Ohayon MM, Vecchierini M. Daytime Sleepiness and Cognitive Impairment in the Elderly Population. Arch Intern Med. 2002;162(2):201-208. doi:10.1001/archinte.162.2.201
Recent findings suggest that there may be a relationship between excessive daytime sleepiness (EDS) and cognitive deficits. This study aims to determine to what extent EDS is predictive of cognitive impairment in an elderly population.
A total of 1026 individuals 60 years or older representative of the general population living in the metropolitan area of Paris, France, were interviewed by telephone using the Sleep-EVAL expert system. To find these individuals, 7010 randomly selected households were called: 1269 had at least 1 household member in this age range (participation rate, 80.9%). In addition to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, and International Classification of Sleep Disorders diagnoses, the system administered to participants the Psychological General Well-being Schedule, the Cognitive Difficulties Scale (MacNair-R), and an independent living scale.
Excessive daytime sleepiness was reported by 13.6% of the sample, with no significant difference among age groups. Compared with nonsleepy participants, those with EDS were at increased risk of cognitive impairment on all the dimensions of the MacNair-R scale after controlling for age, sex, physical activity, occupation, organic diseases, use of sleep or anxiety medication, sleep duration, and psychological well-being. The odd ratios were 2.1 for attention-concentration deficits, 1.7 for praxis, 2.0 for delayed recall, 2.5 for difficulties in orientation for persons, 2.2 for difficulties in temporal orientation, and 1.8 for prospective memory.
Among elderly individuals in the general population, EDS is an important risk factor for cognitive impairment. A complaint of EDS by an elderly patient should signal the possibility of an underlying cognitive impairment in need of evaluation.
IN INDUSTRIALIZED countries, the number of elderly people continues to grow. Health care systems must be prepared to support this fast-growing part of the population, whose needs are specific: the elderly population is vulnerable to a high occurrence of organic diseases and to cognitive impairments that will affect quality of life. The prevalence of mild to severe cognitive deficits is 4% to 10% in the elderly population living in the community.1- 3 Individuals with severe cognitive disorders related to dementia are rarely found in the community because they are quickly losing the autonomy necessary to live in the community. Several longitudinal studies4- 8 have shown an increased risk of mortality in nondemented elderly individuals with cognitive impairments. The adjusted (for age, sex, and health status) relative risk for mortality was 1.7 to 3.6 times higher in elderly individuals with mild to severe cognitive impairments.4- 12 Some longitudinal studies11,13,14 showed that the mortality rate was not statistically different in any age group in individuals with good cognitive performance and that the preservation of cognitive functions was associated with better survival in older individuals.
Decline in cognitive performance has been associated with several factors, including neurological diseases, vascular diseases, depression,15- 17 and diabetes mellitus,18 but not always.19 However, other factors significantly accounted for this decline in the aging process, namely, educational level20,21 and social disengagement.22 More recently, excessive daytime sleepiness (EDS) has been associated with poor cognition and dementia.23 Moreover, an increased mortality risk of 1.73 was found in elderly individuals with cognitive impairments who nap most of the time.24
This study aimed to verify that EDS is an independent predictive factor of cognitive impairment in an elderly population living in the community after controlling for possible confounders, such as obstructive sleep apnea syndrome, organic diseases, anxiety, and depression.
An epidemiological survey of sleep habits and sleep disorders was conducted by telephone in the metropolitan area of Paris, France, between October 13, 1999, and April 28, 2000. The targeted population consisted of noninstitutionalized individuals 60 years or older. This age range represented approximately 4.5 million inhabitants in the metropolitan area of Paris. The sample was drawn according to a 2-stage procedure. In the first stage, official census data were used to divide the population according to its geographical distribution in Paris. Subsequently, telephone numbers were randomly pulled according to this stratification. In the second stage, the Kish method,25 a controlled selection method, was applied to maintain the representation of the sample according to age and sex.
