No , 2 March Prevalence of Sleep Disturbance and Hypnotic Medication Use in Relation to Sociodemographic Factors in the General Japanese Adult Population

This study was the first nationwide population-based study to estimate the prevalence rates of sleep disturbance and hypnotic medication use in the general Japanese adult population. In 1997, 2,800 Japanese adults aged 20 years and over were randomly selected from the 1995 Census and 1,871 were examined using the Pittsburgh Sleep Quality Index. The respective estimated overall prevalences of insomnia (INS), difficulty initiating sleep (DIS), difficulty maintaining sleep (DMS), poor perceived quality of sleep (PQS) and hypnotic medication use (HMU) were 17.3%, 8.6%,12.9%,17.8%, and 3.5% in males and 21.5%,12.6%,16.2%, 20.2% and 5.4% in females. Among males, DIS (OR=2.76) and PQS (OR=2.12) were associated with never having married. DMS was associated with being 60 years and older (OR=2.68) or divorced/separated (OR=3.74). Among females, DMS was associated with being widowed (OR=1.65), unemployed (OR=1.60), 40 to 59 years old (OR=0.57) or never having married (OR=0.39). DIS was associated with being widowed (OR=1.67) or unemployed (OR=1.58). HMU was associated with advancing age (OR=8.26-10.7), being widowed (OR=2.12) or never having married (OR=2.84). PQS was associated with advancing age (OR=0.63-0.50). Our study showed sleep disturbance and hypnotic medication use were prevalent among Japanese adults and some sociodemographic factors contributed to them. J Epidemiol, 2000 ; 10 : 79-86

(HMU) were 17.3%, 8.6%,12.9%,17.8%,and 3.5% in males and 21.5%,12.6%,16.2%,20.2% and 5.4% in females.Among males, DIS (OR=2.76)and PQS (OR=2.12)were associated with never having married.DMS was associated with being 60 years and older (OR=2.68)or divorced/separated (OR=3.74).Among females, DMS was associated with being widowed (OR=1.65),unemployed (OR=1.60),40 to 59 years old (OR=0.57)or never having married (OR=0.39).DIS was associated with being widowed (OR=1.67)or unemployed (OR=1.58).HMU was associated with advancing age (OR=8.26-10.7),being widowed (OR=2.12)or never having married (OR=2.84).PQS was associated with advancing age (OR=0.63-0.50).Our study showed sleep disturbance and hypnotic medication use were prevalent among Japanese adults and some sociodemographic factors contributed to them.J Epidemiol, 2000 ; 10 : 79-86 sleep disturbance, hypnotic medication use, prevalence, sociodemographic factor, Pittsburgh Sleep Quality Index (PSQI) Previous studies have suggested that sleep disturbance is one of the most pervasive health problems in both the general population and general practice 1-24.In addition, sleep deprivation has been noted to place a tremendous burden on our society and life due to its associations with high mortality and co-morbidity, reduced productivity, increased absenteeism and accidents as well as deterioration of personal and professional relationships 10.11,12,23,24.25).
However, with the exception of a few studies that focused on specific groups or regions2-6), little is known about the epidemiology of sleep disturbance and hypnotic medication use in the general adult population of Japan.We collected data from a representative sample (subjects aged 20 years and older) of the entire Japanese adult population and estimated the prevalence rates of sleep disturbance and hypnotic medication use.We also examined age trends, gender differences and associations with sociodemographic factors of sleep disturbance and hypnotic medication use.

