Clinical Equivalence of Generic and Brand-Name Drugs Used in Cardiovascular Disease:  A Systematic Review and Meta-analysis FREE

The problem of rising prescription drug costs has emerged as a critical policy issue, straining the budgets of patients and public/private insurers¹ and directly contributing to adverse health outcomes by reducing adherence to important medications.^2- 3 The primary drivers of elevated drug costs are brand-name drugs, which are sold at high prices during a period of patent protection and market exclusivity after approval by the Food and Drug Administration (FDA).⁴ To control spending, many payers and providers have encouraged substitution of inexpensive bioequivalent generic versions of these drugs, which can legally be marketed by multiple manufacturers after the brand-name manufacturer's market exclusivity period ends.⁵

Generic drugs are chemically equivalent to their brand-name counterparts in terms of active ingredients but may differ in peripheral features, such as pill color or shape, inert binders and fillers, and the specific manufacturing process.⁶ The 1984 Hatch-Waxman Act first authorized the FDA to approve generic drugs demonstrated to be “bioequivalent,” which is defined as absence of a significant difference in the availability of the active ingredient at the site of drug action.⁷ Bioequivalency can be established on the basis of the maximum serum concentration of the drug, the time until maximum concentration is reached, or the area under the curve based on serum concentration as a function of time.

Some physicians and patients have expressed concern that bioequivalent generic and brand-name drugs may not be equivalent in their effects on various clinical parameters, including physiological measures such as heart rate or blood pressure, important laboratory measurements, and outcomes such as health system utilization or mortality.^8- 10 Of particular concern are narrow therapeutic index (NTI) drugs, which are drugs whose effective doses and toxic doses are separated by a small difference in plasma concentration. Brand-name manufacturers have suggested that generic drugs may be less effective and safe than their brand-name counterparts.¹¹ Anecdotes have appeared in the lay press raising doubts about the efficacy and safety of certain generic drugs.^12- 13

Little empirical evidence has been assembled to assess clinical differences resulting from the use of generic medications, so we sought to systematically evaluate comparisons of generic and brand-name drugs on these outcomes. We focused on drugs used primarily to treat cardiovascular disease, which as a group make up the largest portion of outpatient prescription drug spending.¹⁴ We reviewed studies published from 1984 to 2008 comparing clinical characteristics of generic and brand-name drugs in this field and pooled available results. To determine the concurrent expert opinion on the subject of generic substitution, we also systematically reviewed the content of editorials published during this time.

We performed a systematic search of articles published in peer-reviewed health care–related journals between January 1984 and August 2008 using MEDLINE, EMBASE, and International Pharmaceutical Abstracts (IPA) with the help of a professional librarian.

We used 3 main subject heading domains: terms relating to the type of study (for example, clinical study, crossover, equivalen$, effect$, and outcome$), terms relating to the products of interest (for example, brand-name, nonproprietary, generic$, innovator$, patent$, and pharmaceutical drug), and terms relating to cardiovascular medicine. Cardiovascular disease was defined as any condition affecting the heart or blood vessels, including myocardial infarction, hypertension, cardiac arrhythmias, peripheral vascular disease, and heart failure. Under the cardiovascular category, we used search terms addressing general terms (eg, cardiovascular, heart, hematologic), cardiovascular disease (eg, atherosclerosis, hyperlipid, ischemia), and classes of pertinent drugs (eg, β-agonist, anticoagulant). Articles containing at least 1 search term in each of the 3 main categories met criteria for the title/abstract review.

Search terms and parameters were adjusted for each database while maintaining a common overall architecture. Search results from MEDLINE and EMBASE were combined and screened for duplicate entries. Search results from IPA were handled separately because of differences in output organization.

Studies were included if they reported on a comparative evaluation of 1 brand-name drug and at least 1 generic version produced by a distinct manufacturer (biologic products, which are regulated differently, were excluded). The comparative evaluation had to include measurement of at least 1 clinical efficacy or safety end point, including a vital sign (eg, heart rate, blood pressure, urine output), a clinical laboratory study (eg, international normalized ratio [INR], low-density lipoprotein, urine electrolytes), patient morbidity or mortality, or health system utilization. “Clinical laboratory studies” did not include specialized assays of concentrations of the drug or its metabolites used in pharmacokinetic evaluation.

