A Review of Evidence for the Claim that Children are Sleeping Less than in the Past

Abstract

Study Objectives:

The notion that children are sleeping less than they used to is widespread. This study examined the strength of the evidence for this idea by tracing a “scholarly genealogy” of the claims presented within the literature.

Design:

A systematic review of peer-reviewed literature was conducted to identify claims of a secular trend in children's sleep. For each identified claim, the references cited were reviewed.

Measurements and Results:

The review identified 51 studies. Of these, 17 evinced evidence (2 reported increases, 3 reported no change, 6 reported mixed trends, 6 reported decreases) and 34 provided statements without evidence. Although the evidence that sleep duration has declined is contested, all 34 studies reported a decline. Examination of the references cited revealed that 17 papers referred directly to studies which provided evidence, 4 papers referred indirectly to studies which provided evidence, 9 papers did not provide any evidence and 4 papers referred to studies which could not be located. Of the papers that did provide evidence, 85% referred to one of 3 sources of evidence, each of which was of moderate quality.

Conclusions:

The genealogy of the notion of secular declines in children's sleep reveals a limited scientific basis. The apparent evidence base is inflated by repeated references to the same sources of evidence, reference to secondary sources, mis-referencing, and a failure to cite contrary evidence.

Citation:

Matricciani L; Olds T; Williams M. A review of evidence for the claim that children are sleeping less than in the past. SLEEP 2011;34(5):651-659.

INTRODUCTION

The notion that children are sleeping less than they used to is widespread both in the scientific literature and the popular media. This secular decline, variously ascribed to electrification,^1–3 increased use of technology,^2–4 and “modern lifestyle”^2,3 is believed to have resulted in many children not getting enough sleep.^2,3,5,6

Adequate sleep is important for the growth, maturation and health of children and insufficient sleep has been associated with an array of physical and psychosocial health deficits.^7,8 These include an impaired ability to concentrate^9,10 and retain information,^11,12 mood disorders including anxiety, depression, and hyperactivity,^13,14 as well as impaired motor skills¹⁵ and poorer overall health and immune function.¹⁶ Inadequate sleep has also been associated with impaired academic performance,⁹ an increased risk of injuries and accidents,¹⁷ suicide,¹⁸ and drug and alcohol use.¹⁹ It has been suggested that short sleep duration increases the risk of obesity by increasing sympathetic activity, elevating cortisol and ghrelin levels, decreasing leptin levels and/or impairing glucose tolerance.^20,21 Both cross-sectional and prospective cohort studies on children support this association.^22,23 In light of this connection, together with the growing prevalence of obesity worldwide,²⁴ secular declines in sleep duration have been linked to the secular rise in obesity.^23–25

Declines in children's sleep duration over recent decades have been reported in the scholarly literature. Dollman and colleagues²⁶ noted a 30-min decline between the years 1985 and 2004 for 10- to 15-year-old Australian children on school nights while Iglowstein and colleagues²⁷ reported declines for 1- to 16-year-old Swiss children between 1974 and 1993. Similar trends have also been reported for children from Japan²⁸ and Iceland.²⁹

In spite of such findings, there is also evidence to suggest that that the sleep duration of children has not declined over the years. Hofferth and Sandberg³⁰ reported an increase in the sleep duration of 3- to 12-year-old American children between the years 1981 and 1997, while Huysmans³¹ reported no change for 15- to 18-year-old children in the Netherlands between 1980 and 2000. Pääkönen³² found that children in Finland were getting more sleep on non-schooldays and less sleep on schooldays in 2000 than 1987. Randler,³³ on the other hand, noted that the sleep duration of German children had increased from 1907 until the 1970s, at which time a decline commenced.

