Prevalence, Clinical Correlates, and Treatment of Hypertension in Elderly Nursing Home Residents FREE

HYPERTENSION IS one of the most frequent conditions of older people^1- 3 and is an important risk factor for cardiovascular and cerebrovascular disease.^4- 7 Recent data support the notion that the absolute benefits of pharmacological treatment are even more pronounced in the elderly, both in men and in women.^8- 9 Randomized controlled trials have shown that diuretics, and possibly β-blockers and some calcium channel blockers, reduce morbidity and improve survival of older, but otherwise healthy, hypertensive patients.^10- 13 Accordingly, the Joint National Committee has repeatedly recommended these drugs as the preferred choice.^14- 15

Regardless, several calcium channel blockers and angiotensin-converting enzyme (ACE) inhibitors have become widely prescribed, especially among elderly patients.^16- 19 ACE inhibitors are effective in reducing blood pressure, but their effect on morbidity or mortality among hypertensive patients is still unknown.⁴ In the past 2 years, observational studies have questioned the widespread use of calcium channel blockers because of increased morbidity and mortality in selected patients with coronary heart disease,^20- 21 but also in adult²² and elderly²³ hypertensive patients. The Systolic Hypertension in Europe Trial is the only trial that has shown a clear benefit conferred by a calcium channel blocker, but the study enrolled exclusively elderly patients with isolated systolic hypertension.¹³ Yet, no other data from controlled trials are available in this diverse population.

People older than 65 years, despite representing 15% of the US population and using one third of all medications, remain underrepresented in randomized clinical trials.^24- 25 Recently this limitation has been recognized, and all of the studies currently in progress concentrate on older hypertensive patients, some with no upper limit on age.²⁶ However, randomized trials in the elderly can pose serious challenges, and even when targeted to these patients, only the relatively healthier population is included, producing results of limited generalizability.¹¹ In general, the ideal patient for a trial little resembles the clinically complex, frail older patient for whom most of the drugs are prescribed. Such difference is amplified for patients who reside in nursing homes; typically they are the oldest old, are often cognitively and physically impaired, usually have multiple comorbid conditions, and take numerous medications.²⁷ Little is known about hypertensive patients who reside in long-term care facilities, despite the increasing importance of this health sector.^28- 29

The objective of our study was to investigate to what extent antihypertensive drug prescribing patterns conform to consensus guidelines, to evaluate the impact of age and comorbidity, and to identify the independent predictors of receiving pharmacological treatment for hypertension among elderly patients living in nursing homes.

Since 1991, all US nursing homes have been required to use a uniform, comprehensive assessment tool known as the Resident Assessment Instrument (RAI) with its Minimum Data Set (MDS),³⁰ establishing a national, long-term care database. The present study used data from the Health Care Financing Administration's Multi-State, Case-Mix and Quality Demonstration Project involving all 1492 Medicare- and Medicaid-certified nursing homes in Kansas, Maine, Mississippi, New York, and South Dakota. We have merged 3 databases: (1) a computerized, longitudinal MDS data set on nearly 300000 patients residing in, or admitted to, a nursing home in any of the 5 states during the period from January 1, 1992, through December 31, 1994; (2) a longitudinal file containing data on all drugs received by each patient; and (3) the Medicare enrollment files and Medicare Provider Analysis and Review database, which contains information on all persons covered by Medicare Part A. The resulting database (Systematic Assessment of Geriatric Drug Use via Epidemiology [SAGE]) has been described in detail elsewhere,^31- 33 and it is briefly summarized herein.

