INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

When confronted with questions related to their own snoring, many subjects simply state that they "don't know" whether they engage in such nocturnal behavior. The prevalence of such responses varies among surveys and ranges between 2.1 and 14% (1), although at least one study including elderly subjects noted that 36 to 38% of the population employed such a response (10, 11). Previously we have noted that in a sleep clinic population responses of "I don't know" to questions about snoring conferred some risk of having polysomnographically assessed sleep apnea (12). In this study we examined risk factors and disease correlates for the "don't know" response to three different snoring questions. We hypothesized that if subjects disavowed knowledge of their own snoring, this response would be correlated with variables previously shown to be associated with snoring.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The data reported in this study are based on the 1991-1992 follow-up of the Western Collaborative Group Study (WCGS). WCGS was initiated in 1960-1961 as a study of cardiovascular risk factors in a population of 3,152 healthy working men 39 to 59 yr of age (13, 14). Between 1960 and 1970 regular annual examinations were performed in this cohort. Subsequent follow-ups to ascertain vital status occurred in 1982-1983 and again in 1986-1988; the latter data collection also involved follow-up physical examinations and selected laboratory tests. An additional round of physical examinations and selected laboratory tests were conducted in 1991-1992. Data presented below were generated from 1,155 men alive at the time of this most recent follow-up, which represented 70% of the surviving cohort at that time. Those with known stroke at the time of follow-up (see RESULTS) were excluded from all analyses reported below. Further details regarding participation in various phases of the WCGS focusing on the last decade of follow-ups have been published elsewhere (15, 16).

In order to include a population of women of comparable age and demographics to the original male cohort, the 1986-1988 data collection sampled household female coinhabitants for each of the participating WCGS members at that time. The 1991-1992 follow-up (the round of data collection on which the present report is based) included these female subjects. In most cases, these were wives of the WCGS members. Because of their method of recruitment, the women sampled in this study should not be considered to be representative of the same population as the WCGS men.

The final sample on which the present data are based consisted of 1,155 men (X age = 75.9, SD = 4.3; range, 70 to 92 yr) and 560 women (X age = 71.4, SD = 5.3; range, 44 to 88 yr). Of the men, 14% lived alone. Of the women, few (2%) lived alone; however, this small number reflects the aforementioned method of recruitment of the female subjects. The difference in number of male and female participants reflects the fact that wives were recruited many years after enrollment of their husbands and were less inclined to participate. The percentage of women who lived alone reflects widows of men in the WCGS cohort who were enrolled in the 1986-1988 follow-up whose husbands had died in the interim prior to the 1991-1992 follow-up.

The WCGS represents a largely white college-educated population whose vocational achievement tended to be relatively high. Most held "white-collar" or professional positions. Relative to other well-studied characterized populations such as the Established Populations for Epidemiologic Studies of the Elderly (17), the WCGS tends to be healthier both medically and psychiatrically and engages in more healthful behaviors (less smoking and higher levels of physical activity).

Procedures

WCGS subjects and spouses underwent a complete follow-up, including a structured health screening questionnaire, a complete medical history, including a review of all hospitalizations and medical records relevant to cardiovascular disease, and measurements of various components of a cardiovascular status. For the current analyses, in addition to chronologic age and sex, the following risk factors were defined.

Obesity. Obesity was quantified by the body mass index (BMI), defined as weight (kg) divided by [height (m)]² (kg/m²).

Alcohol use. Alcohol use was defined as the number of ounces of alcoholic beverages consumed (beer, wine, hard liquor) weekly divided by seven to derive an average number of ounces consumed daily.

Physical activity. Customary level of physical activity was measured with the Physical Activity Scale for the Elderly (PASE) (18), an activity scale previously validated with activity monitors and shown to be reliable over time for use in an elderly population. As reported by Washburn and colleagues (18), scores on the PASE range from 0 to 360, with higher scores indicating greater degree of customary physical activity. The mean and median score of an independently living ambulatory elderly sample were reported to be about 103 (SD = 64) and 90, respectively.

Smoking. Subjects were categorized into those who reported ever smoking (including current and former) versus those who never smoked.

Pulmonary disease. Subjects were considered to have pulmonary disease if they reported history of any of the following conditions: bronchitis, emphysema, asthma, tuberculosis, or respiratory allergies.

Current medication use was established by having all participants bring medications with them to examination. For those not seen in person, this information was obtained from questionnaires. The following categories of medications were coded.

Respiratory medications. Beta adrenergics, sympathomimetics, xanthines, or steroidal inhalors.

Sinus medications. Any prescription or over-the-counter decongestants, antihistamines, or decongestant/antihistamine combination.

Thyroid medications. Any thyroid replacement medication.

Self-reported Snoring

Snoring was defined by responses to three separate questions: (1) How often do you snore in any way? (2) How often do you snore loudly and disruptively? (3) How often do you hold your breath during sleep? Questions were based on the time frame of the previous 6 mo and could be answered with any of the following responses: "never," "just a few times," "sometimes," "fairly often," "very often," or "don't know." These questions have been used extensively in studies of clinical and nonclinical populations (12, 19) and have been validated polysomnographically (12). Because of suspected links between self-reported snoring and household living status, we determined, for each subject in the WCGS, whether that subject lived alone or lived with a spouse or other companion at time of follow-up.

