The Complexity and Cost of Drug Regimens of Older Patients Hospitalized With Heart Failure in the United States, 1998-2001 FREE

The medical management for patients with heart failure is becoming increasingly complex. For many patients, treatment includes drugs that reduce mortality (eg, angiotensin converting enzyme inhibitors) and others that are used to ameliorate symptoms (eg, digoxin).¹ Simultaneously, the medical therapy of related conditions including hypertension,² coronary artery disease,³ dyslipidemia,⁴ and diabetes⁵ is evolving to include increasing numbers of drugs considered appropriate if not necessary. Among elderly patients with heart failure, noncardiovascular conditions are also common.⁶ These comorbidities may require additional medications, further increasing the likelihood that the comprehensive treatment of the older patient with heart failure will involve several drugs. Not surprisingly, polypharmacy—the prescription of multiple drugs concurrently—has thus been identified as a salient issue in the care of older patients with heart failure.⁷

Despite concerns about polypharmacy, however, surprisingly few data characterizing the drug regimens used to treat patients with heart failure exist.^8- 10 Practice guidelines typically do not acknowledge the issue, providing recommendations that rarely consider the incremental effects on regimen complexity, patient adherence, or the risk of adverse drug reactions (ADRs).^1- 5 With drug costs increasing at a rate greater than any other component of health expenditures,¹¹ the financial impact of larger drug regimens on older persons may also be considerable. More precise knowledge of the costs of drug therapy is particularly important in light of the recently passed Medicare prescription drug benefit, which is designed to pay for a variable proportion of medication costs but may require substantial out-of-pocket expenditures from many beneficiaries.¹²

Therefore, we studied patterns of medication prescription in a nationally representative community-based cohort of older Medicare beneficiaries hospitalized with heart failure between 1998 and 2001. Our goals were to characterize the complexity and cost of chronic drug therapy for patients recently hospitalized with heart failure and to assess trends in the number of medications and estimated costs over time in this patient population.

The National Heart Care (NHC) Project is a Centers for Medicare and Medicaid Services (CMS) initiative designed to improve the quality of care for Medicare beneficiaries hospitalized with heart failure.⁶ The project data consist of 2 national samples of fee-for-service Medicare beneficiaries hospitalized with a principal discharge diagnosis of heart failure (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 402.01, 402.11, 402.91, 404.01, 404.91, or 428).

The first sample included hospitalizations with discharge dates between April 1998 and March 1999, inclusive, and the second between July 2000 and June 2001, inclusive. Thus, the dates of the sampling periods were separated by 27 months. All identified discharged patients in each of the 50 states, Washington, DC, and Puerto Rico within these time frames were sorted by age, sex, race, and hospital. Within each state, up to 800 discharges in each of the 2 sampling frames were randomly selected; a census of records was obtained in states with fewer than 800 eligible discharged patients. Records were reviewed in central data abstraction centers for clinical data. Data quality was ensured through the use of trained abstractors, electronic abstraction instruments, and record reabstraction. Patients without valid social security numbers, receiving long-term hemodialysis, transferred to another hospital, or leaving against medical advice were excluded from the NHC cohorts, which consisted of 39 477 records in 1998-1999 and 39 405 records in 2000-2001.

Because this was a study of discharge prescription drug regimens, patients who died in the hospital (n = 4126) were excluded. Patients younger than 65 years (n = 6214) were also excluded because the younger Medicare population is defined by those with end-stage renal disease or chronic disabilities and is thus not representative. For patients with multiple admissions (n = 3394), a single admission was selected at random for inclusion. Finally, patients with incomplete discharge prescription data (n = 781) were not included. The resulting cohorts consisted of 31 602 patients in 1998-1999 and 30 774 patients in 2000-2001.

The NHC database contains a wide range of demographic and clinical variables, including age, sex, race, cardiac history, noncardiac medical history, characteristics at presentation, laboratory values, imaging data, discharge disposition, and discharge medications. For laboratory and imaging variables with more than 1 measurement (serum creatinine, potassium, and left ventricular systolic function), the value closest to discharge was used. Because serum creatinine level may not accurately reflect renal function in elderly patients, creatinine clearance was estimated using the Modification of Diet in Renal Disease 4-variable formula.¹³ Characteristics of the attending physician responsible for care (specialty and board certification) and of the hospital (teaching status, ownership, availability of cardiac care facilities, and urban/rural designation) were ascertained with the American Medical Association Physician Masterfile and the American Hospital Association Annual Surveys, respectively.

