Analysis · Public data

Site count versus enrollment pace: what adding sites actually buys

The standard response to a behind-schedule study is to activate more sites. The registry shows what that purchases: more total patients, and a per-site pace that falls by a factor of twenty-eight.

Rock Enroll editorial · ClinicalTrials.gov, retrieved September 2026

Using the same set as our timeline analysis — completed interventional studies started 2015–2023 — we took the Phase 2, Phase 3 and Phase 2/3 records that report both an enrollment count and at least one listed location, and computed two rates: total patients per month, and patients per site per month.

Three readings, in order of how often people get them wrong

1. Sites do buy throughput. Total pace rises monotonically from 3.06 patients a month at one site to 24.39 at 151 or more. Anyone claiming site expansion does not work is arguing against the data. If you need 800 patients, you cannot get there on twelve sites.

2. It is nothing like linear. Going from 1 site to 151+ multiplies sites by more than 150 and pace by roughly 8. Per-site productivity falls from 3.06 to 0.11 patients per month — a 28-fold collapse. At the top bucket, the median site enrolls one patient every nine months. Most of those sites are consuming activation cost, monitoring visits, regulatory packets and startup time to produce almost nothing.

3. Adding sites does not shorten studies. Median duration is 21 months at one site and 36 months at 151+. Large multi-site studies take longer, not shorter, because activation is serial, the last sites come online late, and the study waits on the tail.

The number to take into the next enrollment meeting

When a CRO proposes activating twenty more sites, the implicit model is usually that each new site performs like your best current one. The registry says the marginal site in a large study enrolls somewhere between 0.11 and 0.25 patients per month. Twenty sites at 0.2 is four patients a month, arriving three to six months from now after startup, at full activation and monitoring cost each.

Put that against the alternative: the same money spent making the sites you already have produce more. A site running at 0.25 patients a month is rarely short of eligible patients in its catchment — it is short of people walking through the door, or short of the coordinator hours to screen them. Those are different problems with different price tags, and neither is solved by adding a twenty-first site with the same constraint.

The decision rule we would use: expand sites when you have evidence the current sites are saturated — their referral pipeline is full and screening capacity is the binding limit. Fix demand when sites are idle, which is the far more common case. Deciding which one you are in is the whole point of a funnel-stage diagnosis, and it is answerable in weeks with data you already hold.