Renal Association Clinical Practice Guideline on peritoneal dialysis in adults and children

Evidence from observational studies or registry data, with all its limitations, indicate that peritoneal dialysis (PD) used in the context of an integrated dialysis programme is associated with good clinical outcomes, certainly comparable to haemodialysis in the medium term (HD) and potentially better in the first 2 years of dialysis [1‐10]. NICE recommends PD as the initial dialysis treatment of choice of chronic kidney disease stage 5 for children aged 2 years or older, people with residual renal function and adults without significant associated comorbidities (NICE Clinical Guideline 125, 2011). The only randomised study (NECOSAD), comparing HD to PD as a first treatment showed no differences in 2 year quality adjusted life years or 5 year mortality, but the number randomised was insufficient to generalize this observation; notably, most patients in this national study had sufficient life-style preferences related to one modality to decline randomisation [11]. PD has a significant technique failure rate however, so patients need to be able to switch treatment modality (to either temporary or permanent HD) in a timely manner, which has implications for HD capacity and the timing for HD access creation.

PD modalities (CAPD v. APD) have a different impact on life-style; one randomised study found that APD creates more time for the patient to spend with family or continue employment but is associated with reduced quality of sleep [12]. APD is usually the preferred modality for children [13]. There are medical indications for APD (see sections 2, 3 and 4), but generally initial modality choice is a lifestyle issue. Studies suggest no difference in outcomes resulting from selection of CAPD or APD as initial PD modality [14‐16].

The success of a PD programme is dependent upon specialised nurses with appropriate skills in assessing and training patients for PD, monitoring of treatment and with sufficient resources to provide continued care in the community. A randomised trial of more intensive training has shown that this reduces peritonitis risk [17] and there is some evidence to support the benefit of regular home reviews of PD technique [18] (see section 5). Several studies have documented the benefits of home visits in identifying new problems, reducing peritonitis and non-compliance [19‐21]. The National Renal Workforce Planning Group, (2002), recommended a caseload of up to 20 PD patients per nurse. It is important to note that this was a minimum recommendation. For smaller adult units, and paediatric units, a significantly greater number of nurses than determined by this ratio will be required to maintain a critical number to provide adequate specialist nurse cover across the year and to cover periods of absence. This is increasingly relevant now with the decline in PD patient numbers and unit sizes that has occurred since the publication of the Workforce Planning document. It is also of note that the responsibilities of PD nurses vary significantly between units, for example in some additionally being responsible for inpatient PD care, such that the required staffing level will be higher than this minimum. Greater numbers of nurses will be required where assisted PD is performed by staff from the PD unit rather than other external organisations. The requirement for specialist nurses with the skills to deal with complex patient educational issues is highlighted by the ISPD Guideline (2016) for teaching PD to patients and caregivers [22]. Having a designated lead clinician for PD in each unit may help to promote PD as a therapy option and to develop clinical management policies.

Assisted PD, with provision of nursing support in the community to help with part of the workload and procedures associated with PD, is a useful option to overcome an important barrier to home dialysis therapy [23]. Assisted APD should be available for patients, who are often but not always elderly, wishing to have dialysis at home, but are unable to perform self-care PD [24] and may also be used as a temporary measure for established patients temporarily unable to perform PD independently or for those unable to start PD alone but may later become independent. Assisted PD provides at least equivalent outcomes to in-centre haemodialysis for older patients [25‐27], and higher treatment satisfaction [27] and is a viable option for expanding home care in more dependent patients [25, 26].

Disconnect systems have been shown through randomised trials to be associated with a lower peritonitis risk, especially in infections due to touch contamination [28].

A minority of patients commencing PD will experience infusion pain, often severe enough to consider discontinuing the therapy. A double blind randomised study demonstrated that pain could be prevented by using a normal pH, bicarbonate-lactate buffered dialysis fluid (Physioneal) [29]. Subsequent clinical experience has found that the benefit of this more biocompatible solution on infusion pain results in immediate and sustained benefit, and is probably applicable to other biocompatible solutions.

The evidence of other forms of clinical benefit from the routine use of biocompatible solutions is more controversial. Standard solutions are clearly bio-incompatible, with low pH (~5.2), lactate rather than bicarbonate buffer, high osmolality and high concentrations of glucose which also result in high concentrations of glucose degradation products (GDPs). Many in vitro and ex vivo studies have demonstrated the relative toxicity of these solutions, with all of the bioincompatible features playing their part [30‐35]. There is also strong observational evidence that firstly detrimental functional changes to the peritoneal membrane occur with time on treatment, which are more exaggerated in patients using solutions with high glucose concentration early in their time on therapy [36, 37] and secondly, that morphological changes occur that are related to time on treatment which include membrane thickening and vascular scarring [38]. Time on treatment is also the greatest risk factor for encapsulating peritoneal sclerosis (EPS) [39, 40].

