# Bloodstream Infection: Causes and Diagnosis
A bloodstream infection means live microorganisms are present in circulating blood, and the diagnosis rests almost entirely on blood cultures drawn with strict aseptic technique. The single hardest judgment in the whole process is deciding whether a positive culture represents a true pathogen invading from a real source or a skin organism that fell into the bottle during collection.
## Bacteremia, Sepsis, and Septicemia Are Not the Same Thing
The words get used interchangeably in casual conversation, and that causes real confusion at the bedside. They describe three different concepts.
**Bacteremia** is simply the presence of viable bacteria in the bloodstream. It can be transient. Brushing your teeth, chewing hard food, or having a wound debrided can push oral or skin flora into the blood for minutes to hours, and a healthy immune system clears it without symptoms. Bacteremia becomes clinically important when it is persistent, when the organism is a true pathogen such as *Staphylococcus aureus*, or when the patient cannot clear it.
**Septicemia** is an older term that loosely meant "blood poisoning." Modern microbiology and critical care have largely retired it because it blurred the line between the presence of organisms and the host response. When you see it in a chart, read it as "bacteremia with clinical concern."
**Sepsis** is defined by the host response, not the culture. It is life-threatening organ dysfunction caused by a dysregulated host response to infection. A patient can be septic with a negative blood culture, and a patient can have bacteremia without sepsis. Sepsis is scored clinically, for example with the Sequential Organ Failure Assessment (SOFA) score, which grades respiratory, coagulation, liver, cardiovascular, renal, and neurologic function. The SOFA score is a clinical tool, and it is used in the structured models that separate pathogens from contaminants in culture results [1].
The practical consequence: a positive blood culture is a laboratory finding, and sepsis is a clinical syndrome. You need both the microbiology and the clinical picture to act correctly.
## How Bacteria Get Into the Bloodstream
Bacteria in the bloodstream arrive by a limited number of routes. Understanding them is what lets a clinician predict which organism is likely and where it came from.
### Translocation from a mucosal surface
The gut is the largest reservoir of bacteria in the body. When the intestinal barrier fails, through ischemia, severe inflammation, chemotherapy-induced mucositis, or perforation, enteric organisms cross into the portal circulation. *Escherichia coli* and *Klebsiella* species are the classic results. In patients with hematologic malignancies, neutrophil counts at or below 0.5 × 10⁹/L are a strong predictor of bloodstream infection because mucosal defenses are compromised [2]. A related route is translocation from the female genital tract. *Gardnerella vaginalis* is a fastidious organism that rarely invades, but it has been documented causing bacteremia with an extragenital source in a patient with no relevant history, and it was only identified by metagenomic sequencing after routine cultures were negative [3].
### Direct inoculation
Intravascular catheters, injection drug use, surgical wounds, and indwelling devices give organisms a direct path into the blood. Central venous catheters are a leading source in intensive care. A study of blood cultures drawn from the catheter hub during central line insertion found roughly 2.6 times higher odds of contamination compared with cultures drawn from other sites, which is why sampling from a fresh peripheral site is preferred whenever it is feasible [4].
### Focal infection with spillover
Pneumonia, pyelonephritis, cholangitis, endocarditis, and soft tissue infection all seed the blood intermittently. This is why the same organism often appears in both the blood and the presumed source, and why "site concordance" is one of the criteria used to call an isolate a true pathogen [1].
### Zoonotic and occupational exposure
Bloodstream infection is not always hospital-acquired. A pig farmer developed bacteremia with *Helicobacter trogontum*, an organism first isolated from rat colonic mucosa and known from pig feces, after presenting with fever and headache. [Whole genome sequencing](/blog/guides/whole-genome-sequencing) of the blood isolate confirmed the species [5]. This is a reminder that occupational history belongs in the assessment.