The MDS includes sociodemographic information, numerous clinical items ranging from the degree of physical dependence to cognitive functioning, and all active clinical diagnoses.^30- 31 The MDS also includes an extensive array of signs, symptoms, syndromes, treatments, and indicators that describe each resident's behavior and mood.^30- 31 A variety of multi-item, summary scales are embedded in the MDS to examine the performance on Activities of Daily Living, cognition, mood status, behavioral problems, social engagement, communication, mobility, and urinary continence.^32- 33

In addition to MDS data, staff recorded up to 18 different drugs received by each resident in the 7 days preceding the assessment. Information on the resident's drug therapy included brand or generic name, dosage, route and frequency of administration, and whether it was given on a standing or as-needed order.³² Drugs were coded according to the National Drug Code system. We used the Master Drug Data Base (MediSpan Inc, Indianapolis, Ind) to translate National Drug Codes into therapeutic classes and subclasses.³³

From an initial population of 296379 unique individuals (January 1, 1992, through December 31, 1994), we excluded patients younger than 65 years (n=25506) and those in a comatose state (n=725). We then identified 80206 patients based on the presence of a diagnosis of hypertension at their initial MDS assessment. Staff physicians coded a diagnosis of hypertension by means of information obtained from the medical record, including the physical examination of the resident, medication and other treatment orders, and hospital discharge documentation (if any).³⁴ In interrater trials, the κ coefficient for the diagnosis of hypertension was found to be excellent (0.80).³⁵ In addition, a diagnosis of hypertension on the MDS record could be confirmed through the Medicare hospital discharge claims in more than 83% of cases with a hospitalization shortly preceding the MDS.³³

We conducted a cross-sectional study based on the initial MDS assessment of the 80206 hypertensive patients. We evaluated age differences by stratifying the sample into 3 age categories: 65 to 74 years (n=13285), 75 to 84 years (n=31707), and 85 years or older (n=35214). Performance in physical function was expressed with a 5-item, 6-level Activities of Daily Living Scale, and cognition was measured with a 6-item, 7-level Cognitive Performance Scale, both of which have been previously validated.^36- 37

Antihypertensive medications were classified by the following MediSpan classification codes: 33.10, β-blockers, nonselective; 33.20, β-blockers, selective; 34.00, calcium channel blockers (subsequently distinguished as dihydropyridines, diphenylalkylamine, or benzothiazepine); 36.10, ACE inhibitors; 36.20, adrenolytic agents (ie, methyldopa, clonidine, reserpine, guanabenz); 36.30, α-blockers; 36.40, hydralazine; 37.20, loop diuretics; 37.50, potassium-sparing diuretics; and 37.60, thiazides. Nitrates (32.10) were also computed.

To evaluate age trends, we performed Mantel-Haenszel χ² tests.³⁸ We evaluated predictors of pharmacological treatment of hypertension by developing a multiple logistic regression model. Based on crude analyses of the association between pharmacological treatment and various independent variables, including sociodemographic variables, measures of physical and cognitive functioning, and indexes of comorbid conditions, we identified potential predictors for the model. Before constructing the model, we evaluated (and ruled out) multicollinearity. From the final model, we derived odds ratios and corresponding 95% confidence intervals. All analyses were performed with SAS software (version 6.12, SAS Institute, Cary, NC).

Overall, nearly 32% of the nursing home residents (mean age, 82.7±7.8 years; range, 65-115 years) had a diagnosis of hypertension. The prevalence of hypertension was higher among women (33%) than men (27%) and among African Americans (43%) and other minorities (34%) than among whites (30%). As age increased, the prevalence of hypertension declined slightly (35% in the 65- to 74-year-old group vs 33% and 30% in the 75- to 84-year-old and ≥85-year-old groups, respectively), which is accounted for by the decline among men and African Americans. Prevalence of hypertension among women and white patients did not differ by age.

Table 1 summarizes the principal sociodemographic characteristics and several functional variables of hypertensive patients. The proportion of women and white patients increased with age. Overall, two thirds of patients were admitted from an acute care hospital, and 60% of them stayed in the nursing home for less than 6 months. As many as 30% of patients had a body mass index less than 20 kg/m². While 16% of patients overall required no, or minimal, assistance performing basic daily activities, 40% showed severe impairment. No major age-related differences were observed. On the other hand, more than 40% of patients (39%-48% by age group) had only minimally impaired cognitive function. Overall, 46% of patients had urinary incontinence and one third had experienced at least 1 fall in the preceding 6 months.

