Health status and hypertension: A population-based study

Health status and hypertension: A population-based study

J Clin Epidemiol Vol. 49, No. 11, pp. 1239-1245, 0895-4356/96/$15.00 PII SO895~4356(96)00220-X 1996 Copyright 0 1996 Elsevier Science Inc. ELSEVI...

949KB Sizes 0 Downloads 7 Views

J Clin

Epidemiol

Vol. 49, No. 11, pp. 1239-1245,

0895-4356/96/$15.00 PII SO895~4356(96)00220-X

1996

Copyright 0 1996 Elsevier Science Inc. ELSEVIER

Health

Status and Hypertension:

A Population-Based

Study

William F. Lawrence, Dennis G. Fryback, Patricia A. Martin, Ronald Klein, and Barbara E.K. Klein DEPARTMENTS

OF MEDICINE,

PREVENTIVE

MEDICINE, MADISON,

AND

OPHTHALMOLOGY,

UNIVERSITY

OF WISCONSIN,

WISCONSIN

ABSTRACT. We describe the relation between self-reported hypertension and measures of health-related quality of life (HRQOL) in a community-dwelling population. In a cross-sectional study, 1430 randomly selected adults, aged 45 to 89 years, were interviewed to obtain a medical history and health status measures, including the SF-36 questionnaire, the Quality of Well Being (QWB) index, and time trade-off (TTO) assessments. A total of 5 19 participants reported being affected by hypertension (HTN group). The HTN group, compared to the No HTN group, had significantly lower age-adjusted health status scores measured by the between groups for each measure General Health (GH) scale of the SF-36 and by TTO, with differences comprising approximately 5% of the total scale. HTNs also had a significant decline in general health status measures associated with increasing numbers of antihypertensive medications but not with specific classes of medications. We conclude that hypertension and hypertension drug therapy are associated with clinically meanineful decreases in reuorted health status. I CLIN EPIDEMIOL 49:11:1239-1245. 1996. KEY WORDS. Health status, quality of life (health-related), study, cross-sectional study INTRODUCTION It is difficult

to make

the asymptomatic

patvat feel better. Hoerr’s Law [l]

Clinicians often have to deal with trade-offs between quality of life and quantity of life when making decisions in diagnosis and therapy. Hypertension, especially in mild to moderate stages, is usually considered an asymptomatic condition. Treatment of hypertension, particularly with pharmacologic therapy, may be associated with adverse effects that potentially make an asymptomatic person symptomatic. Detection of hypertension may subject people to a labeling effect; people who are otherwise “healthy” now have an “illness.” Several studies have shown that the diagnosis of hypertension was associated with increased absenteeism from work due to illness [2], and with increased number of visits to physicians [3]. Avoiding adverse effects in treating an asymptomatic condition can be difficult, as S. 0. Hoerr has noted. Thus diagnosis and treatment of hypertension potentially involve balancing a decreased future risk of cardiovascular morbidity and mortality with a decreased current health-related quality of life. As many as 50 million people in the United States either have an elevated blood pressure or are taking antihypertensive medications [4]; given the frequency of this condition, it is important that clinicians and researchers understand that the impacts of hypertension and hypertension therapy are on quality of life. Address

reprint

requests w: William

Lawrence,

Section

of General

Internal

Medicine, L>epartment of Medicine, IS/210 CSC, Umversity of Wisconsin Flospitals and Clinics, 600 Highland Ave.. Ma&son, Wisconsin 53792. This pper was presented in part at rhe Sixteenth Annual Meermg of the Society of General Internal Medicme, Arlington, Virginia, April 1993, and ar the Fifteenth Annual Meeting of the Sociery for Medical Decision Making, Research Triangle Park, North Carolma, Ocr&er 1993.

Accepted for puhlicarion IXI 30 April 1996.

hypertension,

antihypertensive

agents, population

While the long-term benefits of treating hypertension have been extensively studied, the effects of hypertension diagnosis and treatment are not as well known. Several randomized clinical trials of antihypertensive medications have examined health and functional status measures [5-131. While these studies have been helpful in comparing health status effects of specific antihypertensive medications, most tend to be relatively short-term studies, and may not reflect clinical practice well due to the necessities of study design. The Medical Outcomes Study examined the cross-sectional changes in health status associated with hypertension [ 141, showing a lower general health perception score in hypertensives compared to those with no chronic conditions. Since the Medical Outcomes Study was based on patients visiting physicians’ offices, it may have been subject to selection bias. The Beaver Dam Health Outcomes Study (BDHOS) represents a unique opportunity to understand further the effects of diagnosis and treatment of hypertension on health-related quality of life (HRQOL). The BDHOS is a population-based epidemiologic study that evaluates the health status of randomly selected adults over age 45 years in Beaver Dam, Wisconsin [15]. Health status is measured using two widely used health questionnaires, as well as a preferencebased assessment technique to measure overall HRQOL. We observed the relationship between hypertension diagnosis and therapy on HRQOL within this community, where adults with hypertension are likely to have had this condition on a relatively long-term basis, allowing them time to adjust and accommodate to their hypertension and to their antihypertensive medications. We hypothesized that the diagnosis and treatment of hypertension in this cohort would be associated with lower health status. METHODS

