Abstract
BACKGROUND
One-third of older adults fall each year. Orthostatic hypotension (OH) has been hypothesized as an important risk factor for falls, but findings from prior studies have been inconsistent.
METHODS
We conducted a prospective study of the association between baseline OH (1987–1989) and risk of falls in the Atherosclerosis Risk in Communities (ARIC) Study. Falls were ascertained during follow-up via ICD-9 hospital discharge codes or Centers for Medicare & Medicaid Services claims data. OH was defined as a drop in systolic blood pressure (SBP) ≥20mm Hg or diastolic blood pressure (DBP) ≥10mm Hg within 2 minutes of moving from the supine to standing position. Changes in SBP or DBP during OH assessments were also examined as continuous variables.
RESULTS
During a median follow-up of 23 years, there were 2,384 falls among 12,661 participants (mean age 54 years, 55% women, 26% black). OH was associated with risk of falls even after adjustment for demographic characteristics and other risk factors (hazard ratio (HR): 1.30; 95% confidence interval (CI): 1.10, 1.54; P = 0.002). Postural change in DBP was more significantly associated with risk of falls (HR 1.09 per −5mm Hg change in DBP; 95% CI: 1.05, 1.13; P < 0.001) than postural change in SBP (HR 1.03 per −5mm Hg change in SBP; 95% CI: 1.01, 1.05; P = 0.002).
CONCLUSIONS
In a community-based, middle-aged population, OH, and in particular, postural change in DBP, were independent risk factors for falls over 2 decades of follow-up. Future studies are needed to examine OH thresholds associated with increased risk of falls.
Keywords: ARIC, blood pressure, epidemiology, fall, hypertension, orthostatic hypotension, prospective cohort.
Falls are a major cause of morbidity and mortality among older adults.1,2 Nearly, one-third of persons aged 65 or older falls each year and the rate of falls in the United States is rising.1,2 Moreover, falls are a significant reason for emergency room visits3 and predispose to the risk of fracture,4 posing a substantial burden on the US health system, estimated at approximately $23 billion yearly.5 Given the aging US population, identifying risk factors for falls and tailoring interventions (e.g., physical therapy, prescription medication choices, or environment-based interventions like handrails) to prevent falls are recognized public health priorities.
Orthostatic hypotension (OH) is a controversial risk factor for falls that has important implications for clinical guidelines on blood pressure management. The recently published SPRINT trial suggested that more intensive blood pressure management was associated with a lower risk of OH, but not a decreased risk of injurious falls.6 OH has been traditionally thought to disproportionately impact older adults. The prevalence of OH increases with age,7 reaching as high as 20% in adults age 75 years and older.8 Furthermore, OH is associated with a history of falls as well as greater morbidity from falls in older adults.8–10 Despite the focus in the literature on older adults, OH is also quite prevalent among middle-aged adults (as high as 15.9% in 1 study11), who have a notably lower fall rate. While in the short-term, middle-aged OH may not be a major contributor to risk of serious falls, it is unknown whether it might be associated with risk of falls over time as people age.
Thus, the aims of this prospective study were to: (i) determine whether OH in middle age was associated with risk of falls over the lifespan; (ii) characterize the continuous relationship between postural changes in systolic blood pressure (SBP) or diastolic blood pressure (DBP) and fall risk; and (iii) identify whether participant demographic characteristics or cardiovascular risk factors (obesity, hypertension, or diabetes) modified the association between OH and falls. We hypothesized that OH would be significantly associated with falls and this association would be more pronounced among older persons with cardiovascular risk factors.
METHODS
Study population
The Atherosclerosis Risk in Communities (ARIC) Study is a community-based prospective cohort of 15,792 adults originally enrolled from 1987 to 1989 from 4 US communities and followed for over 2 decades.12–14 Participants were ages 45 to 64 at the time of enrollment. Physical examinations, medical interviews, and laboratory tests were conducted as part of the original ARIC protocol. Of the original 15,792 baseline participants, our study population was limited to 12,661 participants, after excluding those who did not have an OH assessment at baseline (1987–1989) (N = 2,599) and who were missing relevant covariate data at baseline (N = 532).
Written informed consent was obtained from all participants and the study protocol was approved by institutional review boards at all study sites.
Exposure: orthostatic hypotension
During the baseline visit, supine SBPs and DBPs were measured with a Dinamap 1846 SX oscillometric device in participants, who had been lying for 20 minutes.15,16 Measurements were taken approximately every 30 seconds for 2 minutes (range of 2–5 supine measurements with at least 4 measurements obtained for 90% of participants). Participants were subsequently asked to stand and once both their feet were planted on the ground, standing SBP and DBP were measured approximately every 30 seconds for 2 minutes (range of 2–5 standing measurements with at least 4 measurements recorded for 91% of participants). Change in SBP or DBP was defined as the difference between the average of the standing and the supine blood pressure measurements, excluding the first standing measurement to avoid restabilization effects that occur immediately after standing.17 OH was defined using the consensus definition as a decrease in either SBP or DBP of at least 20 or 10mm Hg, respectively.18,19
Outcome: falls resulting in hospitalization or a physician visit
A fall was defined as the first occurrence of any fall-related hospitalization or claim for inpatient or outpatient services after the baseline visit. Falls were identified via 2 sources: (i) active surveillance of all hospitalizations for all ARIC participants and (ii) linkage to Centers for Medicare and Medicaid Services (CMS) claims data from 1991 to 2013. Details related to falls ascertainment may be found in the Supplement (Supplementary Methods). Falls were grouped in the following categories using ICD-9 codes: (i) falls unspecified, defined using codes E888.0, E888.1, E888.8, and E888.9; (ii) falls on the same level (including slipping and collision), defined using codes E885.0–E885.4, E885.9, E886.0, and E886.9; and (iii) falls from a different level (including stairs/steps from both slipping and collision), defined using codes E880.0, E880.1, E880.9, E881.0, E881.1, E882.0, E883.0–E883.2, E883.9, E884.0–E884.6, and E884.9.