Although the evidence of secular trends in children's sleep appears conflicting, claims of a decline as a result of modernisation appear to be dominant. Frequently featured in contemporary journal articles are claims that modern society has led to a reduction in sleep time.^1,6,24,34 Comments such as “…sleep loss is one of the common plagues of modern societies,”³⁵ “You probably won't be surprised to learn that we sleep a lot less than we did 200 years ago,”³⁶ and “…children today are getting less and less sleep over the years”³⁷ are common in both scientific and popular literature. Although these statements suggest that modern society is facing a new and growing problem of sleep loss, such claims are by no means new. In 1894, the British Medical Journal published an editorial titled “Sleeplessness,”³⁸ which stated:

The subject of sleeplessness is once more under discussion. The hurry and excitement of modern life is quite correctly held to be responsible for much of the insomnia of which we hear; and most of the articles and letters are full of good advice to live more quietly and of platitudes concerning the harmfulness of rush and worry.

The belief that children are getting less sleep than they should, and that this deficit is due to modernism, has persisted for at least a century. Given the conflicting evidence around secular trends in children's sleep and the apparent scarcity of scientific evidence,^4,5 the origin and the strength of these claims are of interest. This paper aims to examine the strength of the scientific evidence for this idea by tracing a “scholarly genealogy” of the claims presented within the literature.

METHODS

A systematic search strategy was employed to identify claims of a secular trend in children's sleep duration. Four electronic databases (EbscoHost, Scopus, Ovid, and Web of Science) were searched (21/12/09) using the terms “sleep time,” “sleep duration,” “sleep quantity,” “time spent asleep,” “time spent sleeping,” “time sleeping,” “time asleep,” “sleep length,” “trend*,” “secular,” “saecular,” “chang*,” “increas*,” “decreas*,” “declin*,” “child*,” “teen*,” “youth*,” “boy*,” “girl*,” “adolescen*,” “student*”. No limits were specified, and papers written in a language other than English were translated.

Papers were initially retained and read in full if the abstract, title, or introduction reported a secular trend in sleep duration. For discussion papers, the entire article was read. Papers were then included for analysis if an explicit statement of a secular trend in children's sleep duration appeared anywhere within the article. For the purpose of this study, a “child” was defined as any individual between 5 and 18 years of age. Consequently, statements describing secular trends in “adolescents,” “teens,” “boys,” “girls,” etc., were included for analysis, while papers describing adults, infants, or no specific age group were excluded.

In addition to searching electronic databases, reference lists of articles included in the review were examined for further studies which might meet the inclusion criteria, and professional colleagues were invited to provide any additional papers meeting inclusion criteria.

For each of the claims identified, the references cited were reviewed, and the authors were contacted in cases where evidence was unclear. A scholarly genealogy of claims made was then developed, tracing links between papers through citations to determine the origin for claims. This method involved retrieving all papers referred to as providing evidence for claims made. In doing so, this review distinguished between papers that made claims of a secular trend (i.e., secondary sources of evidence) and papers that evinced evidence of a secular trend (i.e., primary sources of evidence). The strength of the evidence for claims made was then determined by analysing the results of all primary sources of evidence and critically appraising the methodological bias and quality for each of these articles. A modified critical appraisal tool developed by Tooth and colleagues³⁹ was used to determine the methodological bias of reporting. In addition to this, the location of evidence (i.e., where the study was actually available) was also reviewed, that is, whether the study was published in a journal, could be retrieved via Google (abstract or full text) or was identified by searching the gray literature. Gray literature refers to unpublished studies retrieved via communication with authors or reports identified on time-use institution websites.

RESULTS

This review identified 51 papers claiming secular trends in children's sleep duration, of which, 17 (Figure 1) evinced evidence (2^30,40 reported an increase, 3^31,41,42 reported no change, 6^{32,33,43–45,75} reported a mixed trend, 6^{26–29,46,47} reported a decrease), and 34 provided a claim without evincing evidence. Of the 34 providing only claims, all referred to a decline and, although 3^4,5,7 papers acknowledged a scarcity of evidence, not one paper made any mention of the presence of conflicting evidence. Thus, while there appears to be more studies suggesting that sleep duration has not declined, only claims of a decline were identified. It was therefore of interest to determine the references cited for each of the claims identified and to evaluate the evidence and the location for each of the sources that evinced evidence.

