Polysomnographic Respiratory Abnormalities in Asymptomatic Individuals

Abstract

Study objectives:

Polysomnographic respiratory abnormalities have been extensively studied in the general population, but studies have not targeted completely healthy individuals. We aimed to (1) define the frequency of respiratory disturbances (RDI: events per hour of sleep) during sleep in healthy individuals using current techniques and criteria and (2) determine how these abnormalities change with age and sex.

Design and Setting:

Cross-sectional analyses of RDI in healthy volunteers.

Participants:

One hundred sixty-three individuals (106 men) were screened for chronic medical illness, as confirmed by extensive questionnaires, physical examination, electrocardiography, and laboratory analysis. Obese subjects (body mass index > 30 kg/m2) and subjects taking medications were excluded.

Interventions, Measurements and Results:

Subjects underwent full polysomnography using current standard recording and scoring techniques. There was a remarkable increase in RDI with age, particularly over 50 years. Ninety-five percent of currently healthy subjects under 50 years of age had an RDI <15, whereas 50% of subjects older than 65 years had an RDI <15. Men had a higher RDI (median 10) than women (median 5). The effect of age on RDI was similar in men and women.

Conclusions:

RDI increases with age even in healthy individuals without symptoms or signs of obstructive sleep apnea syndrome. We do not know whether these individuals will develop pathophysiologic consequences over time or whether this increase with age reflects a normal aging process. If the former, treatment should be considered regardless of symptoms. If the latter, the criteria for treatment should be adjusted by age.

Citation:

Pavlova MK; Duffy JF; Shea SA. Polysomnographic respiratory abnormalities in asymptomatic individuals. SLEEP 2008;31(2):241–248.

OBSTRUCTIVE SLEEP APNEA-HYPOPNEA SYNDROME (OSAHS) IS COMMON,^1–4 INCREASES WITH ADVANCED AGE,⁴ IS A MAJOR CAUSE OF EXCESSIVE DAYTIME sleepiness,^3,5 is associated with serious cardiovascular events,^6–8 and is a considerable cost to our healthcare system.⁹

In our research group, as part of a study on aging, it was necessary to recruit healthy young and elderly individuals. These volunteers underwent extensive health screening. The most expensive screening test was a clinical polysomnography, so this was performed last and in only those volunteers who passed all other tests. We were surprised that the prevalence of sleep-disordered breathing among the elderly volunteers in this group of currently completely healthy individuals was larger than in any other reported study. We assumed that the recent developments of more-sensitive measurement techniques¹⁰ and scoring criteria¹¹ could explain our observations. Thus, the aims of the current study were to (1) report the frequency of respiratory disturbances during sleep in otherwise completely healthy individuals measured in a typical clinical sleep laboratory accredited by the American Academy of Sleep Medicine while using current measurement techniques and recently established scoring criteria and (2) report how the frequency of respiratory disturbances during sleep changes with age and sex in these asymptomatic individuals.

METHODS

Subjects

We analyzed polysomnograms from 163 consecutive subjects studied over a 4-year period. All subjects were healthy volunteers who had responded to an advertisement to participate in studies of circadian physiology. Their health status was confirmed by extensive medical history questionnaires followed by electrocardiography, blood chemistry profiles, liver function tests, complete blood count, urinalysis, a history and physical examination by a physician, and a psychiatric and psychological examination by a clinical psychologist. Subjects were excluded if they were obese (body mass index [BMI] > 30); taking any medications; or had any chronic or current acute medical or psychiatric disease including depression or other psychopathology (e.g., Beck Depression Inventory score >10 or Minnesota Multiphasic Personality Inventory score ≥ 90), hypertension, anemia or other hematologic, hepatic, or other abnormalities; if they reported any current or chronic sleep disturbances or sleep disorders, recent shift work, or circadian rhythm disturbances; if they had a first-degree relative with psychiatric history; or if they reported substance abuse, caffeine dependence (> 4 caffeinated beverages per day), alcohol dependence (> 14 alcoholic drinks per week), or nicotine dependence (> 4 cigarettes per day). Finally, use of all of these substances had to be stopped 3 weeks prior to the start of the inpatient study, and this cessation generally started 1 to 2 weeks prior to the screening polysomnography night. To verify these self-reports, a urine sample was tested early in the screening process to confirm that the subjects were free from drugs of abuse, alcohol, caffeine, and nicotine. As the final step in the screening process, only subjects who had passed all of the above tests were then scheduled for an overnight clinical polysomnogram. Historically, in our laboratory, approximately 5% of subjects who respond to the initial advertisement reach this final stage in the recruitment process; thus, this represents a highly selected group of currently completely healthy individuals.