In 2013–2014, 6,192 ARIC participants of our analytic dataset were asked whether they had a fall in the preceding 6 months. While these falls events were not included in our primary analysis given the restricted time window embedded in this question, we compared self-reported falls to falls using CMS ICD-9 codes that were identified in the corresponding 6-month period as a validation analysis.
Baseline covariates of interest
Data were collected at baseline by trained study personnel using standardized protocols, including extensive quality control measures.12–14 Covariates were selected a priori based on hypothesized relationships between OH, assessed at the baseline clinical examination and subsequent risk of falls. Age, sex, race, alcohol use (never, former, current), education level (less than high school, high school degree, or vocational school, at least some college or professional school), smoking status (never, former, current), history of stroke, hypertension medication use in last 2 weeks, diuretic medication use, antidepressant medication use, sedative medication use, hypnotic medication use, antipsychotic medication use, and anticholinergic medication use were self-reported. Participants’ leisure time physical activity was assessed via the ARIC/Baecke Physical Activity questionnaire.20
Diabetes status was defined based on a fasting blood glucose ≥126mg/dl, a nonfasting blood glucose ≥200mg/dl, a self-reported diagnosis of diabetes, or self-reported diabetes medication use. Hypertension was defined as a SBP ≥140mm Hg or DBP ≥90mm Hg or self-report of antihypertensive medication use. Congestive heart failure at baseline was defined as active use of medications to treat heart failure or self-reported symptoms associated with heart failure based on the Gothenburg criteria.21 Coronary heart disease at baseline was based on electrocardiograms (silent myocardial infarction), self-report, or self-reported history of previous coronary procedures.
Resting SBP and DBP was estimated from a mean of the second and third of 3 measurements obtained with a random-zero mercury sphygmomanometer. Heart rate was determined from a baseline electrocardiogram taken at rest. Body mass index was calculated using height and weight measurements. Serum creatinine was measured by the modified kinetic Jaffé method. Estimated glomerular filtration rate was determined from serum creatinine using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation.22
Statistical analysis
Baseline characteristics were compared using means and proportions according to the presence of OH as well as according to whether a participant experienced a fall. Absolute risk was determined via the cumulative incidences (at 5-year increments) or incidence rates (per 1,000 person-years) for participants with and without OH. Age was used as the time axis in addition to follow-up time, given its importance as a risk factor for falls. Nested Cox proportional hazard models, using follow-up time (not age) as the time axis were used to examine the independent association of OH, postural change in SBP, and postural change in DBP with risk of falls. Follow-up time (not age) was used as the time axis in these analyses to allow for an age-based stratified analysis. To facilitate interpretation and comparison, postural change in SBP and DBP were presented per −5mm Hg and per –1 SD. Model 1 was adjusted for age, sex, and race–study center. Model 2 was adjusted for all variables in Model 1 plus heart rate, body mass index, estimated glomerular filtration rate, diabetes, hypertension, alcohol use, education level, smoking status, physical activity, coronary heart disease, history of stroke, and congestive heart failure. Model 3 was adjusted for all variables in Model 2 plus hypertension medication use in last 2 weeks, diuretic use, antidepressant use, sedative use, hypnotic use, antipsychotic medication use, and anticholinergic medication use. Model 4 was adjusted for all the variables in Model 3 plus resting SBP and resting DBP. Finally, Model 5 included all the covariates of Model 4, but also included both postural change in SBP and DBP to allow a direct comparison of these 2 exposures. The continuous associations between postural change in SBP or DBP, and risk of fall, were characterized using restricted cubic splines with 4 knots specified via Harrell’s method.23
We also tested for interactions by demographic characteristics (age, sex, race) or cardiovascular risk factors (obesity, hypertension, and diabetes) in Model 4. In addition, we performed sensitivity analyses examining the independent association of OH using different definitions of falls based on ARIC hospitalization surveillance data alone with follow-up starting from baseline, CMS claims alone among those with continuous enrollment with follow-up starting on the day of enrollment, and fee-for-service claims alone among those with continuous enrollment with follow-up starting on the day of enrollment. Furthermore, we repeated analyses using different ICD-9 groupings to define falls: (i) falls unspecified, (ii) falls on the same level (including slipping and collision), and (iii) falls from a different level (including stairs/steps from both slipping and collision).
All analyses were conducted using Stata 14.0 (StataCorp LP, College Station, TX).
