Sleep Hygiene Practices of Good and Poor Sleepers in the United States: An Internet-Based Study

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Sleep Hygiene Practices of Good and Poor Sleepers in the UnitedStates: An Internet-Based Study

Les A. Gellis,a University of Memphis and Philadelphia Veterans Medical Center

Kenneth L. Lichstein,b University of Alabama

This Internet-based investigation assessed the frequency ofmultiple sleep hygiene practices in 128 good sleepers and 92poor sleepers (mean age=41.6, SD=12.8, 61.8% women).The Pittsburgh Sleep Quality Index was used to measuresleep quality (b5=good sleep, N7=poor sleep). Althoughsleep hygiene practices were generally good, poor sleepersshowed increased cognitive activity in the bed, even aftercontrolling for global indices of depression and anxiety.Poor sleepers also reported statistically significant increasesin excessive noise in the bedroom, uncomfortable nighttimetemperature, and activities that were exciting, emotional, ordemanded high concentration near bedtime. Future studiesshould further investigate the prevalence of these variablesand their potential impact on sleep quality.

INSOMNIA IS THE MOST prevalent sleep disorder, and itis characterized by problems initiating and/or main-taining sleep. Prevalence in the United States isestimated at 9% to 16% for chronic insomnia(Ancoli-Israel & Roth, 1999; Ford & Kamerow,1989; Lichstein, Durrence, Riedel, Taylor, & Bush,2004) and 27% for occasional insomnia (Ancoli-Israel & Roth, 1999). Poor sleep is associated withincreased fatigue, psychological distress, risk ofsuicide (Riedel & Lichstein, 2000; Taylor, Lichstein,& Durrence, 2003), decreased immune functioning

(Hall et al., 1998; Savard, Laroche, Simard, Ivers, &Morin, 2003), higher medical costs, increaseddisability, and greater limitations of activity (Simon& VonKorff, 1997).Attempts to reveal factors contributing to insom-

nia have partly focused on the role of sleep hygiene(SH). SH broadly refers to a set of behaviors thatinfluence the quality of oneTs sleep. The formaldiagnostic category, inadequate sleep hygiene, iden-tifies insomnia due to SH (The International Classi-fication of SleepDisorders, Second Edition [ICSD-II],American Academy of Sleep Medicine, 2005). Thisinsomnia subcategory is defined by engaging in oneor more behaviors related to the following fivecategories: (1) improper sleep scheduling, (2) the useof sleep-disturbing products, (3) engaging in activat-ing or arousing activities close to bedtime, (4) the useof the bed for activities other than sleep, and (5)maintaining an uncomfortable sleeping environment.Experimental investigations have revealed that

many of these SH variables disturb sleep (Riedel,2000; Stepanski & Wyatt, 2003), and studies haveshown a significant association between poor globalSH characteristics and poor sleep (Brown &Buboltz, 2002; Gallasch & Gradisar, 2007; Lacks& Rotert, 1986; Mastin, Bryson, & Corwyn, 2006).However, questions remain concerning the overallimpact of SH. Lacks and Rotert (1986) noted thatSH adherence was generally high in both good andpoor sleepers, and concluded that SH is not aprimary cause of insomnia. This conclusion wassupported by The American Psychological Associa-tion/National Institute of Mental Health DSM-IVfield trial, which showed that only 6.2% of patientsin sleep clinics with a presenting complaint ofinsomnia were given a primary diagnosis ofinadequate sleep hygiene (Buysse et al., 1994).That same study also found that, when secondarydiagnoses were included, the disorder was applied to

Available online at www.sciencedirect.com

Behavior Therapy 40 (2009) 1–9www.elsevier.com/locate/bt

Les A. Gellis is now at the Philadelphia Veterans Medical Center.This work was supported by a grant from the University ofMemphis, Department of Psychology.

Address correspondence to Les A. Gellis, Philadelphia VeteransMedical Center, MIRECC (116, 2nd floor), University and Wood-land Avenues, Philadelphia, PA 19104; e-mail: [email protected]/08/0001–0009$1.00/0© 2008 Association for Behavioral and Cognitive Therapies. Published byElsevier Ltd. All rights reserved.


34.2% of the patients. Together, these studiessuggest that SH is not a primary factor drivinginsomnia, yet its impact may be felt in a goodportion of people with the condition. It is currentlyunclear which individual SH practices in thecommunity are more likely to be a problem.Understanding the potential impact of individualSH behaviors in the population is of value in order totarget unhealthy behaviors for community-basededucation programs and to drive the focus of futureresearch.One way of gathering information related to the

impact of individual SH practices is to compare thefrequency of SH behaviors in good and poorsleepers. The few studies that have utilized thismethod have reported mixed results. Initial studiesassessing the impact of sleep hygiene behaviorsfocused on sleep-incompatible behaviors associatedwith the bed and the bedroom (ICSD-II categories3 & 4). Although incompatible sleep behaviors didnot discriminate between college students with andwithout sleeping difficulties (Haynes, Follingstad,& McGowan, 1974), they were positively asso-ciated with sleep-onset insomnia in psychiatricpatients (Kazarian, Howe, Merskey, & Deinum,1978).More recent studies have assessed SH behaviors

