Why Antibiotics Sometimes Fail — And How to Select the Right Treatment

When infections don’t respond to treatment, the issue isn’t always resistance — it’s often that the right antibiotic has yet to be selected. New approaches like AtbFinder aim to change that.

If your patients have ever dealt with an infection that just won’t go away despite repeated rounds of antibiotics, they know from personal experience just how frustrating and confusing it can be. But it can be equally frustrating for the provider as well.

As a microbiologist, researcher, and physician who helps other doctors manage hard-to-treat infections, I’m going to share something that is not yet widely understood in the medical community. The true crisis is not antibiotic resistance alone, but the failure of today’s antibiotic sensitivity tests to reliably identify antibiotics that actually work in patients.

This is the clinical gap AtbFinder test was developed to address: helping physicians and patients evaluate antibiotic options for difficult-to-treat, recurrent, biofilm-associated, and antibiotic-resistant infections when standard testing may not provide enough guidance.

The truth is, very few pathogens (bacteria, fungi) are completely resistant to every available antibiotic. More often, antibiotics fail because existing diagnostic tests don’t always select the antibiotics that actually work for the patient.

Over the past 20 years, I’ve heard the same question from doctors and patients, usually after multiple rounds of treatment: “Why are antibiotics not helping me?”

We tend to frame this problem as antibiotic resistance. Sure, antibiotic resistance is real and appears to be increasing. But that explanation alone doesn’t reflect what I see in clinical practice.

In reality, serious infections remain a major cause of death in the United States. According to national data from the Centers for Disease Control and Prevention (CDC), tens of thousands of people die each year from antibiotic-resistant infections, which include ESKAPE pathogens and superbugs (1). But when you look more broadly at severe infections overall, the numbers are even higher. Sepsis alone is associated with more than 350,000 deaths annually, while pneumonia and other serious infections account for hundreds of thousands more (2) (3) (4) (5).

It’s not because we lack antibiotics in this country, but because we don’t always identify the right antibiotic, or combination of antibiotics, for the specific infection and patient.

So when someone asks why their antibiotics are not working, or why the infection is not responding to antibiotics, the next step isn’t just to try another medication. A better approach would be to take a step back and ask: “What additional tests can help find the right antibiotic?”

Before we dive into antibiotic testing, it’s important to understand why certain infections are harder to treat than others.

What does it mean when your antibiotics are not working?

When treatment fails, it’s natural to assume the bacteria are resistant to the medication, and sometimes that’s absolutely the case. But often, antibiotic failure is more complicated than that.

An infection can persist even when an antibiotic appears “active” in the lab. That’s because what happens in a petri dish isn’t always reflective of what’s happening inside the body.

Treatment can fail for several reasons in patients. For example, the antibiotic may not reach the infection site at a high enough concentration, the infection may involve multiple organisms working together, or the bacteria may be protected in ways that standard testing doesn’t capture. This is why people often experience what gets labeled as:

• Chronic infection
• Persistent infection
• Recurrent infection
• Antibiotic-resistant infection
• “No option” infection

While these labels can be useful, they don’t always explain the underlying problem.

Infections are complicated

In medical training, infections are often presented as straightforward: identify the organism, match it with an antibiotic, and treat. But in reality, infections are rarely that simple.

Almost all “complicated” infections involve more than one microorganism. Different bacteria, sometimes along with fungi, can coexist at the same site, forming complex communities that behave very differently from a single organism in isolation (as you’d see it in a lab). These communities can support each other’s survival, protecting each other from antibiotics and making treatment more difficult.

Another major factor is biofilm formation. Instead of floating freely, microorganisms often attach to surfaces and produce a protective matrix. Within these biofilms, they become dramatically more tolerant to antibiotics — sometimes requiring concentrations hundreds or even thousands of times higher than what would normally be effective (6).

At the same time, not all pathogens involved in an infection are easily detectable. Some don’t grow well in standard laboratory conditions, which means part of the infection may go undetected (7).

Even when bacteria are successfully identified, they can behave differently outside of the body. Lab conditions can’t fully recreate what’s happening inside a patient, so test results don’t always match what happens in real life (8).

