SEARCH

SEARCH BY CITATION

Keywords:

  • obesity;
  • sleep

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Epidemiologic studies
  5. Sleep and appetite
  6. Sleep and media use
  7. Literature limitations
  8. Discussion
  9. Conclusion
  10. Acknowledgements
  11. References

Obesity has become pandemic. In America, as obesity has increased, the amount of sleep Americans get per night has decreased, and studies are now showing an association. Epidemiological studies on short sleep duration (SSD) and obesity have been conducted in children and adults, and show an overall positive association. Leptin and ghrelin, two hormones that control appetite, have been studied as a mechanism for SSD causing obesity. Low leptin and high ghrelin levels have been seen in sleep deprivation, the effect of which is an increase in appetite that could be linked to obesity. Decreasing media use, namely television and computers, could be one way to increase nightly sleep and potentially help people lose weight. Paediatric studies have shown an association with bedroom media use and shorter sleep duration. Adult studies are lacking in this area. Limitations in the literature include self-report in a majority of sleep studies and only a suggested causal link between SSD and obesity among all of the epidemiological studies. In conclusion, obesity is a global problem with great complexity. Encouraging people to get more sleep could be one part of the solution to help them lose weight and gain health.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Epidemiologic studies
  5. Sleep and appetite
  6. Sleep and media use
  7. Literature limitations
  8. Discussion
  9. Conclusion
  10. Acknowledgements
  11. References

Americans are losing sleep, obesity is epidemic, and studies suggest the two may be associated. The National Sleep Foundation (NSF 2007a) recommends that adults get 7–9 hours of sleep per night. According to the ‘Sleep in America’ poll, however, 35% of Americans slept less than 7 hours per night in 1998, and this number increased to 44% in 2007 (NSF 2002, 2008). Alongside this decrease in sleep is an increase in weight among Americans. The prevalence of obesity in the USA rose from 14.5% in the years 1971–1974 to 34.3% in the years 2005–2006 (Flegal et al. 2002; Ogden et al. 2007). The USA is not alone as one in four adults are considered obese in England (McPherson et al. 2007), and globally, obesity has become pandemic, affecting at least 300 million people worldwide (WHO 2002). These two seemingly unconnected problems, short sleep duration (SSD) and obesity, are sequestering a large amount of interest and research.

Sleep deprivation research has roots in the late 19th century, with Marie De Manaceine conducting the first studies showing sleep as essential to life (Bentivoglio & Grassi-Zucconi 1997). Current sleep research is now suggesting an association between SSD and obesity, but scientists are only beginning to understand the involved mechanisms. In this review, epidemiological studies will be discussed along with studies that have evaluated the effect of sleep on appetite and how the use of media (e.g. TV and computers) influences sleep. Finally, potential limitations of the current research will be reviewed, followed by a discussion of further research needs.

Epidemiologic studies

  1. Top of page
  2. Summary
  3. Introduction
  4. Epidemiologic studies
  5. Sleep and appetite
  6. Sleep and media use
  7. Literature limitations
  8. Discussion
  9. Conclusion
  10. Acknowledgements
  11. References

The literature on sleep and obesity consists primarily of cross-sectional and cohort studies. These studies have been conducted in a wide age range – from children to adults. In paediatric studies, overall results show SSD as a positive risk factor for obesity. Cross-sectional studies have found that children less than 10 years old with less than 10 hours of sleep per night have an increased risk for being overweight/obese (Sekine et al. 2002; Von Kries et al. 2002; Chaput et al. 2006). Sekine et al. (2002) also found a clear dose–response relationship between children's hours of nightly sleep and their risk for obesity. They found adjusted odds ratios (OR) for developing obesity were 1.49, 1.89 and 2.87 for 9–10 hours, 8–9 hours and less than 8 hours of sleep in comparison to children with ≥10 hours of sleep per night. A large UK cohort study looked at data from 5493 children for risk factors associated with obesity at age 7 years. They found sleep to be independently associated with obesity. Specifically, those in categories sleeping <10.5 and 10.5–10.9 hours per night were more likely to be obese than those sleeping >12 hours per night (Reilly et al. 2005).

