What makes a site, or a country, recruit?
The evidence behind every number on the site: kinds of site and what each country brings, how each number was settled, and the register of risks it all draws on.
Site profiles
Recognisable kinds of site, used to choose sites for a new study and to explain why sites of the same kind differ. Draft profiles, modelled on types of institution rather than named hospitals. The figure on the left is the rate a profile is expected to recruit at, against the country baseline.
×1.00Comprehensive cancer centre 26 sites, ×0.61 of expected
- What it is
- A large university cancer centre in a major city: high patient volume, every service on site, experienced investigators and many trials competing for the same patients.
- Expected rate
- ×1 of the country baseline: academic hospital ×1, track record average ×1.
- Strengths
- Complex procedures and biopsies on site; Large referral catchment; Experienced at amendments and audits.
- Watch for
- Site type: activation speed (S-06, grade A); Site capacity saturation (S-11, grade B); Competing trials (S-21, grade B); Start-up: contracts and budgets (S-12, grade A).
- In each country
- United States: NCI-designated cancer centre; United Kingdom: NHS teaching trust with an NIHR clinical research facility; Spain: University hospital in Madrid or Barcelona; Poland: National or regional oncology institute; Turkey: University hospital in Istanbul or Ankara; India: Tertiary cancer hospital in a metro city.
- Source
- Academic and hospital sites take a median 8.1 to 9.4 months to activate against 4.4 to 4.8 for independent sites (S-06, WCG); competing trials bite where the eligible pool is scarce (S-21); 45% of sites say capacity limits new studies (S-11)
×1.00Academic medical centre 28 sites, ×0.81 of expected
- What it is
- A large university hospital in a major city, outside cancer care: specialist clinics, every service on site, experienced investigators, and many trials competing for the same patients.
- Expected rate
- ×1 of the country baseline: academic hospital ×1, track record average ×1.
- Strengths
- Specialist clinics with large referral catchments; Every procedure the protocol needs on site; Experienced at amendments and audits.
- Watch for
- Site type: activation speed (S-06, grade A); Site capacity saturation (S-11, grade B); Competing trials (S-21, grade B); Start-up: contracts and budgets (S-12, grade A).
- In each country
- United States: University medical centre; United Kingdom: NHS teaching trust with an NIHR clinical research facility; Spain: University hospital in Madrid or Barcelona; Poland: Medical university hospital; Turkey: University hospital in Istanbul or Ankara; India: Tertiary teaching hospital in a metro city.
- Source
- The comprehensive cancer centre profile applied to other therapeutic areas: academic and hospital sites take a median 8.1 to 9.4 months to activate (S-06, WCG); competing trials bite where the eligible pool is scarce (S-21); 45% of sites say capacity limits new studies (S-11)
×1.00Regional teaching hospital 98 sites, ×1.01 of expected
- What it is
- A general hospital with a university link outside the capital: steady patient flow, imaging on site, a small research team shared across specialties.
- Expected rate
- ×1 of the country baseline: academic hospital ×1, track record average ×1.
- Strengths
- Patients not already in competing trials; Imaging on site.
- Watch for
- Site type: activation speed (S-06, grade A); Coordinator turnover (S-09, grade B); Start-up: contracts and budgets (S-12, grade A).
- In each country
- United States: Regional academic medical centre; United Kingdom: District teaching hospital in an NHS trust; Spain: Provincial university hospital; Poland: Voivodeship (regional) specialist hospital; Turkey: Training and research hospital; India: Medical college hospital in a state capital.
- Source
- epiClinical assumption built from the register: hospital activation lag (S-06), coordinator turnover of 10 to 37% rising to 35 to 61% in 2022 (S-09), contracts and budgets the top start-up delay (S-12)
×1.25Dedicated research centre 51 sites, ×0.96 of expected
- What it is
- An independent site or site network whose business is running trials: fast contracts, trained coordinators, patients found through databases and referrals.
- Expected rate
- ×1.25 of the country baseline: dedicated research centre ×1.25, track record average ×1.
