A Population-Based Study on Nighttime Road Traffic Noise and Insomnia

Transcription

1 pii: zsw055 ORIGINAL ARTICLE A Population-Based Study on Nighttime Road Traffic Noise and Insomnia Jorunn Evandt, MSc 1 ; Bente Oftedal, PhD 1 ; Norun Hjertager Krog, PhD 1 ; Per Nafstad, MD, PhD 2,3 ; Per E. Schwarze, PhD 1 ; Gunn Marit Aasvang, PhD 1 1 Domain for Infection Control and Environmental Health, Department of Air Pollution and Noise, Norwegian Institute of Public Health, Oslo, Norway; 2 Department of Community Medicine, University of Oslo, Oslo, Norway; 3 Domain for Mental and Physical Health, Department of Noncommunicable Diseases, Norwegian Institute of Public Health, Oslo, Norway Study Objectives: The aims of the present study were to investigate how nighttime road traffic noise relates to self-reported symptoms of insomnia and sleep medication use. Methods: We used questionnaire data from the population-based study Health and Environment in Oslo (HELMILO) ( ; n = ). The insomnia symptoms difficulties falling asleep, awakenings during the night, and waking up too early in the morning as well as self-reported sleep medication use were included as outcomes. Modeled noise levels (L night ) were assigned to each participant s home address. For selecting covariates to the statistical model, we used a directed acyclic graph. The associations between noise and sleep were analyzed using logistic regression models. Results: After adjustment for potential confounders, we found an odds ratio (OR) of 1.05 (95% confidence interval [CI]: ) for the association between traffic noise and difficulties falling asleep, in the total study population. For the association between traffic noise and awakenings during the night, the OR was 1.04 (95% CI: ) and for waking up too early, the OR was 1.06 (95% CI: ). The effect estimates are given per 5-dB increase in traffic noise level (L night ). Self-reported sleep medication use was not statistically significantly associated with traffic noise exposure. Conclusions: In an adult population from Oslo, traffic noise was associated with difficulties falling asleep and waking up too early. These findings indicate that sleep quantity may be compromised for individuals living in areas highly exposed to nighttime traffic noise. Keywords: traffic noise, insomnia, sleep disturbance, sleep medication use, environmental epidemiology, population-based study, directed acyclic graph. Statement of Significance Nighttime traffic noise is of increasing public health concern because of its potential to compromise both quality and quantity of sleep in large parts of urban populations. Our large population-based study, investigating the association between traffic noise and insomnia symptoms and sleep medication use, provides new insight into how road traffic noise may interfere with the sleeping process. The study demonstrated exposure response relationships between nighttime road traffic noise and the symptoms of insomnia difficulties falling asleep and waking up too early. These findings indicate that individuals exposed to high levels of nighttime traffic noise might suffer from insufficient sleep. Further, this work could contribute to health impact assessments for nighttime road traffic noise in the future. INTRODUCTION According to the World Health Organization (WHO), nearly a million healthy life-years are lost each year in Western Europe because of sleep disturbances due to environmental noise exposure. 1 Noise is a factor that is present in the everyday life of people living in urban environments, and there is a decreasing availability of quiet places. 2 Furthermore, environmental noise has been recognized as a ubiquitous disruptor of sleep and is therefore of public health concern. 3 Previous studies on traffic noise and sleep have demonstrated physiological effects including sleep stage changes, delayed sleep onset, awakenings, increased body motility, and changes in heart rate. 4 9 Besides the direct impact on sleep, nighttime traffic noise has also been associated with day after effects such as mood changes, 10 decreased cognitive performance, 11 morning tiredness, 12 and fatigue. 13 Also, long-term health effects including obesity, 14,15 diabetes, 16 hypertension, and cardiovascular disease have been associated with traffic noise exposure. Independent of noise exposure, these health effects have also been associated with insufficient sleep, suggesting that the association between long-term traffic noise exposure and these health outcomes is mediated through disturbed sleep. In spite of the well-documented immediate effects of noise on sleep, there are few population-based studies on traffic noise and insomnia. Insomnia is defined as having difficulties initiating and maintaining sleep or waking up too early unable to go back to sleep; these symptoms should have occurred at least 3 times per week for 3 months, and negative daytime consequences should be present. 32 Furthermore, insomnia is the most common sleep disorder among adults. 33 A number of studies have investigated subjectively reported sleep disturbances due to noise, and exposure response curves have been presented in a study having pooled datasets from 24 studies. 34 In these studies, participants were asked how much a particular noise source disturbed their sleep. Thus, noise is suggested as the cause of the sleep disturbances, and the participants may therefore overattribute sleep disturbances to traffic noise. 35 By rather asking participants about sleep disturbances in general without any reference to noise, or symptoms of insomnia, this bias could be reduced. Population-based studies having used this approach have indicated an association between road traffic noise and sleep disturbances. 36,37 Sleep medication use can also be employed as an outcome reflecting sleeping problems. 38 Few studies have assessed the association between road traffic noise and sleep medication use, most of which found no association. 12,39,40 However, one study reported that road traffic noise could be associated with register-based sleep medication use. 41 In this large population-based study, we investigated the association between nighttime road traffic noise and symptoms of insomnia. In addition, we assessed the association between traffic noise and self-reported use of sleep medication. METHODS Study Population We used data from the Health and Environment in Oslo (HELMILO) study, conducted in In this study, a questionnaire was received by Oslo inhabitants born in 1

