PD in intensive care units (ICUs) is used in two main clinical situations: emergency initiation of PD (within 72 hours of PD catheter placement) for severe ARF (or CRF with late presentation) in a patient not already on dialysis (emergency PD) and continuation of PD as a chronic modality in patients already on dialysis (chronic PD continuing in a “critical” situation).

In intensive care, ARF is mainly linked to shock, severe sepsis, multiple organ failure, and drug-induced nephrotoxicity. ¹ AKI is associated with substantial mortality in patients admitted to the ICU. [8–11] Although these patients are more frequently admitted to intensive care units, they also have an increased risk of morbidity and mortality compared to those with preserved renal function. ¹²

In patients with AKI requiring extra renal purification, PD may be considered. This is referred to as “urgent PD” (urgent start peritoneal dialysis or USPD). When initiating PD in an acute setting, a healing period (“break-in”) after placement of the intraperitoneal catheter is normally desirable. This period, which can last up to 72 hours, reduces the risk of dialysate leakage, particularly along the percutaneous tunnel. The catheter is usually inserted percutaneously under local or regional anesthesia, without general anesthesia, and can be performed by an experienced nephrologist, interventional radiologist, or surgeon, preferably under ultrasound guidance to avoid visceral perforation or puncture of the epigastric artery. However, in certain situations of acute renal failure requiring immediate treatment, this healing time is not possible. PD can then be started within 24 hours of catheter implantation (“very urgent start” or “emergency start”), despite a slightly increased risk of mechanical complications, particularly leaks. In these situations (urgent or very urgent start), an appropriate prescription strategy is essential to limit mechanical complications while ensuring adequate purification. Gradually increasing small infusion volumes reduces the risk of dialysate leakage and limits intra-abdominal pressure. To maintain a sufficient dialysis dose despite these low volumes, the regimen is based on increased numbers of short, repeated cycles (45 to 120 minutes), up to 12 exchanges per day, ideally via an automated cycler in the supine position.

Physiologically, this approach meets several requirements. Short cycles combined with highly hypertonic solutions (with a high percentage of glucose) increase the risk of sodium sieving, i.e., initial ultrafiltration consisting mainly of free water, low in solutes. However, in urgent PD situations, it is preferable to achieve ultrafiltration comprising water and solutes, which is more effective for volume management and correction of metabolic disorders. Adjusting the volume, cycle duration, and osmolarity of the solutions therefore aims to optimize this balance between mechanical safety and purification efficiency (Table 1). ³ If the patient’s metabolic (or catabolic-uremic) state allows it, “dry” periods (intermittent PD where the abdominal cavity is empty) are recommended to promote healing of the percutaneous catheter pathway. ³

Data on USPD highlight an increased incidence of minor mechanical complications, with no significant difference in the survival of the technique or transfer to HD. ^14151617 In a retrospective study, patients started within 48 hours required more catheter repositioning than those started between 2 and 13 days. ¹³ However, more robust clinical data from a randomized controlled trial will demonstrate favorable outcomes for USPD compared to temporary “urgent start” HD in patients transitioning to PD. ¹⁸

In line with these observations, experience with late-stage CKD patients (“crashlanders”) without vascular access also supports the safety and efficacy of urgent PD initiation as an alternative to temporary HD via central venous catheter. Available studies and systematic reviews indicate that urgent initiation of PD is associated with an increased risk of mechanical complications, particularly dialysate leaks, without increasing the risk of infection or short- or long-term mortality compared to conventional PD ¹⁹²⁰.Technical survival and patient survival remain comparable between the two modalities, while the risk of bacteremia appears to be significantly lower in urgent PD than in catheter HD ²¹²²²³. Although no specific data are available for ARF, the principle of urgent initiation is considered safe by extrapolation from the results obtained in CRF ²⁴²⁵. Clinical experience reported, notably by Lobbedez and Povlsen, confirms the feasibility and safety of this strategy in crashlanders, provided that close monitoring and an experienced team are available. ¹⁹

The second situation concerns patients with PD who are admitted to intensive care, for example, following septic shock, respiratory distress, or major surgery. In this case, it is often possible to maintain PD rather than inserting a vascular catheter, which avoids systemic anticoagulation and reduces the risk of bacteremia. However, certain circumstances, such as complex abdominal surgery or severe peritonitis, may require temporary use of HD or CRRT until the abdominal problem is resolved. The decision must then be made collectively by nephrologists and intensive care specialists, weighing the benefits and risks. Continuity of care using a dialysis technique familiar to the patient and the healthcare team can, in some cases, limit errors and facilitate daily management, especially in the context of excessive workload in intensive care.