Figure 1.

Secular trends in children's sleep as reported within the literature. The graph above illustrates findings of 15 of the 17 studies identified for selected ages. The results of two studies are not illustrated as sleep duration values were not explicitly reported²⁹ or did not report the sleep times for children in the age groups graphed.²⁸

*Zuzanek examined secular trends in children sleep from 7 different countries. While the years of measurement and average sleep duration were specified for Finland, Germany, Canada, and Norway, they were not specified for Australia, France, and the USA, all of which were identified as showing increases. For this reason, only trends for Finland, Germany, Canada, and Norway have been included in the graph above.

The Evidence

As shown in Table 1, 17 studies evinced evidence of a secular trend in children's sleep duration. Of these, 2^30,40 showed increases, 3^31,41,42 showed no change, 6^{32,33,43–45,75} showed mixed results, and 6^{26–29,46,47} showed decreases. Of the 6 studies that showed mixed results, 2^32,43 found opposite trends on the different day types (i.e., schooldays versus non-schooldays), one⁴⁵ found opposite trends for children of different age groups, one⁴⁴ found opposite trends for different countries, and 2^33,75 found opposite trends for different time periods. All of the 17 studies which evinced evidence were specific for children of certain ages, from specific countries, on particular day types and often, over relatively short periods of time. Such considerations are important since age, country, and day type are variables known to affect sleep duration, while the examination of secular trends over short time periods increase the risk of statistical volatility. In light of this, the current available evidence is unable to provide support for claims of a secular trend.

Table 1.

Summary of studies that evince evidence for a secular trend in children's sleep duration

Trend	Study	Country	Study design	Sampling method	Reporting	Sleep definition	Method	Day type	Years	Sample (n)	Age (yr)	CAT score
Decrease	26a	Australia	Comparisons of 2 cross-sectional studies	Stratified random sample (state)	Self-report	TIB	Questionnaire	SD	1985–2004	900	10–15	13
	27	Switzerland	Comparisons of 3 longitudinal cohorts	Zurich Longitudinal Study	Proxy-report	TIB & TST	Questionnaire	Week	1974–1993	493	1–16	13
	28b	Japan	Unknown	—	—	—	—	—	1955–1995	Unknown	1	—
		Japan	Unknown	—	—	—	—	—	1965–2000	Unknown	Grade 3–6
	29	Iceland	Comparisons of 2 cross-sectional studies	Stratified random sample (country)	Self-report	TIB & TST	Diary	Week	1985–1995	1,308	1–19	15
	46	USA	Comparisons of 2 cross-sectional studies	Unknown	Self-report	Unknown	Questionnaire	Week	1910–1964	2,911	8–17	6
	47b	Finland	Unknown	—	—	—	Questionnaire	SD	1994–1998	5,646	13&15	—
Mixed	32	Finland	Comparisons of 2 cross-sectional studies	Stratified random sample (country)	Self-report	TST*	Diary	SD & NSD	1988–2000	2,348	10–18	12
	33	Germany	Comparisons of 7 cross-sectional studies	Multiple sampling frames	Self-report	TIB	Questionnaire	SD	1907–2007	4,048	10–14	10
	43	Netherlands	Comparisons of 2 cross-sectional studies	Stratified random sample (country)	Self-report	Unknown	Diary	SD & NSD	1990–2000	398	16–19	14
	44c	Canada	Comparisons of 2 cross-sectional studies	Stratified random sample (country)	Self-report	Unknown	Diary	Week	1986–1998	1,376	15–19	8
		Finland					Diary	Week	1987–2000	1,375	15–19
		Germany					Diary	Week	1991–2001	1,986	15–19
		Norway					Diary	Week	1980–2000	683	15–19
	45	Japan	Comparisons of 2 cross-sectional studies	Unknown	Self-report	Unknown	Questionnaire	Week	2001–2006	Unknown	10–19	4
	75	Finland	Comparisons of 2 cross-sectional studies	Stratified random sample (country)	Self-report	TIB	Questionnaire	SD	1994–2006	12,528	13&15	10
No change	31	Netherlands	Comparisons of 2 cross-sectional studies	Stratified random sample (country)	Self-report	Unknown	Diary	Week	1980–2000	898	15–18	8
	41	USA	Comparisons of 3 cross-sectional studies	Multiple sampling frames	Self-report	Unknown	Questionnaire	Week	1911–1980	5,791	1–16	12
	42	USA	Comparisons of 2 cross-sectional studies	Stratified random sample (country)	Self-report	TIB	Diary	SD – NSD	1981–2006	3,108	15–17	8
Increase	30	USA	Comparisons of 2 cross-sectional studies	Stratified random sample (country)	Self-report	Unknown	Diary	Week	1981–1997	2,341	3–12	13
	40	Australia	Comparisons of 2 cross-sectional studies	Stratified sample (country)	Self-report	Unknown	Diary	SD – NSD	1992–1997	2,639	15–19	8