Two meta-analyses found SSD to be a risk factor for obesity in children. Chen et al. (2008) examined studies researching children and adolescents with ages ranging from 3 to 19 years. These studies found a pooled OR of 1.58 for SSD and overweight/obesity compared to children/adolescents with a recommended level of sleep. Sleep duration categories used to pool the ORs were evaluated on a study-by-study basis. This paper found that boys had a higher risk (OR = 2.50) for SSD and overweight/obesity than girls (OR = 1.24) when compared to those with recommended sleep durations. The authors suggest more research is needed to understand this potential gender difference. Cappuccio et al. (2008) also conducted a meta-analysis looking at studies in children/adolescents with ages ranging from 2 to 20 years. Comparing those with SSD and those without, the pooled OR was 1.89 for children/adolescents with SSD and obesity.

Numerous studies have examined SSD and obesity in adults. In cross-sectional analyses, the majority of the literature from around the world has found SSD to be a positive risk factor for obesity (Vioque et al. 2000; Hasler et al. 2004; Taheri et al. 2004; Singh et al. 2005; Kohatsu et al. 2006; Moreno et al. 2006; Bjorvatn et al. 2007; Stranges et al. 2008). In addition to the meta-analysis conducted using studies based on children, Cappuccio et al. (2008) also conducted a meta-analysis using studies based on adults. They compared those with SSD, most often considered <5 or ≤5 hours of sleep, to those without and reported a pooled OR of 1.55 for adults with SSD and obesity.

There are prospective studies looking at SSD and obesity, but the results are mixed. A 6-year prospective study conducted in Canada found that short-duration sleepers (defined as 5–6 hours per day) had a 27% increased risk for becoming obese than average-duration sleepers (defined as 7–8 hours per day) (Chaput et al. 2008). Another study examined 16 years of data from the Harvard-based Nurses' Health Study cohort. They found sleeping less than 5 and 6 hours per 24-hour period increased a woman's risk for becoming obese by 15% and 6%, respectively. Women sleeping more than 7 hours per 24-hour period showed no risk of becoming obese (Patel et al. 2006). A more recent study conducted cross-sectional and prospective data analyses of SSD and obesity. The investigators found a significant association between obesity and nightly sleep duration of less than 5 hours per night. However, prospective data did not confirm this relationship (Stranges et al. 2008).

Age may be a significant confounder when looking at SSD and obesity. One prospective cohort study was conducted over a 13-year time frame in Zurich, Switzerland. Using a cross-sectional analysis of the data, the investigators found that SSD was positively associated with obesity for age time points 27, 29 and 34, but the association disappeared at age 40 (Hasler et al. 2004). A small cross-sectional study looked specifically at SSD in women ages >50 years (Chaput et al. 2007). SSD was considered <7 hours in this study, and the investigators found no increased risk for overweight/obesity. Limitations to this study included the small population of 90 women, as well as a small body mass index (BMI) (kg/m2) range (including mostly overweight women and few healthy weight or obese women for comparison). In The Netherlands, a large cross-sectional study looked at 983 people >55 years and did find an association between SSD and obesity. Participants had an OR of 2.76 for obesity with an SSD of <5 hours per night compared to participants sleeping 7–<8 hours per night (van den Berg et al. 2008). Sleep was measured by an actigraph, a device worn by participants that determined sleep time through lack of physical movement, rather than by self-reporting as other studies have done, which was a strength of this study. Sleep's association with obesity in older populations is still varied and further study is needed.

Two major questions remain. The first is whether there is a causal relationship, as the studies only suggest an association of sleep and obesity and not that SSD causes or contributes to a person being obese. Second, if SSD does lead to obesity, these studies do not make the mechanisms clear. Other studies have been conducted, however, which start to delve into what the reasons may be for this association.