SUBJECTS AND METHODS
The subjects comprised Japanese adults aged 20 years and over who were living in Japan.The sample size was 2,800 (200 subjects of each gender in each of seven 10-year age groups).
First, all 3,371 regions in the country were listed in order after being weighted according to their population sizes in the 1995 Census 26) and then 100 of these regions were randomly selected from this list.Next, each of 28 subjects in 100 regions was randomly selected from 28 blocks evenly divided in each residential registry, by which subjects were not accumulated in a certain geographical part of each region.This sampling method was used to select all 2,800 subjects for this study.
A questionnaire about sleep and its related variables was mailed to every subject at the end of September 1997.Two reminders were sent at intervals of a few weeks to subjects who had not replied within two weeks.
The self-administered questionnaire used for this study consisted of the Japanese version of the Pittsburgh Sleep Quality Index (PSQI)27,28) and the General Health Questionnaire (GHQ12) 29), together with questions about sociodemographic variables such as age, gender, marital status, educational status, job status and other putative variables associated with sleep habits and sleep problems.We developed the Japanese version of the PSQI after obtaining the permission of the original authors to use this instrument 28).
The PSQI assesses sleep duration, sleep latency and the frequency and severity of specific sleep-related problems during the previous month 27).Four items from the PSQI were used to estimate prevalence rates: 1) how often have you had trouble sleeping because you can not get to sleep within 30 minutes? (DIS, difficulty initiating sleep); 2) how often have you had trouble sleeping because you wake up in the middle of the night or early morning?(DMS, difficulty maintaining sleep); 3) how would you rate your sleep quality overall?(PQS, poor perceived quality of sleep) and 4) how often have you taken medicine to help you sleep?(HMU, hypnotic medication use).The subjects were asked to choose one response from the following four answers on a Likert scale: not at all in one week, less than once a week, once or twice a week and three or more times a week to questions 1), 2) and 4) and from very good, fairly good, fairly poor and very poor to question 3).Those who replied three or more times a week to questions 1), 2) and 4) and very poor or fairly poor to question 3) were categorized as DIS, DMS, HMU and PQS cases, respectively.Insomnia (INS) was defined as DIS and/or DMS occurring three or more times a week in the past month.
The prevalence rates of INS, DIS, DMS, PQS and HMU were calculated by gender and 10-year age groups .In addition, the overall prevalence rates obtained were weighted by a percentage according to the age-group distribution determined by the 1995 Census and 95% confidence intervals were calculated 30).
The Cochran-Armitage Method was used to detect any age trends for INS, DIS, DMS, PQS and HMU by gender 30Wender differences in the prevalence rates of INS , DIS, DMS, PQS and HMU were compared using a chi-square test (significance was set at a level of P<0.05).Differences in gender and age between responders and non-responders were also examined using a chi-square test (the significance was P<0 .05).
The overall prevalence rates of INS, DIS, DMS, PQS and HMU were computed for each gender, after being weighted according to the proportion of each age group in the total adult population determined by the 1995 Census 26).
For each gender, logistic regression analysis 31) was used to determine the odds ratios (ORs) of sociodemographic factors in relation to each type of sleep disturbance and to hypnotic medication use.The seven 10-year age groups were regrouped into three: younger (20-39 years), middle-aged (40-59 years) and older (60 years and older) and job status was categorized as employed and unemployed.Goodness of fit was assessed using the test of Hosmer-Lemeshow 32).Effect modification was tested by including interaction terms.
The data were analyzed using version 8.0 of the Statistical Package of the Social Sciences software33).

RESULTS
Responses were returned by 1,889 (67.5%) subjects, 18 of whom were excluded because of missing data about gender and/or age or extreme outliers of sleep latency and/or sleep duration.The data for 1,871 (66.8%) of the subjects (920 males (49.2%) and 951 females (50.8%)) were analyzed.No gender or age differences between responders and non-responders were found.The sociodemographic characteristics of the subjects who participated in this study are shown in Tablel.Unmarried individuals and unemployed males were the least likely to respond to our survey.
The Cochran-Armitage method revealed significant positive age trends for INS, DMS and HMU in both males and females and for DIS only in females (P<0.05).The PQS prevalence rate in males decreased with increasing age (P<0.05),whereas no significant age trend for this complaint in females was demonstrated.
Chi-square tests were performed to evaluate gender differences.Females in the 20-29 year (DMS P<0.05), 50-59 year (DIS P<0.05, HMU P<0.01), 70-79 year (PQS P<0.05) and 80 years and older (DIS P<0.01, PQS P<0.05, HMU P<0.01) age groups were more likely to suffer from sleep disturbance and use hypnotic medication than males in the corresponding groups.No male preponderance of any type of sleep disturbance or hypnotic medication use was identified.
An inverse relationship between PQS and the age of 60 years and over was identified (OR=0.50,95% CI: 0.29-0.86).HMU correlated significantly with advancing age (OR=8.26,95% CI: 2.15-31.7 for the 40-59 year-old group and OR=10.7,95% CI: 2.64-42.9for those aged 60 years and over), being widowed (OR=2.12,95% CI: 1.14-3.95)and never having married (OR= 2.84, 95% CI: and estimated the prevalence rates of DIS and DMS by each group of poor sleepers and good sleepers22).Their results of DIS and DMS are, however, excluded from further discussion because of difference in denominators used for their study and ours.
The differences and similarities of the findings across crosssectional studies are probably due to methodological issues such as the choice of study subjects, sample sizes, methods of sampling, case ascertainment and methods of data collection (self-administered questionnaires or interviews).The similar results obtained by Karacan et al. 9) and us may be attributable to the similar survey methods used for sampling, data collection and case ascertainment, whereas possible explanations for the slightly higher prevalences in the studies of Bixler et al. 8), Klink et al. 11) and Ohayon et al.14) may that their criteria for sleep disturbance were more lenient than those we used.However, in general, we conclude there is no major disagreement between our results and those of previous studies with respect to the prevalence of sleep disturbance in the general adult population.
Most previous studies have shown that females are more susceptible to sleep disturbance than males [8][9][10][11]13,16,18,20,21).We found that females in certain age groups were more likely to suffer from some types of sleep disturbance and to use hypnotic medication than others and no male preponderance of any type of sleep disturbance or hypnotic medication use was identified.
In the previous studies, statistical analyses of age trends for in the French population aged 15 years and over 13).His results were similar to our PQS prevalence rates of 17.8% in males and 20.2% in females, although the definitions of UQS and PQS are not quite the same.In our study, we identified a negative age trend for PQS in males and multivariate analyses revealed that being single, but not age, is associated with PQS.Multivariate analyses indicated that middle-aged and older females were less likely than younger females to perceive their quality of sleep as poor, suggesting that the perception of the quality of sleep is related to age to some extent, that is, older persons of both genders are less likely to perceive their sleep as poor.As Ohayon et al. pointed out 14), the lifestyles of single males, such as an irregular sleep-wake schedule, may partly contribute to the poor quality of their sleep.Further studies are needed to investigate whether the natures of insomnia with and without PQS differ and whether poor quality and/or dissatisfaction with sleep could be a better indicator of sleep pathology.
The multivariate analyses we used revealed that that sociodemographic factors affected insomnia (DIS and DMS) and hypnotic medication use quite differently by gender .For males, being 60 years of age and older and being divorced/separated correlated positively with DMS.For females , never hav- INS: Insomnia, defined as DIS and/or DMS DIS: difficulty initiating sleep within 30 minutes three or more times a week in the last month DMS: difficulty maintaining sleep three or more times a week in the last month PSQ: poor perceived quality of sleep in the last month HMU: hypnotic medication use Uni and multi indicate univariate analyses and multivariate analyses , respectively.ing married and being 40 to 59 years old were negatively related to DMS, whereas being widowed was positively associated with DMS.Therefore, we speculate that the considerable contribution of age to DMS in males reflects pathophysiological factors such as medical conditions and the aging process as well as psychosocial factors 19).As demonstrated by other studies 2, 3,9,11,12,13,34,35) our study showed a strong association between increasing age and HMU and a modest association of DIS with being widowed and being unemployed in females .
The following limitations of our study must be taken into consideration.We could not differentiate acute/transient versus chronic/stable sleep disturbances in our study, because we used a cross-sectional design and the time frame applied was just one month prior to the study.Furthermore, the case ascertainment of sleep disturbance was based solely on self-reported symptoms.We did not use the criteria described in the ICSD-90 36) and DMS-IV 37) or polysomnographic recordings to diagnose specifically sleep disorders, because such clinical diag-noses could not be incorporated into our community-based epidemiological study.
In conclusion, our nationwide population study in Japan, the first of its kind, has provided evidence that sleep disturbance is prevalent in the general Japanese adult population and that sociodemographic factors have various forms of association with certain types of sleep disturbance and hypnotic mediation use.