These observations have led dialysis companies to develop and market ‘biocompatible’ solutions, with normalization of pH, and/or reduction of GDPs and variable approaches to buffering. In randomised clinical trials these solutions have been shown to improve the dialysate concentrations of biomarkers considered to be indicators of mesothelial cell and possibly membrane health [41‐44]. Systemic benefits possibly include reduced circulating advanced glycation end-products [44] and better glycaemic control in diabetics [45]. Data is currently lacking on hard clinical endpoints including technique failure or patient survival. One non-randomised, retrospective observational study has found an improved patient but not technique survival; patients in this study using biocompatible solutions were younger, suggesting a selection bias that may not be fully adjusted for, so caution should be exercised in the interpretation of this study [46]. Similar findings have been reported in a subsequent observational study, which has the advantage of including analysis of cohorts matched for factors including cardiovascular comorbidity, socioeconomic status and centre experience [47].

However, the limitations of being a non-randomised study with no fixed indication for prescription of biocompatible fluid, with potential for selection bias, and with differences in characteristics of the unmatched groups still apply [47]. Non-randomised, observational studies have also suggested a beneficial effect of biocompatible solutions on peritonitis rates [48, 49], but the strength of the conclusions are limited by the non-randomised study design and possibility of other factors contributing to observed differences in infection rates. A secondary outcome of the randomised balANZ trial was of a reduction in peritonitis rates in group receiving biocompatible PD fluid [50]. However, the most recent and largest registry study reported an increased risk of peritonitis with biocompatible fluids [51] and a recent systematic review has not demonstrated a benefit of low-GDP biocompatible solutions on peritonitis rates, patient or technique survival [52]. Thus further studies are required to answer the question regarding the potential effect of biocompatible fluids on PD peritonitis. The balANZ study also demonstrated interesting differences in effect on peritoneal membrane function. The biocompatible fluid group had a higher initial transport state one month after starting the trial, but transport status was then stable, unlike the standard fluid group where transport sate increased progressively [53]. The impact of this effect on outcomes including technique survival warrants further study.

The area with the strongest evidence for clinical benefit of biocompatible solutions is in the preservation of residual renal function. Several studies have suggested a benefit of low-GDP biocompatible fluids on residual function, with the largest being the balANZ trial [54]. Whilst differences in ultrafiltration between groups (which may indirectly affect residual urine via effects on hydration), make interpretation of the actual effect of the fluids on residual renal function more difficult in some studies [55], three systematic reviews of existing trials demonstrate a benefit of biocompatible solutions on residual renal function, when used for more than 12 months [52, 56, 57]. We suggest that biocompatible solutions be considered for preservation of residual kidney function. Currently there is insufficient evidence to recommend that all patients should be treated with biocompatible solutions, especially as this may have a significant cost implication. The argument for their use may be stronger if there was not an economic disadvantage. However, we note that routine clinical practice in UK is for children receiving PD to routinely be treated with biocompatible solutions.

The arguments and rationale for this guideline relate to the National Service Framework for Renal Services, Part 1. The reader is referred to standard 2, Preparation and Choice pp. 21–23. The vast majority of patients commencing dialysis are medically suitable to receive PD if they select it. Some commonly perceived medical “contraindications” to PD are overstated. The majority of patients with a previous history of major abdominal surgery may successfully be treated with PD [58]. It is also unusual to be unable to achieve target small solute clearances in the majority of larger patients (with the availability of APD, even when anuric).

When patients present needing prompt, unplanned start to renal replacement therapy, rapid insertion of a PD catheter with acute start of PD, along with fast track education regarding dialysis modalities, may allow a proportion to commence directly on PD, avoiding temporary vascular access and urgent haemodialysis [59‐61]. Such patients who initially receive acute start of haemodialysis should receive follow up education regarding RRT options.

The arguments and rationale for this guideline relate to the National Service Framework for Renal Services, Part 1. The reader is referred to standard 3, Elective Dialysis Access Surgery, pp. 24–26. The Moncrief catheter is buried subcutaneously and is designed to be left in this position, where it can remain for many months, until required [62].