Participant

Selection

The protocol for the BDHOS 1s ‘. d escribed in detail elsewhere [15]. Briefly, the BDHOS draws from the cohort of participants in the

1240

W. F. Lawrence

Beaver Dam Eye Study (BDES), a population-based study of eye disease prevalence and risk factors [16]. The city and township of Beaver Dam, Wisconsin, has approximately 17,000 residents. Residents are predominantly (99.3%) white, and most residents are employed in agriculture or light industry. A private enumeration of Beaver Dam in 1987 found 5925 persons between the ages of 43 and 84 years. Of this group, 4926 consented to an extensive eye examination and interview. Of those participating in the BDES, 1817 were randomly selected and invited to participate in the BDHOS interview, of whom 1676 were eligible. One hundred and ninety-four people (11.6%) refused to participate in the study, and 52 people (3.1 “0 /) could not be contacted. The BDHOS cohort comprises 1430 subjects who were not institutionalized or in an acute care facility, and who agreed to participate in the 1-hr interview between January 15, 1991 and October 31, 1992.

Data Collection Participants underwent an interview, which included a history of chronic medical conditions, an inventory of current prescription and nonprescription medications (subjects were asked to bring their medications to the interview), and health status assessments. For each potential health condition, participants were asked whether a physician had ever told them that they have the condition. If the answer to this question was “yes,” then the participants were asked if the condition had affected them in the past year, to which a participant could respond “yes” or “no.” Interviews were administered orally by trained nurse-interviewers. Blood pressure was measured with a calibrated random-zero sphygmomanometer. Participants were seated for 5 min prior to taking blood pressures, and two pressures were taken. The results are based on the average of the two blood pressure measurements. All participants in the BDHOS were interviewed 3 years earlier on average as part of the BDES. In the BDES, participants were also asked if they had hypertension. Thus, information on duration of hypertension for those reporting hypertension in the BDHOS was reported as hypertension for more or less than 3 years.

HRQOL

Measures

We chose four instruments for the BDHOS to measure overall perceptions of health and health-related quality of life. The SF-36 questionnaire (version 1; Interstudy, Inc., Excelsior, MN) contains 36 questions, the answers to which are scored on 8 scales, including general health perceptions (GH), physical function, general mental health, social function, role limitations due to physical problems (role physical), role limitations due to emotional problems (role emotional), bodily pain, and vitality [17]. These scales are scored from 0 to 100, with 100 representing the best health for each scale. This investigation focuses on the GH scale as a measure of general health status from the SF-36. One question from the GH scale of the SF-36 was also examined separately. The question states: “In general would you say your health is,” and subjects complete the statement with a response chosen from: “Excellent,” “Very Good,” “Good,” “Fair,” or “Poor.” This question will be referred to as EVGFP, and was scored numerically for separate analysis as Poor = 1, Fair = 2, Good = 3, Very Good = 4, Excellent = 5. The Quality of Well Being (QWB) index [18] was also used for health status assessments. This instrument is divided into four domains: Symptom-problem complex, Mobility, Physical Activity, and Social Activity. The results of the four sections are combined

et al.

