What Is FICI? Calculation and Interpretation

The Fractional Inhibitory Concentration Index, or FICI, is a numerical score that tells researchers whether two antimicrobial drugs work better together than either one works alone. It is calculated by comparing how much of each drug is needed to stop microbial growth when the drugs are combined versus when each is used solo. The math is straightforward, but interpreting the result is where things get contentious, because the traditional cutoff values used to label a combination “synergistic” or “antagonistic” have been questioned for decades.

How the FICI Is Calculated

The formula itself is simple. You test each drug alone to find its minimum inhibitory concentration (MIC), which is the lowest concentration that stops the microbe from growing. Then you test the two drugs together at various concentration ratios and find the MIC of each drug in the presence of the other. The FICI adds two fractions: the MIC of drug A in combination divided by the MIC of drug A alone, plus the MIC of drug B in combination divided by the MIC of drug B alone.¹

If both drugs in combination need only half their individual MIC to do the job, each fraction equals 0.5, and the FICI comes out to 1.0. That is what you would expect if two drugs simply add their effects without helping or hindering each other. A FICI well below 1.0 means the drugs are doing something more together than you would predict from their individual potencies, which researchers call synergy. A FICI well above 1.0 suggests they are interfering with each other, which is antagonism.

The Standard Interpretation Thresholds

Most published studies use the same set of breakpoints that have been around since the 1990s. A FICI of 0.5 or lower is classified as synergy. A FICI between 0.5 and 4.0 is called indifference, meaning the drugs neither help nor hurt each other beyond simple addition. A FICI above 4.0 is labeled antagonism.¹ These same thresholds appear in studies using gradient diffusion methods like the Etest, where synergy is likewise defined as a FICI of 0.5 or lower.²

The problem is that the “indifference” zone between 0.5 and 4.0 is enormous. It swallows almost every result. When antifungal combinations were tested against Aspergillus fumigatus, the standard range of 0.5 to 4.0 for additivity classified virtually everything as having no interaction, even combinations that clearly performed differently in animal models.³ This has led some researchers to argue for much tighter cutoffs.

Why the Cutoffs Are Debated

One influential study took a clever approach to this problem. The researchers tested drugs combined with themselves, since a drug paired with itself should, by definition, produce a purely additive result with a FICI near 1.0. When amphotericin B was combined with amphotericin B, and voriconazole with voriconazole, the FICI values clustered between 0.5 and 1.25. Meanwhile, a known antagonistic combination (amphotericin B plus voriconazole) produced FICI values between 2.25 and 4.25. Using Monte Carlo simulation, the researchers found that an additivity range of 1.0 to 1.25 correctly classified self-drug combinations as additive and the antagonistic pair as antagonistic in more than 85% of simulated results. They also showed that the 95% confidence intervals for an in vivo synergistic combination fell below 1.0, and those for an in vivo antagonistic combination exceeded 1.25.³

Earlier work had already hinted at this. Blind experiments combining the antibacterials alafosfalin and cephalexin with themselves found that even “weak” interactions, with FICI values between 0.5 and 0.99, were statistically significant.⁴ In other words, combinations falling in the 0.5 to 1.0 zone that the standard thresholds dismiss as “indifferent” may genuinely be interacting in a meaningful way. The field has not reached consensus on whether to adopt tighter cutoffs, so you will see both the traditional and narrower schemes cited in recent literature, sometimes within the same journal.

The Checkerboard Assay

The checkerboard assay is the most common lab method for generating the data that feeds into a FICI calculation. It uses a standard microplate, typically 96 wells arranged in 8 rows and 12 columns. One drug is diluted in serial twofold steps across the rows, the other across the columns, so every well contains a unique concentration pair. Each well is then inoculated with the test organism and incubated, usually at 35°C for about 18 to 20 hours for bacteria.⁵ After incubation, the researcher reads which wells show growth and which do not, then picks out the combination with the lowest total drug exposure that still prevented growth. That is the data point used to calculate the FICI.

The setup is elegant in principle, but the twofold dilution scheme introduces a fundamental limitation. Because each step doubles or halves the concentration, a single dilution step difference in the MIC reading changes the FICI by a factor of two. That means the difference between a FICI of 0.5 (synergy) and 1.0 (indifference) can come down to one well on the plate being read slightly differently. This is not a trivial concern; reproducibility studies of the checkerboard method have documented multiple sources of variability, including pipetting accuracy, inoculum inconsistency, evaporation from outer wells, and reader disagreement on whether borderline wells show growth.⁶

When Checkerboard Results Disagree with Time-Kill Studies

The checkerboard assay gives you a single snapshot: after overnight incubation, did the combination inhibit growth or not? It does not tell you anything about how quickly killing happened or whether the effect was bacteriostatic (stopping growth) versus bactericidal (actually destroying cells). Time-kill studies fill that gap by sampling the culture at intervals, often every few hours over 24 to 48 hours, and counting how many organisms remain alive. The two methods do not always agree.

