Antibiotic stewardship metrics, and what a program actually measures

Antibiotic stewardship metrics fall into three groups: how much antibiotic is used, how appropriately it is used, and what the organisms are doing in response. Days of therapy and defined daily doses measure the first, audit and review measure the second, and the cumulative antibiogram measures the third.

Days of therapy, the working unit of consumption

Days of therapy, usually written DOT, counts one day for each antibiotic a patient receives on a given day, regardless of dose. A patient on two agents for three days contributes six days of therapy. It is normalised per 1,000 patient days present, and it is the standard consumption measure in United States hospital stewardship.

Its strength is that it does not depend on dose, so renal adjustment and paediatric dosing do not distort it. Its weakness is the same thing: a patient on a low dose and a patient on a high dose look identical, and combination therapy inflates the count even when it is the right decision.

Defined daily doses, and why the two disagree

The defined daily dose, or DDD, is the assumed average maintenance dose per day for an adult using the drug for its main indication, as maintained by the World Health Organization Collaborating Centre for Drug Statistics Methodology. Consumption is total grams divided by the DDD.

It is the international comparison unit and it is genuinely useful across countries. In a hospital it has a specific failure mode: any population that is dosed away from the assumed adult dose, notably paediatrics and renal impairment, produces a distorted number. If your DOT and your DDD trends disagree, look at the dosing before you look for a stewardship explanation.

SAAR, the risk-adjusted comparison

The standardised antimicrobial administration ratio, published through the antimicrobial use option of the CDC National Healthcare Safety Network, divides observed antibiotic use by predicted use, where the prediction is modelled from the characteristics of participating locations.

A ratio above one means more use than the model predicts for a unit like yours. It is the closest thing to a fair between-hospital comparison, and it deserves the standard caution that comes with any risk adjustment: it is only as good as the model and the accuracy of your location mapping. A SAAR that moves sharply is more often a mapping change than a prescribing change.

Process measures, which are where behaviour actually changes

These are harder to collect than consumption, and they are the ones that describe what the program does rather than what the hospital consumes.

• Time to appropriate therapy for bloodstream infection, measured from culture collection
• Time to de-escalation once susceptibility results are final
• Rate of intravenous to oral conversion where the agent has good oral availability
• Documented indication and planned duration at the point of prescription
• Acceptance rate of stewardship recommendations, which measures the program as much as the prescriber

Resistance metrics, and the trap in them

The cumulative antibiogram is the resistance side of the report, and it is a lagging measure of prescribing, not a direct one. Susceptibility moves for reasons that have little to do with your program: referral patterns, a change in the population served, a new long term care relationship, a breakpoint revision.

The honest way to report it is with the method attached. A percentage that moved three points is not a finding until you can say that the de-duplication rule, the reporting threshold and the breakpoint version were the same in both periods. That is the entire reason a rule receipt belongs under every matrix.

A reporting set that fits on one page

• Total days of therapy per 1,000 patient days present, with a trend line
• Days of therapy for the two or three agents your program is targeting this year
• The SAAR for your reporting locations, if you submit to NHSN
• The cumulative antibiogram for the period, with the rule receipt
• Movement in the organisms on your watch list, as counts rather than percentages

What not to report

Avoid any single number that claims to summarise stewardship quality. There isn't one, and inventing one invites a committee to manage the number instead of the prescribing. Avoid cost per patient day as a headline, because it moves with drug pricing and contract cycles more than with practice. And avoid resistance percentages computed on cells too small to report, which is the most common way a stewardship report loses its credibility with the microbiology laboratory.

---

Related on this site: the accreditation artefacts, building the cumulative report.