To find the participants, 11 650 telephone numbers were used; 7010 corresponded to households, and 4640 either were not a household or were not in service. Of the households, 1269 included at least 1 resident 60 years or older, and 1026 individuals agreed to be interviewed. The participation rate (80.9%) was calculated based on the number of completed interviews (n = 1026) divided by the number of eligible telephone numbers, which included all residential numbers not meeting any of the exclusion criteria (N = 1269).
Overall, 10 381 telephone numbers were rejected mainly because (1) the household did not include an individual 60 years or older (51.9%); (2) the telephone number did not correspond to a household (ie, it was a business or fax number) (14.8%); (3) the telephone number was not in service (24.8%); (4) the eligible respondent was ill, was deaf, or had a speech impairment (1.8%); or (5) we could not communicate in French with the person who answered the telephone (1.6%).
Interviewers explained the goals of the study to potential participants before requesting verbal consent. Excluded from the study were individuals younger than 60 years, those who did not speak sufficient French, and those who had a hearing or speech impairment or an illness that precluded interview. The ethical committee of the Xavier Bichat Hospital in Paris approved the study.
Individuals who declined to participate were telephoned a second time at least 3 weeks later and were asked again if they were willing to be part of the study. Individuals were classified as refusers if they declined a second time or when it was impossible to reach them a second time. Telephone 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 telephone number by 1, was used to control for unlisted telephone numbers.26 The final sample consisted of 15.9% unlisted numbers.
Interviews were conducted by telephone using the Sleep-EVAL system. The study was conducted at the Xavier Bichat Faculty of Medicine of the Paris 7 University. Interviews were performed by 16 university students who were inexperienced in psychiatric assessment but who received special training on how to use the Sleep-EVAL system. The average (SD) duration of the interviews was 64.3 (25.6) minutes. An average of 5 telephone calls were made to complete an interview. The team of interviewers was monitored daily by one supervisor to ensure that questions were asked correctly and that data were entered properly.
The Sleep-EVAL system was specifically designed to administer questionnaires and conduct epidemiological studies on mental and sleep disorders in the general population.27- 29 It managed the telephone calls (generation of new numbers, management of appointments, and management of numbers that had to be called back) and the Kish selection procedure. It also kept track of all telephone calls made (date and time, interviewer who made the call, issue of the call, duration of the call, elapsed time between each call, number of questions asked during the call, and number of times the number was dialed).
The Sleep-EVAL system includes a nonmonotonic, level-2 inference engine endowed with a causal reasoning mode. These features enable the Sleep-EVAL system to formulate a series of diagnostic hypotheses based on the responses provided by a participant (causal reasoning). The nonmonotonic, level-2 inference engine examines these hypotheses and confirms or rejects them through further questions and deductions. Two classifications are implemented in the knowledge base of Sleep-EVAL: the International Classification of Sleep Disorders30 and the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV).31
The system formulates initial diagnostic hypotheses on the basis of responses to a standard set of questions posed to all participants. Concurrent mental diagnoses are allowed in accordance with the DSM-IV. The system terminates the interview once all International Classification of Sleep Disorders and DSM-IV diagnostic possibilities are exhausted. The system selects and phrases the questions to be administered and provides examples and instructions on how to ask them. The interviewer simply reads the questions as they appear on the monitor and enters the responses. Questions can be close ended (eg, yes-no, present-absent-unknown, or 5-point scale) or open ended (eg, name of illness or duration of illness).
The system has been validated in various contexts and has been demonstrated to be reliable and valid. Five validation studies have been conducted.27,32,33
The standard questionnaire of the Sleep-EVAL system covered (1) sociodemographic information; (2) the sleep-wake schedule; (3) symptoms of sleep disorders; (4) sleep hygiene; (5) current and past consumption of alcohol, tobacco, and coffee; (6) current and past use of medication for sleep, anxiety, and depression; (7) any other medication use; (8) medical information, including organic diseases, hospitalizations, medical consultations, and blood pressure; (9) height and weight; and (10) DSM-IV and International Classification of Sleep Disorders diagnoses.