Polysomnography

Subjects underwent full nocturnal polysomnography, and scoring was performed by registered polysomnographic technologists at a laboratory accredited by the American Academy of Sleep Medicine. Overnight recordings were performed with “lights out” generally between 22:00 and 23:00, and recordings continued for 7 to 8 hours. Data included sleep stages (using 4 electroencephalograms, 2 electrooculograms, and a submental electromyogram), arterial oxyhemoglobin saturation (pulse oximeter), airflow (snoring microphone, nasal pressure transducer, oronasal thermistors), abdominal and thoracic breathing movements (strain gauges wrapped around the thorax and abdomen), electrocardiography, and periodic limb movements (surface electromyography of anterior tibialis activity of both legs). Data were sampled at 100 Hz and stored on a computer using a digital data acquisition system (Alice software versions 3 and 4, Respironics, Murrysville, PA). At the time of the polysomnogram, most subjects completed an additional clinical sleep questionnaire concerning sleep habits and symptoms, the Epworth Sleepiness Scale, and the presence or absence of any snoring. Snoring was reported in answer to the question, “Do you snore?” Answers were selected from “frequently,” “occasionally,” “never,” or “don't know.” Subjects who answered “frequently” or “occasionally” were classified as snorers. In addition, to objectively assess snoring, the polysomnographic technician reported whether or not any snoring was heard overnight via a microphone placed next to the subject's bed.

Scoring of Sleep Stages, Arousals, Respiratory Events, and Leg Movements

Sleep stages, apneas, hypopneas, arousals, and leg movements were scored visually from the computer screen using Alice software versions 3 and 4. Sleep stages were scored according to standard criteria.¹² Arousals were detected from 3-second or longer changes in electroencephalography and electromyography using standardized criteria.¹³ Respiratory events were scored based on the recommendations of the Task Force of the American Academy of Sleep Medicine (the “Chicago Criteria”; 1999). Thus, apnea was scored when cessation of airflow for 10 seconds or longer was observed. Apneas were further classified as obstructive or central depending on the presence or absence of chest-wall breathing movements. Hypopneas were identified based on a discernible decrease in breathing for at least 10 seconds (observed in the respiratory strain gauge, nasal pressure, or thermistor recordings), followed by either arterial oxyhemoglobin desaturation of at least 3% or an arousal. Hypopneas were not categorized as obstructive or central events. Quality control data for this laboratory revealed the interscorer reliability is greater than 90% for scoring respiratory events and for sleep staging. The resting, awake, and supine baseline arterial oxyhemoglobin saturation (SaO₂) was recorded, as was the minimum SaO₂ overnight during sleep. Periodic limb movements were scored when at least 4 consecutive bursts of anterior tibialis activity (> 25% of the calibration activity with a duration of 0.5-5 seconds) occurred within 5 to 90 seconds. The total number of individual leg movements was recorded. If leg movements accompanied arousals at the termination of a Chicago Criteria-defined respiratory event, then these events were classified as respiratory disturbances, and, in such cases, the accompanying leg movements were not counted in the periodic limb movements of sleep index (PLMI). The nasal-pressure tracing, the results of the snoring-detection microphone on the neck, and alterations in the phase relationship between thoracic and abdominal breathing motions (e.g., paradoxical motion) were used qualitatively to help identify airflow limitation. If a leg movement caused an arousal with accompanying larger breaths and there were subsequent smaller unobstructed breaths at sleep onset, then these events were classified as leg movements and counted in the PLMI, rather than the respiratory disturbance index (RDI).

After being scored by a registered polysomnographic technologist, as occurs in most accredited laboratories, all records were reviewed by a sleep specialist, who confirmed the adequacy of the scoring (adjusting scoring where necessary was < 2% of events).