2 the years , , 1955, 1960, and The response rate was 48% (n = ). Using the national identification number, the study population was linked to data from Statistics Norway. A total of 881 participants were excluded because they had been living at their current address for less than 1 year, had moved out of Oslo before returning the questionnaire, or had not been assigned a noise exposure level due to missing or uncertain information about geographic coordinates of the residential address. Outcome Variables Symptoms of insomnia were indicated by the three types of sleep problems, difficulties falling asleep, awakenings during the night, and waking up too early, which are similar to the criteria of an insomnia diagnosis according to the Diagnostic and Statistical Manual of Mental Disorders (DSM). 32 The participants reported how frequently they had experienced these problems during the past 12 months according to the alternatives never/rarely, less than once per week, 1 2 times per week, 3 5 times per week, and always/almost every night. We considered a frequency of 3 to 5 times per week or more as presence of insomnia symptoms and dichotomized this 5-point scale into 3 5 times per week versus <3 5 times per week. The questionnaire included a single item on sleep medication use ( Do you use sleep medication? ). The participants reported whether they currently, previously, or never had used sleep medications. We dichotomized sleep medication use into current versus previous or never use. Noise Exposure Assessment Nighttime road traffic noise (L night, A-weighted nighttime equivalent noise level, hours) was modeled for the most exposed facade of the building and assigned to each participant s home address using geographic coordinates. The noise levels were calculated according to The Nordic prediction method for road traffic noise 42,43 using GIS methodology in the software package CadnaA. 44 The grids for the noise calculations were 5 5 m, and calculation height was 4 m above terrain. Within each grid, the noise level was interpolated at points along the facade with 3 m distance. Road traffic data included in the model (traffic counts, percentage heavy vehicles, speed limits, and diurnal distribution) were obtained from the Norwegian Public Roads Administration and the City of Oslo. Other inputs to the model were digitalized terrain data in 3D including topography, soft versus hard ground, location of buildings, and noise screens. The noise exposure assessment was based on the data used for strategic noise mapping for Oslo, conducted every 5 years in accordance with the European Environmental Noise Directive. 45 We had access to noise data calculated for the years 2006 and The noise mapping for was the one performed closest in time to the study period ( ), and therefore, most participants were assigned a noise level for However, in the time between the study period and 2011, some buildings and noise screens had either been erected or demolished, and the road network had been restructured in some areas. Thus, in some cases, the road traffic noise data assessed for 2006 were evaluated to be more representative than the 2011 data. We therefore assigned the noise 2 level calculated for 2006 to the dwellings where the noise level had either increased or decreased substantially (±3 db) following these local changes (2.2%). Covariates In the procedure of selecting covariates to the statistical models, we used a directed acyclic graph (DAG). 47,48 The DAG was developed using the Web-based software DAGitty 49 by including factors relevant for the association between traffic noise exposure and sleep disturbances, based on literature review of previous research. In Figure 1, we have presented a simplified version of the DAG. The variables identified by the DAG were age, sex, marital status, alcohol use, smoking, physical activity, and night-shift work, in which age and sex were specified as compulsory variables for the final adjustment set. Further, it has been found that socially deprived individuals are exposed to higher noise levels than more socially advantaged individuals. 50 Thus, we added the socioeconomic factors educational level and household income to the final adjustment set. Covariate data were obtained from both Statistics Norway and the HELMILO questionnaire. Information on age and sex was provided by Statistics Norway. For constructing the variable on marital status, we mainly used data from Statistics Norway but added information on cohabitation obtained from the questionnaire. Marital status was categorized as married/cohabiting, unmarried (never married), divorced/separated, and widow(er). Information on socioeconomic status was obtained from Statistics Norway. Educational level was categorized as low (elementary school/no formal education), medium (upper secondary education), and high (higher education). Pretax household income per year was categorized as Norwegian kroner (NOK) < , NOK NOK < , and NOK The following lifestyle variables were obtained from the questionnaire: alcohol use during the previous year (never been drinking/not been drinking the last year, been drinking 3 Figure 1 Simplified directed acyclic graph for the association between road traffic noise and sleep disturbances where some variables have been grouped for legibility. a Includes the variables age, sex, marital status, and having children 5 years. b Includes the variables educational level, employment status, occupation, and household income. c Includes the variables smoking status, alcohol use, caffeine use, physical activity, and night-shift work.