While large multicenter trials such as AKIKI have helped to determine the optimal time to initiate extra renal clearance in intensive care units, ²⁶ the majority of these studies excluded PD, highlighting the need for data specific to this modality. Although dialysis societies ³ position PD as a method that is as valid as HD in the treatment of AKI in various contexts, the efficacy (and predictability) of PD in terms of solute clearance and ultrafiltration for the management of these patients seems controversial among nephrologists and intensive care physicians. Several randomized controlled single-center studies and meta-analyses have highlighted the comparability of its results with those of HD or CRRT, provided that experienced teams are available and certain limitations on the application of PD compared to HD are respected ⁶²⁷²⁸.

Meta-analyses, including a 2017 Cochrane systematic review, involving six randomized controlled trials and 487 patients, mainly from resource-limited countries, indicate that PD is not inferior to extracorporeal HD methods (conventional HD, daily HD, or CVVHDF) in the management of stage 3 ARF in terms of patient survival and renal recovery at 1 month. ²⁸ It should be noted that only one study, the oldest (including patients with ARF from 1993–1998), conducted in Vietnam ²⁹ and subject to numerous biases, had to be stopped prematurely due to a significant survival benefit in favor of extracorporeal purification techniques such as CVVHD(F). The other five studies, all single-center and slightly more recent, demonstrated the non-inferiority of continuous PD at high doses (24 hours/day, 25–44 L/day) compared to HDi/CVV(HD)F with respect to hard endpoints such as 1-month patient survival, as well as other secondary endpoints such as 1-month renal recovery (1-month survival without dialysis), fluid balance (daily UF between 1 and 3 L, comparable with both techniques although with a lower trend in PD), and the rate of complications related to the technique used (mainly infections, hypokalemia, and hemorrhage) ^30313233. A more recent Thai multicenter trial from 2024, including 157 patients suffering mainly from stage 3 septic ARF, confirmed comparable survival and renal recovery at 1 month between conventional HDi and high-dose continuous PD (18–24 L per day), with a comparable weekly UF rate of around 3,000 mL. Better hemodynamic tolerance with less intradialytic hypotension was also reported in the PD group, but this was offset by a need for more frequent supplementation for hypokalemia ³⁰.

In these studies, it should be noted that dialysis prescriptions were very heterogeneous, with daily exchange volumes generally very high, between 18 and 44 liters, reflecting the significant metabolic needs of these stage 3 ESRD patients. Furthermore, most of these randomized trials were conducted in low-income countries with a higher prevalence of PD (Brazil, Thailand, India, Vietnam), with favorable technical expertise and organizational culture, which may have led to a bias in favor of PD. The applicability of these results therefore remains to be demonstrated in our industrialized European countries ³⁴.

PD can offer significant advantages in the specific context of patients with renal failure admitted to intensive care (Table 1). ³⁴ It avoids the risks associated with central venous catheters and extracorporeal blood circuits, such as infections, bacteremia, or catheter or circuit thrombosis (prothrombotic state), and does not require systemic anticoagulation, which is a key advantage for patients at high risk of bleeding or, conversely, coagulopathy. Finally, it is also better tolerated hemodynamically in certain unstable or cardio-renal patients, which may promote faster renal recovery according to some studies. ⁵⁷²⁷³⁵ Its benefits are particularly evident in patients with advanced heart failure or on ECMO (extracorporeal membrane oxygenation), for whom hemodynamic tolerance is a key point, as well as in certain cases of multiple organ failure where vascular access is complex. Numerous studies and experience reports [2,36,37] highlight the feasibility of PD in these situations, provided that organizational obstacles are overcome and trained teams are available.

The main advantages of PD in the intensive care setting also lie in its logistical implementation, which requires less infrastructure (no need for a system to supply osmosis water and discharge it into the sewer system), generally lower costs (supplies, dialysis machine and purification system, staff), and ease of technical learning for healthcare staff compared to HD/CRRT. ³⁸ These advantages are particularly valuable in cases of limited access to HD (low-income countries) or situations of shortages such as during a pandemic (COVID-19) or natural disasters.