RESULTS
The study population (N = 12,661) included 55% women and 26% Black participants with a mean baseline age of 54 (SD, 6) years (Table 1). Five percent of the study population had OH at baseline. During a median of 23 years of follow-up, there were 2,384 (19%) falls (occurring at 2.6 months to 26.4 years from date of the baseline examination); 22% of falls were identified from ARIC hospital surveillance, while 78% were identified from CMS Medicare claims (Supplementary Table S2). Of the falls identified from claims, 76% represented claims for outpatient services. The median crude time to a fall was 16.9 years (interquartile interval of 12.5–20.8) among participants without OH (number of falls was 2,234 of 12,010 participants) and 14.2 years (interquartile interval of 10.9–18.7) among participants with OH at baseline (number of falls was 150 of 651 participants). The most common ICD-9 code for fall was E885.9, i.e., “accidental fall from other slipping, tripping, or stumbling” (Supplementary Table S1). In our validation study comparing self-reported falls to those identified using CMS ICD-9 codes, we determined the sensitivity and specificity of CMS ICD-9 codes to be 5.3% and 99.6%, respectively (Supplementary Table S3).
Table 1.
Baseline population characteristics by OH, mean (SD), or N (%)
| Overall (N = 12,661) | No OH (N = 12,010) | OH (N = 651) | |
|---|---|---|---|
| Age, year | 54.1 (5.7) | 53.9 (5.7) | 57.4 (5.3) |
| Female, N (%) | 6,961 (55.0) | 6,604 (55.0) | 357 (54.8) |
| Race–study center, N (%) | |||
| Washington county (White) | 3,121 (24.7) | 2,935 (24.4) | 186 (28.6) |
| Jackson (Black) | 2,920 (23.1) | 2,747 (22.9) | 173 (26.6) |
| Minneapolis (White) | 3,268 (25.8) | 3,161 (26.3) | 107 (16.4) |
| Forsyth (Black) | 402 (3.2) | 364 (3.0) | 38 (5.8) |
| Forsyth (White) | 2,950 (23.3) | 2,803 (23.3) | 147 (22.6) |
| Systolic blood pressure, mm Hg | 121.2 (19.0) | 120.7 (18.6) | 130.7 (22.4) |
| Diastolic blood pressure, mm Hg | 73.5 (11.2) | 73.4 (11.1) | 75.5 (12.7) |
| Resting heart rate, beats per minute | 66.7 (10.3) | 66.6 (10.1) | 68.9 (12.8) |
| Estimated glomerular filtration rate, ml/min/1.73 m2 | 102.3 (15.8) | 102.6 (15.4) | 96.3 (20.4) |
| Body mass index, kg/m2 | 27.6 (5.3) | 27.6 (5.3) | 27.7 (6.1) |
| Leisure index, units | 2.4 (0.6) | 2.4 (0.6) | 2.3 (0.6) |
| Diabetes, N (%) | 1,502 (11.9) | 1,356 (11.3) | 146 (22.4) |
| Hypertension, N (%) | 4,335 (34.2) | 3,952 (32.9) | 383 (58.8) |
| Hypertensive medication use in last 2 weeks, N (%) | 3,810 (30.1) | 3,469 (28.9) | 341 (52.4) |
| CHD at baseline, N (%) | 607 (4.8) | 550 (4.6) | 57 (8.8) |
| Prior stroke at baseline, N (%) | 226 (1.8) | 188 (1.6) | 38 (5.8) |
| CHF at baseline, N (%) | 567 (4.5) | 511 (4.3) | 56 (8.6) |
| Diuretic use, N (%) | 2,196 (17.3) | 2,008 (16.7) | 188 (28.9) |
| Antidepressant use, N (%) | 351 (2.8) | 317 (2.6) | 34 (5.2) |
| Sedative use, N (%) | 186 (1.5) | 179 (1.5) | 7 (1.1) |
| Hypnotic medication use, N (%) | 258 (2.0) | 247 (2.1) | 11 (1.7) |
| Antipsychotic medication use, N (%) | 88 (0.7) | 80 (0.7) | 8 (1.2) |
| Anticholinergic use, N (%) | 247 (2.0) | 226 (1.9) | 21 (3.2) |
| Alcohol use, N (%) | |||
| Never | 3,155 (24.9) | 2,962 (24.7) | 193 (29.6) |
| Former | 2,371 (18.7) | 2,203 (18.3) | 168 (25.8) |
| Current | 7,135 (56.4) | 6,845 (57.0) | 290 (44.5) |
| Education attainment, N (%) | |||
| Low | 2,900 (22.9) | 2,657 (22.1) | 243 (37.3) |
| Medium | 5,192 (41.0) | 4,955 (41.3) | 237 (36.4) |
| High | 4,569 (36.1) | 4,398 (36.6) | 171 (26.3) |
| Smoking status, N (%) | |||
| Never | 5,228 (41.3) | 4,985 (41.5) | 243 (37.3) |
| Former | 4,140 (32.7) | 3,939 (32.8) | 201 (30.9) |
| Current | 3,293 (26.0) | 3,086 (25.7) | 207 (31.8) |
Abbreviation: CHD, coronary heart disease; CHF, congestive heart failure; OH, orthostatic hypotension.
The overall incidence rate of falls (identified from all available data sources), using age as the time axis, was 2.6 per 1,000 person-years. This varied significantly by the presence of OH with participants with OH at baseline having an incidence rate of 3.2 per 1,000 person-years vs. 2.5 per 1,000 person-years among those without OH at baseline (P = 0.03) (Supplementary Table S4).
