Skip to content

CPsolvers fam, I’m excited to introduce you to our newest schema (and my first!) – an approach to “Antibiotic Failure.”  

To make our learning stick, try to remember a patient encounter in which you asked, “why aren’t they getting better on antibiotics?” Here’s a case I saw last month where this very question came up.

Case: Mrs. G is a 66-year-old woman with a past medical history of insulin-dependent diabetes and peripheral artery disease admitted from clinic with concern for left ankle and foot cellulitis. She’s had left ankle and  foot redness, swelling, and pain for the past month and a painful callus on her third toe. She has been evaluated three times over this period, and each time she was diagnosed with cellulitis. This didn’t improve despite two 7-day courses of oral antibiotics. What’s going on? Why isn’t she getting better? 


Dx Issues: 

The first step in investigating potential “antibiotic failure” is to reevaluate our working diagnosis. Let’s split this category initially into two subgroups: non-infectious and “wrong bug.”  


The signs and symptoms that lead us towards diagnosing an infection are often non-specific, so it’s important that we consider mimics. Of course, when inflammation is present, autoimmune diseases and cancer must be considered. In addition, specific clinical scenarios require other considerations. 

In Episode 9, when considering lower extremity cellulitis, Dr. Dhaliwal reminded us that nearly 30% of  patients hospitalized for “cellulitis” were later found to have a different diagnosis. The most common alternative diagnosis in this study was venous stasis dermatitis followed by DVT and contact dermatitis. He also reminded us that, with few exceptions, one should almost never diagnose bilateral cellulitis. 

Mrs. G’s case: The main non infectious causes we considered in Mrs. G’s case were DVT, venous stasis, lymphedema, contact dermatitis, and charcot arthropathy. 

“Wrong bug”: 

Infections are common, but sometimes we don’t get the culprit organism quite right. While typical bacterial causes are common, we still need to consider viral, fungal, parasitic, and atypical bacterial pathogens (such as mycobacteria or one of the “doxy deficiency” pathogens).  

 Mrs. G’s case: Viral, fungal, parasitic, and atypical pathogens don’t frequently cause cellulitis, making them a lower concern for Mrs. G. While no evidence of tinea pedis was found on her exam, it is important to examine the intertriginous areas of the foot for fungal disease and treat accordingly. Skin breakdown from untreated tinea serve as an entry port for skin flora such as staph and strep that most frequently cause cellulitis. 

Tx Issues:  

Next, we need to consider potential issues related to our current treatment regimen. Let’s subdivide this group into “antibiotic issues” and “tissue issues.”  

Antibiotic issues:  

It’s always worth reevaluating our initial antibiotic choice if things aren’t moving in the right direction. In the hospital, this often requires thinking about risk factors for multidrug resistant organisms (e.g.,., Pseudomonas, MRSA, ESBL). We have included  a very helpful antibiotic spectrum framework at the bottom of the page which was made by my mentor, Dr. Sam Ives.

 Beyond spectrum, it is important to consider whether our antibiotic is being dosed appropriately for our specific clinical scenario (i.e., meningitic dosing, prophylactic vs. treatment dosing). In the clinic it is especially important to make sure adherence isn’t an issue by taking time to consider frequency of dosing, side effects, costs, and patient understanding.  

 Mrs. G’s case: Lack of response could certainly be due to a resistant organism. In cellulitis this would most commonly be MRSA, and given Mrs. G’s diabetes, Pseudomonas and ESBL producing gram negative rods are possible as well. However, the month of symptoms without significant worsening make infection by such organisms unlikely. Unfortunately, she also has multiple potential barriers to being able to take her antibiotics as prescribed – she is non-english speaking, uninsured, and currently experiencing homelessness. All are important considerations here.

Tissue issues:  

Sometimes we have the right diagnosis and bug/drug combination, but the patient isn’t improving. This can either be from lack of source control or insufficient tissue penetration. Some common culprits to consider for source control are lines/hardware/fistulas, abscesses, and concomitant osteomyelitis or endocarditis. The “Infection in the Inpatient” schema can be helpful here. 

 Tissue penetration is always important to consider when treating CNS infections, given many commonly used antibiotics don’t cross the blood-brain-barriers. There are other drug specific considerations such as nitrofurantoin’s lack of efficacy in upper UTIs or daptomycin’s inactivation by surfactant in the lung. When in doubt, ask pharmacy and/or ID (sage advice in many clinical scenarios).  

 Mrs G’s Case: Could her painful callus have an underlying ulcer giving rise to osteomyelitis? Is there an abscess we couldn’t feel on exam? Is her peripheral arterial disease impairing  adequate tissue delivery of the drugs? 

 Natural Hx Issues:  

After adequately excluding diagnostic and treatment issues, our last category to consider is  whether the apparent lack of response is just reflective of the natural history of the disease. The main considerations here are whether the infection is too aggressive or if it simply too soon to see the desired treatment effect.  

 Too aggressive:  

This is an especially difficult area. Sometimes we do everything right, but the infection is too advanced, the immune system too compromised, or pathogens are too virulent for our therapy to work. These patients are often very sick, and we must rely on other supportive measures. In certain cases, specific adjunct antimicrobials are needed despite appropriate coverage. For example, we occasionally consider salvage therapy with multiple agents in staph bacteremia or adding clindamycin in toxin mediated disease like S. pyogenes bacteremia and necrotizing fasciitis. 

Mrs. G’s case: Given the time course these are unlikely. 

Too soon: 

Finally, we are left with the simple (and common) answer that we simply haven’t allowed our antibiotics enough time to exert their effect. For common infections like pneumonia, pyelonephritis, pharyngitis, and cellulitis, symptoms can continue for up to 48 to 72 hours after initiation of appropriate therapy. Lastly, it’s important to remember Jarish-Herxheimer reaction, especially with spirochetes such as syphilis and Lyme.   

 Case wrap up: A lower extremity ultrasound showed severe peripheral arterial disease but no DVT, and there were no other signs of volume overload or exposures to suggest a contact dermatitis. IV vancomycin was initiated. Ultrasound showed no abscess. Due to concern that the callus on her toes was actually an ulcer, a foot XR and then MRI was obtained revealing osteomyelitis. Vascular surgery and podiatry were consulted, and a toe amputation was performed. Post-operatively an antibiotic course was completed, and her diabetes and vascular risk factors were optimized. At her two-week follow-up her symptoms had abated.    


To download our schema PDF, click here

Check out this very helpful antibiotic spectrum pocket card.