## The Organisms That Matter Most
The typical causes of bacterial blood infections are consistent across human and [veterinary medicine](/blog/careers/veterinary-medicine-careers-from-clinical-practice-to-public-health). In people, the most common pathogens recovered in a prospective series were *Klebsiella* (20.0%), *Acinetobacter* (9.3%), and *Pseudomonas* (6.3%), with *E. coli*, *S. aureus*, and streptococci also prominent [1]. In dogs, a German teaching hospital series of 750 blood cultures found clinically relevant growth in 13.6% of samples, dominated by Enterobacterales, coagulase-positive staphylococci, beta-hemolytic streptococci, and obligate anaerobes [6]. In neonatal foals, blood cultures were positive in 43% of cases, with Gram-positive organisms predominating at 65% and the most common isolates being *Staphylococcus* species (25%), *E. coli* (11%), and *Streptococcus* species (11%) [7].
Two patterns are worth internalizing. Gram-negative rods, especially *E. coli* and *Klebsiella*, usually arrive from the gut or urinary tract. Gram-positive cocci, especially staphylococci and streptococci, usually arrive from skin, soft tissue, or a device.
### Why some organisms are harder to grow
Fastidious organisms do not grow on standard media. *Gardnerella vaginalis* is a good example: blood, cerebrospinal fluid, and bone marrow cultures were all negative in a documented case, and only plasma metagenomic next-generation sequencing (mNGS), an unbiased sequencing method that reads all DNA in a sample, identified the organism [3]. This is the niche where molecular methods add value, not as a replacement for culture but as a supplement when culture is negative and suspicion remains high.
## Blood Culture: The Reference Standard
Blood culture and agar plates like these are the reference standard for detecting bloodstream infection. Image: Ajay Kumar Chaurasiya, CC BY-SA 4.0, via Wikimedia Commons.
Blood culture is still the gold standard for identifying bacterial bloodstream infections and for directing antimicrobial treatment [6]. The method is straightforward in principle. A volume of blood is inoculated into bottles containing nutrient broth, one optimized for aerobic organisms and one for anaerobes, and the bottles are incubated in an automated instrument that detects bacterial growth, usually by sensing changes in gas pressure or fluorescence. When a bottle flags positive, the laboratory performs Gram stain, subculture, and identification, followed by antimicrobial susceptibility testing.
### Volume and number of sets matter
The sensitivity of blood culture depends heavily on how much blood is cultured. A single set is not enough, because a low-density bacteremia may only be captured in one of several bottles. Multiple sets drawn from different sites increase yield and, crucially, provide the growth pattern data that lets the laboratory distinguish pathogens from contaminants.
### The bottle shortage problem
Supply chains affect diagnosis. During a 2024 shortage of BACTEC blood culture bottles in a São Paulo laboratory network covering 936,981 culture events, utilization dropped from a median of 78.55 events per week to 48.02, and true-positive counts fell while the contamination proportion rose to 3.12% during the shortage before falling to 1.96% afterward [8]. Fewer bottles meant fewer diagnoses and a higher proportion of noise.
## Contamination: The Central Interpretation Problem
Every blood culture has a chance of growing something that was never in the bloodstream. Skin flora introduced at the puncture site is the usual culprit. The organisms responsible are predictable: coagulase-negative staphylococci (CoNS), diphtheroids (coryneform bacteria), *Cutibacterium acnes*, and *Bacillus* species. Guidelines generally recommend keeping contamination below 3% of cultures [9].
Contamination is not a trivial annoyance. It drives unnecessary antibiotic exposure, prolongs hospitalization, and creates diagnostic uncertainty [10]. It also distorts surveillance data, which is why laboratories track it as a quality metric.
### Growth pattern is the strongest single clue
The most useful laboratory signal is whether one bottle or both bottles in a set grow. In a retrospective analysis of 38,216 blood culture sets, discordant sets (one bottle positive) grew 71.1% contaminants and 28.9% true pathogens, while concordant sets (both bottles positive) grew 88.5% true pathogens and only 11.4% contaminants [10].