3 times/month, 1 3 times/week, and 4 7 times/week), smoking status (current, previous, or never daily smoker), and physical activity (sedentary [mostly sedentary activities], moderately active [light physical activity at least 2 4 hours a week], and highly active [heavy physical activity at least four hours a week]). Night-shift work was categorized according to number of years the participants had worked regular night shifts (0 years, 1 <5 years, and 5 years). Statistical Analyses We used logistic regression for modeling the associations between road traffic noise and symptoms of insomnia and sleep medication use, respectively. For the analyses of the total study population, we ran two models for each association: In Model 1, we adjusted for age and sex, and in Model 2 we adjusted for the variables identified in the DAG and the socioeconomic factors educational level and income. For each analysis, we excluded observations with missing values on any of the variables in Model 2. The noise levels were modeled at the most exposed facade of each participant s residential address. We therefore performed additional analyses for participants who reported to have their bedroom facing a road, assuming they had a more accurate noise exposure than participants with their bedroom located at a shielded facade. The tendency to keep the bedroom window open or closed affects the actual noise level the participants are exposed to. To address this issue, we performed sensitivity analyses, where we included only the participants who reported to sleep with the bedroom window open (n = ). We performed these sensitivity analyses both for the total study population and among individuals with their bedroom facing a road. Working regular night shifts over a prolonged period of time may have severe consequences for sleep quality. 51 Night-shift workers may therefore experience sleep problems for other reasons than the exposure to traffic noise. In addition, regular night-shift workers have their sleeping period during daytime rather than nighttime which does not fit the applied exposure data as the noise levels are estimated for nighttime. Thus, we performed sensitivity analyses, where we excluded individuals who reported regular night shifts for 1 year or more (n = 2046). Sleep medication use could potentially affect how the participants respond to questions concerning sleep quality. A sleep medication user could, for instance, report to sleep well due to the effect of the medications. However, it is also possible that someone using sleep medications could report to have disturbed sleep because of being dependent on medications in order to sleep. Because of these uncertainties in interpreting questions on sleep disturbances, we performed additional analyses where we stratified the study population into sleep medication users and nonusers. Indications that noise exposure may affect sleep differently in men and women have been reported. 52 Thus, we tested the interaction between road traffic noise and sex on the insomnia symptoms and sleep medication use by means of the log-likelihood test. The results are reported as changes in the effect estimate per 5-dB increase in noise level. We used the 5% level of 3 significance for all statistical analyses, except for the test of interaction for which we used the 10% level. Statistical analyses were carried out in STATA version 13 (StataCorp, College Station, Texas). In order to visualize the associations between traffic noise and each insomnia symptom and sleep medication use, we estimated a smooth function of the associations using nonparametric regression spline as smoother. Such models are named Generalized Additive Models (GAM). We used the function gam in library mgcv in the R statistical software version (The R Project for Statistical Computing, Vienna, Austria) to estimate the splines with 95% confidence limits. RESULTS The modeled noise levels in the total study population ranged from L night 7.6 db to L night 70.8 db with a mean of L night 47.2 db (SD = ±8.0). Table 1 shows characteristics of the study population in total and by three categories of traffic noise exposure. There was a larger proportion of women than men participating in the study. Individuals in the category of high noise exposure were characterized as being in the youngest age-group, unmarried, divorced or separated, having no formal education or elementary school only, and having low household income, compared to those exposed to lower noise levels. Regarding lifestyle, we found higher proportions of smokers and sedentary individuals in the highest noise category. In this study population of adults, the prevalence of the various insomnia symptoms was 10.6% for difficulties falling asleep, 12.8% for awakenings during the night, and 11.0% for waking up too early. Furthermore, 11.0% reported to currently use sleep medications. Of the total study population, 31.6% (n = 3803) reported to have their bedroom facing a road. In this subpopulation, the proportion of individuals reporting insomnia symptoms and sleep medication use was higher than in the total study population (Table S1). In the total study population, there were statistically significant associations between road traffic noise and all three symptoms of insomnia (Table 2). The strongest associations were observed in the age- and sex-adjusted model (Model 1). In the model adjusted for potential confounders (Model 2), we found that all three symptoms of insomnia, difficulties falling asleep, awakenings during the night, and waking up too early, were positively associated with a 5-dB increase in traffic noise level. However, the association between traffic noise and awakenings during the night was somewhat weaker and only borderline statistically significant. Regarding sleep medication use, the results indicated no association with traffic noise exposure. Among individuals with their bedroom facing a road, we found a higher effect estimate for the association between traffic noise exposure and difficulties falling asleep, compared to the total study population. There was, however, a decrease in the estimates for awakenings during the night, waking up too early, and sleep medication use, respectively (Table 2). The splines of the association between traffic noise and each insomnia symptom and sleep medication use are shown in Figure 2. An exposure response relationship was shown for the association between traffic noise and difficulties falling asleep and waking up too early, respectively. The shape of the spline for awakenings during the night, however, indicated a more