We included both randomized controlled trials (RCTs) and observational studies. We excluded case studies as well as qualitative analyses of effectiveness, pharmacoeconomic evaluations, or surveys. For this part of the study, we also excluded commentaries, essays, legal analyses, consensus statements, and letters to the editor. Studies were excluded if they were written in a language other than English or they were conducted in vitro or in animals. Although the study could take place in any location, the brand-name drug used (or an identical formulation of it) must have been approved by the FDA. Manual reference mining of articles, letters, and commentaries supplemented the search results.

Data were extracted (A.S.K.) and checked (W.H.S.), with disagreements resolved by consensus. We assessed a number of variables related to the organization and outcome of the studies: the study design, listed source of funding, the setting (US vs non-US), the characteristics of the population studied, the number of participants, the mean age (or age range) of the participants, the clinical end points, and the self-identified source of funding (where listed). The methodological quality of the randomized clinical trials (RCTs) was assessed using the 5-point scale developed by Jadad et al.¹⁵ The methodological quality of nonrandomized trials was assessed using the 9-star Newcastle-Ottawa scale.¹⁶ This was done independently by 2 of us (A.S.K. and W.H.S.), with differences resolved by consensus.

Drugs were further subdivided based on whether they had a wide therapeutic index (WTI) or NTI. The federal definition of an NTI drug follows: “(a) There is less than a 2-fold difference in median lethal dose (LD₅₀) and median effective dose (ED₅₀) values, or (b) There is less than a 2-fold difference in the minimum toxic concentrations and minimum effective concentrations in the blood, and (c) Safe and effective use of the drug products require careful titration and patient monitoring.”^17- 18 The FDA does not formally designate the therapeutic index of drugs, but according to this definition (confirmed with review of the cardiovascular literature), relevant drugs with an NTI include the anticoagulant warfarin (Coumadin; DuPont Pharmaceuticals, Wilmington, Delaware) and antiarrhythmic drugs affecting the sodium and potassium channels (class I and class III).

To conduct a meta-analysis of included studies, we identified those RCTs where means and standard deviations for clinical outcomes were presented or could be derived from the published results. If the correlation was not reported for a crossover design, we assumed a coefficient of 0.5. We calculated a Cohen D effect size for each study with a 95% confidence interval (CI) according to established methods from information provided in the article.^19- 22 The effect sizes compare the difference in effect between the study groups divided by the standard deviation of this difference. We considered an effect size of less than 0.2 to be very small, an effect size of 0.2 to 0.5 to be small, an effect size of 0.5 to 0.8 to be medium, and an effect size of greater than 0.8 to be large. Since this measure is independent of the measurement used, sample size, and standard deviation of the outcome measure, we aggregated different end points across studies to obtain effect sizes with 95% CIs for each cardiovascular drug class as well as an aggregate effect size for all studies included in the meta-analysis.²³

We assessed the perspectives presented in editorials about the appropriateness of using generic drugs in treating cardiovascular disease during the same time period covered by our systematic review of the data. We repeated the MEDLINE and EMBASE searches using the same criteria. Two of us (A.S.K. and A.S.M.) then reviewed each title and abstract. Editorials were defined as articles expressing perspectives or viewpoints that did not include direct pharmacokinetic or clinical comparisons of generic and brand-name drugs. We also excluded systematic literature reviews, reports of surveys, case reports without substantial additional discussion, and letters to the editor.

Using content analysis,²⁴ 2 of us (A.S.K. and W.H.S.) then coded themes in the commentaries. We focused on the examples used (if any), sources cited (if any), and ultimate conclusions reached to categorize the editorial's viewpoint within 1 of 3 main categories: (1) those presenting a generally negative opinion discouraging generic drug substitution, (2) those presenting a generally positive opinion encouraging generic drug substitution, and (3) those presenting a neutral analysis or that otherwise made no recommendations on the issue. We determined whether the editorial addressed generic and/or cardiovascular drugs broadly or focused on a subset of drugs, such as NTI drugs or drugs in a particular class. Investigators reconciled differences in coding by consensus.

The search done in September 2008 identified 8556 records, 3932 records from EMBASE, 2848 records from MEDLINE, and 1776 records from IPA. After removing overlapping citations and applying our exclusion criteria, 71 articles were prioritized from those 3 sources. We added 2 studies from evaluation of citations from prioritized articles. A total of 26 citations were excluded after full review. In total, our review identified 47 articles for detailed analysis (Figure 1), covering 9 different subclasses of cardiovascular drugs.