into a composite score anchored between 1 (representing perfect health) and 0 (representing death). The final measure of health status was a health-state utility assessment using the time trade-off (TTO) technique [19,20], which quantifies a person’s preference for his/her current HRQOL. For this assessment, the respondents are asked to choose between two hypothetical alternatives. The respondents can choose either (1) living the remainder of their life expectancy (age and sex specific) in their current state of health, or (2) living a shorter period of time in excellent health. The time in excellent health is varied until the participants judge the two alternatives to be equally preferred. Results of the TTO are presented as scores ranging from 0 (the person would take a very short time, e.g., 1 day, in excellent health) to 100 (the person would not be willing to have a shorter life expectancy for excellent health), and are expressed as the percentage of current age- and sex-specific life expectancy spent in excellent health that the participant would consider equivalent to living their full life expectancy in their current state of health. As an example, if a person had a 20syear life expectancy, and is indifferent to living 15 years in excellent health or 20 years in his or her current health, this person’s TTO would be 75%, and this score would represent a better self-perceived HRQOL than if he or she had a TTO of 50%. We chose these instruments as four archetypal measures of health status and health-related quality of life. Validity of each measure has been well discussed in the literature [20&22]. The TTO assessment has been widely used [20,23,24], and represents an individual’s valuation of his/her own state of health. This measure represents a tie to the decision-making literature. The EVGFP question is the most widely used single-question measure of general health perceptions [23,25], and is asked in the National Health Interview Survey (NHIS). The QWB index, the first multiattribute health utility index developed, has been used in patients with chronic diseases such as acquired immunodeficiency syndrome (AIDS) and arthritis [26]; this index formed the framework of the health state assessment used in the Oregon Basic Health Services Act 1271. The SF-36 is also frequently used [22], and represents a health profile, or a multidimensional description of current health using summated rating scales. Thus, our four instruments were chosen as leading representatives of four distinct types of measures of general HRQOL.

Statistical

Analysis

Data were analyzed using the SAS System for Windows release 6.08 statistical analysis package (SAS Institute, Inc., Cary, NC), and the Minitab for Microsoft Windows release 9 statistical analysis package (Minitab, Inc., State College, PA). Data on medications used by participants were coded using the Standard Drug Data Base-Plus (licensed from Medi-Span, Inc., Indianapolis, IN). The cross-sectional analysis compared participants who reported being affected by hypertension (self-reported hypertension, or HTN) in the past year to those who reported not being affected (No HTN). Continuous variables were compared using two-tailed Student’s t tests. Models were created for health status measures as functions of age using multiple linear regressions; reported results are age adjusted to the mean age of the BDHOS sample. Sex was entered as a variable into several models for health status scores, and found to have no significant effect, so sex was not controlled for in the final analyses. Age was examined as an independent variable both as a continuous variable, which assumes a linear relationship between age and the dependent variable, and as a categorical variable of age by decade, which does not assume a linear response with age. Since EVGFP is

Health

1241

Status and Hypertension

TABLE 1. Characteristics comes Study participants

of the Beaver

Variable N Age’ (years) Sex Other conditions (count)’ Average number of antihypertensive agents’ Average numher of other medications’ Measured blood pressure (average of 2 measurements)‘

Dam

Health

Out*

HTN”

No HTN”

519 66.6 59.9% female 3.48

911 62.5 58.1% female 2.69

MO250 -

2001.33

0.12

3.23

2.66

142/79 mmHg

E 3 lso-

128/75 mmHg

loo -

,‘Parttctpants reporting being affected hy hypertension in the past year. HTN, Hypertension. ‘Parttctpants repurtmg not being affected by hypertension in the past year. 'I, < 0.05.

50O--

an ordinal variable, we used a polytomous logistic regression [28] as a confirmatory test of significance for age-adjusted differences in the cumulative distribution functions of this variable. Comorbid illness may he a confounding factor in this study. To account for this, we chose prior to analysis to perform a subgroup analysis examining health status measures of participants who have no reported chronic medical conditions compared to those reporting hypertension as their only chronic condition. Results are reported here for this analysis. An alternative way to account for this potential confounding would have been to adjust statistically for the presence of other conditions. Using multivariate regression to adjust for age, number of comorbid illnesses, and number of medications other than antihypertensive medications did not change the conclusions of the analysis. Antihypertensive medication (AHM) effects were examined using multivariate linear regressions to determine the change in health status per medication, adjusted for age, the presence of hypertension, and the presence of other reported chronic medical conditions. Approximately 320 chronic medical conditions were reported by participants in the BDHOS; the adjustment for comorhid illness was performed using a count of participants’ self-reported conditions. To examine the effect of specific classes of antihypertensive medications on health status measures, we performed a subgroup analysis on hypertensives taking only one antihypertensive medication. Analysis of covariance was used to test for significant differences of age- and comorbidity-adjusted health status scores associated with different classes of AHMs.

1

0

Status

Measures

in Hypertension

The characteristics of the participants in the BDHOS can be seen in Table 1, which shows the demographic characteristics for the entire BDHOS sample of 1430 participants. The age range for the entire sample was 45 to 89 years, and averaged 64.0 years. Six hundred and nineteen participants (43.3%) reported ever being told by a physician that they had hypertension. Self-reported hypertensives affected in the past year comprised 36.3% (N = 5 19) of the sample; this group tended to be older, to have more chronic conditions, and to take more medications not associated with hypertension than the No HTN group. HTN group participants were taking an average of

1

I

I

I

I

2

3

4

5

d Anbhypettensd

h&diMions

FIGURE 1. Histogram of the number of antihypertensive medications taken by participants reporting hypertension.