Table 2 shows that two thirds of patients had more than 3 comorbid medical conditions in addition to hypertension, and 25% had 6 or more. Increased age was associated with a significantly greater number of comorbid conditions. Twenty-six percent, 22%, and 29% of patients had a concomitant diagnosis of coronary heart disease (CHD), congestive heart failure (CHF), or cerebrovascular disease, respectively. While the proportion of hypertensive patients with coexisting CHD and CHF increased by 50% or more in the group aged 85 years or older as compared with the 65- to 74-year-old group, a significant decline in the proportion of patients affected by cerebrovascular conditions was evident. There was a steady increase in the proportion of patients diagnosed as having dementia (23% in the 65- to 74-year-old group as compared with 35% in the ≥85-year-old group; P<.001). Clinical conditions known to decrease survival, such as diabetes mellitus, chronic obstructive pulmonary disease, and renal failure, were less frequent in the older age strata. Across age groups, 8% of hypertensive patients had a diagnosis of cancer.

Nearly 60% of patients received more than 5 medications daily, and as many as 23% patients aged 65 to 74 years received 11 or more. In contrast to the trend observed with most clinical diagnoses, increased age was associated with reduced drug use. Table 3 summarizes the overall use of antihypertensive agents. The proportion of hypertensive patients receiving at least 1 antihypertensive medication (70%) was similar in each age category. Of the patients receiving antihypertensive medications, 54% received a single agent, while less than 20% received a combination of 3 or more antihypertensive agents.

Calcium channel blockers were the most commonly used therapeutic agents (26% of patients overall), followed by diuretics (25%) and ACE inhibitors (22%); β-blockers were used by less than 10% of patients. The use of adrenolytic agents, α-blockers, and vasodilators such as hydralazine was relatively infrequent. Significant, and opposite in some instances, age-related trends were observed for virtually all antihypertensive agents. The use of antiadrenergic agents showed a decline with increasing age; only 7% of patients aged 85 years or older used β-blockers as compared with 11% in the younger group (P<.001). The proportion of patients treated with calcium channel blockers decreased from 30% in the 65-to 74-year-old group to 27% and 22% in the groups aged 75 to 84 years and 85 years or older, respectively (P<.001 for trend). This difference is explained by a parallel decrease in the use of the dihydropyridine calcium channel blockers. In contrast, the overall use of diuretics increased significantly among older hypertensive patients. In patients aged 85 years or older, diuretics were used more frequently than any other agent (29% of patients vs 22% of patients receiving a calcium channel blocker). Minor age-related differences were observed in the use of ACE inhibitors.

Calcium channel blockers were more likely to be prescribed as single agents than were diuretics, ACE inhibitors, and β-blockers. Among patients using a single antihypertensive agent (n=30383), calcium channel blockers were the preferred option (35% of patients), followed by diuretics (29%) and ACE inhibitors (26%). Among patients receiving 2 antihypertensive drugs (n=16859), diuretics were used more frequently than any other class (34% of patients compared with 29% and 26% receiving an ACE inhibitor or a calcium channel blocker, respectively).

Figure 1 illustrates the proportion of patients treated with antihypertensive medications as a function of the presence of relevant cardiovascular comorbidities. As compared with patients with hypertension alone (n=17297), the use of diuretics and ACE inhibitors increased among hypertensive patients with CHD (n=20887), while it doubled in those with a concomitant CHF diagnosis (from 21% to 44%, and from 18% to 35% of patients in the case of diuretics and ACE inhibitors, respectively). In contrast, the proportion of patients treated with a calcium channel blocker was slightly decreased only if concomitant with a CHF diagnosis (26% vs 23%; P<.005). Similar changes were observed for the use of β-blockers. As a result, calcium channel blockers remained the most commonly used antihypertensive agents among patients with hypertension alone or in association with CHD, while the use of diuretics exceeded that of any other agent when CHF coexisted with hypertension.

