Postural hand tremor and incident hypertension in young to middle-aged adults: the Bogalusa heart study

Abstract

Background:

Hand tremor and blood pressure (BP) are both increased by adrenergic stimulation and reduced by β-blockade, indicating that they may share a common underlying pathophysiology.

Methods:

We prospectively examined the relationship between postural hand tremor and incident hypertension in a community-based cohort of 715 (184 blacks and 531 whites) adults without hypertension and not using medications to control tremor (e.g. β-blockers). At baseline, tremor was measured with participants holding a laser pointer aimed at a sheet of Polaroid film 8 feet away with arm outstretched for 8 s in a darkened room, and characterized by the width of the circle diameter encompassing all exposures and enumeration of exposure dots in the same area. Incident hypertension was defined as new elevation of BP (SBP≥140 or DBP≥90 mmHg, based on an average of six readings over two visits) or antihypertensive medication use.

Results:

During a median follow-up of 6.4 years, 198 (69 blacks and 129 whites) participants developed hypertension. Tremor measurements (by quartile) were positively associated with incident hypertension after adjustment for baseline demographics, lifestyle characteristics, and BP. There was significant interaction by race (P=0.01). Among whites, tremor was positively associated with incident hypertension [hazard ratio highest vs. lowest quartile: 2.50 (95% confidence interval: 1.40–4.48) dot method and 3.24 (1.78–5.90) circular method; both P trend <0.01]. Among blacks, tremor was not associated with hypertension.

Conclusion:

In this community-based cohort, postural hand tremor was strongly associated with the risk of incident hypertension among whites and merits further study as a potential indicator of risk for hypertension.

INTRODUCTION

Physiologic tremor is a normal oscillatory motion present in every healthy person, which can be amplified with the outstretched hand held against gravity in the postural position [1,2]. Physiologic tremor has a low-amplitude and moderate-to-high frequency and is increased by sympathetic nervous stimulation and decreased by blockade of adrenergic receptors [3–5]. A large body of evidence has also shown that sympathetic stimulation is associated with elevated blood pressure (BP) and hypertension [1,2,4–6], suggesting a potentially common underlying pathophysiology.

If the presence of a postural hand tremor is associated with hypertension, such a marker could potentially be a simple indicator of the need for more intensive preventive measures in individuals at risk for hypertension and its consequences; however, that relationship has not been examined in population-based settings. We examined the association between postural hand tremor and incident hypertension in a biracial, community-based population. Additionally, we explored that relationship by race, given the unequal geographic distributions of tremor and hypertension [7–9].

METHODS

Study population

The Bogalusa Heart Study was begun in 1973 by Dr Gerald Berenson as a series of long-term prospective studies of a semirural, biracial (65% white and 35% black) community in Bogalusa, Louisiana [10,11]. We used data from men and women who were examined in 2000–2002 and participated in follow-up examinations in 2003–2006 and 2007–2010. We initially included 914 participants at baseline who were not using medications related to tremor or hypertension (e.g. β-blockers) and who were free of prevalent hypertension or coronary heart disease. All participants had left ventricular ejection fraction at least 45%. Six participants who lacked tremor measurements and 193 participants who did not attend a follow-up examination were excluded, resulting in an analysis cohort of 715 participants. The demographic characteristics and cardiovascular profile of the 715 participants who were included and 199 who were excluded were similar. The study was approved by the Tulane University institutional review board, and signed informed consent was obtained from all participants.

Tremor measurement

All study data were collected by trained examiners using standardized Bogalusa Heart Study protocols [11,12]. Participants underwent an assessment of postural hand tremor at the examination in 2000 that served as the baseline for this analysis. They were brought to a dark room that was equipped with a red safe light for photo development and were allowed to remain in the room until their eyes became accustomed to the lighting. With arm outstretched at the shoulder in a 90° angle to the body, participants held a laser pointer aimed at a circle in the center of an 8.5 by 10.8 cm sheet of Polaroid 669 film held in a73B film holder, outlined by glowing photographic tape, attached firmly to a wall located 8 feet away. Participants were allowed to practice holding the laser pointer at the center of the film holder for 5 s with the right, and then the left arm extended straight out, with the tip of the laser pointer 8 feet from the target. After this initial practice, the film cover was removed, exposing the film, and the participant was instructed to hold the laser pointer at the center of the film for 8 s with the right hand. The trial was repeated after a break of 75–90 s [13–15]. During the interim between tests, a trained technician slid the film cover to close the film holder and pulled the white tab to develop the film. The same process was repeated with the left hand for a total of two trials on each side.

