NTUH · Yunlin Branch
Nephrology Grand Rounds
For Nephrologists · 40 min
Benefits of PD
for the Elderly
老年人適合腹膜透析嗎？ — A Shared Decision
Agenda02 / 36
Outline
A forty-minute conversation,
in eight movements.
01IntroductionTaiwan's dialysis landscape · TWRDS 2024
02ASN Geriatric CurriculumWhat the canonical reference still tells us
03From Survival to Quality of LifeThree decades of HD vs PD comparisons
04Recent EvidenceCheng 2023 meta-analysis & companion studies
05The Case for PD in the ElderlyHemodynamics · independence · assisted PD
06Shared Decision MakingA four-step model for the dialysis choice
07My PD ExperienceNTUH Yunlin Branch · 6-year cohort
08Closing ThoughtsA teacher's lesson, a patient's verdict
Part One03 / 36
PART · 01
Introduction.
The Taiwanese landscape.
Before we ask whether the elderly should choose PD, it helps to see how few of them actually do — and to ask why.
Introduction04 / 36
TWRDS · prevalent dialysis population
Dialysis in Taiwan — an HD-dominated ecology.
HD grows by nearly 9,000 patients over five years.
PD stays effectively flat.
Taiwan's PD share — already among the lowest in Asia — is trending downward as new HD prescriptions outpace new PD starts.
Among incident patients the picture is a little better: of 12,459 new dialysis starts in 2022, 1,182 (9.5 %) began on PD. The NHI's 2026 PD programme sets an explicit target — incident > 10 %, prevalent > 8 %.
TWRDS 2024 annual report (2022 incident data); prevalent series 2019 – 2023. NHIA PD promotion programme, 2026 edition.
Introduction05 / 36
TWRDS · age structure
Our dialysis population is aging quickly.
67.9
years — the mean age at first dialysis in Taiwan (2022). Mean age at death on dialysis is 73.6. We are, in effect, a geriatric specialty.
≥ 75
years is the fastest-growing band; this is precisely the group most likely to be frail and most likely to benefit from a gentler modality.
9.5%
of all incident patients started on PD in 2022 — and in the ≥ 65 group the share is lower still, far below Hong Kong, France, or the UK.
TWRDS 2024 annual report — 2022 incident cohort (n = 12,459; HD 11,277 / PD 1,182).
Introduction06 / 36
TWRDS · PD utilisation by age band
PD utilisation falls sharply with age.
HK · 82% incident PD
France · assisted PD funded
UK · 17% incident elderly
Canada · 12% incident elderly
“PD is rare and probably underused in elderly patients.”
— ASN Geriatric Nephrology Curriculum, Ch. 22.
Taiwan age bands: TWRDS, approximate. International figures: ASN Curriculum Ch. 22 (UK / Canada, incident elderly); Hong Kong Renal Registry 2022.
Part Two07 / 36
PART · 02
The ASN Geriatric
Nephrology Curriculum.
Wright & Danziger's Chapter 22 has been our anchor reference since 2009. Its conclusions, surprisingly, have aged well — let's walk through them.
ASN Curriculum08 / 36
International comparison
PD penetration is a policy choice, not a clinical limit.
United States · 2009
12 %
→ 4 %
USRDS at the time of the ASN chapter: 12 % PD at ages 20–55, ~4 % beyond 75. The age gradient persists — but the level has moved: incident PD reached 13.7 % by 2022.
Hong Kong
PD-first
82 %
PD-first since 1985. In 2022, 82 % of new dialysis starts and 73 % of prevalent dialysis patients were on PD. A policy mandate, not a clinical recommendation.
France
> 50 %
of PD pts ≥ 70
Read the famous number carefully: over half of French PD patients are ≥ 70 — not over half of elderly dialysis patients (incident PD is ~14 %). What is exceptional is assisted PD.
United Kingdom
17 %
Of incident elderly patients. National framework routes modality choice through structured pre-dialysis education and SDM; assisted PD is available.