The pattern of antihypertensive drug use varied consistently with age whether hypertension was the only diagnosis or was associated with CHD. There was a sustained, age-related reduction in the proportion of patients receiving any of the medications, especially β-blockers (reduced by more than 40%).The only exception was diuretics, which exhibited a sharp increase (on average a 44% increase when the 2 extreme age groups were compared). Similar changes were observed when hypertension and CHF coexisted, although to a lesser extent. Diuretic use was prevalent in any instance (42% of patients overall) and was only slightly higher among patients aged 85 years or older.

The independent predictors of receiving at least 1 antihypertensive medication are listed in Table 4. The relative odds of receiving therapy were significantly decreased for older subjects and men. The presence of a concurrent cardiovascular condition and already being a recipient of a greater number of medications were both independently associated with an increased likelihood of receiving antihypertensive therapy. On the other hand, severe physical and cognitive impairment were inversely related to antihypertensive pharmacological treatment.

Our study documents that hypertension is a common clinical condition among very old residents of long-term care facilities, especially among women and African Americans and other minority groups. The overall picture is that of frail patients with complex medical conditions who have a burden of cardiovascular and neurologic comorbid conditions. Polypharmacy is a distinguishing feature of these patients; 70% receive at least 1 antihypertensive drug, most commonly a calcium channel blocker, followed by diuretics and then ACE inhibitors.

Numerous large, placebo-controlled, randomized clinical trials have investigated the benefit conferred by treatment of combined systolic-diastolic or isolated systolic hypertension in the elderly. In the European Working Party on High Blood Pressure in the Elderly trial,⁴⁰ patients treated with a thiazide diuretic experienced a statistically significant reduction (approximately 39%) in all cumulative cardiovascular events relative to patients taking placebo. The Swedish Trial in Old Patients with Hypertension⁴¹ was interrupted prematurely because, after an average follow-up of 2 years, patients taking a thiazide diuretic and/or a β-blocker experienced a 40% reduction in cumulative cardiovascular events, 47% reduction in all strokes (fatal and nonfatal), and a 43% reduction in total mortality relative to placebo. About two thirds of patients were receiving both medications, and results for the individual drug treatment arms were not reported, although diuretic appeared to be significantly more effective in reducing blood pressure. In the Medical Research Council trial,⁴² patients receiving active treatment took either a thiazide diuretic or atenolol. After an average follow-up of 5.8 years, in the diuretic arm there was a 31%, 44%, and 35% reduction in the incidence of stroke, coronary events, and all cumulative cardiovascular events, respectively, whereas the β-blocker group showed no difference relative to placebo. Two trials, the Systolic Hypertension in the Elderly Program and the Systolic Hypertension in Europe trial, have focused on patients with isolated systolic hypertension. The Systolic Hypertension in the Elderly Program¹² demonstrated that the incidence of all cumulative cardiovascular events was reduced by more than 40% with a thiazide diuretic treatment, and the incidence of the other prespecified end points was also substantially lower. The Systolic Hypertension in Europe trial,¹³ the only trial to investigate antihypertensive medications other than diuretics or β-blockers, has shown that the dihydropyridine calcium channel blocker nitrendipine (with the possible addition of an ACE inhibitor and/or a diuretic) decreased overall stroke rate by 42% and the incidence of all cardiac events by approximately 30%.

Thus, randomized clinical trials have demonstrated that only the use of diuretics, and possibly of β-blockers and some calcium channel blockers, can reduce the incidence of CHF, as well as stroke, coronary disease, and overall cardiovascular mortality.^8- 13 While newer agents (different calcium channel blockers and ACE inhibitors) are effective in reducing blood pressure and ameliorating some surrogate end points, and while they may also be better tolerated,^43- 44 there is no conclusive evidence of an effect on morbidity and mortality.^45- 46 According to the evidence available, the Joint National Committee and the National High Blood Pressure Education Program Working Group have continued to recommend diuretics and β-blockers as the preferred first-line agents.^14- 15