However, the use of PD in critically ill patients has absolute or relative contraindications that limit its more widespread use in ICUs, which should be noted: recent abdominal surgery with a risk of peritoneal breach (laparotomy, cesarean section and other gynecological surgery, urological surgery, laparoscopy, retroperitoneal surgery), carcinosis, active peritonitis, uncontrolled intra-abdominal infection or inflammatory process, uremic pericarditis or life-threatening hyperkalemia, pulmonary edema with respiratory failure, and finally, drug intoxication, ³⁷ a combination of situations that are otherwise common in intensive care units (Table 1). It should be noted that non-biocompatible solutions buffered with lactate can “artificially” raise lactate levels without necessarily reflecting tissue hypoperfusion, but rather a delay in hepatic metabolism. ⁷³⁹

One of the main challenges of using PD in intensive care is its effect on respiratory biomechanics, particularly in mechanically ventilated patients. Several studies have shown that the increase in intra-abdominal pressure (IAP) induced by the instillation of 1.5 to 2 L of dialysate (with an increase in IAP of 0.5 to 10 cm H₂O) does not cause significant deterioration in respiratory compliance, resistance, or oxygenation index, provided that the intraperitoneal volume and patient position are adjusted to maintain controlled intra-abdominal pressure. Ponce et al. recommend the use of short cycles and limited exchange volumes to prevent diaphragmatic compression, ⁵ while Almeida et al. confirmed that a volume of 2 L does not reach a critical threshold likely to alter respiratory mechanics in intubated patients. ³¹

In non-intubated patients, data are more limited; studies conducted in healthy volunteers undergoing PD did not show any changes in FEV1/FVC or diffusion capacity, although a reduction in expiratory reserve volume and residual functional capacity was observed. [32,33] In patients with fragile respiratory balance, dialysate instillation could theoretically compromise this balance, particularly in the pre-intubation phase. ²⁷

The experience gained during the COVID-19 pandemic, particularly in patients with acute respiratory distress syndrome (ARDS) and reduced lung compliance, has provided reassuring evidence. ²³⁴³⁵ Several teams have reported the possibility of using infusion volumes of 2 to 2.5 L without impairing respiratory function or desynchronizing ventilation. Nevertheless, it is recommended to stop exchanges during prone positioning phases to avoid an increase in PIP that could interfere with mechanical ventilation. The feasibility of PD in the prone position remains poorly documented, with some isolated reports mentioning technical constraints, such as compression of the tubing, intra-abdominal hyperpressure, or diaphragmatic restriction, ³⁶ requiring consideration of a more lateral catheter exit site than in normal situations.

Although studies in healthy subjects confirm a theoretical reduction in residual functional capacity after a 2 L dialysate infusion, no worsening of hypoxemia (PaO2/F_iO2 r_atio) or increase in the need for intubation after initiation of PD has been reported, including in patients on spontaneous-assisted ventilation. In the reported series, PD effectively corrected fluid and electrolyte imbalances with daily volumes of 15 to 25 L (8 to 20 exchanges per day) and weekly Kt/V values >2.2, sometimes in combination with HD or continuous hemofiltration during periods of prolonged prone positioning. ³⁴³⁶

In summary, when instillation volumes and patient positioning are appropriate, PD does not appear to compromise respiratory mechanics and remains a safe modality in mechanically ventilated patients, provided intra-abdominal pressure and respiratory status are carefully monitored.

The slower clearance kinetics compared to HD for reducing life-threatening hyperkalemia or drug intoxication are at the top of the list of limitations of PD in critical situations. PD remains generally less effective (can reach 25% of HD clearance) than HD or continuous hemodiafiltration techniques for eliminating many toxic substances in cases of acute drug poisoning. ²⁸ In the absence of an alternative or in cases of moderate poisoning, it can be used with partial effectiveness compared to HD in acute poisoning by substances <500 Da that are highly water-soluble and have low protein binding (<80%). Since the elimination of toxins in PD is mainly diffusive, it will therefore depend on the flow rate of the successively infused volumes (number of cycles and stasis time), the molecular weight and serum concentration of the toxic molecule, and its binding to proteins. For example, certain toxins—such as lithium, acyclovir, phenobarbital, ethylene glycol, and methanol—can still be effectively eliminated by high-volume PD within 48–72 hours. ⁴² Due to the faster clearance rate in HD, PD should not be offered in life-threatening situations or unless other renal replacement methods are unavailable⁴⁰.

