Author + information
- Received August 23, 2000
- Revision received March 19, 2001
- Accepted March 29, 2001
- Published online July 1, 2001.
- Kazuomi Kario, MD, PhD, FACC, FACP∗,b,
- Jonathan N Tobin, PhD∗,‡,* (, )
- Leslie I Wolfson, MD§∥,
- Robert Whipple, MA, RPT§∥,
- Carol A Derby, PhD‡,
- Devender Singh, MD§,
- Paul R Marantz, MD, MPH‡ and
- Sylvia Wassertheil-Smoller, PhD‡
- ↵*Reprint requests and correspondence:
Dr. Jonathan N. Tobin, Clinical Directors Network, Inc., 54 West 39th Street, 11th Floor, New York, New York 10018
We investigated prospectively the relationships among falls, physical balance, and standing and supine blood pressure (BP) in elderly persons.
Falls occur often and adversely affect the activities of daily living in the elderly; however, their relationship to BP has not been clarified thoroughly.
A total of 266 community-dwelling elderly persons age 65 years or over (123 men and 143 women, mean age of 76 years) were selected from among residents of Coop City, Bronx, New York. Balance was evaluated at baseline using computerized dynamic posturography (DPG). During a one-year follow-up, we collected information on subsequent falls on a monthly basis by postcard and telephone follow-up.
One or more falls occurred in 60 subjects (22%) during the one-year follow-up. Women fell more frequently than men (28% vs. 16%, p < 0.03), and fallers were younger than nonfallers. Fallers (n = 60) had lower systolic BP (SBP) levels when compared with nonfallers (n = 206) (128 ± 17 vs. 137 ± 22 mm Hg for standing, p < 0.006; 137 ± 16 vs. 144 ± 22 mm Hg for lying, p < 0.02), whereas diastolic BP was not related to falls. Falls occurred 2.8 times more often in the lower BP subgroup (<140 mm Hg for standing SBP) than in the higher BP subgroup (≥140 mm Hg, p < 0.0003), and gender-related differences were observed (p = 0.006): 3.4 times for women (p < 0.0001) versus 1.9 times for men (p = 0.30). Loss of balance, as detected by DPG, did not predict future falls and was also not associated with baseline BP levels. Multiple logistic regression analysis demonstrated that female gender (relative risk [RR] = 2.1, p = 0.02), history of falls (RR = 2.5, p = 0.008) and lower standing SBP level (RR = 0.78 for 10 mm Hg increase, p = 0.005) were independent predictors of falls during one year of follow-up.
Lower standing SBP, even within normotensive ranges, was an independent predictor of falls in the community-dwelling elderly. Elderly women with a history of falls and with lower SBP levels should have more attention paid to the prevention of falls and related accidents.
One third of the elderly (age 65 years or more) experience one or more falls each year (1–5). In the elderly population, fall-related injuries are the leading cause of death and are increasing (6). Even falls that do not lead to injury often begin a cycle of fear that leads to limitations of activities in daily living and eventually results in loss of independence in normal activities of self-care (7). Previous studies of the elderly have identified several risk factors for falls, which have been classified as intrinsic (those related to the individual) and extrinsic (those associated with environmental factors) (3). However, the relationship between one intrinsic risk factor, blood pressure (BP), and falls in the elderly has not been thoroughly investigated.
Transient cerebral ischemia due to diminished cerebral blood flow is one of the recognized causes of falls in the elderly (8–10), whose cerebral blood flow and reserve capacity are markedly decreased by aging or hypertension. Ischemic brain white matter lesions detected by computer-assisted tomography scanning are present more often in frequent fallers than in nonfallers (11). A recent study using brain magnetic resonance imaging (MRI) in an elderly hypertensive population showed that extreme-dippers, those with nocturnal hypotension (marked nocturnal BP decline), are more likely to have advanced silent ischemic cerebrovascular disease than are dippers, those with appropriate nocturnal BP decline (12). Other brain MRI studies have also demonstrated silent cerebrovascular disease in elderly subjects with relative and transient hypotension (orthostatic hypotension and postprandial hypotension) (13,14). A prospective study of elderly subjects demonstrated a J-curve relationship between BP levels and stroke events (15,16). These studies indicate that hypotension in the elderly may be a risk factor for cerebral ischemia, which might contribute to falls.
To test this hypothesis, we examined the prospective relation of BP and falls in the elderly in a community-based cohort study.