Statistical Methods

Subjects were categorized by age and sex. Subjects were divided into 4 age groups, with each group spanning 15 to 17 years: 18 to 34, 35 to 49, 50 to 64, and 65 to 81 years. We used analysis of variance to compare different age groups and sexes, and posthoc comparisons were performed using the Tukey HSD test. Comparisons of frequency distributions between all pairs of age groups were performed using the Kruskal-Wallis test (essentially a comparison of medians for ranked data, which is suitable for non-Gaussian distributions). We corrected these posthoc tests for multiple comparisons (Bonferroni). No formal analyses were performed on the interaction between sex and age, as there were insufficient data. Physiologically different mechanisms may lead to obstructive and central respiratory events, so separate analyses were performed on the total RDI, as well as on he number of clearly obstructive events, namely the obstructive apnea index.

RESULTS

Effect of Age

The demographic and polysomnographic characteristics of the subjects in these 4 age groups are presented in Table 1. Across all subjects, there was a dramatic increase of RDI with age, particularly above 50 years (Figure 1, top panel). The cumulative relative frequency distributions of RDI and minimum SaO₂ for each age group are presented in Figure 2. It can be seen that 95% of the subjects younger than 50 years of age have an RDI of less than 15 events per hour of sleep, whereas fewer than half of the subjects older than 65 years of age have an RDI of less than 15. In the oldest group, an RDI of 32.5 events per hour of sleep defines the upper quartile. When comparing the frequency distributions of RDI between age groups, 5 of the 6 comparisons between pairs of age groups revealed significant increases in RDI with age (Figure 2 upper panel; all P values < 0.05, except for comparison between age groups 35–49 vs 50–65).

Table 1.

Characteristics of all Subjects When Classified by Age Group

	Age range, y				P Value
	< 35 (n = 63)	35–49 (n = 15)	50–65 (n = 42)	> 65 (n = 43)
Age, y	25.1 (4.2)	40.2 (2.9)	58.0 (4.6)	71.1 (4.1)
Sex, no, M/F	39/24	12/3	22/20	33/10
BMI, kg/m2	23.4 (2.7)	25.8 (2.4)	24.8 (3.1)	24.8 (2.3)	0.034
ESS, score	4.0 (2.8)	2.8 (1.7)	3.5 (2.9)	4.5 (2.7)	NS
RDI, no./h	4.3 (4.2)	7.7 (4.1)	12.8 (12.4)	22.0 (17.4)	< 0.00001
PLMI, no./h	1.3 (3.1)	1.6 (3.6)	7.4 (15.5)	20.2 (32.3)	< 0.00001
Sleep efficiency, %	83.2 (11.6)	86.5 (8.1)	74.5 (10.8)	70.6 (17.3)	0.0005
Arousal index, no./h	16.4 (7.4)	23.3 (7.0)	22.0 (9.6)	26.6 (12.9)	< 0.00001
OAI, no./h	0.2 (1.1)	0.1 (0.3)	1.3 (2.4)	5.4 (9.1)	< 0.00001
Minimum SaO2, %	91.6 (3.9)	90.5 (5.0)	89.1 (5.7)	84.9 (6.9)	< 0.00001
Reporting snoringa
At home	17	15	30	35
In lab	27	54	49	40
Awake SaO2< 95%a	3	0	10	26
Minimum SaO2 < 90%a	19	31	46	74

Data are presented as mean ± SD, except sex, which is provided as the number of men and women and those categories marked with ^a, which are presented as percentage. BMI refers to body mass index; ESS, Epworth Sleepiness Scale; RDI, respiratory disturbance index (the number of events per hour of sleep); PLMI; periodic limb movement index (the number of events per hour of sleep); OAI, obstructive apnea index (the number of obstructive apneas per hour of sleep).

Figure 1.

Scatter plot of the respiratory disturbance index (RDI: events per hour of sleep) by age (years) of men and women. Upper panel, all subjects: the distribution has a similar pattern in men and women, with a dramatic increase in RDI after age 50. There is a similar effect of age in the subgroup of nonsnoring, nonsleepy, nonoverweight subjects (lower panel).

Figure 2.

Cumulative relative frequency of respiratory disturbance index (RDI: events per hour of sleep; upper plot) and minimum arterial oxyhemoglobin saturation recorded overnight (%: lower plot) according to age group. More than 95% of the subjects younger than 50 had an RDI < 12.5, whereas fewer than 50% of the subjects older than 65 years had RDI < 12.5. The increasing RDI with age was associated with concomitant lower minimum arterial oxyhemoglobin saturation with age (1 age group was omitted from the arterial oxyhemoglobin saturation plot due to missing or insufficient data).