Canada
12 %
Of incident elderly patients. Provincially funded assisted-PD programmes — Ontario pays ~CAD 20,600/year for nurse-assisted connections — drive slow growth in seniors.
Taiwan · updated
< 10 %
≥ 65 y
The policy gap has narrowed: the NHI has run a dedicated PD programme since 2023, and now pays an assisted-PD visit fee. What has not changed is that pre-dialysis education is still HD-default.
ASN Curriculum Ch. 22 (2009), updated with USRDS 2024, HKRR 2022, RDPLF, ISPD 2024, NHIA 2026.
ASN Curriculum09 / 36
Considerations specific to older patients
Real concerns — and what the evidence actually says.
The legitimate worries
- Peritoneal aging — proinflammatory mesothelial changes, fibrosis, membrane failure.
- Higher prevalence of diverticular disease and prior abdominal surgery.
- Visual, manual, and cognitive demands of exchanges.
- In patients > 80 y, about 60 % may require some form of assistance.
- ISPD 2024: 63 % of patients with no medical contraindication to PD still had a physical or cognitive barrier to self-care.
What the data has shown
- The proinflammatory profile is real in the laboratory; clinical significance remains untested.
- Catheter placement is feasible after prior abdominal surgery in experienced hands.
- APD cyclers, assistive devices and trained assistants bridge the gap — in the French registry, nurse-assisted PD carried lower peritonitis risk (HR 0.85, 0.78 – 0.94).
- Overall mortality in elderly PD is broadly similar to HD.
- Assisted-PD programmes raise incident PD use and reduce transfers to HD.
Wright & Danziger, ASN Curriculum Ch. 22; Duquennoy et al., PDI 2016; Oliver et al., ISPD assisted-PD position paper, PDI 2024.
ASN Curriculum10 / 36
Peritonitis & infection
Peritonitis risk in the elderly: conflicting evidence.
2008
Szeto et al., PDI — no difference in 12-month peritonitis-free probability (76.6 % vs 76.5 %, ≥ 65 vs < 65).
Hong Kong — incident patients performing their own exchanges.
2016
Duquennoy et al., PDI — age > 75 not associated with peritonitis (HR 0.97); nurse-assisted PD lower risk (HR 0.85).
RDPLF registry, 8,396 incident PD patients 2003 – 2010.
2020
Wu et al., PDI — elderly had a higher peritonitis rate: 0.203 vs 0.145 episodes / patient-year.
Single centre, China, n = 1,953 — and here assisted PD was a risk factor, not a protector.
2022
Jiang & Zheng, PLOS ONE — peritonitis RR 1.56 (1.18 – 2.07) in elderly PD; technique survival unchanged (RR 0.95).
Meta-analysis, 14 studies (12 pooled).
2024
ISPD position paper — peritonitis rates in assisted PD are similar to self-care PD where programmes are properly funded and staffed.
Oliver et al., PDI 2024 — assisted peritoneal dialysis.
Synthesis: who assists, and how well they are trained, moderates the risk far more than age itself.
ASN Curriculum11 / 36
Quality of life
In the elderly, QoL is similar — and that is the whole point.
- In younger patients PD is generally associated with better HRQoL than HD — partly self-selection.
- In older patients the published data are sparse, but consistently report no QoL difference between HD and PD.
- Two contemporary cohorts sharpen this: BOLDE found less depression and less illness intrusion on PD; FEPOD found higher treatment satisfaction on assisted PD than on in-centre HD, with overall QoL similar.
- For an older patient who places a high value on independence, on not travelling three times a week, or on night-time dialysis, PD remains the obviously appropriate first choice.
“If two paths give the same outcome,
the right path is the one the patient would choose.”
— A working principle for SDM.
Wright & Danziger, ASN Curriculum Ch. 22; BOLDE and FEPOD as reviewed in Selwood et al., Clin Kidney J 2025;18:sfaf020.
ASN Curriculum12 / 36
Take-home points
Five take-home points
from the ASN curriculum.
01
Underused
PD is rare and probably underused in elderly patients — in the US, and even more so in Taiwan.