- Strengths
- Activates in about half the time of a hospital; Research-focused sites recruit about 41% better.
- Watch for
- Site type: enrolment rate (S-07, grade C); Competing trials (S-21, grade B); Sites declining new studies for lack of capacity (S-26, grade C).
- In each country
- United States: Independent research site or site network; United Kingdom: Commercial clinical research centre; Spain: Private research unit attached to a hospital group; Poland: Private multi-specialty research centre; Turkey: Private hospital clinical research unit; India: Site management organisation (SMO) site.
- Source
- Research-focused sites 41% better enrollers across 72 protocols (S-07); independent sites activate in 4.4 to 4.8 months (S-06); oncology patients still need referral routes, an assumption
×0.85Community hospital 46 sites, ×1.00 of expected
- What it is
- A district general hospital close to where patients live: good for retention, but few research staff and fewer patients who meet strict criteria.
- Expected rate
- ×0.85 of the country baseline: community hospital ×0.85, track record average ×1.
- Strengths
- Patients close to home, which helps retention; Reaches patients academic centres miss.
- Watch for
- Coordinator turnover (S-09, grade B); Investigator attrition and one-and-done investigators (S-08, grade C); Fewer physicians willing to be investigators (S-30, grade C).
- In each country
- United States: Community hospital in an NCORP network; United Kingdom: District general hospital; Spain: Comarcal (district) hospital; Poland: Powiat (county) hospital; Turkey: State hospital; India: District hospital or mid-size private hospital.
- Source
- UK GP and community sites recruit 0.52 to 0.71 against hospital 0.90 to 1.22 per centre per month2; US community enrolment 7.0% of patients against 15.9% at academic centres (D10); single-filing investigators often do one trial only (S-08)
×0.75Community specialist practice 34 sites, ×0.70 of expected
- What it is
- A private specialist practice, in oncology, cardiology or psychiatry, treating many patients with standard care. It is quick to start and close to patients, but complex procedures and imaging happen elsewhere.
- Expected rate
- ×0.75 of the country baseline: private practice ×0.75, track record average ×1.
- Strengths
- Fast contracts; Many patients at the point of diagnosis.
- Watch for
- Protocol complexity and site burden (S-24, grade B); Site capacity saturation (S-11, grade B); Slow site payment and holdbacks (S-17, grade B).
- In each country
- United States: Community oncology or cardiology practice; United Kingdom: Independent sector clinic; Spain: Private specialist clinic; Poland: Private outpatient clinic (NZOZ); Turkey: Private outpatient clinic; India: Standalone specialist clinic.
- Source
- Private practice x0.75, range 0.7 to 0.75 (D10); procedures per protocol up 40.6% since 2010, which weighs on small teams (S-24); slow payments hit small sites hardest (S-17)
×0.75GP practice No sites yet
- What it is
- A family doctor practice in a town or city running smaller, lower-complexity studies: questionnaire-based studies, atopic dermatitis, respiratory and vaccine trials. Common in Europe, far rarer in the US.
- Expected rate
- ×0.75 of the country baseline: private practice ×0.75, track record average ×1.
- Strengths
- Patients with common conditions at the point of care; Long-term relationships that help consent and retention.
- Suited to
- Questionnaire and observational studies; Dermatology, such as atopic dermatitis; Respiratory and vaccine trials.
- Watch for
- Protocol complexity and site burden (S-24, grade B); Fewer physicians willing to be investigators (S-30, grade C); Coordinator turnover (S-09, grade B).
- In each country
- United States: Family medicine practice in a primary care research network (rare); United Kingdom: GP practice in an NIHR research network; Spain: Primary care centre (centro de salud); Poland: Primary care clinic (POZ); Turkey: Family health centre (aile sağlığı merkezi); India: General practitioner clinic in a city.