The data abstracted from the medical record included a complete listing of the names of all medications prescribed at hospital discharge. The summary list of medications from all members of the study cohort was reviewed by 2 clinicians, who independently attributed misspellings. If this attribution was discordant, the medication was not included in the analyses (<0.5% of all drug names and <0.01% of prescriptions). The database of drug names was subsequently linked to the Mosby’s Drug Consult 2003,¹⁴ which provides the typical dosing interval and the average wholesale price of a medication. To assess differences between time periods independent of inflation of drug prices, both samples used the same pricing standard (year 2003).

The outcome measures of this analysis on the patient level were as follows: (1) the total number of discharge medications, (2) the estimated number of medication doses per day, and (3) the estimated annual cost of the discharge drug regimen. The most frequently prescribed classes of medications in the cohort were also tabulated. All measures were also analyzed in subgroups of cardiovascular agents and noncardiovascular drugs using the classification of the American Society of Health-Systems Pharmacists Drug Information Classification System,¹⁵ except that diuretics, warfarin, aspirin, other antiplatelet agents, and niacin were classified as cardiovascular agents.

The primary purpose of this study was to assess the complexity and cost of long-term medication regimens. Therefore, drugs typically used for short courses (eg, antibiotics), on an as-needed basis (eg, sublingual nitroglycerin), or administered intravenously (eg, dobutamine) were excluded. Because documentation of use is likely to be inconsistent for vitamins, other nutritional supplements, lotions, and powders, these were also excluded.

Because the intervals of administration were not abstracted from the medical record, it was assumed that all medications were prescribed with the frequency identified as the typical interval considered appropriate for treatment for approved indications. The cost used for each medication was the 2003 average wholesale price (a benchmark price used for reimbursement, representing what a retail pharmacist or a dispensing physician might pay for a product without special discounts). The same cost standard was applied to both samples to eliminate the influence of drug cost inflation. For drugs with both proprietary and generic versions available, the name provided in the discharge record was used to determine the cost estimate.

The distributions of the numbers of medications, doses, and estimated annual costs were compared with Wilcoxon rank sum tests. Adjusted values for the number of medications and estimated adjusted annual costs adjusting for patient, physician, and hospital characteristics were derived from multivariable regression models, including a variable for the sampling time frame (1998-1999 compared with 2000-2001) to assess for temporal changes. Because the distributions of the outcome variables were not normal, simple least squares regression was not used. To assess the number of drugs, a negative binomial model and log-link function were used.¹⁶ This approach is appropriate for count data characterized by overdispersion. In such cases, the use of Poisson regression will result in erroneous error estimates.¹⁷ For the cost models, estimated cost was log-transformed. To correct for retransformation bias due to heteroscedasticity, a smearing estimator was used, which provided estimates of the adjusted arithmetic means of the annual costs on a linear scale for greater interpretability.¹⁸ Because all of the patients in the sample were discharged with at least 1 medication, assessment of nonuse was not required.

All P values are 2-tailed. The analysis was performed with SAS software, version 8.0 (SAS Institute, Cary, NC). The analysis of the NHC Project data was approved by the institutional review board of the University of Colorado.

Of the 62 376 subjects, 17 097 (27%) were at least 85 years old (Table 1). Most were female (58%) and white (87%). Comorbid conditions were common, including hypertension (64%), myocardial infarction (29%), chronic lung disease (34%), and diabetes mellitus (40%). One quarter had preserved left ventricular systolic function documented. More than half (54%) were either cared for primarily by a cardiologist or were seen by a cardiologist in consultation. Most received care in nonteaching institutions (65%) with a range of cardiac care facilities.

Between the 2 time periods, the mean number of chronic medications prescribed increased from 6.8 to 7.5 (11% relative increase; P<.001) (Table 2; Figure, A), and the mean daily number of doses increased from 10.1 to 11.1 (9% relative increase; P<.001) (Figure), B. The mean estimated annual cost of medications between the 2 periods, however, rose by 22%, from $3142 to $3823 (P<.001) (Figure, C), and the median cost increased by 27% from $2745 to $3479. Thus, the cost per prescribed drug increased from $462 to $510 between samples, or 10%. The increase in the number of medications, cost, and mean estimated cost per prescription was greater for noncardiovascular drugs than for cardiovascular drugs.

In 1998-1999, the highest decile of patients was prescribed at least 10 medications, requiring at least 17 doses daily at an estimated annual cost of at least $5880. By 2000-2001, the highest decile had increased to at least 11 medications and at least 18 doses daily with an estimated annual cost of $6891.