The Chicago Criteria do not distinguish between hypopneas associated with arousal and those associated with desaturation, but, with the majority of events, both desaturation and arousal occurred. We did not test whether or not there are different distributions in arousal-related versus desaturation-related hypopneas across the age groups, as this distinction is generally not done clinically. However, we did assess baseline and minimal overnight SaO₂. The baseline awake, supine resting SaO₂ was above 91% in all individuals. Baseline awake SaO₂ was no different between the first 3 age groups (medians = 97%) and was 2% lower in the oldest group (median = 95%). Approximately one quarter of the oldest group, and only 5% of the combined younger 3 groups, had a baseline awake SaO₂ below 95% (Table 1). More than 80% of the youngest group had an overnight SaO₂ that remained above 90%, but more than half of the total subjects older than 50 years of age had an overnight minimum SaO₂ below 90%, and 40% of the oldest group had a minimum SaO₂ below 85%, representing notable sleep-related desaturation in this age group. In comparing distributions (posthoc Kruskal-Wallis tests, Bonferroni corrected), all 3 of the comparisons of minimum SaO₂ between pairs of age groups revealed significant decreases in minimum SaO₂ with age (Figure 2 lower panel). Thus, there was a negative correlation between age and minimum SaO₂ (r = −0.54; P < 0.0001). Subjects with a higher BMI tended to have lower minimum SaO₂ overnight (r = 0.27; P < 0.001). There was no significant correlation between minimum SaO₂ and daytime sleepiness as judged by Epworth Sleepiness Scale (r = 0.09). Older subjects also had significantly higher indexes of obstructive apneas, periodic limb movements of sleep, and arousal and significantly reduced sleep efficiency (Table 1).

Effect of Sex

The demographic and polysomnographic characteristics of the 106 male and 57 female subjects are presented in Table 2, and the cumulative relative frequency distributions of RDI and minimum SaO₂ for each sex are presented in Figure 3. It can be seen that men had a higher RDI and lower minimum SaO₂ than did the women (Table 2; Figure 3). The median RDI for men was 10, and the median RDI for women was 5. The median of minimum SaO₂ was less than 90% for men and greater than 90% for women. The dramatic effect of age on RDI was similar in men and women (Figure 1), although no inferential statistics were performed on this interaction term (see Methods). Male subjects also had significantly higher periodic limb movement and arousal indexes (Table 2).

Table 2.

Characteristics of All Subjects when Classified by Sex

	Women (n = 57)		Men (n = 106)		P Value
	mean ± SD	range, y	mean ± SD	range, y
Age, y	44.6 (18)	18-72	48.3 (21)	18-81	NS
BMI, kg/m2	23.6 (2.9)	16-30	24.8 (2.6)	19-30	0.008
ESS, score	3.7 (2.8)	0-12	4.07 (2.7)	0-13	NS
RDI, no./h	7.5 (8.4)	0-44	13.7 (15)	0-73	0.004
PLMI, no./h)	3.1 (8.3)	0-46	10.5 (23.7)	0-128	0.03
Sleep efficiency, %	78.4 (15.5)	10-95	77.6 (13.3)	25-97	NS
Arousal index, no./h	18.8 (8.9)	3-43	22.4 (10.1)	0-57	0.04
Minimum SaO2, %	91.3 (4.7)	76-97	87.9 (6.3)	61-97	0.0007

Data are presented as mean ± SD and range. BMI refers to body mass index; ESS, Epworth Sleepiness Scale; RDI, respiratory disturbance index (the number of events per hour of sleep); PLMI; periodic limb movement index (the number of events per hour of sleep).

Figure 3.

Cumulative relative frequency distributions of respiratory disturbance index (RDI: events per hour of sleep; upper plot) and minimum arterial oxyhemoglobin saturation recorded overnight (%: lower plot) in men and women. Women had an overall lower RDI and less severe desaturation than men, although the shapes of the distributions were similar.

Snoring and Excessive Sleepiness

Questionnaires were collected from 136 subjects. Forty of these subjects reported some snoring. In comparison with the 96 subjects who denied snoring, the reported snorers had a higher mean RDI (17.6 vs 10.1; P = 0.004), a higher mean obstructive apnea index, (3.7 vs 1.5; P = 0.03), and a slightly higher mean BMI (25 vs 24, P = 0.03). There was no significant difference in age between individuals who reported or denied snoring. Snoring was confirmed by microphone during the polysomnogram in 22 of the 40 reported snorers as well as in 20 of the 96 individuals who denied snoring (Table 1). The sensitivity of reported snoring for predicting recorded snoring was 0.48, and the specificity was 0.79.