Table 3 and
4. As shown in Table3, multivariate analyses revealed that, in males, DIS and PQS were significantly associ-Adults

Table 2 .
Prevalence rates (%) of sleep disturbance and hypnotic medication use by gender and age groups in 1,871 adults aged 20 years and older, Japan, 1997.4% in the females.Several epidemiological studies outside Japan have reported prevalence rates of sleep disturbance and hypnotic medication use in the general population.On a sample of a population aged 18 years and older in the Houston Standard Metropolitan Area, Karacan et al. found that the prevalence rates of DIS, DMS and HMU were 6.0%, 12.9% and 3.3% in males, and 11.2%, 17.4% and 4.0% in females, respectively 9).Accordingly toBixler etal.8), the respective prevalence rates of INS, DIS and DMS were 32.2%, 14.4% and 22.9% in a population aged 18 years and older of Los Angels.Klink et al. also reported 33.9% for INS in an adult sample aged 18 years and older of Tucson (28.2% in males and 38.2% in females) 11.Ohayon et al. found respective male and female prevalence rates for DIS and DMS were 10.5% and 16.4%, 15.3% and 24.8% in a representative sample of French subjects of 15 years of age or more 14).Lugaresi et al. reported 19.1% of poor sleepers (14.7% males, 23.1% females) in a population over 20 years of age in the Republic of San Marino 1.02-7.86).INS: Insomnia, defined as DIS and/or DMS DIS: difficulty initiating sleep within 30 minutes three or more times a week in the last month DMS: difficulty maintaining sleep three or more times a week in the last month PSQ: poor perceived quality of sleep in the last month HMU: hypnotic medication use The figures in parentheses are 95 percent confidence intervals.

Table 3 .
Associations of sleep disturbance and hypnotic medication use with sociodemographic variables in males aged 20 years and older, Japan, 1997.INS: Insomnia, defined as DIS and/or DMS DIS: difficulty initiating sleep within 30 minutes three or more times a week in the last month DMS: difficulty maintaining sleep three or more times a week in the last month PSQ: poor perceived quality of sleep in the last month HMU: hypnotic medication use Uni and multi indicate univariate analyses and multivariate analyses, respectively.

Table 4 .
Associations of sleep disturbance and hypnotic medication use with sociodemographic variables in females aged 20 years and older, Japan, 1997.