epiClinicalA fictional CRO

epiClinical is fictional. It is a portfolio demonstration by Adam Daniels. Its studies, sponsors, sites and enrolment figures are synthetic. The evidence behind its assumptions is real, and every number says how far it can be trusted.

Risk monitor

What could slow our studies down, and where?

The risks across every running study: recruitment, the KRIs and QTLs that watch it, and the regional shocks that can hit it from outside.

Real events from 2013 to 2025, read against today's fictional studies: which countries to watch, which sites each event reaches, and what past shocks taught.

Where to look on March 12th 2020

Time to act somewhere

Early-warning score by study country0 to 100, on March 12th 2020

Show data
Early-warning score by country on March 12th 2020
CountryScoreLevel
United States100Act
United Kingdom100Act
Spain100Act
Poland100Act
Turkey100Act
India100Act

Thresholds: Watch 30, Act 60. No published calibration exists; to be tuned by a backtest against Medidata’s monthly series that reports lead time against false alarms (Medidata Releases 2.0 and 9.02; D42).

What moves each country's score on March 12th 2020, and the studies with sites there
CountryScoreDriven byStudies with sites there
United States100 ActWHO declares COVID-19 a pandemic (+49); Italy locks down towns in Lombardy and Veneto as COVID-19 cases rise (+32); WHO declares the coronavirus outbreak a public health emergency of international concern (+18)EPI-118 (60 sites), EPI-214 (40 sites), EPI-301 (50 sites), EPI-405 (6 sites)
United Kingdom100 ActWHO declares COVID-19 a pandemic (+49); Italy locks down towns in Lombardy and Veneto as COVID-19 cases rise (+32); WHO declares the coronavirus outbreak a public health emergency of international concern (+18)EPI-118 (25 sites), EPI-214 (15 sites), EPI-301 (25 sites), EPI-405 (4 sites)
Spain100 ActWHO declares COVID-19 a pandemic (+49); Italy locks down towns in Lombardy and Veneto as COVID-19 cases rise (+32); WHO declares the coronavirus outbreak a public health emergency of international concern (+18)EPI-118 (25 sites), EPI-214 (20 sites), EPI-301 (35 sites), EPI-405 (3 sites)
Poland100 ActWHO declares COVID-19 a pandemic (+49); Italy locks down towns in Lombardy and Veneto as COVID-19 cases rise (+32); WHO declares the coronavirus outbreak a public health emergency of international concern (+18)EPI-118 (25 sites), EPI-214 (20 sites), EPI-301 (25 sites), EPI-405 (2 sites)
Turkey100 ActTurkey begins a military operation in north-east Syria (+60); WHO declares COVID-19 a pandemic (+49); Italy locks down towns in Lombardy and Veneto as COVID-19 cases rise (+32)EPI-214 (10 sites), EPI-301 (12 sites), EPI-405 (1 site)
India100 ActWHO declares COVID-19 a pandemic (+49); Italy locks down towns in Lombardy and Veneto as COVID-19 cases rise (+32); WHO declares the coronavirus outbreak a public health emergency of international concern (+18)EPI-214 (15 sites), EPI-301 (12 sites), EPI-405 (2 sites)

The newsfeed

Six out of 17 events had happened by March 12th 2020; one waits for review.

  1. March 11th 2020All study countriesPublic healthHeadwind

    WHO declares COVID-19 a pandemic

    Where
    Global
    Reaches
    All 432 open and starting sites, in every study.
    Effect on recruitment
    ×0.5 at the sites it reaches (illustrative).
    Score on March 12th 2020
    +49.4 to every country.
    Source
    WHO 2020.

    Drafted recommendation. Activate the continuity plan: move to remote monitoring, protect enrolled patients first, expect screening to fall in every country, and brief the sponsor on a re-forecast with ranges.

    Confidence: high. Reviewer's decision: Accepted.

  2. February 23rd 2020All study countriesPublic healthHeadwind

    Italy locks down towns in Lombardy and Veneto as COVID-19 cases rise

    Where
    Lombardy and Veneto, Italy (outside the study)
    Reaches
    All 432 open and starting sites, in every study.
    Effect on recruitment
    ×0.8 at the sites it reaches (illustrative).
    Score on March 12th 2020
    +32.5 to every country.
    Source
    Decree-law 6 of 23 February 2020 (Gazzetta Ufficiale 2020).

    Drafted recommendation. Brief every site on remote-visit procedures, prepare country-by-country re-forecast scenarios, and front-load screening at sites that are still unaffected.

    Confidence: medium.