The pattern is even sharper for specific organisms. Among CoNS isolates, 98.0% of discordant sets were contaminants. For *S. aureus*, only 30.4% of discordant sets were contaminants, meaning a single positive bottle of *S. aureus* is still very likely to be real [10]. The negative predictive value of a discordant first CoNS set to exclude true CoNS bacteremia was 98.1% (95% CI 96.7% to 98.9%) [10]. That number is the reason a single CoNS bottle in a stable patient is usually watched rather than treated.
### Time to positivity adds information
Time to positivity (TTP) is the interval from incubation start to the instrument flagging the bottle as positive. A shorter TTP generally means a higher bacterial load, which correlates with true infection. In a prospective cohort of 205 positive cultures, mean TTP was 16.3 ± 8.0 hours for pathogens, with bacterial pathogens averaging 15.01 hours and fungal pathogens averaging 26.33 hours [1]. In children with *E. coli* bloodstream infection, each hour of shorter TTP was independently associated with higher in-hospital mortality (OR 0.71 per hour increase, 95% CI 0.56 to 0.89) [11].
### A structured approach beats a gut call
The most defensible way to classify an isolate is a stepwise algorithm that combines SOFA score, TTP, and site concordance. Applied prospectively to 205 isolates, a seven-step model classified 78% as pathogens and 22% as non-pathogens. The non-pathogens were dominated by CoNS (18.5%) and *Stenotrophomonas* (8.3%) [1]. This is the framework to keep in mind when a result looks ambiguous.
### Table: Common contaminants versus true pathogens
| Feature | Typical contaminant | Typical true pathogen |
|--|--|--|
| Common organisms | Coagulase-negative staphylococci, diphtheroids, *Cutibacterium acnes*, *Bacillus* species | *E. coli*, *Klebsiella* species, *S. aureus*, *Streptococcus* species, *Pseudomonas*, *Acinetobacter* |
| Bottles positive per set | Usually one of two (discordant) | Usually both (concordant) |
| Time to positivity | Often prolonged | Often shorter, roughly 15 hours for bacterial pathogens [1] |
| Clinical picture | Often stable, no clear source | Fever, organ dysfunction, identifiable source |
| Repeat culture | Usually negative | Frequently positive again |
| Proportion in discordant sets | 71.1% of discordant sets [10] | 28.9% of discordant sets [10] |
| Proportion in concordant sets | 11.4% of concordant sets [10] | 88.5% of concordant sets [10] |
## Reducing Contamination at the Bench and Bedside
Contamination is a preanalytical problem, which means it is fixable.
**Aseptic technique.** Skin antisepsis, allowing the disinfectant to dry, sterile gloves, and not repalpating the site after preparation are the fundamentals. These steps are not optional and they are the single largest determinant of culture quality.
**Avoid catheter hubs when possible.** Blood cultures drawn from a central venous catheter hub during insertion carried about 2.6 times the odds of contamination compared with cultures from other sites in a same-day within-patient paired analysis [4]. Peripheral sampling is preferred when it is feasible.
**Diversion devices.** A device that discards the first 0.15 mL of blood, which is the fraction most likely to carry skin flora, reduced contamination from 10.4% to 2.7% in an emergency department pilot, a 74% relative reduction, with staff adherence averaging 28.5% [9]. Even imperfect adherence produced a large effect.
**Training and system fixes.** An operational research program at a Ghanaian tertiary hospital combined physician sensitization, laboratory staff training, automated incubation, and better consumable supply. Blood culture requests rose fivefold from 8% to 40% of eligible inpatients, time from admission to request fell from two days to one, laboratory turnaround fell from seven days to five, and contamination dropped from 16% to 14% [12]. The diagnostic yield improved only modestly, from 7% to 10%, which shows that utilization and quality are separate problems that need separate solutions.
## The Diagnostic Workflow
The path from suspicion to answer follows a predictable sequence.