In one study testing tigecycline combined with thirteen other antibiotics against various bacteria, the checkerboard method classified several combinations as synergistic, but time-kill analysis showed no interaction for all but one of those same pairs.⁷ Other work has found the opposite pattern, with time-kill studies confirming the checkerboard’s call in roughly seven out of ten isolates where the checkerboard detected synergy.⁸ A third study evaluating the antifungal agent Mycograb combined with amphotericin B against Cryptococcus neoformans found that time-kill data added nuance that the checkerboard alone could not capture, revealing shifts from static to fungicidal effects at specific concentration thresholds.⁹

The takeaway is that a FICI value from a checkerboard assay is a useful screening tool, not a definitive verdict. Researchers who find a promising FICI typically follow up with time-kill curves before drawing strong conclusions, and even then the path to clinical relevance is uncertain.

Gradient Diffusion Methods and Newer Alternatives

Not every lab has the equipment or throughput for full checkerboard assays. Gradient diffusion methods, which use antibiotic-impregnated strips placed on agar plates at right angles so their concentration gradients overlap, offer a simpler workflow. When compared to the checkerboard as a reference, gradient methods agreed on the FICI classification between 60% and 77% of the time in one study of tobramycin and ceftazidime combinations against multidrug-resistant Pseudomonas aeruginosa from cystic fibrosis patients.¹⁰ That is reasonable agreement for a simpler method, but it means roughly a quarter to a third of results will be classified differently depending on which technique you use.

A newer commercial system, the MTS Synergy Application System, was evaluated across multiple laboratories and showed 96.2% concordance with the checkerboard method, even without specialized operator training at each site.¹¹ That level of reproducibility suggests the field is moving toward more standardized platforms, which would help resolve some of the inter-lab variability that has plagued checkerboard testing. On the computational side, tools like CombiANT automate image analysis of agar-based synergy tests and produce FICI values with high accuracy and precision.¹² Software packages built on Hill function modeling have also been developed for analyzing checkerboard datasets, offering statistical output beyond a single FICI number.¹³

Loewe Additivity Versus Bliss Independence

The FICI is grounded in a reference model called Loewe additivity, which assumes that a drug combined with itself should produce a perfectly predictable additive effect. Any combination that beats this prediction is synergistic; any that falls short is antagonistic. But Loewe additivity is not the only framework for defining “no interaction.” An alternative, Bliss independence, assumes the two drugs hit entirely different cellular targets and calculates expected combined effects as probabilities of independent events rather than dose substitution.

Mechanistic modeling work has shown that Bliss independence is the more appropriate reference when two drugs genuinely act on different cellular components, while Loewe additivity better describes drugs that share the same target pathway.¹⁴ The choice of reference model can change the conclusion. A large multicenter study of antifungal combinations against Candida species found strong overall concordance between the two approaches when specific cutoffs were applied: roughly 89% of combinations with a FICI below 1.0 were also classified as synergistic under Bliss, and about 81% of combinations with a FICI at or above 2.0 were Bliss antagonistic. But in the middle zone, where the FICI fell between 1.0 and 2.0, only 61% were Bliss independent, with about a third classified as Bliss antagonistic.¹⁵ For most practical purposes in clinical microbiology labs, the FICI using Loewe additivity remains the default, but researchers working with drugs that have clearly independent mechanisms should be aware that this default may misclassify the interaction.

Special Challenges with Filamentous Fungi

The FICI was originally developed and validated mostly with bacteria and yeasts, organisms that have relatively clear-cut MIC endpoints. Filamentous fungi like Aspergillus species present a different problem. These organisms do not grow uniformly in broth culture; instead of a clean transition from “growth” to “no growth,” you get partial inhibition that makes MIC endpoints ambiguous. The choice of which endpoint to use, complete inhibition versus partial inhibition, can dramatically change the result. In one study of antifungal combinations against both itraconazole-susceptible and itraconazole-resistant Aspergillus fumigatus strains, the FICI for the same drug pair varied by as much as twofold depending on which MIC endpoint was selected.¹⁶ Response surface analysis, a three-dimensional modeling approach that uses the entire dataset rather than a single point, correlated well with the FICI trend but gave more interpretable results in these difficult-to-read assays.

