(See the Major Article by Yahav et al on pages 1968–74.) In this issue of Clinical Infectious Diseases, Yahav and colleagues examine data from clinical trials supporting marketing approvals of new antimicrobials by the US Food and Drug Administration (FDA) [1]. In summary, these trials do not evaluate whether new drugs improve efficacy in patients who lack effective options, and they exclude patients most in need, often do not enroll sufficient numbers of participants, and often evaluate outcomes that are not direct measures of patient benefit. Postmarketing studies largely were not performed or enrolled insufficient numbers of participants. While none of the drugs received “accelerated approval,” which requires follow-up studies to confirm clinical benefits on surrogate endpoints such as microbiological results, many antibiotics are approved based on surrogate endpoints whose relationship to mortality/symptoms remains unclear [2]. The new drugs often cost more than older effective agents. There are several obvious questions: How did this disconnect occur between marketing approval data and evidence patients and clinicians need to justify use and payment? What are the consequences of this lack of evidence? What should be done to get the answers to questions that are needed clinically? How we got here involves scientific and regulatory issues. Regulators are under pressure to address the issues of developing therapeutics for infectious diseases. Therefore, regulators have accepted this data, seemingly prioritizing “feasibility” of trials, incentivizing drug developers, and drug availability over clinical applicability of trial design and results. The belief seems to be that new drugs might have superior efficacy for unstudied future patients who lack effective options, assuming that in vitro biological activity against resistant pathogens directly translates into added benefits for sicker populations. Congressional incentives for infectious diseases interventions lack requirements for added benefit, differing from all other therapeutic areas, which require added benefits over available therapies. The Centers for Medicare and Medicaid Services has recently removed the added benefit requirement for new antibiotics to qualify for increased payments [3]. The end result is the type of evidence presented by Yahav et al (Table 1). Current Model of Drug Development in Infectious Diseases and Current Incentives Current Model of Drug Development in Infectious Diseases and Current Incentives The consequences have been increased quantities and cost of drugs approved for a limited number of diseases (eg, skin and urinary tract infections), realizing that these drugs will be used in other diseases and types of patients [4]. These drugs are described as “new options [5], but evidence of added clinical benefits for patients is lacking [1, 6]. People will not spend more for products without demonstrated added benefits regardless of assumptions about their future potential. It is not surprising that companies manufacturing such drugs have declared bankruptcy. The notion that “many hospital pharmacies will dispense cheaper generics even when a newer drug is far superior” [7] is inconsistent with evidence showing lack of superiority of new drugs on patient-centered outcomes [1, 6]. Another byproduct is lack of venture capital firms’ investment, painting the entire field with the same brush even when sponsors attempt to evaluate superiority of their interventions. Is this current methodology for developing interventions scientifically sound and clinically rational? All studies used noninferiority (NI) hypotheses. NI hypotheses allow for lesser efficacy compared to older effective drugs, among patients in whom the older drug is effective, in trade-off for superiority on other benefits such as fewer adverse effects where loss of efficacy does not result in irreversible harm for patients [8]. They do not evaluate whether new drugs have superior efficacy. Yet the clinical need is for improved efficacy in patients without effective options. Furthermore, NI trials exclude patients with disease resistant in vitro to comparators or who have received prior therapy to avoid bias, but these exclusions are unnecessary in superiority studies enrolling participants without effective options. Last, noninferiority trials raise important ethical issues. Even research that is scientifically valid in design must be valuable to patients in order to be ethical [9]. Noninferiority studies often leave unanswered basic questions of whether the new intervention is better or worse in efficacy. NI trials also raise issues regarding respect for persons, justice, and beneficence in clinical research as they potentially expose patients who have effective options to increased harm for unclear benefits to themselves [8, 10]. Evidence on what amount of loss of effectiveness and what trade-offs patients are willing to accept often is missing [11]. Would patients enroll in a trial in which 1 in 5 persons might do worse compared to available standard of care (a 20% loss of effect used in some trials) if they understood the study’s purpose? Potential superior efficacy results from NI trials do not justify their hypotheses post hoc as trial results obviously are not available at the time of consent [12]. Trials with no hypothesis or appropriate inferential statistical testing raise questions of scientific validity and the ethics of underpowered trials, and also leave clinically relevant questions unanswered [9, 13]. Trials that cannot answer clinically relevant questions are a waste of resources, result in low-value care, and squander the goodwill of research volunteers. While developing drugs for the future is laudable, this cannot be done at the expense of current patients. Patients who are sick now are