And finally, antibiotics don’t always distribute evenly throughout the body. A drug that works well in the bloodstream may not reach sufficient levels in your lungs, urinary tract, or soft tissues, for example. So an antibiotic can look effective on paper but still fail where it matters most.

All of these factors contribute to a simple but important reality: The infection we treat in the lab is not always the infection the patient actually has.

Why standard laboratory tests can be misleading

When an infection is not responding to antibiotics, clinicians rely on laboratory testing to guide the next step. These tests are essential, but they have limitations that are often overlooked.

Traditional phenotypic tests measure how bacteria respond to antibiotics under controlled conditions. These methods have been used for decades and remain the foundation for choosing an antibiotic.

But they typically evaluate one organism at a time, ignoring that many infections involve multiple species. They test bacteria in a free-floating state rather than within biofilms. And they interpret results based on antibiotic concentrations achievable in blood, not necessarily at the actual site of infection.

There is also a timing issue. In serious infections, waiting several days for results can mean that patients continue receiving ineffective treatment while their condition worsens.

The limitations of these tests are reflected in what’s known as the “90–60 rule.” In simple terms, even when a bacterium is labeled “susceptible,” treatment works as expected only about 90% of the time. And surprisingly, even when it’s labeled “resistant,” treatment can still work in roughly 60% of cases (9).

That’s a significant gap, and it highlights a key problem: standard laboratory categories don’t always predict what actually happens in the patient.

Molecular tests, on the other hand, offer a different approach by identifying pathogens or resistance genes more quickly. These can be valuable tools, especially in urgent situations.

But they answer a different question. They tell us what might be present, not necessarily what will respond to treatment.

Identifying a resistance gene does not always mean that resistance is active in that patient. And identifying a pathogen does not guarantee that a particular antibiotic will be effective in the patient’s body (10). So while these tests are useful, they often stop short of selecting the antibiotics that would really work for the patient.

Clinical situations where infections are not responding to antibiotics

Treating doctors and families often notice when something isn’t right. A loved one is receiving treatment, but the condition isn’t improving. The same concerns come up again and again. Some of these include:

Why is pneumonia not responding to antibiotics?

Why is sepsis not improving despite treatment?

Why does the infection keep coming back after antibiotics?

These are not rare scenarios — they’re common occurrences in hospitals, especially in patients with complex medical conditions, such as ventilator-associated pneumonia, cystic fibrosis, or in immunocompromised patients.

In these situations, simply changing antibiotics or increasing doses may not be enough. Without better insight into what is actually happening at the infection site, treatment can become a repetitious (and defeating) cycle of trial and error.

What does a better test look like?

If we want to improve health outcomes, we need to rethink how we evaluate infections.

A more useful diagnostic approach? Start by looking at the infection as a whole rather than focusing on a single “primary” or “lead” pathogen. It would consider how microorganisms interact, how they behave within biofilms, and how antibiotics perform under conditions that resemble the patient’s actual body.

It would also take into account where the infection is located. The concentration of an antibiotic in the lungs or urinary tract differs significantly from its level in the bloodstream. Testing should reflect those differences.

Another equally important factor is speed. In serious infections, waiting days for results can limit the ability to intervene effectively. Faster, more relevant data could change treatment decisions earlier in the course of illness.

Finally, a better test would evaluate a broader range of antibiotics, including combinations when needed, rather than limiting the analysis to a small predefined panel.

In other words, we need diagnostics that move closer to real-life treatments — not further away from them.

How AtbFinder selects antibiotics against antibiotic-resistant microorganisms

AtbFinder is an alternative approach to antibiotic selection, specifically in cases where standard susceptibility-guided treatment doesn’t lead to clinical resolution. This test was developed with this exact challenge in mind: how to choose the right antibiotic when standard treatment is not working. It helps identify effective antibiotics for resistant, chronic, persistent, recurrent, and hard-to-treat infections.

Unlike traditional methods, AtbFinder evaluates antibiotics against all microorganisms present at the infection site, including “primary” pathogen and supporting microorganisms, rather than isolating a single pathogen. It also incorporates biofilm conditions, recognizing that many infections exist in this protected state and maintaining the patient-mediated microbial “memory.”