Two measurement methods were used to characterize postural tremor in both dominant and nondominant hands. The first method measured the widest circle diameter encompassing all exposures on the light-sensitive Polaroid film and was termed the circle diameter method. The second method enumerated exposure dots (dots representing longer exposures in one location) from a preset grid in the contained area. This second method was termed the dot enumeration method. A larger circle diameter or a greater dot number represented a greater level of tremor. The four values of circle diameters or dot numbers, two trials per hand, were averaged and then categorized into quartiles for the analysis. Correlation coefficients for the circular diameter method were 0.93 and 0.70 in the left and right hands, respectively. Corresponding numbers for the dot enumeration method are 0.97 and 0.96, respectively (all P<0.001).

Hypertension diagnosis

Identical protocols were used to measure BP across all surveys in the Bogalusa Heart Study [11,12]. At each follow-up examination, BP measurements were obtained between 0800 and 1000 h by trained observers on the right arm of the participants in a relaxed, sitting position. Participants were required to be seated quietly for 5 min in a chair prior to BP measurement. Arm measurements, length and circumference, were made during the examination to ensure proper cuff size. SBP and DBP levels were recorded using a mercury sphygmomanometer. BP levels were reported as the mean of six replicate readings, taken by each of two randomly assigned observers (three replicates each). Incident hypertension was defined as a new measurement of average SBP at least 140mmHg or average DBP at least 90mmHg or a new self-report of hypertension or new treatment with an antihypertensive medication.

Covariates

At baseline, height and weight were measured twice to the nearest 0.1 cm and nearest 0.1 kg, respectively. Averaged values were used to calculate BMI as weight in kilograms divided by height in meters squared. Waist circumference was measured twice in the horizontal plane at 1 cm above the navel. Venipuncture was performed after an overnight fast. Serum cholesterol and triglycerides levels were assayed using an enzymatic procedure as part of a lipid panel (Laboratory Corporation of America, Burlington, North Carolina, USA). Fasting plasma glucose levels were measured by a glucose oxidase method (Laboratory Corporation of America). Self-reported current smoking status (yes/no) was defined as smoking one or more cigarettes within the past year. Self-reported alcohol consumption was categorized as drinking alcoholic beverages at least three times a week, drinking less than twice a week, or no alcohol consumption. Self-reported leisure-time physical activity was assessed on a scale of 1 (very inactive) to 5 (very active) on the basis of a validated questionnaire [16,17]. Heart rates (HRs) were measured three times by counting the radial pulse at the BP station and then averaged.

Statistical analysis

Baseline characteristics were presented as mean (SD) or N (%) among persons with or without hypertension, separately, and were compared using Student’s t test or χ² test when appropriate. Cumulative rates for hypertension were calculated using the Kaplan–Meier method. Incident rates were also calculated using Poisson regression model after adjustment for age, sex, and race. For the primary analysis, we used complementary log–log regression analysis to estimate hazard ratios for the association between tremor variables (by quartile) and incident hypertension, as the exact date of hypertension development or diagnosis was not known with greater precision than the follow-up interval. This approach assumes that the hazard is constant within each follow-up interval but allows for variation of the hazard between intervals [18]. For these complementary log–log regression models, person-time accrued until the development of incident hypertension or the date of the participant’s last visit, whichever came first. The basic model included age, sex, and race as covariates. The main model was further adjusted for educational level, SBP, BMI, LDL cholesterol, HDL cholesterol, fasting glucose, self-reported current smoking status, alcohol consumption, and leisure-time physical activity at baseline. Exploratory analysis stratified by race was performed to evaluate for possible effect modification. Secondary analysis was performed using traditional Cox proportional hazard regression models. Another exploratory, secondary analysis was performed using mixed-effects repeated measures regression models to examine the association between tremor variables and SBP or DBP as a continuous outcome. These mixed-effects models were adjusted for time and aforementioned covariates; if the interaction term for tremor variable by time was significant, this interaction term was added to the models. Sensitivity analyses were performed to examine the potential effect of thyroid medication use and HR levels at baseline. Analyses were performed using SAS 9.4 for Windows (SAS Institute, Cary, North Carolina, USA).