“Study” refers to papers that evince evidence and is denoted by the reference list number. “Reporting” refers to whether questions regarding sleep duration were self- or proxy-reported. “Sleep definition” refers to how sleep duration was defined in the study (“TST” refers to nightly total sleep duration, excluding latency periods; “TIB” refers to time in bed; “TIB & TST” refers to studies that assessed trends in children's TIB and TST; “TST*” refers to studies that assessed day and night sleep; “unknown” was allocated to studies that did not explicitly state how sleep duration was defined). “Method” refers to the method used to collect sleep data; “Day type” refers to the day type examined (“SD” refers to schooldays, “NSD” refers to non-schooldays, “week” refers to weekly average). “CAT score” refers to the “critical appraisal tool score.” The tool used in this study was a modified version of the tool developed by Tooth and colleagues.³⁹ The highest possible score that could be attained using was 25.

^aStudy 26 examined secular trends for a mixed sample of boys and girls as well as boys and girls separately. All other studies examined samples of children consisting of an approximately equal number of boys and girls.

^bStudy 28 and 47 were written in a language other than English. Although the results of these studies were provided by the authors, a complete translation was not provided and thus a CAT score was not provided for these studies. Methodological details not obtained for these studies are denoted by “—”.

^cStudy 44 examined secular trends in children's sleep from seven different countries. While the years of measurement and average sleep duration were specified for Finland, Germany, Canada and Norway, they were not specified for Australia, France, and the USA. For this reason, only trends for Finland, Germany, Canada, and Norway have been included in the table above.

NB: Efforts were made to contact authors in cases where details were not provided.

To critically appraise the studies that evinced evidence of methodological bias, a modified tool developed by Tooth and colleagues³⁹ was used. As shown in Table 1, there were no substantial differences in the quality of reporting for studies that evinced evidence of a decrease, mixed trend, no change, or an increase (average score ˜10). Sample sizes were also similar, averaging about 2,000, as were the study designs, sampling methods, reporting methods, and definitions of sleep.

Location of Evidence

As presented in Figure 2, although most of the studies that evinced evidence of a secular trend in children's sleep duration were published in a journal, those that showed decreases were cited more often than those that showed an increase, no change, or a mixed trend. Similarly, a Google search for each of the studies that evinced evidence revealed that studies showing a decrease were more likely to be identified and were generally cited more. It was also found that freely available studies were cited more than those that were not.

Figure 2.