Another key difference is how antibiotic exposure is modeled. Instead of relying only on blood-based assumptions, AtbFinder tests antibiotics at concentrations that reflect what is actually achievable in specific tissues, such as the lungs, urinary tract, or wounds.

The test is also designed to be both exhaustive and fast. It evaluates a large panel of antibiotics, often between 90 and 180, and delivers results within hours rather than days.

Clinical studies have shown that this method can improve antibiotic selection in difficult-to-treat infections, including pulmonary infections and chronic urinary tract infections (11) (12).

The Bottom Line

When antibiotics are not working, it’s easy to fall into a pattern of trying one medication after another. But without more information from better testing, that approach can prolong the problem rather than solve it.

A more effective step is to pause and reassess. Consider asking yourself, and your care team, the following questions:

1. Is the infection being evaluated in a way that reflects how it actually behaves in the body?
2. Are we relying on tests that may not capture the full picture?
3. What additional tests can help find the right antibiotic?

The goal is to move away from testing systems that oversimplify the biology of complex infections and toward approaches that better capture what’s happening at the site of infection.

This is exactly the gap newer technologies, such as AtbFinder, are designed to address. In fact, this approach has been recognized at a national level, including as a first-place winner in an FDA-organized competition in collaboration with Children’s National Hospital (13).

The future of infection treatment depends not only on developing new antibiotics, but on learning how to select the right existing antibiotics more accurately, more rapidly, and in ways that mirror real-life conditions.

This is the major gap that current testing often fails to close, and the gap that better diagnostics must address.

FAQ

What should I do if antibiotics did not help?

If antibiotics are not working, the next step is not simply switching medications. It’s important to reassess how the infection is being evaluated and whether additional testing can help identify the right antibiotic based on real-life infection conditions.

What can doctors do if antibiotics fail?

When antibiotics fail, clinicians should consider whether standard susceptibility testing reflects the actual infection environment. More advanced diagnostic approaches may help improve antibiotic selection, especially in complex or persistent infections.

Why is my infection not responding to antibiotics?

An infection may not respond to antibiotics for several reasons. It may involve multiple microorganisms, exist within biofilms, or require antibiotic concentrations that are not achieved at the site of infection.

How does AtbFinder help identify effective antibiotics?

AtbFinder is designed to address one of the most important questions in infection treatment: can an antibiotic, at the concentration that actually reaches the site of infection, eliminate the infection?

Instead of relying only on standard laboratory conditions, AtbFinder evaluates how antibiotics perform under conditions that more closely reflect what happens in the body. This includes testing at tissue-level antibiotic concentrations — such as those found in the lungs, urinary tract, blood, skin, or wounds — rather than relying only on blood-based assumptions.

It also assesses how antibiotics act against all microorganisms present at the infection site, including within biofilms and microbial communities, rather than testing a single isolated pathogen.

By combining these factors, AtbFinder helps improve antibiotic selection in cases where infections are resistant, chronic, persistent, or not responding to standard treatment.

How does AtbFinder differ from standard diagnostic methods?

Different diagnostic methods answer different clinical questions. The key difference is whether the method evaluates isolated organisms, genetic markers, biofilms, or the functional behavior of the full microbial community.

What is the difference between Culture + Microbroth Dilution / MIC / Disk-Diffusion Methods?

Culture-based AST, including MIC and disk-diffusion methods, identifies how an isolated primary pathogen appears to respond to an antibiotic under standardized in vitro conditions. These conditions fail to replicate the real-life infection environment.

This approach does not assess microbial communities, biofilms, polymicrobial interactions, or tissue-level antibiotic exposure. As a result, an antibiotic reported as active by MIC testing may fail clinically if the real antibiotic concentration at the infection site is lower than the concentration used in the test.

What is the purpose of molecular and genetic tests (PCR and NGS)?

PCR and NGS identify which microorganisms or antibiotic-resistance genes are present in a sample. However, this information is often insufficient to select effective therapy for antibiotic-resistant infections.

These methods do not directly show which antibiotic will work clinically. The presence of an organism or resistance gene does not always translate into phenotypic resistance or treatment failure.