1.33 antihypertensive medications (range O-5) (see Fig. 1). Cuff blood pressure, measured at the beginning of the interview, tended to be higher in the HTNs. Of those reporting hypertension, 92.7% had a duration of hypertension of at least 3 years. Eighty-three percent of the HTN group who were taking antihypertensive medications at the time of the BDHOS also reported taking medication 3 years earlier. The general health status measures TTO, GH, and EVGFP, adjusted to the mean age of the cohort, were all significantly lower (p < 0.001 for each) for the HTNs versus the No HTNs (Table 2). For these three measures, the mean difference between the HTNs and No HTNs comprised 4.8% of the total range of the scale for TTO, 6.1% for GH, and 7.8% for EVGFP. The finding of a tendency

TABLE samnle

2. Mean

Health status measure’ TTO

Health

I

Number

RESULTS

General

I

GH EVGFP

QWB

age-adjusted

mean

health

HTNh (95%

83.1 (81.1-85.1) 70.8 (69.3-72.3) 3.34 (3.27-3.42) 0.719 (0.710-0.728)

CI)

status scores for general No HTN’ mean (95% CI) 87.9 (86.4-89.4) 76.9 (75.7-78.0) 3.65 (3.60-3.71) 0.729 (0.723-0.736)

“TTO, Time Trade-Off (scored as percentage of life expectancy, maximum score is 100); GH, General Health Perceptton (maximum IS 100); EVGFP, Excellent, Very Good, Good, Fan, Poor (maximum is 5); QWB, Qualtty uf Well-Being Index (maximum is 1.0). “Participants reporting being affected by hypertension m the past year. ‘Partictpants reportmg not being affected hy hypertension m the past year.

1

1242

W. F. Lawrence

et al.

FIGURE 2. Relation of age to health status measures for the HTN and No HTN groups (closed squares, HTN group; open squares, No HTN group), with predicted mea* sures by linear regression (solid line, HTN group;

50

so

70

so

90

60

50

M

80

90

QWB.



Age (years)

toward lower responses on the EVGFP question in the HTN group compared to the No HTN group was confirmed by polytomous logistic regression. The difference between the QWB scores for the HTNs and the No HTNs was only 1.0% of the total scale (Table 2), and did not quite achieve statistical significance (p = 0.064). Results did not change appreciably when performed comparing the 619 people reporting ever being told they had hypertension with the 811 people never having this diagnosis. Figure 2 shows a linear regression of health status on age for the HTN and No HTN groups. In all four measures of general health status, there was a significant decrease in health status associated with age. The GH and EVGFP showed significant interactions (GH, p = 0.004; EVGFP, t, = 0.038) between age and hypertension on predicted scores, such that No HTNs compared with HTNs had larger decreases in scores associated with advancing age. Nonlinear modeling of age did not appreciably increase variance accounted for by age. We also examined a subgroup of participants whose only selfreported chronic condition was hypertension (HTN Only), and compared this group to those who reported no chronic conditions (No Conditions). The demographics of the hypertension-only suh-

TABLE 3. Characteristics comes Study participants:

of the Beaver Dam Subgroup analysis

Variable

HTN

Only”

N Age’ (years) Sex Average number of anti hypertensive agents’ Average number of other medications Measured hlood pressure (average of 2 measurements)J

45 60.6 5 1.1% female

Health

Out.

No Conditions’ 167 57.5 52.1% female

Hypertension

TABLE analysis

TTO GH EVGFP

1.09

1.24

QWB

128/78 mmHg

blood pressure only.

4. Mean

Health status measure’

0.03

134/80 mmHg

and SF-36 Scores

In the general BDHOS sample, th ree other scales of the SF-36 in addition to the GH showed a significant decrease in the HTN group compared to the No HTN group. These scales were the physical

1.20

~‘Participants who reported hyperrension as their only chronic medical con&non. “Participants who reported no chronic medical conditions. ‘p < 0.05. ,‘p < 0.05 for systolic

group can he seen in Table 3. Again, hypertensives tended to he older. There was no significant difference in the proportion of the sexes between groups or in the average number of nonantihypertensive medications taken. On average, the HTN Only participants were treated with slightly more than one antihypertensive medication. Only the systolic blood pressure was signihcantly increased in the HTN Only group over the No Conditions group; the diastolic blood pressure was not significantly different. Comparisons in the subgroups HTN Only versus No Conditions showed no significant differences in the QWB scores; the TTO (p = 0.002), GH (p = 0.009), and EVGFP (p = 0.003) scores did show significant decrements (Table 4). Th e magnitudes of the decrements in TTO (4.5% of the total range of the scale), GH (5.10/o), and EVGFP (9.3%) scores were similar to the larger HTN group, hut with baseline health status scores being higher.

