What Procalcitonin and CRP Can and Cannot Decide About Antibiotics

A fever, cough, elevated white-cell count, or abnormal chest image can arise from more than one cause. The same is true of inflammatory biomarkers. Procalcitonin and C-reactive protein, usually abbreviated PCT and CRP, can contribute information about infection and inflammation, but they do not contain a hidden label saying bacterial, viral, treat, or do not treat.

Their clinical value comes from combining a result with pretest probability, illness severity, the suspected site, timing, microbiology, organ function, and repeated examination. Their research value comes from a different question: does a biomarker-guided care pathway reduce antibiotic use without worsening outcomes? That pathway question cannot be answered by quoting a threshold in isolation.

What each marker represents#

CRP is an acute-phase protein produced by the liver in response to inflammatory signaling, especially interleukin-6, and it can rise with bacterial or viral infection, tissue injury, autoimmune disease, malignancy, and other inflammatory states. That broad responsiveness is exactly what makes it useful for following inflammation and what limits its specificity. It moves; it does not say why.

Procalcitonin is the precursor of calcitonin. During many systemic bacterial infections, tissues outside the thyroid release more of it into the circulation. Viral immune responses can suppress some of that production, and this biological pattern explains why PCT may separate bacterial from viral illness better than a general inflammatory marker in some settings. It does not create a perfect boundary.

The two measures also have different kinetics. PCT may rise within hours of a systemic bacterial insult and fall as the stimulus resolves. CRP commonly rises more slowly and can remain elevated after clinical improvement. Timing relative to symptom onset matters. A sample taken very early can look reassuring before either marker has fully responded.

Why no threshold is a diagnosis#

A diagnostic result has meaning only after you consider the probability before testing. If bacterial pneumonia is already very likely because of hypoxemia, a compatible infiltrate, focal examination findings, and systemic illness, a low PCT may lower the probability without making it safely negligible. If bacterial infection is unlikely in a stable person with a typical viral syndrome, a modestly abnormal CRP may raise probability without making antibiotics beneficial.

Thresholds trade sensitivity against specificity. Lowering a PCT cutoff captures more bacterial infections but also labels more nonbacterial illness as concerning. Raising it avoids more unnecessary antibiotic use but misses more infections. The best trade is not universal. Missing early sepsis in an unstable patient has a different consequence from prescribing an avoidable course for an uncomplicated outpatient illness.

Assay calibration, specimen handling, laboratory turnaround, kidney dysfunction, recent surgery, major trauma, burns, and immune status can alter interpretation. Localized infections can produce less systemic PCT than bloodstream infection. Atypical organisms and some fungal infections do not follow a single predictable pattern. These are not obscure exceptions. They are why the result belongs inside a clinical process.

Starting antibiotics and stopping them are different decisions#

The evidence for a biomarker may be stronger for stopping treatment than for deciding never to start it. At presentation, uncertainty and the cost of delay can be high. After 48 or 72 hours, clinicians may have cultures, imaging, a treatment response, source-control information, and serial biomarkers. The decision has changed.

The ATS and IDSA community-acquired pneumonia guideline recommends initiating empiric antibiotics in adults with clinically suspected and radiographically confirmed pneumonia regardless of the initial serum PCT. Reported sensitivities are too variable to make a low value a reliable exclusion rule.

The Surviving Sepsis Campaign guideline similarly does not support using PCT plus clinical evaluation to decide when to start antimicrobials instead of clinical evaluation alone. It suggests that PCT plus clinical evaluation may help decide when to discontinue them when an adequate duration is unclear and source control is adequate, and the recommendation is weak, which accurately reflects uncertainty and variation among studies. Keep the two decisions apart, because evidence that serial values can help end an unnecessary extra day does not prove that the initial value can safely cancel urgent treatment.

What respiratory-infection trials found#

An individual-participant Cochrane review pooled trials that embedded PCT in treatment algorithms for acute respiratory infections. The PCT-guided groups received fewer antibiotic days and had fewer antibiotic-related adverse effects; outcomes were not worse in the pooled analysis, and mortality was lower, although interpretation must account for differences among trials and settings.

The result is evidence for an algorithm, not for handing a laboratory number to a clinician without guidance; many trials specified repeat testing, recommended actions at several concentration ranges, exceptions for unstable patients, and permission to override the recommendation. Adherence and usual-care prescribing influenced the size of benefit.