After adjustment for age, sex, and race–study center (Model 1), OH in middle age was significantly associated with risk of falls (hazard ratio (HR): 1.43; 95% confidence interval (CI): 1.21, 1.69) (Table 2). After full adjustment (Model 4), this relationship was attenuated, but still significant (HR: 1.30; 95% CI: 1.10, 1.54). Similarly, both postural change in SBP or DBP were associated with falls, demonstrating adjusted HRs of 1.03 (95% CI: 1.01, 1.05) and 1.09 (95% CI: 1.05, 1.13) per −5mm Hg, respectively. Interestingly, postural changes in DBP were more significantly associated with falls than postural changes in SBP (Model 5). The relationship between postural change in SBP and DBP appeared linear with no obvious threshold at the standard cut points (Figure 1). There was no evidence of effect modification across strata of age, sex, race, obesity, hypertension, or diabetes (Table 3).
Table 2.
Association between orthostatic hypotension and risk of fallsa
| Falls (N = 2,384)a | ||
|---|---|---|
| HR (95% CI) | P | |
| Orthostatic hypotension | ||
| Unadjusted | 1.97 (1.67, 2.33) | <0.001 |
| Model 1 | 1.43 (1.21, 1.69) | <0.001 |
| Model 2 | 1.33 (1.12, 1.57) | 0.001 |
| Model 3 | 1.31 (1.10, 1.54) | 0.002 |
| Model 4 | 1.30 (1.10, 1.54) | 0.002 |
| Postural change in SBP, per −5mm Hg | ||
| Unadjusted | 1.07 (1.05, 1.09) | <0.001 |
| Model 1 | 1.03 (1.02, 1.05) | <0.001 |
| Model 2 | 1.03 (1.01, 1.05) | 0.001 |
| Model 3 | 1.03 (1.01, 1.05) | 0.002 |
| Model 4 | 1.03 (1.01, 1.05) | 0.002 |
| Postural change in DBP, per −5mm Hg | ||
| Unadjusted | 1.14 (1.10, 1.18) | <0.001 |
| Model 1 | 1.11 (1.07, 1.16) | <0.001 |
| Model 2 | 1.09 (1.05, 1.13) | <0.001 |
| Model 3 | 1.09 (1.05, 1.13) | <0.001 |
| Model 4 | 1.09 (1.05, 1.13) | <0.001 |
| Postural change in SBP, per −1 SD (Model 4) | 1.07 (1.02, 1.11) | 0.002 |
| Postural change in DBP, per −1 SD (Model 4) | 1.10 (1.05, 1.15) | <0.001 |
| Model 5 both postural changes in SBP and DBP | ||
| SBP per −5mm Hg | 1.01 (0.98, 1.03) | 0.53 |
| DBP per −5mm Hg | 1.08 (1.03, 1.13) | 0.002 |
Model 1: adjusted for age, sex, race–study center. Model 2: Model 1 + heart rate, body mass index, estimated glomerular filtration rate, diabetes, hypertension, alcohol use, education level, smoking status, leisure index, prior CHD, prior CHD, prior stroke. Model 3: Model 2 + hypertension medication use in last 2 weeks, diuretic use, antidepressant use, sedative use, hypnotic use, antipsychotic medication use, anticholinergic medication use. Model 4: Model 3 + resting systolic blood pressure, resting diastolic blood pressure. Model 5: Model 4 including both postural change in SBP (per −5 mmHg) and postural change in DBP (per −5mm Hg). Abbreviations: ARIC, Atherosclerosis Risk in Communities; CHD, coronary heart disease; CI, confidence interval; DBP, diastolic blood pressure; HR, hazard ratio; SBP, systolic blood pressure.
aFalls were ascertained from diagnostic codes from either ARIC or CMS sources.
Figure 1.
Adjusted hazard ratios (solid line) for falls between baseline and 1 January 2012, according to postural change in (A) SBP or (B) DBP, using a restricted cubic spline model. Gray shade represents 95% confidence intervals. Vertical lines represent orthostatic hypotension thresholds (−20mm Hg for SBP; −10mm Hg for DBP). The models were expressed relative to the 50th percentile of postural change in SBP or DBP with 4 knots specified using Harrell’s method.23 Models were adjusted for age, sex, race–study center, heart rate, body mass index, estimated glomerular filtration rate, diabetes, hypertension, alcohol use, education level, smoking status, leisure index, prior coronary heart disease, prior stroke, prior congestive heart failure, hypertension medication use in last 2 weeks, diuretic use, antidepressant use, sedative use, hypnotic use, antipsychotic medication use, anticholinergic medication use, resting SBP, and resting DBP. The plot was truncated at the 0.5st and 99.5th percentiles of postural change in SBP or DBP (standing blood pressure minus supine blood pressure; note that a negative value represents a drop in blood pressure with standing). The hazard ratios are shown on a natural log scale. In addition, overlaid are kernel density plots depicting the distribution of postural change in SBP or DBP by participants who fell (dash) vs. those who did not fall (solid). Abbreviations: DBP, diastolic blood pressure; SBP, systolic blood pressure.
Table 3.