age-adjusted

health

HTN Only” mean (95% CI) 93.0 (90.4-95.5) 81.4 (78.0-84.8) 3.72 (3.50-3.94) 0.804 (0.780-0.829)

status scores:

Subgroup

No Conditions’ mean (95% CI) 97.5 (96.0-99.0) 86.5 (84.4-88.5) 4.09 (3.96-4.23) 0.806 (0.792-0.821)

“TTO, Time Trade-Off (scored as percenrage of life expectancy, maximum score 100); GH, General Health Perception (maximum 100); EVGFP, Excellent, Very Good, Good, Fair, Poor (maximum 5); QWB, Qualrty of WellBeing Index (maximum 1.0). “Participants reporting hyperrension as their only chronic medical con&tion. ‘Participanrs who reporred no chronic medical condlriuns.

Health TABLE cation

1243

Status and Hypertension 5. Change

in health

status scores per additional Mean

Health

status

change

mediin score’

-2.7

-l-TO

(-5.1, -0.29) -4.4 (-6.1, -2.8)

OH

-0.19 (-0.27, -0.10) -0.013 (-0.022, -0.004)

QWB

AHM

class

Definition

N(%)

(95% CI)

measure

EVGFP

TABLE 6. Antihypertensive medication classes used by hypertensive participants on one antihypertensive medication

BETA CA ACEI CENT COMBO

DIUR

P-Adrenergic blocking agents Calcium channel-blocking agents Angiotensin-converting enzyme inhibitors Centrally acting agents Combination medications of two or more pharmacologic classes of AHMs Diuretics

54 (16.2) 57 (17.1) 61 (18.3) 12 (3.6) 37 (11.1)

109 (32.7)

t’Scores are given as a mean change in baseline score per additional medicanon, adjusted for age and number of comorhid illnesses.

function scale (mean age-adjusted scores, HTN vs. No HTN: 77.6 vs. 82.8, p < O.OOl), the mental health scale (80.2 vs. 82.5, p = 0.004), and the vitality scale (62.7 vs. 66.5, p < 0.001). The other f&r scales of the SF-36 showed no significant change between HTNs and No HTNs. In the hypertension-only subgroup analysis, the only statistically significant change in the SF-36 scales was the GH scale.

Antihypertensive Measures

Medications

and Health

Status

In the BDHOS sample, HTNs reported taking up to five antihypertensive medications (Fig. 1). Significant decrements in all four general health status measures were associated with increasing number of AHMs. Table 5 shows the average decrease in health status per and addition;ll AHM, adjusted for age, the presence of hypertension, number of other reported illnesses (TTO, p = 0.028; GH, e < 0.001; EVGFP, p < 0.001; QWB, e = 0.006). We compared the change in health status associated with AHMs to the change in health status associated with age (average change in the BDHOS cohort); the mean change in health status per additional medication was equivalent to a change in health status associated with an increase in age of 7.5 years for TTO, 15.2 years for GH, 10.6 years for EVGFP, and 4.6 years for QWB. We studied the 333 members (64.2%)) of the HTN group who were taking a single antihypertensive medication to examine the

effects of specific pharmacologic classes of antihypertensive medications on the health status measures. Data from the two participants taking a-adrenergic blocking agents and the one participant taking reserpine are not reported owing to low numbers of participants using these classes of medications. The types of AHMs examined and the number of participants taking these medications can he seen in Table 6. Figure 3 shows the mean health status measures for people taking different classes of antihypertensives, adjusted for age and number of comorbid illnesses. Analysis of covariance showed no significant differences in health status measures for the six classes of medication taken by this sample (TTO, 1, = 0.62; GH, p = 0.87; EVGFP, 0 = 0.85; QWB, p = 1 .O). For the two largest groups, those taking diuretics and those taking angiotensin-converting enzyme (ACE) inhibitors, our analysis had 80% power to detect a lo-point difference in TTO, a 7-point difference in GH, a 0.36-point difference in EVGFP, and a 0.04-point difference in QWB. DISCUSSION We studied the relation between self-reported health and the diagnosis and pharmacologic treatment of hypertension in a crosssectional analysis of a community population. Our population was randomly sampled from adults over age 45 years and contributes data from not only a population-based study, but a population in which the diagnosis and treatment of hypertension is likely to be in “steady state” for the majority of participants. This cohort was not assembled from a clinic visit inception point, nor are the participants

100

FIGURE 3. Mean health status measures (and 95% confidence intervals) by AHM class for participants on one AHM, adjusted for age and number of comorbid illnesses. (A) TTO results; (B) GH rev sults; (C) EVGFP results; (D) QWB results. Analyses of covariance adjusting for age and comorbidities were not significant for any of the health status measures (TTO, p = 0.62; GH, p = 0.87; EVGFP, p = 0.85; QWB, p = 1.0).

