Electrolyte balance of CRRT depends on baseline plasma concentrations, the composition of the dialysate or replacement fluids, CRRT dose and type of anticoagulation. Hypokalemia and hypophosphatemia are the most faced electrolyte disturbances during RRT. During citrate CRRT, calcium balance depends on the target plasma ionized calcium concentration of the protocol, but is often negative. Some algorithms provide a zero calcium balance option. During heparin CRRT, calcium balance is generally slightly positive. Magnesium balance is generally negative, more so during citrate than during other anticoagulation, due to the low magnesium concentration of commercial fluids. During CRRT, plasma concentrations of potassium, phosphate, calcium and magnesium should be monitored. Choose replacement fluids with concentrations close to target plasma concentrations or supplement these electrolytes separately.

The most common electrolyte disorders observed during CRRT are hypokalemia and hypophosphatemia [10–12]. Due to frequent hyperkalemia in patients with acute kidney injury before start of CRRT, solutions with a low potassium concentration (0–2 mmol/L) are developed. Then, hyperkalemia is rapidly normalized by CRRT. However, after some hours of treatment, hypokalemia occurs. This is more so for phosphate, because most replacement fluids are phosphate-free. Hypophosphatemia can occur early during hemofiltration treatment, especially with higher CRRT dose. Although few disturbances are registered in some studies (in particular with pediatrics patients [11, 13], other electrolytes are correctly regulated, because concentrations in the substitution fluids are close to normal [14, 15].

There are clearly some differences between the different modalities of RRT treatment in terms of electrolytes removal and acid–base equilibrium. The diffusion principle used in the dialysis mode is more effective for small molecule removal (as electrolytes) than hemofiltration at standard dose (25 ml/kg/h). However, the convection principle used in the hemofiltration mode can remove much more electrolytes when used at high volume (more than 35 ml/kg/h) [16]. However, the most important parameter for electrolyte removal remains the dose of renal replacement (effluent volume). Electrolyte deficiencies are more frequent when high effluent volumes are used [17]. In two large randomized controlled trials comparing low- and high-intensity RRT, more electrolyte imbalances were found in the high-intensity group [2, 3].

Furthermore, continuous techniques are more able to normalize hyperkalemia or hyperphosphatemia than intermittent hemodialysis, but are also more likely to induce electrolyte deficits like hypophosphatemia [9]. Accordingly, CVVH achieves better acid–base balance and electrolyte removal than extended daily dialysis, but can also induce severe hypophosphatemia [18].

The big challenge for the industry is to provide fluids as close as possible to the plasma composition [15, 17].

The typical calcium concentration of the bicarbonate- and lactate-buffered fluids is about 1.75 mmol/L, substantially higher than the plasma ionized calcium concentration, which is the amount accessible for filtration or dialysis. Thus, calcium balance during conventional CRRT is generally slightly positive (about 5–12 mmol/day) [19, 20]. However, during citrate CRRT, calcium handling is different and calcium balance of many protocols is negative [19, 20]. Most fluids used for citrate CRRT are calcium-free and calcium is generally replaced by separate infusion. For calcium replacement, each protocol has its own ionized plasma calcium target, ranging between 1.0 and 1.35 mmol/L. Especially with lower plasma calcium targets, calcium balance is negative, 12–96 mmol/day is reported [19, 20]. Of note, a negative calcium balance does not necessarily cause systemic hypocalcemia, due to a rapid parathormone response, which leads to unwanted calcium release from bone [21]. Targeting a higher plasma calcium concentration seems to suppress PTH release, which may limit calcium mobilization from bone. However, PTH measurements in critically ill patients should be interpreted with caution, because the normal assay measures oxidized PTH as well, which is biologically inactive [22]. Some modern CRRT devices incorporate an algorithm with a 100 % calcium compensation option, which seems attractive. Up to now, optimal calcium targets are not known. Hypocalcemia is common in patients with sepsis, and it is not known whether correction is beneficial [23]. However, a prolonged negative calcium balance for days should likely be avoided.

Magnesium balance during CRRT has gained little attention. The typical magnesium concentration of commercial CRRT fluids is about 0.5 mmol/L, while high normal or increased magnesium concentrations are targeted in ICU patients to prevent arrhythmias, promote vasodilatation and anti-inflammatory and anti-oxidant defense [24]. When using these fluids, mean magnesium balance during CVVHDF was −17 mmol/day in heparin circuits and −26 mmol/day in circuits anticoagulated with citrate [25]. Apart from calcium, citrate chelates magnesium, explaining the higher magnesium loss during citrate CRRT in this study. Hypomagnesemia is reported by others as well [26] and is even associated with non-recovery of renal function [27]. Given the important role of magnesium during critical illness, additional magnesium should be supplemented, the amount depending on the fluids in use, the local CRRT protocol and the plasma magnesium concentration.