Sleep and appetite

  1. Top of page
  2. Summary
  3. Introduction
  4. Epidemiologic studies
  5. Sleep and appetite
  6. Sleep and media use
  7. Literature limitations
  8. Discussion
  9. Conclusion
  10. Acknowledgements
  11. References

One suggested mechanism in the literature relates to two appetite hormones, ghrelin and leptin. Ghrelin is a hormone that is produced mostly in the stomach and small bowel. One of the observed roles ghrelin plays is to stimulate appetite (Van der Lely et al. 2004). Leptin, on the other hand, is primarily secreted by adipose tissue into the blood and decreases appetite by signaling satiety (Ahima & Osei 2004). These two opposing hormones should control appetite in such a way that optimal caloric intake and energy homeostasis is obtained (Valassi et al. 2008).

Studies looking at leptin and ghrelin levels in patients with induced sleep deprivation are few, as the nature of these studies is invasive. Nevertheless, sleep deprivation's effect on leptin and ghrelin levels has been investigated. One study looked specifically at leptin and ghrelin levels in ten healthy men before, during and after an induction of 88 hours of continuous sleep deprivation. With this sleep deprivation, circulating leptin levels decreased in the participants (Mullington et al. 2003). Saad et al. (1998) studied 31 healthy men and women, and found leptin levels among these study subjects peaked on average at 01:20 hours with the lowest levels at approximately 10:33 hours. Peaking at night would signal satiety as a person sleeps, while the decrease in the morning would trigger appetite and food intake. If sleep debt decreases leptin levels, it is essentially prompting an increase in appetite.

Spiegel et al. (2004) looked at both leptin and ghrelin levels before, during and after induced sleep deprivation. Subjects in this study were 12 young men with healthy weights who normally got 7–9 hours of sleep per night. After baseline acclimation, sleep was restricted to 4 hours in bed for two consecutive nights and was extended to 10 hours per night for two consecutive nights. Half of the participants received the 4 hours per night before the 10 hours, and half got the reverse. On average, leptin levels were 18% lower and ghrelin levels were 28% higher when the men had 4 hours in bed per night compared to 10 hours. Taheri et al. (2004) conducted a Wisconsin cohort study, which confirms these data. They combined one night of sleep in a laboratory with a morning blood draw for leptin and with a self-report sleep diary and questionnaire for ghrelin levels to obtain normal sleep times. They found that people with a habitual sleep of 5 hours per night compared with 8 hours per night had 15.5% lower morning leptin levels. When comparing 5 and 8 hours of nightly sleep measured in the laboratory setting (one night of measurement by polysomnography) there was a 14.9% increase in morning ghrelin levels.

Chemically, a decrease in leptin and an increase in ghrelin would increase appetite. Spiegel et al. (2004) also had their participants rate their hunger and appetite, both of which increased after forced sleep debt. This agrees with the hypothesis that the combination of decreased leptin and increased ghrelin increases appetite overall. What is also interesting about this study were the foods the participants craved. Overall, they saw a 33–45% increase in craving for carbohydrate-rich foods including sweets, salty snacks and starchy foods.

Two recent studies have attempted to look at SSD and specific food intake. A study of 31 healthy Greek women found no association between SSD and energy, total fat or carbohydrate intakes. Diet was analyzed with a 24-hour diet recall, but the investigators only looked at two 24-hour intakes and did not specify whether diet was recorded for a week or weekend day. This could have potentially introduced bias into the results (Rontoyanni et al. 2007). Another study in a rural Midwestern US population found an increase in fast food and high-fat food consumption along with a decrease in fruit and vegetable consumption in people who slept <7 hours per night vs. those who slept 7–8 hours. However, there was no association between SSD and obesity (Stamatakis & Brownson 2008). All of these studies have limitations, indicating the need for further research on SSD's influence on people's appetites and food choices.