in good and poor sleepers across a more diverserange of variables. Adult volunteers with sleep-onset insomnia were compared to 30 individualswithout insomnia on a number of different SHbehaviors, and people with insomnia reportedworse sleep efficiency, the ratio of total sleep timeto the time spend in bed, and an indicator of usingthe bed for activities other than sleep (Harvey,2000a). Another investigation focused on womenvolunteers between the ages of 40 and 55 andreported that people with insomnia were morelikely to abstain from alcohol and use less caffeine(Cheek, Shaver, & Lentz, 2004), suggesting thatthose with insomnia may have better SH thannormal sleepers. A recent study used random digitdialing to recruit 258 people with and withoutinsomnia and found significantly greater napping,alcohol usage, and smoking behaviors in peoplewith insomnia (Jefferson et al., 2005). Individualswith insomnia were also more likely to sleep in ondays they did not work and reported worse sleepefficiency. Among older adults, napping frequencywas greater in those who complained aboutinsomnia symptoms as compared to noncomplai-ners in a sample generated in Memphis, TN;however, the same investigation showed no differ-ences in any SH variable using a sample of olderadults from North Central Florida (McCrae et al.,2006).

It is unclear why SH differences existed betweenstudies. Cheek et al. (2004) excluded individualswith sleep-disordered breathing, other significantmedical conditions and psychiatric disorders, whichmay have minimized the relation between SH andinsomnia. That study also only included women ina particular age range, and it is possible that thisspecific demographic is not susceptible to problemswith sleep hygiene. None of the studies reported afull array of behaviors as reflected in ICSD-IIcriteria for inadequate sleep hygiene. For instance,none of the recent investigations assessed emotionalor physically arousing behaviors at bedtime. OnlyHarvey (2000a) assessed environmental conditionsthat may have affected sleep in adults, and theauthor's sample was relatively small for an epide-miologic investigation.One way to efficiently collect large amounts of

information about community behaviors is throughthe Internet. Several studies have noted that onlinequestionnaire data share similar reliability andvalidity characteristics with traditional methods(Eysenbach & Wyatt, 2002; Gosling, Vazire,Srivastava, & John, 2004). Although initial datasuggested that Internet users were more malad-justed and unrepresentative of the general popula-tion, more recent studies have disputed this claim(Gosling et al., 2004). The disadvantages to internetsamples include self-selected volunteer populationsand limited representation by individuals of lowersocioeconomic status and the elderly; however,these problems are consistent with the majority oftraditional studies (Gosling et al., 2004). Fewinvestigations have used an Internet sample tocollect health behaviors, and only one known studyhas attempted to assess sleep-related characteristicsusing an online sample (Gallasch & Gradisar,2007).The purpose of this study was to assess the

impact of SH characteristics in the community.Because global SH characteristics have been shownto be generally good, we hypothesized that bothgood and poor sleeper would show good overalladherence to SH behaviors. We also explored thefrequency of individual SH in good and poorsleepers to assess which variables may be a problemin the community. We were able to assess a widedemographic range from individuals across thecountry. We compared good and poor sleepersacross a variety of SH variables using the ICSD-IIdiagnosis of inadequate sleep hygiene as a guide fordetermining important SH items. Thus, we assessedan inclusive set of pertinent SH behaviors. We alsocollected health information and were able toexclude individuals with suspected other sleepdisorders.

2 gell i s & l i chst e in


Table 1Sample demographics compared to estimated united statesdemographics in 2005

Demographic Sample United states

EthnicityCaucasian 180 (81.8.7%) 74.7%African-American 16 (7.3%) 12.1%Hispanic or Latin American 8 (3.6%) 14.5%Asian or Pacific Islander 13 (5.9%) 4.4%

Education levelHigh school dropouts 14 (6.4%) 16%High school graduates 36 (16.4%) 30%Some college 31 (14.1%) 27%College graduates 139 (63.2%) 27%

GenderMales 82 (37.3%)Females 136 (61.8%)

Age Mean=41.6SD=12.8

Note. United States ethnicity and educational attainment esti-mates were obtained from http://factfinder.census.gov.