Why Synergy in the Lab Does Not Always Mean Synergy in the Patient

Even a clean FICI of 0.25, solidly in the synergy zone, does not guarantee that the combination will outperform monotherapy in a living organism. Translating lab results to clinical benefit has been described as a “paramount challenge.”¹⁷ The reasons are numerous. The checkerboard assay tests a single strain in nutrient broth under constant conditions, while an actual infection involves tissue barriers, immune responses, drug metabolism, and mixed microbial populations. Drug concentrations at the site of infection fluctuate with each dose and may never reach the levels used in the plate wells.

There are also biological mechanisms at play in vivo that simply cannot appear in a plastic well. When antimicrobial peptides are combined with antibiotics, for example, in vivo studies have revealed synergistic effects that involve modulated immune responses, reduced inflammation, and improved tissue regeneration, none of which exist in a checkerboard assay.¹⁸ Conversely, combinations that look inert in vitro might produce meaningful benefit through immune-mediated pathways the lab test cannot detect.

Why the Mechanism Matters for Predicting Synergy

The FICI number alone does not explain why two drugs interact the way they do, but the underlying mechanism often determines whether the synergy is robust or fragile. One well-studied pattern involves membrane disruption: if one agent damages the bacterial membrane, it can let a second agent that normally struggles to get inside the cell reach its intracellular target more easily. Silver nanoparticles, for instance, showed synergistic or additive effects with antibiotics like kanamycin and chloramphenicol that act on ribosomes inside the cell, because the nanoparticles destabilized the membrane and promoted drug entry. But the same nanoparticles showed no synergy with beta-lactam antibiotics, which act on the cell wall rather than needing to penetrate the membrane.¹⁹ Knowing the mechanism behind a low FICI helps predict whether the combination will hold up against different bacterial species or resistance profiles, rather than treating the number as a universal verdict.

Synergy, Resistance, and the Paradox of Being Too Effective

There is a counterintuitive wrinkle that rarely makes it into the standard FICI discussion. Computational modeling of how drug interactions affect resistance evolution has found that synergy has two opposing effects. On one hand, a synergistic combination clears an infection faster, giving resistant mutants less time to appear. On the other hand, synergy increases the selective advantage those mutants enjoy over normal (wild-type) cells, because the wild-type population is being crushed more aggressively by the combination. When microbes compete strongly for resources, a tradeoff emerges: greater synergy speeds up clearance but also raises the probability that multi-drug-resistant mutants will take over. Above a certain threshold of synergy, pushing the interaction strength higher no longer improves clearance but continues to increase resistance risk.²⁰

This modeling suggests that the lowest possible FICI is not always the best therapeutic outcome. In some scenarios, a mildly antagonistic combination, one that kills more slowly, might actually suppress resistance evolution more effectively than a powerfully synergistic one. It is a finding that challenges the common assumption that synergy is always the goal, and it highlights why the FICI, useful as it is for characterizing drug interactions in the lab, cannot be the sole guide for treatment decisions.

Common Pitfalls When Reading FICI Data in Published Papers

If you are reviewing published literature that reports FICI values, a few recurring issues are worth watching for:

Unstated cutoffs: Some papers define synergy as FICI ≤0.5, others use ≤0.75, and still others adopt the narrower scheme with additivity defined as 1.0 to 1.25. If the paper does not explicitly state its thresholds, the classification is difficult to evaluate or compare with other studies.
Single replicates: A FICI based on a single checkerboard run is unreliable. Given the inherent variability of the method, triplicate testing with confidence intervals provides much more trustworthy classification.
Incubation time: For slow-growing organisms, the incubation period affects the result. FICI values determined at 24 hours can differ meaningfully from those at 48 hours, and some studies do not report which time point was used.
Endpoint ambiguity: Particularly in antifungal work, the choice between complete and partial growth inhibition endpoints can swing the FICI by a factor of two or more, as described above for Aspergillus.
Missing monotherapy MICs: The FICI formula depends on accurate individual MICs. If those monotherapy values are off by a single dilution step, the entire index shifts accordingly.

None of these pitfalls invalidate the FICI as a metric. They do mean that a single FICI number, pulled from a single study, should be treated as preliminary evidence rather than a definitive characterization of how two drugs interact. The most reliable conclusions come from studies that run multiple replicates, use clearly stated cutoffs, and ideally confirm results with a second method like time-kill analysis.