Nearly half of included studies (23/47, 49%) were primarily bioequivalency studies, in which pharmacokinetic comparisons occurred along with clinical end points, and more than a third (18/47, 38%) involved only healthy, young subjects. Less than half of the articles (21/47, 45%) were published since 2000 and only 17 (36%) were conducted in the United States. Table 1, Table 2, Table 3, and Table 4 include all categories of WTI cardiovascular drugs while Table 5 highlights the 2 NTI categories, warfarin (Coumadin) and class I antiarrhythmic drugs.

Nearly all trials (31/34, 91%) comparing generic and brand-name cardiovascular drugs with a WTI were RCTs with a crossover design. These articles encompassed 7 different drug classes, although more than three-fourths (27/34, 79%) involved β-blockers, diuretics, or calcium channel blockers.

We identified 9 articles that compared clinical outcomes in generic and brand-name β-blockers.^25- 33 These studies involved 4 different β-blockers: long-acting metoprolol (Toprol XL; AstraZeneca, Wilmington, Delaware), atenolol (Tenormin; AstraZeneca), carvedilol (Coreg; GlaxoSmithKline, London, England), and propranolol (Inderal; Ayerst Laboratories, Radnor, Pennsylvania). Long-acting metoprolol was evaluated in 1 double-blind RCT in outpatients with stable angina and 1 retrospective cohort study involving nearly 50 000 German patients over 4 years.²⁵ The cohort study identified users of β-blockers from provincial administrative data in Germany and found no differences in clinical outcomes after controlling for patient sociodemographic characteristics and their comorbidities. In 1 RCT in outpatients with hypertension and 2 bioequivalency studies in healthy volunteers, Tenormin was not found to be superior to the generic version in lowering heart rate and blood pressure.^27,29- 30 In a retrospective cohort study of patients switching from short- to long-acting β-blocker preparations, self-reported adverse effects occurred more frequently at baseline in patients taking generic propranolol than in those taking Inderal (34.6% vs 24.8%, P < .001), and the difference was noted to be extinguished after all were switched to Inderal LA (Long-Acting) (20.5% vs 17.6%, P = .15).³³ These patients were not randomly assigned to different preparations, and recipients of the generic formulation may have been different from recipients of the brand. An RCT later conducted in hypertensive patients found no clinical differences, including rates of observed adverse effects, among these 3 versions of propranolol.³¹

Eleven articles compared outcomes among patients using diuretics: 10 with the loop diuretic furosemide (Lasix; Sanofi-Aventis, Paris, France)^{34- 36,38- 44} and 1 with the combination diuretic triamterene-hydrochlorothiazide (Dyazide; GlaxoSmithKline).³⁷ The furosemide studies were of lower quality, and 7 were bioequivalency studies performed in a total of 82 generally young, healthy subjects who received only 1 dose of each brand-name or generic formulation.^{35- 36,39- 41,43- 44} The clinical end points for these studies were primarily urine output and urine electrolytes. However, only 1 study, conducted in South Africa in 1985, found significant differences.³⁹

Three studies of furosemide involved patients with volume overload. In these studies, generic and brand-name formulations of furosemide showed no significant clinical differences.^34,38,42 A 1997 open-label RCT with crossover in 17 outpatients with congestive heart failure who received Lasix, 3 versions of generic furosemide, and intravenous furosemide for a week's time noted wide intraindividual variability in patients' urine electrolytes that the authors hypothesized might overwhelm any minor differences in bioavailability.³⁴ The study of triamterene-hydrochlorothiazide was a prospective RCT in 30 patients with hypertension.³⁷ It demonstrated no statistically significant differences on blood pressure and serum electrolytes in patients using the medication for 3-week blocks.

Seven articles evaluated generic and brand-name versions of calcium channel blockers.^45- 51 The largest, a multicenter, double-blind, parallel-group RCT in 189 patients with hypertension, found improvements in blood pressure and no significant differences between brand-name and generic versions of amlodipine (Norvasc; Pfizer, New York, New York) over 8 weeks.⁴⁵ Two studies reported slight, but statistically significant, differences in 1 measured clinical outcome (the PR interval on electrocardiogram), although there were no associated changes in heart rate or other clinical outcomes in either of those studies.^50- 51

