Sleep Duration and Hypertension Are Not Associated in the Elderly
Several large studies have shown that both short and long average sleep durations increase the risk of hypertension in adults. We investigated whether sleep duration is also associated with hypertension in the elderly. This cross-sectional study was conducted in 5058 participants of the population-based Rotterdam Study, aged 58 to 98 years. Blood pressure was measured at the research center. Hypertension was defined as a systolic blood pressure of ≥160 mm Hg and/or a diastolic blood pressure of ≥100 mm Hg or current use of antihypertensive medication. In all of the participants, sleep duration was assessed by self-report. In a subsample of 975 subjects, it was additionally measured with actigraphy, a validated method that infers wakefulness and sleep from the presence or absence of limb movement. After adjustment for age and gender and additionally for body mass index, smoking, depressive symptoms, sleep medication use, diabetes mellitus, myocardial infarction, and stroke, none of the odds ratios (varying from 0.54; 95% CI: 0.27 to 1.08; to 1.19; 95% CI: 0.89 to 1.58) reflected a significant association between sleep duration and hypertension, whether measured by self-report or actigraphy. This study strongly suggests that sleep duration is not associated with hypertension in the elderly.
Epidemiologic studies have consistently shown that self-reported short or long sleep duration is associated with increased mortality.1,2 Whether this association can be explained by elevated blood pressure or other cardiac factors remains uncertain. Gottlieb et al3 found that self-reported habitual sleep duration above or below the median of 7 to 8 hours per night is associated with an increased prevalence of hypertension, particularly at the extreme of <6 hours per night. Gangwisch et al4 reported that sleep durations of ≤5 hours per night were associated with a significantly increased risk of hypertension in subjects 32 to 59 years. The association between sleep duration and hypertension in the elderly is uncertain.
Our study aimed to investigate the associations of sleep duration with hypertension in an elderly community-dwelling population. We hypothesized that a markedly shorter or longer sleep duration than the median duration would be associated with a higher prevalence of hypertension. Our study was based on self-reported sleep duration, and in a subsample, sleep duration was also assessed with actigraphy.
This study is embedded in the Rotterdam Study, a population-based cohort study aimed at assessing the occurrence of and risk factors for chronic diseases in the elderly.5 In 1990, all of the inhabitants of a district of Rotterdam aged ≥55 years were invited, and 7983 agreed to participate (response: 78%). In 2000, the cohort was extended with 3011 participants from the same district (response: 67%), also aged ≥55 years. Between January 2002 and December 2005, 3115 participants from the original cohort and 2249 participants of the extended cohort underwent a home interview and visited the research center. For 5058 of these 5364 participants, valid data on both sleep duration and hypertension were available. From December 2004 onward, 1515 of these persons were asked to participate in the actigraphy study, 1076 (71%) of whom agreed. Both valid actigraphy data and hypertension data were available for 975 of these subjects. Together they contributed 5999 valid nights (mean: 6.15; SD: 1.1). The medical ethics committee of the Erasmus University Rotterdam approved the Rotterdam Study, and written informed consent was obtained from all of the participants.
Ascertainment of Hypertension
At the research center, trained examiners took 2 blood pressure measurements at the right brachial artery using a random-0 sphygmomanometer with appropriate adult cuff size and the participant in sitting position. The average of these 2 measurements was taken. Hypertension was defined as a systolic blood pressure of ≥160 mm Hg, a diastolic blood pressure of ≥100 mm Hg, or current use of antihypertensive medication.6 Medication use was ascertained by means of a standardized interview.
Sleep duration was assessed by the following question in the home interview: “During the past month, how many hours of actual sleep did you get at night?” The results were then divided in 5 categories of sleep duration. Because polysomnographic measurements for >1 night are not feasible in large studies, we used actigraphy to obtain objective sleep parameters. Actigraphy is a method that infers wakefulness and sleep from the presence or absence of limb movement. It estimates sleep parameters more accurately than sleep diaries and agrees reasonably with polysomnography.7 With actigraphy, it is possible to measure several nights in the home environment with little burden for the participants and with acceptable costs.