Subjects and methods
The study was approved by the Institutional Review Board of the Albert Einstein College of Medicine. We first sent recruitment invitations with a brief self-administered medical history questionnaire to a total of 6,486 elderly subjects from Bronx, New York, at least 65 years of age, who had previously been screened and found ineligible for the Systolic Hypertension in the Elderly Project (SHEP) (17). Of the 867 subjects who responded, 695 met preliminary eligibility criteria (18). Of these 695 subjects, 411 were scheduled for a screening visit. Subjects were excluded from further evaluation if any of the following conditions or treatments applied to them: terminal illness; radiation therapy or chemotherapy; severe orthopedic or arthritic problems requiring joint replacement, braces, cane or walker; severe joint disease or pain or limp; claudication secondary to vascular disease or spinal stenosis; amputation of lower extremities; visual acuity worse than 20/50 or visual field defect on examination; history of findings consistent with neurological diseases that impair sensory and/or motor function (Parkinson’s disease, stroke, and so on); history of transient ischemic attacks or seizures; arrhythmia on resting electrocardiography (bradycardia below 45 beats/min; narrow QRS tachycardia over 120 beats/min; wide QRS tachycardia over 100 beats/min; frequent premature ventricular contraction; second- and third-degree heart block); symptomatic orthostatic hypotension (≥20 mm Hg drop in systolic BP [SBP] with accompanying complaints of dizziness upon rising by history or examination); or regular usage of sedatives, tranquilizers, hypnotics, butyrophenones or tricyclic or atypical antidepressants.
From the 411 initial subjects, 266 were recruited and completed the baseline evaluation. There were no differences in age, gender, ethnicity and proportion of subjects who reported a history of hypertension among the total subjects, the initial subjects and the final subjects (data not shown).
At the Balance and Gait Evaluation Laboratory, all subjects underwent a standardized evaluation that included a self-administered medical history and prescription drugs questionnaire followed by an interview to verify self-reports. Physical, neurological and arthritic examinations were performed by a trained clinician. Activities of daily living (19), mobility (20)and cognitive function using the Blessed Mental Status (21)were also evaluated.
BP measurement and hypertension status
All BPs were measured by one of two observers who were trained and certified in BP measurement using standard mercury sphygmomanometers in accordance with the procedures of the National Heart, Lung, and Blood Institute Hypertension Detection and Follow-up Project modified for SHEP (17). Subjects had their BP measured supine, followed by two standing BP measurements, immediately after standing and 2 min after standing. The average standing BP levels and postural BP change (average of standing BP and supine BP) was calculated from these three BP measures.
Individuals were classified on the basis of their baseline observed supine BP measurement and BP medication status. Subjects who were not hypertensive on examination and who were not taking any antihypertensive medication were considered normotensives. Subjects who were taking antihypertensive medications were considered treated hypertensives and those with supine BP ≥140 mm Hg systolic and/or ≥90 mm Hg diastolic in an untreated condition were considered untreated hypertensives.
Computerized dynamic posturography
Subjects were tested on a computerized dynamic posturography (DPG) platform (EquiTest, Clackamas, Oregon). Computerized DPG measures the timing and magnitude of surface forces produced in response to perturbations of the support surface and visual surround (18). The protocol consists of the Sensory Organization Test (SOT), which assesses the three sensory components of balance under a variety of altered visual and support surface conditions, and the Motor Coordination Test (MCT), which evaluates the automatic motor reactions provoked by unexpected support surface perturbations (22,23). During the SOT, subjects were tested in six visual and support surface conditions, and during the MCT automatic postural reactions were provoked through a series of sudden anterior and posterior support surface translations as well as toe-up and toe-down rotations. Loss of balance is determined when a subject’s sway exceeds the limits of stability (8.5° anteriorly and 4° posteriorly), requiring a step and/or support by the examiner to prevent a loss of balance, fall or injury, at least once during the SOT (six different conditions) or MCT.
Follow-up of falls
All subjects were given a set of 12 monthly falls report postcards and instructed to send one card at the end of each month, or sooner if they fell. We completed the collection of fall information every month by postcard or by telephone. The stamped, pre-addressed postcards contained questions about whether the subjects had experienced any falls in the previous month, and if so, to indicate the circumstances of the fall. Subjects who failed to return a postcard during a given month were telephoned and interviewed to obtain the information. All subjects who reported a fall during a month were contacted by telephone to verify the recorded information. Fallers were defined as those who experienced one or more falls during the first one-year follow-up period, and nonfallers were those who did not report any fall in all 12 months of follow-up.
