BACKGROUND: Minimising antimicrobial overuse is needed to limit antimicrobial resistance. There is little evidence on how often microbiological testing informs antimicrobial de-escalation (e.g. stopping, shortening duration, switching to narrower spectrum or intravenous to oral switch) at 48-72 h "review and revise". We performed a patient level analysis of diagnostic microbiology and antimicrobial prescribing to determine the impact of microbiology results on antimicrobial review outcomes. METHODS: Antimicrobial prescribing data were collected for hospitalised adults from across Brighton and Sussex University Hospitals NHS Trust using routine monthly audits of prescribing practice from July 2016 to April 2017. Microbiology testing data for cultures of blood, urine, sputum and cerebrospinal fluid (CSF) were gathered from the hospital pathology database and linked to prescriptions with matching patient identification codes. Antimicrobial prescriptions were grouped into "prescription episodes" (PEs), defined as one or more antimicrobials prescribed to the same patient for the same indication. Medical records were reviewed for all PEs with positive microbiology and a randomised sample of those with negative results to assess the impact of the microbiology result on the antimicrobial prescription(s). RESULTS: After excluding topical and prophylactic prescriptions, data were available for 382 inpatient antimicrobial prescriptions grouped into 276 prescription episodes. 162/276 (59%) had contemporaneous microbiology sent. After filtering likely contaminants, 33/276 (12%) returned relevant positive results, of which 20/33 (61%) had antimicrobials changed from empiric therapy as a result with 6/33 (18%) prompting de-escalation. Positive blood and CSF tended to have greater impact than urine or sputum cultures. 124/276 (45%) PEs returned only negative microbiology, and this was documented in the medical notes less often (9/40, 23%) than positive results (28/33, 85%). Out of 40 reviewed PEs with negative microbiology, we identified just one (~ 3%) in which antimicrobials were unambiguously de-escalated following the negative result. CONCLUSIONS: The majority of diagnostic microbiology tests sent to inform clinical management yielded negative results. However, negative microbiology contributed little to clinical decision making about antimicrobial de-escalation, perhaps reflecting a lack of trust in negative results by treating clinicians. Improving the negative predictive value of currently available diagnostic microbiology could help hospital prescribers in de-escalating antimicrobial therapy.