- Source
- UK GP sites recruit 0.52 to 0.71 per centre per month against hospital 0.90 to 1.222 (D10), so suited to studies that need no hospital procedures; complex protocols weigh on small teams (S-24); Adam Daniels, CRA experience: GP sites commonly run questionnaire and atopic dermatitis studies
×0.64Site new to research 84 sites, ×0.81 of expected
- What it is
- Any setting running its first trial with this sponsor or at all: brings new patients, but needs about ten extra weeks to start and more support.
- Expected rate
- ×0.637 of the country baseline: community hospital ×0.85, track record new site, no history ×0.75.
- Strengths
- Untapped patients; Widens the investigator pool.
- Watch for
- New-to-sponsor site (S-05, grade B); Non-enrolling activated sites (S-01, grade A); Investigator attrition and one-and-done investigators (S-08, grade C).
- In each country
- United States: Hospital joining its first industry trial; United Kingdom: Trust new to commercial research; Spain: Hospital new to the sponsor; Poland: Hospital new to the sponsor; Turkey: Hospital new to the sponsor; India: Hospital new to the sponsor.
- Source
- Sites new to the sponsor take about 10 weeks longer to initiate (35.6 against 25.7 weeks, Tufts; S-05); 11% of activated sites enrol nobody (S-01); x0.75 for no history stays Illustrative (D11)
What each country brings
The structure behind each country's recruitment, and how its sites are doing across the portfolio. Observed ÷ expected pools every open site in the country; the range is where an ordinary group of that size would land.
| Country | Rate against the area baseline | Sites open between | Site cap | Sites judged | Observed ÷ expected |
|---|---|---|---|---|---|
| United States | ×0.75 | month 4.1 to 10.2 | 100 | 116 | ×0.80 0.78 to 1.24 |
| United Kingdom | ×0.85 | month 4 to 10 | 44 | 52 | ×0.81 0.69 to 1.37 |
| Spain | ×1 | month 7.6 to 13.6 | 71 | 68 | ×1.04 0.72 to 1.32 |
| Poland | ×1.4 | month 11.1 to 17.1 | 41 | 63 | ×0.84 0.71 to 1.34 |
| Turkey | ×1 | month 6 to 12 | 18 | 31 | ×1.05 0.62 to 1.47 |
| India | ×1.1 | month 6 to 12 | 15 | 37 | ×1.01 0.65 to 1.41 |
Issues that slow recruitment, and which can be fixed
The evidence document's risk register lists 34 design, site-selection and sponsor issues that slow recruitment. Of these, 27 can be improved, most of them before the first site opens. The rest come with the study's question or with trials in general, and have to be planned for. Grade A is the strongest evidence and D the weakest.
Study design
EB-01Overestimated eligible pool (Lasagna's law)Improvable at design Grade B
- How to improve it
- Check the eligible pool against real-world data before the protocol is final, not against investigators' estimates.
- Effect
- Negative; the commonest cause of recruitment-related discontinuation, first-quarter recruitment slower than planned in 63% of UK trials, and realistic yield quoted second-hand as one-tenth to one-third of the estimate.
- What helped
- Internal pilot with an explicit projection equation and stop-go review (Bertram 2019, descriptive); pre-trial accrual assessment (Schroen 2010: 67% against 47% sufficient accrual); evidence-based feasibility (CTTI 2018, no numbers).
- What did not
- A formal pilot phase as such (McDonald 2006: OR 1.09, no association); time extensions alone (Sully 2013: 55% against 54% reached target).
- Sources
- Briel 2016; McDonald 2006; van der Wouden 2007; Lasagna 1979; Feinstein 2001 via Bogin 2022
PD-01Restrictive eligibility criteria (organ function, brain metastases, prior or concurrent malignancy, HIV, laboratory thresholds, performance status)Improvable at design Grade A
- How to improve it
- Criteria that protect neither the patient nor the result can be removed; the rest are the science.
- Effect
- Negative; eligible pool ×0.48 to 0.53 against broadened criteria, with the hazard ratio unchanged or better.
- What helped
- ASCO-Friends modernised criteria and the FDA 2020 to 2026 guidances; data-driven criteria; adoption of 54 to 93% by criterion in NCI CTEP protocols.