4 Table 1 Characteristics of the Study Population by Nighttime Road Traffic Noise (L night ) Exposure. Characteristic (No. missing) Nighttime road traffic noise (L night ) No. (%) <45 db, 4304 (35.5) 45 <55 db, 5919 (48.8) 55 db, 1915 (15.8) Total, (100) Age (0) 39 years 793 (18.4) 1196 (20.2) 457 (23.9) 2446 (20.2) 49 years 1035 (24.0) 1257 (21.2) 393 (20.5) 2685 (22.1) 55 years 942 (21.9) 1215 (20.5) 354 (18.5) 2511 (20.7) years 1123 (26.1) 1562 (26.4) 504 (26.3) 3189 (26.3) years 411 (9.5) 689 (11.6) 207 (10.8) 1307 (10.8) Sex (0) Women 2331 (54.2) 3179 (53.7) 1049 (54.8) 6,559 (54.0) Men 1973 (45.8) 2740 (46.3) 866 (45.2) 5579 (46.0) Marital status (1) Married/cohabiting 3244 (75.4) 3957 (66.9) 1062 (55.5) 8263 (68.1) Unmarried 355 (8.2) 810 (13.7) 413 (21.6) 1578 (13.0) Divorced/separated 438 (10.2) 731 (12.4) 284 (14.8) 1453 (12.0) Widow(er) 267 (6.2) 420 (7.1) 156 (8.2) 843 (6.9) Educational level (89) Low 527 (12.3) 931 (15.8) 325 (17.1) 1783 (14.8) Medium 1505 (35.2) 2146 (36.5) 677 (35.6) 4328 (35.9) High 2,240 (52.4) 2,799 (47.6) 899 (47.3) 5938 (49.3) Household income (13) NOK < 500k 972 (22.6) 1,824 (30.8) 737 (38.6) 3533 (29.1) NOK 500k <1000k 1621 (37.7) 2,366 (40.0) 745 (39.0) 4732 (39.0) NOK 1000k 1708 (39.7) 1724 (29.2) 428 (22.4) 3860 (31.8) Alcohol use (79) Never/ not last year 318 (7.4) 464 (7.9) 182 (9.6) 964 (8.0) 3 times / month 1273 (29.8) 1805 (30.7) 619 (32.6) 3697 (30.7) 1 3 times / week 2134 (49.9) 2799 (47.5) 807 (42.5) 5740 (47.6) 4 7 times / week 548 (12.8) 820 (13.9) 290 (15.3) 1658 (13.7) Smoking status (112) Current 582 (13.6) 950 (16.2) 357 (18.8) 1889 (15.7) Former 1570 (36.8) 2,235 (38.2) 713 (37.5) 4518 (37.6) Never 2117 (49.6) 2673 (45.6) 829 (43.7) 5619 (46.7) Physical activity (272) Sedentary 367 (8.7) 646 (11.2) 239 (12.8) 1,252 (10.6) Moderately active 2776 (65.8) 3800 (65.7) 1254 (67.2) 7830 (66.0) Highly active 1074 (25.5) 1337 (23.1) 373 (20.0) 2784 (23.5) Years with night shifts (197) 0 years 3548 (83.7) 4821 (82.9) 1526 (80.9) 9895 (82.9) 1 <5 years 385 (9.1) 495 (8.5) 181 (9.6) 1061 (8.9) 5 years 306 (7.2) 499 (8.6) 180 (9.5) 985 (8.2) Difficulties falling asleep (244) 3 5 times / week 381 (9.0) 649 (11.2) 233 (12.4) 1263 (10.6) <3 5 times week 3846 (91.0) 5142 (88.8) 1643 (87.6) (89.4) Awakenings during the night (396) 3 5 times / week 467 (11.1) 768 (13.5) 270 (14.6) 1505 (12.8) <3 5 times week 3722 (88.9) 4936 (86.5) 1579 (85,4) (87.2) Waking up too early (447) 3 5 times / week 391 (9.4) 653 (11.5) 246 (13.4) 1290 (11.0) <3 5 times week 3777 (90.6) 5029 (88.5) 1595 (86.6) (89.0) Sleep medication use (335) Yes 433 (10.3) 633 (11.0) 228 (12.3) 1294 (11.0) Abbreviations: db, decibel; NOK, Norwegian kroner. No 3777 (89.7) 5104 (89.0) 1628 (87.7) (89.0) 4

5 Table 2 ORs and 95% CIs for the Association Between Nighttime Road Traffic Noise (L night ) and Insomnia Symptoms and Sleep Medication Use for the Total Study Population and Among Individuals With Bedroom Facing a Road by 5-dB increase in Noise Level. Total Study Population n Model 1 a, OR (95% CI) Model 2 b, OR (95% CI) Difficulties falling asleep ( ) 1.05 ( ) Awakenings during the night ( ) 1.04 ( ) Waking up too early ( ) 1.06 ( ) Use of sleep medication ( ) 0.99 ( ) Bedroom facing road Difficulties falling asleep ( ) 1.08 ( ) Awakenings during the night ( ) 1.03 ( ) Waking up too early ( ) 1.00 ( ) Use of sleep medication ( ) 0.93 ( ) a Model adjusted for age and sex. b Model adjusted for age, sex, marital status, alcohol use, smoking status, physical activity, night-shift work, educational level, and household income. Abbreviations: OR, odds ratio; CI, confidence interval; db, decibel. uncertain relationship with road traffic noise. The spline of the association between traffic noise and difficulties falling asleep was steeper among those having their bedroom facing a road than for the total population (Figure S1). In the sensitivity analyses where only individuals reporting to sleep with their bedroom window open were included, we observed only a slight increase in the effect estimate for waking up to early in the total study population. However, for individuals having their bedroom facing a road, the effect estimate increased for difficulties falling asleep and awakenings during the night (Table S2). After having excluded night-shift workers in the sensitivity analyses, we observed a slight increase in the effect estimate in Model 2 for the association between traffic noise exposure and difficulties falling asleep (odds ratio [OR] = 1.06; 95% confidence interval [CI]: ), compared to the total study population. For the other two insomnia symptoms, there were no changes in the effect estimates (Table S3). In Table 3, we have presented the results from the analyses stratified on sleep medication use. We found no statistically significant association between traffic noise and insomnia symptoms among the sleep medication users. For the nonusers, on the other hand, higher effect estimates were shown compared to the total study population (Table 2), except for waking up too early. The largest increase in effect estimate, compared with the total study population, was observed for the outcome difficulties falling asleep. In Figure 3, we have presented splines of the association between traffic noise and difficulties falling asleep among both users and nonusers of sleep medications. An exposure response relationship was indicated among nonusers but not among sleep medication users. Furthermore, the spline for nonusers of sleep medication (Figure 3) was steeper than the spline for the total study population (Figure 2). The tests of interaction between traffic noise exposure and sex on insomnia symptoms and sleep medication use were not statistically significant with p values ranging from.16 to DISCUSSION In this population-based study, we observed statistically significant associations between road traffic noise and insomnia symptoms. Visual assessment of the splines of these associations revealed exposure response relationships between traffic noise and difficulties falling asleep and waking up too early, respectively. We found no association between traffic noise and sleep medication use. In the present study, we found that the proportion reporting the different insomnia symptoms ranged from 10.6% to 12.8% which is close to the estimated proportion having insomnia in Norway (11.7%). 