Relative risks (RR) of falling during the first year of follow-up and 95% confidence intervals were calculated. All categorical variables were compared using the continuity-adjusted chi-square (or Fisher exact test) and continuous variables were compared using ttests. Two-way repeated measures analysis of variance was performed to detect differences among groups, and Bonferroni’s test was used for multiple pairwise comparisons of means among groups. Pearson’s correlation coefficient was used to measure the relationships between the standing BP and other variables. Multiple logistic regression analysis was performed to examine the impact of standing SBP, antihypertensive medication status, previous history of falls, age and gender upon the probability of falling during the first year of follow-up period. In a prospective study, odds ratios calculated from coefficients of multiple logistic regression analysis provide an unbiased estimate of the RR. All significance tests were two-tailed, with a p < 0.05 considered significant.
A total of 266 elderly subjects, 123 men (46%) and 143 women (54%), with mean ± SD age of 75.5 ± 5.0 years, were recruited and followed for one year. The prevalence of hypertension was 62%, and of these hypertensive subjects, 72% were aware of their hypertension, 35% were taking regular antihypertensive medication, and 38% were well controlled (<140/90 mm Hg). Overall, the mean supine BP was 143 ± 21/81 ± 9.7 mm Hg, and the mean of two standing BPs was 136 ± 20/81 ± 10 mm Hg. Supine BP ranged from 100 to 220 mm Hg systolic and from 60 to 110 mm Hg diastolic.
Table 1shows the measures of falls in this sample. All study subjects were followed up for at least 12 months (21 months of the average follow-up period). With 5,583 person-months of follow-up, 31% of the subjects experienced one or more falls, and of those experiencing a fall, approximately 45% had two or more episodes. During the first one year follow-up period, falls occurred in 22% of subjects (n = 60). The incidence rate was higher than the baseline prevalence rate of fall history in the last year (14%). The following activities were reported during falls (32 reports were available): using steps (n = 9); crossing the street (n = 10); in the bathroom (n = 1), other activity (n = 5); unknown (n = 12). Fifteen falls resulted in an injury. Prevalence of loss of balance on DPG was 28%.
Table 2compares the clinical characteristics and medical conditions of those who did and did not fall during the one-year follow-up (fallers and nonfallers). Younger elderly subjects (age 65 to 74 years) experienced falls more often than older elderly subjects (75 or more years) (p < 0.01). The younger elderly group had lower levels of mobility impairment (score ≤27) than the older elderly group (35% vs. 60%, p < 0.0001). Other factors did not differ between the younger elderly and older elderly subgroups. Falls were more common in women than in men (p < 0.03). Of the various medical conditions measured, only a history of cancer or previous falls were significant predictors of subsequent falls in the first year of follow-up. The number of medical conditions, activities of daily living, mobility and cognitive function were not associated with future falls. Age, gender, the number of medical conditions, routines of daily living, mobility, cognitive function, loss of balance by DPG and history of falling were not associated with baseline standing SBP or diastolic BP (DBP) (data not shown), except for a negative association between age and standing DBP (r = −0.162, p < 0.01).
Table 3shows the relationship of falls during follow-up with previous fall history and loss of balance on DPG. The history of falls in the past year was the strongest predictor of falls during the following one year (p < 0.001). The relative risk for falls associated with loss of balance triggered by DPG was smaller (1.5) and did not achieve statistical significance (p = 0.13).
Figure 1shows BP levels during the active standing test. Fallers compared with nonfallers were lower in supine SBP just after standing and 2 min after standing (p < 0.001 for SBP for three measures, p = 0.01 for faller status and p = 0.69 for the interaction between systolic BP and faller status). There was no significant difference in DBP during the standing test between the fallers and nonfallers. Figure 2shows the incidence of falls in each subgroup classified according to standing SBP. Falls occurred 2.8 times more often in the lower BP subgroup (<140 mm Hg for standing SBP) than in the higher BP subgroups (≥140 mm Hg, p < 0.0003) and gender-related differences were also observed (p = 0.006): Falls occurred 3.4 times more often in women (p < 0.0001) and only 1.9 times more often for men (p = 0.30). The multiple logistic analysis showed that falls were significantly associated with female gender (RR = 2.8, p = 0.006) but not with higher standing SBP (>140 mm Hg) (RR = 0.49, p = 0.20), nor was the interaction of female gender and higher standing SBP significant (p = 0.24).
There were no associations observed between the degree of change from supine SBP and DBP levels to those just after standing between fallers and nonfallers (Table 4). When we defined asymptomatic orthostatic hypotension as a ≥20 mm Hg for SBP decline or ≥10 mm Hg for DBP decline (without any accompanying symptoms), the incidence of falls was not different.
