Understanding insomnia complaints requires a careful examination of factors that may cause or maintain the sleep problem. Insomnia was defined in different ways between the epidemiological surveys. Global sleep dissatisfaction (GSD) is not part of the habitual insomnia symptoms in epidemiological studies. Furthermore, none of these studies has examined the relative importance of the various factors correlated to sleep dissatisfaction. This study aims to examine the links between GSD and insomnia and to find the factors contributing to GSD.

In a previous study we have showed that insomnia subjects dissatisfied with their sleep are more likely to seek medical help, to report daytime impaired functioning and to be diagnosed with a sleep or a mental disorder than those who have insomnia symptoms without a complaint of sleep dissatisfaction (1). These results were confirmed in a study performed in Hong-Kong (2).

Several factors were found to be related to sleep quality and to increase sleep dissatisfaction and insomnia:

• First, the sociodemographic characteristics of subjects dissatisfied with their sleep are often involved. Epidemiological studies have reported an increased prevalence of sleep dissatisfaction in women (1,4). The variation of prevalence with age had mixed results (1,3). Marital and economical status, educational level and occupation were also found to be associated with sleep dissatisfaction, but not always (1).
• Second, environmental factors, such as the temperature of the bedroom or the level of noise when the subject is sleeping, also may play a role in the genesis of insomnia (5). Some of these factors were studied in specific populations such as shiftworkers (6), but were rarely assessed in the general population.
• Third, life habits can also contribute to the appearance or maintenance of sleep problems. In the literature, use of alcohol, especially as a sleeping aid, coffee and tobacco were non-systematically reported to be associated with insomnia. Doing regularly physical activities, on the other hand, would help to maintain a good sleep (7).
• Fourth, the health of subjects with insomnia symptoms and of those dissatisfied with their sleep was often described as being poorer than that of normal sleepers. They were making a more negative subjective evaluation of their health. Some studies have reported a greater number of insomnia subjects suffering musculo-skeletal diseases, cardio-vascular diseases and diabetes (8,9).
• Fifth, psychological factors were also frequently identified in insomnia subjects. Life stress, depressive and anxious mood and stressful events were commonly found to be associated with insomnia in the general population (1,9,10).

Unfortunately, few epidemiological surveys have used all these factors to examine their contribution to sleep dissatisfaction within a same general population. Most of the studies limited their analyses to one or two of these factors. Therefore, this study aims to examine the importance of these different factors in relation to sleep dissatisfaction and to examine the links between GSD and insomnia.

METHODS

A - SAMPLE

This epidemiological study on sleeping habits, sleep and mental disorders was performed from January to October 1996 in Germany. The targeted population was non-institutionalized German residents 15 years of age or older. This represented 66,649,000 inhabitants.

A representative sample of this population was drawn using a two-stage sampling design:

• At the first stage, a random sample of telephone numbers was drawn based upon the population distribution within the 16 states of Germany.
• At the second stage, during the telephone contact, the Kish method (12) was used to select one respondent in the household.

This method allowed selecting a respondent based on age and gender to maintain the representation of the sample according to these two parameters. If the household member thus chosen refused to participate, the household was dropped and replaced by another, and the process repeated.

During the telephone contact, interviewers explained the goals of the study to potential participants and then requested verbal consent before interviewing the participants. A letter was sent to those who requested further details before agreeing to participate. Excluded from the study were subjects who did not speak sufficient German, who suffered from a hearing or speech impairment or an illness that precluded being interviewed.

The study was approved by ethical and research committees of Montreal (Canada) and Regensburg (Germany). The interviews were conducted with the help of a pool service company specialized in conducting telephone surveys. Subjects who refused to participate or who asked to stop before completing at least half the interview were classified as refusals. Phone numbers were dropped and replaced only after a minimum of 10 unsuccessful dial attempts were made at different times and on different days, including weekdays and weekends.

An added-digit technique, that is, increasing the last digit of a number by one, was employed to control for unlisted telephone numbers. The final sample included 10.4% of unlisted telephone numbers. The participation rate (68.1%) was calculated based on the number of interviews (n=4115) divided by the number of eligible telephone numbers, which included all residential numbers not meeting any of the exclusion criteria (N=6047). The final sample included 28.9% of subjects who initially refused to participate but accepted upon a second request.