Methodsample and procedures

Data were collected through a national Internet-based investigation, with the assistance of Study-Response, a nonprofit agency based at SyracuseUniversity (Stanton & Weiss, 2002). StudyRe-sponse facilitates online research by maintaining adatabase of individuals interested in participating inInternet-based studies. In the present investigation,randomly chosen panelists associated with Study-Response were sent email messages regarding anongoing study about “sleeping patterns and sleep-related behaviors” and were given a link to thestudy’s informed consent. Interested persons weredirected to the study materials, and completed thesurveys online. After completing the questionnairesthe participants were entered into a raffle for ten$79 gift certificates for Amazon.com. Email mes-sages were sent to 4,945 residents of the UnitedStates, and 817 individuals completed the surveys.Of the participants who completed the surveys, 116had either incomplete or inconsistent data, repre-senting a final return rate of 14.2%. We surmisethat the possibility of receiving no compensation ledto the low response rate. Of the 701 individualswho completed the study materials, 32 (4.6%)individuals were excluded because of nightshiftemployment. Additionally, 310 (44.2%) wereexcluded because of the suspicion of a sleepingdisorder other than insomnia. Any of the followingled to exclusion from the study: (a) a report offrequent leg jerks during sleep or restless legs beforesleep onset, (b) admitting to difficulty breathing orgasping for breath during sleep, (c) a report ofheavy snoring and general sleepiness during theday, or (d) admitting to heavy snoring with a bodymass index N30. Participants admitted into thestudy were classified as good or poor sleepers basedon scores from the Pittsburgh Sleep Quality Index(PSQI; Buysse, Reynolds, Monk, Berman, &Kupfer, 1989). Individuals with PSQI scores b5were categorized as good sleepers, and those withscores N7 were classified as poor sleepers. Thus,data from individuals with scores 5, 6, and 7 on theborder of good and poor sleep were not analyzed.The completed sample was composed of 220

participants. The demographics of the study sampleare described in Table 1, which also includesprevalence estimates of ethnicity and educationalattainment in the United States in 2005. Analysis ofTable 1 reveals that Caucasian individuals wereoversampled and African Americans and indivi-duals of Hispanic or Latin American descent wereundersampled. Further, this sample overrepre-sented college graduates, and underrepresented all

other education groups. This sample also included agreater proportion of females.

measures

Sleep hygiene. An investigator-designed question-naire was used to assess SH (see Table 4). This 19-item measure was based on ICSD-II criteria forinadequate sleep hygiene and lists a variety ofactivities that are characteristic of poor SH. Respon-dents state the average number of days per week inwhich they engaged in these poor SH activitiesduring the previous month. Frequency scores(number of days per week) are calculated for eachitem, and higher frequency scores indicate worse SH.Item 2 (Woke up at approximately the same time)and Item 3 (Went to bed at approximately the sametime) are features of good sleep hygiene, and theseitems are scored in reverse.This instrument was used because no other

known measures of SH assess a full range ofbehaviors as reflected in current diagnostic criteria.We established cutoffs on a number of variables toidentify responses of poor SH. For instance,exercise between 0 to 4 hours before bedtime wasclassified as poor SH. This cutoff was chosenbecause exercise within 4 hours of bedtime waslinked to increased sleep-onset latency, whereasexercise between 4 to 8 hours before bedtime wasassociated with decreased sleep onset latency(Youngstedt, O'Conner, & Dishman, 1997).Because caffeine may exert its influence up to 10hours after consumption (Roehrs & Roth, 1997),caffeine use 0 to 10 hours before bedtime was usedto identify poor SH. Most individuals will be moreaffected by caffeine closer to bedtime so we divided

3s l ee p hyg i ene pract i c e s


caffeine usage into two variables: caffeine usagebetween 0 and 5 hours before bedtime and caffeineusage between 5 and 10 hours before bedtime.Cutoffs related to alcohol usage (3 hours beforebedtime) and smoking (2 hours before bedtime)were derived from our clinical experience in work-ing with insomnia patients. Because individualsshow different sensitivities to SH variables, thesecutoff points are considered estimates of problembehaviors.Five thematically related groups were created

based on categories for the ICSD diagnosis ofinadequate sleep hygiene. Items 1–3 refer toimproper sleep scheduling behaviors. Nappedduring the day and Woke up at approximately thesame time are two examples of these behaviors.Items 4–7 refer to the usage of sleep-disruptingproducts. Drank caffeinated beverages such ascoffee or tea or soft drinks between 5 to 10 hoursbefore bedtime and Drank alcohol within 3 hoursof bedtime are examples of these behaviors. Items8–11 refer to activating or arousing activities nearbedtime. Sample items includeWorried, planned, orthought about important matters at bedtime andEngaged in exciting or emotionally upsettingactivities near bedtime. Items 12–15 involve theuse of the bed for activities other than sleep:Read inbed and Watched TV in bed are examples of thesebehaviors. Items 16–19 assess environmental char-acteristics. Slept on an uncomfortable mattress andSlept in a room with an uncomfortable nighttimetemperature are sample items.