Recommendations for management of PD catheter insertion in adults are contained in the Renal Association Peritoneal Access Guidelines. The same principles apply in paediatric practice, except that procedures in children will routinely be performed under general anaesthetic. For management of the catheter in the peri-operative period, for catheter related problems including leak (internal and external), poor flow, obstruction and hernias, the guidelines developed by the International Society of Peritoneal Dialysis, www.​ispd.​org [63, 64] and the European Elective Chronic Peritoneal Guideline [13] should be used. Catheter problems due to increased intra-peritoneal pressure, especially leaks, hernias and prolapse are an important medical indication for the use of APD either temporarily or permanently; poor flow or catheter related flow pain should be treated with tidal APD. In the majority of cases where surgical repair for mechanical complications is required (e.g. catheter replacement, hernia repair) it is possible to avoid the need to temporary haemodialysis. In many PD patients, remaining residual renal function may permit an adequate period post-surgery before dialysis needs to be recommenced. Where PD does need to start soon after surgery, in many cases this may be safely achieved by initial use of APD with small volume exchanges and avoiding a day dwell in ambulant patients [65].

Small solute clearance is one of the measurements of adequate dialysis treatment. Salt and water removal and acid-base balance are considered in sections 4 and 6 respectively. There are two issues in measuring small solute clearance that need to be taken into consideration.

First, the relationship to clinical outcomes of residual renal versus peritoneal small solute clearance is quantitatively different. Observational studies have shown that preserved renal clearance, in fact just urine volume, is associated with improved survival, independent of other known factors such as age and comorbidity [66, 67]. Randomised controlled trials designed to replace this residual renal function with peritoneal clearance did not show a proportional survival benefit [68, 69]. The recommendation to measure solute clearance six-monthly is driven primarily by the residual renal function component; indeed if dialysis dose has not been changed the peritoneal component will not be different and it would be acceptable just to measure the residual renal function. Indeed RRF can fall rapidly in some patients, certainly within a few weeks. If there are clinical concerns (e.g. if changes in symptoms, blood biochemistry, reported fall in urine output or after potential insults to residual renal function), or if achievement of solute clearance target is dependent on residual renal function, this should be undertaken more frequently.

Second, there are two potential surrogate solutes, urea and creatinine, that can be used to measure solute clearance in PD patients. There is no clear evidence as to which is the more useful clinically, and both have their problems. Current advice, therefore, is that either one or both can be used, ensuring that minimal clearances are achieved for at least one, but clinicians should be aware of their differing limitations. Urea clearances are limited by the difficulty in PD patients of estimating V accurately, whilst peritoneal creatinine clearances are affected by membrane transport characteristics (see Appendix).

Two randomised controlled trials (ADEMEX and Hong Kong) have evaluated the impact of peritoneal solute clearances on clinical endpoints [68, 69]. Neither found that an increase of peritoneal Kt/V_urea > 1.7 was associated with an improvement in survival. Only one of these studies (ADEMEX) measured creatinine clearance, which was the solute used to make decisions in this case; patients in the control group achieved an average peritoneal creatinine clearance of 46 L/1.73m²/week and a total (urine plus renal) of 54 L/1.73m²/week. In setting a recommendation for minimal peritoneal clearances, to be achieved in anuric patients, the previous Renal Association guideline of Kt/V > 1.7 and creatinine clearance >50 L/1.73m²/week is supported by both the randomised and observational data. In the Hong Kong study, patients randomised to a Kt/V < 1.7, whilst their mortality was not significantly worse they had a significantly higher drop out rate, more clinical complications and worse anaemia. One observational longitudinal study demonstrated that patients develop malnutrition once the Kt/V falls below 1.7 with a three-fold increase in the death rate [70]. The NECOSAD study found that a creatinine clearance of <40 L/week or a Kt/V urea <1.5 was associated with increased mortality in anuric patients [71].

The vast majority of PD patients will be able to reach these clearance targets, especially if APD is employed [72]. These guidelines must however be viewed as recommendations for minimal overall clearance. In patients with residual renal function this renal clearance can be subtracted from the peritoneal clearance with confidence that the value of equivalent renal clearances is greater. Equally, in a patient achieving these clearances but experiencing uraemic symptoms, including reduced appetite or nutritional decline, or failing to achieve adequate acid base balance (see section 6) then the dialysis dose should be increased. Drop out due to uraemia or death associated with hyperkalaemia and acidosis was significantly more common in the control patients in the ADEMEX study [68]. In patients with borderline clearances, who fail to achieve these clearance targets, other aspects of patient wellbeing, long-term prognosis from other comorbidity and patient perspective should be considered in deciding whether switch of modality to haemodialysis is appropriate. It is important to note that spuriously low Kt/V urea may arise due to overestimation of V in patients with significant obesity (see Appendix).

ADEMEX randomised patients between a “standard” CAPD regime of 4 × 2 l exchanges (rather than a specific clearance value) vs enhanced prescription to obtain specified clearance targets [68]. Thus this study should not be used to justify routine reduction of dialysis prescription down to minimum clearance targets. The large ANZDATA observational study suggested a lower survival with low peritoneal Kt/V [73]. One possible interpretation of the data is that low peritoneal clearances were linked to reduced dialysis prescription in patients with good residual renal function.

