Global Sleep Dissatisfaction and its links with Insomnia

The term "insomnia" is often felt to be inadequate because it is used to cover a variety of sleep problems, including abnormal sleep latency, difficulty maintaining sleep, non-restorative sleep, and early-morning awakening with the inability to go back to sleep.

Specialists usually classify insomnia complaints according to these four "indicators of insomnia", and subjects participating in therapeutic trials for the investigation of a new hypnotic, for example, are usually chosen using these classifications, which are based on the patient's subjective report of sleep quality.

The assumption is that these complaints indicate a major impairment in quality of life for the affected individuals, and that treatment is needed.

In the 70's and 80's, sleep researchers demonstrated that subjective, sleep-related complaints do not always correlate with polygraphic findings. Very little information exists, however, on the relationship between these complaints in the general population and self-assessed satisfaction with nocturnal sleep.

The present epidemiological study investigated the sleep habits and disorders of the French-speaking population in Montreal, Quebec.

We questioned:

  • how frequently these indicators of insomnia were reported in the general population,
  • how useful these indicators were in identifying people dissatisfied with the quality of their sleep,
  • what the relationship was between these indicators and the intake of sleep-enhancing drugs, and
  • whether socio-economic factors played a role in how subjects viewed their sleep.



In 1993, an epidemiological study of sleep habits and sleep disorders was conducted by telephone in the metropolitan area of Montreal (Quebec, Canada). The target population was the French speaking residents 15 years of age and older, representing approximately 1,761,121 inhabitants.

A representative sample was obtained using a stratified probability approach with a sample selection based upon the geographic distribution according to the results of the 1991 census survey of the area using the quota method and according to age and gender.

The Kish selection method was applied to select the subject to interview in the household. Briefly, the Kish method is based on the use of 8 selection tables randomly assigned among households. Each table offers a predetermined proportion of family attributes. Of 2,117 subjects who were solicited, 1,722 interviews (81.3%) were completed.

Subjects who did not speak French or suffered from a hearing or speech impairment, or who had a sickness that prevented them from participating in the interview were excluded. Subjects who did not want to participate at the first phone call were telephoned a second time. Subjects who refused to participate were classified as "refusal". Any selected phone number with no answer was dialled again at least ten times at different evening hours and on different days, including both weekdays and week-ends, before being replaced due to lack of response.



Telephone interviews were conducted by 12 interviewers inexperienced in psychiatric assessment but specially trained to use the "Sleep-Eval" Knowledge Based System.

This system is a validated non-monotonic, level-2 expert system with a causal reasoning mode and serves to manage epidemiological surveys and to administer questionnaires ensuring homogeneous and standardised evaluations. Comparisons between diagnoses obtained by lay interviewers using the Sleep-Eval system and clinicial psychologists shown a good agreement (kappa 0.85 for recognition of sleep problems and 0.70 for insomnia disorders) between the interviews (unpublished data).

Briefly, an expert system is a software that exhibits a degree of expertise in problem solving (here the attribution of sleep and mental disorders diagnoses) comparable to that of a human expert. The system managed the epidemiological study and administered the questionnaire via three primary "modules":


The interviewer entered the number of the Kish table and the gender and age of each elligible individual in the household. At the end of this data capture, the system indicated the household member to be interviewed.


The system began with a series of standard questions which were applied to the whole sample. This standard set covered sociodemographic information and sleeping habits. Subsequently, the interview was adapted depending on the answers provided by the subject on a series of key questions. Positive or negative answers to these questions fired a reasoning process similar to the differential diagnosis applied by a psychiatrist.

To perform this operation, the system formulated a series of diagnostic hypotheses (causal reasoning process) that it attempted to confirm or reject through further questioning or deductions (non-monotonic, level-2 feature). The system ended the interview once all the diagnostic possibilities were exhausted. Co-occurrence of two diagnoses was allowed only when respecting the conditions described in the DSM-IV classification. This reasoning process enabled the system to economise the interview by pre-emptively eliminating superfluous questions that might otherwise be put to a given subject.

For respondents without sleep and psychiatric problems, the interview was brief (10 to 15 minutes). Conversely, for subjects who presented with a history of sleep or mental disorder(s), the interview was designed to be far more comprehensive, and lasted up to 90 minutes in some cases.