For the purposes of this study, validated assessment scales were added to the knowledge base of the Sleep-EVAL system:
Cognitive Difficulties Scale (MacNair-R) (26-item French version).34 This scale assesses 6 dimensions of cognitive difficulties: attention-concentration deficits, praxis, delayed recall, difficulties in orientation for persons, difficulties in temporal orientation, and difficulties in prospective memory. Higher scores correspond to greater cognitive difficulties.
Psychological General Well-being Schedule.35,36 This index measures subjective well-being. The individual self-reports on 22 items that are indicators of 6 affective states: anxiety, depressed mood, sense of positive well-being, self-control, general health, and vitality. Low scores on each of these states indicate impairment.
Mini-Mental State Examination.37,38 This evaluation scale assesses cognitive efficiency. Questions about orientation and memory were asked. Praxis could not be assessed because it requires a face-to-face meeting.
Functional Assessment. Instrumental activities of daily living39 that measures the older person's abilities to perform simple everyday tasks.
The recruited sample matched the official census data for the metropolitan area of Paris in terms of age and sex distribution. Therefore, it was not necessary to apply a weighting procedure that would correct for evident disparities.
Bivariate analyses were performed using the χ2 test with Yates correction or the Fisher exact test when n values were smaller than 5. Reported differences were significant at P≤.05. Analysis of variance and independent samples t test were used to analyze continuous variables. When basic assumptions for the use of these statistical methods were violated, nonparametric tests were also performed (Kruskal-Wallis and Mann-Whitney tests). Logistic regressions40 were used to compute the odds ratios associated with each type of cognitive difficulty. Colinearity between variables (ie, information redundancy) was verified beforehand.
Overall, 1026 individuals participated in the study: 28.7% were aged 60 to 64 years; 25.4% were aged 65 to 69 years; 21.2% were aged 70 to 74 years; and 24.7% were 75 years and older. Women represented 59.8% of the sample. The proportion of women increased with age: 51.6% of participants aged 60 to 69 years were women, and 66.9% of those 75 years and older were women. More than half of the participants (52.0%) aged 60 to 64 years were active (ie, working or doing an activity at least 3 days per week). This rate was 43.2% in those aged 65 to 69 years, 40.0% in those aged 70 to 74 years, and 28.2% in those 75 years and older.
Overall, the sleep-wake schedules of the different age groups were comparable (Table 1). The only significant difference was observed in bedtime between individuals 75 years and older and those aged 70 to 74 years, the latter being in bed with the intention of sleeping 35 minutes later than the older participants (Table 1).
Generally speaking, men and women had comparable scores on all dimensions of the Cognitive Difficulties Scale; only praxis was higher in women than in men, indicating a deterioration in manual dexterity and efficiency of movement in women. Similarly, age groups had comparable scores on all the dimensions of the Cognitive Difficulties Scale except praxis, which was higher in the oldest participants compared with the other age groups (Table 2).
Overall, 5.3% of participants claimed that they fell asleep easily during the day and almost everywhere at least weekly. This rate was comparable among the age groups.
Almost 6% of participants reported feeling moderately sleepy during the day, and 5.2% said they felt sleepy a lot during the day. Again, there were no significant differences among the age groups. Therefore, 13.6% of the sample reported having daytime sleepiness.
Individuals with daytime sleepiness had significantly higher scores on all the dimensions of the Cognitive Difficulties Scale (Table 3). These differences were not present in all age groups: for participants 75 years and older, the 6 dimensions of the MacNair-R scale were all significantly higher in individuals with daytime sleepiness compared with those without sleepiness. In individuals aged 60 to 64 years and those aged 70 to 74 years, only praxis and prospective memory did not differ significantly between individuals with and without daytime sleepiness. Individuals aged 65 to 69 years with daytime sleepiness differed significantly from those without daytime sleepiness only on difficulties in temporal orientation (P = .02).
