Microbiology Culture Reporting and the Hospital Antibiogram

A preliminary report should be visibly labelled, timestamped, attributed, and clearly incomplete. The clinician reading it must be able to tell what is known so far and what is still pending, so the report should state the pending components rather than simply omitting them. A Gram stain result with no indication that culture is still incubating reads as a finished answer.

When the final report supersedes the preliminary, the previous version should remain retrievable but should not compete for attention in the chart. The failure mode to avoid is two apparently equal reports where the clinician has to work out which is current, particularly when the preliminary and final differ.

Amendments after final are a separate case and need explicit handling. A corrected organism identification or a revised susceptibility must be flagged as an amendment, notified actively rather than passively, and recorded with the reason. Silent amendment of a microbiology report is one of the more dangerous behaviours a laboratory system can permit.

Which antimicrobials are tested and which are reported are two different decisions. Testing panels follow the organism and specimen source according to established standards. Reporting is filtered — the laboratory deliberately suppresses agents that are inappropriate for the site, unnecessarily broad for the organism, or reserved for situations where narrower options have failed.

Selective reporting exists because clinicians reasonably prescribe from what they are shown. Displaying a carbapenem susceptibility alongside a narrow agent that would treat the infection perfectly well invites unnecessary escalation. Cascade reporting takes this further: broader agents are revealed only when the organism is resistant to the first-line options.

The suppression rules must be agreed with the infectious disease and stewardship team, documented, and reviewed as local resistance patterns change. They also need an override path, because there are legitimate cases — allergy, penetration to a specific site, pregnancy — where a clinician needs the suppressed result. An override that requires a phone call to the laboratory is usually the right level of friction.

A cumulative antibiogram summarises the percentage of isolates of each organism susceptible to each agent over a defined period, and it is what guides empirical therapy before a patient's own culture returns. Its usefulness depends entirely on how it is constructed, and constructed carelessly it will point clinicians in the wrong direction with the authority of data.

The core rules are consistent: include only the first isolate per patient per organism per period, exclude surveillance and screening isolates, report only organisms with enough isolates to support a percentage, and use consistent interpretive criteria throughout. Duplicate isolates from a single colonised patient will otherwise dominate the denominator and distort the whole picture.

Stratification is where the antibiogram becomes genuinely useful. Resistance in an ICU is not resistance in OPD, and community-acquired urinary isolates behave differently from catheter-associated ones. Producing separate antibiograms for critical care, inpatient, and outpatient populations gives clinicians something applicable to the patient in front of them rather than a hospital-wide average that fits nobody.

The antibiogram is one input to stewardship; the daily flow of microbiology results is the other and the more actionable. Stewardship teams need a working list of patients on antimicrobials with positive cultures, patients on broad-spectrum therapy with a susceptibility result permitting de-escalation, and patients on empirical therapy with no culture sent at all.

That last category is often the largest and the least visible. Linking prescribing data with microbiology ordering makes it apparent, and it is a straightforward query once the two datasets share a patient and encounter identity. HealUDoc dashboards can bring pharmacy antimicrobial dispensing alongside microbiology results so a stewardship round works from one list rather than from three reports reconciled by hand.

Feedback needs to reach prescribers as well as committees. Unit-level and prescriber-level summaries of de-escalation opportunities acted on, culture-before-antibiotic rates, and duration of broad-spectrum therapy influence behaviour in a way that hospital-wide consumption figures do not.

Microbiology has its own critical results, and they are not numeric thresholds. A positive blood culture Gram stain, a positive cerebrospinal fluid finding, growth of an organism with significant infection control implications, and detection of certain notifiable pathogens all require active communication to a clinician who can act, not simply publication to the record.

The communication record should capture what was communicated, to whom, by whom, at what time, and with read-back confirmation. This is a standing accreditation expectation and a standing weak point, because the call happens at three in the morning and the documentation is completed later or not at all. Building the record into the same screen that produces the result is the only reliable fix.

Escalation matters when the first contact cannot be reached. A positive blood culture on a patient whose treating team has changed shift needs a defined second and third contact, with the escalation itself recorded. Laboratories that leave this to individual judgement discover the gap during an incident review.

Everything downstream of microbiology reporting — antibiogram, stewardship analytics, infection control surveillance — rests on consistent organism naming and consistent specimen source coding. Free-text organism names, or the same organism recorded under three spellings, make aggregate analysis unreliable in ways that are hard to detect from the output.

Coded organism dictionaries and standardised specimen source lists are therefore a prerequisite rather than an enhancement. Specimen source is particularly prone to drift because it is often entered by the collecting clinician rather than the laboratory, and a urine sample recorded variously as urine, mid-stream urine, and catheter urine cannot be stratified.

Retrospective cleanup is expensive and imperfect, so the controls belong at entry: constrained pick lists, mapping of legacy values, and a periodic review of what is actually being recorded. A laboratory that fixes this once gets years of usable surveillance data; one that does not will re-derive its antibiogram by hand every year.