Distribution and Observed Associations of Orthostatic Blood Pressure Changes in Elderly General Medicine Outpatients

Distribution and Observed Associations of Orthostatic Blood Pressure Changes in Elderly General Medicine Outpatients DAVID ROBERTSON, MD,* JEFF A. DESJARDIN, MD,t MICHAEL j. LICHTENSTEIN, MD, MSct

ABSTRACT: Factors associated with orthostatic blood pressure change in elderly outpatients were determined by surveying 398 medical clinical outpatients aged 65 years and older. Blood pressure was measured with random-zero sphygmomanometers after patients were 5 minutes in a supine and 5 minutes in a standing position. Orthostatic blood pressure changes were at normally distributed levels with systolic and diastolic pressures dropping an average of 4 mm Hg (standard deviation [SD] = 15 mm Hg) and 2 mm Hg (SD = 11 mm Hg), respectively. Orthostatic blood pressure changes were unassociated with age, race, sex, body mass, time since eating, symptoms, or other factors. According to multiple linear regression analysis, supine systolic pressure, chronic obstructive pulmonary disease (COPD), and diabetes mellitus were associated with a decrease in systolic pressure on standing. Hypertension, antiarthritic drugs, and abnormal heartbeat were associated with an increase in systolic pressure on standing. For orthostatic diastolic pressure changes, supine diastolic pressure and COPD were associated with a decrease in diastolic pressure on standing. Congestive heart failure was associated with an increase in standing diastolic pressure. Using logistic regression analysis, only supine systolic pressure was associated with a greater than 20-mm Hg drop in systolic From the *Autonomic Dysfunction Center, Vanderbilt University, Nashville, Tennessee, and the tGeriatrics, Research, Educa· tion and Clinical Center, Audie L. Murphy VA Hospital, San Anto· nio, Texas. Submitted October 20, 1997; accepted in revised form December 24, 1997. Supported in part by National Institutes of Health grants NS33460, HL56373, RR00095, and RR05424, and NASA grants NAS9-19483. Correspondence: David Robertson, MD, Autonomic Dysfunction Center, Vanderbilt University, AA-3228 Medical Center North, Nashville, TN 37232-2195. Email: david.robertson®mcmail. vanderbilt.edu THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES

pressure (n = 53, prevalence = 13%). Supine diastolic pressure and COPD were the only variables associated with a greater than 20-mm Hg drop in diastolic pressure (n = 16, prevalence = 4%). These factors may help physicians in identifying older persons at risk for having orthostatic hypotension. KEY INDEXING TERMS: Blood pressure; Heart failure; Autonomic; Hypotension; Syncope. [Am J Med Sci 1998;315(5):287295.]

L

arge declines in blood pressure on standing are likely to be risk factors for syncope and falls in elderly patients. 1 - 4 Orthostatic changes in diastolic pressure have also been associated with an increased risk of myocardial infarction. 5 In spite of their importance, the epidemiology, etiology, and symptoms of orthostatic blood pressure changes in elderly persons remain incompletely defined. Recent reviews have summarized epidemiologic studies and reported that the prevalence of orthostatic hypotension, depending on the definition used, varies from 4% to 33% in a variety of clinical settings. 6 ,7 Several reports have suggested a high prevalence of hypotension associated with upright or seated posture among elderly patients. 8 - 11 These and many other studies have driven an interest in defining orthostatic hypotension. In 1995, the American Autonomic Society defined orthostatic hypotension as a fall in blood pressure of 20/10 mm Hg with upright posture. 12 It seemed important to examine this definition of orthostatic hypotension among elderly individuals in a primary care practice. Few clinical studies have looked at associations between clinical factors and orthostatic blood pressure changes, considered either as a continuous variable or as an arbitrarily chosen cut-off point, and divided the distribution into those with and those without large orthostatic declines in blood pressure. The present

287

Orthostatic Hypotension in the Elderly

survey was conducted to attempt to answer two questions. First, what is the distribution of orthostatic blood pressure changes among elderly individuals using a general medical practice? Second, what clinical features are associated with orthostatic blood pressure changes and what are the magnitude of these associations? Patients and Methods Patient Recruitment. We sought to examine a consec-

