How to Evaluate New Drugs: An Evidence-Based Reading of Coronavirus Oral Antivirals, Weight-Loss Drugs, and Alzheimer’s Antibodies

How to Evaluate New Drugs: Efficacy, Risk, and Scientific Progress

The conclusion in one sentence

A new drug should not be judged solely by whether it is “effective” or “approved.” Decisions must integrate the eligible population, clinical benefit, safety risks, alternatives, and continually evolving evidence. PMID: 42060802

The central question is not how new a drug is, but what its evidence answers

Pre-approval trials cannot identify every rare, delayed, or population-specific adverse effect. Post-approval safety surveillance is therefore not merely a patch for gaps; it is part of the life cycle of drug evidence. A new signal emerging after marketing does not by itself mean that someone previously overlooked or concealed a problem, but neither does it rule out the need to investigate individual events. PMID: 42060802、PMID: 41033707

Professor Lin has long emphasized balancing benefits and risks, an approach consistent with modern drug decision-making. In clinical practice, disease severity, available alternatives, patient preferences, and uncertainty in the evidence must also be considered. There is no single answer for everyone that can be obtained simply by subtracting one column of numbers from another. PMID: 42359102、PMID: 41796415

Coronavirus oral antivirals: cross-trial figures are not a head-to-head contest

Nirmatrelvir inhibits the SARS-CoV-2 main protease, while ritonavir increases nirmatrelvir exposure by inhibiting CYP3A4. This combination provides antiviral benefit, but it also creates drug-interaction concerns that require a medication-by-medication review. PMID: 38177893

In the pivotal Paxlovid trial, the risk of hospitalization or death was reduced by approximately 89% among unvaccinated, nonhospitalized adults at high risk of severe disease who received early treatment. This finding cannot be applied without qualification to every infected person. PMID: 35172054

The full analysis of the Molnupiravir trial found an approximately 30% relative reduction in hospitalization or death among unvaccinated, high-risk outpatients. The magnitudes of the effect estimates in the original trials clearly differed, but the drugs were not compared directly head to head, and differences in population, time, and setting must also be considered. The figures 89% and 30% therefore cannot be treated as a precise ratio of drug efficacy. PMID: 35172054、PMID: 34914868

The active metabolite of Molnupiravir disrupts viral replication through lethal mutagenesis, and cellular and mechanistic studies have also raised concerns about host mutagenesis. These studies, however, are insufficient to quantify the actual risk of human genetic mutations from a short course in typical adults. The most accurate formulation is that “there is a mechanistic warning, but the human risk still cannot be established precisely from current data,” rather than presenting an unknown as conclusively safe or dangerous. PMID: 33961695、PMID: 37953355

Viral or symptomatic rebound can occur after treatment, but it is not unique to people who take Paxlovid and cannot be equated directly with drug resistance. Later evidence indicates that post-treatment resistance mutations are not impossible, but they are rare. This is consistent with Professor Lin’s cautious description at the time: “rebound exists, but its cause remains uncertain.” PMID: 38814880、PMID: 40592258

Weight-loss drugs: substantial weight effects do not erase adverse effects or eligibility criteria

Among adults with overweight or obesity who did not have diabetes, weekly semaglutide combined with lifestyle intervention produced substantial mean weight loss. Common adverse effects included gastrointestinal events such as nausea, diarrhea, vomiting, and constipation. Efficacy figures should be cited together with the trial population, treatment duration, and lifestyle intervention, rather than leaving only an impressive percentage of weight loss. PMID: 33567185

Trials of Tirzepatide likewise showed marked weight reduction and improvement in several cardiometabolic measures. The most common adverse events were mainly mild-to-moderate gastrointestinal symptoms during dose escalation. Whether pancreatitis and cholecystitis are “slightly more common” remains unsettled across the available pooled evidence. Serious events are rare, so conclusions should retain the language of surveillance and uncertainty rather than becoming absolute. PMID: 35658024、PMID: 41927408、PMID: 41884101

Alzheimer’s antibodies: clearing plaques is not the same as reversing disease

Donanemab can remove amyloid plaques effectively, and a subsequent phase III trial showed some slowing of clinical progression in patients with early symptomatic disease. The broad claim in earlier articles that “all secondary functional outcomes other than plaque removal were ineffective” is therefore no longer appropriate after the literature update. This is a refinement of the evidentiary scope through scientific progress, not a rewriting of Professor Lin’s original article or historical position. PMID: 37459141、PMID: 42128444

At the same time, efficacy cannot be discussed separately from ARIA and serious treatment-related events, nor can these findings be generalized to mean that the antibodies are safe and effective for every person with Alzheimer’s disease. Eligibility, imaging surveillance, and risk stratification are indispensable conditions for understanding this drug class. PMID: 37459141、PMID: 40063015

Lecanemab also lowers the brain amyloid-β burden and reduced the degree of cognitive and functional decline relative to placebo in patients with early disease. Later pooled evidence rated the overall clinical benefit as small and confirmed an increased risk of ARIA. In other words, whether a biomarker changes, whether a clinical effect exists, and whether that effect is sufficiently meaningful are three separate questions. PMID: 38095619、PMID: 41985900

Antimicrobial resistance and other new drugs: preclinical promise is not human efficacy

Antimicrobial resistance is indeed a major contributor to the global burden of death. Newer systematic analyses, however, have updated the estimation framework used in earlier reports, so old projections should not be treated as facts that remain unchanged today. PMID: 35065702、PMID: 39299261

Brazilian pepper-tree extracts interfere with quorum sensing without inhibiting the growth of Staphylococcus aureus and reduce MRSA skin necrosis in mouse models. These findings support an “antivirulence” research strategy, but they do not prove that the approach can never promote resistance, much less establish it as a treatment for humans. PMID: 28186134、PMID: 32415287

A triply modified vancomycin derivative showed potent, multimodal preclinical activity against vancomycin-resistant enterococci. The available abstract, however, is insufficient to verify fully the potency multiples that have been widely repeated, and it provides no evidence of clinical efficacy in humans. This case illustrates why a news report describing a “breakthrough” should first be classified according to whether the research was conducted in a test tube, in animals, or in patients. PMID: 28559345

Other examples follow the same principle. Randomized trials of lifitegrast, a newer treatment for dry eye disease, support improvement in symptoms. By contrast, the older claim that cyclosporine only increases tear production and does not improve symptoms at all has been revised by later, low-certainty evidence suggesting a possible symptomatic benefit. PMID: 30844466、PMID: 42464966

A practical sequence for readers evaluating drug news

When you encounter news about a new drug, first ask whether the study participants resemble you. Next, determine whether the endpoint was a biomarker, symptom, function, hospitalization, or death. Then identify whether the comparator was placebo, standard treatment, or another drug; only after that should you examine the effect size. At the same time, check common and serious adverse effects, drug interactions, contraindications, and follow-up requirements. For conditions such as suspected giant cell arteritis, which can rapidly cause blindness, evidence supports immediate emergency treatment with high-dose corticosteroids. Tocilizumab can improve sustained remission and reduce cumulative corticosteroid exposure, showing that “when treatment must begin” and “how long-term harm can be reduced” are questions at different levels. PMID: 36788362、PMID: 28745999

Truly reliable information about new drugs neither reports only good news nor merely lists risks. It states clearly where the evidence applies, what remains unknown, and how subsequent research has changed the interpretation.