not concerned about ‘insurance policies’ for the future. Experienced clinicians know drugs have different effects in different types of patients; therefore, it cannot be assumed that noninferior drugs in less sick patients will be superior in sicker/immunocompromised patients. Drugs with promising in vitro activity routinely fail to fulfill their promise for patients [1, 6, 8]. Conversely, superiority has been shown even when in vitro susceptibility to both drugs is present [14]. Achieving the goal of helping patients live longer and better will require realigning research from questions focused solely on “pathogens” and refocusing on patients and improving their outcomes [15] (see Table 2). This has implications for research questions asked, types of patients enrolled, types of interventions evaluated, and the outcomes measured in clinical trials as well as providing evidence justifying reimbursement. The goal is to help patients regardless of whether they have disease due to susceptible or resistant organisms. In absolute numbers, more patients die with susceptible than resistant disease. For instance, 1.1 million persons die annually with drug-susceptible tuberculosis, compared to 100 000 with drug-resistant tuberculosis [16]. Studies that focus on interventions benefiting patients where current therapies are ineffective, regardless of an organism’s in vitro resistance profile (which is often unknown at presentation), allow for enrolling more patients and have greater public health impact. This means enrolling patients who lack effective options who have concomitant illness, severe disease on presentation, and immunocompromise. Randomization distributes these factors between the study groups to protect from selection bias and confounding. This would make trials better reflect real-world practice. Rapid point-of-care diagnostics evaluating host as well as microbiological factors could help select patients who might benefit from interventions and avoid prescribing to those who do not. Types of interventions should be expanded beyond antimicrobials to include host-directed therapies, as dysregulated immune responses may factor into treatment failure as is being seen and studied in the current coronavirus pandemic. Outcomes measured should be direct measures of all-cause mortality or symptoms/function using appropriately developed patient-reported outcome tools instead of solely focusing on clinician reported outcomes like hospitalization and prescribing interventions which are indirect measures of patient benefit [17]. Surrogate endpoints often are not needed in acute potentially fatal diseases [2]. Superiority trials often need fewer participants to evaluate if new interventions are better if they are highly effective. Large safety databases are not needed if interventions improve survival, as patients may be willing to trade off increased adverse effects for decreased mortality. Making Progress in Developing Better Interventions for Patients Making Progress in Developing Better Interventions for Patients Incentives should be realigned so the primary goal of enabling drug development is improved patient outcomes, focusing on quality of interventions, not just quantity. Incentives should support development of interventions with demonstrated, rather than assumed, added benefits for patients. Following the model for human immunodeficiency virus (HIV), first interventions with demonstrated improved effectiveness are needed from superiority trials, then interventions with better adverse effect profiles and better convenience, evidence of which might come from validly designed NI trials in appropriate settings. But even in HIV, patients still struggle with affordability. Improving patient outcomes must include containing drug costs for all patients, not shifting costs to patients who might not have infections. Consumers should receive something of value in return for their investments of time, effort, and money. Obtaining better evidence is a necessity for informed patient care, as well as using resources wisely as healthcare costs increase unsustainably. Clinicians cannot practice in an evidentiary vacuum based on assumptions. We need better, not just more research [18]. Patients and clinicians should not have to choose between no drugs and drugs incompletely evaluated to answer clinically relevant questions. Innovation should mean better, not just new. As Einstein said, the definition of insanity is doing the same thing over and over again and expecting different results. Yahav et al have shown that current methods of development and incentives fail to answer clinically relevant questions, and therefore do not justify increased costs. It will take time and effort to answer these questions, with basic science formulating hypotheses that can be evaluated in adequate and well-controlled studies in and for patients. New classes or mechanisms of action should translate into added patient benefits. We need to work together to support and hold accountable regulators, funders, researchers, and drug sponsors and advocate for better incentives and affordability from legislators. Settings of public health urgency are when valid science serves us best, as in recent trials in Ebola where asking the right questions in the right patients resulted in valid scientific evidence to help patients live longer and live better [19, 20]. Potential conflicts of interest. The author has received consulting fees from Arrevus, Corbus, DaVolterra, Eicos, Eli Lilly, Fuji, Gilead, GlaxoSmithKline, Johnson & Johnson, Microbion, Otsuka, Romark, and Shinogi, outside the submitted work. The author has submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
No takes yet. Share an insight, caveat, or question.
John H. Powers (2020) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: