Industries / Health / Healthcare / Dialysis
Vascular access
Vascular access is the connection to a blood vessel that lets blood leave the body for the dialyzer and come back, several times a week for years; this is the work of making these connections, using them and keeping them open.
7
Milestones achieved, from 1960 to 2020.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1960
A shunt left in place
Before 1960 each treatment needed fresh cuts into an artery and a vein. Wayne Quinton and Belding Scribner made an arteriovenous shunt, tubes of Teflon (a slippery plastic) left in an artery and a vein so they could be joined to the machine again and again. The Kidney Project's chronology says it changed hemodialysis by doing away with repeated incisions; the same chronology credits Nils Alwall in Sweden with glass shunts in 1948.
Credit
Wayne Quinton, D. Dillard and Belding Scribner
1966
A fistula under the skin
M. J. Brescia, J. E. Cimino, K. Appel and B. J. Hurwich reported long-term hemodialysis "using venipuncture and a surgically created arteriovenous fistula". A surgeon joins an artery to a vein under the skin, and at each session needles in that vein carry blood to the machine and back, with nothing left outside the skin. The fistula became the preferred access because, in a 2008 trial report's words, it has "lower thrombosis and infection rates and lower health care expenditures" than synthetic grafts or catheters.
Credit
M. J. Brescia, J. E. Cimino, K. Appel and B. J. Hurwich
1976
Plastic grafts for poor veins
When a patient's veins are not suited to a fistula, a surgeon can join the artery and vein with a tube under the skin instead, called a graft. L. D. Baker, J. M. Johnson and D. Goldfarb reported 84 grafts of expanded PTFE, a stretched form of Teflon, in 72 patients. In early follow-up they worked at least as well as the grafts made from cow arteries then in use, and the authors noted they were easier to handle and cheaper.
Credit
L. D. Baker, J. M. Johnson and D. Goldfarb
2003
A campaign for fistulas
Only 24% of US patients had fistulas in the 1990s. The US Fistula First Breakthrough Initiative began in 2003, by R. P. Franco's account; M. Allon and T. Lee date it to 2004. US Renal Data System figures cited by Franco show fistula use rising from 32% of patients in 2003 to 63% in 2018, while grafts fell from 40% to 17.5% and catheters only from 27% to 19.6%.
LessonThe rise came mostly from fewer grafts, not fewer catheters, so the campaign later took the slogan "Fistula first, catheter last".
Credit
Fistula First Breakthrough Initiative; the 2006 KDOQI guidelines, which took up its aims
2008
Most new fistulas not usable
In a trial at 9 US centers from 2003 to 2007, 877 people had new fistulas made. A fistula counted as usable if it ran at a blood pump rate of 300 ml/min or more in 8 of 12 sessions. By the end of follow-up, about six months after surgery, 61.8% of fistulas in one group and 59.5% in the other had not reached that mark. An anti-clotting drug cut early clotting but made no difference to this.
LessonPreventing clots did not make fistulas usable; failure to mature, a new fistula not growing large and strong enough for needles, was the bigger barrier.
Credit
L. M. Dember and the Dialysis Access Consortium
2018
Fistulas made through catheters
On 22 June 2018 the FDA granted De Novo authorization, its route for new kinds of device with nothing comparable on the market, to two systems that join an artery and a vein from inside the vessels, without open surgery: TVA Medical's everlinQ and Avenu Medical's Ellipsys. Ellipsys seals the two vessels together with heat from the tip of a catheter. Its single-arm study enrolled 117 people at five US sites; 89.3% of the fistulas counted in the main analysis had matured at 90 days.
LessonNewer models were cleared through the 510(k) route as equivalent to earlier ones, and a 2024 study of FDA problem reports found the newer WavelinQ had fewer failures to align but more reports of high blood pressure than its predecessor.
accessdata.fda.govaccessdata.fda.govaccessdata.fda.goveutils.ncbi.nlm.nih.gov
Credit
TVA Medical; Avenu Medical; US Food and Drug Administration
2020
Right access, right patient
The 2019 update of the National Kidney Foundation's KDOQI vascular access guideline, published in 2020, moved care from "Fistula First" to an "ESKD Life-Plan", in which each patient's access is chosen to fit their plans for treatment and life, with new targets for fistulas, grafts and catheters. In Allon and Lee's summary it asks clinicians to "place the right access in the right patient at the right time for the right reason", after studies found that 30% to 40% of fistulas placed aggressively fail to mature.
LessonAllon and Lee call that phrasing "too vague to be helpful to clinicians", so the change of goal came before guidance on how to meet it.