The remaining 7 studies evaluated antiplatelet agents (clopidogrel; [Plavix; Bristol-Myers Squibb, New York, New York] and enteric-coated aspirin [acetylsalicylic acid]),^52- 54 the angiotensin- converting enzyme (ACE) inhibitor enalapril (Vasotec; Merck, Whitehouse Station, New Jersey),⁵⁵ the statin simvastatin (Zocor; Merck),^56- 57 and the α-blocker terazosin (Hytrin; Abbott Laboratories, Abbott Park, Illinois).⁵⁸ None of these studies reported significant clinical differences between the generic and brand-name versions. Two longer-term RCTs of simvastatin were conducted in Thailand. Both of these studies, of high methodological quality, showed no statistically significant differences in lowering low-density lipoprotein levels.^56- 57 However, there were a number of important limitations in the studies. The 2 studies of clopidogrel used clinical outcomes related to platelet aggregation and bleeding time, not incidence of cardiovascular disease such as myocardial infarction.^52- 53 The study involving enalapril was well designed but measured bioequivalency in 24 healthy subjects who received only 1 dose of the generic and brand-name forms.⁵⁵ The terazosin study, which was conducted in outpatients with benign prostatic hypertrophy, found no significant differences in heart rate and blood pressure and was of relatively high quality.⁵⁸

Thirteen articles analyzed generic and brand-name versions of cardiovascular drugs with an NTI. Two addressed clinical end points in treatment with class I antiarrhythmic drugs (propafenone [Rythmex; Knoll Pharmaceuticals, Delkenheim, Germany] and procainamide [Pronestyl; E. R. Squibb & Sons, New Brunswick, New Jersey]).^59- 60 The study of propafenone used a pre/post design of 114 patients with atrial fibrillation receiving stable doses of brand-name propafenone for at least 18 months who were required by their insurer to switch to a generic version of the drug. This study, which included no concurrent controls, found no differences in rates of health system utilization such as clinic visits, coprescription with other medications, or rates of cardioversion in the 18 months after switching to a generic drug and a slight reduction in emergency department visits with the generic version (P < .01).⁵⁹ Procainamide was studied in a bioequivalency study of patients with ventricular dysrhythmias; no differences in telemetry output were found between the generic and brand-name versions.⁶⁰

The remaining 11 articles studied warfarin (Coumadin).^61- 71 In 6 RCTs or prospective studies, generic and brand-name warfarin performed similarly with respect to clinical end points such as INR, frequency of adverse events, and number of required dose adjustments.^{61- 62,64,68- 70} Five retrospective observational studies evaluated patient INRs and clinical outcomes in patients who were required to switch from Coumadin to warfarin because of changes in coverage in diverse settings: nationwide in Israel, a Canadian province, a staff model health maintenance organization (HMO), a commercial HMO, and a municipal hospital in the United States. All of these studies used pre/post designs and found results similar to the RCTs; no significant differences were seen in clinical outcomes, including hemorrhagic adverse events or thromboembolic disease.^63,65- 67 One of the cohort studies found a small but significant decrease in INR in patients using the generic drug, although it did not translate into differences in morbidity or mortality.⁶⁶ A fourth retrospective cohort study found increased health care system utilization in patients not taking Coumadin (although no differences in morbidity/mortality), but the drug used as a comparator in that study was not rated as bioequivalent by the FDA.⁷¹

Data from 30 studies contributed to the effect sizes of the outcomes. As seen in Figure 2, when data were pooled by drug class, in each case, the 95% CI crossed zero, and the effect size was “very small” (except for statins and antiplatelet agents, where the effect size was “small”). The aggregate effect size (n = 837) was −0.03 (95% CI, −0.15 to 0.08), which indicates nearly complete overlap of the generic and brand-name distributions. These data suggest no evidence of superiority of brand-name to generic drugs in measured clinical outcomes among these studies.

Forty-three editorials and commentaries met our criteria during the study period. The greatest number (19, 44%) were published from 1993 to 1999^9,72- 89 while 14 (33%) were published from 2000 to 2008.^90- 103 Twenty-five (58%) discussed cardiovascular and generic drugs broadly^{72,76- 77,80- 84,86- 87,90,93- 95,97,101- 110} while 18 (42%) focused only on cardiovascular NTI drugs.^{9,73- 75,78- 79,85,88- 89,91- 92,96,98- 100,111- 113}

Of these editorials, 23 (53%) expressed a negative view of the interchangeability of generic drugs compared with 12 (28%) that encouraged substitution of generic drugs (the remaining 8 did not reach a conclusion on interchangeability). Among editorials addressing NTI drugs specifically, 12 (67%) expressed a negative view while only 4 (22%) supported generic drug substitution.