We used the Actiwatch (Cambridge Neurotechnology Ltd), an actigraph that can be worn like a watch. Participants wore the actigraph over a period of 5 to 7 consecutive nights. To calculate sleep parameters from the raw actigraphy data, we used the Actiwatch algorithm, which has been validated against polysomnography by Kushida et al.8 With this algorithm, a score is calculated for each 30-second epoch, taking into account the weighted value of previous and following epochs. We used a threshold of 20 to distinguish sleep from waking (high-sensitivity setting), because this yielded the best agreement with polysomnography with regard to total sleep time in the validation study by Kushida et al.8
Weight and height were measured at the research center, and body mass index was calculated (weight in kilograms/height in meters squared). Smoking (never, former, or current) and the use of sleep medication were determined in the home interview. Depressive symptoms were assessed with the Center for Epidemiologic Studies Depression Scale.9 Napping during daytime was assessed with a dichotomous question (“did you take a nap today?”) in the sleep diary that accompanied the actigraphy study. Diabetes was defined as a fasting serum glucose level of ≥7 mmol/L (measured at the research center) or the use of antidiabetic medication. Information on myocardial infarction and stroke was obtained from medical charts.
To assess the association of self-reported and actigraphically measured sleep duration with hypertension, we performed logistic regression analyses. We adjusted all of the analyses for age and gender (model 1). In addition, we adjusted for body mass index, smoking, depressive symptoms, napping during daytime (only actigraphy subgroup), sleep medication use, diabetes mellitus, myocardial infarction, and stroke, with imputed means for missing values (study population: n=223; actigraphy subgroup: n=20; model 2). We used the median sleep duration category as the reference category in all of the logistic regression analyses and repeated all of the analyses with lower cutoff points for the definition of hypertension (eg, 140 and 90 mm Hg, as used by Gottlieb et al3). Further analyses were performed in subjects who did not use antihypertensive medication, with systolic and diastolic blood pressure as continuous outcome variables in multiple linear regression models. In addition, we repeated all of the analyses for men and women separately and again after exclusion of participants who still had a paid job for ≥20 hours per week (4.6% of our study population and 6.2% of our actigraphy subgroup).
Of the 5058 participants in our study (57.5% female; mean age: 72.1 years; SD: 7.5 years; range: 58 to 98 years), 2485 (49.1%) had hypertension. Of the subsample of 975 subjects (52.1% female; mean age: 68.4 years; SD: 6.9 years; range: 59 to 97 years) who participated in the actigraphy study, 417 (42.8%) had hypertension.
In our actigraphy study subgroup, Pearson’s correlation coefficient between total sleep time as measured by questionnaire and mean actigraphically measured total sleep time was 0.25 (P<0.001). This indicates that perceived sleep duration is not necessarily the same as actigraphically measured sleep duration. It should be noted, however, that the questionnaire was administered a few weeks before the actigraphy study.
Nonresponders, who visited the research center but refused to participate in the actigraphy study, were, on average, 1.9 years older than responders (P<0.001) and were more likely to be female (P<0.001). Refusal to participate was neither associated with the risk of hypertension (P=0.08) nor with average self-reported sleep duration (P=0.42) but was often related to the unattractive appearance of the watch. The characteristics of the study population are presented in Table 1.
Table 2 shows the results of the logistic regression analyses. Because self-reported sleep duration was, on average, longer than actigraphically measured sleep duration, the actigraphy subgroup and the total study population have different reference categories for sleep duration. We did not find an association between either self-reported or actigraphically measured sleep duration and hypertension. Lower cutoff points for the definition of hypertension yielded very similar nonsignificant results (data not shown).
Sleep duration, measured by either self-report or actigraphy, was not associated with either systolic or diastolic blood pressure in a multiple linear regression analysis, both without and with a quadratic term to assess a possible U-shaped curve (Table 3). Repeating the analyses for men and women separately, and after exclusion of working participants, did not change the estimates (results not shown).
In a cross-sectional study of 5058 community-dwelling elderly subjects, we found no association between either self-reported or actigraphically measured sleep duration and hypertension or blood pressure. This adds important information to the findings of ≥2 other recently published, large population-based studies.3,4
Gangwisch et al4 reported that sleep durations of ≤5 hours per night were associated with a significantly increased risk of hypertension in subjects 32 to 59 years but that sleep duration and hypertension were unrelated, in an adjusted model, in people aged ≥60 years. Our findings are in agreement with this last result. Gangwisch et al4 suggested that the lack of an association between short sleep duration and hypertension incidence in older subjects could result from a lack of statistical power. In our study with 2485 subjects with hypertension aged ≥58 years, this is certainly not an appropriate explanation.