B

1244 mainly newly diagnosed or newly treated. The large majority (93%) of the study population had reported the diagnosis of hypertension 3 years previously, and the majority also reported taking some medication for hypertension at that time. Thus, we anticipate that for the most part the participants with hypertension have stabilized into long-term patterns of management and that our data provide a crosssectional view of their health status after the participants have accommodated to their condition. We examined the effects of being diagnosed with and treated for hypertension, using instruments representing three distinct types of instruments used for measurement of health status (the SF-36, QWB, and the TTO) as well as one widely reported single-question item, the EVGFP. Our analysis showed a lower age-adjusted health status associated with self-reported hypertension, with this result evident on three of our four measurement instruments. Prior studies that were not community based have shown an increase in illness behavior associated with the diagnosis of hypertension, suggestive of a lower health status associated with this condition [2,3]. The Medical Outcomes Study [14] f ound lower genera1 health perceptions scores in hypertensive patients compared to those patients without chronic conditions. The Edgecombe County High Blood Pressure Control Program investigators found a similar effect for males but not for females [29]. We found in our community population that HTNs tended to be older and to report more chronic health conditions than No HTNs, thus differences in health smtus may not be attributable directly to hypertension. To minimize this potential confounding, we also studied the subgroup of people reporting only hypertension and compared them to the population subgroup reporting no chronic conditions. The results still showed a health status decrement of the same absolute magnitude for the HTN Only subgroup compared to the No Conditions subgroup, albeit the decrement was from a higher baseline health status for those with no conditions. We believe that this analysis considerably strengthens the evidence for more than a casual relation between the hypertension diagnosis and difference in health status. One scale, the QWB, did not detect a statistically significant decrement in health status associated with hypertension. In the subgroup analysis (and thus without effect of comorbid illness), the power to detect a small difference on the order of 0.02 points on the QWB scale was 0.70 for this measure. We interpret the failure to detect the change on this measure as a problem of both modest power and insensitivity of the QWB to small changes in basically healthy persons. It is interesting to note that the EVGFP, a rather gross measure of self-perceived health status, did show a decrement in health status in both the whole-sample and subgroup analyses. Age was a highly significant predictor of health status for all health status measures, although sex was not in this analysis. Age adjustment is thus important in comparisons of health status between groups. We also examined the association between antihypertensive medication use and health status in this population. Controlling for age and the number of comorbid illnesses, we found in cross-sectional analyses an inverse relationship hetwcen the number of antihypertensive medications and health status measures. The average decrease in general health status scores per addItional medication in the regression analysis was approximately equivalent to the decrease in scores associated with an Increase in age of 5 to 15 years. The evidence of a dose-response relation between the numher of medications and health status suggests that the therapy is at least in part responsible for the decrement, although we cannot exclude severity

W. F. Lawrence

et al.