utive sample of patients older than 65 years who attended a university-based primary care general internal medicine practice. Observers were present for a half-day at a time and located in a single quadrant of the clinic. Logs were kept of visits made in that quadrant and the patients' names and dates of birth were recorded. The volume of the practice and limited number of observers precluded complete ascertainment of a consecutive sample. Although we had no a priori reason to suspect bias in patient selection, we did not test this possibility formally. Blood Pressure Measurement. Blood pressure was recorded as a single reading in the right arm after each patient had been quietly supine for 5 minutes and again after each patient had been standing for 5 minutes. We chose to measure blood pressure in the upright position after 5 minutes to ensure that orthostatic change due to standing had stabilized. The disappearance of the Korotkoff sounds was recorded as diastolic pressure (phase 5). Cuffs appropriately sized to each patient's arm were used. Blood pressure was measured using random-zero sphygmomanometers to avoid potential observer expectation bias that the blood pressure would llecrease on standing. 13 To reduce interobserver variability, observers were trained in blood pressure measurement using a videotaped recording of sphygmomanometers and Korotkoff sounds. 14 Observers were tested for accuracy in use of the random-zero sphygmomanometer before the study was initiated. Ascertainment of Other Variables. Mter the observer obtained informed consent, a closed questionnaire was administered. In addition to blood pressure, the following variables were assessed: 1. Demographic information: sex, race, age, educational level. 2. Time of most recent meal and time of examination. 3. Height (feet and inches with shoes off) and weight (pounds). Body mass index was subsequently calculated as kg/m2.15 4. Supine and standing heart rate (counted for 30 seconds). 5. Supine and standing respiratory rate (counted for 60 seconds). 6. Presence or absence of Osler's sign.16 7. Varicose veins assessed when standing; noted

288

as absent, superficial, deep, or superficial and deep. 8. Symptoms on standing. These were assessed between blood pressure measurements by asking specific questions, for example, ''When you stood up just now, did you feel dizzy?" Patients' responses were recorded as yes or no. 9. History of falls in the past month or year. 10. Mental status, assessed using the Short Portable Mental Status Questionnaire. 17 11. Functional status, assessed using a modified version of the Healthy Activity Questionnaire/ 8 a measure of instrumental activities of daily living. 12. Alcohol consumption, assessed using questions from the Behavioral Risk Factor Survey, recording self-reported type, frequency, and amount of alcohol consumed per occasion. 19 13. Salt ingestion, crudely assessed by asking patients about use of salt in cooking and at the table. 14. Cigarette and other tobacco use, assessed by self-report. 15. First seven current diagnoses listed on the clinic problem list. These were subsequently coded and grouped using the International Classification of Health Problems in Primary Care. 20 16. First seven currently taken medications listed in the clinic's medication flow sheet. These drugs were assigned codes and grouped using chapter headings from theAMA Drug Evaluations, 5th edition,21 as a means of classifying drugs. Current drug use was confirmed by the patient. Full assessment of each participant (obtaining informed consent, measuring blood pressures, and completing the questionnaire) took approximately 40 minutes. Data Analysis. The standing blood pressure was subtracted from the supine pressure to determine the orthostatic change. In the tables and text, a negative value denotes a decrease in blood pressure on standing. The frequency distributions for orthostatic changes in systolic and diastolic pressure were determined. Univariate associations with orthostatic changes in pressure were assessed by calculation of Pearson product-moment correlations and through stratification by levels of each independent variable. Two-tailed t-tests or analysis of variance was used as appropriate to test levels of statistical significance. 22 Simple univariate odds ratios and 95% confidence intervals (CI) were calculated for the associations between level of orthostatic blood pressure change (>20-mm Hg fall in systolic or > 10-mm Hg fall in diastolic pressure) and presence of symptoms on May 1998 Volume 315 Number 5

Robertson, Desjardin, and lichtenstein

standinlf3 using the American Autonomic Society criteria. 12 Both stepwise linear and logistic regression analyses were performed for three reasons. First, because the distribution of orthostatic blood pressure changes is continuous, we used linear regression to determine associations across the spectrum of orthostatic changes. In contrast, logistic regression was used to determine the clinical associations of orthostatic changes at levels often used to clinically define orthostatic hypotension. Second, the regressions were used to adjust for the phenomenon of regression to the mean to account for the orthostatic changes. Third, these analyses adjust for the problems of multiple comparisons seen with repeated univariate analyses. Stepwise multiple linear regression was used to assess the adjusted associations between orthostatic blood pressure changes (as continuous dependent variables) and the independent variables that were univariately associated with orthostatic pressure change. 22 Because supine blood pressure readings are used to determine orthostatic change and hence are not independent measures, adjustments were made to account for potential biases (such as regression to the mean) that may result from such observations. 24 In general, the lack of independence between the two measures tends to underestimate their true association as determined through linear regression.24 Because clinically orthostatic hypotension is somewhat arbitrarily defined as either a greater than 20-mm Hg fall in systolic or a greater than 10mm Hg fall in diastolic pressure on standing, stepwise multiple logistic regression analysis was used to assess the associations for orthostatic decreases of these magnitudes and the univariately associated independent variables. 25 Adjusted odds ratio and 95% CI were determined from the logistic regression coefficients. 25 For all statistical tests, P values less than 0.05 were considered to indicate statistical significance. Results Response Rate. A total of 4,092 visits were made

on the 120 survey dates during the times when a study observer was present in the clinic. Of these, 1,444 visits were made by persons aged 65 years or older. Of the visits made by eligible patients, 643 (44.5%) were made by 398 persons who participated in the survey. These 398 patients formed the basis for this report. No patient was studied more than once. The mean age of the sample was 73.9 years with 87 patients older than 80 years (SD = 6.5 years, range = 65-94 years). Of participating patients, 68.8% were white. For the remaining 801 visits for which data were not collected, 99 patients refused to participate, 39 THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES

could not stand, 97 did not have the time, and 71 were too ill. Of the incomplete visits, 495 (61.8%) occurred because the patients were missed by the observers due to site logistics and practice volume. Distribution of Orthostatic Blood Pressure Changes.