There is a discrepancy between clearance of small solutes and larger molecules, which are more dependent on time of contact of dialysate with the peritoneal membrane than dialysate volume [74]. Thus continuous regimes are preferred to those with “dry” periods (e.g. NIPD), particularly in anuric patients, even if small solute clearance targets can be achieved without continuous therapy. An exception to this is in the situation where a patient still has a large residual renal function.

In paediatrics there is a lack of high quality evidence to determine clearance targets for children on PD. In small children and infants, Kt/V is likely to be disproportionately high compared with creatinine clearance and adult targets are particularly inadequate in these patients [75]. It is suggested by British Association of Paediatric Nephrology that the adult targets should be considered as minimum desirable, with an increase in PD prescription in the presence of features of uraemia, including inadequate growth [76]. Evidence in small numbers of subjects has suggested that in children increasing dialysis prescription may reach a point of no further benefit or adverse effects on nutrition due to increased dialysate protein removal [77].

Assessment of membrane function, specifically solute transport rate and ultrafiltration capacity) is fundamental to PD prescription. (See Appendix for methodological description of membrane function tests). This is for the following reasons:

The European Renal Best Practice advisory board have produced detailed recommendations for the methodology of evaluation of peritoneal membrane function in clinical practice, and for utilising the results in PD prescription [89].

Residual renal function, as discussed above, is one of the most important factors, along with age, comorbidity, nutritional status, plasma albumin and membrane function that predict survival in PD patients. Its rate of loss is variable and clinically significant changes can occur within 6 months. Total fluid removal is associated with patient survival, especially once anuric [85, 90, 91].

Increased solute transport has been repeatedly shown to be associated with worse survival, especially in CAPD patients [82‐84, 86]. The explanation for this association is most likely to be because of its effect on ultrafiltration when this is achieved with an osmotic gradient (using glucose or amino-acid dialysis fluids). The reason is twofold: first, due to more rapid absorption of glucose, the osmotic gradient is lost earlier in the cycle resulting in reduced ultrafiltration capacity. Second, once the osmotic gradient is dissipated the rate of fluid reabsorption in high transport patients is more rapid. This will result in significant fluid absorption, contributing to a positive fluid balance, during the long exchange.

These problems associated with high transport can be avoided by using APD to shorten dwell length and by using icodextrin for the long exchange to prevent fluid reabsorption. Several randomised controlled trials have shown that icodextrin can achieve sustained ultrafiltration in the long dwell [92‐96] and that this translates into a reduction in extracellular fluid volume [97, 98]. Observational studies indicate that high solute transport is not associated with increased mortality or technique failure in APD patients, especially when there is also a high use of icodextrin [84, 85, 99]. Results from the ANZDATA Registry show that in high transport patients, treatment with APD was associated with a superior patient survival compared with CAPD [100]. Survival in low transport patients in contrast was lower with APD. A Korean registry study reported a benefit of icodextrin on patient and PD technique survival [101] but adequately powered randomised trials to confirm this are still needed [102].

A difference in practice for paediatrics is that patients with an underlying diagnosis of renal dysplasia are often polyuric, and so not so dependent on peritoneal ultrafiltration for maintenance of euvolaemia.

There is growing evidence that regular use of hypertonic glucose dialysis fluid (3.86%), and where possible glucose 2.27%, is to be avoided as far as possible. It is associated with acceleration in the detrimental changes in membrane function that occur with time on treatment [80, 103], as well as several undesirable systemic effects including weight gain [94, 104], poor diabetic control, delayed gastric emptying [105], hyperinsulinaemia [106] and adverse haemodynamic effects [107]. In addition to patient education to avoid excessive salt and fluid intake, where possible the use of hypertonic glucose should be minimised by enhancing residual diureses with the use of diuretics (e.g. frusemide 250 mg daily) [108]. Substituting icodextrin for glucose solutions during the long exchange will result in equivalent ultrafiltration whilst avoiding the systemic effects of the glucose load [94, 98, 107]. Observational evidence would suggest that icodextrin is associated with less functional deterioration in the membrane in APD patients [103].