UF in PD remains less predictable and slower than in HD, although more than 2 L per day can be removed using hypertonic solutions and/or successive short cycles ³. In cycles of less than 120 minutes, the removal of free water may predominate over the removal of salt-coupled water, and care should be taken to avoid the risk of hypernatremia ⁴¹. Hydraulic catheter dysfunction may also occur as a result of constipation or even ileus, which is very common in patients who are bedridden for long periods, leading to additional mechanical ultrafiltration failure. Critical patients should therefore be treated with laxatives as a preventive measure, and drainage should be performed in the supine position rather than the prone position.

Among the mechanical complications associated with USPD, intra- or extraluminal catheter obstructions are common. They can be caused by the formation of intraluminal fibrin clots, which is promoted by the procoagulant state of patients in intensive care or major fecal stasis. Careful monitoring of the catheter is essential, along with corrective measures such as rigorous aseptic flushing and the systematic addition of heparin (500 to 1,000 U/L) in the first exchanges and mild laxatives to prevent obstruction⁴².

Catheter migration (often in the upper position) or mechanical twisting can also cause hydraulic dysfunction. An abdominal X-ray can usually identify these abnormalities, and targeted repositioning using interventional radiology or laparoscopy, depending on the type of anesthesia used, may be necessary⁵.

Early leaks (along the subcutaneous path or around the insertion area of the inner sleeve) are another common complication, especially if too large volumes are instilled immediately after catheter placement. This situation can be avoided by limiting exchange volumes to less than 1–1.5L L during the first 72 hours of treatment, maintaining strict supine position during exchanges and periods of empty abdomen (8 hours/24 hours min) to promote healing ³.

Experience gained during the COVID-19 pandemic has demonstrated the feasibility of PD in UDR. Catheter insertion was successful in 85–95% of cases in selected patients, although those with abdominal obesity, a history of complex subumbilical surgery, or severe thrombocytopenia/coagulopathy were excluded. [2,43,44] The rate of major complications, including gastrointestinal perforations and dialysate leaks, remained low (<2–5%). This apparent safety can be partly explained by a center effect (technical expertise of teams in applying emergency PD protocols), a bias inherent in this type of observational report.

While PD spares patients from the risk of central catheter-related bacteremia, peritonitis (infection of the dialysate fluid), whether hand-carried or enteric in origin, remains a feared complication in critically ill patients. As the usual signs of abdominal pain are masked, it is necessary to regularly monitor the nucleated elements (NE) in the dialysate, ³ especially in the context of prolonged antibiotic therapy, so as not to miss a secondary fungal peritonitis. In addition, antibiotic prophylaxis with vancomycin or first-generation cephalosporin, depending on the center’s microbiology, ³ is recommended during PD catheter placement to prevent early infections of the exit site or tunnel after placement (percutaneous or laparoscopic).

All of these mechanical or infectious complications specific to PD warrant close multidisciplinary management involving nephrologists, intensive care specialists, and dedicated nursing staff. ^727293345

Furthermore, in the context of acute peritoneal dialysis in intensive care, the pharmacokinetics of intraperitoneally administered antibiotics differ from those observed in chronic settings. Recent data show that the elimination of certain molecules, particularly cephalosporins and vancomycin, is clinically significant in acute PD, with significant variability in serum and intraperitoneal concentrations. In particular, therapeutic vancomycin concentrations are not always achieved and are associated with clinical success, suggesting that plasma concentration monitoring is warranted to adjust dosages and optimize infection cure in critically ill patients. ⁴⁶

L.J. wrote the first draft of the manuscript. A.-L.C. supervised the development and writing of the text. M.T., J.L., A.L., M.T.S., K.F., M.C., and G.G. participated in the critical revision of the manuscript and made substantial intellectual contributions to its improvement. All authors read and approved the final version.

NA

NA

NA

The authors received no financial support for the research, writing, and/or publication of this article. They declare that they have no conflicts of interest with respect to the conduct of this study.

The authors declare that they have not conflict of interest with this work.

Lucas jacobs : https://orcid.org/0000-0001-8148-0349

Maxime Taghavi : https://orcid.org/0000-0003-1442-9716

Jabber Laouni : https://orcid.org/0009-0006-8771-7072

Adrien Lengelé : https://orcid.org/0000-0003-1724-8830

Mohamed Tayeb Salaouatchi : https://orcid.org/0000-0003-3730-772X

Karlien François : https://orcid.org/0000-0003-0071-8402

Giuseppe Gambino : https://orcid.org/0009-0007-2138-5834

Anne-Lorraine Clause : https://orcid.org/0000-0003-2469-1910