53 In the total study population, the effect estimates for the association between road traffic noise and difficulties falling asleep and waking up too early were similar. The estimate for awakenings during the night was somewhat lower. This may be due to the fact that traffic volumes are lower during the middle of the night than in the late evening and early morning. The spline for the association between traffic noise and awakenings during the night indicated a tendency toward an inverse association from traffic noise levels around 55 db. This may indicate that awakenings during the central part of the night are not as sensitive to traffic noise as the other two outcomes. This further suggests that the sleep phases in the start and end of the night are particularly vulnerable to noise. An experimental study investigating the effect of traffic curfews in order to prevent sleep disturbances reported that traffic curfews were most beneficial in the end of the night. 54 Furthermore, the noise indicator L night represents the average noise level of the period from 23:00 to 07:00, while the sleep outcomes difficulties falling asleep, awakenings during the night, and waking up too early relate to different time periods of the night. Ideally, more specific time frames of the night could have been used for each outcome in the estimation of noise exposure. However, no such data were available, and the average noise level like L night is commonly used in scientific studies and is currently used by WHO for providing guidelines for nighttime noise. 55

6 Figure 2 Splines with 95% confidence limits of the associations between nighttime road traffic noise from 35 db and insomnia symptoms and sleep medication use for the total study population. The vertical lines on the x-axis indicate number of observations. General Additive Model adjusted for age, sex, marital status, alcohol use, smoking status, physical activity, night-shift work, educational level, and household income. db, decibel. In the present study, no association between traffic noise and self-reported sleep medication use was found. Although sleep medication use may be a proxy for sleep problems, this outcome may also cover more severe sleep disorders as well as anxiety and depression. 56 Thus, other mechanisms than those induced by traffic noise may be more important when it comes to sleep medication use. Previous research shows conflicting results with respect to a possible association between traffic noise and sleep medication use. A large register-based study reported an association among individuals living in areas low in social deprivation. 41 Furthermore, in line with the present study, null findings have also been reported. 12,39,40 Some of the inconsistency in the results of the previous research may relate to 6 that some studies have used self-reports of sleep medications as outcome and others have applied register-based medication use. Another issue regarding self-reported sleep medication use as an outcome is that the respondents definition of sleep medications may vary. 38 Some might define sleep medications as prescription required medications only, such as hypnotics, while others might define it less strictly and include over-the-counter sleep aids in their definition. A more specific outcome on sleep medication use could have been obtained by having the participants report what type of sleep medication they used and the reasons for using it. Although sleep medication use could certainly affect the association between noise and sleep disturbances, 57 individual

7 Table 3 ORs and 95% CIs for the Association Between Nighttime Road Traffic Noise (L night ) and Insomnia Symptoms Stratified by Sleep Medication Use by 5-dB Increase in Noise Level. Sleep medications n Model 1 a, OR (95% CI) Model 2 b, OR (95% CI) Difficulties falling asleep No ( ) 1.09 ( ) Yes ( ) 1.00 ( ) Awakenings during the night No ( ) 1.05 ( ) Yes ( ) 1.01 ( ) Waking up too early No ( ) 1.08 ( ) Yes ( ) 1.04 ( ) a Model adjusted for age and sex. b Model adjusted for age, sex, marital status, alcohol use, smoking status, physical activity, night-shift work, educational level, and household income. Abbreviations: OR, odds ratio; CI, confidence interval; db, decibel. Figure 3 Splines with 95% confidence limits of the associations between nighttime road traffic noise from 35 db and difficulties falling asleep among nonusers (left) and users of sleep medications (right). The vertical lines on the x-axis indicate number of observations. General Additive Model adjusted for age, sex, marital status, alcohol use, smoking status, physical activity, night-shift work, educational level, and household income. db, decibel. use of sleep medications has rarely been taken into account in population-based studies on traffic noise and sleep disturbances. Exclusion of sleep medication users has been a more common practice in experimental studies on noise and sleep. 8,58,59 To our knowledge, only one population-based study on self-reported sleep disturbances has taken sleep medication use into consideration by excluding regular sleep medication users from the analyses. 37 In the present study, instead of solely excluding sleep medication users from the analyses, we performed stratified analyses, which enabled us to compare the effect estimates for nonusers with the total study population as well as with the users of sleep medication. The results from the stratified analyses showed a clear increase in 7 the effect estimate for difficulties falling asleep among nonusers, compared to the total study population. For the other insomnia symptoms, there were only minor changes. No associations were found among the users. That we only observed a clear increase for difficulties falling asleep is not surprising, as the majority of sleep medications involve agents working to enhance sleep-promoting systems. 60 In large-scale studies, noise exposure is commonly modeled for the most exposed side of the residential building, although the bedroom may be located on a shielded side. Therefore, by not accounting for bedroom location, the effect of traffic noise on sleep could be underestimated. 34,37,61,62 Furthermore, not being able to account for bedroom location has been reported