Stratified analysis of the association between orthostatic hypotension and risk of fallsa
| N | HR (95% CI) | P interaction | |
|---|---|---|---|
| Age (years)b | |||
| <50 | 3,408 | 1.60 (0.85, 3.02) | 0.72 |
| 50–60 | 6,952 | 1.23 (0.97, 1.58) | |
| >60 | 2,294 | 1.35 (1.04, 1.75) | |
| Sex | |||
| Male | 5,694 | 1.61 (1.23, 2.11) | 0.11 |
| Female | 6,960 | 1.18 (0.95, 1.46) | |
| Racec | |||
| White | 9,336 | 1.27 (1.04, 1.54) | 0.40 |
| Black | 3,318 | 1.51 (1.08, 2.13) | |
| Obese (body mass index ≥30kg/m2) | |||
| No | 9,242 | 1.28 (1.04, 1.57) | 0.76 |
| Yes | 3,412 | 1.36 (1.01, 1.84) | |
| Hypertension | |||
| No | 8,323 | 1.29 (1.00, 1.66) | 0.72 |
| Yes | 4,331 | 1.25 (0.99, 1.56) | |
| Diabetes | |||
| No | 11,152 | 1.28 (1.06, 1.55) | 0.87 |
| Yes | 1,502 | 1.35 (0.93, 1.96) |
Models adjusted for all covariates with exception of covariate related to associated strata. Abbreviations: ARIC, Atherosclerosis Risk in Communities; CI, confidence interval; HR, hazard ratio.
aFalls were ascertained from diagnostic codes from either ARIC or CMS sources.
bWhen evaluating an interaction between strata of age, the age variable was replaced with a simplified variable (<50, 50–60, >60).
cIn analysis of strata by race, the race–ARIC center variable was replaced by a dichotomous Black vs. White variable to reflect the strata presented in this table and assess for an interaction.
Sensitivity analyses were performed, restricting our definition of falls to either ARIC cohort surveillance of hospitalized events, and CMS Medicare claims, or continuous FFS alone. Despite notable attenuation, we observed similar patterns of association regardless of the source of falls (Supplementary Table S5). Similarly, defining falls by more specific categories of ICD-9 codes merely attenuated our findings (Supplementary Table S6).
DISCUSSION
We found that baseline OH assessed in a middle-aged, community-based population was associated with risk of hospital- or health care-related falls over a substantial time period. Postural reductions in either SBP or DBP both showed inverse associations with falls; although, changes in DBP were more strongly and significantly associated with risk of falls. Furthermore, similar associations between OH and fall risk were observed in demographic subgroups and in persons with and without cardiovascular risk factors. Overall, our study suggests that OH and, more specifically, postural changes in DBP are important risk factors for falls over a long period of time.
Our finding that OH precedes falls in a community-based, middle-aged population represents an important contribution to the literature. Prior cross-sectional studies have shown that OH is associated with history of falls among older adults8,10,24 with few exceptions.25 However, prospective studies examining this relationship are small, of short duration, and inconsistent with some showing OH to be associated with risk of falls26–30 and others showing no association.31–38 One of the larger prospective studies, conducted in 844 nursing home residents, reported that OH was associated with a higher risk of falls (relative risk: 2.1, 95% CI: 1.4, 3.1) during 1.2 years of follow-up.26 In contrast, a larger prospective study of 1,517 ambulatory Chinese patients, showed that OH was not associated with falls after adjustment for other risk factors.31 These inconsistencies have led to the conclusion that OH is an important predictor of falls in institutionalized, but not ambulatory adults.39 In contrast, our study of 12,661 ambulatory middle-aged adults, showed that OH was not only associated with falls, but also that the association was independent of other risk factors well in advance of a fall event.
The hypothesis that OH is causal in the pathogenesis of falls is biologically plausible. In healthy persons, change from a recumbent to standing position causes venous pooling, decreased venous return, and reduced stroke volume.40 The cardiovascular system responds with increased heart rate and vasoconstriction to maintain cerebral perfusion.40 As 1 ages, these responses become blunted—heart rate is more fixed, blood vessels are less responsive to changes in pressure and hormonal signaling pathways, and often there is reduced blood volume due to dehydration or reduced endogenous blood production.40 Many of these acquired conditions are irreversible and worsen over time. Furthermore, adults are commonly introduced to medications as they age, e.g., diuretics, beta-blockers, that may attenuate the normal physiologic response to standing. This contributes to a number of symptoms—lightheadedness, vision problems, weakness, fatigue, trouble concentrating, and head or neck discomfort,41 which can increase the risk of falling.
In this study, for the first time, we characterized the continuous association between postural changes in SBP or DBP and falls. Not only were they both associated with falls, but the inverse relationships between postural change in SBP or DBP extended above and below the cut points used in the consensus definition of OH with no threshold effect. These findings highlight a need for further research to establish a definition of OH based on more objective evidence, such as clinically relevant outcomes (e.g., fall vs. fracture vs. mortality).
We found postural change in DBP to be a stronger predictor of falls than postural changes in SBP. While prior studies have shown postural changes in DBP to be associated with an increased risk of coronary events,42 myocardial infarction,43 and mortality,44 its role in the development of falls is unclear. In 1 study of 446 diabetic participants with mean follow-up of 5 years, postural change in DBP (not SBP), was associated with more frequent falls.45 However, the mechanism behind this association is beyond the scope of this study.