INTERVIEWERS

Thirty-seven native German-speaking interviewers inexperienced in psychiatric assessment but who received special training on how to use the Sleep-EVAL knowledge-based system performed the interviews. The training was made by the Principal Investigator (MMO). It consisted mainly of role-playing, during which the interviewers practiced how to introduce the study and how to ask and answer the questions. They were instructed to read all the answer choices and to never decide for the subject what was the most appropriate answer. The average duration of the interviews was 48.0 ±18 minutes.

The team of interviewers was monitored daily by two supervisors, who listened on calls to ensure that questions were asked correctly and data entered properly. The supervisors reported daily the progress of the study to the principal investigator or to one of his staff members.

INSTRUMENT

Interviewers used the Sleep-EVAL system (12,13) to conduct the telephone interviews. This system is an artificial intelligence tool, specifically designed to administer questionnaires and conduct epidemiological studies on mental and sleep disorders in the general population (13). The system is composed of a non-monotonic, level-2 inference engine endowed with a causal reasoning mode that simulates the reasoning process employed in a psychiatric or sleep consultation.

Interviews started first with a standard questionnaire administrated to all the participants. After analysis of these first answers, the Sleep-EVAL system formulated a series of diagnostic hypotheses (done using causal reasoning mode). These hypotheses were examined by the Sleep-EVAL's non-monotonic, level-2 inference engine and were confirmed or rejected through further questions and deductions. The system rendered its diagnoses using two classifications: the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) (14) and the International Classification of Sleep Disorders (ICSD) (15). Concurrent diagnoses were possible as long as the diagnoses respected the recommendations of the used classifications and were explored as long as a symptom, criterion or syndrome particular to that diagnosis was present.

The system selected and phrased the questions to be administered and provided examples and instructions on how to ask them during the interview. The interviewer simply read out the questions as they appeared on the monitor and entered the responses. Questions were closed-ended (e.g., yes-no, present-absent-unknown, five-point scale, fuzzy sets) or open-ended (e.g., name of illness, duration). The system has been validated in various contexts and has been demonstrated to be reliable and valid (16,17).

VARIABLES

Global dissatisfaction with sleep (GSD) was defined using the following question read by the interviewer (including the alternative phrasing in brackets):
In general, do you find your sleep is:

• Very satisfactory? [Or if you prefer, would you say "I sleep well"]
• Rather satisfactory? [Or if you prefer, would you say "occasionally I do not sleep well, but I am generally happy about my sleep"]
• Rather unsatisfactory? [Or if you prefer, would you say "my sleep has already caused me problems, I think I have a problem with my sleep"]
• Quite unsatisfactory?
• Completely unsatisfactory?

Subjects were classified as GSD if they answered rather, quite or completely unsatisfactory. Dissatisfaction with sleep was analyzed in association with six classes of variables identified as potential associated factors in the literature:

• Sociodemographic factors: age, gender, marital status, occupation, income, educational level, states and size of the settlement.
• Environmental factors: quality of the bedroom: temperature, mattress, furniture, level of noise.
• Life habit factors: use of alcohol, tobacco, caffeine (drinking caffeinated beverages close to bedtime), eating before going to bed, activities performed in bed before falling asleep, required objects or particular environment needed to fall asleep (e.g., a special object, listening to music, etc), doing physical exercise at least three times per week for at least 15 minutes per session.
• Health factors: physical disease (cardiovascular disease, musculo-skeletal disease, diabetes, hypertension, etc.), body mass index and the number of medical consultations in the previous year.
• Psychological factors: level of stress, the occurrence of a stressful event during the previous year, satisfaction with one's social network, depressive mood, anxious mood, depressive disorders, anxiety disorders.
• Sleep-wake factors: bedtime hour, wake-up hour, regularity of the bedtime and wake-up hours, sleep latency, sleep duration and extra sleep on the weekend and days off.

Consequences of dissatisfaction with sleep were also analyzed: tiredness upon awakening, daytime sleepiness, naps, accidents and medical consultations for sleep problems. DSM-IV sleep and mental diagnoses were also analyzed.