8 as a limitation in previous research. 34,36,37 In the present study, we had self-reported data on whether the participants bedrooms were facing a road or a shielded side of their homes. We further defined a subpopulation of individuals having their bedroom facing a road. Assuming that the side of the home facing a road is the most noise exposed side this subpopulation was likely to have a more precisely assigned noise exposure than the rest of the study population. We observed an increase in the effect estimate for the association between traffic noise and difficulties falling asleep in this subpopulation, which may indicate an underestimation of the same association in the total study population. This increase in effect estimate further suggests that the process of falling asleep may be particularly vulnerable to road traffic noise. This is in line with previous research reporting that individuals sleeping on the street side of the building have more difficulties falling asleep and have longer sleep onset latency than those sleeping on a more quiet side. 58 From another perspective, if the bedroom window is facing a yard, water, or green space, sleep problems have been found to be less likely. 63 The effect estimate for waking up too early was reduced among individuals with the bedroom facing a road, compared with the total study population, whereas the effect estimate for awakenings during the night remained similar. The reduction in the effect estimate for waking up too early may indicate a more uncertain association with traffic noise for this outcome. Among the participants keeping their bedroom window open, we observed only slight changes in the effect estimates, compared to the total study population. However, in the subpopulation having the bedroom facing a road, there was an increase in the effect estimates for difficulties falling asleep and awakenings during the night. Assuming a more precisely assigned nighttime noise exposure for those having the bedroom facing a road, it is not surprising that the associations were even stronger among individuals sleeping with the bedroom window open. After having excluded night-shift workers for the sensitivity analysis, we observed a slight increase in the effect estimate for the association between traffic noise and difficulties falling asleep. The estimates for the other insomnia symptoms and for sleep medication use remained unchanged after exclusion of night-shift workers. However, a limitation with our data on night-shift work is that we could not separate current nightshift work from previous night-shift work. If a participant had worked regular night shifts many years ago, it may not be relevant for sleep disturbances at the time of the present study. Strengths Strengths of the present study include that we used data from a large population-based study (HELMILO) specifically designed to examine health effects from environmental exposures. We used modeled levels of noise exposure, obtained independent of the respondents report of sleep disturbances and other questionnaire information, preventing information bias. There was a broad range in the noise exposure level, which strengthened the opportunity to detect associations. We had information on whether the participants bedrooms were facing a road or a shielded side of the building and could therefore define a subpopulation with noise exposure outside the bedroom. In order to further reduce exposure misclassification, we performed 8 sensitivity analyses for individuals reporting to sleep with their bedroom window open. By collecting information about both difficulties falling asleep, and awakenings, both during the night and in the early morning, we were able to assess how nighttime traffic noise was related to sleep during different periods of the night. Furthermore, we had information on a large set of potential confounders. By using a DAG for confounder selection, we went through a thorough process for defining the statistical model and thereby minimized bias. 64 Limitations Some limitations apply to this study. First, the study is crosssectional; hence, we cannot know whether the sleep problems were already present prior to the noise exposure. However, sleep disturbances as a response to noise can be fairly acute, so a longitudinal study for assessing this association is not necessarily needed. Furthermore, by excluding those having lived at their home address for less than 1 year, we can assume that the noise exposure level for the respondents has been stable for some period of time. A response rate of 48% is not optimal regarding representativeness. Although the generalizability of the study may be affected by a low response rate, this is only the case if the relationship between exposure and outcome is different among the study participants and those who did not participate. 65 Another limitation is that the questionnaire did not include information on daytime functioning following sleep disturbances. Having information on daytime functioning would have enabled us to generate an outcome more in line with the diagnostic criteria for insomnia according to DSM-V. 32 Regarding potential inaccuracy in noise exposure assessment, we did not have information on sound insulation factors such as type of bedroom window, ventilation, and facade insulation, which affects the transmission of outdoor traffic noise to inside the bedroom. Random error in estimates of exposure would probably dilute the associations between exposure and outcomes. CONCLUSION We observed exposure response relationships between nighttime road traffic noise and difficulties falling asleep and waking up too early. These findings indicate that individuals exposed to high levels of nighttime traffic noise might suffer from insufficient sleep. Among individuals with the bedroom facing a road, the effect estimate for difficulties falling asleep was increased, compared to the total study population. Of the insomnia symptoms we explored, difficulties falling asleep seemed to be the most vulnerable to traffic noise. Our results did, however, not indicate an association between traffic noise and sleep medication use. In order to improve the validity of the results in future population-based studies, standardized questionnaires on sleep disturbances should be applied. Furthermore, traffic noise exposure should be estimated for the bedroom facade of the participants homes. The findings of the present study may contribute to future health impact assessments for nighttime road traffic noise, which can in turn be used for noise abatements and thereby improve sleep quality for people living in areas highly exposed to traffic noise.