- What did not
- Laboratory and washout exclusions carried forward unchanged across phases and years.
- Sources
- Harvey 2021; Liu 2021; Kim 2017 and 20212; Unger 2019
PD-02Number of eligibility criteria and complexity growthImprovable at design Grade B
- How to improve it
- Each extra criterion narrows the pool; a feasibility review can ask what each one is for.
- Effect
- Negative; more criteria independently associated with poor accrual (p = 0.013); complex trials randomise 59% of screened against about 75%.
- What helped
- Upstream protocol feasibility review and benchmarking against Tufts norms.
- What did not
- not found
- Sources
- Carlisle 2015; Getz 2014; Getz 2023
PD-03Procedural and visit burden (visits, blood draws, questionnaires)Improvable at design Grade B
- How to improve it
- Visits, tests and questionnaires that do not serve an endpoint can go, or move closer to home.
- Effect
- Negative; burden score predicts screen failure, trial duration, and amendments (p < 0.01); oncology plans 29 visits against 16.
- What helped
- Score burden at design; prune non-core procedures, which are 22 to 25% of all procedures.
- What did not
- Sites rate decentralised trials as more burdensome for patient-facing procedures.
- Sources
- Smith 2021; Dirks 2025; Getz 2023
PD-10Run-in period before randomisationImprovable at design Grade C
- How to improve it
- A run-in loses patients before randomisation; keep it only when the question needs it.
- Effect
- Negative for randomised yield; narrows the population and overstates benefit.
- What helped
- Use only where there is a scientific need.
- What did not
- not found
- Sources
- Pablos-Méndez 1998
PD-15Comorbidity exclusionsImprovable at design Grade B
- How to improve it
- Comorbidity exclusions copied from earlier protocols often exclude the patients the drug is for.
- Effect
- Negative; 65.6% of cancer patients have at least one comorbidity; participation OR 0.76.
- What helped
- Remove ASCO-flagged comorbidity restrictions, simulated to add up to 6,317 US registrations a year.
- What did not
- not found
- Sources
- Unger 2019 JAMA Oncology
PD-16Consent document length and reading levelImprovable at design Grade C
- How to improve it
- A shorter, plainer consent document costs nothing and is read more often.
- Effect
- Unclear for enrolment; mean 10.8 pages and Gunning Fog grade 11.9.
- What helped
- Shorter, plain-language materials improve comprehension.
- What did not
- User-tested leaflets did not change recruitment.
- Sources
- Sharp 2004; Beardsley 2007
S-14Protocol amendments before and after activationImprovable at design Grade B
- How to improve it
- Most amendments are foreseeable; a feasibility review before the protocol is final avoids many of them.
- Effect
- Negative on time: 24% of studies amended before activation, a 4 to 6 week delay typical and 10 weeks or more for about 20% of centres, 260 days to implement an amendment, and 82% of Phase III protocols amended.
- What helped
- Finalise the protocol before feasibility. Take site input on design (46% of sites ask for it).
- What did not
- not found
- Sources
- Clinical Trial Vanguard 2024; AACI 2026
PD-05Placebo or no-added-treatment control armCannot be avoided Grade B
- Why it cannot be avoided
- Where regulators need a placebo comparison, the control arm is part of the question.
- Effect
- Negative; willingness 30% against 39%; acceptance 60.5% against 80.6%.
- What helped
- All-active arms; add-on designs; equipoise-focused recruiter training through the QRI.
- What did not
- not found
- Sources
- Welton 1999; Jenkins 2000; Mills 2006; Uzzan 2023
PD-06Blinded placebo-controlled against open designCannot be avoided Grade A
- Why it cannot be avoided
- Blinding protects the result; it is rarely worth trading for faster recruitment.
- Effect
- Negative for blinding; an open design adds 10 percentage points (95% CI 7 to 13), high certainty.
- What helped
- Open or PROBE design where scientifically acceptable.
- What did not
- User-tested information leaflets added 1 point, no effect.