Table 5shows the relationship between antihypertensive therapy and falls on follow-up. Falls were significantly less common in both treated (17%) and untreated (20%) hypertensive subjects compared with normotensives (34%) (p < 0.003 and p < 0.02, respectively). Within hypertensive subjects, there were no differences in falls between the treated and the untreated groups, and there was no difference in the incidence of falls among different antihypertensive agents.
In the multiple logistic regression analysis (Table 6, Model 1), female gender (RR = 2.1, p = 0.02) and history of falls (RR = 2.5, p = 0.008) were independent positive predictors of falls. Standing SBP was a negative predictor of falls, and a 10 mm Hg increase in standing SBP reduced falls by 22% (p = 0.005). Age, medical conditions and antihypertensive medication were not independent predictors of falls during the first year of follow-up. Excluding the antihypertensive medication term (Model 2), the associations were essentially the same as Model 1, except that age was significant (p < 0.05). When we added the interaction term between supine SBP and antihypertensive medication in Model 1, this interaction was not significant. When the supine SBP was substituted for standing SBP in the same models of Table 6, a 10 mm Hg increase in supine SBP reduced falls by 18% (p = 0.02) for Model 1 and by 19% (p = 0.02) for Model 2. When the mobility score was entered into Model 2, the RR of age became insignificant (RR = 0.94, p = 0.17).
In this prospective study of a relatively healthy elderly, urban community-dwelling population, we found that lower SBP is an independent risk for falls, especially for women. We carefully excluded subjects at high risk for falls, such as those with symptomatic orthostatic hypotension, visual impairment, and history of cerebrovascular disease, etc.
The unique features of this study are the prospective assessment of falls and standardized BP measurement. Prior reports indicate that the recall of falls is poor, even in cognitively intact elderly adults, and approximately 30% of cognitively normal elderly subjects are unable to recall documented falls three months later (10,23). To improve the reliability of measuring falls, we collected information on falls each month.
Although there are numerous causes of falls in the elderly (1–3), abnormalities in BP regulation and syncope (8,10)are recognized as significant contributors to falls. Hypotensive responses to common daily activity may contribute to the high incidence of falls among the elderly (9). Orthostatic hypotension (24,25)and postprandial hypotension (26,27)are established risk factors for both syncope and falls in the elderly, probably through reduced cerebral perfusion. Recent studies using brain MRI have demonstrated that these hypotensive conditions are commonly accompanied by silent ischemic cerebral disease (cerebral infarction and/or advanced deep white matter lesions) (13,14). Brain MRI studies have shown that nocturnal hypotension is also associated by silent ischemic cerebrovascular disease in the elderly (12). Because silent ischemic cerebral disease is more advanced in the elderly with frequent falls than in subjects without falls (11), cerebral hypoperfusion due to hypotension may be associated with falls in the elderly.
High BP is an important risk factor for both cerebral atherosclerosis and silent cerebrovascular disease (28,29). However, the mechanisms of syncope and cerebral target organ damage may be different. The hypertension-related ischemic target organ damage may be attributed to chronic atherosclerosis through long-term pressure overload and sheer stress on the blood vessels. However, syncope can occur from a single low BP episode when the person is awake. Thus, high BP level is a chronic risk factor for ischemic organ damage, and low BP level is an acute risk factor for syncope or an ischemic episode, especially in the elderly. The hypotensive effects on cerebral ischemic damage may be greater in elderly subjects who are likely to have advanced cerebral atherosclerosis and diminished cerebral perfusion reserve capacity. Both high and low BP predict subsequent dementia status (30–32). In this study, the falls incidence in those with standing SBP <120 mm Hg was lower than in those with SBP 120 to 139 mm Hg. It is possible that those with the lowest standing SBP may have less advanced cerebral atherosclerosis.
Symptomatic orthostatic hypotension is a well-recognized risk factor for falls in the elderly. We did not find an increase in the falls rate in subjects with either asymptomatic SBP (≥20 mm Hg decline) or DBP (≥10 mm Hg decline) orthostatic hypotension. As the aim of the present study was to find determinants of falls in relatively healthy community-dwelling elderly subjects, we excluded patients with symptomatic orthostatic hypotension. The occurrence of orthostatic symptoms with BP decline may be an important for a risk predictor for falls in this population. As one of the limitations of our study, our standing BP was measured 2 min after standing, and there is the possibility that fallers might have experienced a further, more severe reduction in SBP after more prolonged standing (delayed orthostatic hypotension) (33). Delayed orthostatic hypotension, one cause of syncope in the elderly, is more common than acute orthostatic hypotension in elderly subjects (34). When syncope is induced in those with delayed orthostatic hypotension during tilt-table studies, it usually occurs more than 10 min after induction of the head-up tilt (35).
