Antimicrobial Stewardship Programme Setup for a 200-Bed Hospital

The core team in a hospital this size is usually five people, none of them full-time on stewardship. A clinical champion, typically an intensivist or a physician with credibility among prescribers, carries the conversations that a pharmacist cannot. The clinical microbiologist owns the antibiogram and the interpretation rules. A designated pharmacist does the daily work: pulling the list, checking doses, tracking consumption. The infection control nurse links stewardship to device and surgical site data. An administrator with budget authority makes decisions stick when a consultant escalates.

Be specific about hours rather than titles. The pharmacist needs a protected block each working morning, not spare time between dispensing peaks. The microbiologist needs a fixed half-day a month for antibiogram and policy work. The champion needs a weekly slot to walk the ICU and the two heaviest medical wards. Write these into duty rosters and job descriptions, because a stewardship role that exists only in a committee charter is the first thing dropped when the pharmacy is short-staffed.

The trade-off nobody warns you about is authority. A pharmacist-led programme is the affordable model and it works, but only where prescribers accept the recommendation as clinical rather than administrative. Early on, route contested recommendations through the physician champion instead of letting a junior pharmacist argue with a senior consultant in a corridor. That costs the champion time in the first quarter and saves the programme from being written off as pharmacy interference, which is very hard to recover from.

An antibiotic policy is a set of syndrome pages, not a list of drugs. Community-acquired pneumonia, hospital-acquired pneumonia, complicated urinary tract infection, skin and soft tissue infection, febrile neutropenia, intra-abdominal sepsis and surgical prophylaxis will cover the large majority of prescribing in a general hospital. Each page should be short enough to read on a phone at the bedside. If a page runs past one screen, prescribers will fall back on habit, and habit in most Indian hospitals means a carbapenem or a piperacillin-tazobactam started because it feels safe.

Every syndrome page needs the same six fields: the empirical first choice, an alternative for documented beta-lactam allergy, the dose with renal adjustment, the intended total duration, the de-escalation trigger, and the instruction to draw cultures before the first dose. The duration field does more good than any other. A policy that says five days for uncomplicated community-acquired pneumonia, with a named reviewer at day five, changes more consumption than a restriction list, because most excess use is not the wrong drug but the right drug continued too long.

The honest problem is that a first-year programme rarely has enough local susceptibility data to justify its own empirical choices. The workable compromise is to publish an interim policy built on ICMR antimicrobial resistance surveillance network data and your state or regional reports, label it clearly as interim, and commit to a revision date once your own cumulative antibiogram covers twelve months. Do not quietly present borrowed data as local. Prescribers notice, and the credibility of the whole policy rests on it being visibly your hospital's numbers.

Sort the formulary into three tiers rather than two. Unrestricted agents any prescriber may start. Monitored agents that anyone may start but that trigger a stewardship review at day three. Restricted agents that need pre-authorisation. The WHO AWaRe classification is a reasonable starting frame, with most Access agents unrestricted, Watch agents monitored, and Reserve agents such as colistin, polymyxin B, ceftazidime-avibactam, daptomycin and linezolid restricted. Keep the restricted list short. Ten molecules you genuinely control beats thirty on paper.

The mechanics matter more than the list. Authorisation must be reachable at every hour, which in practice means a named on-call number, a documented verbal approval with a written note entered the same shift, and an automatic stop at 72 hours unless the approval is renewed with a reason. Out-of-hours starts that could not reach an approver should route to post-prescription review the next morning rather than being blocked. Build the stop dates into the medication order itself so the control does not depend on someone remembering.

The rule that keeps this safe is absolute: approval never gates the first dose in suspected sepsis or septic shock. Delay in effective therapy is a mortality risk, and a stewardship control that introduces it has caused more harm than the resistance it prevents. Say this in the policy in one sentence, in bold, and train the nursing staff on it, because nurses are the people who will otherwise hold a syringe waiting for a call-back. The other predictable failure is the rubber stamp, where approval becomes a phone formality; auditing a sample of approvals against the documented indication is the only fix.