- Sources
- Treweek 2018
PD-11Advanced line of therapyCannot be avoided Grade B
- Why it cannot be avoided
- A later line of therapy means fewer, sicker patients; it comes with the indication.
- Effect
- Negative; 40 to 57% of first-line patients reach second line and attrition is about 60% per line (pool evidence grade B; per-site rate evidence grade D).
- What helped
- Allow any number of prior lines where scientifically justified.
- What did not
- not found
- Sources
- Ran 2025; Sheffield 2026; Lee 2026; Schoenfeld 2025; Osarogiagbon 2021
PD-12Biomarker-selected population with low marker prevalenceCannot be avoided Grade B
- Why it cannot be avoided
- A rare biomarker sets the pool; the fix is more sites and pre-screening, not fewer criteria.
- Effect
- Negative; 37.6% actionable and 17.8% assigned in NCI-MATCH; 39 to 70% amyloid-negative at screening; MASH screen failure 56%.
- What helped
- Central screening at scale (1,117 sites); EHR and registry pre-screening.
- What did not
- not found
- Sources
- Flaherty 2020; Sevigny 2016; Sperling 2020; Souza 2025
PD-13Rare or orphan populationCannot be avoided Grade B
- Why it cannot be avoided
- A rare disease has few patients anywhere; plan for wide geography and long timelines.
- Effect
- Negative; 27 against 64 sites; enrolment 603 against 444 days; completion 50.8% against 72.5%; early termination 13.7% against 6.3%.
- What helped
- Expert-centre networks; single-arm and open designs; natural-history controls.
- What did not
- not found
- Sources
- Getz 2023; Bell 2014
SH-29Standard-of-care changeCannot be avoided Grade B
- Why it cannot be avoided
- A new standard of care can overtake a comparator mid-study; it can be watched for, not prevented.
- Effect
- 11% of non-data terminations on ClinicalTrials.gov came from external information or product withdrawal.
- What helped
- Protocol amendment of the comparator.
- What did not
- not found
- Sources
- Williams et al. 2015
Site selection
S-01Non-enrolling activated sitesImprovable at site selection Grade A
- How to improve it
- Sites that enrol nobody can often be spotted at feasibility: no patients seen, no staff, no track record.
- Effect
- Negative: 11% of activated sites enrol nobody industry-wide (2008 to 2010), zero-patient sites equal 27% of enrolling sites in recent oncology trials, the median benchmark site had zero completers, and the share has not changed in 20 years.
- What helped
- Plan for non-performers explicitly (Phesi: about 150 sites to get 120 enrolling). One sponsor cut non-delivering sites from about a third to 10% with dedicated start-up staff.
- What did not
- Treating sites as interchangeable. Extensions alone did not raise the chance of reaching target.
- Sources
- Tufts CSDD 2013; Silverman 2013; Getz 2012; Phesi 2023; Smith and Getz 2026; Miseta 2017
S-04Past performance as a predictorImprovable at site selection Grade B
- How to improve it
- A site's own record in similar studies predicts better than its feasibility answers.
- Effect
- Positive but modest: Spearman 0.13 to 0.16 for historical rate alone and 0.37 to 0.46 with patient-volume data; well-performing sites have a 70% higher chance of out-performing; trial-manager predictions reach 82% sensitivity but 32% specificity.
- What helped
- Combine historical rate with real-world patient volume, referral flow, and research experience. Guided prediction forms with red flags.
- What did not
- Historical performance alone. Unguided judgement after the initiation visit.
- Sources
- Getz 2011; Hulstaert 2024; Bruhn 2019; Yang 2018; Ninh 2026
S-05New-to-sponsor siteImprovable at site selection Grade B
- How to improve it
- New sites bring new patients but start slower; choose them on purpose and support them.
- Effect
- Negative on time: 28 to 30% of sites are new to the sponsor and initiation takes about 10 weeks longer; sites with 6 to 10 trials' experience are more likely to hit target; no measured rate penalty.
- What helped
- Repeat-site preference where the patient pool allows. Master agreements and central IRB.