The program formulated questions, where the interviewer was required to glance at the monitor as the prompts appeared and to read them aloud to the interviewee. Additional samples and instructions on how to answer a question were provided to the interviewers. The questions were either closed-ended (e.g., yes/no, 5-point scale, multiple choice) or open-ended (e.g., duration of symptom, description of illness).


The classification included:
a) subjects with sleep problems,
b) subjects without sleep problems,
c) subjects who refused the interview,
d) telephone numbers called more than ten times unsuccessfully, and
e) telephone numbers to be renewed.

Once a file was definitively classified, it was impossible for the interviewer to gain access to it. Ten percent of the studied population was called again the following day to verify the responses provided.


Responses to questions related to sleep complaints were analyzed in relation to the following major categories explored during the investigation: sociodemographic information; current medication intake; medical consultations and treatments during the past 12 months; sleep/wake schedule for work-days, week-ends or days off, and vacation periods; evening activities just preceding bedtime; quality of nocturnal sleep or longest sleep period; assessment of social life; and DSM-IV mental disorders10.


Subjects were classified as "Dissatisfied with Quality of Sleep" (DQS) if they reported dissatisfaction with their sleep or the intake of sleep-enhancing medications at the time of the study. They were classified as "Satisfied with Quality of Sleep" (SQS) if they reported satisfaction with their sleep and did not take sleep-enhancing drugs.

The indicators of insomnia were defined as follows:

  1. Difficulty Initiating Sleep (DIS)
    Strong dissatisfaction with "long sleep latency" at sleep onset (i.e. a sleep latency longer than 30 minutes identified by the subject as a major problem .)
  2. Difficulty Maintaining Sleep (DMS)
    Nocturnal awakenings with great difficulty or the inability to fall back to sleep, or identification of difficulty maintaining sleep as a major problem.
  3. Non-restorative Sleep (NRS)
    Sleep of normal duration but associated with a complaint of tiredness at awakening, lack of normal, rested feeling after nocturnal sleep, or inability to get going in the morning.
  4. Early Morning Awakening (EMA)
    Complaint of short nocturnal sleep duration due to abnormal awakening before 5:00 am (or at least 1-1/2 hours prior to the desired wake-up time) without the possibility of going back to sleep or identification of early morning awakening as a major problem.
The definition of the term "insomnia" was also based on the definition from the Diagnostic and Statistical Manuals of mental Disorders - DSM-IV and DSM-III-R and included the requirement of at least one month's duration of the complaint at the time of the interview. The two groups (SQS and DQS) were further divided into satisfied or dissatisfied with sleep plus "indicators of insomnia" (SQS+I and DQS+I respectively), and satisfied or dissatisfied with sleep "without indicators of insomnia" (SQS-I and DQS-I), based on the presence or absence of one or more indicators of insomnia. Consequently, four major subgroups were identified: DSQ+I, DSQ-I, SQS+I, and SQS-I. These 4 subgroups were further analysed by taking into consideration the number and type(s) of insomnia indicator(s).


Data from the 1991 Canadian census of the non-institutionalised population aged 15 years and older living in the metropolitan area of Montreal were used to represent the standard population. To compensate for potential bias (from such factors as an uneven response rate in different demographic groups or the absence of telephones in a fraction of households), a weighting procedure was used to adjust the sample.

The standard population and the sample were distributed within 12 demographic cells (6 age groups (x) gender). For each demographic cell, a weight was derived from the following equation: Wi=nexpected/nobserved, where the nexpected represents the number of subjects expected in a sample of 1722 subjects for a given age group and gender, and the nobserved represents the number obtained in the sample for this cell.

Results are presented with weighted percentages and 95% confidence intervals (95% C.I.). Bivariate analyses were carried out using the Chi Square statistics. The Bonferroni corrected significance for p-values was also applied. Predictive variables of indicators of insomnia were determined using logistic regression. Models were calculated using SQS+I, SQS-I, DQS+I or DQS-I as the dependent variables.

The INDICATOR Contrasts Method was used to determine which categories of independent variables were significantly associated with a given subgroup. For each variable considered, the reference category was the one that contributed least to the presence of the considered subgroup.