Pittsburgh Sleep Quality Index (PSQI). ThePSQI is a retrospective self-report questionnairethat measures sleeping patterns and sleep distur-bances that existed during the previous month(Buysse et al., 1989). The PSQI yields a global scoreranging from 0 to 21, with higher scores indicatingworse sleep quality. The questionnaire measuresmany components of the sleep experience, such assleep onset latency, total sleep time, habitual sleepefficiency, the use of sleep medications, daytimeimpairment, and subjective sleep quality. The PSQIhad strong test-retest reliability, and using a cutoffscore of 5, the measure demonstrated a sensitivityof 89.6% and specificity of 86.5% in separatingpeople with and without insomnia (Buysse et al.,1989). Recent studies, however, have suggested thata cutoff of 6 maximized sensitivity and specificity(Backhaus, Junghaans, Broocks, Riemann, &Hohagen, 2002; Tsai et al., 2005). Other studieshave used scores N7 to identify clinically significantinsomnia (Galasch & Gradisar, 2007; Nofzinger etal., 2004). Because the PSQI assesses sleep qualityas opposed to pure insomnia, increases in scores arenot necessarily a reflection of increased insomnia

severity. Thus, in analyzing the data we separatedindividuals into two groups. Individuals with PSQIscores b5 were categorized as good sleepers, and inattempt to approximate insomnia, those with scoresN7 were classified as poor sleepers. These cutoffswere arbitrarily generated. It stands to reason thatpeople who score on the boundary of poor andnormal sleep may share characteristics more incommon with subjects in the opposing group thanthe majority of members in their own group. Inorder to more sharply illuminate differencesbetween groups, it makes sense to eliminate border-line subjects who do not unambiguously representthe groups of interest.

General information. An investigator-designedquestionnaire constructed specifically for this inves-tigation was used to collect demographics andhealth-related information. Participants reportedtheir ethnicity, age, education level, gender, heightand weight. Height and weight were used tocompute body mass index. Information related toother sleep disorders such as sleep apnea, restlesslegs syndrome, and periodic leg movement disorderwere also collected. For these items respondentsanswered yes, no, or don't know to symptomsindicative of these conditions. Are you a heavysnorer? and Are you generally sleepy during the day?are examples of these items. Volunteers also reportedtheir overall physical health, and frequency andseverity of anxiety and depression experienced in thepast month. Physical health was measured using aone-item ordinal scale across 5 points, ranging fromexcellent to poor. The frequency of anxiety anddepression were each assessed using a one-itemordinal scale across 4 positions, ranging from rarelyor none of the time to most or all of the time. Theseverity of anxiety and depression were alsomeasured on a 4-point ordinal scale ranging fromnot at all to severe. These measures are indices ofglobal anxiety and depression and were used toindicate a general level of arousal or distress.

Resultsdemographics of good and poor sleepersand sample health characteristics

The final sample (N=220) included 128 (58.2%)good sleepers and 92 (41.8%) poor sleepers. Goodsleepers reported an average sleep onset latency of13.0 minutes (SD=9.7), a mean total sleep time of7.5 hours (SD=0.91), and an average global PSQIscore of 3.0 (SD=1.0). Among good sleepers, 28(21.9%) had trouble sleeping because they woke upduring the middle of the night or early morning 3 ormore times per week. Poor sleepers noted an averagesleep onset latency of 38 minutes (SD=26.6), a



mean total sleep time of 5.7 hours (SD=1.4), and anaverage global PSQI score of 10.2 (SD=2.3).Among poor sleepers, 58 (63%) had troublesleeping because they woke up during the middleof the night or early morning 3 or more times perweek.Sleep status (good or poor sleepers) was not

correlated with gender or age, and poor sleeperswere more likely to have a lower education level(ϕ=.16, pb .05). Poor sleepers were more likely toreport increased anxiety frequency (ϕ=.47, pb .01),depression frequency (ϕ=.35, pb .01), and worsephysical health (ϕ=.43, pb .01). Physical health,depression, and anxiety levels by good and poorsleepers are described in Table 2. We used one-itemscales to measure both the severity and frequency ofdepression. These items were highly correlated, andonly the frequency indicators are reported.

sleep hygiene frequencies

Mean totals for good and poor sleepers on theindividual SH behaviors are listed in Table 4. Poorsleepers engaged in only two behaviors more thanthree times per week. Poor sleepers drank caffeinebetween 5 and 10 hours before bedtime 3.42(SD=2.90) days per week and worried, planned,or had important thoughts in bed 3.63 (SD=2.51)days per week. Good sleepers did not engage in anypoor SH practice greater than 3 times per week.

comparing sh frequencies by sleep status

One-way multivariate analysis of variance (MAN-OVA) was used to evaluate whether individual SHbehaviors differ by sleep status (good and poorsleepers). Five one-way MANOVAs were con-ducted. In each MANOVA, the individual itemmeans of the variables from each SH category (i.e.,

Items 1–3, 4–7, etc.) were entered as the dependentvariables, and sleep status was entered as theindependent variable. Significant findings from theMANOVAs were followed up with univariateANOVAs. By using 5 MANOVA analyses insteadof 19 comparisons, we were able to minimize Type Ierror.MANOVA revealed significant differences

between good and poor sleepers in regards toimproper sleep scheduling behaviors (Items 1–3).All MANOVA results are included in Table 3, andunivariate results are presented in Table 4. Follow-up univariate ANOVAs showed that poor sleeperswere more likely to nap during the day. MANOVAshowed that poor sleepers were more likely toengage in activating or arousing activities nearbedtime (Items 8–11). Univariate ANOVAs foundthat poor sleepers were more likely to worry, plan,or think about important matters at bedtime,engage in exciting or emotional activities nearbedtime, and engage in activities of high concentra-tion near bedtime. A third MANOVA showed thatbed activities other than sleep (Items 12–15) weregreater among poor sleepers. Univariate resultsdemonstrated that poor sleepers were more likely toworry, plan, or think about important matters inbed. Finally, MANOVA revealed that environmen-tal conditions (Items 16–19) were worse amongpoor sleepers. Univariate ANOVAs showed thatpoor sleepers slept in a noisy environment and sleptin an uncomfortable temperature.