Of 96 subjects who reported an Epworth Sleepiness Scale score, only 6 (6%) had a score higher than 10, suggestive of excessive daytime sleepiness, and 3 of these 6 had an RDI less than 5.

Secondary Analyses on the Subset of Subjects Who Denied Snoring and Excessive Sleepiness, and Who Were Not Overweight

Reported snoring along with reported daytime sleepiness and obesity are among the primary signs of sleep disorders utilized by clinicians. Although we had already excluded obesity (BMI > 30), many subjects would be considered overweight (25 < BMI> 30). Thus, we performed a secondary analysis on only those subjects (n = 51) who reported no snoring and little or no significant daytime sleepiness (Epworth Sleepiness Scale score < 10) and who were not overweight (BMI < 25). The correlation between age and RDI remained similar after excluding subjects with subjective sleepiness or any reported or recorded snoring (r = 0.56; P < 0.001), as well as after excluding overweight individuals with these symptoms (BMI > 25; r = 0.51; P < 0.001). These data are presented in Figure 1 (lower panel).

DISCUSSION

We sought to determine the degree of sleep-disordered breathing in otherwise currently completely healthy, nonobese individuals when measured in a typical clinical sleep laboratory accredited by the American Academy of Sleep Medicine. Strikingly, we found that the majority of currently healthy older subjects, particularly if they were men, would be considered “abnormal,” based on commonly used criteria (e.g., an RDI cutoff of 10 events per hours of sleep). The data suggest that respiratory disturbances during sleep naturally accompany aging and raise questions about how to account for differences in the RDI across age group when this is used as a metric for therapeutic intervention. Although these subjects are currently healthy, it is unclear from these cross-sectional data whether those subjects with higher RDI have preclinical OSA that will later manifest with sleepiness of if they have subclinical disturbances that may eventually manifest with cardiovascular or other morbidity. The strengths and limitations of this study are presented below along with an overall interpretation and a discussion of the clinical implication of the results.

The population studied

Although all subjects passed an extensive health screening, it remains possible that there is some socioeconomic class bias in the population studied because participants answered an advertisement for a research study that involved remuneration. Nonetheless, there is no reason to suspect that individuals with sleep disorders would be overrepresented in the older group or among the men. Although obesity is clearly associated with OSAHS, we limited our population to nonobese subjects (BMI < 30). Thus, we believe that these data can be generalized to the broad population of healthy nonobese individuals.

High RDI has been reported in other studies of older subjects.^1–4,14 Individuals included in our study had a larger age span, as compared with other studies such as the Wisconsin Study cohort⁴ (30 to 60 years old) and the Sleep Heart Health Study¹⁵ (> 40 years old). In our study, the differences in RDI in different age groups were more dramatic, perhaps because the age span was larger, because more-sensitive recording techniques and scoring criteria were used, or both. Thus, our study extends the previously reported normative data in terms of including a wider age span and excluding current comorbidity and obesity, and our study provides a further update on preexisting data because we used commonly used current recording techniques and scoring criteria.

It is possible that our results are influenced by a tendency for older individuals with sleep disorders to underreport their symptoms, in comparison with younger subjects with sleep disorders who may have been screened out prior to the polysomnogram due to reported symptoms. For example, complaints of insomnia increase with age until around age 65 and then decrease after age 65,¹⁶ despite equivalent or greater insomnia in the older group. In a similar way, older individuals may underestimate the significance of snoring or may be less troubled by fatigue after retirement. However, the most dramatic increase in the rate of respiratory disturbances was found at a younger age, in individuals in their 40s and 50s. Thus, underestimation of symptoms is not as likely to cause underreporting of symptoms in this age group.

The arousal index appears high in our subjects, even in the younger group. These arousal rates may be high due to a first-night effect (e.g., anxiety, miscellaneous arousing stimuli, equipment, etc.). The arousal rates that we found are similar to the arousal rates reported for healthy subjects staying in the sleep laboratory for the first time (e.g., 21 per hour [American Sleep Disorders Association 3-seconds criteria]¹⁷). Excluding almost half of their subjects who displayed some snoring or apnea did not alter the mean arousal frequency in the remaining subjects in that study.¹⁷ Those authors also found that the arousal rate significantly increased with age, as occurred in our data.