STATISTICAL ANALYSES

Data were weighted to compensate for disparities between the final sample and the data of the official German census. Bivariate analyses involving categorical or qualitative variables were carried out with chi-square statistics. Each class of variables was analyzed using logistic regressions (18). Ninety-five percent confidence intervals were calculated for prevalence rates and odds ratios. Reported differences were significant at .05 or less.

RESULTS

The final sample was composed of 4,115 subjects aged from 15 to 99 years. The unweighted sample was composed of 2,216 women and 1,899 men. After adjusting for gender, age and region, the weighted sample consisted of 52.1% women. The demographic characteristics of the sample are presented in Table 1. Overall, 67% of subjects were between 25 and 64 years old. Half of them were women and half were married. More than half of the sample was employed; 7.9% worked on shifts or during the night (Table 1).

PREVALENCE

Overall, 7.0% of the sample reported a GSD. This prevalence was higher in women than in men (8.2% vs. 5.6%; p<.001). The prevalence also varied with age (p<.0001); the highest rates were found in subjects between 45 and 64 years of age (9.5%) and in those between 65 and 74 years (9.7%). The youngest subjects had the lowest prevalence (2.6%). The subjects between 25 and 44 years of age had a prevalence of GSD (6.1%) comparable to that of the oldest subjects (5.7%).

Almost all GSD subjects reported at least one insomnia symptom (95.4%). A complaint of disrupted sleep was reported by 80.9% of GSD subjects, difficulties initiating sleep was reported by 61.5%, early morning awakenings by 54.2% and non-restorative sleep by 60.9%. The duration of insomnia symptoms was longer in GSD subjects (average of 110.9 months) compared to subjects who reported insomnia symptoms but without GSD (average of 107.6 months; p< .0001).

ASSOCIATION WITH SLEEP AND MENTAL DISORDERS

In this sample, the prevalence of DSM-IV insomnia diagnoses was 6.2%. In the GSD group, 37.1% had a DSM-IV insomnia diagnosis compared to 9.8% in subjects with at least one insomnia symptom without GSD. The most frequent diagnosis encountered in the GSD group was insomnia related to another mental disorder (12.6%) followed by primary insomnia (9.5%) (Figure 1).

A quarter (25.7%) of GSD subjects had a sleep disorder diagnosis other than insomnia, mainly a diagnosis of circadian rhythm disorder (16.9%) (Figure 1). A third (37.0%) of GSD subjects had a DSM-IV diagnosis of mental disorder, mainly an anxiety disorder alone (12.0%) or associated with a mood disorder (12.6%) (Figure 1). As shown in Fig. 1, proportion of GSD subjects with any diagnosis was always significantly higher than in subjects with at least one insomnia symptom without GSD.

RELATED FACTORS

SOCIODEMOGRAPHIC FACTORS

Sociodemographic variables listed in the method section were entered into a logistic regression model. Marital status, educational level, income, states and size of the settlement were unrelated to GSD. The risk of GSD increased with age with the highest odds ratio in subjects between 65 and 74 years of age (OR=6.7), and with being a woman (OR=1.3) (Table 2).

ENVIRONMENTAL FACTORS

Another model was calculated for environmental factors. In addition to variables listed in the method section, age and gender were also introduced in the model. Non-significant variables were: noisy bedroom, cluttered bedroom and poor quality of the mattress. Variables significantly associated with GSD were: a bedroom that was too bright (OR=1.9 [1.1-3.2]; p<.05), too stuffy (OR=2.4 [1.4-4.3]; p<.05) or with inappropriate temperature (too hot (OR=2.5 [1.6-4.0]; p<.001) or too cold (OR=1.9 [1.2-3.1]; p<.01)).

LIFE HABIT FACTORS

Nine of the 16 variables in the category were non-significant: alcohol, drinking coffee, eating before going to sleep, use of the bed to watch TV, to read or to play games (like crossword puzzles), the need to have someone in the bed, and the need to listen to music or a familiar sound.

Factors associated with GSD were:

• smoking more than 25 cigarettes daily (OR=2.1 [1.3-3.6]; p<.01),
• using the bed to study or work (OR=2.1 [1.4-3.2]; p<.001);
• in order to fall asleep, the need of either a light on (OR=2.3 [1.4-3.7]; p<.01), a particular object (OR=2.4 [1.6-3.7]; p<.001) or a glass of water on the bedside table (OR=1.5 [1.1-2.1]; p<.05).
• Doing physical activity in the evening was a protective factor (OR=0.5 [0.3-0.8]; p<.001).