This is the single most important parameter in PD patients, and also the one most likely to change with time. Clinically significant changes can occur within three months. Because secretion of creatinine by the kidney at low levels of function overestimates residual creatinine clearance, it is recommended to express this as the mean of the urea and creatinine clearances. Observational and randomised studies have shown that episodes of volume depletion, whether unintentional or in response to active fluid removal with the intent of changing blood pressure or fluid status, are associated with increased risk of loss in residual renal function [97, 98, 109]. Care should be taken not to volume deplete a PD patient too rapidly or excessively. The need to determine an appropriate target weight to avoid the cardiac complications of occult fluid overload, whilst avoiding loss of residual renal function due to excessive fluid removal is a major challenge in the management of the PD patient who has still has a significant residual urine output. The use of diuretics to maintain urine volume is not associated with a risk to renal clearances [108]. ACE inhibitors, (Ramipril 5 mg) [110] and ARBs (valsartan) [111] have been shown in randomised studies in adults to maintain residual diuresis. A Cochrane review also suggested superior preservation of residual function in PD with ACEis or ARBs [112]. Evidence for a benefit of ACE inhibitors or ARBs to preserve residual renal function in children is lacking, and a recent report from the International Pediatric Peritoneal Dialysis Network registry suggested that renin-angiotensin blockade could be associated with an increased risk of loss of residual renal function in children [113], and so these drugs are not recommended for preservation of kidney function in paediatric PD patients. Paediatric practice may also differ with the management of a subgroup of patients with renal dysplasia and a tendency to polyuria.

Observational studies have consistently shown that reduced peritoneal ultrafiltration is associated with worse survival rates; whilst this is seen in studies with or without residual urine [90], this effect is most marked in anuric patients [85]. In the only prospective study to have pre-set an ultrafiltration target (750 ml/day), patients who remained below this had higher mortality after correcting for age, time on dialysis, comorbidity and nutritional status. It is likely this association is multifactorial, but failure to prescribe sufficient glucose or icodextrin and a lower ultrafiltration capacity of the membrane were factors in this study and should be considered [85, 114]. The European guidelines have suggested a 1 l minimal daily ultrafiltration target [115] but there is insufficient evidence to say that such a target must be met at this stage. It is possible that in some patients with low ultrafiltration, this is appropriate to their low fluid intake, and that in these cases decreased survival possibly results from poor nutrition rather than fluid excess, and that increasing ultrafiltration would simply result in dehydration with its adverse effects. Blood pressure, salt (and fluid) intake, nutritional and fluid status, and presence of any features of uraemia should be taken into account. Nevertheless patients with less than 750 ml ultrafiltration once anuric should be very closely monitored and the potential benefits of modality switch considered.

The rationale underpinning the guidelines in this section is laid out in a series of documents published by the International Society of Peritoneal Dialysis, available on their web-site: www.​ispd.​org.

Prevention strategies: The ISPD 2016 PD-related infections guideline, the ISPD 2011 position statement on reducing the incidence of PD-related infections, 2017 ISPD catheter-related infection recommendations and the 2012 ISPD guideline for prevention and treatment of catheter-related infections and peritonitis in paediatric patients receiving PD [116‐119] place increasing emphasis on prevention strategies. Regular audit is essential to this progress with a team approach to quality improvement [117] and the following standards should be considered as minimal:

Patient training to perform PD technique by experienced PD nurses trained to do this as part of a formalised training programme is essential in patients commencing PD [120]. Greater experience of nurses providing training is associated with greater time to initial episode of peritonitis [121]. It is recommended that review of patient PD technique is performed on a regular basis, at least annually, or more frequently if there is evidence of inadequate technique or development of PD –related infection, or a significant interruption in the performing PD e.g. after a significant period of hospitalisation). Approaches that have been shown to reduce infection rates in randomised studies include increased intensity of training, use of flush before fill systems, antibiotic prophylaxis to cover catheter insertion and prevention of exit-site infections [116, 117]. Several studies have addressed the latter issue; following demonstration that the risk of Staph aureus exit site infection (the organism most frequently responsible) is associated with pre-existing skin carriage, several randomised studies demonstrated that clinical exit-site infection and associated peritonitis could be reduced by either nasal or exit-site application of mupirocin. This has led to the practice of applying mupirocin to all patients [122, 123] and this approach should be discussed with the local microbiology and infection control team. A systematic review has confirmed the benefits of prophylactic mupirocin in preventing exit-site infections and Staph aureus peritonitis [124] A more recent study, comparing mupirocin with gentamicin cream, found that the latter prevented both Staph aureus and Pseudomonas exit-site infections and peritonitis episodes [125]. This approach should be considered in patients with a known history of Pseudomonas infections; again the policy should be discussed and agreed with the local microbiology team.

The International Society of Peritoneal Dialysis (ISPD) has developed a simple scoring system for exit site signs and symptoms which is easy to use and gives guidance on when to treat immediately rather than waiting for a swab result. Purulent discharge is an absolute indicator for antibiotic treatment [126].