Location of evidence of a secular trend in children's sleep duration according to study's results and times cited. The location of the evidence, that is, where the study is actually available was determined by assessing the source of publication and by searching Google. Whether an article appeared in Google was determined by typing the title of the article in single quotation marks. Citation frequency was calculated through the use of Google Scholar. Study reference: 26 Dollman, 27 Iglowstein, 28 Kohyama, 29 Thorleifsdottir, 30 Hofferth, 31 Huysmans, 32 Pääkönen, 33 Randler, 40 Soupourmas, 41 Weissbluth, 42 Knutson, 43 Vaage, 44 Zuzanek, 45 Statistics Bureau, 46 Webb, 47 Tynjälä (2002), 75 Tynjälä (2010).

References Cited

As shown in Figure 3, in reviewing the references cited for each of the 34 claims identified, 17^{4,6–8,48–59,62} papers referred directly to studies which provided evidence (open circles), 4^3,5,25,65 papers (gray-shaded circles) referred indirectly to evidence (i.e., by referring to a study which itself contained a reference), 9^{2,24,34,63,66–70} papers (black-shaded circles) did not refer to a specific study that evinced evidence of secular trend in children's sleep duration, and 4^60,61,64,71 papers (double-edged circle) referred to studies^72–74 which could not be located despite efforts to contact the authors and searching overseas libraries.

Figure 3.

Scholarly genealogy of claims within the literature.

Of the 17 papers that referred to studies that evinced evidence, 13^{4,6,7,48–57} referred to at least one of two studies: Dollman et al.²⁶ and Iglowstein et al.²⁷ In the remaining 4 papers,^8,58,59,62 4 different sources of evidence were referred to.^28,29,46,47 Of these, 2^28,47 were obtained and translated with assistance from the authors, and 2^29,46 were retrieved from database and library searches.

Of the 4 papers that referred indirectly to studies which provided evidence, 2^3,5 referred to evidence originally provided by Webb⁴⁶ via reference to Webb and Agnew.⁶² Of the remaining 2 papers, one²⁵ referred to evidence originally provided by Iglowstein et al.²⁷ via reference to Keith et al.,⁴⁸ and the other paper⁶⁵ referred to a study²⁴ that made an explicit claim of a decline in children's sleep duration, but which in fact referred to evidence⁸⁰ describing changes in the sleep duration of adults. Thus, of the 4^3,5,25,65 papers that referred indirectly to studies which provided evidence, only 3^3,5,25 supported claims made for children.

Of the 9 papers that did not refer to a specific study that evinced evidence of secular trend in children's sleep duration for claims made, 3^63,67,68 did not provide a reference, one⁶⁶ referred to a paper that did not contain any statement of a secular trend, and 5^{2,24,34,69,70} referred to cross-sectional data. Of the 5 studies that referred to cross-sectional data, 3^24,69,70 referred to the sleep of adults, and 2^2,34 referred to the sleep of children. Of the studies that referred to children, cross-sectional data from 3 different points in time were provided. However, one study³⁴ compared the sleep duration of children from different countries, while the other study² referred to the sleep of children at different ages.

Thus, of the 34 papers that presented claims, 20 (17^{4,6–8,48–59,62} referred to papers that evinced evidence and 3^3,5,25 referred indirectly to studies which provided evidence) provided a reference to support claims made. Of these, 17 (85%) referred to Iglowstein and colleagues,²⁷ Dollman and colleagues,²⁶ or Webb.⁴⁶ It therefore appears that the widespread notion of a secular decline in children's sleep duration is based largely upon the circulation of the same references rather than a critical evaluation of the available evidence. To gain a better understanding of why these sources were cited more than others, all sources of evidence were evaluated and the location for each source was determined.

DISCUSSION

This review reveals a characteristic pattern of referencing that favors the propagation of widely accepted beliefs rather than the available evidence. Despite contemporary efforts to produce “evidence-based” research, this review reveals selective citation of evidence, a lack of attention to the appropriateness of evidence cited, a preference for citing secondary rather than papers that evinced evidence and perhaps a misunderstanding or misinterpretation of studies or results. Although it may be argued that authors are largely responsible for ensuring that they accurately represent concepts and notions and offer an unbiased view of the available evidence, our study identifies a similar pattern in the dissemination of evidence which also favors the propagation of widely accepted beliefs.