Technique Used for Recording and Scoring Breathing Abnormalities

The techniques we used to record breathing abnormalities are more sensitive than those used in many prior studies on the prevalence of OSAHS (for example, Sleep Heart Health Study,^1,2,16,18,19 Wisconsin cohort⁴). For instance, we used nasal cannula pressure as a measure of airflow obstruction, which is more sensitive in detecting subtle respiratory events,^12,18 especially in patients who have less severe apnea,²⁰ and especially when the presence of airflow limitation pattern from a nasal pressure recording is used to aid with the scoring.²¹ In our own laboratory, we found that scoring events using nasal pressure estimates of airflow when compared with oronasal thermistor estimates of airflow resulted in an increased RDI by an average of 9 events per hour of sleep among 40 consecutive patients entering the laboratory with a presumptive diagnosis of sleep-disordered breathing.²²

Different criteria exist for the definition of hypopnea.¹⁹ Using the Medicare definition of hypopneas (which requires a ≥ 4% desaturation) is very likely to result in a lower RDI than when using the Chicago criteria (which also counts hypopneas terminated by arousal but without desaturation).²³ Nonetheless, it is widely believed among sleep medicine professionals that respiratory-related arousals have a significant effect on sleepiness, which is the major symptom of OSAHS.²⁴ Thus, although it is clear that the use of nasal pressure recordings and the use of the Chicago criteria lead to a higher reported RDI, as compared with some the results of other studies that have utilized less sensitive techniques and criteria, many clinical sleep laboratories across the United States do utilize both nasal pressure recordings as well as the Chicago criteria. Thus, we believe that our findings have current relevance.

One quarter of the oldest group had a baseline awake SaO₂ below 95%, compared with only 5% of the other 3 age groups (Table 1). This difference likely represents a natural decrease in baseline SaO₂ with age. Since the oldest group tends to live on a slightly steeper part of the oxyhemoglobin desaturation curve, compared with the younger groups, equivalent changes in partial pressure of oxygen would result in slightly larger changes in arterial oxyhemoglobin saturation in the older group. Because the Chicago criteria do not distinguish between hypopneas associated with arousal and those associated with arterial oxyhemoglobin desaturation alone, it remains a possibility that a change in baseline oxygenation could contribute to a higher RDI in the older group—particularly in those individuals with the lowest baseline saturation.

We note that the distinction between central and obstructive hypopneas cannot be determined with absolute certainty. A predominantly obstructive component can be assumed if, in the presence of snoring, a flattening airflow limitation pattern in the nasal pressure trace or alteration in the phase relationship between thoracic and abdominal breathing motions occurs (e.g., paradoxical motion). The interpreting sleep specialist reviewed all records and estimated that a qualitative sign of airflow limitation occurred in the vast majority of the scored respiratory disturbances and that this did not noticeably differ among age groups. However, we did not perform a quantitative analysis on these records because that is not done clinically due to frequent lack of clear distinction between obstructive and central hypopneas. However, in terms of the obstructive apnea index, which is derived from unambiguous obstructive apneas, we saw a clear increase in obstructive apnea index with age that mirrored the RDI (Table 1). In addition, both reported and recorded snoring increased substantially from the youngest to the oldest age groups (Table 1). Thus, overall evidence suggests that the increase in RDI with age in this study was mostly due to an increase in obstructive rather than central respiratory disturbances.

The polysomnograms were scored by registered polysomnographic technologists who knew that the subjects were being screened as healthy recruits for a subsequent circadian research study and that the subjects did not necessarily have any symptoms of sleep disorders. Nor were the personnel blinded to the age of the individual. Thus, it is possible that the technicians imparted some bias, but it seems unlikely that any such bias would result in the overscoring of events in these ostensibly healthy individuals. Moreover, all scoring undergoes regular quality control, and each of these records was subsequently reviewed by a specialist in sleep medicine, who agreed with or modified the scoring, as occurs in most accredited clinical laboratories. Finally, although some individuals with severe OSAHS experience prolonged apneas in all lying postures and sleep stages, a larger group of individuals experience obstructed breathing during sleep only when they are supine,²⁵ or only in specific sleep stages,²⁶ but sleep stage and posture cannot be well controlled throughout the clinical polysomnogram test and were not reported in the current study.