Orthostatic changes in both systolic and diastolic blood pressures were normally distributed. For systolic pressures the average change was a -4-mm Hg decline (SD = 15 mm Hg) in pressure with a range of 46 to -50 mm Hg. For diastolic pressures the average change was a -2-mm Hg decline (SD = 11 mm Hg) in pressure with a range of 36 to -50 mm Hg. Of the 398 patients, 111 (28%) and 72 (18%) had drops of greater than 10 mm Hg in systolic and diastolic pressures, respectively. Fifty-three (13%) and 16 (4%) had drops of greater than 20 mm Hg in respective systolic and diastolic pressure. An orthostatic change in systolic pressure was correlated with an orthostatic change in diastolic pressure (r = 0.33, P < 0.01). Dissociation Between Symptoms and Orthostatic Blood Pressure Changes. Although symptoms on

standing were common (reported by a third of study participants), they were infrequently associated with orthostatic changes in blood pressure after 5 minutes in an upright position. In Table 1 patients have been stratified by whether or not they had an orthostatic decline of greater than 20 mm Hg in either systolic or diastolic pressure. For the more common symptoms (unsteadiness, light-headedness, dizziness, and weakness) the odds ratios for these symptoms being associated with a greater than 20mm Hg drop in systolic or diastolic pressure ranged from 0.9 to 2.2. The 95% CIs all included the value of I, indicating no significant association between these symptoms and an orthostatic drop of greater than 20 mm Hg in blood pressure. For the less common symptoms the odds ratios ranged from 0.29 to 3.61 with most 95% CIs including the value 1. In addition, no association was observed between presence of orthostatic symptoms and other levels of blood pressure change on standing. Univariate Associations with Other Variables. Univariately, orthostatic systolic or diastolic blood pressure changes were not associated with age, race, sex, body mass index (kg/m2 ), heart rate (supine, standing, or orthostatic change), respiratory rate (supine, standing, or orthostatic change), mental status (most patients were cognitively intact), functional status (most patients were functionally independent), varicose veins, time since last meal, history of falling in the past month or year, questions on salt intake, or self-reported alcohol consumption. With the exceptions of the variables discussed, diagnoses and medication classes were not associated with orthostatic changes in blood pressure. The variables univariately associated with orthostatic changes in either

289

Orthostatic Hypotension in the Elderly

Table 1. Persons With Postural Symptoms Who Have a Greater Than 20 mm Hg Decrease in Blood Pressure Decrease in SBP >20 mm Hg Symptom

Percent With Symptom

Odds Ratio

Unsteady Light-headed Dizziness Weak Breathless Dim vision Neck discomfort Throbbing sensation Faint Headache Dark vision Nauseated

33 31 27 20 9 9 8 7 6 6 4 3

1.80 1.04 1.23 1.22 1.33 1.12 1.68 0.79 2.17 0.29 1.64 3.90

CI

=

confidence interval; DBP

=

diastolic blood pressure; SBP

=

systolic or diastolic pressure are listed in Tables 2 and 3, respectively. Supine systolic pressure was associated with orthostatic changes in systolic pressure (r = 0.24, P < 0.01) and supine diastolic pressure (r = 0.32, P < 0.01). The higher the pressure, the greater the fall on standing. Cigarette smoking, diagnosis of chronic obstructive pulmonary disease (COPD), and medication use for bronchial disorders were all associated with orthostatic declines in systolic or diastolic pressure or both. However, these variables were interrelated such that accounting for the presence of COPD reduced the association attributable to the other two factors. A comparison of the 32 persons listed as having COPD showed they had a mean orthostatic decrease of -9 mm Hg in both systolic and diastolic pressures on standing compared with mean decreases of -4 mm Hg and -1 mm Hg in systolic and diastolic pressures, respectively, for persons without COPD. Hypertension, the most commonly coded diagnosis, was present in 219 (55%) of patients. Patients with hypertension were associated with smaller declines in orthostatic pressure for both systolic (- 2 mm Hg) and diastolic (-1 mm Hg) pressures compared with patients without this diagnosis (-4 mm Hg and - 2 mm Hg for systolic and diastolic pressures, respectively). This was observed although hypertensive patients had higher measured supine pressures than normotensive patients (157/83 versus 142/75 mm Hg, respectively). It is possible that this unexpected finding arose from the fact that most of the patients had satisfactorily controlled blood pressure at the time of the study. Diabetes was present in 58 patients, 40 (73%) of whom also suffered from hypertension. Although diabetes was not univariately associated with orthostatic changes in blood pressure, this may have