The proportion of participants taking a nap at least 2 days per week was 30.1%. The percentage of nappers increased with age: 22.4% of individuals aged 60 to 64 years; 29.3%, aged 65 to 69 years; 34.3%, aged 70 to 74 years; and 33.8%, 75 years and older (χ23 = 11.05; P = .01). Most of them were taking only 1 nap per day (90.4%). Among nappers, 14.6% had unintentional naps, that is, they were not planned. The proportion of unintentional nappers was comparable in each age group.
Intentional napping was associated with a higher score on difficulties in orientation for persons compared with never napping. Participants who took unintentional naps had higher scores on attention-concentration deficits, delayed recall, difficulties in orientation for persons, and difficulties in temporal orientation (Table 3).
Intentional nappers reported slightly but not significantly more frequently feeling sleepy during the day (14.0% vs 9.5%; P = .06). The association between intentional naps and EDS was significant only in the 2 younger groups. In individuals aged 60 to 64 years, 19.5% of intentional nappers reported EDS compared with 8.7% of nonnappers (P = .04). In participants aged 65 to 69 years, 15.8% of intentional nappers also reported having EDS compared with 5.6% of nonnappers (P = .02).
Overall, 27.0% of the participants in the study had an organic disease. The most frequently reported diseases were arthritic diseases (20.2%), hypertension (17.3%), and heart diseases (10.5%). The prevalence of organic diseases was higher in the 2 oldest groups (70-74 years, 34.6%; ≥75 years, 31.7%) compared with the 2 younger groups (60-64 years, 18.7%; 65-69 years, 25.2%; χ23 = 18.83; P<.001).
These participants achieved higher scores on 5 of the 6 dimensions of the Cognitive Difficulties Scale: attention-concentration deficits, praxis, difficulties in orientation for persons, difficulties in temporal orientation, and prospective memory (Table 4).
Obstructive sleep apnea syndrome was found in 2.6% of the sample, with no significant difference among age groups. In general, participants with obstructive sleep apnea syndrome obtained comparable cognitive difficulty scores, except on prospective memory, where the scores were significantly higher (Table 4).
Approximately 20% of the participants reported using a medication to promote sleep, with no significant difference among age groups. The most frequently used medications were lorazepam (22.1%), zolpidem tartrate (15.4%), zopiclone (10.6%), and bromazepam (10.1%). Use of sleep medication was associated with higher scores on attention-concentration deficits, praxis, and prospective memory (Table 4).
The use of medication to reduce anxiety was reported by 7.3% of the sample. The proportion of individuals using such medication was higher in the 2 oldest groups (70-74 years, 11.2%; ≥75 years, 9.1%) compared with the 2 younger age groups (60-64 years, 4.9%; 65-69 years, 5.6%; χ23 = 8.85; P = .03).
The most frequently used medication was bromazepam (24.0%). Users of anxiety medication had higher scores than nonusers on attention-concentration deficits and praxis (Table 4).
Finally, 2.6% of the sample reported taking an antidepressant medication, with no significant difference among age groups. This consumption, however, was not associated with cognitive deficits.
The anxiety and general health dimensions of the Psychological General Well-being Schedule were negatively correlated with scores on the Cognitive Difficulties Scale for all dimensions except prospective memory (Table 5). This means that greater anxiety and poor health perception were positively correlated with poor cognitive performance. The depressed mood dimension was negatively correlated with attention-concentration deficits and difficulties in temporal orientation (ie, greater depressive mood was positively correlated with increased difficulties in attention-concentration and temporal orientation). The self-control dimension was positively correlated with 4 of the 6 dimensions of the Cognitive Difficulties Scale (attention-concentration deficits, delayed recall, difficulties in temporal orientation, and prospective memory). Positive well-being was correlated only with attention-concentration deficits, and vitality was correlated with all the dimensions of the Cognitive Difficulties Scale (Table 5).