cognitive activityatbedtime andpoor sleepcontrolling for anxiety and depression

These data suggest that excessive cognitive activityin the bed is frequent and associated with poorsleep; however, this relationship may be explained

Table 2Frequencies of health characteristics of good and poor sleepers

Health measure Good sleepers Poor sleepers Total

Physical healthExcellent 34 (87.2%) 5 (12.8%) 39 (17.7%)Very good 58 (68.2%) 27 (31.8%) 85 (38.6%)Good 32 (45.1%) 39 (54.9%) 71 (32.3%)Fair 2 (11.8%) 15 (88.2%) 17 (7.7%)Poor 1 (14.3%) 6 (85.7%) 7 (3.2%)

Anxiety frequencyRarely or none 69 (79.3%) 18 (20.7%) 87 (39.5%)Some or little 43 (64.2%) 24 (35.8%) 67 (30.5%)Occasional or moderate 13 (28.9%) 32 (71.1%) 45 (20.5%)Most or all 2 (11.1%) 16 (88.9%) 18 (8.2%)

Depression frequencyRarely or none 72 (72.0%) 28 (28.0%) 100 (45.5%)Some or little 45 (57.7%) 33 (42.3%) 78 (35.5%)Occasional or moderate 10 (32.3%) 21 (67.7%) 31 (14.1%)Most or all 1 (9.1%) 10 (90.9%) 11 (5.0%)



by anxiety and depression rather than the practiceof poor SH. Two hierarchal logistic regressionswere conducted to assess whether this SH beha-vior was associated with poor sleep after control-ling for the frequency and severity of anxiety anddepression. In each analysis, sleep status wasentered as the dependent variable. The firstanalysis included the frequency of anxiety anddepression on step one and the SH variable onstep two. The second analysis included the severityof anxiety and depression on step one and the SHvariable on step two. Excessive cognitive activityat bedtime remained significantly associated withpoor sleep after accounting for the severity of

anxiety and depression (Odds Ratio = 1.42,pb .001; 95% confidence interval=1.21–1.66),and the frequency of anxiety and depression(Odds Ratio=1.31, pb .01; 95% confidence inter-val=1.12–1.53).

DiscussionThe present study aimed to explore the frequency ofSH behaviors among good and poor sleepers toassess the potential impact of these variables in thecommunity. Overall, the results indicate that sleephygiene practices for both good and poor sleepersare generally good. This finding is consistent withprevious studies suggesting that SH is not a primarycause of insomnia.Various individual SH behaviors are increased

in poor sleepers, and these variables may representrisk factors for poor sleep in the community. Themost robust finding among individuals’ SHpractices points to increased amounts of worry,planning, and thinking about important matters inbed among poor sleepers. This result is notsurprising, as excessive cognitive activity is astrong correlate of insomnia (Harvey, 2000b).Increases in this activity among poor sleepersexists even after controlling for global indices ofanxiety and depression, suggesting that that thelink between cognitive activity and insomnia may

Table 3Sleep hygiene categories by sleep status

Sleep hygiene category MANOVA results

Improper sleepscheduling behaviors

Wilks′ Λ=.963, F(3, 213)=2.71⁎

Sleep-disruptingproducts

Wilks′ Λ=.977, F(4, 212)=1.26

Activating or arousingactivities near bedtime

Wilks′ Λ=.803, F(4, 214)=13.12⁎⁎⁎

Bed activities otherthan sleep

Wilks′ Λ=.790, F(4, 214)=14.26⁎⁎⁎

Environmentalconditions

Wilks′ Λ=.913, F(4, 215)=5.13⁎⁎

Note. ⁎=pb .05, ⁎⁎=pb .01, ⁎⁎⁎=pb .001.

Table 4Sleep behavior scores across sleep status

SH item Good sleepers Poor sleepers Results

Mean (SD) Mean (SD)

1. Napped during the day 1.19 (1.73) 1.72 (2.05) F(1, 215)=4.21⁎2. Woke up at approximately the same time 1.52 (1.65) 1.84 (1.82) F(1, 215)=1.773. Went to bed at approximately the same time 1.55 (1.58) 1.99 (1.83) F(1, 215)=3.524. Drank caffeinated beverages such as coffee or tea or soft drinksbetween 5 to 10 hours before bedtime.

2.72 (2.88) 3.42 (2.90)

5. Drank caffeinated beverages such as coffee or tea or soft drinkswithin 5 hours of bedtime.

1.94 (2.65) 2.75 (2.82)

6. Drank alcohol within three hours of bedtime. 1.10 (1.75) 1.19 (2.03)7. Smoked a cigarette or chewed tobacco within 2 hours of bedtime orin the middle of the night.