Gottlieb et al3 found that self-reported habitual sleep duration above as well as below the median of 7 to 8 hours per night is associated with an increased prevalence of hypertension in subjects aged 40 to 100 years (mean age: 63.1 years; SD: 10.7 years). Our study seems to contradict these findings, although the higher age of our participants should be noted.
The biological mechanisms behind the earlier reported association of short sleep duration and hypertension in adults are still unclear. It has been hypothesized that short sleep duration causes metabolic changes that are associated with increased body mass index and elevated blood pressure. Possibly, this mechanism occurs in particular as a result of voluntary sleep restriction because of exogenous factors, such as a fixed time to get up and go to work. It might not occur when short sleep duration is the result of insomnia or a decreased need of sleep. That would explain the lack of association between short sleep duration and hypertension in older adults, who are often retired. Another possibility is that older people compensate for their nightly short sleep duration by daytime napping. However, in the actigraphy subgroup, we adjusted for napping, and this did not markedly change the results.
There is evidence for a strong association between sleep apnea and hypertension,10 and sleep apnea may also cause variation in reported or observed sleep durations. However, the association between sleep apnea and hypertension seems to be age dependent; in the Sleep Heart Health Study, it was not found among those aged ≥60 years.11 This may to some extent, explain the lack of association in the relationship between sleep duration and hypertension in the elderly, if sleep apnea is an important mechanism underlying this association in younger people.
Neither the study by Gangwisch et al4 nor the study by Gottlieb et al3 used objective measures of sleep duration. If the perception of sleep duration is distorted by factors that are unrelated to hypertension, this may cause nondifferential misclassification and, thus, bias the results toward the null value. If, however, misperception is related to risk factors for hypertension, the resulting differential misclassification can bias the results in either direction. Our actigraphic measurements of sleep duration confirmed the findings of the self-report data. This adds to the credibility of the results, because actigraphic measurements are not influenced by an individual’s perception of sleep duration. Likewise, other strengths of our study are the measured (as opposed to self-reported) blood pressure, body weight, and height.
Several limitations of our study need to be discussed. First, the participation rate in our study was not 100%. However, the occurrence of selection bias is unlikely, because (non)participation was neither associated with the self-reported average sleep duration nor with the risk of hypertension. Second, wrist actigraphy does not perfectly measure sleep duration when compared with the gold standard of polysomnography. However, an important advantage of actigraphy over polysomnography is that it is unlikely to affect bedtime, sleep latency, and sleep duration. Third, we could not rule out the possibility of residual confounding, which may have obscured the associations under study.
Because of the cross-sectional design of this study, we cannot determine causal relationships or rule out bidirectional relationships. The problems with cross-sectional data, however, tend to be with spurious associations. So, the finding of a lack of any association between sleep duration and hypertension or blood pressure makes it unlikely that there actually is a causal association in the elderly.
The results from this study suggest that sleep duration and hypertension are not related in an elderly population. This implies that, above the age of 60, hypertension probably does not contribute to the higher mortality found in short and long sleepers. Future research into the associations of sleep duration and mortality should, therefore, preferably focus on mechanisms other than hypertension.
The contributions of all of the participants, general practitioners, and pharmacists of the Ommoord district to the Rotterdam Study are gratefully acknowledged.
Sources of Funding
The Rotterdam Study is supported by the Erasmus Medical Center and Erasmus University Rotterdam; the Netherlands Organization for Scientific Research; the Netherlands Organization for Health Research and Development; the Research Institute for Diseases in the Elderly; the Ministry of Education, Culture and Science; the Ministry of Health, Welfare and Sports; the European Commission (DG XII); and the Municipality of Rotterdam. Additional funding for the work on sleep duration was provided by Netherlands Organization for Health Research and Development grant 4200.0019.
- Received April 23, 2007.
- Revision received May 9, 2007.
- Accepted June 22, 2007.
Gangwisch JE, Heymsfield SB, Boden-Albala B, Buijs RM, Kreier F, Pickering TG, Rundle AG, Zammit GK, Malaspina D. Short sleep duration as a risk factor for hypertension: analyses of the First National Health and Nutrition Examination Survey. Hypertension. 2006; 47: 833–839.
Radloff LS. The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Meas. 1977; 1: 385–401.
Haas DC, Foster GL, Nieto FJ, Redline S, Resnick HE, Robbins JA, Young T, Pickering TG. Age-dependent associations between sleep-disordered breathing and hypertension: importance of discriminating between systolic/diastolic hypertension and isolated systolic hypertension in the Sleep Heart Health Study. Circulation. 2005; 111: 614–621.