Of Clinical Microbiology

Setting Up a Routine Culture and Susceptibility Workflow in a Clinical Lab

The first thing most people get wrong about clinical microbiology is the assumption that automation removes the need for hands-on technique. It doesn't. Automated systems like Vitek 2, Phoenix, and MicroScan do a good job of reading color changes and turfidity, but they still depend on properly prepared inocula, correct media, and someone who knows when a result looks wrong. I spent years watching technologists trust machines to do the thinking for them, then spend three hours afterward trying to explain discrepant results to attending physicians. Here is how I actually run the workflow. You start with the specimen. The type of specimen dictates everything — what you plate, what you incubate, what you skip. A clean-catch midstream urine is straightforward: calibrated loop onto blood agar and MacConkey, 35°C in ambient air, read at 16 to 18 hours. A sputum specimen from a good-quality sample (fewer than 10 epithelial cells per low-power field) gets the same plates, but you also run a quadrant streak to encourage isolation. Wound swabs are where things get messy, because swabs are inherently suboptimal for recovering fastidious organisms. I switch to transport media or the wound margin onto the plate surface whenever possible.

Practical Approaches Of Clinical Microbiology

Susceptibility testing follows culture isolation. I use disk diffusion for most routine organisms because it is cheaper, faster to set up, and gives you visual confirmation of zone morphology that can flag heteroresistance or edge artifacts. Broth microdilution or automated AST cards are fine for high-throughput labs, but they hide problems. A Vitek 2 card will give you a clean S/I/R call and move on, even if the growth curve looks slightly irregular. Disk diffusion forces you to look at the plate. One detail beginners consistently overlook: inoculum standardization. The 0.5 McFarland standard is not a suggestion. Over-inoculation produces falsely resistant results because there is so much organism that the disk cannot create a concentration gradient across the expected distance. Under-inoculation does the opposite — zones expand artificially and you call a resistant organism susceptible. I verify my suspensions against a fresh McFarland standard every morning, and I remake any suspension that looks too cloudy or too thin. A single bad inoculum can flip a vancomycin-susceptible Enterococcus call to intermediate and trigger unnecessary linezolid ordering. Incubation conditions matter more than most lab manuals admit. Most pathogens grow fine in ambient air at 35°C, but certain organisms need CO2. Neisseria gonorrhoeae and Streptococcus pneumoniae require 5 to 10 percent CO2 for reliable recovery, and if you plate them on a shelf in a standard incubator you will miss them or get tiny fragile colonies that look like contaminants. I keep a CO2 jar with verified indicators in the corner of my workbench for these cases. Haemophilus influenzae needs both X and V factors — defibrinated blood on Chocolate agar, not on standard blood agar where the factor is destroyed by NADase in the red cells.

I ran into a specific problem last year that illustrates why experience matters more than protocol checkboxes. A blood culture from an ICU patient flagged positive for gram-positive cocci in clusters. The automated system identified Staphylococcus aureus and reported oxacillin susceptible. The colony morphology looked right, the catalase and coagulase tests were positive, and the PBP2a latex test was negative. Everything said methicillin-susceptible S. aureus. But the zone around the oxacillin disk had a faint fuzzy halo — heteroresistance. The machine read it as susceptible because the majority population was sensitive, but a subpopulation was expressingmecA-mediated resistance. I retried using a higher inoculum density and a longer incubation of 24 hours, ran a cefoxitin disk instead of oxacillin since cefoxitin is a better inducer of mecA expression in staphylococci, and ordered a PCR for mecA. The PCR was positive. The patient was on nafcillin. We switched to vancomycin the same day. That heteroresistant population would have progressed under nafcillin monotherapy and the patient likely would have developed a persistent bacteremia within 48 to 72 hours. That kind of edge case is why I always run a cefoxitin disk as a surrogate marker for oxacillin resistance in staphylococci now, even though the CLSI guidelines list both as acceptable. Cefoxitin detectsmecA-mediated resistance more reliably than oxacillin because it induces expression at lower concentrations and the zone interpretation is more consistent across different inoculum densities and incubation times. The difference is subtle but clinically meaningful. Another practical consideration: quality control strains. You run ATCC 25922 for E. coli, 29213 for S. aureus, and 49607 for P. aeruginosa on every batch of disks and every new lot of Mueller-Hinton agar. These are not optional. I have seen labs skip QC because "the machine passed its internal check," which conflates machine function with media and reagent quality. A Vitek 2 can read cards perfectly and still produce wrong results if the inoculum is wrong or the agar is too thick or too thin. The standard Mueller-Hinton depth is 4 millimeters. If your poured plates are consistently 3 millimeters, adjust your incubation time or your disk diffusions will read larger than they should.