Selectively Citing Evidence

While there appear to be more studies to suggest that sleep duration has not decreased amongst children, papers citing secondary evidence referred exclusively to declines. Not one author acknowledged the presence of contrary evidence and few presented a case that the evidence was equivocal. Although methodological quality may influence citation rates, there did not appear to be any methodological differences between sources of evidence that may have been responsible for the differences in citation rates in this study. Specifically, most studies compared at least two cross-sectional studies, utilized a stratified random sampling method and required self-report measures of sleep to determine secular trends. While there are slight differences in the definition of sleep [time in bed (TIB) vs. total sleep time (TST)] between studies, this does not appear to be related to the trend identified. In contrast, noticeable differences in the methods of data collection (diary vs. questionnaire) are present, with diary data more strongly associated with mixed, no change, or increased sleep, and questionnaire data more strongly associated with decreases. Although diary data are often regarded as a superior method of measuring sleep, studies^77,76 indicate no difference between the two methods. While it may be possible that the two methods are not equivocal in estimating children's sleep time, there is no evidence to suggest that the two methods track changes in sleep differently. Thus, although, there is an apparent difference, it is unlikely that these differences are responsible for the differences in the trends identified.

Such selectivity in citing evidence does not appear unique to claims on sleep. Ravnskov,⁷⁸ for example, showed that although there were an equal number of supportive and unsupportive trials to suggest that a cholesterol lowering diet is protective against heart disease, the supportive trials were cited more frequently than the unsupportive trials. For example, in 16 trial reports published after 1970, a total of 40 supportive or inconclusive trials were cited but only a single unsupportive trial was cited. Similarly, Cope and Allison⁷⁹ have identified “white hat bias” in the reporting of the relationships between obesity and breastfeeding and soft drink consumption.

The volume of literature available today requires authors to conscientiously and diligently make decisions concerning which references to include or exclude, both from the perspective of searching and reviewing the literature and requirements of peer reviewed journals. Such selective pressure, according to Leimu and Koricheva,⁸⁰ leaves room for secondary motives such as citing evidence in order to persuade readers to believe particular ideas and selectively citing studies that are more likely to provide supporting evidence.

Citing the Same Sources of Evidence

Although 59% of the papers identified in this review cited evidence for claims made, 85% of these referred to just one of three studies (Dollman et al.,²⁶ Iglowstein et al.,²⁷ or Webb⁴⁶). Although these studies provide evidence of a decline in sleep duration, Dollman and colleagues²⁶ examined the sleep duration of Australian children between the ages of 10 and 15 years of age only, and only school nights were examined. Similarly, Iglowstein and colleagues²⁷ examined the sleep duration of Swiss children only and did not examine the different day types. Webb⁴⁶ also reported a secular decline by comparing cross-sectional data of 311 children in 1964 with 2000 children in 1910/11, despite authors of the 1910 study warning that that their values were unusually high compared with other data at the time. None of the claims in this review acknowledged these limitations. Interestingly, Weissbluth and colleagues,⁴¹ who compared the data from 1910 with those of 1980 and reported no change in children's sleep duration, were not cited by any of the papers.