Given the public health burden of falls, this study has important implications. Not only was OH associated with subsequent fall, but it was associated with falls many years after the assessment of OH. This latency period between middle age OH and falls supports the notion that the pathogenesis of a fall is a multifactorial process. Prior research studies have identified multiple factors that contribute to falls—age, muscle strength, gait speed, etc.31,39 It is possible that when OH is developed at a younger age, a person can compensate for this destabilizing factor, but with age the ability to compensate for OH is lost, which we speculate could be related to the acquisition of additional functional and neurologic deficits. However, further research is needed to confirm this hypothesis. At the very least, adults identified as having OH should receive special consideration when prescribed medications known to worsen alertness, affect balance, or increase the risk of falls.34,37,39 Moreover, interventions useful in preventing falls may be worth implementing in adults with OH such as physical therapy.46
This study has a number of important limitations. First, fall history was not assessed at baseline. As a result, we are unable to differentiate between participants with falls prior to baseline vs. participants who never had a fall. Falls are estimated to occur in about 1% of middle-aged adults each year.47 Second, while external cause of injury codes are reportedly valid,48 there is likely substantial under-ascertainment of falls given the fact that we only capture falls reported to health care providers, not the large number of falls that are not reported and do not result in significant injury. This is supported by the low sensitivity we observed from CMS claims data (for further elaboration on CMS limitations see Supplementary Methods). Third, OH was measured in a subgroup of ARIC participants due to late implementation of this protocol and was not re-measured at a later time. As a result, our sample size was smaller than the original ARIC cohort; furthermore, we could not evaluate change in OH over time. Fourth, the OH assessment reflected the first 1–2 minutes after standing, but not 3 minutes or more, as suggested by the consensus definition. As a result, it is possible the OH status of some participants was misclassified. Fifth, our analyses of types of falls according to broader groupings of ICD-9 codes, namely, falls from the same level and falls from a different level had fewer events. As a result, these results may be underpowered to show an association between OH and falls. Finally, residual confounding is always a concern with observational studies. In particular, we lack data on important contributing factors such as vision, neuropathy, long-term glucose-lowering medication use, vitamin D deficiency, vitamin B12 deficiency, osteoarthritis, and strength.
This study also has a number of strengths. Its biracial, large sample is representative of a general US population. Further, participants were followed for a substantial time period. Hospitalization records were reviewed in a detailed fashion by trained ARIC staff following a rigorous protocol. Use of ICD-9 codes to ascertain falls ensures that these events had some degree of clinical relevance. Finally, OH was assessed via a standardized protocol. This along with other covariate measures afforded the opportunity to account for multiple potential confounders in our models.
In conclusion, the traditional definition of OH was associated with risk of fall irrespective of participant characteristics or comorbid conditions. Postural change in SBP and DBP were associated with fall above and below traditional cut points with DBP being more strongly associated with risk of falls. Future studies should examine alternative definitions of OH as well as exposures and outcomes associated with change in OH over time. Finally, future studies should determine whether targeted therapies can reduce falls in persons with OH.
SUPPLEMENTARY MATERIAL
Supplementary materials are available at American Journal of Hypertension (http://ajh.oxfordjournals.org).
DISCLOSURE
The authors declared no conflict of interest.
ACKNOWLEDGMENTS
S.P.J. is supported by a NIH/NIDDK T32DK007732-20 Renal Disease Epidemiology Training Grant. The Atherosclerosis Risk in Communities Study is carried out as a collaborative study supported by National Heart, Lung, and Blood Institute contracts (HHSN268201100005C, HHSN 268201100006C, HHSN268201100007C, HHSN2682 01100008C, HHSN268201100009C, HHSN268201100010C, HHSN268201100011C, and HHSN268201100012C). The authors thank the staff and participants of the ARIC study for their important contributions. Parts of this study were presented as a moderated poster at the American Heart Association, Epidemiology & Nutrition Conference in Phoenix, March 2016.
REFERENCES
- 1. Stevens JA, Mack KA, Paulozzi LJ, Ballesteros MF. Self-reported falls and fall-related injuries among persons aged>or=65 years–United States, 2006. J Safety Res 2008; 39:345–349. [DOI] [PubMed] [Google Scholar]
- 2. Hu G, Baker SP. Recent increases in fatal and non-fatal injury among people aged 65 years and over in the USA. Inj Prev 2010; 16:26–30. [DOI] [PubMed] [Google Scholar]
- 3. Burt CW, Fingerhut LA. Injury visits to hospital emergency departments: United States, 1992–95. Vital Health Stat 13 1998; 131:1–76. [PubMed] [Google Scholar]