The severity of OSAHS as judged by symptoms of excessive daytime sleepiness is subject to variability because both subjective and objective measurements of sleepiness are relatively imprecise measures^27–29 and sleepiness can be caused by other sleep disorders or lifestyle choices that are not specific to OSAHS.³⁰ A disparity between RDI and clinical symptoms has been observed in other studies. In the slightly younger population studied by Young et al,⁴ a population that was not prescreened for health status, 24% of men and 9% women had an abnormal RDI, but only 9% of men and 4% of women met the diagnostic criteria for OSAHS, which includes the symptom of sleepiness. Our secondary analyses in subjects who denied sleepiness and snoring and who were not overweight showed similar effects of age on the RDI. Thus, in our group of highly selected nonobese subjects, these symptoms had considerably less effect in predicting abnormal RDI than did age and less effect in predicting abnormal RDI that occurs in the population who attend sleep laboratories for clinical reasons. In the current study, only 6 subjects reported excessive daytime sleepiness, and the high RDI in the older individuals persisted even after excluding individuals who reported snoring or sleepiness. Thus, a high proportion of elderly men, a smaller proportion of younger men, and some women have a high RDI plus notable desaturation, which is suggestive of moderate to severe OSAHS but without symptoms.

What are the Clinical Implications of a High RDI without Symptoms?

RDI increases dramatically with age even in healthy individuals without symptoms or signs of obstructive sleep apnea syndrome. If a high RDI inevitably or often leads to pathophysiologic consequences, then treatment for sleep apnea should be considered based on RDI alone—regardless of symptoms. On the other hand, if an increase in RDI with age reflects a normal aging process without adverse consequences, then the clinical interpretation of abnormal breathing during sleep should be adjusted by age. These 2 possibilities are perhaps not mutually exclusive and are considered below.

More than 95% of the individuals younger than 50 years of age had an RDI of less than 15, but fewer than 50% of the older asymptomatic individuals had an RDI of less than 15. Simplistically, this effect of age on RDI suggests that a high RDI in a younger individual may be more clinically relevant than the same RDI in an older subject. Implicit in this assumption is the fact that a higher RDI with age is part of the normal aging process and does not necessarily reflect pathologic breathing changes. For example, it is possible that the sleep fragmentation that naturally increases with age^{31, 32} may lead to irregular breathing with larger breaths at the time of arousal and smaller breaths or even central apneas at sleep onset. These irregularities in breathing pattern could potentially be misinterpreted as obstructive hypopneas. Whether or not such natural fluctuations in breathing that accompany sleep fragmentation is of clinical concern beyond the sleep fragmentation itself is not known. In addition, as noted above, most evidence in the current study suggests that the increase in RDI with age was at least partly due to an increase in obstructive rather than central respiratory disturbances.

Although all subjects in this study were currently completely healthy, it is unclear from these cross-sectional data whether those subjects with higher RDIs have subclinical abnormalities that will eventually adversely affect health. Certainly, many clinicians would be concerned if their patient's nocturnal SaO₂ falls below, say 80%, especially in a patient with cardiac disease, even in the absence of excessive daytime sleepiness. Recurrent transient hypoxia and arousals, triggered by respiratory obstructive events, could potentially contribute to long-term pathologic changes with OSAHS, such as adverse cardiovascular consequences, facets of metabolic syndrome, or both. In the absence of any longitudinal therapeutic studies in asymptomatic individuals having signs of OSAHS, it is not known whether such individuals would benefit from therapy for OSAHS. The benefits from treatment of OSA include control of symptoms of sleepiness³³ or improvement of cardiovascular outcome.³⁴ On the other hand, many patients with obstructive sleep apnea and cardiac disease do not have sleepiness.³⁵ It is possible that obstructive sleep apnea in these patients could have contributed to development of cardiac morbidity and that early treatment of the obstructive sleep apnea may help delay or prevent it in a way similar to the early treatment of hypertension. Thus, detection of sleep apnea even in the absence of the classic symptoms may benefit certain patients. Our subjects did not have sleepiness and thus would not have been referred for evaluation.

Definitive answers to the clinical necessity for obstructive sleep apnea screening and therapy in asymptomatic individuals and the age- and sex-specific normal RDI threshold will require longitudinal studies of health outcomes in untreated and treated individuals with well-characterized RDIs. The Sleep Heart Health study is one example of a study that will help answer such questions, although caution with interpretation of current clinical polysomnograms will be needed because recording techniques and scoring criteria have been widely updated since the start of the Sleep Heart Health Study.