1.32 (2.62) 1.78 (2.99)

8. Engaged in exciting or emotionally upsetting activities near bedtime. 0.66 (1.24) 1.31 (1.66) F(1, 217)=11.05⁎⁎9. Performed activities demanding high levels of concentration near bedtime. 0.56 (1.35) 1.07 (1.58) F(1, 217)=6.62⁎10. Exercised within 4 hours of bedtime 0.56 (1.25) 0.67 (1.39) F(1, 217)=0.3511. Worried, planned, or thought about important matters at bedtime. 1.48 (1.72) 3.53 (2.49) F(1, 217)=51.97⁎⁎⁎12. Read in bed 1.91 (2.60) 1.63 (2.27) F(1, 217)=0.6513. Watched television in bed 2.35 (2.86) 2.89 (3.07) F(1, 217)=1.7814. Lounged around in bed 1.42 (2.23) 1.91 (2.51) F(1, 217)=2.2915. Worried, planned, or thought about important matters in bed. 1.56 (1.71) 3.63 (2.51) F(1, 217)=52.60⁎⁎⁎16. Slept on an uncomfortable mattress 0.52 (1.71) .78 (1.90) F(1, 218)=1.1617. Slept in a room with an uncomfortable nighttime temperature. 0.64 (1.41) 1.78 (2.44) F(1, 218)=18.86⁎⁎⁎18. Slept in a noisy environment 0.34 (1.20) 0.84 (1.88) F(1, 218)=5.69⁎19. Slept in a room that was too bright 0.11 (0.77) 0.26 (1.13) F(1, 218)=1.41

Note. ⁎=pb .05, ⁎⁎=pb .01, ⁎⁎⁎=pb .001.



be partly due to the SH behavior and not justpsychological distress or trait arousal.Other SH variables are also increased in poor

sleepers, yet these practices are relatively low infrequency, and the overall impact of these char-acteristics in the community are likely to be small.Nevertheless, the select few with these character-istics may be at risk for poor sleep. For example,poor sleepers more often sleep in a noisy environ-ment and in an uncomfortable nighttime tempera-ture. These data are consistent with one study thatshowed increased rates of insomnia in womenliving in areas of high nighttime traffic volume(Kageyama et al., 1997). To our knowledge, noother studies have assessed the relationship betweenuncomfortable nighttime temperatures and sleepquality. These data were accumulated during themonth of August, when temperatures are generallyhigh, and it is possible that individuals lacking airconditioning are negatively affected. Poor sleepersare also more likely to engage in behaviors that mayincrease arousal at bedtime, such as activitiesrequiring increased concentration, and activitiesthat are emotional or exciting. Future studiesshould assess the importance of “winding down”before attempting sleep.Poor sleepers also show a small but statistically

significant increase in the frequency of napping.This finding is consistent with Jefferson et al.(2005), yet not in line with the results from Cheeket al. (2004). The low effect size obtained in thisstudy, in addition to previous mixed findingscomparing napping among good and poor sleepers,calls into question the impact of napping on sleepquality in the community.There are no differences in the usage of sleep-

disrupting products between good and poor slee-pers. These findings are consistent with Cheek et al.(2004) and show that, after excluding for othersleep disorders, there is not a positive relationshipbetween these variables and sleep quality. Thesedata conflict with Jefferson et al. (2005), and it ispossible that sampling and measurement differ-ences between the two studies contribute to thediscrepancy in findings. Our data include a greaterproportion of female participants who are lessprone to excessive drinking (Grant et al., 2004),and thus, less likely to self-medicate with alcohol,become reliant on alcohol for sleep, or experiencesleeping difficulties due to consistent drinking.Consistent with this hypothesis, Johnson, Roehrs,Roth, and Breslau (1998) showed that poor sleeperswere more likely to use alcohol as a sleep aid, andtheir sample included gender proportions morecomparable to the national average. The SHinstrument used in Jefferson et al. also detected

the overall severity of alcohol usage by assessing thenumber of alcoholic beverages per week, whereasour study assessed the number of days per week inwhich the participant drank alcohol within 3 hoursof bedtime. It is possible that the extent of alcoholuse is more important than the mere existence ofdrinking in contributing to poor sleep. Jeffersonet al. also measured smoking as a dichotomousvariable and asked participants whether they areregular smokers, or smoked within 5 minutes ofbedtime. It is possible that cigarettes are onlydisruptive when taken immediately before bedtime,or taken regularly throughout the day. Even thoughcaffeine usage was somewhat high, its intake is notpositively correlated with poor sleep, and this isconsistent with Jefferson et al., Cheek et al., andMcCrae et al. (2006). Thus, it appears that a goodpercentage of individuals who use caffeine are notengaging in sleep-disrupting behaviors.These results need to be interpreted within the