Sleep and media use

  1. Top of page
  2. Summary
  3. Introduction
  4. Epidemiologic studies
  5. Sleep and appetite
  6. Sleep and media use
  7. Literature limitations
  8. Discussion
  9. Conclusion
  10. Acknowledgements
  11. References

In China, a large study examined questionnaires completed by parents of 9-year-old children regarding sleep and media use. ‘Media’ was defined as televisions and computers. When the researchers compared children with television sets in their bedroom to those without, they found that children with television sets in their bedroom slept less overall. They found the same to be true for children with computers in their bedroom (Li et al. 2007). Another study looking at adolescents in Belgium found similar results. In this study, adolescents who had either a television or computer in their bedroom spent less time in bed on weekdays (Van den Bulck 2004). Similarly, a study of Finnish adolescents found that high media use led to irregular and less sleep. A gender difference in type of media usage was also found. Boys used the computer and internet more, while girls used a mobile phone more often (Punamaki et al. 2006). All of these studies suggest that media use may play a role in the amount of sleep children and adolescents obtain.

Studies have also shown that children with a television set in their bedroom have an increased risk of being overweight or obese. Dennison et al. (2002) studied children, ages 1–4, from low-income families. The investigators found that children with a television set in their bedroom had an OR of 1.3 for a BMI >85th percentile in comparison with children who did not have a bedroom television set. Another study looking at 9- to 12-year-olds also found that in comparison to children without a television set in their bedroom, children with a television had an OR of 1.3 for being overweight (Adachi-Mejia et al. 2007). A similar study by Barr-Anderson et al. (2008) did not show the same results for adolescents; having a television in the bedroom was associated with other obesity risk factors including poor diet and low physical activity in girls, but it was not associated with being overweight or obese. When combining these results with previous studies on sleep and media, it strengthens the hypothesis that media affects sleep and possibly weight as well. In the USA, the American Academy of Pediatrics (AAP 2001) recommends that parents take the television out of their children's bedrooms.

The literature mentioned thus far on sleep, obesity and media use in the bedroom is limited to paediatric and adolescent populations. Unfortunately, the effects of sleep and media use on obesity in adults have not been looked at in nearly as much depth. Vioque et al. (2000) examined the association between obesity and television viewing as well as sleep duration. Although they did find dose-dependent data showing that less sleep and greater amounts of television viewed per day increased the prevalence for obesity, they were independent associations. A Japanese study using an internet survey examined sleep and media in adults aged 15–79 years. The study looked at self-perceived insufficient sleep and what the person thought the reason for their lack of sleep was. Of those surveyed, 45% ascribed their insufficient sleep to media use before sleep (Suganuma et al. 2007). This is the only study focusing on media use directly before sleep and its possible effect on sleep in adults, but it had flaws, namely there was only a 24.1% response rate to the survey questionnaire, which potentially introduced a selection bias into the data. Overall, the question of media use and sleep is an interesting one, and it is hoped that there will be more research on the topic in the future.

Literature limitations

  1. Top of page
  2. Summary
  3. Introduction
  4. Epidemiologic studies
  5. Sleep and appetite
  6. Sleep and media use
  7. Literature limitations
  8. Discussion
  9. Conclusion
  10. Acknowledgements
  11. References

Although the literature is building with regard to sleep and obesity, the same study limitations appear repeatedly. Most of the sleep studies relied on questionnaires and self-report, for example. The most accurate way to obtain information about a person's sleep, however, would be a device that could accurately record when a person falls asleep, awakens during the night and wakes up after sleep. Studies have evaluated the accuracy of self-reported sleep diaries (logs) compared with wrist actigraphy. A wrist actigraph is a device that people can wear at home that will record their physical activity (Wilson et al. 1998). Activity should be limited while a person sleeps, allowing the actigraph to record such time as sleep for use in sleep studies. The investigators of this study (Wilson et al. 1998) compared sleep diary data with wrist actigraph data. Sleep diaries were filled out when the subjects awoke in the morning and included when they went to bed, fell asleep and arose in the morning, how many times they awoke during the night, and their total sleep. The sleep diaries were found to be consistent with the wrist actigraph total sleep measure, but not in the number of awakenings during the sleep time. Lockley et al. (1999) reported similar results.