  3. January 30th 2020All study countriesPublic healthHeadwind

    WHO declares the coronavirus outbreak a public health emergency of international concern

    Where
    Global
    Reaches
    All 432 open and starting sites, in every study.
    Effect on recruitment
    ×0.95 at the sites it reaches (illustrative).
    Score on March 12th 2020
    +18.5 to every country.
    Source
    WHO 2020.

    Drafted recommendation. Check drug-supply chains that run through China and stock levels at every depot; confirm which countries can switch to direct-to-patient shipment.

    Confidence: medium. Reviewer's decision: Accepted.

  4. January 23rd 2020All study countriesPublic healthHeadwind

    Wuhan is placed under lockdown as a novel coronavirus spreads

    Where
    Wuhan, China (outside the study)
    Reaches
    All 432 open and starting sites, in every study.
    Effect on recruitment
    ×0.95 at the sites it reaches (illustrative).
    Score on March 12th 2020
    +17.0 to every country.
    Source
    Tian et al. 2020 (Wuhan’s travel shutdown began on 23 January).

    Drafted recommendation. Start contingency planning now: confirm remote-visit options, direct-to-patient drug shipment and a three-month drug buffer at every site, and agree a re-forecast trigger with the sponsor.

    Confidence: low. Reviewer's decision: Accepted.

  5. October 9th 2019TurkeyConflict & securityHeadwind

    Turkey begins a military operation in north-east Syria, across the border from the study’s south-eastern sites

    Where
    Gaziantep, near the Syrian border
    Reaches
    EPI-214: one out of ten sites in Turkey; EPI-301: one out of 12 sites in Turkey.
    Effect on recruitment
    ×0.3 at the sites it reaches (illustrative).
    Score on March 12th 2020
    +59.6 to Turkey.
    Source
    UN Special Advisers 2019 (the operation began on 9 October). Turkey is among the countries whose trials are most exposed to conflict in the region (Phesi 2026).

    Drafted recommendation. Pause new screening at the Gaziantep sites for four weeks; confirm the safety and dosing of every enrolled patient and check drug-supply routes; offer travel support to the Adana sites for patients who want to move. Keep Turkey’s country target; reassess in two weeks.

    Confidence: medium. Reviewer's decision: Accepted.

  6. September 2013IndiaRegulatoryHeadwind

    India’s Supreme Court stops approvals of new-drug trials until a monitoring mechanism is in place; 162 approved trials are put on hold

    Where
    National
    Reaches
    EPI-214: 15 out of 15 sites in India; EPI-301: 12 out of 12 sites in India; EPI-405: two out of two sites in India.
    Effect on recruitment
    ×0.3 at the sites it reaches (illustrative).
    Score on March 12th 2020
    +1.0 to India.
    Source
    Nair 2015: trial approvals fell from 253 in 2012 to 73 in 2013.

    Drafted recommendation. Pause plans for new Indian sites; for running studies, confirm each site’s approval status with the regulator and move India’s open target to other countries until the new rules are in force.

    Confidence: high. Reviewer's decision: Accepted.

Would the score have warned us?

Each chart has two lines on one 0 to 100 scale. Screening, in colour, is a percentage of plan; the week it falls below 80 is the hit. The early-warning score, in grey, rises with the news: it puts the country on Watch at 30 and calls for action at 60. A warning works when Watch and Act come before the hit; the markers along the top show when each came. Each replay scores the events above, with your decisions, week by week.

United Kingdom: COVID-19, 2020

Watch came seven weeks and Act four weeks before screening fell.Screening as % of plan and the early-warning score, by week