Do Biofilm Inhibitory Concentration and Minimal Biofilm Eradication Concentration testing methods accurately represent real-life biofilm infections in patients?

These methods partially address biofilm biology, but only for selected primary pathogens, not for the entire microbial community.

They remain primarily research-use approaches. They are not broadly clinically validated and do not assess real-life polymicrobial communities as they exist in patient samples.

What sets AtbFinder apart from other tests?

AtbFinder is designed to identify which antibiotics or antibiotic combinations functionally suppress the microbial community under clinically relevant conditions.

Unlike standard MIC-based testing, AtbFinder evaluates microbial communities, biofilm-associated behavior, and real-life polymicrobial interactions using antibiotic concentrations achievable at the site of infection.

References

1. Centers for Disease Control and Prevention. 2019 Antibiotic Resistance Threats Report. Antimicrobial Resistance. Published May 7, 2024. https://www.cdc.gov/antimicrobial-resistance/data-research/threats/index.html
2. Rhee C, Jones TM, Hamad Y, et al. Prevalence, Underlying Causes, and Preventability of Sepsis-Associated Mortality in US Acute Care Hospitals. JAMA Network Open. 2019;2(2):e187571. doi:10.1001/jamanetworkopen.2018.7571
3. Carey MG, Valcin EK, Lent D, White M. Nursing Care for the Initial Resuscitation of Severe Sepsis Patients. Critical Care Nursing Clinics of North America. 2021;33(3):263–274. doi:10.1016/j.cnc.2021.05.003
4. La Via L, Ferlito S, Di Modica MS, et al. The Global Impact of Sepsis: Epidemiology, Recognition, Management, and Health System Challenges. Epidemiologia. 2026;7(1):20. doi:10.3390/epidemiologia7010020
5. CDC. FastStats — Pneumonia. Centers for Disease Control and Prevention. Published 2019. https://www.cdc.gov/nchs/fastats/pneumonia.htm
6. Thöming JG, Häussler S. Pseudomonas aeruginosa Is More Tolerant Under Biofilm Than Under Planktonic Growth Conditions: A Multi-Isolate Survey. Frontiers in Cellular and Infection Microbiology. 2022;12. doi:10.3389/fcimb.2022.851784
7. Tetz G, Tetz V, Vecherkovskaya M. Genomic characterization and assessment of the virulence and antibiotic resistance of the novel species Paenibacillus sp. strain VT-400, a potentially pathogenic bacterium in the oral cavity of patients with hematological malignancies. Gut Pathogens. 2016;8(1). doi:10.1186/s13099–016–0089–1
8. Tetz V, Tetz G. Novel prokaryotic system employing previously unknown nucleic acids-based receptors. Microbial Cell Factories. 2022;21(1). doi:10.1186/s12934–022–01923–0
9. Doern GV, Brecher SM. The Clinical Predictive Value (or Lack Thereof) of the Results of In Vitro Antimicrobial Susceptibility Tests. Journal of Clinical Microbiology. 2011;49(9 Supplement):S11-S14. doi:10.1128/jcm.00580–11
10. Tetz G, Tetz V. Overcoming Antibiotic Resistance with Novel Paradigms of Antibiotic Selection. Microorganisms. 2022;10(12):2383. doi:10.3390/microorganisms10122383
11. Tetz G, Kardava K, Vecherkovskaya M, et al. AtbFinder Diagnostic Test System Improves Optimal Selection of Antibiotic Therapy in Persons with Cystic Fibrosis. McAdam AJ, ed. Journal of Clinical Microbiology. 2023;61(1). doi:10.1128/jcm.01558–22
12. Tetz GV, Kardava KM, Vecherkovskaya MF, Tsifansky MD, Tetz VV. Treatment of chronic relapsing urinary tract infection with antibiotics selected by AtbFinder. Urology Case Reports. 2022;46:102312. doi:10.1016/j.eucr.2022.102312
13. COVID-19-edition of pediatric medical device competition announces finalists | Children’s National Hospital. Children’s National Hospital. Published 2020. Accessed April 13, 2026. https://www.childrensnational.org/about-us/newsroom/2020/covid-19-edition-of-pediatric-medical-device-competition-announces-finalists