RESULTS

Baseline characteristics

At baseline, the mean age of the study cohort was 36 (SD: 4.3) years. Of the 715 participants included in the analysis, 184 (25.7%) were black and 278 (38.9%) were man. Compared with participants who were included in the analysis, those who were lost to follow-up or did not attend exams were slightly younger (mean age at baseline: 35, SD: 4.9), but were similar in their distribution by sex and race. The participants who developed hypertension had significantly higher baseline levels of SBP and DBP, BMI, waist circumference, and fasting glucose (Table 1).

TABLE 1.

Baseline characteristics by hypertension status at follow-up

Variablesa	Hypertension N = 198	Normal N = 517	P valueb	Overall N = 715
Age (year)	36.2 (4.5)	36.2 (4.3)	0.95	36.2 (4.3)
Man	78 (39.4%)	200 (38.7%)	0.86	278 (38.9%)
Black	69 (34.9%)	115 (22.2%)	<0.001	184 (25.7%)
Less than college	122 (61.9%)	276 (54.1%)	0.06	398 (56.3%)
BMI (kg/m2)	31.0 (7.3)	27.6 (5.9)	<0.001	28.5 (6.5)
Waist circumference (cm)	97.1 (17.2)	88.9 (14.8)	<0.001	91.2 (15.9)
SBP (mmHg)	118.6 (8.5)	110.2 (8.7)	<0.001	112.6 (9.4)
DBP (mmHg)	80.0 (6.0)	74.5 (6.7)	<0.001	76.0 (7.0)
Total cholesterol (mg/dl)	189.8 (34.0)	187.2 (36.4)	0.39	188.0 (35.8)
LDL cholesterol (mg/dl)	123.6 (29.4)	122.1 (32.1)	0.56	122.5 (31.3)
HDL cholesterol (mg/dl)	47.3 (14.5)	48.4 (13.1)	0.30	48.1 (13.5)
Fasting glucose (mg/dl)	89.3 (23.5)	84.0 (22.0)	0.005	85.5 (22.5)
Diabetes	9 (4.6%)	11 (2.1%)	0.08	20 (2.8%)
Alcohol consumption			0.96
None	123 (62.1%)	324 (62.7%)		447 (62.5%)
Less than twice a week	50 (25.3%)	126 (24.4%)		176 (24.6%)
At least 3–4 times a week	25 (12.6%)	67 (13.0%)		92 (12.9%)
Current smoking	64 (32.3%)	137 (26.7%)	0.13	201 (28.2%)
Leisure-time physically inactivec	139 (70.2%)	342 (66.2%)	0.30	481 (67.3%)
Tremor variables
Dot numbers (right hand)	72 (38)	67 (33)	0.009	68 (34)
Dot numbers (left hand)	81 (42)	72 (38)	0.02	75 (39)
Dot numbers (both hands)	76 (38)	69 (33)	0.01	71 (35)
Maximized circle diameter (right hand) (cm)	3.5 (1.0)	3.3 (0.9)	0.04	3.4 (1.0)
Maximized circle diameter (left hand) (cm)	3.7 (1.0)	3.5 (1.0)	0.02	3.6 (1.0)
Maximized circle diameter (both hands) (cm)	3.6 (1.0)	3.4 (0.9)	0.006	3.5 (0.9)

^aData were presented as mean (SD) or N (%) when appropriate.

^bP values for the comparison between hypertensive and normotensive participants were from Student’s t test or χ² test when appropriate.

^cLeisure-time physically inactive was defined as self-reported physical ‘very inactive’, ‘inactive’ or ‘unknown’.

Level of tremor, as measured by either the number of dots or the circle diameter, was presented at baseline. The mean number of dots was 71 (SD: 35) and the mean maximized circle diameter was 3.5 (SD: 0.9) cm. Compared with whites, blacks tended to have a greater number of dots and a larger maximized circle diameter [mean number of dots: 83 (SD: 42) in blacks vs. 67 (SD: 31) in whites; mean circle diameter: 3.7 (SD: 1.0) cm in blacks vs. 3.4 (SD: 0.9) cm in whites]. The participants who developed hypertension also had higher baseline levels of tremor for both hands (P<0.05) (Table 1).