- 4. Edwards MH, Jameson K, Denison H, Harvey NC, Sayer AA, Dennison EM, Cooper C. Clinical risk factors, bone density and fall history in the prediction of incident fracture among men and women. Bone 2013; 52:541–547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Davis JC, Robertson MC, Ashe MC, Liu-Ambrose T, Khan KM, Marra CA. International comparison of cost of falls in older adults living in the community: a systematic review. Osteoporos Int 2010; 21:1295–1306. [DOI] [PubMed] [Google Scholar]
- 6. Wright JT, Jr, Williamson JD, Whelton PK, Snyder JK, Sink KM, Rocco MV, Reboussin DM, Rahman M, Oparil S, Lewis CE, Kimmel PL, Johnson KC, Goff DC, Jr, Fine LJ, Cutler JA, Cushman WC, Cheung AK, Ambrosius WT; SPRINT Research Group A randomized trial of intensive versus standard blood-pressure control. N Engl J Med 2015; 373:2103–2116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Finucane C, O’Connell MD, Fan CW, Savva GM, Soraghan CJ, Nolan H, Cronin H, Kenny RA. Age-related normative changes in phasic orthostatic blood pressure in a large population study: findings from The Irish Longitudinal Study on Ageing (TILDA). Circulation 2014; 130:1780–1789. [DOI] [PubMed] [Google Scholar]
- 8. Rutan GH, Hermanson B, Bild DE, Kittner SJ, LaBaw F, Tell GS. Orthostatic hypotension in older adults. The Cardiovascular Health Study. CHS Collaborative Research Group. Hypertension 1992; 19:508–519. [DOI] [PubMed] [Google Scholar]
- 9. Allcock LM, O’Shea D. Diagnostic yield and development of a neurocardiovascular investigation unit for older adults in a district hospital. J Gerontol A Biol Sci Med Sci 2000; 55:M458–M462. [DOI] [PubMed] [Google Scholar]
- 10. Passant U, Warkentin S, Gustafson L. Orthostatic hypotension and low blood pressure in organic dementia: a study of prevalence and related clinical characteristics. Int J Geriatr Psychiatry 1997; 12:395–403. [PubMed] [Google Scholar]
- 11. Wu JS, Yang YC, Lu FH, Wu CH, Chang CJ. Population-based study on the prevalence and correlates of orthostatic hypotension/hypertension and orthostatic dizziness. Hypertens Res 2008; 31:897–904. [DOI] [PubMed] [Google Scholar]
- 12. The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. The ARIC investigators. Am J Epidemiol 1989; 129:687–702. [PubMed] [Google Scholar]
- 13. Jackson R, Chambless LE, Yang K, Byrne T, Watson R, Folsom A, Shahar E, Kalsbeek W. Differences between respondents and nonrespondents in a multicenter community-based study vary by gender ethnicity. The Atherosclerosis Risk in Communities (ARIC) Study Investigators. J Clin Epidemiol 1996; 49:1441–1446. [DOI] [PubMed] [Google Scholar]
- 14. White AD, Folsom AR, Chambless LE, Sharret AR, Yang K, Conwill D, Higgins M, Williams OD, Tyroler HA. Community surveillance of coronary heart disease in the Atherosclerosis Risk in Communities (ARIC) Study: methods and initial two years’ experience. J Clin Epidemiol 1996; 49:223–233. [DOI] [PubMed] [Google Scholar]
- 15. Mundt KA, Chambless LE, Burnham CB, Heiss G. Measuring ankle systolic blood pressure: validation of the Dinamap 1846 SX. Angiology 1992; 43:555–566. [DOI] [PubMed] [Google Scholar]
- 16. The National Heart, Lung, and Blood Institute. ARIC Manual 11: Sitting Blood Pressure http://www.cscc.unc.edu/aric/visit/Sitting_Blood_Pressure_and_Postural_Changes_in_Blood_Pressure_and_Heart_Rate.1_11.pdf Accessed April 4, 2016.
- 17. Smith JJ, Porth CM, Erickson M. Hemodynamic response to the upright posture. J Clin Pharmacol 1994; 34:375–386. [DOI] [PubMed] [Google Scholar]
- 18. Freeman R, Wieling W, Axelrod FB, Benditt DG, Benarroch E, Biaggioni I, Cheshire WP, Chelimsky T, Cortelli P, Gibbons CH, Goldstein DS, Hainsworth R, Hilz MJ, Jacob G, Kaufmann H, Jordan J, Lipsitz LA, Levine BD, Low PA, Mathias C, Raj SR, Robertson D, Sandroni P, Schatz I, Schondorff R, Stewart JM, van Dijk JG. Consensus statement on the definition of orthostatic hypotension, neurally mediated syncope and the postural tachycardia syndrome. Clin Auton Res 2011; 21:69–72. [DOI] [PubMed] [Google Scholar]
- 19. Consensus statement on the definition of orthostatic hypotension, pure autonomic failure, and multiple system atrophy. The Consensus Committee of the American Autonomic Society and the American Academy of Neurology. Neurology 1996; 46:1470. [DOI] [PubMed] [Google Scholar]
- 20. Richardson MT, Ainsworth BE, Wu HC, Jacobs DR, Jr, Leon AS. Ability of the Atherosclerosis Risk in Communities (ARIC)/Baecke Questionnaire to assess leisure-time physical activity. Int J Epidemiol 1995; 24:685–693. [DOI] [PubMed] [Google Scholar]
- 21. Eriksson H, Caidahl K, Larsson B, Ohlson LO, Welin L, Wilhelmsen L, Svärdsudd K. Cardiac and pulmonary causes of dyspnoea–validation of a scoring test for clinical-epidemiological use: the Study of Men Born in 1913. Eur Heart J 1987; 8:1007–1014. [DOI] [PubMed] [Google Scholar]
- 22. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF, 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J; CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) A new equation to estimate glomerular filtration rate. Ann Intern Med 2009; 150:604–612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Harrell FE, Jr, Lee KL, Pollock BG. Regression models in clinical studies: determining relationships between predictors and response. J Natl Cancer Inst 1988; 80:1198–1202. [DOI] [PubMed] [Google Scholar]