The patterns of drug use that we have documented suggest that these recommendations have little, if at all, informed the practice of nursing home physicians. These findings are in agreement with several recent reports describing the progressive decline of diuretic use and the parallel increase of use of ACE inhibitors and calcium channel blockers that have occurred in other settings.^16- 19 Several reasons can possibly explain the apparent nonadherence to widely accepted recommendations. Some authors have suggested that clinicians may be disappointed with the results of completed clinical trials in which blood pressure reduction has yielded less benefit on coronary artery disease outcomes than expected.¹⁶ Such consideration implies a widespread dissemination of the recommendations; although possible, this has been shown not to occur in different settings.^16,47- 48 Instead, it is likely that the effectiveness of pharmaceutical promotion practices relative to the guidelines or the attractiveness of using new therapies has had the greatest impact.^49- 50 Alternatively, the present findings may reflect appropriate prescribing by an enlightened physician facing the difficult applicability of the guidelines to the frail, very old nursing home resident with multiple comorbid conditions and concomitant, complex pharmacological regimens.

Indeed, the results of the published trials are hardly applicable to the treatment of all older hypertensive patients, and apparently less so to institutionalized individuals.^11,51 In seeking internal validity, randomized controlled trials have shown efficacy of treatment for only an average, "eligible" patient, and not for pertinent subgroups characterized by such cogent clinical features as severity of symptoms, illness, comorbidity, and other clinical nuances. Generally, the patients enrolled were healthy cohorts of younger people living in the community.¹¹ For example, the Systolic Hypertension in the Elderly Program trial⁵² included only 2% demented, 10% depressed, and 7% physically disabled patients, whereas in an age-matched general population these conditions have been reported to occur in 6%, 23%, and 19% of people, respectively. These estimates are far lower than those found among the oldest old, frail residents of long-term care facilities, which were specifically excluded by any trial. For example, in the present study, 29% to 43% of patients were affected by dementia and, indirectly confirming the validity of the diagnosis of hypertension, the vascular type was far more common. This circumstance merits careful consideration, because, although there is evidence that antihypertensive medications commonly used do not adversely affect cognition,^53- 55 overall the issue remains controversial.⁵⁶ The relatively high prevalence of diabetes mellitus and the additional presence of chronic obstructive pulmonary disease might have justified a reduced use of diuretics and β-blockers, which have been shown to exacerbate insulin resistance, adversely affect plasma lipid levels, and worsen symptoms of obstructive pulmonary disease.¹⁵ Conversely, in a more empirical way, ACE inhibitors and calcium channel blockers could have had better safety profiles in these conditions. Finally, the high prevalence of urinary incontinence, frequent falls, and depression may have justified the use of certain medications in favor of others.

We found that about 30% of hypertensive patients were not receiving any traditionally known antihypertensive agent, and this is consistent with previous reports.^57- 60 Even after adjusting for many relevant variables, increased age was strongly associated with an increased likelihood of not being treated, whereas high-risk individuals (those with multiple cardiovascular comorbidities) seemed more likely to receive a pharmacological treatment. Because of the limitations of our data, no inferences can be made regarding quality or appropriateness of treatment. However, it is completely unknown whether the risks of therapy may outweigh the benefits among severely impaired patients.

Also, the possibility that the benefit of antihypertensive treatment does not extend to patients beyond a certain age threshold needs to be carefully considered. In the randomized controlled trials of antihypertensive drug treatment in the elderly, age at recruitment ranged from 60 to 97 years. In the European Working Party on High Blood Pressure in the Elderly trial,⁴⁰ the reduction in cardiovascular mortality was progressively lessened in the oldest patients, especially beyond 80 years. There was similar evidence in the Swedish Trial in Old Patients with Hypertension.⁴¹ Older patients randomized to receive active treatment, compared with placebo, experienced less reduction of stroke, myocardial infarction, or other cardiovascular death than younger patients. A new report from the Systolic Hypertension in Europe Trial reinforces the warning that the effect of antihypertensive drug treatment may be attenuated in very old patients.⁶¹ Above 75 years of age there was no treatment benefit in terms of total and cardiovascular mortality. Nonetheless, the authors documented that stroke and cardiac end points may be still prevented, at least until 78 to 80 years of age (eg, when the upper 95% confidence limit included unity). Yet, these do not represent conclusive evidence, because trials differed in their treatment regimens and outcome definitions, and may have included too few patients older than 80 years to permit meaningful interpretation of the risk in this age stratum. In this regard, in the Systolic Hypertension in the Elderly Program, which included 650 patients aged 80 years or older, the reduction in the risk of stroke was completely preserved in this cohort. The remaining uncertainties could be resolved by the results of the Hypertension in the Very Elderly Trial⁶² currently ongoing.