of hypertension as a possible causal component. Severity of hypertension could not be measured in our study of community-dwelling persons as we do not have premeditation blood pressure measurements. Interestingly, we found that the number and not the pharmacologic class of antihypertensive medications was associated with changes in health status. Combination medications consisting of two or more classes of AHMs in a single pill led to no worse decrements in health status measures than did a single medication. This suggests that a combination medication may have a better qualityof-life effect than the same two medications taken separately. However, our power for detecting differences among the pharmacologic classes was modest. Prior observational studies of the effects of AHMs on health status are conflicting. Siscovick et al. [30], found patients taking AHMs were more likely to be impaired on a measure of mental health; the Edgecombe County study [29] and the Batelle Center study [31] did not find a significant relation between treatment for hypertension and general health status measures. Prospective cohort studies have examined the effect of AHMs on health status [5-131. Several studies have no placebo or untreated arm to compare with treatment arms [5-lo], and most of these studies have time horizons on the order of 1 year or less [5-l 11. In these studies participant5 were typically assigned to take one specific medication, as opposed to clinical practice where a patient may have medications changed if the first prescribed one is not well tolerated. Also, participants had the ability to withdraw from the study completely and not be included in the final report of the data; in practice, although medications can be switched, patients may not have the option of withdrawing from therapy completely. Under these conditions, several prospective studies have shown no change or an improvement in measures of health status with initiation of therapy. Goldstein et al. [lo] found no significant changes between measures of activities of daily living between people taking one of four different AHMs in addition to hydrochlorothiazide, but one of the medication arms had a 40% dropout rate and the overall dropout rate of the study was 23%. The Treatment of Mild Hypertension Study (TOMHS) followed 902 patients, treating with one of five AHMs or placebo, all in addition to intensive nonpharmacologic therapy [12]. With only a 6% dropout rate at 1 year, the TOMHS found an improvement in several health status dimensions with medical therapy and with placebo, and two medications performed better than placebo. Four-year results were similar [13], but at that time 21% were taking AHMs in addition to or instead of their study medication. A prospective trial such as the TOMHS lnay be affected by recruitment of particularly motivated participants, and its results are confounded by the concomitant intensive nonpharmacologic therapy. We believe our epidemiologic study design and results give a better picture of the long-term, steady state health status effects of antihypertensive medications in a community population, although there are limits to our data as well. In particular, we do not have a prior measure of severity of hypertension or doynge information for those persons reporting taking AHMs in theu medication inventory. Our power for detecting interclass differences among AHMs is limited. Within these limitations, our data raise an important LSSLK. Our results indicate that people having the diagnosis of hypertension, and those treated pharmacologically for hypertension, lnay be experiencing a present reduction in health-related quality of life in order to achieve future longevity benefits of hypertension reduction. Pharmacologic therapy of hypertension reduces the risk of stroke, and

Health

Status

1245

and Hypertension

may reduce the risk of coronary heart disease [32]. The longevity benefit of reducing diastolic blood pressure to 88 mmHg in 35-yearold hypertensives has been estimated to be 0.4 to 1.1 additional years of life [33], and may be expected to be somewhat less for people in the age range of our study. Our results imply that a person treated pharmacologically for hypertension has a health status similar to that of an otherwise similar person 5 to 15 years older. Since it has been shown that physicians can be poor judges of quality of life in hypertensives [34], perhaps the inherent trade-offs between the health costs and benefits of AHMs should be presented prospectively to patients to allow them to participate better in the treatment decision. For some patients, the longevity benefit of antihypertensive

therapy

may not

outweigh

the adverse

effects

on quality

of

13. 14.

15.

16. 17.

18.

life. 19. The authors acknowledge the assistance of Ann Varda and Lisa Goetzfor computer datamanagement; Norma Darn, R.N. and Kathy Peterson, R.N., fordatacollection; and Helen Sokfner for participant contact and follow-up. Erik J. Dasbach, Ph.D., assisted in the design of the Beaver Dam Health Outcomes Study. This project is supported by Grant No. HS 06491 from the Agency for Health Care Policy and Resepch, and National Eye htitute Grant 1OUO6594.

20. 21. 22.

References 1. Hoerr SO. Hoerr’s 2.

3.

4.

5. 6.

7.

8.

9.

10.

11.

12

law. Am J Surg 1962; 103: 41 I. Haynes RB, Sackett DL, Taylor W, Gibson ES, Johnson AL. Increased absenteeism from work after detection and lahelling of hypertensive patients. N Engl J Med 1978; 299: 741-744. Wagner EH, Strogatz DS. Hypertenston labeling and well-being: Alternative explanations in cross-sectional data. J Chron Dis 1984; 12: 943947. The Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure. The fifth report of the Joinr National Committee on detection, evaluation, and treatment of high blood presSure (JNC V). Arch Intern Med 1993; 152: 154-183. Croog SH, Levine S, Testa MA, et al. The effects of antihypertensive therapy on rhe quality of life. N Engl J Med 1986; 314: 1357-1364. Croog SH, Kong W, Levine S, Weir MR, et al. Hypertensive black men and women: Quality of life and effects of antihyperrensive medicatmns. Arch Intern Med 1990; 150: 1733-1741. Applegate WB, Phillips HL, Schnaper H, er al. A randomized controlled trial of rhree antihypertensive agents on blood pressure control and quality of life in older women. Arch Intern Med 1991; 151: 1817-1823. Testa MA, Hollenherg NK, Anderson RB, Williams GH. Assessment of quality of life by patient and spouse during antihyperrensive therapy with atenolol and nifedipine gastrointestinal therapeutic system. Am J Hypertens 1991; 4: 363-373. Testa MA, Anderson RB, Nackley JF, et al. Quality of life and antihypertcnsive therapy in men: A comparison of captopril with enalapril. N Engl J Med 1993; 328: 907-913. Goldstem G, Materson BJ, Cushman WC, Reda DJ, Freis ED, Ramirez EA, et al. Treatment of hypertension m the elderly. II. Cognitive and hehavloral function: Results of a Veterans Affairs Cooperarive Srudy. Hypertension 1990; 15: 1413-1423. Wasserrhell-Smoller S, Blaufox D, Oherman A, et al. Effect of annhypertensives on sexual function and quality of life: The TAIM Study. Ann Intern Med 1991; 114: 613-620. The Treatment of Mild Hypertension Research Group. The Treatment of Mild Hypertensron Study: A randomized, placeho conrrolled trial of

23. 24.