Show data
United Kingdom: COVID-19, 2020: by week
Week ofScreening, % of planScoreBasis
30 Dec 20191000Synthetic shape: Medidata’s monthly UK series starts in April
6 Jan 20201010Synthetic shape: Medidata’s monthly UK series starts in April
13 Jan 2020990Synthetic shape: Medidata’s monthly UK series starts in April
20 Jan 202010029Synthetic shape: Medidata’s monthly UK series starts in April
27 Jan 20209856Synthetic shape: Medidata’s monthly UK series starts in April
3 Feb 202010051Synthetic shape: Medidata’s monthly UK series starts in April
10 Feb 20209947Synthetic shape: Medidata’s monthly UK series starts in April
17 Feb 202010184Synthetic shape: Medidata’s monthly UK series starts in April
24 Feb 202010077Synthetic shape: Medidata’s monthly UK series starts in April
2 Mar 20209771Synthetic shape: Medidata’s monthly UK series starts in April
9 Mar 202090100Synthetic shape: Medidata’s monthly UK series starts in April
16 Mar 202055100Synthetic shape: Medidata’s monthly UK series starts in April
23 Mar 202020100Synthetic shape: Medidata’s monthly UK series starts in April
30 Mar 202014100Synthetic shape: Medidata’s monthly UK series starts in April
6 Apr 202013100Synthetic shape: Medidata’s monthly UK series starts in April
13 Apr 202013100Medidata’s April figure for the UK: new patients 87% below 2019
20 Apr 202014100Between Medidata’s April and May figures
27 Apr 202014100Between Medidata’s April and May figures
4 May 202014100Between Medidata’s April and May figures
11 May 202015100Medidata’s May figure for the UK: new patients 85% below 2019
18 May 20201794Between Medidata’s May and June figures
25 May 20202087Between Medidata’s May and June figures
1 Jun 20202280Between Medidata’s May and June figures
8 Jun 20202574Between Medidata’s May and June figures
15 Jun 20202668Medidata’s June figure for the UK: new patients 74% below 2019
22 Jun 20203163Between Medidata’s June and July figures
29 Jun 20203458Between Medidata’s June and July figures
6 Jul 20203754Between Medidata’s June and July figures
13 Jul 20204049Medidata’s July figure for the UK: new patients 60% below 2019
20 Jul 20204546Between Medidata’s July and August figures
27 Jul 20205042Between Medidata’s July and August figures
3 Aug 20205439Between Medidata’s July and August figures
10 Aug 20206036Medidata’s August figure for the UK: new patients 40% below 2019
17 Aug 20206033Held at Medidata’s August figure, the last in the series
24 Aug 20206030Held at Medidata’s August figure, the last in the series
31 Aug 20206028Held at Medidata’s August figure, the last in the series

Dots are Medidata’s monthly figures, with the weeks between interpolated: Medidata Releases 6.0 and 9.02. Synthetic shape: Medidata’s monthly UK series starts in April.

Turkey south-east sites: earthquake, 2023

Act came only in the week screening fell: there was nothing to warn of.Screening as % of plan and the early-warning score, by week

Show data
Turkey south-east sites: earthquake, 2023: by week
Week ofScreening, % of planScoreBasis
26 Dec 20221009Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
2 Jan 20231028Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
9 Jan 2023998Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
16 Jan 20231018Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
23 Jan 20231008Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
30 Jan 2023988Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
6 Feb 202315100Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
13 Feb 2023594Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
20 Feb 2023887Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
27 Feb 20231281Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
6 Mar 20231875Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
13 Mar 20232570Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
20 Mar 20233265Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
27 Mar 20233860Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
3 Apr 20234456Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
10 Apr 20235052Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
17 Apr 20235548Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
24 Apr 20235845Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
1 May 20236142Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
8 May 20236439Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
15 May 20236637Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
22 May 20236834Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
29 May 20237032Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
5 Jun 20237130Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
12 Jun 20237228Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
19 Jun 20237326Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
26 Jun 20237425Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)
3 Jul 20237423Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45)

Synthetic: no trial-level recruitment data for the 2023 earthquake was found in English or Turkish sources (D45).

How each kind of news is scored

How fast each category fades and how far it reaches. The weights are assumptions to be tuned against past events; the pace of fading has more evidence behind it.