290

95% CI 1.00, 0.56, 0.65, 0.61, 0.52, 0.41, 0.65, 0.23, 0.82, 0.04, 0.45, 1.10,

3.24 1.94 2.33 2.46 3.36 3.03 4.33 2.71 5.71 2.21 6.02 13.79

Decrease in DBP >20 mm Hg Odds Ratio

95% CI

1.54 1.20 2.24 0.94 3.61 0.69 0.80 2.00

0.58, 4.36 0.41, 3.54 0.81, 6.17 0.26, 3.37 1.10,11.86 0.09, 5.43 0.10, 6.27 0.43, 9.32

1.13 1.73 2.45

0.14, 8.94 0.21,14.02 0.29,20.43

systolic blood pressure.

been a result of its being confounded with the diagnosis of hypertension. There was an insignificant trend in persons with diabetes to have greater falls in systolic pressure whether they had hypertension (-6 mm Hg versus -2 mm Hg) or not (-9 mm Hg versus -6 mm Hg) (F[1,394] = 3.22, P = 0.073 for main effect of diabetes). The presence of an abnormal heartbeat (29 patients) was associated with an increase in both systolic (1 mm Hg) and diastolic (2 mm Hg) pressures on standing. This broad rubric included all patients with atrial fibrillation and flutter, paroxysmal tachycardia, or ectopic beats (all types) but is an accepted ICD-9 equivalent for coding this group of outpatient diagnoses. 2o The use of antiarrhythmic drugs was neither prominent in this subgroup nor univariately associated with orthostatic changes in blood pressure. Similarly, the diagnosis of congestive heart failure was significantly associated with an increase in diastolic pressure on standing (2 mm Hg). This increase was not observed for systolic pressure changes. For the other medication classes, current use of antiarthritic drugs was associated with smaller orthostatic decreases in systolic but not diastolic pressure. In patients using electrolyte supplements there was no change in orthostatic systolic pressure and a small increase in diastolic pressure (1 mm Hg) compared with persons not using these agents. Similarly, current users of diuretics had smaller orthostatic declines in systolic and diastolic pressures compared with nondiuretic users. As in the case of diabetes, these findings for electrolytes and diuretics may be explained by their being confounded with the diagnosis of hypertension. Of the 194 persons who were taking diuretics, 145 (75%) had hypertension compared with 74 (36%) of 204 persons not taking diuretics. Examination of orthostatic changes in systolic pressure showed that normotensive patients May 1998 Volume 315 Number 5

Robertson, DesJardin, and Lichtenstein

Table 2. Variables Univariately Associated with a Postural Change in Systolic Blood Pressure

Variable Quintile supine systolic pressure (F [4,373] = 4.71; P = 0.001) 0-128 130-140 142-154 156-170 172Cigarette use (t [395] = 6.96; P = 0.009) Nonsmoker Any current cigarette use Medications Antiarthritic (t = [397] = 2.36; P = 0.019) Current use Nonuser Diuretic (t = [397] = 2.59; P = 0.010) Current use Nonuser Electrolyte supplements (t = [397] = 2.14; P = 0.038) Current use Nonuser Bronchial disorder (t = [397] = 0.06; P = 0.953) Current use Nonuser Diagnosis Hypertension (t = [397] = 2.69; P = 0.007) Yes No Diabetes mellitus (t = [397] = 1.46; P = 0.145) Yes No Congestive heart failure (t = [397] = 1.32; P = 0.188) Yes No COPD (t = [397] = 1.82; P = 0.069) Yes No Abnormal heart beat (t = [397] = 2.10; P = 0.036) Yes No

Number

Mean Postural Change in Systolic mm Hg*

83 76 85 75 79

1 -3 -4 -10 -7

(14) (13) (13) (17) (16)

5.54 (P = 0.019). Use of other antihypertensive agents was not associated with orthostatic blood pressure changes. Multivariate Analyses. Results ofthe stepwise multiple linear regression analyses are given in Table 4. When orthostatic systolic blood pressure change was

Table 3. Variables Univariately Associated with a Postural Change in Systolic Blood Pressure

Variable 337 59

-4 (14) -9 (17)

88 310

-1 (14) -5 (15)

194 204

-2 (16) -6 (14)

35 363

0(11) -5 (15)

31 367

-4 (14) -5 (15)

219 179 58 340

-2 (16) -6 (14) -7 (15) -4 (15)

34 364

-1 (17) -5 (15)

32 366

-9 (13) -5 (15)

29 369

1 (14) -5 (15)

* Values are mean (standard error). COPD = chronic obstructive pulmonary disease. using diuretics had a -1-mm Hg decline in pressure compared with a -8-mm Hg decline in normotensive patients not taking diuretics. Hypertensive patients taking diuretics had a -3-mm Hg decline compared with a -2-mm Hg decline in hypertensive patients not taking diuretics. The F(I,394) value for an interaction between diuretic use and hypertension was THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES

Quintile supine diastolic pressure (F [4,373] = 4.71; P = 0.001) 0-68 70-77 78-82 84-89 90Cigarette use (t [395] = 2.06; P = 0.045) Nonsmoker Any current cigarette use Medication Antiarthritic (t = [397] = 0.82; P = 0.415) Current use Nonuser Diuretic (t = [397] = 2.17; P = 0.031) Current use Nonuser Electrolyte supplements (t = [397] = 1.77; P = 0.078) Current use Nonuser Bronchial disorder (t = [397] = 2.58; P = 0.010) Current use Nonuser Diagnosis Hypertension (t = [397] = 1.58; P = 0.114) Yes No 'Diabetes mellitus (t = [397] = 1.32; P = 0.186) Yes No Congestive heart failure (t = [397] = 2.45; P = 0.015) Yes No COPD (t = [397] = 1.82; P = 0.0001) Yes No Abnormal heart beat (t = [397] = 1.89; P = 0.059) Yes No

Number

Mean Postural Change in Systolic mm Hg*

78 87 85 64 84

4 (10) 0(10) -5 (10) -2 (9) -7 (13)

337 59

-2(11) -5(11)

88 310

-1 (12) -2(11)

194 204

-1 (13) -3 (13)

35 363

1(11) -2 (11)

31 367

-7(11) -2 (11)

219 179

-1 (11) -2 (10)

58 340

0(11) -2 (11)

34 364

2 (10) -5 (15)

32 366

-9 (12) -1 (11)

29 369

2 (13) -2 (13)

* Values are mean (standard error). COPD = chronic obstructive pulmonary disease. 291

Orthostatic Hypotension in the Elderly

Table 4. Stepwise Multiple Linear Equations for Factors Associated With Postural Changes in Blood Pressure Variable Entered

Regression Coefficient

Systolic pressure* Supine Hypertension COPD Arthritis drugs Abnormal heartbeat Diabetes Diastolic pressure:j: Supine COPD Hypertension Congestive heart failure

95% CI

-1.79t 6.79 -5.80 3.80 6.11 4.14

-1.22, +9.71, -0.70, +7.12, 11.43, +0.21,

-2.36 +3.88 -10.91 +0.49 +0.78 +8.08

-3.21t -8.89 3.72 5.13

-2.43, -5.26, +5.77, +8.65,

-3.99 -12.53 +1.66 +1.61

* Variance explained equals 14%. t Postural change in blood pressure per 10-mm Hg increase in supine pressure. :j: Variance explained equals 19%. CI = confidence interval; COPD = chronic obstructive pulmonary disease.

treated as a continuous dependent variable, six independent variables remained that were related to observed orthostatic changes. For each 10-mm Hg increase in supine systolic pressure, a -1. 79-mm Hg drop in orthostatic systolic pressure was anticipated. After further adjustment for the fact that orthostatic declines in pressure are based in part on the supine measurements,24 the drop increased to -4.00 mm Hg (standard error [SEl = 0.41 mm Hg). Systolic pressure on standing was increased by the presence of hypertension, decreased by the presence of COPD, increased if the patient took arthritis medications, increased by the presence of an abnormal heart beat, and decreased by the presence of diabetes. For orthostatic diastolic changes, supine diastolic pressure was the first variable to enter the model. For a 10-mm Hg increase in diastolic pressure in the supine position, diastolic pressure dropped -3.21mm Hg on standing. After adjustment for the calculation including supine pressure,24 the drop increased to -5.60 mm Hg (SE = 0.56 mm Hg). Diastolic pressure on standing was decreased by COPD, increased by hypertension, and increased by congestive heart failure. Stepwise multiple logistic regression analyses were performed to determine which variables were associated with clinically used definitions of orthostatic hypotension and with orthostatic declines in systolic or diastolic pressure of greater than 10 or 20 mm Hg, respectively. The adjusted odds ratios are given in Table 5. These odds ratios represent the change in likelihood that a patient would have orthostatic hypotension given the presence of the independent variable. 292