02
Mortality parity
No substantial mortality difference between PD and HD in elderly patients eligible for both.
03
QoL parity
Quality of life is not reported to differ between elderly patients on PD and HD.
04
Infection parity
"Infection rates are not higher in elderly versus younger PD patients." This is the point that has aged least well — see slide 10.
05
Assistance helps
Assistants and cyclers "may make PD more accessible to some elderly patients." Since 2024, ISPD states it more firmly than that.
Wright & Danziger, ASN Curriculum Ch. 22 — Take Home Points (2009). Points 1 – 3 and 5 hold up; point 4 needs the qualifier "when assistance is properly resourced".
Part Three13 / 36
PART · 03
From Survival
to Quality of Life.
For thirty years the field litigated a single question: which modality keeps patients alive longer? It was the wrong question — or at least, the wrong only question.
HD vs PD14 / 36
Three decades, one debate
The survival war: no winner declared.
1995
Bloembergen et al., JASN — USRDS analysis concludes HD survival is better.
Foundational, but observational; severe selection bias.
2003
Korevaar et al., KI — the only randomised trial of PD vs HD was stopped early: 38 patients randomised.
A field-defining null result. The RCT will not be done.
2005
Jaar et al., Ann Intern Med — the CHOICE study finds no difference.
Prospective US cohort, modern era.
2015
Han et al., CJASN — Korean ≥ 65 registry plus meta-analysis: PD HR 1.20 (1.13 – 1.28), but no adjustment for frailty.
The frailty omission is the single most important caveat in this literature.
2025
Nardelli et al., BMC Nephrol — 27 studies, 1.03 million incident patients: HR 1.01 (0.93 – 1.10). It is a tie.
PD better under 65 (HR 0.75). Cochrane 2024: the evidence remains uncertain.
Lee & Bargman, "Survival by Dialysis Modality — Who Cares?" Semin Dial; Ethier et al., Cochrane Database Syst Rev 2024.
HD vs PD15 / 36
The hidden variable
Most "HD wins" studies are really measuring frailty.
In real-world cohorts, PD recipients in the elderly are often self-selected for frailty — the patient who cannot tolerate centre HD, who is bedbound, who comes from a nursing home.
Frailty is reported in up to 82 % of patients with CKD stage 5. When studies fail to adjust for it, the modality variable simply absorbs the frailty signal.
Restrict to patients eligible for both modalities, and the survival difference largely disappears.
“Survival of the fittest,
not survival of the modality.”
— Lee & Bargman, on how to read the literature honestly.
Wong et al., AJKD 2018 — after excluding the ~one third of incident patients ineligible for PD, survival was similar regardless of age. Frailty prevalence: Selwood et al., Clin Kidney J 2025.
HD vs PD16 / 36
Morton et al., 2012 · AJKD
When asked, patients say QoL — not months.
15 – 23 mo
of life expectancy patients were willing to trade away for greater freedom to travel.
7 mo
of life expectancy patients were willing to trade for fewer hospital visits.
“If we measure success only in months,
we are not measuring what our patients came here for.”
Morton et al., AJKD 2012 — discrete-choice experiment, n ≈ 100 dialysis-eligible patients.
Part Four17 / 36
PART · 04
Recent Evidence.
The 2023 meta-analysis.
Cheng et al. assembled thirty-one studies and 770,000 patients to revisit the survival question one more time — with the modern era, diabetes, and dialysis duration as the moderators.
Recent Evidence18 / 36
Cheng et al., 2023 · Updated systematic review & meta-analysis
Mortality of PD vs HD in older adults — study design.
Studies
31
Observational cohorts only; databases searched from inception to June 2022.
Patients
> 774k
More than 74,000 on PD and 680,000 on HD, in an older dialysis population.
Outcome
All-cause mortality
Hazard ratios pooled with random-effects; stratified by era, DM, age, RRT duration.
Quality
Moderate
No RCT in the pool; residual confounding for frailty and dialysis access remains.