- What did not
- not found
- Sources
- Getz 2011; WCG 2018; Getz 2016; Miseta 2017
S-06Site type: activation speedImprovable at site selection Grade A
- How to improve it
- Hospitals activate slower than research centres; the mix sets the start-up curve.
- Effect
- Negative for academic and hospital sites: median 8.1 months versus 4.4 months for independent sites; 9% versus 54% start within 60 days; 21%, 34%, and 80% within 60 days for AMCs, hospital systems, and investigative sites.
- What helped
- Standardised start-up workflow with a site navigator (133 versus 277 days). Master contracts. UK National Contract Value Review (305 to 194 days).
- What did not
- Sequential review layers. Pre-activation amendments, which add 4 to 6 weeks each.
- Sources
- Smith (WCG) 2024; Advarra 2020; WCG 2025; Dilts and Sandler 2006
S-21Competing trialsImprovable at site selection Grade B
- How to improve it
- Ask each site which trials it already runs for the same patients, and avoid saturated pools.
- Effect
- Negative when the eligible pool is scarce: OR 1.88 (1.32 to 2.68) for low accrual per competing trial per 10,000 eligible patients per year; no effect when the pool is large; 3,042 immuno-oncology trials sought 577,076 patients.
- What helped
- Redirect referrals to non-overlapping sites and re-engage the PI (anecdotal). Choose indications and regions with slack.
- What did not
- Adding sites into the same saturated pool.
- Sources
- Bennette 2016; Hutchinson 2022; Tang 2018; ClinOops 2026
S-11Site capacity saturationImprovable at site selection Grade B
- How to improve it
- A site at capacity says yes and then cannot staff the study; ask about staff, not interest.
- Effect
- Negative on available capacity: 45 to 47% of sites say challenges restrict new-study participation and 50% of US sites agreed to fewer studies in 2024.
- What helped
- Fund pre-screening, screen-failure, and start-up costs (site-advocated; no measured enrolment effect).
- What did not
- not found
- Sources
- WCG 2024; WCG 2025
PB-01No trial open at the patient's institution (structural access)Improvable at site selection Grade A
- How to improve it
- Most patients are treated where no trial is open; referral routes reach them.
- Effect
- Removes 55.6% (95% CI 43.7 to 67.3) of cancer patients before any offer; community sites enrol 7.0% of patients against 15.9% at academic centres.
- What helped
- Opening trials at community sites; EHR alerts to find eligible patients multiplied referrals by 10.44 at one site (Embi 2005).
- What did not
- Mass media alone has no low-bias evidence (Cochrane).
- Sources
- Unger 2019 JNCI
PD-08Lack of clinician equipoise and physician gatekeepingImprovable at study conduct Grade B
- How to improve it
- Investigators who doubt the question do not offer the trial; engage them before and during.
- Effect
- Negative; 73% of non-entering surgeons cite the doctor-patient relationship; decliners prefer doctor choice 54% against 39%; comorbid patients are less often offered trials (OR 0.82).
- What helped
- The QRI supported five RCTs to target, with randomisation rates up in three.
- What did not
- QRI-Two performed less well in 6 of 14 start-up trials.
- Sources
- Taylor 1984; Jenkins 2013; Unger 2019; Rooshenas 2019
S-09Coordinator turnoverImprovable at study conduct Grade B
- How to improve it
- Coordinator turnover cannot be prevented, but a handover plan and a back-up keep the study going.
- Effect
- Negative: typical turnover of 10 to 37% rose to 35 to 61% in 2022, Duke's voluntary rate runs 15 to 19%, 51% of sites reported under 10% in 2024, SCRS estimates 6 to 12 months to get a study back on track, and a vendor reports a six-week enrolment pause and 3 to 7 lost enrolments per affected study.
- What helped
- Study continuity plans, renegotiated budgets, flexible schedules, and career paths. Capacity-pod staffing (vendor claim, 41% less disruption).
- What did not
- Matching sponsor and CRO salaries, which sites say they cannot do.