Statistical Analyses

The relative distributions of responses to the three questions are shown by sex in Figure 1. We performed multiple logistic regression, with all risk factors predicting snoring status. For each analysis, "never" was the referent category. "Low" snoring frequency consisted of responses of "a few times" or "sometimes." "High" snoring frequency consisted of responses of "often" or "always." "Don't know" responses were entered separately into logistic models and also referenced to "never." For use in logistic models, smoking, presence of pulmonary disease, and use of respiratory, sinus, and thyroid medications were entered as categorical variables; age, BMI, alcohol consumption, and physical activity were all entered as continuous variables.

View larger version (25K): [in this window] [in a new window]   Figure 1.   Relative frequency distributions of WCGS subjects' responses to three questions about snoring.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Snoring, as shown in Figure 1, was the most commonly endorsed response in both men and women, followed by snoring loudly and then breathholding. "Don't know" responses were relatively common in this population. Between 28 and 44% of the cohort were unwilling or unable to answer at least one of these questions. Men were more likely than women to employ a "don't know" response to questions about snoring loudly and breathholding. Because some (14%) of the WCGS men lived alone and because this might have conceivably affected their frequency of use of the "don't know" response, we analyzed the effect of household living status. The proportion of men living alone indicating "don't know" was significantly higher than for men residing with a spouse or significant other for snoring (62 versus 22%, chi-square = 99.5, p < 0.0001), snoring loudly (65 versus 26%, chi-square = 90.7, p < 0.0001), and breathholding (66 versus 39%, chi-square = 38.1, p < 0.0001). In women, household living status was not significantly associated with "don't know" responses for any of the three questions; however, this may have reflected the small number of women in the WCGS who lived alone (see METHODS).

Results of the multiple logistic regressions are shown in Tables 1, 2, and 3. Analysis of both low and high snoring responses indicated that men reported snoring more often than women did for all three items. Age and BMI appeared in most of the models as well, indicating a decreased likelihood of reporting snoring with advanced age and increased likelihood with increased BMI. Other risk factors associated with low or high snoring (relative to never snoring) included: smoking, use of thyroid or sinus medication, and history of respiratory disease.

Similar to the risk conferred by both low and high levels of reported snoring, "don't know" responses were associated with male sex and higher BMI. Among other risk factors, subjects stating that they "don't know" whether they snored were more likely to use sinus medication relative to those who never snored (11.5 versus 7.7%). Similar associations with use of sinus medication were seen for snoring loudly (11.3 versus 9.2%) and breathholding (11.7 versus 9.8%). "Don't know" responses to the breathholding question were also more likely to be positively associated with smoking history relative to "never" breathholding (8.0 versus 5.4%).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The implications of subjects stating "I don't know" when confronted with questions about their snoring may be relevant for epidemiologic research. The prevalence of this response in the WCGS data, ranging from 28 to 44%, is substantially higher than figures reported for middle-age populations: 6.8% (9), 2.1% (1), 11.2% (8), 5.0% (4), 1.0% (5), 5.0% (2), 7.0% (7) and in some: 5.7% (3), 5 to 14% (6), but not all: 36 to 38% (10, 11) elderly populations as well. Curiously, these relatively high figures for lack of endorsement in the WCGS take place in the context of the relatively small proportion of subjects living alone (14%), a factor that has been shown to be associated with the use of the "don't know" response (6, 8, 20), particularly in the one study of the elderly reporting a comparable prevalence (10, 11).

Apart from such issues of prevalence, our data show that factors associated with use of the "don't know" response are similar to factors associated with snoring (male sex, younger age, higher BMI) in this study. Additionally, other risk factors such as use of sinus medication and positive smoking history, shown to be risk factors for snoring in other studies (2, 5, 10, 21), were also associated with the "don't know" response here. Taken together, these data imply that epidemiologic studies should not discard or discount such responses when encountered in populations. In fact, previously we have noted that, in a sleep clinic population, responses of "I don't know" to any of the three questions asked here conferred at least some risk of having polysomnographically assessed sleep apnea (12). This was seen in men and, to some extent, in women as well and was relatively independent of the number of people living in the household or sleeping in the bedroom. Why subjects elect to answer in this way is unclear, though it is conceivable that an individual subject's purported lack of knowledge regarding this aspect of their sleep may constitute some element of unwillingness to admit to or, conceivably, denial of their behavior.

This study is clearly limited by the nature of the population under investigation. For example, the women presently incorporated into the cohort are, strictly speaking, not directly comparable to the men because of a different sampling strategy. Additionally, the population under study was largely Caucasian, upper middle-class, well-educated, and probably healthier than most typical elderly populations, not only because these subjects are survivors from the original WCGS cohort but also because the rates of smoking and level of physical activity in this group tend to place them below and above national averages, respectively. Scant data exist on responses to snoring questions in other racial/ethnic minorities such as Hispanic (5, 24) and African-American (24). How ethnicity impacts upon use of the "don't know" response remains to be determined and is worthy of further investigation.