Significant Odds Ratios (OR) were calculated according to these different categories. The entry level was set at 15%. The SPSS statistical computerised package, version 6.1 for the Macintosh, was used.



The unweighted sample included 955 women and 768 men with an age ranging from 15 to 100 years.

Subjects "dissatisfied with their sleep", independent of the presence or absence of "indicators of insomnia" (i.e., DQS+I and DQS-I) represented 17.8% of the sample: 15.9% (95% C.I.: 13.4% to 18.4%) of men and 19.5% (16.9% to 22.1%) of women.

Eleven percent (11.2%) of the subjects were classified as DQS+I and 6.5% as DQS-I. Eighty percent of the subjects reporting ingestion of sleep-enhancing medication complained also of "indicators of insomnia".


of subjects classified as DQS+I, DQS-I and SQS+I across age groups and gender were fairly close, as shown in table 1. There were, however, significantly more DQS+I (11.6%) than DQS-I (4.9%) among subjects between 45 to 64 years old and among subjects older than 64 (15.1% vs. 5.7%; p=0.05).

Investigations of marital status and educational level (table 2) indicated that more DQS+I women were separated, divorced or widowed compared to other woman groups. On the other hand, DQS-I women were more often married than other groups. While SQS+I and DQS+I had comparable educational levels, more DQS-I subjects had higher levels of education. The significant difference associated with educational level was not reflected in the annual income calculation nor in the employment status.

About eleven percent (11.3%) of employed subjects were shift workers, but the difference was not significant between groups (SQS-I= 10.9%, SQS+I= 12%, DQS-I= 6.2%, p=0.13).


Distribution of subjects displayed little variability across categories in terms of usual bedtime, time of morning awakenings and number of additional hours spent sleeping on weekends or days off, except that the SQS+I subjects woke up earlier in the morning (<6 a.m., 51.1%) compared to other groups (38.0% of DQS+I; 30.9% of DQS-I and 28.5% of SQS-I; p<0.001).

A significantly higher number of DQS+I women and SQS+I subjects (independent of gender) had a total sleep time of less than 7 hours as compared with DQS-I and SQS-I subjects (table 3). As expected, the DQS+I group had a larger proportion of subjects dissatisfied with their sleep duration. Similarly, in DQS+I and DQS-I groups, "light sleep" was a concern for over half and over a third of each group, respectively (table 3).

About 25% of the population reported at least one awakening in the middle of the night at least three nights per week, but only 10.8% of the population considered these awakenings a problem. Compared to the SQS-I and SQS+I subjects, among whom awakenings occurred at rates of 18% and 48.6% respectively, DQS+I subjects were significantly different, with 60.8% of the group reporting awakenings (p<0.0001). Of the DQS+I subjects who complained of awakenings, a higher proportion reported waking up three or more times per night (39.7%) than SQS-I subjects (14.8%) and SQS+I (13.8%; p<0.0001).


Only one "indicator of insomnia" was reported by 14% of the studied population and two indicators or more by 7.7%. As shown in Table 4, "Difficulty initiating sleep" (DIS) and "Difficulty maintaining sleep" (DMS) were nearly evenly reported in the sample and were more frequent in women than in men.

"Non-restorative sleep" (NRS) was mentioned by 4.1% of the total sample while "early morning awakenings" (EMA) was reported by 8.1%. DIS, DMS and EMA were more frequently mentioned in DQS+I than in SQS+I subjects. DIS and EMA were reported with similar frequencies among men in the DQS+I and SQS+I groups. NRS was mentioned in nearly equal proportions among the different groups.

The number of indicators of insomnia was strongly associated with sleep satisfaction. Subjects with only one indicator were more likely to be satisfied with their sleep (SQS+I: 70% vs. DQS+I: 30%) while those with two indicators or more were mostly dissatisfied (DQS+I: 89.2% vs. SQS+I: 10.8%; p<0.0001).


The intake of sleep-enhancing medication was reported by 3.8% (2.9% to 4.7%) of the sample. There were no significant gender differences (men: 2.8% [1.7% to 3.9%]; women: 4.7% [3.3% to 6.1%]).