```mermaid
flowchart TD
A[Suspected bloodstream infection] --> B[Aseptic blood collection]
B --> C[Incubate aerobic and anaerobic bottles]
C --> D{Bottle flags positive}
D -->|No growth| E[Assess non culture causes]
D -->|One bottle only| F[Weigh contaminant versus pathogen]
D -->|Both bottles| G[Likely true pathogen]
F --> H[Check organism and clinical context]
G --> I[Identify organism and susceptibility]
H --> I
I --> J[Correlate with source and organ function]
J --> K[Decide treatment and repeat culture]
```
## Molecular and Culture-Independent Methods
Culture takes time, and in septic patients time matters. Several technologies shorten or bypass the growth step.
**[Multiplex PCR](/knowledge/diagnostics/molecular/multiplex-pcr-design-optimization-and-troubleshooting) panels.** The BioFire FilmArray blood culture identification panel runs directly on positive blood culture broth and returns organism identities within hours. It is not perfect. In a 12-month study of 454 positive blood cultures, 28 cases of fungemia were identified by culture, and agreement between culture and the panel was 50.0%. Eight results were false-positive fungemia, seven of them *Candida tropicalis* [13]. A positive panel result still needs culture confirmation.
**Metagenomic next-generation sequencing (mNGS).** This sequences all nucleic acid in a sample without targeting specific organisms. It has diagnosed infections that culture missed, including *Gardnerella vaginalis* bacteremia [3] and *Staphylococcus pettenkoferi* bloodstream infection where mNGS and two sets of peripheral cultures agreed, supporting true infection rather than contamination [14]. The limitation is interpretation. In a cohort of 329 episodes, 232 were mNGS-positive but culture-negative, and adjudication classified 124 as bloodstream infection and 108 as non-bloodstream infection, with nonviral signals present in 78.2% and 42.6% respectively [15]. Most mNGS organisms in that discordant group were judged plausible rather than confirmed. mNGS generates signal, not automatically a diagnosis.
**Droplet digital PCR (ddPCR).** A study of 336 patients found that adding bloodstream pathogen ddPCR to a conventional clinical model raised the area under the curve from 0.716 to 0.764, a modest improvement that did not reach [statistical significance](/blog/guides/statistical-significance) by the DeLong test (P = 0.152) [16].
**Shotgun metagenomics on whole blood.** In a study of 51 samples, only 36 yielded usable data, and 15 were excluded for low DNA library yield or low sequencing output. Only two results clearly matched blood culture, and most reads were suspected contamination [17]. This method is not ready for routine use.
## Special Populations and Recurrence
Bloodstream infection behaves differently in patients whose defenses are compromised.
In hematologic malignancy, bloodstream infection is a major complication. Among 3,014 patients, 725 developed bloodstream infection, and a risk model incorporating 12 variables, including neutrophil count at or below 0.5 × 10⁹/L, achieved an area under the curve of 0.775 in the derivation cohort [2]. Recurrence is common and dangerous. Among 1,153 patients with hematologic malignancy and bloodstream infection, 173 (15.0%) developed recurrent infection, and mortality climbed from 10.7% to 19.0% after the first recurrence and 30.5% after the second [18]. A Charlson Comorbidity Index above 3 at the initial episode was associated with later recurrence [18].
In children with *E. coli* bloodstream infection, in-hospital mortality was 12.32% in a series of 211 patients. Independent factors included invasive mechanical ventilation (OR 46.40), secondary hypoalbuminemia (OR 10.59), higher PRISM III score, and shorter time to positivity [11].
In veterinary medicine, parallels are direct. In dogs, multidrug-resistant Enterobacterales were common, and polymicrobial growth occurred in 7.8% of positive cultures [6]. In neonatal foals, multidrug-resistant bacteria made up 26% of isolates, predominantly Gram-positive, with resistance to penicillin at 53% and ceftiofur at 33%, while resistance to amikacin and gentamicin stayed low [7]. The diagnostic logic, aseptic collection, paired bottles, growth pattern interpretation, and susceptibility testing, is identical across species.
## Common Mistakes and Limitations
**Treating a single positive CoNS bottle as infection.** A discordant CoNS set has a 98.1% negative predictive value for excluding true CoNS bacteremia [10]. Acting on it exposes the patient to unnecessary antimicrobials.