Half-life and weights by distance, by category of news
CategoryHalf-life, daysWeight by distance from our sitesWhy
Site regionNationalNeighbouring countryGlobal
Public health 6010.90.70.5Half-life: The pandemic’s hit to new patients per site halved in about two months (Medidata Release 9.0; D40). Weights: Distance barely weakens the signal: WHO’s global declarations came weeks before recruitment collapsed country by country (WHO 2020; Medidata Release 2.0; D41).
Conflict & security 36510.50.20Half-life: Ukraine’s recruiting sites were 48% down at seven months and 62% down at 13 months, with no recovery (Phesi via Lab News 2022 and 20232; D40). Weights: The effect falls steeply with distance, and next door it can be a gain: Poland’s recruiting sites rose 10% (Clinical Research News 2022; Phesi via Lab News 2022; D41).
Natural disaster 6010.30.10Half-life: No study measured how quickly recruitment recovers after a disaster; 60 days is an epiClinical assumption. Weights: The damage is local: hospitals near the epicentre were lost while those further away kept working (IFRC 2023; D41).
Health system capacity 2110.800Half-life: Strikes and cyberattacks fade in two to four weeks: hospital activity dipped on strike days and recovered after them (Wyatt et al. 2026; Borchert et al. 2025; D40). Weights: A strike reaches the sites of the health system on strike, and no further; the weights are epiClinical assumptions.
Policy & funding 365 from the effective date10.800Half-life: Funding cuts and new laws act for a year or more, and a law counts from the day it takes effect, however old the news (D40, D43). Weights: A national law or funder reaches that country’s sites; the weights are epiClinical assumptions.
Regulatory 365 from the effective date10.900Half-life: A regulator’s or court’s rule stays live from publication until it takes effect, then fades over a year or more (D40, D43). Weights: A national rule applies to every site in the country; the weights are epiClinical assumptions.
Trust & access 36510.600Half-life: Mistrust is a slow headwind: the effect of the Tuskegee study on Black men’s use of care lasted decades (Alsan and Wanamaker 2018; D43). Weights: The weights are epiClinical assumptions.
Population movement 6010.80.40Half-life: People who move need continuity of care, not a change to the forecast, so these events carry no score; the half-life is an epiClinical assumption. Weights: The weights are epiClinical assumptions.
  • Tailwind factor 0.2. A favourable event offsets risk at a fifth of its weight. One measured case: Poland’s recruiting sites rose 10% while Ukraine’s fell 48% (Phesi via Lab News 2022; D41).
  • Watch threshold 30. Score at which the country goes on watch. No published calibration exists; to be tuned by a backtest against Medidata’s monthly series that reports lead time against false alarms (Medidata Releases 2.0 and 9.02; D42).
  • Act threshold 60. Score at which a strategy change is recommended. No published calibration exists; to be tuned by the same backtest (Medidata Releases 2.0 and 9.02; D42).
  • Impact visible below 80. Screening index (plan = 100) treated as a visible hit. A convenient rule that the UK met in the first month of the pandemic, when new patients fell more than 20% below 2019 in March 2020 (Medidata Release 2.0; D42).

Three shocks, and what they taught

Rare events that stopped or slowed recruitment, told for what they teach if something like them happens again. The events are real; the studies they touch here are fictional.

The COVID-19 pandemic

2020 to 2022

In spring 2020 the first lockdowns closed clinics and moved research staff to COVID-19 care. New patients per site fell by more than half worldwide in the worst month. Recovery came in months, with reversals in each new wave, and the shortage of research staff it left lasted for years.

Pandemic wave, first national lockdown
Headwind: new patients per site fell 59% globally in the trough month against the 2019 baseline (UK −87%, India −98%, US −45%, Spain −60%, France −77%, Germany −68%).What helped: Remote visits, direct-to-patient IMP, pausing non-essential screening, and protecting enrolled patients; oncology continued. What did not: Remote SDV (regulators diverged and sponsors reverted within months); the UK's non-COVID triage cost 44% of commercial recruitment. (SH-01, grade A)
Pandemic recovery and second waves
Partial recovery over 3 to 5 months with reversals: global −10% in July and −20% in August 2020; UK −40%, India −71%, US −22%, Spain +25% in August.What helped: Phased restart frameworks (NIHR levels); reducing recruitment before pausing. What did not: Staff attrition prolonged recovery; the winter wave cut SWOG enrolment 2% a week. (SH-02, grade A)
Therapeutic-area sensitivity to a pandemic
Oncology least hit (April −41%, baseline by June); cardiovascular −97%, respiratory −92%, endocrine −82%, dermatology −74% in April; control and prevention trials −73% against treatment trials −29% (SWOG).What helped: Prioritise trials where treatment cannot be deferred. What did not: Deferrable trials (prevention, dermatology) stayed depressed for months. (SH-03, grade A)
Research-staff redeployment and burnout (long tail)
Headwind for 2 to 3 years; coordinator shortages at 64.6% of sites; 80% of cancer centres with activation delays and 64% with declining accrual.What helped: Flexible working and retention measures; AACI turnover fell 29% from 2021 to 2022. What did not: Hiring inexperienced coordinators; a six-month cost per replacement is claimed by SCRS (anecdotal, no dataset). (SH-04, grade B)
National research prioritisation during an emergency
Headwind for non-priority studies: UK commercial recruitment −44% from 2017/18 to 2021/22; UK rank for trials initiated fell from 4th to 10th.What helped: Explicit triage protected the platform trials (RECOVERY and Solidarity). What did not: A reputational spiral of reduced allocations that had not reversed by 2022/23. (SH-05, grade A)

If it happens again

  • Look after patients already in the study first: remote visits and medicine delivered to the home kept them on treatment and in follow-up.
  • Expect the effect to differ by therapeutic area. Treatment that cannot wait, such as cancer care, recovered within months; prevention and deferrable studies stayed down far longer.
  • Slow recruitment before pausing it, and restart in phases.
  • Plan for a long staffing tail, not a quick rebound: coordinator shortages outlasted the lockdowns by two to three years.

