Survival by Dialysis Era
Do recent starters live shorter than earlier starters — or is it just follow-up time? Live query onptdatacomplete.
Generated: 2026-09-21 23:14
Patients analysed
535
Deaths recorded
270
1-yr survival change
-5.6 pp
newest vs earliest era
2-yr survival change
-8.4 pp
newest vs earliest era
Kaplan–Meier survival curves by era
Each curve starts at 100% at dialysis initiation. Steps down = deaths. Line ends when the last patient of that era is censored. Compare curves at 12, 24 and 60 months — do not compare total height alone.
Survival at fixed landmarks (only eras that reached each landmark)
This is the fairest comparison. If the 2024 era has not yet reached 2 years of follow-up for all patients, only the 1-year bar appears for it.
Automatic analysis
Eras compared: 2012–2015, 2016–2019, 2020–2023, 2024–now. Total patients analysed: 535.
Suggested action: Landmark survival (1/2/5-year) is the only fair comparison because recent eras have not yet had time to reach later landmarks.
1-year survival has fallen by 5.6 percentage points in the newest era (2024–now) compared with the earliest (2012–2015).
Suggested action: Investigate: are recent starters older, more comorbid, more often catheter-based or starting dialysis as emergencies? Compare against the "Age at initiation" and "Mortality by access" pages.
2-year survival is lower in the newest era by 8.4 percentage points.
Suggested action: Audit comorbidity, dialysis dose and access type in recent starters. 2-year survival is dominated by cardiovascular events and infection.
Do not read the 'mean survival' column naively: it only counts patients who have already died, so a recent era that still has many patients alive will look artificially short (or long, depending on the pattern).
Suggested action: Use the Kaplan–Meier curves and the 1/2/5-year landmark table for real comparisons. Mean survival is shown for reference only.
Newest era (2024–now) has a maximum follow-up of only 32 months — 5-year survival cannot be computed for it yet.
Suggested action: Wait until this cohort accumulates 5 years of follow-up, or compare only on 1- and 2-year landmarks.
Note: an apparent 'newer patients die sooner' pattern can be produced by survivor bias in the older cohort — long-term survivors accumulate in the earliest era because they simply had more time.
Suggested action: Disentangle by re-running this page with the 1-year landmark only (fair to all eras) and with an age- and access-adjusted comparison.
What this page measures
- Era — group of patients by the year they started dialysis (SOT).
- Kaplan–Meier curve — the probability of still being alive at any time after starting, properly accounting for patients still alive (censored).
- 1 / 2 / 5-year survival — % of the era still alive at that landmark.
- Median survival — the month at which half the era has died (only shown if the curve actually drops below 50%).
- Mean survival (reference only) — average among the deceased only; severely biased by surviving patients and shown for completeness.
Why this matters — and the trap
It feels natural to say "newer patients live shorter on dialysis." But if a patient started in 2024, we simply don't yet know whether they will live 1, 5 or 10 years. The naïve mean of duration among deceased patients will make the recent era look artificially bad — or artificially good — depending on the pattern.
The correct questions are: at 12 months after starting, what fraction of each era was still alive? and at 24 months? If those fractions fall in recent eras, the signal is real. If they are stable, the "shorter average" you see is only follow-up time, not prognosis.
Action plan
- Read the landmark chart first. Only if 1-year and 2-year survival fall in recent eras is there a real problem to investigate.
- Adjust for case-mix. If recent eras are older, more diabetic or more catheter-heavy, part of the drop is expected. Cross-check with the "Age at initiation" and "Mortality by access" pages.
- Audit recent early deaths. For every 2023–2026 death within 12 months of start: referral timing, planned vs emergency start, access at initiation, cause of death.
- Strengthen the first-year bundle. Weekly nurse review for the first 90 days; 3/6/12-month multidisciplinary reviews.
- Plan for the long tail. If early survival is stable but 5-year survival is falling, the problem is chronic care — cardiovascular risk, dialysis dose, access survival, mineral-bone disease.
- Re-run this page yearly. Trends matter more than any single snapshot; the newest era should be read with growing confidence each year.
- Guard the data. Rows with missing or invalid SOT/EOT are silently dropped — audit them if the totals differ from your register.
Landmark survival table
| Era | Patients | Deaths | Max follow-up (mo) | 1-yr survival | 2-yr survival | 5-yr survival | Median survival (mo) | Mean (deceased only, biased) |
|---|---|---|---|---|---|---|---|---|
| 2012–2015 | 116 | 74 | 168.5 | 84.5% | 72.4% | 55.2% | 70.6 | 40 |
| 2016–2019 | 134 | 89 | 126.9 | 81.3% | 67.9% | 44.8% | 51.4 | 32.5 |
| 2020–2023 | 203 | 84 | 80.6 | 73.9% | 65.5% | 58.1% | not reached | 12 |
| 2024–now | 82 | 23 | 32 | 78.9% | 64% | – | not reached | 9.1 |
"–" means the era has not yet accumulated enough follow-up to compute that landmark. "Not reached" means fewer than half the era has died.