In both 1998-1999 and 2000-2001, loop diuretics, angiotensin-converting enzyme inhibitors, potassium supplements, digoxin, and aspirin were the most commonly prescribed cardiovascular medications (Table 3). In 2000-2001, prescriptions for β-blockers, hyroxymethyl glutaryl coenzyme A reductase inhibitors, and potassium-sparing diuretics had increased substantially. The most common noncardiovascular drug classes included those for lung disease, hypothyroidism, diabetes, gastrointestinal, and psychiatric conditions. The most substantial change in prescriptions between the 2 time frames was for proton-pump inhibitors, which increased from 14.4% of the population in 1998-1999 to 22.3% in 2000-2001 (Table 3).

In the multivariable models, adjusting for patient, physician, and hospital characteristics, younger age was associated with larger medication regimens. Patients aged 65-74 years received a mean of 1.5 more drugs than patients older than 84 years, with an additional estimated annual cost of $1166 (P<.001 for both comparisons, Table 1). Black patients received significantly fewer medicines (7.2) with lower associated estimated annual costs ($3634) than white patients (7.9 medicines, $4161 estimated annual cost; P<.001). Cardiovascular conditions associated with the greatest incremental complexity and cost after adjustment included a history of myocardial infarction (0.8 additional medications, $754 additional cost) and a history of revascularization (1.3 medications, $1154). The most important comorbidities included diabetes mellitus (1.6 medications, $1094), chronic lung disease (1.2 medications, $814), and reduced estimated glomerular filtration rate (0.9 medications, $830 for <30 mL/min per 1.72 m² compared with >90 mL/min per 1.72 m²). Patients receiving care from a cardiologist or from a board-certified physician were treated with more medications at a higher cost. Hospital characteristics were also associated with differences in regimen complexity and cost.

After adjustment, the mean number of medications increased from 7.4 to 8.3 (12% relative increase; P<.001) between 1998-1999 and 2000-2001, while estimated adjusted mean annual cost increased from $3649 to $4526 (24% relative increase; P<.001). The adjusted estimated cost per prescription also increased by $52, or 10%.

This study illustrates the prescription drug regimens provided to a representative community-based population of older persons hospitalized with heart failure. The size of these regimens, considerable in both time periods, increased significantly over 27 months. Estimated costs increased by approximately 25% without consideration of drug cost inflation. In 2000-2001, the mean number of medications exceeded 8; regimens required a mean of more than 11 doses daily; and the estimated mean annual cost exceeded $3800. Noncardiovascular agents constituted a substantial minority of the regimens, and common comorbidities contributed significantly to the complexity and cost of drug therapy.

Although the importance of polypharmacy in the contemporary management of patients with heart failure has been acknowledged,⁷ relatively little has been known about the complexity and estimated costs of drug therapy among elderly patients with heart failure. Survey data of randomly selected ambulatory adults 65 years or older in the United States in 1998-1999 indicate that 6% of men and 23% of women used at least 5 prescription drugs.¹⁹ Not surprisingly, the treatment of heart failure and highly prevalent comorbidity in this elderly hospitalized population results in markedly more complex medication regimens.

The cost of prescription drugs in the elderly has recently been a central focus of policy debate. Despite the high prevalence of heart failure, however, existing studies of the estimated costs of drug therapy for heart failure have not been conducted in the United States or do not reflect contemporary practice.^8- 10 In the present study, the adjusted costs of treatment increased over a 27-month period by 24%, out of proportion to the number of drugs prescribed. Because the inflation rate for prescription drugs was not reflected in these estimates, this finding indicates that cost increases are due not only to the larger number of drugs prescribed, but also to the use of more expensive medications. This phenomenon of substitution of more costly agents has been identified in other populations as an important contributor to the rapid increases in the cost of prescription drugs.^11,20

Polypharmacy also raises concerns of the risks of ADRs and nonadherence. Adverse drug reactions are important causes of preventable iatrogenic injury and hospitalization in older persons,²¹ are related to the number of drugs taken,^22- 23 and are particularly common with cardiovascular drugs.^21- 22,24 Thus, the patients represented in the present study may be at particularly high risk for ADRs.

Although nonadherence has many root causes, higher cost and greater complexity of drug regimens have been identified as important contributors.^25- 27 The recently enacted Medicare Prescription Drug, Improvement, and Modernization Act of 2003 includes a drug benefit (Part D) that will be implemented in 2006.¹² This is intended to reduce the burden of the cost of medications to older persons. Although this plan does not provide complete prescription coverage, it is possible that it will enhance adherence by improving access to medications for older patients. Because nonadherence diminishes the possible benefits of drug treatment, the findings of the present study highlight the importance of interventions to minimize the cost and complexity of the medical regimen in older patients with heart failure wherever possible.

