And Then You're Dead

The Autopsy / Case File · Chronic Disease Progression

The First 90 Days

Kidney failure's most dangerous window isn't a diagnosis or a decade of decline — it's the three months right after dialysis starts.

Roughly 800,000 Americans are living with kidney failure severe enough to require dialysis or a transplant, and for the majority who start on dialysis, the national data shows a jarring pattern: mortality risk isn't spread evenly across the first year. It spikes hard, immediately, and then eases. Here's what the first 90 days actually look like, and why the way a patient arrives at dialysis often shapes whether they survive it.

And Then You're Dead · September 2026 · 8 min read

Join the discussion →
An older man receiving hemodialysis treatment in a clinical chair, connected to a dialysis machine
A patient during a hemodialysis session. National data shows the period right after treatment begins, not years later, carries the sharpest mortality risk.

Kidney failure is usually described, correctly, as a slow disease — years of declining kidney function, often tied to diabetes or high blood pressure, tracked through a lab value called estimated glomerular filtration rate that creeps downward for a decade or more before a patient reaches the threshold requiring dialysis or a transplant. That slow arc is why the sharpness of what happens next surprises people outside nephrology: the mortality curve for new dialysis patients doesn't continue that gradual decline. It spikes.

A Registry Built to See This

The United States Renal Data System, the federal registry that has tracked every American with kidney failure since the 1980s through its Annual Data Report, is one of the most complete disease-specific mortality datasets the government maintains — a byproduct of the fact that nearly all dialysis and transplant care in the U.S. is covered by Medicare regardless of a patient's age, which means USRDS captures close to the entire treated population rather than a sample. That completeness is what makes the early-mortality pattern so clearly visible in the data, rather than being an artifact of a small or skewed study population.

0
Americans living with kidney failure (end-stage renal disease) as of the most recent USRDS reporting, roughly two-thirds of them on dialysis and the remainder living with a transplant
0
adjusted deaths per 1,000 patient-years among the U.S. dialysis population overall — an annual mortality rate far higher than almost any other chronically managed condition

Against that already-elevated baseline, the first 90 days after a patient's very first dialysis session stand out sharply. Multiple analyses of large dialysis cohorts — including a widely cited USRDS-linked study of patients starting dialysis in the mid-2000s — found that somewhere between a third and a little over a third of all deaths occurring in a patient's first year on dialysis happen within just the first 90 days of that year, a window that represents only a quarter of the twelve-month period being measured.

0%
of incident dialysis patients who died within the first 90 days of starting treatment, in a large North American registry cohort
0%
of all first-year dialysis deaths that occurred within that same 90-day window — more than a third, compressed into a quarter of the year
The risk of death in the first 120 days after starting dialysis runs measurably higher than the risk during the remainder of a patient's first year — a pattern researchers have called a "universal phenomenon" across dialysis populations studied internationally.

What's Actually Killing Patients This Early

The cause-of-death breakdown for deaths in this early window looks different from dialysis mortality more broadly, and points toward a specific clinical story: a patient's body, and the medical team managing it, both adjusting in real time to an entirely new physiological state, often without much lead time to prepare for it.

0%
of early (90-day) dialysis deaths attributed to cardiovascular causes — heart attack, sudden cardiac death, and related events
0%
of early dialysis deaths attributed to infection, with catheter-associated bloodstream infection a recurring specific cause

Cardiovascular causes dominate for a reason rooted in the disease's own progression: by the time a patient's kidney function has declined far enough to require dialysis, years of the metabolic strain that caused the kidney failure in the first place — commonly diabetes and hypertension — have usually already done significant damage to the heart and blood vessels. Starting dialysis itself adds an acute stress on top of that existing damage: rapid fluid and electrolyte shifts during and between treatment sessions, blood pressure swings, and the hemodynamic strain of establishing a new pattern of removing several liters of fluid two or three times a week. For a cardiovascular system already compromised, that adjustment period is where a meaningful share of early deaths concentrate.