Cheng, Hu, Song & Chen. "Mortality of Peritoneal Dialysis versus Hemodialysis in Older Adults: An Updated Systematic Review and Meta-Analysis." Gerontology 2024;70(5):461 – 472 (online 2023). PMID 37742209.
Recent Evidence19 / 36
Cheng et al., 2023 · overall mortality
Overall, PD carries a 17 % higher mortality.
HR 1.17
95 % CI 1.10 – 1.25 · PD vs HD, all-cause mortality, age ≥ 65.
The headline number is real — but it is a population average across three decades and many health systems.
As we will see, the average dissolves once you stratify by era, comorbidity, and dialysis vintage.
And it is one estimate, not the estimate: a 2025 meta-analysis of 1.03 million incident patients puts the overall figure at HR 1.01 (0.93 – 1.10).
For most contemporary elderly patients eligible for both modalities, the choice is not between living and dying — it is between two ways of living.
Cheng et al., Gerontology 2024 · PMID 37742209. Comparator: Nardelli et al., BMC Nephrol 2025;26:588.
Recent Evidence20 / 36
Cheng et al., 2023 · subgroup analysis
The average dissolves on subgroup analysis.
Cheng et al., Gerontology 2024 · PMID 37742209. Coloured rows are where PD carries measurable excess risk. Note ≥ 75: the point estimates are raised, not null — the confidence intervals simply cross 1.
Recent Evidence21 / 36
Cheng et al., 2023 · what to do on Monday
What this means in clinic.
Time horizon
Match modality to expected vintage
For patients whose realistic horizon is under 2 years, PD and HD are equivalent for survival — and PD is gentler. Where residual function persists, start incremental.
Diabetes
Diabetic patients need closer follow-up
DM elderly carry the largest excess risk on PD. Tight glycaemic and volume control; revisit modality at the 3-year mark, which is where the excess actually begins.
Era effects
Contemporary PD has narrowed the gap
Pre-2010 data should no longer dominate counselling. We have biocompatible fluids, APD, remote monitoring and assisted PD — and in Taiwan, NHI payment for all three.
Cheng et al. conclude: modality choice should weigh diabetes status, comorbidity burden, expected RRT duration — and incorporate patient preference and quality of life through shared decision making.
Cheng et al., Gerontology 2024 · PMID 37742209.
Part Five22 / 36
PART · 05
The case for PD
in the elderly.
If survival is parity-class, the conversation shifts to fit. Three reasons recur in every careful review — and one operational answer to the "but who will do the exchanges?" objection.
The case for PD23 / 36
Reason one
PD is a kinder haemodynamic option.
- Continuous, gentle ultrafiltration — fluid is removed minute by minute, not in a 4-hour bolus.
- No intradialytic hypotension, no post-HD stunning of brain and myocardium.
- Steady blood-pressure control; less reliance on IV inotropes and reactive volume management.
- Particularly attractive in older patients with heart failure, autonomic dysfunction, or labile blood pressure.
- Preserves residual kidney function longer — a meaningful survival and QoL signal.
- Taiwanese national data agree: in incident patients ≥ 65, PD had lower MACCE (HR 0.74) and lower MACCE-related death (HR 0.62) than HD.
“For a frail elderly heart, three weekly shocks are not therapy. They are a stressor we have learned to tolerate.”
— A working argument for PD-first in the elderly.
Peng et al., Sci Rep 2023;13:16199 — NHIRD 2013 – 2019, ≥ 65 incident ESKD, 1,628 PD propensity-matched to 6,512 HD. All-cause mortality was equal (HR 0.98); infection was higher on PD (HR 1.28).
The case for PD24 / 36
Reason two
PD is performed at home — and at home is where older people heal.
Time
No 3× weekly transit
Eliminates the four- to six-hour day spent on transport, waiting, and recovery — the hidden cost of centre HD.
Autonomy
Schedule on patient terms
Day-time CAPD or overnight APD allows for family meals, religious schedules, grandchildren, travel — the texture of late life.