The drugs used fell into four categories: hypnotics: 21.5%; anxiolytics: 52.3%; other psychotropic drugs (mostly antidepressants): 7.7%; and other nonpsychotropic drugs: 18.5%, according to the classification of the Canadian Pharmaceutical Association. Nearly 76% of consumers had taken psychotropic drugs for at least one year. Most had received their prescription from a general practitioner (77.4%).

Three percent (2.2% to 3.8%) of the sample reported current use of anxiety-reducing medications. In that subgroup, 29.7% had no sleep complaints, and women were significantly more represented than men (4.3% [3% to 5.6%] of all women compared to 1.6% [0.7% to 2.5%; p<0.05] of all men).

Of the drugs used, 53.8% were anxiolytics, 26.9% other psychotropic drugs, and 19.2% other non-psychotropic drugs (More extensive results regarding drug consumption in this group are presented elsewhere16.)

Of the "Insomnia disorder" subjects, 55.3% reported daily use of coffee, 10% of alcohol, and 28.1% of tobacco. Of those who consumed coffee, 38% drank more than 2 cups of coffee per day and of those who consumed alcohol, 35% had more than 3 alcoholic beverages per day.


Subjects were presented with a list of activities and asked to indicate those they usually engaged in before bedtime.

The large majority of SQS (+I and -I) subjects reported watching television (over 70%), compared with 58% of DQS+I subjects and 19% of DQS-I subjects (p<0.0001). Reading was also mentioned by about half of SQS (+I and -I) and DQS+I subjects, but by only 15.8% of DQS-I subjects (p<0.0001).

Interestingly, 56.3% of DQS-I subjects, who did not watch TV or read, indicated performing demanding intellectual activities prior to going to bed (studying, filing papers, etc.) compared to 14.9%, 21.2% and 24% of SQS-I, SQS+I and DQS+I subjects (p<0.0001).

The percentage of each group having a snack before bedtime was similar (about 40% in each group).


Table 3 outlines the results obtained. The DQS-I group had the greatest proportion of subjects dissatisfied with socio-familial interactions (friends, spouse, and children) and their overall social life (Table 3). SQS+I and SQS-I subjects were mostly satisfied with socio-familial interactions. Both DQS-I and DQS+I reported less frequent social contacts than SQS-I subjects.


Multivariate analyses were performed using a stepwise logistic regression.

Since the risk of interrelated variables increases with multiple measures, we first checked for the presence of collinear effects between variables.

Not surprisingly, such a problem was found between age groups and employment status; employment status was removed because age is expected to have a greater clinical relevance. Also, by suppressing the "time of morning awakening" variable, problems of colinearity between the variables assessing sleep/wake schedule were avoided. Subsequently, significant variables were introduced into the models; gender and age, though not significant variables, were "forced" into the model due to their clinical relevance.


The following were predictive factors associated with DQS:

  • dissatisfaction with sleep duration (OR: 24.2 [13.1-44.7]);
  • aged 15-44 years (OR:1.5 [1.0-2.3]);
  • dissatisfaction with sociofamilial life (OR: 10.9 [7.4-14.8]);
  • being a light sleeper (OR: 5.8 [4.0-8.4]);
  • reporting never or rarely dreaming (OR: 1.7 [1.2-2.4]);
  • having previously used a sleep enhancing medication (OR: 1.6 [1.0-2.7]); and
  • having been treated for a physical illness (OR: 1.5 [1.0-2.3]).

When compared to SQS+I, the following factors were significantly associated with DQS+I:

  • reporting two or more insomnia indicators (OR: 35.9 [35.1-36.8]);
  • dissatisfaction with sociofamilial life (OR: 13.7 [12.9-14.5]);
  • dissatisfaction with sleep duration (OR: 9.2 [8.1-10.3]); aged 15-45 years (OR:2.7 [1.9-3.5]);
  • reporting never or rarely dreaming (OR: 2.5 [1.8-3.1]);
  • going to sleep after 11 P.M. (OR: 2.4 [1.6-3.1]); and
  • being separated, divorced, or widowed (OR: 2.3 [1.4-3.3]).

Sleeping less than 7 hours was a higher risk for SQS+I subjects (OR: 4.0[1.5-10.7]).