HEALTH FACTORS

Three of the nine health variables were non-significant in the multivariate model:

• the presence and number of medical consultations in the previous year,
• diabetes and thyroid disease.

The strongest factor associated with GSD was:

• to have an upper airway disease (asthma, chronic bronchitis) (OR=7.1),
• followed by musculo-skeletal disease, heart disease, hypertension or another type of disease.

A body mass index between 25 and 29 kg/m2 appeared to be a protective factor. Odds ratios and their 95% confidence intervals are presented in Table 3.

PSYCHOLOGICAL FACTORS

All variables entered in this model were significant.

The strongest factor associated with GSD was reporting:

• anxiety symptoms (OR=3.0), followed by
• having a very stressful life,
• a depressive mood,
• having experienced a stressful event in the previous year and
• being dissatisfied with ones social life.

Odds ratios and their 95% confidence intervals are presented in Table 4.

SLEEP-WAKE FACTORS

Of the seven variables entered into this model, sleep duration shorter than five hours (OR=9.7), shorter than 6 hours, and extended sleep latency (30 minutes or more) were significantly associated with GSD. Odds ratios and their 95% confidence intervals are presented in Table 5. Therefore, bedtime hour, wake-up hour, regularity of the sleep-wake schedule and extra sleep on weekends and days off were non-significant.

Finally, all the significant variables previously identified were introduced in a single model in order to identify the most meaningful factors associated with GSD. Age remained a strong predictor of GSD, but gender was no longer significant.

Other associated factors were:

• having an upper airway disease (OR:4.8) or
• a musculo-skeletal disease (OR=2.9);
• sleeping in a hot bedroom (OR=2.1);
• needing a particular object to fall asleep (OR=1.9);
• having an average (OR=1.8) or very stressful life (OR=2.9);
• reporting anxiety (OR=2.4) or
• depressive symptoms (OR=1.9);
• sleeping less than seven hours per night (OR from 1.8 to 9.2); and
• having a sleep latency of 30 minutes or more (30-59 min OR=2.9; 60 min and over OR=8.7).
• Doing physical activities close to bedtime remained a protective factor (OR: 0.5).

NAPS, DAYTIME SLEEPINESS AND ROAD ACCIDENTS

Naps were comparable between GSD subjects (39.9%) and subjects with at least one insomnia symptom without GSD (37.2%).

Excessive daytime sleepiness was more frequent in GSD subjects:

• 28.9% reported feeling moderately sleepy during the day and
• 11.3% said they felt sleepy a lot. By comparison, insomnia subjects without GSD reported to be moderately sleepy in 17.8% of cases and sleepy a lot in 4.8% of cases (p<.0001).

Overall, 61.3% of GSD subjects and 65.4% of subjects with at least one insomnia symptom but without GSD drove a car. However, GSD drivers were twice as likely to have a car accident in the previous year than drivers with at least one insomnia symptom but without GSD (13.0% vs. 6.9%; OR: 2.0 [1.2-3.4]; p<.01).

MEDICAL CONSULTATIONS AND SLEEP MEDICATION

Subjects were asked if they ever consulted a physician specifically about their sleep difficulties, or if they spoke with their doctor of their sleep difficulties during a medical consultation for another purpose, or if a medical visit was motivated by the consequences of their sleep problems.

Overall, 41.8% of GSD subjects answered positively to one of these questions compared with 6.0% in subjects with at least one insomnia symptom without GSD (p<.001). Use of a medication for sleep was reported by 15.1% of GSD subjects as compared to 3.6% in subjects with at least one insomnia symptom without GSD. Therefore, the likelihood of taking a sleep medication was five times higher in GSD subjects (OR: 4.8 [3.1-7.3]).

DISCUSSION

This epidemiological study was performed in the most populated European country. By the time of the study, over 66 millions people 15 years or older were living in Germany. It was found that 7% of the sample was globally dissatisfied with sleep (GSD). The question of the reliability of sleep data collected by telephone could be raised. However, the literature suggests telephone interviews in general are appropriate and yield results comparable to other strategies. A good inter-rater reliability is obtained between face-to-face and telephone interviews assessing DSM-IV psychiatric disorders (19). We did not collect physiological parameters on sleep. While such data would be useful to have, self-reports and interview-based measures remain the most widely used measures in community surveys.

