton & South West Hants LREC Ethics submission no. 234/03/w). Both were questionnaires. There were 3 main independent variables explored using the sleep diary for this study – total sleep time in minutes, total time taken for sleep to onset in minutes and number of wakings per night. The sleep diary also recorded time when sleep was first attempted, how long the wakings during the night lasted in total, what time the participant awoke and what time they got up. At the bottom was a subjective rating scale that had 5 points scored from 0 (very good etc) to 4 (very poor) for sleep quality, sleep ease, refreshment after sleep and whether it was enough sleep. The dependent variable was the anxiety level as measured by the GADI. The

3

GADI contains 22 questions about feelings, e.g. I find it difficult to relax. All questions must be answered with a tick in the option box that the participant feels most applies to themselves. The options are not at all, a little, somewhat, very much and extremely and are scored in that order from 0 to 4. A high total score indicates a high anxiety level etc. Participants 21 first year undergraduate students were used. 6 were male and 15 were female. Students were recruited from within the university via solicitations before or after lectures, word-of-mouth between students, and in the latter stages flyers were used to increase recruitment in last few weeks, (see appendix 5). They were left in random places around campus e.g. cafeteria, library, law atrium and student pin-board. Materials Participants were each given a handout containing 2 copies of the GADI, 7 copies of the sleep diary and a consent form. A pen or pencil was used to complete the questionnaires. SPSS was used to analyse the data. Procedure Participants were initially briefed about the nature of the experiment (see appendix) and asked to sign a consent form. Each was given the questionnaire handout (see appendix) and was told to complete the initial GADI prior to completing the first sleep diary. They were to complete one sleep diary as soon after waking from the previous nights sleep as possible for 7 days. After completing all 7 sleep diaries, Participants were told to complete the final GADI and return for debriefing. Participants were debriefed by informing them again that the study was looking into the link between sleep and anxiety in first year undergraduates. They were given an email address if they did decide to withdraw from the study after having completed the forms. Participants were told how the research was going and what was beginning to emerge at the time of their completion, if anything, and how it fit with the previous research. They were also advised that if they had been affected by anything within the study (e.g. realising how stressed they were or how little they were sleeping) or generally that there were counsellors within the university who could help them deal with their problems. Participants were thanked again for taking part. RESULTS The initial and final GADI scores were combined and the average taken for each participant. Averages were calculated for sleep onset, total sleep time and number of wakings for each participant. The median was calculated for sleep onset time in minutes (15 minutes), total sleep time in minutes (480 minutes) and number of wakings (0 wakings). Each participant’s average score for each variable was compared to the median and given a score of 1 for good and 2 for bad, see Appendix 7. The data was analysed using parametric (t-test and ANOVA) and nonparametric (Mann Whitney U) tests due to the normality of the data. The data was close enough to the normal to use a parametric test but a nonparametric test was used in order to clarify the reliability. For all SPSS output, see appendix 6. N GADI v Total GADI v Onset GADI v No. Of Wakings

Good Poor Good Poor Good Poor

Mean Rank 8.62 12.46 7.83 12.27 6.80 14.82

8 13 6 15 10 11

4

Sum of Ranks 69.00 162.00 47.00 184.00 68.00 163.00

Table 1. Mann Whitney U Test Mean Ranks and Sum of Ranks. Mann Whitney U test was run due to the normality of the data and Table 1. Shows the mean ranks and sums of ranks used for this nonparametric test. There was no significant difference in GADI scores between the two sleep total groups, Mann Whitney U (n1=8, n2=13) = 33, p>0.05, (two-tailed). There was no significant difference in GADI scores between the two sleep onset groups, Mann Whitney U (n1=6, n2=15) = 26, p>0.05, (two-tailed). There was a significant difference in GADI scores between the two number of wakings groups, Mann Whitney U (n1=10, n2=11) = 13, p<0.01, (two-tailed). N GADI v Total GADI v Onset GADI v No. Of Wakings