Credit
C. E. Lok, T. S. Huber, T. Lee and colleagues, for the National Kidney Foundation
Watch for
Lab-grown vessel against fistulas
Humacyte's acellular tissue engineered vessel (ATEV) is grown from human vascular cells, which are then removed, leaving a tube that the patient's own cells move into. The FDA approved it in December 2024 for repairing injured arteries in the limbs. In a phase 3 trial at 31 US centers, 242 patients on dialysis were randomly given the vessel or a fistula; at 6 months 81.3% of the vessels and 66.4% of the fistulas were working, though the vessel needed more procedures to keep it open.
Achieved whenthe FDA approves the vessel for dialysis access, or a trial not funded by the company repeats the result.
Claimed by
Humacyte, which funded the trial, and its investigators, August 2026
Narrowing heard by software
A team in Taichung, Taiwan, recorded access sounds with an electronic stethoscope from 30 patients being treated for narrowed access. It trained a neural network on recordings from 10 of them and tested it on 40 recordings from the other 20. It reports 100% sensitivity and specificity for narrowing of more than 50%, checked against X-ray images of the vessels.
Achieved whena different group tests the method on hundreds of patients at several centers and publishes its accuracy.
Claimed by
M. Y. Kang, B. Y. Liau and colleagues, Taichung Veterans General Hospital and Feng Chia University, June 2025
Open
01
A public fistula sound set
An open library of recordings from silicone fistula phantoms, bench models with set narrowings from 0% to 90% and flows of 400 to 1,200 ml/min, recorded with several electronic stethoscopes and low-cost microphones. It would come with a fixed split into training and test files, so that published sound-reading methods can be compared on the same data.
The briefCredit · unclaimed02
Needle slip alarms tested
A bench test of whether, and how fast, dialysis machines from different makers raise an alarm when the needle that returns blood slips out of the access. It would run on a sham circuit with blood-mimicking fluid and a model arm, with no patient, across set blood flows and pressures, and would also test moisture sensors made to detect blood at the needle site.
The briefCredit · unclaimed
Open 01 · the brief
A public fistula sound set
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open library of recordings from silicone fistula phantoms, bench models with set narrowings from 0% to 90% and flows of 400 to 1,200 ml/min, recorded with several electronic stethoscopes and low-cost microphones. It would come with a fixed split into training and test files, so that published sound-reading methods can be compared on the same data.
Why it is still open
Each method is tested on its own lab's recordings. The Taichung team used 30 patients; S. J. A. Majerus's lab built its own vascular phantom and reported up to 100% accuracy; a Rome group built a 3D-printed fistula phantom for testing optical sensors. Searches of PubMed and GitHub found no public set of phantom recordings, and the one open-source fistula monitor found, DSFAV, runs on synthetic signals.
Built with what exists
3D-printed molds and silicone casting, following the Rome group's published method; a pulsing pump and blood-mimicking fluid; electronic stethoscopes sold to clinicians and MEMS microphones; Python and scikit-learn for baseline classifiers; Zenodo to host the files with a DOI.
First step
Cast one straight-tube phantom with three swappable narrowings and record 100 clips with one stethoscope.
Done looks like
at least 2,000 labeled recordings across at least 6 narrowing levels, 4 flow rates and 3 sensor types, published with a DOI; a second lab builds the phantom from the published files and matches the first lab's baseline accuracy within 5 percentage points.
Open 02 · the brief
Needle slip alarms tested
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A bench test of whether, and how fast, dialysis machines from different makers raise an alarm when the needle that returns blood slips out of the access. It would run on a sham circuit with blood-mimicking fluid and a model arm, with no patient, across set blood flows and pressures, and would also test moisture sensors made to detect blood at the needle site.
Why it is still open
A needle that slips out unnoticed can cause severe injury or death. Frinak and colleagues ran 19 simulated slips on one machine model, the Fresenius 2008K, and only 8 set off the machine's alarm; their better alarm came from Vasc-Alert, a company that took part in the study. Searches found no test across makers.
Built with what exists
used or loaned machines from at least three makers run in sham mode; blood-mimicking fluid; a silicone arm with a vessel channel; pressure loggers; a camera with an on-screen timer; research or commercial moisture sensors.
First step
Rebuild the Frinak sham circuit on one machine and repeat five of the 19 published scenarios.
Done looks like
at least 3 machine models and 2 sensors, each run through the same 20 scripted slips at set blood flows and pressures, with alarm times published as raw data and video.