9 REFERENCES 1. World Health Organization. Burden of disease from environmental noise. In: Fritschi L, Brown AL, Kim R, Schwela D, Kephalopoulos S (eds). Quantification of healthy life years lost in Europe. Copenhagen, Denmark: WHO Regional Office for Europe; 2011: V. 2. Basner M, Babisch W, Davis A, et al. Auditory and non-auditory effects of noise on health. The Lancet 2014;383(9925): Hume KI. Noise pollution: a ubiquitous unrecognized disruptor of sleep? Sleep. 2011; 34(1): Pirrera S, De Valck E, Cluydts R. Nocturnal road traffic noise: A review on its assessment and consequences on sleep and health. Environ Int. 2010; 36(5): Muzet A. Environmental noise, sleep and health. Sleep Med Rev. 2007; 11(2): Basner M, Müller U, Elmenhorst EM. Single and combined effects of air, road, and rail traffic noise on sleep and recuperation. Sleep. 2011; 34(1): Griefahn B, Bröde P, Marks A, Basner M. Autonomic arousals related to traffic noise during sleep. Sleep. 2008; 31(4): Aasvang GM, Øverland B, Ursin R, Moum T. A field study of effects of road traffic and railway noise on polysomnographic sleep parameters. J Acoust Soc Am. 2011; 129(6): Öhrström E, Hadzibajramovic E, Holmes M, Svensson H. Effects of road traffic noise on sleep: Studies on children and adults. J Environ Psychol. 2006; 26(2): Skånberg A, Öhrström E. Sleep disturbances from road traffic noise: a comparison between laboratory and field settings. J Sound Vib. 2006; 290(1-2): Elmenhorst EM, Elmenhorst D, Wenzel J, et al. Effects of nocturnal aircraft noise on cognitive performance in the following morning: dose-response relationships in laboratory and field. Int Arch Occup Environ Health. 2010; 83(7): de Kluizenaar Y, Janssen SA, van Lenthe FJ, Miedema HM, Mackenbach JP. Long-term road traffic noise exposure is associated with an increase in morning tiredness. J Acoust Soc Am. 2009; 126(2): Griefahn B, Marks A, Robens S. Noise emitted from road, rail and air traffic and their effects on sleep. J Sound Vib. 2006; 295: Oftedal B, Krog NH, Pyko A, et al. Road traffic noise and markers of obesity - a population-based study. Environ Res. 2015; 138: Pyko A, Eriksson C, Oftedal B, et al. Exposure to traffic noise and markers of obesity. Occup Environ Med. 2015; 72(8): Sørensen M, Andersen ZJ, Nordsborg RB, et al. Long-term exposure to road traffic noise and incident diabetes: a cohort study. Environ Health Perspect. 2013; 121(2): Bluhm GL, Berglind N, Nordling E, Rosenlund M. Road traffic noise and hypertension. Occup Environ Med 2007;64(2): Barregard L, Bonde E, Ohrström E. Risk of hypertension from exposure to road traffic noise in a population-based sample. Occup Environ Med. 2009; 66(6): de Kluizenaar Y, Gansevoort RT, Miedema HM, de Jong PE. Hypertension and road traffic noise exposure. J Occup Environ Med. 2007; 49(5): Bodin T, Albin M, Ardö J, Stroh E, Ostergren PO, Björk J. Road traffic noise and hypertension: results from a cross-sectional public health survey in southern Sweden. Environ Health. 2009; 8: Sørensen M, Hvidberg M, Hoffmann B, et al. Exposure to road traffic and railway noise and associations with blood pressure and self-reported hypertension: a cohort study. Environ Health. 2011; 10: Babisch W, Beule B, Schust M, Kersten N, Ising H. Traffic noise and risk of myocardial infarction. Epidemiology. 2005; 16(1): Selander J, Nilsson ME, Bluhm G, et al. Long-term exposure to road traffic noise and myocardial infarction. Epidemiology. 2009; 20(2): Sørensen M, Andersen ZJ, Nordsborg RB, et al. Road traffic noise and incident myocardial infarction: a prospective cohort study. PLoS One. 2012; 7(6): e Grandner MA, Jackson NJ, Pak VM, Gehrman PR. Sleep disturbance is associated with cardiovascular and metabolic disorders. J Sleep Res. 2012; 21(4): Vgontzas AN, Liao D, Bixler EO, Chrousos GP, Vela-Bueno A. Insomnia with objective short sleep duration is associated with a high risk for hypertension. Sleep. 2009; 32(4): Cappuccio FP, D Elia L, Strazzullo P, Miller MA. Quantity and quality of sleep and incidence of type 2 diabetes: a systematic review and meta-analysis. Diabetes Care. 2010; 33(2) Buxton OM, Marcelli E. Short and long sleep are positively associated with obesity, diabetes, hypertension, and cardiovascular disease among adults in the United States. Soc Sci Med. 2010; 71(5): Vgontzas AN, Liao D, Pejovic S, Calhoun S, Karataraki M, Bixler EO. Insomnia with objective short sleep duration is associated with type 2 diabetes: A population-based study. Diabetes Care. 2009; 32(11): Laugsand LE, Vatten LJ, Platou C, Janszky I. Insomnia and the risk of acute myocardial infarction: a population study. Circulation. 2011; 124(19): Hoevenaar-Blom MP, Spijkerman AM, Kromhout D, van den Berg JF, Verschuren WM. Sleep duration and sleep quality in relation to 12-year cardiovascular disease incidence: the MORGEN study. Sleep. 2011; 34(11): American Psychiatric Association. Diagnostics and statistical manual of mental disorders, fifth edition (DSM-V). Arlington, VA: American Psychiatric Association; Wickwire EM, Shaya FT, Scharf SM. Health economics of insomnia treatments: The return on investment for a good night s sleep. Sleep Med Rev. 2015; 30: Miedema HM, Vos H. Associations between self-reported sleep disturbance and environmental noise based on reanalyses of pooled data from 24 studies. Behav Sleep Med. 2007; 5(1): van den Berg F, Verhagen C, Uitenbroek D. The relation between scores on noise annoyance and noise disturbed sleep in a public health survey. Int J Environ Res Public Health. 