Many studies use survey questionnaires asking how much sleep a person usually gets per night, or how much sleep a person gets in a 24-hour period. Knutson and Lauderdale (2007) studied the relationship between self-reported sleep time in adolescents and obesity. The two methods used were sleep duration based on one question, ‘How many hours of sleep do you usually get a night?’, and sleep duration calculated from a time diary where participants recorded all activities throughout a 24-hour period. The time diary is similar to the sleep diary described earlier, and the researchers were able to determine total sleep time for data analysis. Results showed that the time diary and survey self-reported sleep time to differ by >1 hour. Survey self-reported sleep showed a significant association with being overweight, and data from the time diary did not. Methods for assessing sleep differ throughout the literature. Outcomes from studies may be stronger with consistent and valid measures of sleep.

Another limitation with the current literature is the inability of cross-sectional and cohort studies to show direct causal evidence. It is true that the majority of the literature shows an association between SSD and obesity in adults and children, which in turn makes the argument stronger for causality. Spiegel et al. (2004) seem to come close, looking at causality between SSD and leptin, ghrelin and appetite. Unfortunately, the generalisability of the study is small because only 12 healthy young men participated. Furthermore, having participants sleep in a laboratory and not in their natural sleep setting could introduce bias. It would also be unethical to conduct a long-term intervention study comparing people with induced habitual sleep deprivation and people with normal sleep to see if less sleep significantly increases people's weight.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Epidemiologic studies
  5. Sleep and appetite
  6. Sleep and media use
  7. Literature limitations
  8. Discussion
  9. Conclusion
  10. Acknowledgements
  11. References

Although there have been no large randomised controlled trials demonstrating that SSD is a cause for obesity, the positive associations found among the numerous cross-sectional and prospective studies provide a strong suggestion of causality. Even with all of these data, a practical question still remains: Do naps increase or decrease the association between SSD and obesity? Questionnaires among studies provide a mixture of sleep data, with questions to participants including sleep per day, sleep per 24 hours and nightly sleep duration. The NSF (2007b) advises that naps are good for relieving sleepiness and improving mood. They also warn that naps can be detrimental to a good night's sleep if taken too late in the day or for too long. Examining cross-sectional and prospective sleep data with and without naps would add to the data suggesting that SSD causes obesity.

Another interesting and practical question is that of sleep quality. Studies are strengthened when they incorporate physical measurements of sleep including time in bed, time of sleep onset, awakenings, sleep end, and ‘out of bed’ times. Resta et al. (2003) compared people of healthy weight with obese people who did not have obstructive sleep apnoea syndrome. They found that obese people had less sleep efficiency, defined as the ratio of total sleep time to time spent in bed, compared to people with healthy weight. They also found that obese people spent less time in the rapid eye movement (REM) stage of sleep compared with people of healthy weight, which suggests that more detailed studies of sleep and obesity are needed. Vgontzas et al. (1998) found similar results.

Conclusion

  1. Top of page
  2. Summary
  3. Introduction
  4. Epidemiologic studies
  5. Sleep and appetite
  6. Sleep and media use
  7. Literature limitations
  8. Discussion
  9. Conclusion
  10. Acknowledgements
  11. References

The association between sleep and obesity has become a topic of great interest for good reason. Thompson et al. (1999) modelled health consequences of obesity and found increased risks for hypertension, hypercholesterolaemia and type 2 diabetes, as well as increased lifetime risks of coronary heart disease and stroke. If something as simple as getting more sleep at night could alter body weight, the public health message would be simple and clear. Sturm (2002) found obesity in the USA related to 36% more in health-care costs compared to people with a healthy weight. In the UK, obesity costs to the National Health Service were £2.3 billion/year in 2007 and are projected to reach £7.1 billion/year by the year 2050 (with the model assuming no inflation for a direct comparison) (McPherson et al. 2007). By helping to prevent obesity, it should in turn help bring health-care costs down. Based on the current research, clinicians should recommend habitual nightly sleep of 7–9 hours. If a person is consistently getting less than 7 hours of sleep per night, they may not just be tired the next day. They may be increasing their weight and ultimately harming their health.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Epidemiologic studies
  5. Sleep and appetite
  6. Sleep and media use
  7. Literature limitations
  8. Discussion
  9. Conclusion
  10. Acknowledgements
  11. References