The Access Problem

Infection risk in this window traces back heavily to a specific, well-documented clinical detail: how a patient's body is actually connected to the dialysis machine. The preferred long-term option, an arteriovenous fistula — a surgically created connection between an artery and vein, usually in the arm — takes weeks to months to mature to the point it can be safely used, and ideally is placed well before a patient's kidney function reaches the point dialysis becomes necessary. Patients who reach kidney failure without that advance planning, which nephrology research has repeatedly found to be a large share of the incident dialysis population, start treatment instead with a central venous catheter — a tube inserted directly into a large vein, usable immediately but carrying substantially higher infection risk the longer it remains in place.

0%
of patients, in USRDS-tracked data, who began dialysis using a catheter rather than a more mature, lower-risk vascular access
0%
of early dialysis deaths attributed to treatment withdrawal — patients or families choosing to stop dialysis, most often after a difficult early course

That withdrawal figure is its own quiet finding. A meaningful share of early dialysis deaths aren't a medical crisis in the acute sense — they're the outcome of a patient or family deciding, after a difficult start on treatment, that continuing isn't the right choice. Nephrology palliative-care researchers have pointed to that pattern as evidence that the first 90 days function as a genuine decision point for many patients, not just a physiological gauntlet, and that earlier, more thorough conversations about what dialysis actually involves — before a patient is in a crash-start situation with a fresh catheter and no advance plan — could change how many patients arrive at that decision point already exhausted by it.

The Care Gap Before Day One

Both the catheter-access problem and the withdrawal pattern trace back to a step that happens, or doesn't happen, well before a patient's first dialysis session: pre-ESRD care, the standard package of nephrology follow-up, dietary counseling, and advance planning for vascular access that's supposed to happen in the months or years a patient's kidney function is declining toward the dialysis threshold. National data on how often that actually happens is not encouraging.

0%
of patients starting maintenance hemodialysis who had received none of the three recommended pre-ESRD care elements — nephrologist visits, dietitian care, and a mature vascular access already in place
0
percent of patients who received all three recommended elements of pre-dialysis care before their first treatment

A patient who reaches dialysis without any of that groundwork is, almost by definition, more likely to start with a catheter instead of a fistula, more likely to be unprepared for the physical experience of treatment, and more likely to be making decisions about whether to continue care under the kind of acute, unplanned crisis conditions that nephrology researchers associate with the highest early-withdrawal rates. Nothing about the first-90-days spike is fixed in the biology of kidney failure the way the underlying disease progression is — a meaningful share of it traces back to a healthcare-system gap that shows up long before a patient ever sits down in a dialysis chair.

A Pattern Seen Everywhere Dialysis Is Studied

The early-mortality spike isn't a quirk of American healthcare specifically. Registry studies from Canada, Australia, and multiple European dialysis networks have documented the same basic shape — sharply elevated risk in the first several months, tapering toward the steadier, still-elevated baseline rate that defines dialysis mortality more broadly for the rest of a patient's first year and beyond. That international consistency is part of why nephrology researchers increasingly describe the phenomenon as close to universal rather than tied to any one country's specific healthcare-delivery model — though the United States' comparatively high rate of catheter-based dialysis starts, tied to lower rates of pre-emptive vascular access planning, has been flagged in the American data specifically as a modifiable piece of the picture, rather than an unavoidable feature of the disease itself.

And Then You're Dead

Kidney failure gives most patients years of warning before dialysis becomes unavoidable — declining lab values, nephrology visits, a slow-motion countdown most people manage for a long time before crossing the threshold. And then, for a meaningful share of them, the most dangerous window of the entire disease turns out to be the twelve weeks right after the machine gets turned on, a period during which the heart, the blood vessels, and a still-immature vascular access are all being asked to do something new at the same time. The disease had been slow for years. The part that kills people fastest is the part that starts all at once.

That's it. That's the whole thing.

Comments

Loading…

Comments are anonymous and moderated.

Sources

Photo: "Patient receiving dialysis," by Anna Frodesiak, via Wikimedia Commons (CC0 1.0 Universal Public Domain Dedication).