012022
Journal of Patient Safety
US malpractice claims, national database, 2001 to 2011
49 percent of claims involved a communication failure; 40 percent of those, a failed handoff.
The library
The evidence is public. We cite it.
Every number Lumen quotes traces to a published study you can pull yourself. This page is the shelf: eighteen studies across five sections, each with its year, its cohort, its finding, and its PubMed record one click away.
Catalogue slip
Studies shelved18
Years spanned2002 to 2025
Largest cohort337,025 patients
Smallest cohort12 clinicians
Our own outcome numbers0, by commitment
Links outPubMed, every card
Scroll
Shelf 01
The failure, measured
022007
Journal of the American College of Surgeons
444 surgical malpractice claims, 4 insurers
92 percent of harmful communication breakdowns were verbal; 43 percent occurred at handoffs.
032019
BMJ
Meta-analysis, 70 studies, 337,025 patients
About 6 percent of patients suffer preventable harm; 12 percent of that harm is severe or fatal.
Shelf 02
The cost of the hour
042006
Critical Care Medicine
2,731 septic-shock patients, 14 ICUs
Each hour of delay in effective antimicrobial therapy cut survival by an average 7.6 percent across the first six hours.
052017
New England Journal of Medicine
49,331 patients, 149 hospitals
Each hour to completion of the sepsis bundle raised the odds of in-hospital death by 4 percent.
062024
European Journal of Internal Medicine
Meta-analysis, 42 studies, 190,896 patients
Antibiotics within three hours cut the odds of death by 20 percent; the one-hour threshold did not reach significance. We quote the conservative reading.
072016
Critical Care Medicine
1,725 rapid-response activations, one centre
57 percent of activations came more than an hour late; the delayed cases died at 15 percent against 8. Delay was more common in the most-monitored units.
082007
Critical Care Medicine
50,322 ICU admissions, multicentre US database
Six or more hours boarded in the emergency department: in-hospital mortality 17.4 percent against 12.9.
Shelf 03
Where the time goes
092008
The Permanente Journal
767 nurses, 36 hospitals, time and motion
Documentation consumed 35.3 percent of nursing practice time, more than any other activity; care coordination took another 20.6.
102016
Annals of Internal Medicine
57 physicians, 430 observed hours, 4 specialties
49.2 percent of the office day went to the record and desk work, 27.0 percent to the patient.
112013
Journal of General Internal Medicine
29 interns, 873 observed hours, 2 centres
12 percent of intern time went to direct patient care; 40 percent went to a computer.
Shelf 04
The interruption tax
122010
Archives of Internal Medicine
98 nurses, 4,271 medication administrations
Each interruption raised clinical errors by 12.7 percent; at four interruptions, the risk of a major error doubled.
132017
BMJ Quality and Safety
227 nurses, 4,781 administrations, cluster randomised
87.9 percent of interruptions during medication work had nothing to do with the task at hand.
142018
BMJ Quality and Safety
36 emergency physicians, 120 shadowed hours
Prescribing while interrupted carried 2.82 times the error rate.
152002
Medical Journal of Australia
12 clinical staff, 35 observed hours, 2 EDs
36.5 communication events per person per hour, nearly a third of them interruptions.
Shelf 05
The fix, tested
162014
New England Journal of Medicine
Nine hospitals, 10,740 admissions
30 percent fewer preventable adverse events after structured handoffs, with no added handoff time.
172025
BMJ Quality and Safety
Systematic review, 12 handoff tools, 10 years
Moderate-certainty evidence that structured handoffs reduce errors and adverse events. We quote the grading with the finding.
182011
Academic Emergency Medicine
One hospital, matched 40-day windows
Timed automatic escalation of consults cut the median emergency-department stay by 106 minutes.
How the shelf is read
Three reading rules govern every card.
Published means published.
Every figure above belongs to its authors, is presented under their names, and describes hospitals in general. None of it is a Lumen result, and no card claims otherwise.
The conservative reading wins.
Where a literature is contested, the card quotes the most conservative synthesis, and it names certainty grades where reviews assign them. Card 06 and card 17 carry theirs.
Our own number will be measured first.
We publish no clinical outcome number of our own until we measure one at a reference site. That commitment is dated on the trust page.
The research feature
The discipline has a working instrument.
MedScribe, the manuscript workflow inside Lumen.
Software the founder built before Lumen, preserved inside Lumen as its research feature: a workflow that takes a study from structure to submission, every claim cited to source. Six roles carry a manuscript end to end, and one of them exists only to hold the line this page is built on.
StructureDoctor · PubMedScout · DataAnalyst · ClarityPolisher · IntegrityGuard · JournalMatch
IntegrityGuard flags overclaim, missing disclosure, and citation drift: nothing ships above its evidence. The rule that governs a manuscript governs every number this company publishes. MedScribe is not in clinical use, and no revenue is claimed for it.
Pull any card. Check us.
The clinical page reads this shelf against one night in the ICU. The trust page holds the commitments the shelf is quoted under.