Safety
Lower infection exposure
Avoids the dialysis-unit waiting room — relevant in respiratory-virus seasons and for immunocompromised patients.
Cognition
No dialysis-day stunning
Less post-HD fatigue and cognitive fog; older patients report more usable hours per day.
Economics
Lower system cost
Consistently cheaper than centre HD across health systems. Taiwan's NHI now also pays for APD with remote patient management — about NT$2,500 a month back to the patient.
Dignity
A patient, not a slot
Home dialysis preserves the social role of the older adult inside the family, rather than as a tri-weekly outpatient.
Synthesis of registry, QoL, and cost-effectiveness literature 2010 – 2024.
The case for PD25 / 36
Reason three · operational
"But who will do the exchanges?" — assisted PD.
The three operational models
- Family-assisted PD — spouse, adult child, or live-in caregiver trained by the PD nurse.
- Home-care assisted PD — visiting nurse performs CAPD exchanges or sets up overnight APD.
- APD with telemonitoring — cycler does the work overnight; clinic reads compliance and ultrafiltration remotely.
Taiwan is no longer the outlier: the NHI now pays an assisted-PD visit fee (P8117C), and age ≥ 65 alone qualifies.
“The risk of peritonitis is not increased in elderly patients on PD — in a country where assisted PD is available.”
— Duquennoy et al., PDI 2016.
FR · Reimbursed
UK · Available
HK · Routine
TW · Fee P8117C
France funds assisted PD nationally — ~35 % of PD patients, most of those ≥ 75. ISPD position paper (Oliver et al., PDI 2024): fund it, staff it with redundancy, monitor it like self-care PD.
Part Six26 / 36
PART · 06
Shared Decision
Making.
When the survival curves overlap, the decision is no longer a medical fact — it is a personal one. Our job is to make sure the patient is the one making it.
SDM27 / 36
Charles, Gafni & Whelan · Soc Sci Med, 1997
What we mean when we say shared decision making.
Criterion 01
Two participants
At minimum, both the physician and the patient are actively involved in the decision-making process.
Criterion 02
Information flows both ways
The physician shares evidence; the patient shares values, preferences, and life circumstances.
Criterion 03
Deliberation is shared
Both parties take steps to build a treatment preference — neither imposes nor abdicates.
Criterion 04
A decision is agreed
An explicit decision is reached, jointly owned by patient and physician. Documentation matters.
“SDM is the meeting of two experts — one in evidence,
one in the life that will live with the consequences.”
Charles, Gafni & Whelan, 1997 — the foundational four-criterion definition.
SDM28 / 36
A four-step model for the dialysis modality choice
SDM in practice — four steps, two voices.
Step 01 · 醫
Present the options
The nephrologist lays out HD, PD, conservative care, and transplantation — fairly, with personalised survival and QoL framing.
Step 02 · 病
Explore values
The patient — and family — names what matters: independence, travel, family role, fear of needles, fear of operations.
Step 03 · 病
Make the choice
A decision is articulated by the patient, supported by the physician. Reversibility is named explicitly.
Step 04 · 醫
Create the access
The nephrologist plans and executes the chosen access — PD catheter, AVF, AVG, or tunnelled catheter (Permcath).
Adapted for dialysis modality decision-making from Charles et al., 1997; consistent with KDIGO and ASN SDM frameworks.
SDM · Step 0129 / 36
STEP · 01
01
醫師The physician presents
Present all options — honestly, evenly.
- HD, PD, kidney transplantation, and conservative kidney management — name all four.
- Personalise survival and QoL framing using patient age, frailty, comorbidities — not pooled hazard ratios.
- Where you can, use a comprehensive geriatric assessment — it speaks to whether the patient will still be alive and at home in 3 – 12 months, which is the question actually being asked.
- Acknowledge what the evidence shows and what it does not — the residual uncertainty is the patient's to weigh.
- Avoid the default-HD framing ("we'll just put in a Permcath") that is endemic in Taiwanese nephrology practice.
- Where possible, use a visual decision aid (Option Grid, NKF tools, KDIGO leaflets).