Discussion

This study confirms and quantifies the normal distribution of orthostatic blood pressure changes in elderly outpatients. The mean changes after 5 minutes of standing were small, a -4-mm Hg decline in systolic and a -2-mm Hg decline in diastolic pressure. The changes in pressure were unrelated to age, race, sex, symptoms, or other clinical factors noted in the results section. The mean changes in orthostatic blood pressure are similar to others reported in ambulatory elderly subjects8- 11 and are also similar to the orthostatic changes observed in surveys of healthy young persons. 26,27 The study confirms previous reports that supine blood pressure is a determinant of observed orthostatic changes. 8,28-33 The higher the supine pressure, the more likely the pressure was to fall on standing. The lower the supine pressure, the more likely the pressure was to rise on standing. Although a high measured level of supine blood pressure predicted a greater fall in pressure on standing, the label of hypertension was associated with smaller observed mean declines and an increase (in the multivariate analyses) in both standing systolic and diastolic pressures. This was observed even though hypertensives as a group had higher resting pressures than healthy subjects without hypertension. The observed association was not removed when the type of drug therapy used to treat the hypertension was factored in. Although this study cannot offer a definitive explanation for this seemingly paradoxical finding, an answer may be that elderly hypertensive patients have higher levels of circulating norepinephrine34 with attenuated catecholamine responses to tilt tests. 35 It is recognized that there is increased circulating norepinephrine in patients with congestive heart failure, which could explain the observed increase in standing pressures among persons labeled as having congestive heart failure or an abnormal heartbeat. In addition to peripheral edema, there may also be physical support of the venous capacitance vessels. There may also be effects on stroke volume together with decreased venous compliance and increased central blood volume. 36 - 38 The diagnosis of diabetes mellitus, a recognized cause of autonomic neuropathy and orthostatic hypotension, was not strongly associated with orthostatic changes in systolic or diastolic pressure because of its confounding with hypertension. In the linear regression analyses, after adjusting for hypertension, the presence of diabetes did predict a further 4-mm Hg fall in systolic pressure. In the logistic regressions, diabetes did not predict who would have large orthostatic decreases in blood pressure. The diagnosis of COPD was persistently and strongly associated with orthostatic changes in both systolic and diastolic pressures. This association was May 1998 Volume 315 Number 5

Robertson, Desjardin, and lichtenstein

Table 5. Stepwise Multiple Logistic Equations for Factors Associated With Postural Changes in Blood Pressure

Variable

Drop in Pressure> 10 mm Hg Odds Ratio (95% CD*

Drop in Pressure >20 mm Hg Odds Ratio (95% CDt

1.25 (1.14, 1.38) 0.75 (0.59, 0.96) 1.52 (1.03, 2.24)

1.23 (1.11, 1.37)

1.48 (1.21, 1.81) 1.75 (1.15, 2.67) 0.47 (0.23, 0.99)

2.21 (1.50, 3.25) 2.70 (1.40, 5.19)

Systolic blood pressure Supinet Hypertension COPD Diastolic blood pressure Supinet COPD Congestive heart failure

Values greater than 1 increase the odds of a postural blood pressure decline. Values greater than 1 decrease the odds of a postural blood pressure decline. * For systolic blood pressure, n = 111 and variance explained = 5%; for diastolic blood pressure; n = 72 and variance explained = 6%. t For systolic blood pressure, n = 53 and variance explained = 4%; for diastolic blood pressure; n = 16 and variance explained = 15%. t Odds ratio for a 10-mm Hg increase in supine pressure. COPD = chronic obstructive pulmonary disease.

not explained by associations with heart rate, respiratory rate, body mass, or supine blood pressure. In fact, the 32 patients with COPD had lower average supine pressures (140176 mm Hg) than the other 367 patients (151/80 mm Hg). The association persisted in both the linear and logistic regressions. To our knowledge, the association ofCOPD with orthostatic hypotension has not been previously reported. It is possible that increased amplitude of intrathoracic pressure changes may compromise venous return in the more severely affected patients but no truly satisfactory explanation for this apparent association is obvious. The results of this study must be interpreted within the limitations ofthe methodology. First, the response rate, due to logistic factors in conducting the survey, was low. We did not collect data on the elderly patients missed during the study and do not know if their clinical features would bias our results. However, the observed distribution and the magnitude of orthostatic blood pressure changes reported here are in accord with the work of other investigators. 8 ,9 Second, single blood pressure measurements were made before standing. Mader et al. has demonstrated that repeated supine measurements result in lower baseline pressures and consequently the level of orthostatic change. 39 The effect on our results would be to decrease the degree of observed orthostatic change by approximately 2 mm Hg. If the effect were constant across all levels of supine pressure, the linear regressions and associations would be unaffected. Because the further decrease in orthostatic pressure changes would alter the prevalence of decreases of greater than 10 or 20 mm Hg, the logistic regression analyses might be changed. Third, standing blood pressure was measured after 5 minutes. In most patients, blood pressure THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES

falls on standing and reaches a nadir between 30 and 90 seconds. 40 - 42 The pressure then rises and stabilizes after 3 to 5 minutes. We chose the stable 5minute reading because we felt measuring blood pressure at 1 minute with mercury sphygmomanometers would be inaccurate. For a research study it takes 30 to 45 seconds to measure blood pressure carefully using mercury sphygmomanometers, and use of this technique to assess orthostatic changes after 1 minute would not accurately assess the blood pressure nadir. Although we found few associations among clinical measures, symptoms, and orthostatic changes at 5 minutes, clinically important associations might be detected ifthe orthostatic nadir could be accurately indirectly assessed, as with tonomet ry43 or a beat-to-beat tracking system. 44 Our effort was directed toward making a few careful indirect blood pressure measurements that were as free of potential observer bias as possible. Finally, the data are cross-sectional, and one cannot directly infer causal relationships between the measured variables. 45 The small amount of variance explained by the models indicates that much has yet to be learned about the clinical determinants of orthostatic blood pressure changes. Lipsitz and colleagues pointed out the marked day-to-day variation in supine pressures and orthostatic changes. 29 Perhaps repeated measures would better characterize persons with orthostatic hypotension and account for more of the variance in orthostatic blood pressure changes. In spite of these limitations, our survey provides data that should be clinically useful to physicians. As is the case in making a diagnosis of hypertension, orthostatic changes in blood pressure should be measured on several occasions before a patient is considered to have orthostatic hypotension. On average, the orthostatic drop in pressure is small and unre-

