When Do We Decide That Generic and Brand-Name Drugs Are Clinically Equivalent?
Perfecting Decisions With Imperfect Evidence
The pharmaceutical market accounts for over (US) $1 trillion dollars in revenue per year worldwide, over one third of which is generated from sales of generic drugs.1 Available evidence suggests that on average, 48% of the total volume of this market among Organization for Economic Cooperation and Development nations comprised generic medications.2 Specifically in the United States, generic drugs comprise 84% of the total volume and 28% of the total value of the pharmaceutical market. However, significant variations in the uptake and costs of generic prescriptions are observed, which are partially attributable to differences in formulary management (pharmacy benefit managers), as well as variations in regulatory, substitution, and pricing policies between countries. Although manufactures must prove that generic analogues are bioequivalent to their brand-name comparators, the validity of generic drug substitution policies and their investment returns rest on the assumption that such drugs are clinically interchangeable. The extent to which such assumptions have been tested and proven remains unclear.
See Article by Leclerc et al
It is with this context that we read with interest the article by Leclerc et al3 in this month’s issue of Circulation: Cardiovascular Quality and Outcomes. This study used the Quebec Integrated Chronic Disease Surveillance System to examine the monthly rates of adverse events among 136 177 losartan, valsartan, and candesartan users aged 66 years and beyond, during the 24 months before and 12 months after the commercialization of these generic agents. The authors tracked each patient over time and identified the date at which the patient switched from the brand-name to the generic drug (ie, defined as the commercialization date). Using interrupted time-series analyses, the authors compared monthly adverse event rates (as defined by emergency room visits and hospitalizations) between generics and brands, before and after the commercialization date.
The authors determined that generic losartan, valsartan, and candesartan were associated with a 7.5%, 17.1%, and 16.6% higher rate of adverse events, respectively, in the period of time that followed generic drug commercialization as compared with their corresponding brands (with statistical significance attained for the valsartan and candesartan). Although the adverse event rates remained higher with generic analogs throughout the follow-up period, the dramatic differences in adverse events between generics and brands were ostensibly confined to the 1-month period that followed commercialization. The absolute differences in adverse event rates associated with generic analogues were most pronounced for candesartan, where differences in bioavailability between the generic and brand comparators were greatest.
As with all observational research designs, the study by Leclerc et al3 had both strengths and weaknesses. The study’s strengths included the sample size, the ability to track the actual date at which the patient first received their generic prescription, the employment of several sensitivity analyses (eg, with and without the use of continuous multiple interval measures; high versus low comorbid disease burden), and the fact that hospitalizations more specifically comprised cardiovascular than noncardiovascular outcomes, making biological plausibility more likely.
However, several important limitations also existed. For instance, the study design was ecological and accordingly could not incorporate risk-adjustment methodology. Clinical detail was lacking, which could not identify the reasons for the higher number of emergency room visits and hospitalizations associated with generic drugs. The adverse events associated with generic analogues may have been neither avoidable nor attributable to the generic drugs themselves.
Although others have observed clinical differences between generic and brand-name medications for selective conditions,4,5 the findings by Leclerc et al3 stand in contrast to other studies examining the clinical interchangeability between generic and brand-name drugs for cardiovascular disease. For example, 1 recent meta-analysis, consisting of 74 randomized trials including several which compared generic with brand-name angiotensin-converting enzyme inhibitors, demonstrated no significant differences in adverse event rates between the 2.6 Other observational studies, which have either incorporated propensity-matching methodologies or time-series analyses, have similarly supported clinical interchangeability between generic and brand-name drugs.5,7 In this regard, Leclerc et al3 challenge the assumption that generic and brand-name drugs are clinically equivalent.
Some may view the results of the study of Leclerc et al3 as being the exception rather than the rule; some generics may not be as interchangeable as others. After all, not every generic drug has been compared against its corresponding brand-name counterpart. To the best of our knowledge, the study by Leclerc et al3 is the first to have compared the adverse event rates between generic and brand-name losartan, valsartan, and candesartan.
If we were to assume that these 3 drugs were indeed clinically inferior to their brand-name counterparts, the potential clinical, policy, and economic implications would be nontrivial. For example, the absolute differences in adverse event rates observed by Leclerc et al3 were estimated to range between 15 and 40 additional emergency room visits or hospitalizations per 1000 patients with the use of generic versus brand-name angiotensin II inhibitors—an absolute rate that would comprise ≈5400 additional adverse events among the 135 000 patients in the study. Depending on the costs of each emergency room visit and hospital encounter, the expenditures associated with such adverse events could be high. With differences in acquisition cost of [CAD] $1.00 or more per day between the generic and brand-name drugs examined,8 one would estimate an additional [CAD] $4 million per month in drug expenditures assuming all such patients were to switch from their generic analogues back to brand-name losartan, valsartan, or candesartan. Consolidating the risk–benefit trade-offs may be challenging even for the best of decision makers. Even if the incremental higher risk of adverse events was to have been entirely attributable to the use of generic drug preparations, quantifying the cost-effectiveness and budgetary impact may still be difficult, given that costs, utilization, and outcomes may vary across regions.