Good Poor Good Poor Good Poor

Mean 12.44 24.23 11.75 22.93 9.85 28.73

8 13 6 15 10 11

Std. Deviation 6.13 21.99 6.56 20.71 6.75 21.17

Table 2. Mean and Standard Deviations for T-Test Table 2. shows the means and standard deviations that were used to run the t-test. There was no significant difference in GADI scores between the two sleep total groups, t (19) = -1.47, p>0.05, (two-tailed). There was no significant difference in GADI scores between the two sleep onset groups, t (19) = -1.28, p>0.05, (twotailed). There was a significant difference in GADI scores between the two number of wakings groups, t (19) = -2.69, p<0.05 (two-tailed). Two-way ANOVA was used to look for significance between the three independent variables and the dependant variable, see Table 3. Source Sleep Total Waking No. ST * WN Error

Sums of Sqs 95.68 991.14 58.44 4759.21

1 1 1 17

df

Mean Square 95.68 991.14 58.44 279.95

F

Sleep Total Sleep Onset ST * SO Error

170.11 87.84 1.55 5975.42

1 1 1 17

170.11 87.84 1.55 351.49

0.48 0.25 0.00

Sleep Onset Waking No. SO * WN Error

201.06 978.24 47.11 4670.52

1 1 1 17

201.06 978.24 47.11 274.74

0.73 3.56 0.17

0.34 3.54 0.21

Table 3. ANOVA used to detect significance between sleep total, sleep onset and waking number against GADI scores. There was no significant variation in GADI scores across the two sleep total groups, (F1, 17 = 0.34, p>0.05, MSE = 279.95). The main effect of the number of wakings was not significant, (F1, 17 = 3.540, p>0.05, MSE = 279.95). There was no significant interaction between sleep total and number of wakings (F1, 17 = 0.21, p>0.05, MSE = 279.95). There was no significant variation in GADI scores across the two sleep total groups, (F1, 17 = 0.48, p>0.05, MSE = 351.49). The main effect of sleep onset was not significant, (F1, 17 = 0.25, p>0.05, MSE = 351.49). There was no significant interaction between sleep total and sleep onset (F1, 17 = 0.00, p>0.05, MSE = 351.49).

5

There was no significant variation in GADI scores across the two sleep onset groups, (F1, 17 = 0.73, p>0.05, MSE = 274.74). The main effect of number of wakings was not significant, (F1, 17 = 3.56, p>0.05, MSE = 274.74). There was no significant interaction between sleep onset and number of wakings (F1, 17 = 0.17, p>0.05, MSE = 274.74). DISCUSSION There was found to be no significance between GADI scores, sleep onset and sleep total. This suggests that there is little or no interaction between these variables which means that sleep onset and sleep total are not affected by anxiety in this study. This contradicts the general finding that anxiety affects sleep and can lead to insomnia (http://sleepdisorders.about.com/cs/sleepdestroyers/a/anxiety.htm, 2006, as accessed on April 30th 2008). The scores for the GADI were low across the board and suggests that the students were not that anxious. This does not support Dahlin, Joneborg and Runeson (2005)’s results which found that first year students tended to be very anxious. This could be the reason why there was no significance found linking sleep total with the GADI scores. As anxiety is a common sleep disrupter, a participant needs to be anxious in order to have their sleep disrupted. As the GADI scores are low, meaning participants were not very anxious, it is not surprising then that sleep is not affected. If the GADI scores were higher then it would have been presumed that total sleep time would be lower and that the data would have shown a significant effect for a direct relationship between poor sleep and anxiety. For instance looking at the data in Appendix 7, Participant 12 had an average GADI score of 66 with average sleep duration of 455 minutes sleep per night and participant 16 had a GADI average score of 74 with an average 227 minutes sleep per night. Both of these participants have a high GAD score and have a poor sleep total rating. Participant 16 does appear to conform to the hypothesis that the higher the GADI score, the poorer sleep is, however participant 12 has a high anxiety level yet sleeps longer per night that participant 6 who has an anxiety score of 33. This shows that the data as a whole as analysed by t-test, ANOVA and Mann Whitney and more specifically when looking at individual participant data, does not show a significant relationship between anxiety and sleep disruption for this study. The relationship between the GADI score and the number of wakings is significant for the t-test and Mann Whitney. This means that the anxiety score affects the number of times a person wakes during the night. It can be said that the higher the level of anxiety as measured by the GADI, the more times a person will wake throughout the night. This means that an anxious student is more likely to get a more disrupted night’s sleep.The ANOVA does not show significance for number of wakings, however it is close to significance. It is more of a trend in ANOVA than a significant relationship as found in t-tests and Mann Whitney. This lack of significance in ANOVA could be down to the normality of the data which is only very slightly skewed but would still have an effect on the outcome. It would be a good idea for further research to look at whether the length of the wakings was also significant when compared to the GADI scores. Due to the lack of significance of a direct relationship between anxiety and sleep total or sleep onset time, but does show significance for the number of wakings per night, it is important to consider the limitations of this study and the factors that may have influenced the outcome. It is also important to reflect upon what could be altered if this study was to be replicated in the future or to be a source of information for any similar studies.