confines of several limitations. First, the return ratewas low, which may affect our ability to generalizefrom the sample. Indeed, this study overrepresentedwomen, college graduates, and Caucasians, andunderrepresented African Americans and indivi-duals of Latin American descent. Second, attributesinherent to the online sample may also affect theability to generalize from these findings. Study-Response panelists tend to have greater access toand knowledge of information technology, morefree time, and an interest in browsing for recrea-tional purposes (Stanton, 2006). Thus, the volun-teers in this sample may not be representative of thepopulation. Third, the percentage of poor sleepersin this sample is also high as compared to otherstudies. It is possible that individuals volunteeringfor Internet-based studies show worse sleep ascompared to the population, particularly amongmales. As future studies will no doubt be utilizingthe Internet to assess community attributes, futureresearch should continue to assess the validity ofusing the Internet sample to accurately assess sleepcharacteristics in the population. Fourth, this studyrelied on a self-report measure to assess SH, andparticipants may not have accurately identifiedtheir behaviors. Finally, these self-report datacannot address the possibility that poor sleepersdiffer from good sleepers because they are morevulnerable to the effects of SH and not necessarilybecause of an increased frequency of behaviors. Forexample, poor sleepers may not necessarily drinkmore coffee but they may be more susceptible to theeffects of coffee. Indeed, it is possible that certainpeople with insomnia may be likely to practice goodSH behaviors due to the knowledge that practicinggood SHmay limit disturbed sleep. Adherence to SH



recommendations would also appear to be influ-enced by access to such information and motivationto change. Factors likely to contribute to the practiceof SH should be considered in future research.These data do tell us whether individuals are

likely to practice a certain behavior and whether itis associated with poor sleep and provide anestimate of the potential impact these behaviorsmay have in the community. Activities that increasearousal at bedtime, particularly increased cognitiveactivity in the bed, poor temperature control, andincreased noise at bedtime are associated with poorsleep. Future studies should further investigate theprevalence of these variables and their possibleimpact on poor sleep.

ReferencesAmerican Academy of Sleep Medicine. (2005). International

classification of sleep disorders: Diagnostic and codingmanual, 2nd ed. Westchester, IL: Author.

Ancoli-Israel, S., & Roth, T. (1999). Characteristics of insomniain the United States: Results of the 1991 National SleepFoundations Survey. I. Sleep, 22, 347–353.

Backhaus, J., Junghaans, K., Broocks, A., Riemann, D., &Hohagen, F. (2002). Test-retest reliability and validity of thePittsburgh Sleep Quality Index. Journal of PsychosomaticResearch, 53, 737–740.

Brown, F. C., & BuboltzJr., W. C. (2002). Relationship of sleephygiene awareness, sleep hygiene practices, and sleepquality. Behavioral Medicine, 28, 33–39.

Buysse, D. J., Reynolds, C. F., Monk, T. H., Berman, S. R., &Kupfer, D. J. (1989). The Pittsburgh Sleep Quality Index: Anew instrument for psychiatric practice and research.Psychiatry Research, 28, 193–213.

Buysse, D. J., Reynolds, C. F., Kupfer, D. J., Thorpy, M. J.,Bixler, E., Manfredi, R., et al. (1994). Clinical diagnoses in216 insomnia patients using the International Classificationof Sleep Disorders, DSM-IV and ICD-10 categories: A reportfrom the APA/NIMHDSM-IV field trial. Sleep, 17, 630–637.

Cheek, R. E., Shaver, J. L. F., & Lentz, M. J. (2004). Variationsin sleep hygiene practices of women with and withoutinsomnia. Research in Nursing and Health, 27, 225–236.

Eysenbach, G., & Wyatt, J. (2002, November 22). Using theinternet for surveys and health research. Journal of MedicalInternet Research, 4, e13 Retrieved March 28, 2007, fromhttp://www.jmir.org/2002/2/e13.

Ford, D. E., & Kamerow, K. B. (1989). Epidemiologic study ofsleep disturbances and psychiatric disorders. Journal of theAmerican Medical Association, 262, 1479–1484.

Gallasch, J., & Gradisar, M. (2007). The relationships betweensleep knowledge, sleep practice, and sleep quality. Sleep andBiological Rhythms, 5, 63–73.

Gosling, S. D., Vazire, S., Srivastava, S., & John, O. P. (2004).Should we trust web-based studies? A comparative analysisof 6 preconceptions about internet questionnaires.AmericanPsychologist, 59, 93–104.

Grant, B. F., Dawson, D. A., Stinson, F. S., Chou, S. P., Dofour,M. C., & Pickering, R. P. (2004). The 12-month prevalenceand trends in DSM-IValcohol abuse and dependence: UnitedStates, 1991–1992 and 2001–2002. Drug and AlcoholDependence, 74, 223–234.

Hall, M., Baum, A., Buysse, D. J., Prigerson, H. G., Kupfer,D. J., & Reynolds, C. F. (1998). Sleep as a mediator of the

stress-immune relationship. Psychosomatic Medicine, 60,48–51.

Harvey, A. (2000a). Sleep hygiene and sleep-onset insomnia.Journal of Nervous Mental Disorders, 188, 53–55.