Prospective audit with intervention and feedback is the intervention with the best evidence behind it, and it is also the one most hospitals implement as a paperwork exercise. Done properly it is a conversation. The pharmacist pulls the list of patients on day three of a Watch or Reserve agent, reviews the culture result, the clinical trajectory and the renal function, and then goes to find the treating team. The recommendation is made verbally, at the bedside or in the doctors' room, and only then written in the file.

Written recommendations left in the case sheet have a low acceptance rate because nobody is obliged to read them and no one is embarrassed to ignore them. Face-to-face recommendations get accepted far more often, and the difference is not clinical, it is social. Track acceptance as a named metric from month one: how many recommendations were made, how many were accepted within 24 hours, how many were declined and for what stated reason. Declined recommendations with a good clinical reason are a healthy sign; declined with no reason recorded is a governance problem.

Feedback should run at two levels. Individual feedback happens in the moment, on the ward. Aggregate feedback goes to the department head monthly, showing that unit's consumption and acceptance alongside the hospital position, without naming individual prescribers in a circulated document. Peer comparison works; public shaming produces under-documentation and a quiet refusal to co-operate. Where prescribing is captured electronically, a system such as HealUDoc can generate the day-three review list automatically instead of the pharmacist rebuilding it from dispensing records every morning.

Two consumption metrics are worth maintaining. Defined daily doses per 100 bed days uses the WHO ATC/DDD assigned dose as a denominator and is easy to derive from purchase or dispensing data. Days of therapy per 1000 patient days counts each calendar day on which a patient received at least one dose of a given agent, regardless of dose size. Both are proxies for exposure, and neither tells you whether the prescribing was appropriate. Do not present either as a quality outcome on its own.

The distinction between them is practical, not academic. DDD distorts wherever the actual dose differs systematically from the WHO assigned dose, which is exactly what happens in paediatrics, in renal impairment, and in the higher-dose regimens used for resistant organisms. A paediatric ward can show a flattering DDD figure while treating a large number of children, simply because each child receives a fraction of an adult daily dose. DOT is harder to compute because it needs administration-level data, but it survives those distortions and is the better measure if you can build it.

Be honest about outcome measures. At 200 beds you will not detect a mortality effect, and the resistance trend in your own antibiogram moves too slowly and with too much noise to attribute to the programme in year one. What you can measure credibly is process: policy concordance of empirical therapy, the proportion of positive cultures where therapy was de-escalated within 48 hours of the report, mean duration for a named syndrome, and consumption of your top three restricted agents. Those move within months and they are defensible in front of a committee.

The first thirty days are entirely preparatory and should not involve telling anyone what to prescribe. Constitute the team with named people and rostered hours, get the administrator to sign the charter, and pull twelve months of microbiology and dispensing data to see what you are dealing with. Identify your three heaviest consuming units and your top five agents by volume and by cost. Baseline measurement happens now, before any intervention, because a programme with no baseline can never demonstrate that it did anything.

Days thirty-one to sixty are for the policy and the restriction list. Draft the syndrome pages with the champion and the microbiologist, circulate them to department heads for comment, and hold one meeting where objections are argued out rather than emailed. Agree the restricted list and the approval route in the same meeting, and publish both together. Train nursing and pharmacy staff on the first-dose rule. This month produces documents; resist the urge to start auditing before prescribers have seen the policy they will be audited against.

Days sixty-one to ninety are the first live cycle. Start prospective audit on one unit, usually the ICU, and hold it there for a full month before extending. Report the first month's process metrics to the committee with the baseline alongside. Expect the acceptance rate to be poor at first and expect at least one senior clinician to test whether the restriction is real; how that first confrontation is handled sets the programme's authority for the following two years. Extend to a second unit only when the first one runs without the champion in the room.