2014; 11(2): Halonen JI, Vahtera J, Stansfeld S, et al. Associations between nighttime traffic noise and sleep: the Finnish public sector study. Environ Health Perspect. 2012; 120(10): Frei P, Mohler E, Röösli M. Effect of nocturnal road traffic noise exposure and annoyance on objective and subjective sleep quality. Int J Hyg Environ Health. 2014; 217(2-3): Barregard L, Stansfeld S. Medication use as an outcome variable in environmental (noise) epidemiology. Scand J Work Environ Health. 2014; 40(3): Floud S, Vigna-Taglianti F, Hansell A, et al. Medication use in relation to noise from aircraft and road traffic in six European countries: results of the HYENA study. Occup Environ Med 2011;68(7): Halonen JI, Lanki T, Yli-Tuomi T, et al. Associations of traffic noise with self-rated health and psychotropic medication use. Scand J Work Environ Health. 2014; 40(3): Bocquier A, Cortaredona S, Boutin C, et al. Is exposure to night-time traffic noise a risk factor for purchase of anxiolytic-hypnotic medication? A cohort study. Eur J Public Health. 2014; 24(2): Bendtsen H. The Nordic prediction method for road traffic noise. Sci Total Environ 1999;235: Nordic Council of Ministers. Road traffic noise - Nordic prediction method. TemaNord, Copenhagen: Nordic Council of Ministers; 1996: DataKustik. CadnaA Computer aided noise abatement. Manual version 3.4; European Commission. Directive 2002/49/EC of the European Parliament and of the Coucil of 25 June 2002 relating to the assessment and management of environmental noise. 46. The Municipality of Oslo. Støykartlegging 2012 (Strategic noise mapping in Oslo). Forurensningsforskriften kapittel 5 - Støysituasjon for år Oslo, Norway: Bymiljøetaten; Pearl J. Causality: models, reasoning and inference. New York, NY: Cambridge University Press; Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology. 1999; 10(1): Textor J, Hardt J, Knüppel S. DAGitty: a graphical tool for analyzing causal diagrams. Epidemiology. 2011; 22(5): World Health Organization. Environmental Health Inequalities in Europe. Copenhagen, Denmark: WHO regional office for Europe; Akerstedt T, Wright KP Jr. Sleep Loss and Fatigue in Shift Work and Shift Work Disorder. Sleep Med Clin. 2009; 4(2):

10 52. Röösli M, Mohler E, Frei P, Vienneau D. Noise-related sleep disturbances: does gender matter? Noise Health. 2014; 16(71): Pallesen S, Nordhus IH, Nielsen GH, et al. Prevalence of insomnia in the adult Norwegian population. Sleep. 2001; 24(7): Griefahn B, Marks A, Robens S. Experiments on the time frame of temporally limited traffic curfews to prevent noise induced sleep disturbances. Somnologie-Schlafforschung und Schlafmedizin. 2008; 12: World Health Organization. Night noise guidelines for Europe. Copenhagen, Denmark: WHO regional office for Europe; Omvik S, Pallesen S, Bjorvatn B, Sivertsen B, Havik OE, Nordhus IH. Patient characteristics and predictors of sleep medication use. Int Clin Psychopharmacol. 2010; 25(2): Cluydts R, De Roeck J, Cosyns P, Lacante P. Antagonizing the effects of experimentally induced sleep disturbance in healthy volunteers by lormetazepam and zolpidem. J Clin Psychopharmacol. 1995; 15(2): Pirrera S, De Valck E, Cluydts R. Field study on the impact of nocturnal road traffic noise on sleep: the importance of in- and outdoor noise assessment, the bedroom location and nighttime noise disturbances. Sci Total Environ. 2014; : Griefahn B, Schuemer-Kohrs A, Schuemer R, Moehler U, Mehnert P. Physiological, subjective, and behavioural responses during sleep to noise from rail and road traffic. Noise Health. 2000; 3(9): Krystal AD. Current, emerging, and newly available insomnia medications. The Journal of clinical psychiatry. 2015; 76: e Aasvang GM, Moum T, Engdahl B. Self-reported sleep disturbances due to railway noise: exposure-response relationships for nighttime equivalent and maximum noise levels. J Acoust Soc Am. 2008; 124(1): Fyhri A, Aasvang GM. Noise, sleep and poor health: Modeling the relationship between road traffic noise and cardiovascular problems. Sci Total Environ. 2010; 408(21): Bodin T, Björk J, Ardö J, Albin M. Annoyance, sleep and concentration problems due to combined traffic noise and the benefit of quiet side. Int J Environ Res Public Health. 2015; 12(2): Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol. 2008; 8: Rothman KJ. Epidemiology: An Introduction. 2 ed. New York, NY: Oxford University Press; 2012: SUPPLEMENTARY MATERIAL Supplementary Material is available at SLEEP online. ETHICAL APPROVAL The study was approved by the Regional Committee for Medical and Health Research Ethics and the Norwegian Data Inspectorate. Written informed consent was obtained from all participants. This was not an industry supported study. FINANCIAL SUPPORT Funding was provided by the Research Council of Norway Program for Environmental Exposures and Health Outcomes. Project no.: ACKNOWLEDGMENT The authors would like to thank Erik Aune at the City of Oslo, Agency for Urban Environment for invaluable help with the noise exposure data. No clinical trial is indicated. This work was performed at the Norwegian Institute of Public Health. SUBMISSION & CORRESPONDENCE INFORMATION Submitted for publication August, 2016 Submitted in final revised form October, 2016 Accepted for publication October, 2016 Address correspondence to: Jorunn Evandt, MSc, Norwegian Institute of Public Health, P.O. Box 4404 Nydalen, N-0403 Oslo. Telephone: ; Fax: ; jorunn.evandt@fhi.no DISCLOSURE STATEMENT None disclosed. 10