Antibiotic stewardship integration is another area where the lab has real influence. When you report an organism as resistant, you are directly narrowing the therapeutic options for the treating physician. I make sure my reports include not just the S/I/R designation but also notes about intrinsic resistance patterns and emergent resistance mechanisms. An Enterobacterales isolate that is resistant to ampicillin but susceptible to ceftriaxone is expected — that is basic beta-lactamase activity. An isolate that is resistant to ampicillin, cefazolin, and ceftriaxone but susceptible to cefepime and meropenem is an ESBL producer, and I flag that explicitly. ESBL confirmation requires a double-disk synergy test or a Carba NP test for carbapenemase detection, and I run those on every extended-spectrum cephalosporin-resistant Enterobacterales regardless of how many there are in a given week. The volume of work does not justify skipping the confirmatory test because the treatment implications are significant. Mass spectrometry like MALDI-TOF has changed the identification landscape considerably. It cuts organism ID time from 16 to 24 hours down to 30 minutes after a pure colony is obtained. But it has limitations. It struggles with closely related species within the same genus — Pasteurella multocida versus Pasteurella canis, for example, or differentiating between members of the Streptococcus anginosus group. It also requires a clean, pure colony. If your streak plate has mixed growth, MALDI-TOF will give you a ambiguous or incorrect reading. I still maintain a library of biochemical tests and API strips for these edge cases, and I run MALDI-TOF as the first pass, not the final word. The turnaround time for a complete culture and sensitivity from receipt to final report is typically 48 to 72 hours for most specimen types. Blood cultures can flag positive in as little as 6 to 18 hours, and a Gram stain from the positive bottle can give you a preliminary answer within that window. That preliminary Gram stain result — gram-positive rods, gram-negative coccobacilli, yeast — is often the most clinically actionable piece of information before the full identification and susceptibility data are available. I make sure the microbiologist reviewing positive blood cultures calls the clinical team with that preliminary finding immediately, not after the final report is ready.

Specimen rejection criteria are another practical area. A throat swab in plain transport media without Stuart or Amies medium deteriorates quickly. A stool specimen that has been sitting at room temperature for more than two hours loses viability of Campylobacter and some Salmonella strains. I reject specimens that are clearly compromised and document the reason. This is not bureaucratic pedantry — it is the difference between a false-negative result and a correct diagnosis. Labs that accept every specimen regardless of condition produce more ambiguous or incorrect results, and clinicians lose trust in the laboratory when their positive cultures don't match the clinical picture. Cost is a real constraint in clinical microbiology. Automated AST systems cost significantly more per test than disk diffusion, and the consumables are not cheap. A single Vitek 2 AST card runs about $15 to $20 depending on the configuration. Disk diffusion costs a fraction of that once you account for the reusable incubator space and the relatively low cost of Mueller-Hinton plates and disk packs. For a mid-size hospital lab processing 200 to 400 culture and sensitivity sets per day, the cost difference between automated and manual methods can be tens of thousands of dollars annually. I balance this against throughput needs and staff capacity, but I do not treat automation as an unconditional upgrade. Documentation and traceability matter more than most labs acknowledge. Every plate, every organism ID, every susceptibility result, every QC run, every specimen acceptance or rejection decision should be logged with a timestamp and a technologist identifier. This is not just regulatory compliance — it is the foundation for outbreak investigation and internal quality audits. When a cluster of identical Klebsiella pneumoniae isolates appears across multiple patients in a single week, the ability to trace back exactly who processed each specimen, on what day, with what lot of media, and under what incubation conditions can be the difference between identifying a cross-contamination event and missing it entirely.