Follow-up

During a median follow-up of 6.4 years, 198 (27.7% of the sample, including 129 whites and 69 blacks) participants developed hypertension. Using either characterization method, greater levels of postural hand tremor were positively associated with incident hypertension after adjustment for age, sex, race, educational level, SBP, BMI, LDL cholesterol, HDL cholesterol, fasting glucose, self-reported current smoking status, alcohol consumption, and leisure-time physical inactivity at baseline (Table 2, Fig. 1a and b). There was significant interaction by race in the main model using either the dot enumeration method or the circle diameter method (P=0.01). Among whites, postural hand tremor was positively associated with incident hypertension. In the main model, hazard ratios of highest vs. lowest quartiles of dot numbers and circle diameters were 2.50 [95% confidence interval (CI): 1.40–4.48; P for linear trend=0.002) and 3.24 [95% CI: 1.78–5.90; P for linear trend <0.001], respectively (Fig. 2a and b). Among blacks, postural hand tremor was not associated with risk of incident hypertension. Results were similar when the analysis was performed using Cox proportional hazard regression models (Supplemental Table, http://links.lww.com/HJH/A621). For example, in the overall sample, the hazard ratios for the highest quartile vs. the lowest quartile in the main model were 2.59 (95% CI: 1.65–4.06) for the dot enumeration method and 2.93 (95% CI: 1.86–4.61) for the circle diameter method (Both P for linear trend <0.001). Further adjustment for baseline HR did not change the results. Results were similar after exclusion of 10 participants who were using thyroid medications at baseline (data not shown).

TABLE 2.

Adjusted hazard ratio (95% confidence interval) by complementary log–log analysis for the association between tremor variables and incident hypertension over a median of 6.4 years of follow-up

	Quartile 1 (ref)	Quartile 2	Quartile 3	Quartile 4	P for linear trend
Dot numbers
N	181	176	179	179
Median	39.8	55.3	74.0	107.8
Range	16.5–46.8	47.0–63.5	63.8–87.3	87.8–355.5
Number of events	44	48	47	59
Incidence ratesa (per 1000 person-years)	38.5 (28.7, 51.8)	44.3 (33.4, 58.7)	46.0 (34.6, 61.2)	57.8 (44.8, 74.7)
Model 1a	1	1.13 (0.75, 1.70)	1.12 (0.73, 1.70)	1.45 (0.95, 2.21)	0.07
Model 2b	1	1.30 (0.84, 2.01)	1.45 (0.92, 2.26)	1.82 (1.17, 2.85)	0.008
Maximized circle diameter (cm)
N	178	178	178	178
Median	2.5	3.1	3.7	4.5
Range	1.2–2.8	2.9–3.4	3.5–4.0	4.1–8.2
Number of events	37	42	56	63
Incidence ratesa (per 1000 person-years)	32.5 (23.6, 44.9)	42.5 (31.4, 57.5)	51.6 (39.7, 67.0)	60.6 (47.4, 77.6)
Model 1a	1	1.26 (0.81, 1.98)	1.68 (1.10, 2.55)	1.93 (1.26, 2.95)	0.01
Model 2b	1	1.62 (1.01, 2.60)	1.88 (1.22, 2.90)	2.37 (1.51, 3.73)	<0.001

^aAdjusted for age, sex, and race.

^bAdjusted for age, sex, race, educational level, SBP, BMI, LDL cholesterol, HDL cholesterol, fasting glucose, self-reported current smoking status, alcohol consumption, and leisure-time physical inactivity at baseline.

FIGURE 1.

Cumulative rates for hypertension by quartile of tremor variables. (a) Dot number quartile. (b) Circle diameter. Log rank test P less than 0.001 for both tremor variables.

FIGURE 2.

Adjusted hazard ratios (95% confidence intervals) by complementary log–log analysis for the association between tremor variables and incident hypertension by race over a median of 6.4 years of follow-up. Hazard ratios were adjusted for age, sex, educational level, SBP, BMI, LDL cholesterol, HDL cholesterol, fasting glucose, self-reported current smoking status, alcohol consumption, and leisure-time physical inactivity at baseline. (a) Dot number quartile. (b) Circle diameter.

In the mixed-effects repeated measures analysis, tremor variables were associated with higher levels of SBP and DBP. In the entire cohort, mean SBP for the highest quartile vs. the lowest quartile of postural tremor was 8.3 (95% CI: 5.0–10.6)mmHg greater using the dot enumeration measure and 8.1 (95% CI: 4.9–10.4)mmHg greater using the circle diameter measurement. Corresponding numbers for DBP were 4.2 (95% CI: 3.1–5.4)mmHg for the highest vs. lowest quartile of postural hand tremor using the dot enumeration method and 4.5 (95% CI: 3.3–5.7)mmHg using the circle diameter method. Greater differences in continuous BP measures were seen across quartiles of postural hand tremor among whites as compared with blacks.