The ISPD has become less dogmatic about the initial choice of antibiotic treatment for peritonitis, provided that gram positive and negative infections are covered [116]. It is recognised that patterns of resistance vary considerably and thus a local policy must be developed. Studies do not currently demonstrate a favoured regime [127]. For exit site infections the presence of a tunnel infection should be recognised as it may require more aggressive management. We concur with the ISPD guidelines that suggest suitable antibiotic dosing regimens, including options for intermittent and continuous dosing of intraperitoneal antibiotics. We also note their comment that infections from Gram negative organisms are more likely to lead to refractory or recurrent peritonitis. A single study suggested that treating Gram negative peritonitis with 2 appropriate antibiotics might be associated with better outcomes. It is also important to be aware of the potential for impaired absorption of oral antibiotics in some situations, e.g. co-prescription of ciprofloxacin with some phosphate binders [128].

We would emphasise the ISPD guidelines that it is important that timely PD catheter removal is undertaken in refractory PD peritonitis [116]. PD catheter removal or swap is also required in refractory exit site infections, and may be required earlier where there is a Pseudomonas infection or associated tunnel infection, which can be confirmed by ultrasound imaging [126, 129].

There will be a lower threshold in paediatrics for admission for IV antibiotics (at least for first 48 h), especially in infants and small children where oral antibiotics commonly cause diarrhoea/feed intolerance.

Glycaemic control can be made worse by glucose absorption across the peritoneal membrane. Dialysis regimens that incorporate less glucose and more glucose free (amino acid, icodextrin) solutions have been shown to improve glycaemic control [130, 131]. Diabetes is a rare cause of end-stage renal failure in paediatrics, but these principles would also apply to children on PD who have diabetes. The IMPENDIA-EDEN randomised controlled study compared all-glucose regimes with regimes including both icodextrin and amino acid PD dialysis fluids in diabetic patients on PD demonstrated a 0.5% reduction in glycated haemoglobin [131]. Serum triglyceride, very-low-density lipoprotein, and apolipoprotein B also improved. However it is important to note that the intervention group suffered an increase in adverse events and deaths, including events related to extracellular fluid expansion [131]. It is therefore critical that this approach with use of low-glucose solutions is accompanied by careful monitoring of hydration and is not at the expense of a decline in fluid management. It also should not be an alternative to appropriate use of hypoglycaemic drugs, and monitoring for hypoglycaemia is important in patients where dialysate glucose load is reduced.

Although there is no strong equivalent evidence in paediatrics, it is suggested that the principles of minimisation of peritoneal glucose exposure to avoid obesity and reduce the adverse effects on peritoneal membrane function should also apply to children.

Two randomised controlled trials have suggested that clinical outcomes, including gaining lean body mass and reduced hospital admissions are achieved if the plasma bicarbonate is kept within the upper half of the normal range [132, 133]. Generally this can be achieved by using dialysis fluids with a 40 mmol buffer capacity (lactate or bicarbonate results in similar plasma bicarbonate levels [134] and ensuring that the dialysis dose is adequate (see section 3 (b), above) [135]. However, for solutions with a lower buffering capacity, when patients are switched from an all lactate (35 mmol/l) to a 25 mmol bicarbonate: 10 mmol lactate mix, there is a significant improvement in plasma bicarbonate (24.4 to 26.1 mmol/l), such that a higher proportion of patients will fall within the normal range [136]. Whilst bicarbonate solutions may have a role in biocompatibility (see section 1(e), above), they are generally not required to achieve satisfactory acid-base balance in adults. The main reason for using a 35 mmol buffer capacity solution (25:10 bicarbonate:lactate mix) is to avoid excessive alkalinisation [137]. Plasma bicarbonate will also be affected by some phosphate binders that either increase, or occasionally (sevelamer hydrochloride) decrease concentrations. In paediatric practice in UK, use of neutral pH/low GDP solutions is routine.

Control of acidosis is especially important in malnourished patients who may benefit from the glucose available in dialysis solutions as a calories source. Amino acid solutions were developed in an attempt to address protein calorie malnutrition and several randomised studies have been conducted. In using amino acid solutions it is essential to ensure that acidosis does not develop and to use the solution at the same time as there is a significant intake of carbohydrate [138]. Despite demonstration that amino acids delivered in dialysis fluids are incorporated into tissue protein, the randomised trials have failed to show benefit in terms of hard clinical endpoints [139, 140].