**Dismissing a single positive *S. aureus* bottle.** Only 30.4% of discordant *S. aureus* sets were contaminants [10]. A single bottle of *S. aureus* deserves serious attention.
**Drawing cultures from a catheter hub out of convenience.** This roughly doubles the odds of contamination [4]. Use a peripheral site when you can.
**Culturing after antibiotics without noting it.** Prior antibiotic exposure is enriched in patients who go on to have recurrent bloodstream infection and can sterilize cultures while infection continues [18].
**Assuming a negative culture rules out bloodstream infection.** Fastidious organisms such as *Gardnerella vaginalis* do not grow on standard media and may require molecular methods [3].
**Over-reading molecular results.** False positives occur. Eight false-positive fungemia results appeared in one panel evaluation [13], and most mNGS reads in a whole-blood metagenomics study were suspected contamination [17].
**Ignoring time to positivity.** A short TTP carries prognostic weight, particularly in pediatric *E. coli* bacteremia [11].
**Forgetting that utilization and quality are different problems.** Increasing how often cultures are ordered does not automatically lower contamination, as one operational program demonstrated when requests rose fivefold but contamination only fell from 16% to 14% [12].
Individual cases need a veterinarian or physician who can integrate the culture result with the clinical picture. No algorithm replaces that judgment.
## Frequently Asked Questions
### What is the difference between bacteremia and sepsis?
Bacteremia is the presence of bacteria in the blood, which may be transient and harmless. Sepsis is organ dysfunction caused by a dysregulated host response to infection, and it is defined clinically rather than by culture.
### Can bacteria in the bloodstream clear on their own?
Yes. Transient bacteremia from chewing, dental work, or minor procedures is common and is usually cleared by a healthy immune system within minutes to hours.
### Why does my blood culture say one bottle grew bacteria?
A single positive bottle in a set is called a discordant result. Discordant sets grow contaminants about 71% of the time, so the organism identity and the clinical picture decide the interpretation [10].
### Which bacteria are most often contaminants in blood cultures?
Coagulase-negative staphylococci, diphtheroids, *Cutibacterium acnes*, and *Bacillus* species are the classic skin flora contaminants.
### How long does a blood culture take?
Many positive cultures flag within about 15 hours for bacterial pathogens, while fungal pathogens average closer to 26 hours [1]. Negative cultures are typically held for several days before being reported as no growth.
### What is time to positivity and why does it matter?
Time to positivity is how long the instrument takes to flag a bottle as positive. Shorter times generally indicate a higher bacterial load and correlate with true infection and worse outcomes [11].
### Can a blood culture be negative even with a real infection?
Yes. Fastidious organisms may not grow on standard media, and prior antibiotics can suppress growth. Molecular methods such as metagenomic sequencing can identify organisms that culture misses [3].
### Do dogs and cattle get bloodstream infections too?
Yes. Blood culture is the gold standard in veterinary medicine as well. In dogs, clinically relevant growth occurs in roughly 13.6% of cultures, with Enterobacterales, staphylococci, streptococci, and anaerobes predominating [6].