The U.S. ProACT trial is an important counterpoint. In emergency departments and hospitals with lower respiratory tract infection, providing PCT results and guideline recommendations did not significantly reduce antibiotic days compared with usual care. Clinicians frequently did not follow the PCT recommendation, and baseline practice differed from earlier European studies. A biomarker with positive trials can have little added value when the surrounding workflow does not change.

In critical care, an individual-participant meta-analysis found shorter antibiotic courses with PCT-guided care and did not show a mortality penalty. Yet intensive-care populations, infection sources, stopping rules, and adherence varied. The finding supports carefully governed serial use. It does not establish a universal discharge or stop threshold.

What point-of-care CRP can add#

CRP is often studied in primary care for acute respiratory symptoms, where uncertainty and defensive prescribing can lead to antibiotics for illnesses unlikely to benefit, and a rapid result can make a conversation more concrete, especially when it is paired with communication training and a delayed-prescription or follow-up plan.

The Cochrane review of point-of-care biomarkers found that CRP-guided care probably reduced antibiotic prescribing for acute respiratory infections without a meaningful reduction in recovery. Heterogeneity was substantial. Most evidence came from European and Asian systems, and the test was an adjunct to standard care rather than a stand-alone rule.

The concentration ranges used in trials should not be detached from their original populations. A stable adult with a short-duration lower respiratory illness is not the same as a febrile infant, a person receiving chemotherapy, an older adult with delirium, or someone with severe hypoxemia. A threshold studied in one group may be poorly calibrated in another.

Serial change can be more informative than one result#

A single value is a snapshot. A repeated PCT that falls substantially while blood pressure, oxygenation, fever, and organ function improve supports a different inference from a repeated value that rises while the patient deteriorates. Serial CRP can similarly help describe an inflammatory trajectory, although it often lags behind clinical change.

Trend interpretation still requires discipline. Regression toward the mean, laboratory variation, the expected half-life, renal clearance, and the time between samples all matter. A falling marker does not prove that an abscess has been drained or that a prosthetic infection is eradicated. A persistently high CRP after surgery can be nonspecific. The clinical question has to be stated before you celebrate the trend.

What the tests cannot tell you#

Neither marker identifies the organism; neither reports whether a bacterium is susceptible to a particular drug, neither confirms source control, and neither tells you whether an antibiotic reached the infected compartment. Cultures, molecular tests, imaging, procedures, and local resistance information answer different parts of the problem.

They also cannot calculate net benefit by themselves. The decision includes severity, allergy history, pregnancy, immune compromise, drug interactions, kidney and liver function, prior resistant organisms, and the consequences of delay. Antibiotic stewardship is not simply using fewer drugs. It is using prompt effective treatment when needed and avoiding treatment when expected harm exceeds benefit.

A practical interpretation sequence#

First, define the suspected syndrome and severity. Second, estimate pretest probability using history, examination, vital signs, and whatever imaging you have. Third, ask whether the biomarker was validated for this decision, population, and timing. Fourth, interpret the value as a probability modifier rather than a command. Fifth, obtain microbiology or source-control evaluation when indicated. Finally, set the point at which you will reassess.

In a report or a study, ask whether clinicians received a numeric value, an algorithm, or both, then check the override rules, the adherence, the repeat-testing schedules, the outcome definitions, the antibiotic use in the control group, and the safety endpoints. A study showing fewer prescriptions is incomplete if it does not also measure recovery, escalation, readmission, serious infection, and adverse effects. The prescription count on its own cannot tell you that.

The safest conclusion#

PCT and CRP are most useful when they make reasoning more explicit, and they can expose a mismatch between low probability and habitual prescribing, or support stopping treatment after accumulating evidence. They become dangerous when a threshold erases severity, timing, and uncertainty.

The mature interpretation is conditional: this result, obtained at this time, in this syndrome, changes probability by this much and will be reconsidered alongside the patient's course. That is more demanding than a red or green laboratory flag, and it is closer to what the evidence actually supports.

References#

1. ATS and IDSA community-acquired pneumonia guideline
2. Surviving Sepsis Campaign 2021 guideline
3. Cochrane review of PCT guidance in acute respiratory infection
4. ProACT randomized trial
5. Cochrane review of point-of-care inflammatory biomarkers
6. PCT-guided treatment meta-analysis in intensive care