Weight gain, or regain, is common after starting peritoneal dialysis and this is associated with a worsening in the lipid profile [141, 142], though there may not be a significant difference from haemodialysis [143]. Randomised studies comparing glucose 2.27% with icodextrin in the long exchange have shown that the latter prevents weight gain, which in body composition studies is at least in part fat weight. Substituting icodextrin for 2.27% glucose in the long dwell also improves insulin resistance [144]. There is limited available trial data on the benefit of statins in PD patients with a hard clinical endpoint. The 4D and AURORA studies did not include PD patients, and whilst SHARP included 33% dialysis patients, only 5% of the study patients were receiving PD. There is no data on the effects of lipid-lowering in children on PD. There are good reasons for believing that the lipid abnormalities in the PD patient population may be different to patients on HD, and potentially more atherogenic. The KDIGO guideline for lipid management in CKD suggests that statins and/or ezetimibe are not commenced in dialysis patients, but that they are continued if a patient is receiving them before stating dialysis [145] though it is important to note that the majority of evidence this is based on is derived in haemodialysis patients. Observational data in one trial of adults has suggested a possible benefit of statins in adults receiving PD [146]. The Canadian Society of Nephrology Guidelines suggest that statins and/or ezetimibe should be considered for adult PD patients [147].

Use of icodextrin is associated with circulating levels of metabolites that can interfere with laboratory assays for amylase (or actually suppress amylase activity) [148‐151] and for glucose when finger-prick tests that utilise glucose dehydrogenase as their substrate are employed (manufactured by Boehringer Mannheim) [152‐155]. In the case of amylase, the measured level will be reduced by 90%, leading to the potential failure in the diagnosis of pancreatitis. No adverse events have been reported, but clinicians should be aware of this possibility. If clinical concern remains then plasma lipase can be used. In the case of glucose measurements, the methods using glucose dehydrogenase will overestimate blood glucose levels, leading to a failure to diagnose hypoglycaemia. This has been reported on several occasions in the literature and has contributed to at least one death. Typically these errors occur in places and circumstances in which staff not familiar with peritoneal dialysis work, for example emergency rooms and non-renal wards. A number of solutions to this problem are under active review (e.g. use of alarm bracelets) but it is also the responsibility of health-care professionals to ensure that clinical environments in which their patients using icodextrin may find themselves are notified of this issue on a routine basis.

Encapsulating peritoneal sclerosis (EPS) is rare, but serious complication of long-term PD. It involves formation of an inflammatory, and later fibrotic, “cocoon” surrounding the gastrointestinal tract [156]. This results in features of abdominal inflammation and intestinal obstruction. Symptoms may include abdominal pain, nausea, vomiting and haemoperitoneum and may predate definitive diagnosis by significant time periods in some instances. Typical appearances will be noted at laparotomy or laparoscopy. EPS should be distinguished from the thickening of the peritoneal membrane that typically occurs with time on PD, but which is not associated with obstructive features. Changes in peritoneal membrane small solute transport and ultrafiltration capacity often occur [157‐159], but are also common in long-term PD and not always present in EPS, so are not of diagnostic value for EPS. There is no gold standard for the diagnosis of EPS, and it is recommended that the condition is diagnosed by the presence of the combination of characteristic clinical and radiological features [160, 161]. A challenge in this is that there is significant heterogeneity in the condition with variation of severity and extent of peritoneal involvement [156, 162, 163]. The epidemiology, clinical features, investigation and management of EPS in paediatric patients is similar to that in adults [164, 165]. Recommendations are developed from the UK Encapsulating Peritoneal Sclerosis Clinical Practice Guidelines 2009 [166].

Radiology plays a key role in the diagnosis of EPS. Plain abdominal X-rays may show features of bowel obstruction, but are non-diagnostic, except in cases where peritoneal calcification is present as a feature suggestive of EPS. CT scanning is recommended as the definitive radiological investigation for the diagnosis of EPS [167‐172]. It has high reproducibility and evaluation has provided the basis of a standardised approach to CT diagnosis of EPS [171]. The presence of peritoneal calcification, bowel wall thickening, bowel tethering, and bowel dilatation are the features with greatest agreement between reporting radiologists [171]. Abdominal ultrasound may detect characteristic features in EPS [173]. However, there is a limitation to depth of penetration of sound waves which may limit ability for thorough evaluation of the abdomen, and it is operator-dependent. Small bowel contrast studies may also have a role in defining the presence of strictures prior to surgery.

At present, there are no investigations that can be recommended to monitor or screen patients on long-term PD to identify those who will develop EPS. One study has demonstrated that in patients developing EPS, who had abdominal CT scans for other reasons within a period of a year or less prior to diagnosis of EPS, there were no radiological abnormalities to suggest imminent development of EPS [174].

EPS is a rare and complex condition, whose optimal management requires integrated care from an expert team experienced in managing this condition. Multiple disciplinary input includes PD physicians, nurses, surgeons, dieticians, radiologists and intensive care physicians.