Gender difference and age
In this study, a marked gender difference was found in the incidence of falls and the relationship between falls and lower BP levels. Although clinically overt and silent cerebral infarction is more common in men, elderly women are more likely to fall (5,36). Also, the tolerance of transient cerebral hypoperfusion may be lower in women than in men. A recent study using brain MRI shows that normotensive women with nocturnal hypotension more commonly had silent cerebral ischemic lesions (37). In this study, younger elderly subjects between 65 and 74 years of age developed falls more frequently than the older elderly ≥75 years of age. The relatively healthy older subjects who survive from cardiovascular disease may be selected in this population because of a survivorship effect. As for the elderly in nursing homes, the fall rate increased with advancing age (38), whereas in other community-based studies on falls in the elderly, age was not always a significant predictor (4,5). Additionally, mobility levels may have affected the age-related differences in the falls rate in this study. The older elderly group had more decreased mobility and may have had less opportunity to experience falls than did the younger elderly group.
The overall benefit of antihypertensive therapy is well established in the elderly hypertensive population (39). However, previous studies have suggested that antihypertensive therapy was positively associated with falls (3,8). The volume depletion due to diuretics or fluctuation due to short-acting antihypertensive drugs might trigger a fall in the hypertensive population. In the present study, it is noteworthy that BP was well controlled in only 38% of hypertensives, although this is better than the 27% hypertension control rate reported in the general population (40). Overall, antihypertensive therapy did not increase the falls in the hypertensive study population. Although falls were 2.5 times more common in those treated with short-acting calcium antagonists than in those treated with other antihypertensive agents, this difference did not reach statistical significance, probably because of the small number of subjects.
We found that lower normal standing SBP is a predictor of falls in the elderly population, especially in women. Special care for the prevention of falls should be addressed for elderly subjects with lower standing SBP.
We are grateful to M. Donald Blaufox, MD, PhD, for providing access to SHEP screening data, Zirel Sweezy and Sara Pressel for standardized blood pressure training and certification assistance, and to Thomas G. Pickering, MD, DPhil (Mount Sinai School of Medicine) and Joseph E. Schwartz, PhD (SUNY/Stony Brook School of Medicine) for comments and suggestions, and to the anonymous reviewers for their helpful suggested revisions.
☆ This study was supported in part by grants to the Albert Einstein College of Medicine of Yeshiva University from the National Institute on Aging (#AG03949 [Teaching Nursing Home] and #AG05643), and the National Heart, Lung, and Blood Institute (#1-HC-48055 for the Systolic Hypertension in the Elderly Project [SHEP]).
- blood pressure
- diastolic blood pressure
- dynamic posturography
- Motor Coordination Test
- magnetic resonance imaging
- relative risk
- systolic blood pressure
- Systolic Hypertension in the Elderly Project
- Sensory Organization Test
- Received August 23, 2000.
- Revision received March 19, 2001.
- Accepted March 29, 2001.
- American College of Cardiology
- Campbell A.J.,
- Borrie M.J.,
- Spears G.F.
- Shaw F.E.,
- Kenny R.A.
- Matsubayashi K.,
- Okumiya K.,
- Wada T.,
- et al.
- Vokó Z.,
- Bots M.L.,
- Hofman A.,
- Koudstaal P.J.,
- Witteman J.C.M.,
- Breteler M.M.B.
- Frohlich E.D.
- Labarthe D.R.,
- Blaufox M.D.,
- Smith W.M.,
- et al.
- Brook R.H.,
- Ware J.E.J.,
- Davies-Avery A.,
- et al.
- Blessed G.,
- Tomlinson B.E.,
- Roth M.
- Craig G.M.
- Kwok T.,
- Liddle J.,
- Hastie I.R.
- Kario K.,
- Matsuo T.,
- Kobayashi H.,
- Asada R.,
- Matsuo M.
- Guo Z.,
- Viitanen M.,
- Fratiglioni L.,
- Winblad B.
- O’Neill T.W.,
- Varlow J.,
- Silman A.J.,
- et al.
- Watanabe N.,
- Imai Y.,
- Nagai K.,
- et al.
- ↵National High Blood Pressure Education Program. The sixth report of the Joint National Committee on Prevention, Detection, Evaluation and Treatment of High Blood Pressure. National Heart, Lung, and Blood Institute; NIH Publication No. 98-4080, 1998.