When compared to DQS-I, the following factors were significantly associated with DQS+I:

  • having previously used a sleep enhancing medication (OR: 4.5 [1.4-14.3]),
  • being dissatisfied with sleep duration (OR: 4.4 [2.0-10.0]), and
  • being treated for a physical illness (OR: 2.1 [1.0-4.4]).

Being a light sleeper (OR: 1.7 [1.0-3.1]) and being dissatisfied with sociofamilial life (OR: 4.2 [2.3-7.5]) were higher risks for DQS-I.


Determination of psychiatric disorders was based upon DSM-IV classification. Prior consultations for anxiety symptoms were reported by 30.1% of DQS+I subjects, and prior consultations for depressive thoughts by 23% of them.

The most frequent Mood Disorder, was a current episode of "Major Depressive Disorder" (i.e., 296.2x and 296.3x). It was present in 7% of the DSQ+I subjects. The most frequent Anxiety disorder was "Generalized Anxiety Disorder" (300.02) (19% of DSQ+I) followed by "Panic disorder" (300.1 and 300.32) (11% of DSQ+I).

Twenty percent of the DSQ+I subjects responded to the criteria for "Primary Insomnia" (307.42). Nineteen percent of DSQ+I subjects did not fit the DSM-IV criteria for either a mental or a sleep ("primary insomnia") disorder.


Physician visits that had occurred in the six months prior to the survey were common (58.7% of the population). Interestingly, a significantly higher percentage of DQS+I women (79.6%) reported having consulted a physician than other woman groups (DQS-I: 68.1%; SQS+I: 70.8%; SQS-I: 63.2%; p< 0.005). There was no difference when men were investigated (DQS+I: 62.2%; DQS-I: 44.6%; SQS + I: 50.9%; SQS-I: 49.2%; p=0.156).

Eighty-one percent of the consultations were with a general practitioner. Physical illnesses were more commonly reported by DQS+I subjects (30%) than by subjects in the three other groups (DQS-I: 13.9%, SQS+I: 18.6, SQS-I: 13.4%; p < 0.0001). There were no significant gender differences.


Many western countries are attempting to slow the rising cost of health care. Sleep complaints are frequent and may lead to or be the consequence of serious health problems. As such, they are associated with visits to a physician and the use of pharmacological agents, and clearly have an economic impact. An emphasis on prevention may help reduce healthcare expenses, but prevention requires a good understanding of the extent of the problem in the general population. General surveys and self-report questionnaires are important for attaining this sort of understanding.

Our investigation shows that dissatisfaction with sleep is not uncommon in the general population. Our study agrees with previous work demonstrating a greater degree of dissatisfaction among the elderly, women, and the socially isolated (divorced, widowed, separated).

If we consider previous reports, the results from the "Santé Québec" survey in 1987 are the most similar to ours.For example, 12.9% of the Quebec population complained of having problems, often or very often, initiating or maintaining sleep (compared to 10.8% in our population). By comparison, 37.2% of the subjects surveyed by Bixler et al. had difficulty either initiating sleep (14.4%), or maintaining sleep (22.9%). These results, however, do not include any evaluation of the severity of the problems. Mellinger et al., reported that 17% of their subjects had problems initiating sleep or difficulty maintaining sleep, but the time period investigated (the previous 12 months) was much longer than our own.

With regard to drug intake, our findings are more similar to studies performed in the United States, such as Mellinger et al., than those in France(23,26). Sleep-enhancing medications were used by 3.8% of our population. "Psychotropic" drugs (hypnotics, anxiolytics or antidepressants) were used by 3.1% and other drugs by 0.7% of the population. Mellinger et al. found that 4.3% of the adult population in the US had taken hypnotics, anxiolytics or antidepressants for sleep complaints. In comparison, Ohayon and Quera-Salva et al. reported hypnotic use in over 10% of the adult French population.

These differences in drug intake may represent differences in the medical education and prescription habits of these two countries. Montreal is a North American city, and prescription habits are probably closer to those of the United States than to those of Western Europe.

Interestingly, once sleep-enhancing medications were prescribed, they were very likely to be taken chronically. Seventy-six percent of the medication users in our study had taken their drugs for over one year, despite recent recommendations against long-term use by official agencies and pharmaceutical companies and efforts to encourage re-evaluation of insomniacs who are taking medication. This finding is similar to that reported in other countries.