More likely than not, the incremental risks observed by Leclerc et al3 were not entirely attributable to the use of generic drug preparations. Although most would agree that estimations like numbers needed to harm may be premature and potentially misleading, what remains less clear is how we approach the analytic designs of future research. The extent to which the analytic methods used by Leclerc et al3 help us or hinder us underscores the issue at hand. For example, Leclerc et al3 viewed their use of segmentation analyses as a methodological strength, given their ability to distinguish the use of the generic analogue from the brand-name comparator during the postcommercialization transition period. Others, however, may not view such methodological approaches as favorably, given that systematic differences may have existed that explained why some patients received their generic drugs earlier than others. Simply put, the exposure variable (generic versus brand-name drug) was a nonrandom event. One might hypothesize that physician practice patterns and prescribing behaviors dictated the selection of who received generic drugs and who did not—such selection may have been preference based, whereas others may have been based on clinical factors, affordability factors, or patient preferences themselves. For example, socioeconomically disadvantaged patients or the elderly might get changed to generic first because of coverage on government drug plans and formulary requirements, so this might cause systematic bias. It is also possible that the pre-/postcommercialization time periods were not comparable. Perhaps a different patient pool of new users to the angiotensin II inhibitor class may have followed the generic commercialization period. This might have occurred if, for instance, regulatory bodies imposed explicit utilization or limited-use criteria as a cost-containment strategy before the generic commercialization period. If post commercialization, new users to angiotensin II inhibitors were at higher clinical risk, then those who comprised the angiotensin II inhibitor pool during the patent-protected pregeneric commercialization period and then pre-/postcommercialization comparisons themselves may not have been valid, thereby undermining the use of time-series analyses as a whole. Although admittedly speculative, it is plausible that the segmentation analysis and the differentiation of who used what and when may have introduced selection biases that only further obscured the truth.
Leclerc et al3 are, therefore, correct in advising caution when interpreting their findings. They advocate instead, for further research. However, what sort of future research also remains unclear. The methodological techniques the authors themselves used and promoted underscores uncertainty about the optimal research methods that may be most appropriate when conducting drug surveillance research. Their study, therefore, raises more questions than answers with regard to appropriate next steps:
– Do we have consensus on the methodological designs that should be used?
– When should results from observational research warrant more sophisticated research designs, such as randomized clinical trials?
– What would be considered a minimally clinically important difference that would inform the sample size of such clinical trial designs if applicable?
– What consistency in evidence are we looking for to reassure ourselves that clinical equivalence has been achieved?
– Do brand–generic clinical equivalence vary by drug/drug class and disease?
– Where clinical equivalence remains in question; how do we make regulatory and policy decisions on generic substitution and pricing policies?
These are just some of many questions we are to answer—and this for only 3 drugs examined within 1 class. Yet, we as a community of clinicians, researchers, and policymakers are expected to respond in kind, now that generic–brand clinical equivalence has been challenged.
Pharmaceutical surveillance strategies and data registries are put in place to help answer questions about comparative effectiveness and safety. Accordingly, the evaluation of a drug does not cease on the date of patent expiration. Leclerc et al3 should be commended in their efforts to elucidate the truth when uncertainty exists. In so doing, we gain a better appreciation as to why clinical and scientific thought leaders may remain skeptical about the interchangeability between generics and brand-name drugs even when faced with research data that suggest otherwise.9 Clinical trials so too have their limitations. For example, available evidence suggests that <50% of registered protocols comparing brand-name to generic drugs have ever disseminated their results. Among those that have, most such protocols were inbred trials funded by generic companies themselves, which generally favored interchangeability.10 Nonprofit funding and arm’s length evaluation are clearly needed. But the action plans remain unclear.
If we are to continue to promote drug surveillance research, we may be leaving ourselves vulnerable to having to address clinical uncertainty when faced with imperfect evidence. Solutions will almost certainly necessitate much broader and more comprehensive collaboration between clinical guideline developers, regulatory bodies, policy makers, and the scientific community—perhaps even a conceptual framework to help us map out our next steps, our priorities, and our action plans may be useful. The investment in time, effort, and resources may not be small but may yield a return on investment that approaches or surpasses those associated with generic drugs themselves.
Sources of Funding
Dr Alter was funded by a Chair in Cardiovascular and Metabolic Rehabilitation, University Health Network-Toronto Rehabilitation Institute, University of Toronto. The Institute for Clinical Evaluative Sciences is supported by a grant from the Ministry of Health and Long Term Care.
The opinions expressed in this article are not necessarily those of the editors or of the American Heart Association.
The opinions reported in this article are those of the author and are independent from ICES and the UHN-TRI. No endorsement by ICES, the Ontario MOHLTC, or UHN-TRI is intended or should be inferred.
- © 2017 American Heart Association, Inc.
- 1.↵International Federation of Pharmaceutical Manufacturers and Associations. The Pharmaceutical Industry and Global Health. Facts and Figures 2017. https://www.ifpma.org/wp-content/uploads/2017/02/IFPMA-Facts-And-Figures-2017.pdf. Chapter 3; page 41. Accessed August 31, 2017.
- 2.↵OECD. Health at a Glance 2015: OECD Indicators. Paris, France: OECD Publishing; 2015. http://dx.doi.org/10.1787/health_glance-2015-en.
- Leclerc J,
- Blais C,
- Rochette L,
- Hamel D,
- Guénette L,,
- Poirier P
- Hsu CW,
- Lee SY,
- Wang LJ
- Hellström J,
- Rudholm N
- Manzoli L,
- Flacco ME,
- Boccia S,
- D’Andrea E,
- Panic N,
- Marzuillo C,
- Siliquini R,
- Ricciardi W,
- Villari P,
- Ioannidis JP
- Jackevicius CA,
- Tu JV,
- Krumholz HM,
- Austin PC,
- Ross JS,
- Stukel TA,
- Koh M,
- Chong A,
- Ko DT
- 8.↵Ontario Drug Benefit Program, Ministry of Health and Long Term Care of Ontario. Ontario Drug Benefit Formulary/Comparative Drug Index. https://www.formulary.health.gov.on.ca/formulary/. Accessed August 31, 2017.
- Flacco ME,
- Manzoli L,
- Boccia S,
- Puggina A,
- Rosso A,
- Marzuillo C,
- Scaioli G,
- Gualano MR,
- Ricciardi W,
- Villari P,
- Ioannidis JP