Our study has several limitations. First, it included exclusively institutionalized individuals, and as such the findings are not necessarily generalizable to the majority of the oldest old hypertensive patients living in the community. It is difficult to determine whether cross-sectional findings related to age may be attributed to aging or to age-related selection bias, such as a survivor effect. There is potential for misclassification of the diagnosis of hypertension and for the use of "antihypertensive" medications for indications other than hypertension. We lack blood pressure measurements and we are unable to assess the severity and control of hypertension. However, several considerations lead us to believe that our sample represents "real" hypertensive patients, and the bias, if any, is conservative. First, the prevalence estimates we report are similar to those documented by the recent national nursing home survey⁶³ and by other authors examining patients who live in long-term care facilities, with clinical examination,⁶⁴ chart review,⁶⁵ or analysis of data collected with MDS.⁶⁶ In addition, the age- and gender-related trends are consistent with several National Health Interview and Examination Surveys in noninstitutionalized people,^67- 70 but also with reports on residents of long-term care facilities.^67,71 Second, to ensure quality of care, the nurses in our study had to measure blood pressure before administering an antihypertensive medication, and drug treatment, regardless of specific class, was stable over extended periods (G.G., unpublished observation, 1998). Third, a diagnosis of hypertension was found consistently across all MDS assessments if the patients had such a condition collected. Fourth, the accuracy, reliability, and validity of the diagnosis of hypertension has already been shown to be excellent.^33,35 In addition, in our study as well in others,⁷² few patients (approximately 9%) were using an "antihypertensive" drug for an indication other than hypertension or any plausible cardiovascular condition.

In conclusion, our study shows that current prescription patterns do not concur with widely approved guidelines for the pharmacological treatment of hypertension. Future studies will have to address the precise reasons for this divergence, but age, functional status, and comorbidity appear to be important determinants of treatment choice. Although this is not a phenomenon restricted to the nursing home, the risk-benefit ratio of any antihypertensive treatment remains undefined in the typical nursing home patient.

Accepted for publication March 12, 1998.

Steering Committee: Roberto Bernabei, MD, Università Cattolica del Sacro Cuore, Rome, Italy; Constantine Gatsonis, PhD, Brown University, Providence, RI; Lewis Lipsitz, MD, Harvard Medical School, Boston, Mass; Vincent Mor, PhD, Brown University. Coordination: Giovanni Gambassi, MD, Università Cattolica del Sacro Cuore and Brown University; Kate Lapane, PhD, Brown University. Writing Panel: Daniel Forman, MD, Brown University; David Gifford, MD, Brown University; Francesco Landi, MD, Università Cattolica del Sacro Cuore; Antonio Sgadari, MD, Università Cattolica del Sacro Cuore. Data Management: Jeffrey Hiris, MS, Brown University; Chris Brostrup-Jensen, BA, Brown University; Sharon Gonzales, MSPH, Brown University. Biostatistics: Constantine Gatsonis, PhD, Brown University; Joseph Hogan, PhD, Brown University; Orna Intrator, PhD, Brown University. Participants: Marilyn Barbour, PharmD, University of Rhode Island, Kingston; Katherine Berg, PhD, Brown University; Anne Hume, PharmD, University of Rhode Island; Paul Larratt, PhD, University of Rhode Island; Knight Steel, MD, Hackensack University Medical Center, Hackensack, NJ.

Reprints: Giovanni Gambassi, MD, Center for Gerontology and Health Care Research, Brown University, Box G-B213, Providence, RI 02912 (e-mail: giovanni_gambassi@brown.edu).