To our knowledge, our analysis is the first comprehensive review of the empirical evidence comparing clinical characteristics of generic and brand-name drugs used in cardiovascular disease. The 47 studies in our sample covered 8 different subclasses of cardiovascular drugs, including 2 types of NTI drugs. Measured clinical outcomes included vital signs; clinical laboratory values such as INR and urine electrolytes; adverse effects or other morbidity; and health care system utilization, including clinic and emergency department visits.

The studies in our sample concluded that generic and brand-name cardiovascular drugs are similar in nearly all clinical outcomes. Among WTI drugs, the best evidence for clinical equivalence emerged from high-quality prospective RCTs in patients with cardiovascular disease involving β-blockers, calcium channel blockers, and statins. Fewer trials compared generic and brand-name diuretics, antiplatelet agents, ACE inhibitors, and α-blockers, limiting our ability to reach similar conclusions in these drug classes.

Among NTI drugs, warfarin was the subject of the most studies addressing therapeutic equivalence. The 6 studies with a prospective design (461 patients) demonstrated similar clinical outcomes with brand-name and generic versions of the drug for multiple different outcomes, including INR, required dose adjustments, and adverse events. Among the retrospective reviews, 2 revealed transient differences in INR after changes from brand-name to generic warfarin without any differences in clinical outcomes. The only study showing specific differences in use of health care resources compared Coumadin with a version of warfarin that was not rated as bioequivalent by the FDA. Taken as a whole, these results suggest that switching from brand-name to generic warfarin products rated as bioequivalent by the FDA is safe, although it may be useful to monitor the INR of higher-risk patients more closely during a switch period.

Even though there is little evidence of important clinical differences between generic and brand-name drugs in cardiovascular disease, many editorials expressed a negative view of generic drug interchangeability and urged heightened concern on the part of physicians and patients. This opinion has not changed substantially over time; among the most recent editorials (published 2000-2008), 6 of 14 (43%) expressed a negative view of substitution. One explanation for this discordance between the data and editorial opinion is that commentaries may be more likely to highlight physicians' concerns based on anecdotal experience or other nonclinical trial settings. Another possible explanation is that the conclusions may be skewed by financial relationships of editorialists with brand-name pharmaceutical companies, which are not always disclosed.¹¹⁴ Approximately half of the trials in our sample (23/47, 49%), and nearly all of the editorials and commentaries, did not identify sources of funding.

Our study has several limitations that reflect the underlying literature. The majority of the studies we identified were bioequivalence studies, which included small populations and were powered to assess differences in pharmacokinetic parameters rather than clinical outcomes. For the smaller studies, only large differences in clinical outcomes would have been statistically significant, although our meta-analysis addresses the limitation of small sample size by pooling results across studies. Most clinical outcomes were evaluated by testing a superiority hypothesis rather than noninferiority hypothesis. Statistical insignificance in the context of a superiority study does not allow one to conclude that agents are equivalent, only that there is insufficient evidence available to conclude that the agents are different. In addition, many of the bioequivalence studies included disproportionately young and healthy subjects, and there were limited data comparing generic and brand-name medications in patients with multiple morbidities and taking numerous medications. Such patients may be at greater risk of adverse events if modest clinical differences in medication formulations exist.

Most of the studies were conducted in 4 medication classes: β-blockers, calcium channel blockers, diuretics, and warfarin. The small numbers of studies in other classes limited our ability to draw class-specific conclusions about comparative safety or efficacy. Finally, most studies were short-term evaluations and did not collect the data necessary to compare long-term outcomes associated with generic drug use such as rates of myocardial infarction or death. The lack of studies evaluating clinical outcomes in generic drug use is not altogether surprising, as neither generic drug makers nor brand-name manufacturers are likely to make large financial investments over many years to pursue a research initiative that could adversely affect their business model if their hypotheses are not confirmed.

Despite these limitations, we identified numerous studies that evaluated differences in clinical outcomes with generic and brand-name medications. Our results suggest that it is reasonable for physicians and patients to rely on FDA bioequivalence rating as a proxy for clinical equivalence among a number of important cardiovascular drugs, even in higher-risk contexts such as the NTI drug warfarin. These findings also support the use of formulary designs aimed at stimulating appropriate generic drug use. To limit unfounded distrust of generic medications, popular media and scientific journals could choose to be more selective about publishing perspective pieces based on anecdotal evidence of diminished clinical efficacy or greater risk of adverse effects with generic medications. Such publications may enhance barriers to appropriate generic drug use that increase unnecessary spending without improving clinical outcomes.