## Related Articles
- [Bacterial Infections in Chickens: Clinical Signs and Diagnosis](/knowledge/bacteria/avian-bacteria/bacterial-infections-chickens-clinical-signs)
- [Dog skin infection causes diagnosis treatment prevention](/knowledge/veterinary-medicine/dermatology-and-allergy/dog-skin-infection)
- [Snake Respiratory Infection: Symptoms, Causes, Diagnosis and Veterinary Treatment](/knowledge/veterinary-medicine/reptile-care/snake-respiratory-infection-symptoms-treatment)
- [Poultry Bacteria Infections: Comprehensive Overview of Pathogenesis, Diagnosis, and Antimicrobial Strategies](/knowledge/bacteria/avian-bacteria/poultry-bacteria-infections-pathogenesis-diagnosis-antimicrobial-strategies)
- [Can Worms Cause Infection In Dogs](/knowledge/veterinary-medicine/parasitic-diseases/can-worms-cause-infection-in-dogs)
- [What Can Cause A Urinary Tract Infection In A Cat](/knowledge/veterinary-medicine/clinical-methods/what-can-cause-a-urinary-tract-infection-in-a-cat)
- [E. coli UTI: Causes, Symptoms, and Treatment](/knowledge/veterinary-medicine/digestive-and-urinary-care/e-coli-uti-causes-symptoms-and-treatment)
## Sources
1. [A structured stepwise model to distinguish pathogens from non-pathogens in blood culture isolates: A prospective clinical-microbiological study.](https://pubmed.ncbi.nlm.nih.gov/42419646/)
2. [A multidimensional nomogram for clinical risk stratification of bloodstream infection and mortality in hematologic malignancies.](https://pubmed.ncbi.nlm.nih.gov/42723754/)
3. [Metagenomic next-generation sequencing contributes to the diagnosis of Gardnerella vaginalis bacteremia associated with a rare extragenital infection and diagnostic challenges in a 13-year-old female patient: A case report.](https://pubmed.ncbi.nlm.nih.gov/42719063/)
4. [Risk of blood culture contamination during central venous catheter insertion in critically ill adults: a self-controlled study.](https://pubmed.ncbi.nlm.nih.gov/42594818/)
5. [Bacteremia caused by Helicobacter trogontum indicative of zoonotic infection in a pig farmer: a case report.](https://pubmed.ncbi.nlm.nih.gov/41244967/)
6. [Veterinary blood culture diagnostic testing: methodology and results from a German single-center study on dogs (2014-2022).](https://pubmed.ncbi.nlm.nih.gov/41742499/)
7. [Retrospective study of blood culture isolates, antimicrobial sensitivity, and survival factors from 243 neonatal foals in Normandy (France).](https://pubmed.ncbi.nlm.nih.gov/42302765/)
8. [Blood culture bottle shortage and diagnostic yield: A 5-year interrupted time-series analysis.](https://pubmed.ncbi.nlm.nih.gov/42722427/)
9. [Reducing blood culture contamination in the ED: impact of Kurin Lock® implementation.](https://pubmed.ncbi.nlm.nih.gov/42169718/)
10. [Diagnostic value of blood culture growth patterns in distinguishing contaminants from pathogens.](https://pubmed.ncbi.nlm.nih.gov/41568951/)
11. [Secondary hypoalbuminemia is independently associated with in-hospital mortality in children with Escherichia coli bloodstream infection.](https://pubmed.ncbi.nlm.nih.gov/42720514/)
12. [Improved Utilisation and Quality of Blood Culture Services Following Operational Research in a Tertiary Hospital in Ghana.](https://pubmed.ncbi.nlm.nih.gov/41003579/)
13. [Fungal false positive in BioFire® FilmArray® analysis for bloodstream infection.](https://pubmed.ncbi.nlm.nih.gov/41455529/)
14. [mNGS-Supported Interpretation of Staphylococcus pettenkoferi Bloodstream Infection After Intracerebral Hemorrhage: A Case Report.](https://pubmed.ncbi.nlm.nih.gov/42199698/)
15. [Microbial signal profiles and organism-level concordance between plasma metagenomic sequencing and blood culture in suspected bloodstream infection.](https://pubmed.ncbi.nlm.nih.gov/42678565/)
16. [Incremental diagnostic value of bloodstream pathogen ddPCR for early identification of blood culture-positive bloodstream infection.](https://pubmed.ncbi.nlm.nih.gov/42591506/)
17. [Performance of shotgun metagenomics on whole blood from patients with suspected bloodstream infection: Challenges remain.](https://pubmed.ncbi.nlm.nih.gov/40819729/)
18. [Predictors and outcomes of recurrent bloodstream infection in hematological malignancies: a multicenter real-world study.](https://pubmed.ncbi.nlm.nih.gov/42745743/)