- 24. Azidah AK, Hasniza H, Zunaina E. Prevalence of falls and its associated factors among elderly diabetes in a tertiary center, Malaysia. Curr Gerontol Geriatr Res 2012; 2012:539073. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Ensrud KE, Nevitt MC, Yunis C, Hulley SB, Grimm RH, Cummings SR. Postural hypotension and postural dizziness in elderly women. The study of osteoporotic fractures. The Study of Osteoporotic Fractures Research Group. Arch Intern Med 1992; 152:1058–1064. [PubMed] [Google Scholar]
- 26. Ooi WL, Hossain M, Lipsitz LA. The association between orthostatic hypotension and recurrent falls in nursing home residents. Am J Med 2000; 108:106–111. [DOI] [PubMed] [Google Scholar]
- 27. Tinetti ME, Williams TF, Mayewski R. Fall risk index for elderly patients based on number of chronic disabilities. Am J Med 1986; 80:429–434. [DOI] [PubMed] [Google Scholar]
- 28. Graafmans WC, Ooms ME, Hofstee HM, Bezemer PD, Bouter LM, Lips P. Falls in the elderly: a prospective study of risk factors and risk profiles. Am J Epidemiol 1996; 143:1129–1136. [DOI] [PubMed] [Google Scholar]
- 29. Granek E, Baker SP, Abbey H, Robinson E, Myers AH, Samkoff JS, Klein LE. Medications and diagnoses in relation to falls in a long-term care facility. J Am Geriatr Soc 1987; 35:503–511. [DOI] [PubMed] [Google Scholar]
- 30. Heitterachi E, Lord SR, Meyerkort P, McCloskey I, Fitzpatrick R. Blood pressure changes on upright tilting predict falls in older people. Age Ageing 2002; 31:181–186. [DOI] [PubMed] [Google Scholar]
- 31. Chu LW, Chi I, Chiu AY. Incidence and predictors of falls in the Chinese elderly. Ann Acad Med Singapore 2005; 34:60–72. [PubMed] [Google Scholar]
- 32. Maurer MS, Cohen S, Cheng H. The degree and timing of orthostatic blood pressure changes in relation to falls in nursing home residents. J Am Med Dir Assoc 2004; 5:233–238. [DOI] [PubMed] [Google Scholar]
- 33. Gangavati A, Hajjar I, Quach L, Jones RN, Kiely DK, Gagnon P, Lipsitz LA. Hypertension, orthostatic hypotension, and the risk of falls in a community-dwelling elderly population: the maintenance of balance, independent living, intellect, and zest in the elderly of Boston study. J Am Geriatr Soc 2011; 59:383–389. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Liu BA, Topper AK, Reeves RA, Gryfe C, Maki BE. Falls among older people: relationship to medication use and orthostatic hypotension. J Am Geriatr Soc 1995; 43:1141–1145. [DOI] [PubMed] [Google Scholar]
- 35. Campbell AJ, Borrie MJ, Spears GF. Risk factors for falls in a community-based prospective study of people 70 years and older. J Gerontol 1989; 44:M112–M117. [DOI] [PubMed] [Google Scholar]
- 36. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988; 319:1701–1707. [DOI] [PubMed] [Google Scholar]
- 37. Lord SR, Anstey KJ, Williams P, Ward JA. Psychoactive medication use, sensori-motor function and falls in older women. Br J Clin Pharmacol 1995; 39:227–234. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Clark RD, Lord SR, Webster IW. Clinical parameters associated with falls in an elderly population. Gerontology 1993; 39:117–123. [DOI] [PubMed] [Google Scholar]
- 39. Rubenstein LZ, Josephson KR, Robbins AS. Falls in the nursing home. Ann Intern Med 1994; 121:442–451. [DOI] [PubMed] [Google Scholar]
- 40. Feldstein C, Weder AB. Orthostatic hypotension: a common, serious and underrecognized problem in hospitalized patients. J Am Soc Hypertens 2012; 6:27–39. [DOI] [PubMed] [Google Scholar]
- 41. Kaufmann H, Malamut R, Norcliffe-Kaufmann L, Rosa K, Freeman R. The Orthostatic Hypotension Questionnaire (OHQ): validation of a novel symptom assessment scale. Clin Auton Res 2012; 22:79–90. [DOI] [PubMed] [Google Scholar]
- 42. Fedorowski A, Stavenow L, Hedblad B, Berglund G, Nilsson PM, Melander O. Orthostatic hypotension predicts all-cause mortality and coronary events in middle-aged individuals (The Malmo Preventive Project). Eur Heart J 2010; 31:85–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Luukinen H, Koski K, Laippala P, Airaksinen KE. Orthostatic hypotension and the risk of myocardial infarction in the home-dwelling elderly. J Intern Med 2004; 255:486–493. [DOI] [PubMed] [Google Scholar]
- 44. Lagro J, Laurenssen NC, Schalk BW, Schoon Y, Claassen JA, Olde Rikkert MG. Diastolic blood pressure drop after standing as a clinical sign for increased mortality in older falls clinic patients. J Hypertens 2012; 30:1195–1202. [DOI] [PubMed] [Google Scholar]
- 45. Schwartz AV, Vittinghoff E, Sellmeyer DE, Feingold KR, de Rekeneire N, Strotmeyer ES, Shorr RI, Vinik AI, Odden MC, Park SW, Faulkner KA, Harris TB; Health, Aging, and Body Composition Study Diabetes-related complications, glycemic control, and falls in older adults. Diabetes Care 2008; 31:391–396. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Karinkanta S, Piirtola M, Sievänen H, Uusi-Rasi K, Kannus P. Physical therapy approaches to reduce fall and fracture risk among older adults. Nat Rev Endocrinol 2010; 6:396–407. [DOI] [PubMed] [Google Scholar]
- 47. Verma SK, Willetts JL, Corns HL, Marucci-Wellman HR, Lombardi DA, Courtney TK. Falls and fall-related injuries among community-dwelling adults in the United States. PLoS One 2016; 11:e0150939. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. McKenzie K, Enraght-Moony EL, Walker SM, McClure RJ, Harrison JE. Accuracy of external cause-of-injury coding in hospital records. Inj Prev 2009; 15:60–64. [DOI] [PubMed] [Google Scholar]