Risk register SH-01 to SH-05: Medidata 20202; Unger et al. 2020 and 20212; O'Shaughnessy review 2023; Advarra 2025; AACI 2023

The war in Ukraine

From February 2022

Russia's full-scale invasion of Ukraine put patients, families and hospital staff in danger and forced millions of people from their homes. Sites near the fighting had to stop; many in the west of the country kept working. Research teams moved patients to sites within Ukraine and abroad so that their treatment could continue.

Regional war in a study country
Recruiting sites −48% at 7 months and −62% at 13 months with no recovery; new-study starts about a third of pre-war three years on; the rate at surviving sites is unknown (grade A for sites, C for rate).What helped: Transfers within country (115) and abroad (196) in four months; land IMP routes; direct-to-patient IMP; home blood draws; emergency ethics-committee meetings; 386 amendments. What did not: A one-to-two-month national IMP buffer; narrow visit windows; consent language mismatch; a three-year sponsor absence. (SH-06, grade A)
Proximity to the fighting within a country at war
Steep gradient: the east and north of Ukraine were hit while Kyiv was near normal within two months; about half of sites kept recruiting.What helped: Move patients west inside the country, which GlobalData calls more typical than cross-border transfer. What did not: Assuming a nationwide shutdown. (SH-07, grade B)
Displacement-driven loss to follow-up
Non-participation odds +54% for evacuation outside the municipality and +40% outside the prefecture (Fukushima cohort).What helped: Multiple contact channels; relationship-based retention (TRACER). What did not: Rigid visit windows. (SH-14, grade B)

If it happens again

  • Put continuity of care first. A patient who has had to move should be able to keep their treatment and follow-up at a site near where they now live, in Ukraine or abroad.
  • Prepare transfers before they are needed: agreements between sites, routes for the study medicine, and access to the patient's records, so that moving does not mean leaving the study.
  • Offer consent and study information in the patient's own language at the receiving site, with an interpreter.
  • Judge exposure by distance from the fighting, not by country: about half of Ukraine's sites kept recruiting.
  • Patients who have moved need care and continuity; they are not a new source of recruitment.

Risk register SH-06, SH-07 and SH-14: Phesi via Lab News 2022 and 2023; Clinical Leader 2026; DIA Global Forum 2023; Clinical Research News 2022; on consent in the patient's language, host-country guidance cited in register SH-08

The earthquakes in south-east Turkey

February 2023

On February 6th 2023 two earthquakes struck south-east Turkey and northern Syria. Hospitals in the worst-hit provinces were destroyed or overwhelmed, and Turkey's medicines regulator suspended ethics-committee work in the emergency provinces until May. EPI-512, a fictional study, has sites in those provinces; its dip is illustrative, because no trial-level recruitment data for the earthquakes was found.

Major earthquake in a site region
Headwind, probably near-total in destroyed provinces (three main hospitals destroyed in Hatay); health staff partly back on duty within a day.What helped: Transfer to unaffected cities (the dashboard's assumption; untested). What did not: none recorded. (SH-11, grade C)

If it happens again

  • Contact every enrolled patient first: their safety, medicine supply and follow-up come before new recruitment.
  • Know which sites share a hazard zone, and agree in advance which unaffected site can take their patients.
  • Expect regulatory pauses as well as damage: approvals in the emergency provinces stopped for three months.

Risk register SH-11: IFRC 2023; TİTCK announcements of February 28th and March 3rd 2023

How far each scenario would cut recruitment, country by country

Recruitment-rate multiplier under each shock scenarioEach dot a country, United States highlighted; hollow dots are illustrative

Show data
Recruitment-rate multiplier under each shock scenario, by country
ScenarioUnited StatesUnited KingdomSpainPolandTurkeyIndia
Pandemic peak (April 2020)×0.55×0.13×0.4×0.4×0.35×0.02
Pandemic recovery (August 2020)×0.78×0.6×0.65×0.7×0.65×0.29
Border conflict: site near the border×1×1×1×1×0.3×1
Border conflict: site elsewhere in country×1×1×1×1×0.9×1
Regional war next door (2022 analogue)×1×1×1×1.1×1×1

Source: epiClinical planning scenarios, illustrative shapes anchored to the 2020 pandemic and the 2022 war in Ukraine. Hover a dot for its source.