DISCUSSION

In this community-based biracial cohort, a strong dose–response relationship was observed between levels of postural hand tremor and subsequent risk of incident hypertension. The results from this study suggest that postural hand tremor may be an indicator of risk for hypertension. To our knowledge, this is the first study to document the association between postural hand tremor and hypertension in a population-based cohort. Given the high prevalence of hypertension and its sequelae, this study also has important clinical and public health implications.

Few studies have examined the relationship between hand tremor and hypertension. A cross-sectional study of 32 individuals reported a correlation between physiologic hand tremor and BP levels [19]. A prospective study based on medical records collected between 1935 and 1979 in Olmsted County, Minnesota included 155 cases and age–sex-matched controls without essential tremor and reported no difference in onset of hypertension among individuals with and without essential tremor [20]. Our study, in contrast, was significantly larger and included a biracial population. Tremor was measured and quantified using the average of two trials in both hands, increasing precision. The Minnesota study used information abstracted from the medical record that is likely to be an incomplete representation of exposures and confounding factors.

The mechanisms underlying the association between postural hand tremor and hypertension are unclear but several possible pathways have been suggested. The endogenous catecholamines epinephrine and norepinephrine are likely to represent an important link between the two conditions [3,21]. It is well accepted that β-blockers reduce the work of the heart and, in turn, lower BP. The nonselective β-blocker, propranolol, is effective in reducing tremor, whereas selective β-blockade is not efficacious [22]. Therefore, peripheral β₂ adrenergic receptors appear to play a role. The presence of postural hand tremor may indicate increased levels of sympathetic activity in the periphery, which may result in an elevation in systemic vascular resistance, such as that which occurs in hypertension [23,24]. Evidence suggests that tremor may correlate with white matter structural abnormalities, with persons having white matter lesions also having significantly higher values of SBP and DBP [25,26].

Physiologic studies also provide insight into the apparent racial difference in the relationship of postural hand tremor to risk of hypertension. BP response to sympathetic blockade tends to be greater in whites compared with blacks, suggesting that the role of endogenous catecholamines in the development of hypertension may be more dominant in whites [27]. Genetic polymorphisms in the β₁-adrenergic receptor and associated regulatory proteins have been reported to contribute to racial differences in responses to sympathetic blockade [28,29]. If the degree of postural hand tremor is linked to β adrenergic activity, white as compared with black individuals who have such a tremor may be more prone to develop hypertension [30–32].

The study has several strengths. As a part of the Bogalusa Heart Study, standardized protocols and well established quality control procedures were used for data collection at each of the examinations [33,34]. This study was conducted in a biracial community-dwelling population, which enhances the generalizability of the results and enables exploration of the relationship in each race group. However, the conclusions from this study are subject to some limitations. This is an observational study, so the possibility of residual confounding cannot be eliminated. Given the smaller number of black participants, the power to detect an association in this group was lower in the black compared with white participants. Measurement error is inherent in any technique for BP measurement, although it is reduced with a greater number of measurements such as in 24-h ambulatory BP. Although 24-h ambulatory BP was not measured in our study, given that staff who measured BP were blinded to tremor variable levels, the measurement error in BP and incident hypertension is likely nondifferential with regard to tremor variables, resulting underestimation of the associations. The measurement of postural hand tremor had only moderate-to-high reliability. Thus, we used the average values from multiple measures of hand tremor in examining the associations.

Postural hand tremor is easy to measure in a clinical setting with simple and inexpensive equipment such as the laser pointer technique used here and in other studies [14]. Our findings indicate that postural hand tremor may be a strong predictor of incident hypertension. If this is borne out by further research, measurement of postural hand tremor could potentially serve clinicians as an early indicator of elevated risk for hypertension or could be used as part of a risk stratification algorithm that might signal the need for more intensive dietary and lifestyle interventions to prevent the development of hypertension. In conclusion, postural hand tremor was strongly associated with the risk of incident hypertension among whites in this biracial community-based cohort. Postural hand tremor merits further study as a potentially simple clinical indicator of high risk for the development of hypertension.

Abbreviations:

95% CI

95% confidence interval

BP

blood pressure