293

Orthostatic Hypotension in the Elderly

lated to age. In our elderly clinical patients, the presence of a high-resting blood pressure and COPD increased the odds of having large orthostatic drops in blood pressure (> 20 mm Hg in systolic or diastolic pressure). The presence ofthe diagnoses of hypertension or congestive heart failure decreased the odds of having a greater than 10-mm Hg drop in systolic or diastolic pressure, respectively.

18.

19.

20.

Acknowledgments

The authors thank the members of the Division of General Internal Medicine at Vanderbilt University for allowing them access to their patients. Barbara Forbes, Sue MacArthur, and Sue Sieveking helped with data acquisition. The authors also thank Mrs. Dorothea Boerner and Mrs. Varonica Watts for help with manuscript preparation.

22.

References

23.

1. Kapoor W, Snusted D, Peterson J, Wieand HS, Cha R, Karpf M. Syncope in the elderly. Am J Med. 1986;80:41928. 2. Ashley MJ, Gryfe CI, Amies A. A longitudinal study offalls in an elderly population. II. Some circumstances of falling. Age Ageing. 1977;6:211-20. 3. Liu BA, Topper AK, Reeves RA, Gryfe C, Maki BE. Falls among older people: Relationship to medication use and orthostatic hypotension. JAm Geriatr Soc. 1995;43:1141-5. 4. Vogt TM. Falls, fallacies, and hypertension. J Am Geriatr Soc. 1995;43:1169-73. 5. Sparrow D, Tifft CP, Rosner B, Weiss ST. Postural changes in diastolic blood pressure and the risk of myocardial infarction: The Normative Aging Study. Circulation. 1984; 70:533-7. 6. Lipsitz LA. Orthostatic hypotension in the elderly. N Engl J Med. 1989;321:952-7. 7. Mader SL. Aging and postural hypotension: An update. J Am Geriatr Soc. 1989;37:129-37. 8. Myers MG, Kearns PM, Kennedy DS, Fisher RH. Postural hypotension and diuretic therapy in the elderly. Can Med Assoc J. 1978; 119:581-5. 9. Mader SL, Josephson KR, Rubenstein LZ. Low prevalence of postural hypotension among community-dwelling elderly. JAMA. 1987;258:1511-4. 10. Applegate WB, Davis BR, Black HR, Smith WM, Miller ST, Burlando AJ. Prevalence of postural hypotension at baseline in the Systolic Hypertension in the Elderly Program (SHEP) cohort. JAm Geriatr Soc. 1991;39:1057-64. 11. Harris T, Lipsitz LA, Kleinman JC, Cornoni.Huntley J. Postural change in blood pressure associated with age and systolic blood pressure:· The National Health and Nutrition Examination Survey II. J Gerontol. 1991;46:M159-63. 12. Schatz IJ, Bannister R, Freeman RL, Goetz CG, Jankovic J, Kaufmann HC, et al. Consensus statement on the definition of orthostatic hypotension, pure autonomic failure, and multiple systems atrophy. Neurology. 1996;46:1470-1. 13. Wright BM, Dore CF. A random-zero sphygmomanometer. Lancet. 1970; 1:337 -8. 14. American Heart Association-Minnesota Affiliate. Measuring Blood Pressure [videotape]. 1984. 15. Revicki DA, Israel RG. Relationship between body mass indices and measures of body adiposity. Am J Public Health. 1986; 76:992-94. 16. Messerli FH, Ventura HO, Amodeo C. Osler's maneuver and pseudohypertension. N Engl J Med. 1985;312:1548-51. 17. Pfeiffer E. Short portable mental status questionnaire for

294

21.

24. 25.

26. 27. 28.

29. 30.

31.

32.

33. 34. 35.