It is noteworthy that anxiolytics were more commonly taken (2% of the total population) than hypnotics (0.8% of the population). This may reflect the negative publicity that certain hypnotics have received in recent years and indicates a displacement, but not a disappearance, of the prescription of hypnotics. In comparison, Mellinger et al.21 found prescription of hypnotics in 2.6% of the population in the mid 1980s. When we looked at our subjects according to our four main categories, those dissatisfied with sleep and with indicators of insomnia (DQS+I) reported a significantly higher number of physical illnesses than other subjects at the time of the interview.

Whether this was a cause or a consequence of the sleep disorder and sleep complaint is unknown, but several surveys27,28 have indicated an association between sleep complaints and increased mortality rate within a specific follow-up time period.

Another important finding of our study is that the perception an individual has of his or her sleep quality is not necessarily based on the classic "indicators of insomnia". The DSM-III-R, DSM-IV and ICSD-90 all list non-restorative sleep as a symptom of insomnia, but in our study, non-restorative sleep was reported by nearly the same percentage of subjects satisfied and dissatisfied with their sleep. This may be a result of the sleep deprivation many people inflict on themselves during the work week. The report of non-restorative sleep should lead to further investigation, but should not be taken immediately as an indicator of insomnia.

Ford and Kamerow's study reported a prevalence of 10.2% for "insomnia disorder", according to DSM-III criteria. We applied DSM-III-R and DSM-IV criteria for "insomnia disorder" to our population. Five percent of the population met DSM-III-R and 7% met DSM-IV criteria. These diagnoses were rendered in strict accordance with the classifications.

Furthermore, we found that our "insomnia disorder" subjects, using the above classifications, would meet the criteria for mental disorders (major depression or anxiety disorders) in about 69% of the subgroup using DSM-III-R and in 64% of the subgroup using DSM-IV criteria. These percentages are twice as high as the percentage (37.9%) found by Ford and Kamerow using DSM-III criteria. It must be emphasised that 19% of subjects classified as DSQ+I could not be classified with a mental and/or sleep disorder using DSM-IV criteria, some of them received prescription medication from their general practitioners who are the only physicians contacted in over 80% of the cases. This is a failure of the classification.

Existing classifications are unclear about the presence of mild insomnia "indicators" (or "symptoms"). the International Classification of Sleep Disorders (ICSD-90) describes "mild" insomnia as "an almost nightly complaint of an insufficient amount of sleep, or not feeling rested after the habitual sleep episode accompanied by little or no evidence of impairment of social or occupational functioning".

DSM-III-R defines "non-restorative sleep" as " a feeling that the sleep is restless with an apparently normal sleep duration" and the DSM-IV classification extends this definition to "a feeling that the sleep is restless, light or of poor quality".

It is difficult to fit our group of subjects who were dissatisfied with their sleep but without indicators of insomnia (DQS-I) into any of these classifications. The group, however, is well defined, quite large, and clearly different from the DQS+I and SQS subjects. It includes subjects who tend to be younger (83.1% under 55 years of age), married (67.8%) with a higher level of educational (61.1%).

Over 80% of this group expressed dissatisfaction with most areas of their social life (relations with their spouse, children, friends, and overall social life), and 56.5% reported "light" sleep. The fact that approximately 60% of this group, on a regular basis, engaged in some form of mental activity near bedtime suggests that poor sleep hygiene may be involved. These activities (studying, filing papers, etc.) may induce a hyper-alertness not conducive to sleep onset. Clearly, this group of subjects needs more investigation and should be considered in any revised classification system.

We must also understand the opposite group, i.e., subjects with indicators of insomnia who are satisfied with their sleep (SQS+I). Clearly, there are great differences between subjects in their perceptions of good and poor sleep. A better understanding of sleep complaints and more appropriate classifications will help physicians meet the needs of patients with insomnia, and perhaps lead to better preventive measures that can help reduce the individual and socio-economic costs of this common problem.


Content of this page is extracted from:
Ohayon MM, Caulet M, Guilleminault C. How a general population perceives its sleep, and how this relates to the complaint of insomnia. Sleep 1997;20:715