- Sources
- WCG 2024; Stroo 2025; SCRS 2023; Bourne 2024; Clinitiative 2026; Zoster Eye Disease Study 2024
PM-07Site recruitment rates are highly heterogeneous within a trialCannot be avoided Grade B
- Why it cannot be avoided
- Sites of the same kind recruit at very different rates; plan with the spread, not the average.
- Effect
- Widens the forecast band: gamma shape 1.75 to 2.15 in myeloma trials (coefficient of variation about 0.7), a simulation convention of CV 1.2 (shape about 0.69), and a fit of about 0.4 to 0.5 to Phesi's cancer-trial distribution.
- What helped
- not found
- What did not
- not found
- Sources
- Minois 2017; Urbas 2020; Anisimov 2024; Phesi 2023
Sponsor and CRO
S-16Feasibility over-optimismImprovable at design Grade A
- How to improve it
- Feasibility answers are optimistic; discount them with history, and plan in ranges.
- Effect
- Negative: 31 to 56% of trials reach target, 45 to 53% are extended, 81.5% of NCI trials miss the projected accrual window, regional delivery runs 43 to 104% of forecast, and one pilot delivered 44% of plan.
- What helped
- Internal pilots with monthly re-forecast from screening logs. Pre-trial accrual assessment. Stop or go milestones.
- What did not
- Unverified questionnaires. Extensions without any change in approach.
- Sources
- Altair 2015; McDonald 2006; Sully 2013; Walters 2017; Bertram 2019; Cheng 2011
PD-21No upstream protocol feasibility reviewImprovable at design Grade B
- How to improve it
- A protocol feasibility review before the protocol is final catches burden and amendments early.
- Effect
- Negative; 23 to 45% of amendments avoidable and about 40% occur before first patient first dose.
- What helped
- Cross-functional governance review; the CTTI three-area framework; critical-to-quality factors at planning.
- What did not
- not found
- Sources
- Getz 2014, 2016, and 202423; Huang 2018; ICH E8(R1)
PD-18No patient and public involvement in designImprovable at design Grade A
- How to improve it
- Patients asked during design point out the visits and criteria that will put people off.
- Effect
- Positive when present; enrolment OR 1.16 (1.01 to 1.34); lived-experience involvement OR 3.14, exploratory.
- What helped
- Patient and public involvement in protocol design.
- What did not
- Retention was not improved (OR 1.16, 0.33 to 4.14).
- Sources
- Crocker 2018
FM-06Sites needed derived by a deterministic rule with no confidence levelImprovable at design Grade B
- How to improve it
- Sites needed from a single-number rule carries no confidence; simulate the range instead.
- Effect
- Target divided by expected patients per site plans to roughly an even chance of finishing on time; Carter's unconditional estimate was 18 months against a simulated 23.2 months (SD 1.0), and the probabilistic site count rises from 40 to 49 as confidence rises from 0.5 to 0.9.
- What helped
- The least N with P(T(n, N) ≤ T) ≥ p via the negative binomial identity, the normal criterion, or simulation to a threshold.
- What did not
- Expected-value site counts such as Shipes's 197 sites and Carter's unconditional 18 months.
- Sources
- Carter, Sonne, and Brady 2005 [23]; Shipes et al. 2023 [16]; Anisimov and Austin 2022 [5]
S-12Start-up: contracts and budgetsImprovable at site selection Grade A
- How to improve it
- Contracts and budgets are the commonest start-up delay; templates and fair-market budgets shorten them.
- Effect
- Negative on time: the top delay cause for 73% of sites, 63% say fee negotiation holds activation, identification to initiation has sat at about 31 weeks for a decade, and 40% of engaged North American sites never activate.
- What helped
- Master agreements, standard clauses, and national model contracts (UK NCVR cut 111 days). Dedicated start-up staff.
- What did not
- Slow sponsor budget responses, reported by 50% of sites.