To verify the hypothesis that daytime sleepiness is an independent predictive factor for cognitive difficulties, we used stepwise logistic regression analyses. The models were adjusted for the possible confounding effects of age, sex, physical activity, occupation, organic diseases, use of sleep or anxiety medications, daytime sleepiness, napping, sleep duration, and the dimensions of the Psychological General Well-being Scale. Each cognitive difficulty dimension was dichotomized into presence vs absence of difficulty to meet the requirements of logistic regression analysis. Furthermore, results obtained on the Mini-Mental State Examination were also used as the dependent variable in a logistic regression model. For all the cognitive difficulty dimensions and the Mini-Mental State (memory), daytime sleepiness seemed to be a significant independent predictive factor (Table 6).
The independent predictive factors for attention-concentration deficits were having daytime sleepiness, taking sleep medication, sleeping 5 hours or less per night, and being anxious. Odds ratios and associated 95% confidence intervals are presented in Table 6. Independent predictive factors for praxis were having daytime sleepiness, being 75 years or older, being a woman, having an organic disease, taking a sleep medication, and being physically inactive (ie, absence of physical exercise ≥3 times per week for ≥15 minutes) (Table 6). For delayed recall, the independent predictive factors were having daytime sleepiness, being a man, being depressed, and being anxious (Table 6). Independent predictive factors associated with difficulties in orientation for persons were having daytime sleepiness, having an organic disease, sleeping 5 to 7 hours per night, being anxious, and having an occupation (a job or participating in various activities such as volunteer work, babysitting, or helping his or her children) (Table 6). Independent predictive factors associated with difficulties in temporal orientation were having daytime sleepiness, having an organic disease, being a woman, being depressed, and being anxious (Table 6). Independent predictive factors associated with prospective memory were having daytime sleepiness, having an organic disease, having a diagnosis of obstructive sleep apnea syndrome, and having an occupation (Table 6). Finally, independent predictive factors associated with the presence of memory deficits on the Mini-Mental State Examination were having daytime sleepiness and being 70 years or older (Table 6).
To our knowledge, this is the first study that aimed to verify whether EDS is an independent predictive factor for cognitive difficulties in elderly individuals living in the community. Our results clearly show that EDS is a strong predictive factor for cognitive difficulties in elderly individuals even after controlling for possible confounding effects of age, sex, physical activity, occupation, organic diseases, use of sleep or anxiety medication, napping, and mental diseases.
In our sample, 14.1% of individuals 65 years and older had moderate to severe daytime sleepiness. Napping at least 2 days per week was reported by 32.6% of participants 65 years and older; 14.5% of elderly subjects were taking a nap daily. Excessive daytime sleepiness should be distinguished from intentional naps. The latter may not reflect the presence of a sleep disorder and can be a healthy habit in the elderly. The increasing rate of napping with age is probably because older people have little or no life constraints preventing them from napping whenever they feel the urge, and rarely does this constitute a social problem.
There was no association between intentional napping and EDS in the elderly individuals in this study. The association was, however, significant in younger individuals in our sample. These younger participants were mostly still working and therefore could not nap when they felt the need. If intentional napping is harmless, EDS can be the expression of a more important underlying disorder, for example, a sleep-related breathing disorder or depression.16- 18,41- 44
Some may question the reliability of sleep data collected on elderly people by telephone. However, the literature suggests that telephone interviews in general are appropriate and yield results comparable to those of other strategies.45,46 Rohde et al46 reported good interrater reliability between face-to-face and telephone interviews assessing DSM-IV psychiatric disorders. Another limitation is that such a method does not allow assessment of the most impaired individuals, such as elderly individuals with important cognitive deficits or those with speech or hearing impairments.