36.

assessment of organic brain deficit in elderly patients. J Am Geriatr Soc. 1975;23:433-41. Pincus T, Summey JA, Soraci SA, Wallston KA, Hummon NP. Assessment of patient satisfaction in activities of daily living using modified Stanford Health Assessment Questionnaire. Arthritis Rheum. 1983;26:1346-53. Bradstock MK, Marks JS, Forman MR, Gentry EM, Hogelin GS, Trowbridge FL. The behavioral risk factor surveys. III. Chronic heavy alcohol use in the United States. Am J Prev Med. 1985;1:15-20. World Organization of National Colleges, Academies, and Academic Associations of General Practitioners! Family Physicians and the World Health Organization. ICHPPC-2-Defined: International Classification of Health Problems in Primary Care, 3rd ed. Oxford, England: Oxford University Press; 1983:542-8. AMA Division of Drugs and the American Society for Clinical Pharmacology and Therapeutics. AMA Drug Evaluations, 5th ed. Chicago: American Medical Association; 1983. Armitage P. Regression and correlation. In: Statistical Methods in Medical Research. Oxford, England: Blackwell Scientific Publications; 1971:147-66. Schlesselman JJ. Basic methods of analysis. In: Case-Control Studies: Design, Conduct, Analysis. Oxford, England: Oxford University Press; 1982:171-226. Svardsudd K, Blomqvist M. A new method for investigating the relation between change and initial value in longitudinal blood pressure data. Scand J Soc Med. 1978;6:85-95. Breslow NE, Day NE. Unconditional logistic regression for large strata. In: Statistical Methods in Cancer Research, vol. 1. The Analysis of Case Control Studies. Lyon, France: International Agency for Research on Cancer; 1980:192-247. Schneider EC, Truesdell D. A statistical study ofthe pulse rate and the arterial pressures in recumbency, standing and after a standard exercise. Am J Physiol. 1922;61:429-74. Currens JH. A comparison ofthe blood pressure in the lying and standing positions: A study of five hundred men and five hundred women. Am Heart J. 1948;35:646-54. Harris T, Kleinman J, Lipsitz L, Cornoni-Huntley JI, Garrison R. Is age or level of systolic blood pressure related to positional blood pressure change (abstract)? Gerontologist. 1986; 26(suppl):26a. Lipsitz LA, Storch HA, Minaker KL, Rowe JW. Intraindividual variability in postural blood pressure in the elderly. Clin Sci. 1985;69:337-41. Masuo K, Mikami H, Ogihara T. The frequency of orthostatic hypotension in elderly patients with essential hypertension, isolated systolic hypertension and borderline hypertension. J Hypertens Suppl. 1993; l1(suppl 5):S306-7. Raiha I, Luutonen S, Piha J, Seppanen A, Toikka T, Sourander L. Prevalence, predisposing factors, and prognostic importance of postural hypotension. Arch Intern Med. 1995; 155:930-5. Tilvis RS, Hakala SM, Valvanne J, Erkinjuntti T. Postural hypotension and dizziness is a general aged population: A four-year follow-up of the Helsinki Aging Study. JAm Geriatr Soc. 1996;44:800-14. Ward C, Kenny RA. Reproducibility of orthostatic hypotension in symptomatic elderly. Am J Med. 1996; 100:418-22. Kaplan NM. Primary (essential) hypertension: Pathogenesis. In: Kaplan NM; ed. Clinical Hypertension, 4th ed. Baltimore: Williams and Wilkins; 1986:56-122. Jansen RWMM, Lenders JWM, Thien T, Hoefnagels WHL. The influence of age and blood pressure on the hemodynamic and humoral responses to head-up tilt. J Am Geriatr Soc. 1989; 37:528-32. Abelmann WH, Fareeduddin K. Circulatory response to upright tilt in patients with heart disease. Aerospace Medicine. 1967;38:60-5. May 1998 Volume 315 Number 5

Robertson, Desjardin, and Lichtenstein

37. Murray RHI, Thompson LT, Bowers JA, Albright CPo Hemodynamic effects of graded hypovolemia and vasodepression syncope induced by lower body negative pressure. Am Heart J. 1968; 76:799-811. 38. Jacob G, Biaggioni I, Mosqueda-Garcia R, Robertson RM, Robertson D. Relation of blood volume and blood pressure in orthostatic intolerance. Am J Med Sci. 1998; 315(2):95-100. 39. Mader SL, Palmer RM, Rubenstein LZ. Effect of timing and number of baseline blood pressure determinations on postural blood pressure response. J Am Geriatr Soc. 1989;37:444-6. 40. Thulesius YO. Pathophysiological classification and diagnosis of orthostatic hypotension. Cardiology. 1976;61:180-90. 41. Onrot J, Goldberg MR, Hollister AS, Biaggioni I, Rob-

THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES

42. 43.

44.

45.

ertson RM, Robertson D. Management of chronic orthostatic hypotension. Am J Med. 1986; 80:454-64. Robertson D, Robertson RM. Etiology of chronic orthostatic hypotension. Arch Int Med. 1994; 154:1620-4. Kemmotsu 0, Veda M, Otsuka K, Winter DC, Eckerle JS. Evaluation of arterial tonometry for non-invasive continuous blood pressure monitoring during anesthesia (abstract). Anesthesiology. 1988; 71:A406. Shapiro D, Greenstadt L, Lane JD, Rubinstein E. Tracking-cuff system for beat-to-beat recording of blood pressure. Psychophysiology. 1981; 18:129-36. Hennekens CH, Buring JE. Descriptive studies. In: Mayrent SL; ed. Epidemiology in Medicine. Boston: Little, Brown, and Company; 1987:101-31.

295