6

This study used the GADI at the beginning of the sleep diary period and upon completion of the sleep diary period. The averages were taken and used for the analysis. However if the anxiety level had changed over this period it was not taken into account and the reasons behind the change were not explored. As some of the anxiety levels did decrease or increase and the average score was used, it may have meant that the scores were not an accurate depiction of the anxiety level of the participant and may have confounded the results. It was not necessary to take two scores unless they were used separately. If this study was to be replicated in the future it would only be necessary to take one GADI score. Another irrelevant source of data in this study were the subjective ratings of sleep quality, sleep ease, feeling of refreshment and whether it was enough sleep. These ratings were not explored explicitly within the study as they were not specifically relevant to the hypotheses that were being tested and due to the constraints of time and resources. Rather than changing the scales of the rating system as was done prior to data collection, the ratings should have been eliminated. It was not clear at the beginning of the study that they would not be used which is why they were kept on the sleep diary. However, it could be said that sleep ease was looked at loosely in the form of the variable sleep onset, just not in the subjective manner of the rating syste,. As these results were collected they should have been explored, even if it was only quickly to see if there was any relationship between them and anxiety measured by the GADI score. It could have possibly shown significance between anxiety and one or more of the variable. However, this was not the case and should be the focus if the study is replicated in the future in case they did provide any significant results. This is a severe criticism of this study as it is ignoring potentially significant data. Although this study went on from the work of Dahlin, Joneborg and Runeson (2005) where it was found that first year students had a higher level of stress than third or sixth year students, it could have been a bit limited. If the study was broadened to include first year postgraduates, who may be under greater, equal to or less pressure than first year undergraduates, the results may have indicated more significance in the results. First year postgraduates have survived undergraduate study however the pressure that is heaped upon postgraduate students is probably very similar to first year undergraduates. By including first year postgraduates it would have also made the sample population larger which may have increased the ease of participant recruitment and made for a larger sample. This would be a good comparison to undertake at a later date. It could be possible that even though the postgraduate students are under more pressure to perform and succeed, they may be able to deal with it better as they have been through undergraduate schooling and learnt to deal with it and subsequently sleep better. If the study had been carried out in a similar manner to Dahlin, Joneborg and Runeson (2005), where different university years were compared against each other, the study may have produced more conclusive results regarding the hypotheses being tested. It may have been that although first year students are under pressure, it may not have been to the extent to which third year students who are doing dissertations are under, but more so than second year students. For future work it would be a good idea to replicate their study parameters between the different years of university students whilst studying and applying the results to the effect this anxiety has on sleep. Harvey (2000) looked into the pre-sleep cognitive activity of insomniacs and the types of worries that took place at this time. The more activity in this pre-sleep stage, the more sleep was affected in a negative manner. As sleep onset was a key variable in this study it would have been a good idea to look at the sorts of things