Standard electrolyte monitoring should include potassium (two to four times daily), phosphate and magnesium (twice daily), total calcium (once daily) and ionized calcium (four times daily during citrate anticoagulation). See also the chapter on citrate anticoagulation.

In patients receiving renal replacement therapy, enteral nutrition is preferred. Specific so-called renal feeding preparations are not needed. The dose of protein should be between 1 and 1.5 g/kg/day. Energy provision should likely be between 20 and 25 kcal/kg/day. Amino acid loss through the filter accounts for 10–20 % of administered protein. Water soluble vitamins are lost and should be replaced. Lipid soluble vitamins are not lost and do not require regular replacement. Some trace elements are lost and regular trace element administration seems prudent. However, all recommendations regarding nutrition in these patients are based on weak evidence.

The preferred mode of nutritional support in critical illness is early enteral nutrition (EN) [28, 29]. In the absence of more definite evidence, CRRT patients should receive EN within 24–48 h following admission titrated to gut tolerance. EN is preferred to parenteral nutrition (PN) because of its safety, low cost, and its physiological route and hormonal signalling. The role of supplemental PN in CRRT patients is unknown.

The target goal of nutrition remains ill defined. Providing 20–35 kcal/kg/day of calories with 1.5–1.8 g/kg/day of protein in patients with AKI on CRRT appears to allow a reasonable balance between protein catabolism and nitrogen balance [30]. This empirical caloric amount seems consistent with calculated requirements [31]. Due to water soluble vitamin losses, it seems prudent to administer water soluble vitamins daily or second daily. As some trace elements are also lost, it seems prudent to administer trace elements daily or second daily.

These recommendations for CRRT patients are based on clinical observational studies, small studies with physiological outcomes and physiological reasoning. They are low-level recommendations (level C) and open to challenge. Such lack of evidence suggests the need for level 1 studies in this field.

The optimal protein intake is unclear, and the flexibility of adjusting intake is limited by the composition in standard enteral formulas. For example, Isosource^® 1.5 cal (Nestle) contains 1.5 kcal/ml of calories and 67.6 g of protein per litre. Intake of 30 kcal/kg/day only delivers 1.4 g/kg/day of protein. Higher protein intake of 2.5 g/kg/day results in a near positive or positive nitrogen balance and offers the only active measure to reduce the negative nitrogen balance so commonly seen in these patients [32, 33]. A positive nitrogen balance is associated with improved patient survival in critical illness, but improved survival is not a direct effect of increased protein intake [31]. However, there may be a dose-related association between increased protein intake and clinically significant improvement in renal function in critically ill patients [34]. Supplemental PN does not lead to improved clinical outcomes [35] and, if given early, may lead to clinical problems [36]. There is no evidence that specific administration of essential AA in preference to general AA preparations leads to any improved clinical outcomes.

The molecular weights of proteins range from 55 to 220 kDa, well above the cut-off of standard CRRT membranes, while that of AA is only 110 Da (75–204 Da). AA loss during CRRT has been studied mostly in the context of total parenteral nutrition (PN).

Most AA have a high sieving coefficient (easily filtered) of near 1.0, such as cysteine, arginine, alanine and glutamine, except for glutamic acid (0.25–0.5) [37–39]. AA clearances range from 20 to 45 ml/min in patients on CRRT [40], as compared to about 200 ml/min in patients with AKI on intermittent hemodialysis [41]. In general AA loss represents about 11–12 % of total (parenteral) protein intake. The amount of AA lost in each patient depends on the intensity of CRRT but also in great part on AA blood levels, which in turn depend on the amount administered and route of administration.

Protein intake of more than 1.5 g/kg/day is associated with AA loss of about 12 g/day [39], and AA loss seems less with lower protein intake [41]. AA losses can account for 5 % to a substantial 20 % of daily AA intake [37–39, 41]. Individual blood AA levels correlate well to corresponding losses in CRRT [41], and overall AA loss is higher with higher intensity of treatment [37]. In addition, there is selectivity in individual AA losses, with glutamine and alanine being lost in greater absolute amounts [39, 41], and tyrosine having the highest fractional loss per intake [37]. All three are non-essential AA under normal physiological states.

Total nitrogen loss in CRRT is about 25 g/day [38, 39]—the vast majority of which is due to urea nitrogen with 10–20 % (2–5 g/day) contributed by AA nitrogen loss [37–39]—while the rest is likely from protein catabolism; this results in a negative nitrogen balance. Intermittent hemodialysis has been reported to lead to losses of 6–8 g/session [17].