Harvey, A. (2000b). Pre-sleep cognitive activity: A comparisonof sleep-onset insomniacs and good sleepers. Brittish Journalof Clinical Psychology, 39, 275–286.

Haynes, S. N., Follingstad, D. R., & McGowan, W. T. (1974).Insomnia: Sleep patterns and anxiety level. Journal ofPsychosomatic Research, 18, 69–74.

Jefferson, B. S., Drake, C. L., Scofield, H. M., Myers, E.,McClure, T., Roehrs, T., & Roth, T. (2005). Sleep hygienepractices in a population-based sample of insomniacs. Sleep,28, 611–615.

Johnson, E. O., Roehrs, T., Roth, T., & Breslau, N. (1998).Epidemiology of alcohol and medication as aids to sleep inearly adulthood. Sleep, 21, 178–186.

Kageyama, T., Kabuto, M., Nitta, H., Kurokawa, Y., Taira, K.,Suzuki, S., et al. (1997). A population study on risk factorsfor insomnia among adult Japanese women: A possibleeffect of road traffic volume. Sleep, 20, 963–971.

Kazarian, S. S., Howe, M. G., Merskey, H., & Deinum, E. L.(1978). Insomnia: Anxiety, sleep-incompatible behaviorsand depression. Journal of Clinical Psychology, 34,865–869.

Lacks, P., & Rotert, M. (1986). Knowledge and practice ofgood sleep hygiene techniques in insomniacs and goodsleepers. Behaviour Research and Therapy, 24, 365–368.

Lichstein, K. L., Durrence, H. H., Riedel, B. W., Taylor, D. J., &Bush, A. J. (2004). Epidemiology of sleep: Age, gender, andethnicity. Mahwah, NJ: Erlbaum.

Mastin, D. F., Bryson, J., & Corwyn, R. (2006). Assessing sleephygiene using the sleep hygiene index. Journal of BehavioralMedicine, 29, 223–227.

McCrae, C. S., Row, M. A., Dautovich, N. D., Lichstein, K. L.,Durrence, H. H., Riedel, B. W., et al. (2006). Sleep hygienepractices in two community dwelling samples of olderadults. Sleep, 29, 1551–1560.

Nofzinger, E. A., Buysse, D. J., Germain, A., Price, J. C.,Miewald, J. M., & Kupfer, D. J. (2004). Functionalneuroimaging evidence for hyperarousal in insomnia.American Journal of Psychiatry, 161, 2126–2129.

Riedel, B. W. (2000). Sleep hygiene. In K. L. Lichstein, & C. M.Morin (Eds.), Treatment of late-life insomnia (pp. 125–146).Thousand Oaks, CA: Sage.

Riedel, B. W., & Lichstein, K. L. (2000). Insomnia and daytimefunctioning. Sleep Medicine Reviews, 4, 277–298.

Roehrs, T., & Roth, T. (1997). Hypnotics, alcohol, and caffeine:Relation to insomnia. In M. R. Pressman, & W. C. Orr(Eds.), Understanding sleep: The evaluation and treatmentof sleep disorders (pp. 339–355). Washington DC: AmericanPsychological Association.

Savard, J., Laroche, L., Simard, S., Ivers, H., & Morin, C. M.(2003). Chronic insomnia and immune functioning.Psychosomatic Medicine, 65, 211–221.

Simon, G. E., & VonKorff, M. (1997). Prevalence, burden, andtreatment of insomnia in primary care. American Journal ofPsychiatry, 154, 1417–1423.

Stanton, J. M., & Weiss, E. M. (2002). Online panels for socialscience research: An introduction to the StudyResponseProject (Tech. Rep. No. 13001). Syracuse, NY: SyracuseUniversity, School of Information Studies.

Stanton, J. M. (2006). The StudyResponse open recruitmentmethod (Tech. Rep. No. 13007). Syracuse, NY: SyracuseUniversity, School of Information Studies.

Stepanski, E. J., &Wyatt, J. K. (2003). Use of sleep hygiene in thetreatment of insomnia. Sleep Medicine Reviews, 7, 215–225.



Taylor, D. J., Lichstein, K. L.,&Durrence,H.H. (2003). Insomniaas a health risk factor. Behavioral Sleep Medicine, 1, 227–247.

Tsai, P. S., Wang, S. Y., Wang, M. Y., Su, C. T., Yang, T. T.,Huang, C. J., et al. (2005). Psychometric Evaluation of theChinese version of the Pittsburgh Sleep Quality Index(CPSQI) in primary insomnia and control subjects. Qualityof Life Research, 14, 1943–1952.

Youngstedt, S. D., O'Conner, P. J., & Dishman, R. K. (1997).The effects of acute exercise on sleep: A quantitativesynthesis. Sleep, 20, 203–214.

RECEIVED: August 6, 2007ACCEPTED: February 15, 2008Available online 14 July 2008


Sleep Hygiene Practices of Good and Poor Sleepers in the United States: An Internet-Based Study

Documents