7

that run through the minds of the participants in this sleep onset stage. This could have been important as Harvey (2000) had already recognised that this pre-sleep activity posed problems especially in insomniacs more than it did for good sleepers. It would have indicated who had merely problems with switching off at the end of the day and those who could possibly have insomnia. The participants that scored poorly on the sleep onset variable may have had higher levels of pre-sleep cognitive activity than those who had good sleep onset scores. This should be something to be explored in a future study into anxiety and sleep disruption. Data collection was quite challenging as there were similar studies running at the same time which affected the number of participants that were willing to take part in the study and was also the reason why flyers had to be resorted to in the last few weeks. If this was a stand alone study then it may have been easier to recruit participants and therefore a larger sample could have been used. A larger sample may have affected the outcome of the study in either direction. Ages of the participants were not taken. This was not thought to be an important factor at the beginning of the study; however it may have had some effect on the end result. If this study was to be carried out in the future it would be a good idea to look at age’s effect on anxiety and the subsequent sleep disruption. For instance, it may be possible that the older the undergraduate student is, the more able they are to cope with the stressors that they face much better, which could mean that as they are more able to deal with the stress, and their sleep will be less disrupted. Dahlin, Joneborg and Runeson (2005) found that there was a gender difference between men and women in reporting stress levels. It would be a good idea to re-examine or to replicate this study with an aim of trying to see if this data correlates with their finding as this was not something that was taken into account in this study. The study hypothesised that the more anxious a student was (as shown by a high GADI score), the less and poorer sleep they would get in total per night, the more wakings they would have per night and the longer it would take to get to sleep per night. However the results showed that there was no significant relationship between level of anxiousness and sleep total or sleep onset. There was only a significant relationship between level of anxiousness and the number of wakings per night. These results could have been affected by the low GADI scores of the participants, as well as other confounding factors. If this study was to be replicated there are a variety of things that should be taken into account or modified. A larger sample would be a good place to start and there are other populations that could have been taken into account that may have provided a more significant result.

The initial and final GADI scores were combined and the average taken for each participant. Averages were calculated for sleep onset, total sleep time and number of wakings for each participant. The median was calculated for sleep onset time in minutes (15 minutes), total sleep time in minutes (480 minutes) and number of wakings (0 wakings). Each participant’s average score for each variable was compared to the median and given a score of 1 for good and 2 for bad. The data was analysed using parametric (t-test and ANOVA) and nonparametric (Mann Whitney U) tests due to the normality of the data. The data was close enough to the normal to use a parametric test but a nonparametric test was used in order to clarify the reliability. There was no significant difference in GADI scores between the two sleep total groups, Mann Whitney U (n1=8, n2=13) = 33, p>0.05, (two-tailed). There was no significant difference in GADI scores between the two sleep onset

8

groups, Mann Whitney U (n1=6, n2=15) = 26, p>0.05, (two-tailed). There was a significant difference in GADI scores between the two number of wakings groups, Mann Whitney U (n1=10, n2=11) = 13, p<0.01, (two-tailed). There was no significant difference in GADI scores between the two sleep total groups, t (19) = -1.47, p>0.05, (twotailed). There was no significant difference in GADI scores between the two sleep onset groups, t (19) = -1.28, p>0.05, (two-tailed). There was a significant difference in GADI scores between the two number of wakings groups, t (19) = -2.69, p<0.05 (two-tailed). There was no significant variation in GADI scores across the two sleep total groups, (F1, 17 = 0.34, p>0.05, MSE = 279.95). The main effect of the number of wakings was not significant, (F1, 17 = 3.540, p>0.05, MSE = 279.95). There was no significant interaction between sleep total and number of wakings (F1, 17 = 0.21, p>0.05, MSE = 279.95). There was no significant variation in GADI scores across the two sleep total groups, (F1, 17 = 0.48, p>0.05, MSE = 351.49). The main effect of sleep onset was not significant, (F1, 17 = 0.25, p>0.05, MSE = 351.49). There was no significant interaction between sleep total and sleep onset (F1, 17 = 0.00, p>0.05, MSE = 351.49). There was no significant variation in GADI scores across the two sleep onset groups, (F1, 17 = 0.73, p>0.05, MSE = 274.74). The main effect of number of wakings was not significant, (F1, 17 = 3.56, p>0.05, MSE = 274.74). There was no significant interaction between sleep onset and number of wakings (F1, 17 = 0.17, p>0.05, MSE = 274.74).

9