There is increasingly strong evidence for a central role for surgery in the management of EPS [175‐178]. Whilst earlier experience of EPS reported a high mortality for patients with this condition, and complications following surgery, in experienced hands, surgery results in high rates of resolution of symptoms and survival, and possibly superior relief of obstruction compared with conservative treatment with nutrition and/or drug treatment [178]. Surgery should be performed by a surgical team which has a high level of expertise and experience with EPS, and the appropriate multidisciplinary input and peri-operative renal and intensive care support. Surgical units at Manchester (Mr Titus Augustine), and Cambridge (Mr Chris Watson) are designated as national referral centres for surgery relating to EPS in England by NCG (National Commissioning Group). An early surgical opinion facilitates decisions regarding the need for, preparation and timing of surgery. Indications for surgery include non-responsiveness to medical treatment, bowel obstruction (acute and recurrent subacute), intraperitoneal bleeds, and peritonitis. A proportion of patients with EPS may have a good outcome without surgery so further work to define those most likely to benefit from surgery is needed. Where possible, surgery should be timed to take place electively before the patient is too ill or nutritionally depleted. Surgery involves careful dissection of thickened peritoneum from bowel loops to achieve maximal removal of sclerotic membrane from the bowel wall, whilst avoiding inadvertent perforation [176].

Reduced nutritional intake resulting from intestinal dysfunction, plus an ongoing inflammatory state in EPS, can lead to severe protein energy wasting [179, 180]. Nutritional state is associated with survival in EPS. Patients with EPS should be referred early to a renal dietician to allow nutritional assessment, monitoring and institution of nutritional support where needed. In more severe cases, parenteral nutrition may be required [180], and in patients where intestinal function does not recover, this may be required on a permanent basis. In milder cases, nutrition support may be managed with an energy dense diet or prescription of oral nutritional supplements and anti-emetics. Where patients are unable to tolerate adequate oral intake, nasogastric or nasojejunal feeding may be utilised.

Whilst there has been much interest in drug treatments for EPS, there is no robust evidence to support the use of anti-inflammatory or antifibrotic drugs in this condition. Corticosteroids have been most commonly used, particularly in the Japanese literature [178]. Any benefit is most likely with use in the early inflammatory stage of EPS. However there is not strong objective evidence for their effectiveness, and in EPS side effects of immunosuppression and protein catabolism are a particular concern. There are reports of use of immunosuppressants including azathioprine and cyclosporine in EPS. However evidence is largely as case reports, and as a common setting for development of EPS is following transplantation, in patients taking these drugs, their therapeutic effectiveness is doubtful. There is increasing interest in the role of tamoxifen, which is effective in other fibrotic conditions, in EPS [181, 182]. There is a suggestion from retrospective data of a beneficial effect of tamoxifen on survival [183] or that it could even have a preventative role [184], but robust evidence is currently lacking.

PD is usually discontinued and the PD catheter removed after diagnosis of EPS, with transfer to haemodialysis. However, as some cases are mild, the individual patient’s wishes and clinical state should be considered, as stopping PD may not be appropriate in a patient with mild symptoms and a poor long term prognosis, where continuation of PD and/or later conservative management may be appropriate. Also, there is experience in Japan of leaving the PD catheter in and performing peritoneal lavage after diagnosis of EPS, with observational non-randomised studies suggesting some benefit, though this approach is not widespread in other countries [185, 186].

The risk of developing EPS is extremely low in the first 3 years of PD and low before 5 years of therapy. The overall reported incidence typically varies between 0.5–3% in reported series [187‐190]. Whilst the risk increases with time, the majority of patients on longer term PD will not develop EPS. The Scottish Renal Registry is notable in reporting a steeper rise of incidence with time on PD, with 8.1% risk of EPS after 4–5 years of PD [189]. Thus consideration of management of patients remaining on PD for longer term is warranted. However, it is unknown what impact elective discontinuation of PD after a certain period of time will have on the risk of developing EPS. A significant proportion of cases of EPS occur after discontinuing PD (either transplantation [191, 192] or switching to haemodialysis), so it is even possible that elective switching from PD could increase, rather than decrease, the risk of developing EPS. Discontinuing PD may also have potentially major adverse negative medical and social effects in some patients. Concern about EPS risk on long-term PD should be balanced against reports showing relatively good outcomes for EPS, relative to other competing risks [190], and good outcomes and success rates for EPS surgery when required. Thus routine discontinuation of PD after a fixed period of time cannot be recommended. The risks and benefits of continuing PD or dialysis modality change should be considered and discussed with the individual patient, as recommended in the ISPD 2009 Length of Time on Peritoneal Dialysis and Encapsulating Peritoneal Sclerosis Position Paper [193] (revised position paper will be published 2017).