25.

26. 27.

28. 29.

30.

31.

32.

33.

34.

a nutritional-hygienic regimen along with various drug monotherapies. Arch Intern Med 1991; 151: 1413-1423. Neaton JD, Grimm RH, Prineas RJ, et al. Treatment of Mild Hypertension Study: Final results. JAMA 1993; 270: 713-724. Stewart AL, Greenfield S, Hays RD, et al. Functional status and wellbeing of patients with chronic condinons: Resulrs from the Medical Outcomes Study. JAMA 1989; 262: 907-913. Fryhack DC, Dashach EJ, Klein R, et al. The Beaver Dam Health Outcomes Study: Initial catalog of health state quality factors. Med Decis Making 1993; 13: 89-102. Klein R, Klein BEK, Linton KLP, deMets DL. The Beaver Dam Eye Study: Visual acuity. Ophthalmology 1991; 98: 1310-1315. Ware JE, Sherhourne CD. The MOS 36-item short form health survey (SF-36). 1. Conceptual framework and item selection. Med Care 1992; 30: 473-483. Kaplan RM, Anderson JP. A general policy model: Update and applications. Health Services Res 1988; 23: 203-235. Torrance GW, Thomas WH, Sackett DL. A utdity maximizarion model for evaluation of health care programs. Health Services Res 1972; 7: 118-133. Froherg DC, Kane RL. Methodology for measurement of health-state preferences. II. Scaling methods. J Clin Epidemiol 1989; 42: 459-471. Patnck DL, Erickson P. Health Status and Health Policy. New York: Oxford University Press; 1993. Ware JE. SF-36 Health Survey: Manual and Interpretation Guide. Boston: The Health Instirute, New England MedIcal Center; 1993. Sackett DL, Torrance GW. The utility of different health states as perceived hy the general public. J Chron Dis 1978; 31: 697-704. Nease RF, Kneeiand T, O’Connor CT, et al. Variation in pacienr utilities for ourcomes of the management of chronic stable angina. JAMA 1995; 273: 1185-1190. Ware JE, Nelson EC, Sherhourne CD, Stewart AL. Prelimmary tests of a 6-item general health survey: A patient application. In: Stewart AL, Ware JE, Eds. Measuring Functioning and Well-Being: The Medical Outcomes Study Approach. Durham, North Carolina: L>uke Universxy Press; 1992. Kaplan RM, Anderson JP. A general health p&y model: Update and applications. Health Serv Res 1988; 23: 203-235. U.S. Congress, Office ofTechnology Assessment. Evaluation of rhe Oregon MedIcaid Proposal. OTA-H-531. Washington, D.C.: U.S. Government Printing Office; 1992. Hosmer DW, Lemeshow S. Applied Logistic Regresston. New York: John Wiley & Sons; 1989. Wagner EH, James SA, Beresford SAA, et al. The Edgecomhe County High Blood Pressure Control Program. 1. Correlates of cmcomrolled hypertensIon at baseline. Am J Public Health 1984; 74: 737-742. Siscovick DS, Strogatz DS, Fletcher SW, Leake B, Brook RH. The association between hypertension treatment, control, and functional status. J Gen Intern Med 1987; 2: 406-410. Barr&e Medlcal Technology Assessment and Policy Research Center. Health Utihty and Health Status Outcomes of Antihypertensive Treatment. Washington, DC.: Battelle; 1992. Collins R, Peto R, MacMahon S, etal. Blood pressure, stroke, and corenary heart disease. 2. Short-term reductions in blood pressure: Overview of randomised drug trials in their epidemiological context. Lancet 1990; 335: 827-838. Tsevat J, Weinstein MC, Williams LW, Tosteson ANA, Goldman L. Expected gains in life expectancy form various coronary heart disease risk factor modifications. Circulation 1991; 83: 1194-1201. Jachuck SJ, Brierly H, Jachuck S, W’II1 cox PM. The effect of hypotcnsive drugs on the quahty of life. J R Co11 Gen Pratt 1982; 32: 103-105.