- Sources
- WCG 2018; Miseta 2017; Clinical Trial Vanguard 2024; Advarra 2020; WCG 2025
S-15Sequential or late site activationImprovable at site selection Grade B
- How to improve it
- Opening sites one after another wastes the recruitment window; open them in parallel where possible.
- Effect
- Negative: only 6% of trials activate 50 or more sites within 100 days, the average Site Effectiveness Index is about 50%, linear Phase II to III site scaling cost 5 to 21 months, 30% of sites activated in the last six months were mostly non-performing, and sites starting more than 5 months after the first patient supplied under 10% of enrolment in 11 of 14 large trials.
- What helped
- Staged 70/30 activation based on first-batch performance. Batch activation with surplus sites planned for non-performers.
- What did not
- Proportional site additions. Extensions. Adding sites many months into enrolment.
- Sources
- Clinical Trial Vanguard 2024; Li (Phesi) 2018; Haidich and Ioannidis 2003; Getz, Smith, Peachey, and Li 2022
S-17Slow site payment and holdbacksImprovable at study conduct Grade B
- How to improve it
- Slow payments hit small sites hardest and cost goodwill; pay on time.
- Effect
- Negative, self-reported for enrolment: 83% of sites want payment within 30 days, 37% of North American sites were paid after 90 days (2012), 23% of sites have 31 to 40% of invoices over 90 days late, one-third have receivables over 90 days, 50% have three months or less of cash, and sites say slow payment reduces screening and enrolment effort.
- What helped
- Monthly electronic payment, no holdback, and automated pass-through invoicing (site-preferred; effect self-reported).
- What did not
- Quarterly payment and 10 to 14% holdbacks.
- Sources
- Miseta 2016; SCRS 2023 Site Landscape; Scout 2025; Bourne 2024; SCRS and Greenphire 2017
S-18Technology burdenImprovable at study conduct Grade B
- How to improve it
- Each extra system and log-in takes coordinator time away from patients.
- Effect
- Negative on site time, unquantified for enrolment: 60% of sites use 20 or more systems daily, about 10 to 12 applications per site in 2015, 46% say too many separate technologies and 74% cite login tracking (2020), 40% of sites spend 5 to 15 hours per month per study on unpaid training, and 19% rank sponsor technology a top challenge.
- What helped
- Accept site-preferred systems. Consistent technology across a sponsor's studies. Paid training.
- What did not
- not found
- Sources
- WCG 2024; SCRS 2023 Site Landscape; Bechtel 2024; Florence 2024; SCRS 2020
S-10CRA turnover and monitoring continuityImprovable at study conduct Grade C
- How to improve it
- A change of CRA resets the relationship with the site; keep continuity where possible.
- Effect
- Negative, unquantified for enrolment: CRA turnover 28 to 32% in 2021 to 2022, nine months to proficiency, and sites deprioritise studies when the sponsor is absent (qualitative).
- What helped
- Mentoring, flexible schedules, and pay.
- What did not
- not found
- Sources
- Parrish and Minns 2025; Lamberti 2022
S-22Rescue by adding sitesImprovable at study conduct Grade B
- How to improve it
- Sites added as a rescue open late; plan enough sites at the start.
- Effect
- Weak to negative: extensions are no more likely to succeed, timelines nearly double, STICH grew from 32 to 171 sites with 44 zero-enrollers, late-starter sites supplied under 10% of enrolment in 11 of 14 large trials, and new sites take 4 to 8 months to activate plus about 1 to 3 months to first patient.
- What helped
- Early decision. Staged activation. Poisson-gamma computation of the sites needed at a chosen confidence.
- What did not
- Late, proportional site additions. Adding sites many months after the start.
- Sources
- Tufts CSDD 2013; Smith (WCG) 2024; CRIO 2026; Sully 2013; Fogel 2018; Li (Phesi) 2018; Haidich and Ioannidis 2003
Whether each issue is unavoidable or improvable: epiClinical classification of risk-register entries; each entry's evidence grade and sources are the register's.
Each study's design, plan, sites, KPIs and QTLs are in its workspace, reached from the studies page.