In this study, almost all GSD subjects also reported at least one symptom of insomnia (95.4%) but it resulted in an insomnia diagnosis for only 37.1% of them. The other GSD subjects had either a mental disorder diagnosis (37%) or another diagnosis of sleep disorder (25.7%). Compared to subjects who reported only insomnia symptoms (i.e., without GSD), GSD subjects were 8 times more likely to have a diagnosis of sleep or mental disorder. This confirms that sleep dissatisfaction is a better indicator of an underlying pathology than insomnia symptoms alone (1).

Furthermore, GSD subjects were having insomnia symptoms since a longer time than insomnia subjects without GSD. This finding was not due to age: a 20-month difference was observed between the GSD and insomnia subjects without GSD in all age groups. A multifactorial approach was taken to determine what factors increase the risk of being GSD. These factors concern sociodemographic descriptors, environmental factors, life habits, health status, psychological factors, and sleep/wake factors that may impair the sleep quantity and quality.

Several studies have pointed out that women are more likely than men to have insomnia or to be dissatisfied with sleep (2,4,10). This was also found in this study in bivariate analyses and in the model that analyzed only sociodemographic factors. When all associated factors are analyzed in a single model, being a woman is no longer significant. Therefore, the impact of gender was reduced when we adjusted for other factors and was no longer statistically significant. Advancing in age, however, remains a strong factor associated with GSD in all the models. It is very likely also that variables significantly associated with GSD change with age or, at least, that some factors become more important than others.

For example, in younger subjects, life stress would probably play a more important role than in the elderly and conversely, physical diseases are probably a more meaningful factor in elderly subjects. Environmental factors were rarely investigated in epidemiological studies in the general population. A study with male industrial workers (6) found that limited-space bedroom was related to insomnia. Another study found that dissatisfaction with the bedroom environment was related to insomnia (20). In this study, we found that sleeping in a too hot bedroom was associated with GSD. Among life habits, we found that doing physical activities at least three days per week, for a period of at least 15 minutes, in the two hours before bedtime appeared to be a protective factor for GSD. This finding may come as a surprise since a common belief is that exercise close to bedtime disturbs the sleep.

However, two recent studies have reached a different conclusion. In the elderly, light physical activities and structured social intervention (in the morning and in the evening) increased slow-wave sleep and improved memory (7). In healthy, fit males, vigorous exercise shortly before bedtime did not disturb sleep (21). Also, studying or working in bed was related to sleep dissatisfaction. This habit increases the level of vigilance at a period when it should be reduced and causes a hyperarousal state that may delay the sleep onset.

Physical diseases, especially upper airway disease and musculo-skeletal disease, were related to GSD. This finding was not surprising. Diseases that cause pain during sleep (e.g., arthritis, back pain) or impair the normal breathing (e.g., asthma, chronic bronchitis) are well known to be associated with insomnia symptoms, especially with disrupted sleep (8,9).

The association between sleep complaints and psychological factors was demonstrated in many clinical and epidemiological studies. Of particular interest in this study was that life stress is a stronger factor associated with GSD than other psychological variables, including depressive mood and anxiety symptoms, when all significant variables were introduced into the same model. This indicates that experienced level-of-life stress makes an independent contribution to GSD, while anxiety and depressive symptoms have multiple interactions with other symptoms that are more important in the prediction of GSD.

We also found that our GSD subjects more frequently reported daytime sleepiness. A possible consequence is a decreased vigilance that put these subjects at higher risk to have home, work or road accidents (22,23). In this study, we found that the GSD drivers were two times more likely to have had road accidents in the previous year than insomnia subjects without GSD.

Finally, the results of this study confirm that GSD subjects are more actively seeking help from their physicians than insomnia subjects without GSD. Moreover, they are five times more likely to use a sleep medication. On the other hand, this study also illustrates the importance of using a multifactorial approach when assessing sleep complaints. This approach provides valuable information with regards to the possible causes of the complaint, and it gives indications on what aspects the treatment should be focused on.

REFERENCES