Despite the potential adverse consequences of polypharmacy, the most appropriate regimens for older patients with heart failure are not known and likely vary substantially among individuals. It is known, however, that the patterns of prescription in this study occur in the context of a high prevalence of underuse of some medications and simultaneous overuse or misuse of others. Significant gaps in the use of potentially beneficial agents for cardiovascular conditions, particularly among patients with competing comorbidity, have been well documented.²⁸ Specifically, a recent study of the NHC cohort demonstrated substantial underuse of angiotensin-converting enzyme inhibitors among ideal candidates with left ventricular systolic dysfunction.²⁹ Conversely, other studies have demonstrated high rates of use of potentially harmful agents. For example, an assessment of the NHC cohort demonstrated that spironolactone is often prescribed to patients with heart failure who are at high risk for hyperkalemia.³⁰ Optimizing regimens for older patients with heart failure and multiple comorbidities will therefore require simultaneous efforts to increase the use of potentially beneficial medications and to minimize the use of potentially harmful and contraindicated drugs.

Unfortunately, recommendations in practice guidelines are typically based on the assessment of the efficacy of individual medications in highly selected populations of randomized trials and do not consider the implications of cost, drug interactions, or nonadherence. As a result, practitioners are provided little guidance for managing the complex cases of patients with heart failure and comorbidities comprehensively. Further studies in community-based populations are needed to determine the applicability of clinical trial results and optimal management strategies for older patients with heart failure and multiple competing comorbidities requiring polypharmacy treatment. Future recommendations for additional drug therapy for heart failure or conditions commonly occurring in elderly patients should not only incorporate evidence of effectiveness but should also acknowledge the substantial burden of polypharmacy that characterizes the contemporary medical management of this population.

Certain issues should be considered in the interpretation of these results. First, this study assessed only prescription drugs. The exclusion of other medications was intended to provide conservative estimates but could also lead to underestimates of cost and complexity. Second, the regimens attributed to the patients in this study were based on those prescribed at hospital discharge. We were unable to ascertain patient adherence or changes in the drug regimen after discharge, which might have influenced the results in some patients. However, we characterized the drug regimens that clinicians expect their patients to pay for and to take. Finally, because the hospitalized Medicare population is a particularly ill segment of the larger population with heart failure, these results may not be applicable to other populations with heart failure. However, because older patients account for the majority of heart failure admissions in the United States, these results apply to a large segment of the population with heart failure.

In conclusion, the costs and complexity of drug regimens prescribed to older patients discharged after hospitalization for heart failure are increasing rapidly. This dual increase is attributable not only to the drugs used to treat heart failure and other cardiovascular conditions, but also to those used to treat a wide range of comorbidities. Further study of the optimal armamentarium of drugs will be instrumental in optimizing outcomes and the costs of drug therapy in the growing population of older patients with heart failure and multiple comorbidities.

Correspondence: Frederick A. Masoudi, MD, MSPH, Division of Cardiology, Mail Code 0960, Denver Health Medical Center, 777 Bannock St, Denver, CO 80204 (fred.masoudi@uchsc.edu).

Accepted for Publication: June 9, 2005.

Financial Disclosure: Dr Masoudi has served on speakers’ bureaus of AstraZeneca, Pfizer, and Takeda NA; he has received consultation fees from Takeda NA and Wyeth. Dr Havranek has received both consultation fees and honoraria from Bristol-Myers Squibb and has been on the speakers’ bureau for Takeda NA. Dr Foody has served on speakers’ bureaus for Merck, Pfizer, and Bristol-Myers Squibb. None of these entities provided financial support for or were involved in the conduct of this study.

Funding/Support: Dr Masoudi is supported by National Institutes of Health (NIH) Research Career Award K08-AG01011. Dr Foody is supported by NIH/National Institue on Aging (NIA) Research Career Award K08-AG20623 and NIA/Hartford Foundation Fellowship in Geriatrics. The analyses on which this publication is based were performed under contract No. 500-99-C001 titled “Utilization and Quality Control Peer Review Organization for the State of Colorado” sponsored by the Center for Medicare and Medicaid Services, Department of Health and Human Services. This article is a direct result of the Health Care Quality Improvement Program initiated by the Center for Medicare and Medicaid Services, which has encouraged identification of quality improvement projects from analysis of patterns of care, and therefore required no special funding on the part of this contractor.

Disclaimer: The content of the publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US government.

Previous Presentation: This study was presented as an abstract at the American Heart Association Scientific Sessions; November 2004; New Orleans, La.

Additional Information: Ideas and contributions to the author concerning experiences in engaging with issues presented are welcomed.