Citing evidence on the basis of its ability to support popular ideas, rather than scientific merit, is not uncommon. Several attempts have been made to explain why certain studies are cited more often than others. Lawrence,⁸¹ in investigating whether the availability of papers influence the number of times they are cited, suggested that papers freely accessible online were more likely to be cited than those that were not. Indeed, in the present study, the papers available online, especially those with free full-text, were cited more than others (Figure 3). This explanation may, in part, explain the results of our review. For example, when Googling (18 January 2010) “trends in children's sleep duration,” just under 5 million hits were identified. The first result was Iglowstein et al.,²⁷ which is freely available and the reference most commonly referred to in our review. The second result was Dollman et al.,²⁶ for which the abstract containing the results of the study is freely accessible, and was the second most referenced in our review. The subsequent 200 records all appeared to support a decline in sleep duration, and all referred to either Iglowstein et al.,²⁷ Dollman et al.,²⁶ or Webb and Agnew.⁶² This finding lends further support to the theory that the number of times a paper is cited is related to the ease of accessibility of the paper.

Misinterpretation of Evidence

Papers that cite secondary sources of evidence rather than the original reports may not accurately describe and interpret original findings and may be more likely to result in misinterpretations and inaccurate beliefs.⁷⁹ This was evident in tracing the origins for many of the claims identified in this review. For example, Bixler,⁶⁵ in claiming that American adolescents sleep less than they used to, cited Van Cauter,²⁴ who in turn referred to the National Health Interview Survey,⁸² which provided evidence of a secular decline in the sleep duration of adults rather than children. Similarly, Seicean,⁷⁰ claimed a decline in the sleep duration of children, but citing Kripke,⁸³ National Sleep Foundation,^84–86 and Sugino et al.⁸⁷ who provided evidence relating only to adults or did not mention sleep at all.

Although there may be several reasons why authors misinterpret and reframe evidence, one possible explanation, as described by Antelman,⁸⁸ is “hollow citing,” the process whereby authors do not actually read the paper that they are citing. This is one possible explanation for why, in our study, papers included references that either described adults or did not mention sleep at all, to support their claim of a secular decline in children's sleep duration.

“negative results have never made riveting reading”⁸⁹

Publication bias has been described within the literature as early as 1959⁹⁰ and reflects the tendency of journals to publish studies which report significant findings. This process has been blamed for providing an unequal representation of the available evidence.⁹¹ While the sale of advertising space and journal subscriptions may underpin the viability of a journal, publishing papers on the basis of their ability to capture an audience rather than to disseminate the available evidence may result in an inherent difficulty with obtaining a comprehensive, balanced view of the available evidence. The legacy of publication bias last long after the original publication date. This issue is further exacerbated by the media, which also reports evidence in this manner. Since the general public, health professionals, and scholars heavily rely on these two sources to obtain information, it is clear how ideas with little scientific evidence are able to propagate throughout the literature and media and become part of received wisdom.

Implications for Practice

Claims that consistently appear within the literature with little or no acknowledgement of contradictory findings may be interpreted as factual evidence which in turn has the potential to influence policies, guidelines and practices. This is noticeable in our review, whereby interventions directed towards “curbing the sleep epidemic” are described as “necessary interventions” required to address the coexisting obesity epidemic.⁹² This is concerning since policies and guidelines are less likely to achieve set goals and are more likely to cause harm if they are not based on all available evidence. Given that the patterns identified in this review have been acknowledged by several other studies, it is probable that many of the ideas presented within the literature are based on an unbalanced view of the available literature. Although this is in part, due to selectivity in citing evidence, the findings of our review suggest that even people of goodwill trying to be objective by searching respectable databases and following rigorous scientific processes may still be misled and exposed to a skewed view of the available evidence.

It is important to note that this study did not aim to confirm or refute the claim that sleep has declined in children. The current evidence is conflicting, of low methodological quality and specific to particular populations. Given the implications that this popular belief may have, there is a real need for comprehensive review of all available evidence to quantify secular trends in children's sleep duration. Such a study is important as results may offer epidemiological insight into the prevalence of various conditions such as diabetes, cardiovascular disease and obesity, all of which have been associated with both short and long sleep durations.^93–96 Such information may therefore also be useful in developing more effective prevention and treatment strategies for such conditions.

DISCLOSURE STATEMENT

This was not an industry supported study. The authors have indicated no financial conflicts of interest.