"Honestly and enthusiastically offer PD to all ESRD patients based on clinical appropriateness." — Prof. 黃政文.
SDM · Step 0230 / 36
STEP · 02
02
病人The patient explores
Surface the values beneath the question.
- How important is staying at home?
- Is there a trained family member or willing caregiver?
- What is the patient's relationship to needles, to operations, to hospitals?
- Travel, religious obligations, work, grandchildren — which of these is non-negotiable?
- Family dynamics: who will the decision actually fall to in 12 months, when fatigue sets in?
- What does the patient (and family) understand the prognosis to be?
This is where the nephrologist must listen more than speak. The PD nurse and social worker are essential collaborators.
SDM · Step 0331 / 36
STEP · 03
03
病人The patient decides
The patient names the choice — not us.
- The articulation should come from the patient's own mouth: "I want PD" / "I want HD".
- Acknowledge that the decision is reversible — modality switches are common and acceptable.
- Set a clear re-decision point: a planned conversation at 12 months, or sooner if circumstances change.
- Document the SDM process — not just the outcome — in the chart.
- Affirm the choice without subtle disapproval. Patients sense ambivalence and over-correct toward what they think the physician wants.
A decision documented in the chart as the product of SDM is also the legally and ethically strongest position for the team.
SDM · Step 0432 / 36
STEP · 04
04
醫師The physician creates access
Now — and only now — plan the access.
For PD
Tenckhoff catheter
Open or laparoscopic; plan 2 – 4 weeks before anticipated start; consider rectus-sheath tunnelling in obese or elderly patients.
For HD
AVF / AVG / Permcath
AVF preferred where vessels permit; AVG in elderly with poor venous quality; tunnelled cuffed catheter (Permcath) as planned bridge or by patient choice.
In the elderly, a tunnelled Permcath is not a moral failure — it is sometimes the right access for a defined horizon.
Access planning belongs after the decision, never before it. Pre-emptive AVF in a patient who will choose PD is a system failure, not a clinical victory.
Part Seven33 / 36
PART · 07
My PD experience
at NTUH Yunlin.
Numbers from a six-year cohort in a real Taiwanese county hospital — not a referral centre, not a registry, just what happened.
My PD experience34 / 36
NTUH Yunlin Branch · 2020 – 2026
A small cohort, honestly reported.
Catchment
Yunlin 659k
Changhua 1.21m
Taichung 2.87m
A largely rural population with the oldest age structure in Taiwan.
Centre HD
~ 4,000 /mo
HD volume in the county, distributed across multiple clinics.
PD on programme
60 → 50
Slow decline across the catchment, consistent with the national trend.
My patients
38 in 6 y
Started since my return from NTUH Taipei.
Currently active
15
Mean PD vintage 40 months. Mean current age 49.
Transferred
1 + 1 + 8
1 transplanted · 1 relocated · 8 transitioned to HD.
Deceased
13
Mean age at death 74 — a population we did not surrender to HD.
Mentor
"Trained at NTUH
under Prof. 黃政文."
460+ PD patients across 2 fellows.
A real-world programme. Not a registry, not selection-corrected — what came through the door.
Closing35 / 36
My thoughts on PD for the elderly
What I have learned, in one paragraph.
“No patient is unfit for PD
— only the physician untrained in it.”
— Prof. 黃政文, my teacher.
- Practise shared decision making; centre quality of life, not modality preference.
- Honestly and enthusiastically offer PD to every ESRD patient who is clinically appropriate — and that includes the elderly.
- Embrace technology, embrace AI — let it draft, let us decide.
- Above all: the rest of this, too, I learned from my patients.
A summary of six years of practice, three iterations of this talk, and many evenings reviewing peritoneal effluent.
End36 / 36
Thank you.
Q & A
紀竣議 醫師 Chun-Yi Chi, MD
Division of Nephrology · NTUH Yunlin Branch
2026 · 04 · 30
“To PD or not to PD —
let the patient decide, with us beside them.”