Another limitation is that we did not have objective data on cognitive difficulties and daytime sleepiness apart from questions about orientation and memory from the Mini-Mental State Examination. Some may argue that self-reporting of cognitive difficulties may not be as good as objective measures of cognitive functions. However, a study by Geerlings et al47 showed that self-reported memory difficulties was a strong predictor of incident Alzheimer disease in older persons. Furthermore, 175 of our elderly participants were called back by a physician to undergo a brief interview, and 10 have been invited to come to the sleep laboratory at Xavier Bichat Hospital for a complete examination.
In this study, we found that EDS is an independent predictive factor of a variety of cognitive difficulties that may impair quality of life. Elderly participants in this study were nondemented and had enough autonomy to still live in the community. However, EDS and cognitive difficulties were related in our study to a decreased ability to perform activities of daily living. There was also the possibility that elderly individuals who were retired or who had fewer domestic obligations received less stimulation and therefore were more sleepy during the day. This could be deduced by the increase with age in the number of individuals who took a nap during the day. However, from the results of the logistic regressions that we performed, it seems that age was not a major factor for the presence of cognitive difficulties and that napping was not related at all. Older age was an independent predictive factor on only 2 of the 7 cognitive measures: praxis and memory deficits on the Mini-Mental State Examination. An independent predictive factor that may be surprising at first for difficulties in orientation for persons and prospective memory was the presence of an occupation. However, this is not surprising: the Cognitive Difficulties Scale is based on a self-report of daily living difficulties. An individual who is still active is more likely to be confronted with a decline in cognitive abilities than a person who has retired from active life. This is confirmed by the fact that this factor was not predictive of memory deficits when we used the results on memory from the Mini-Mental State Examination, where memory is more objectively assessed. Furthermore, active individuals at any age in our study less frequently reported being limited in their travels and in their capacity to do their shopping, 2 activities that are narrowly related to the items assessed in the prospective memory (need of a list when shopping and forgetting the things that he or she was planning to buy).
What is the mechanism that explains the predictive value of daytime sleepiness for cognitive difficulties? A possible explanation is that other disorders, such as a mental or organic pathologic condition or an obstructive sleep apnea syndrome, caused daytime sleepiness. Recently, obstructive sleep apnea syndrome, for which daytime sleepiness is a cardinal symptom, has been found to cause cognitive deficits because of the repeated anoxia provoked by breathing pauses during sleep.41,42 However, this explanation does not fully explain our findings. Indeed, the multivariate models controlled for the effects of these 3 types of pathologic mechanisms, and daytime sleepiness still emerged as a strong independent predictor of cognitive difficulties. Another possibility is that daytime sleepiness may be due to a lack of cognitive or social stimulations. Several studies22,48,49 have shown that when elderly people receive cognitive stimulation and are kept socially active, the likelihood of cognitive decline decreases. Another possible explanation is that daytime sleepiness is an early indicator that may predict subsequent cognitive decline. However, longitudinal studies are needed to confirm this hypothesis.
In summary, these data from a community-based sample indicate that EDS is a good predictor of cognitive difficulties. Physicians who treat elderly patients with such complaints should be aware that these patients are at greater risk to have cognitive deficits. As shown in longitudinal studies, it is possible to delay or prevent these cognitive deficits by maintaining intellectual stimulation and by promoting social engagement in these elderly individuals.
Accepted for publication May 8, 2001.
This study was supported by an unrestricted grant from the Laboratoire L. Lafon, Maisons Alfort, France.
We thank Serge Lubin, MD, for his help in this project and Kenza Ejbari, PhD student, for her outstanding work in the monitoring of the study.
Corresponding author and reprints: Maurice M. Ohayon, MD, DSc, PhD, Sleep Disorders Center, Stanford University School of Medicine, 401 Quarry Rd, Suite 3301, Stanford, CA 94305 (e-mail: email@example.com).