The dose of RRT is a measure of the quantity of a solute that is removed from the patient during extracorporeal treatment and is reasonably representative of other solutes which require removal [1, 2]. In patients on chronic hemodialysis, dose of treatment is expressed as URR or Kt/V (K = dialyzer clearance of urea, t = dialysis time and V = volume of distribution of urea). Both parameters have important limitations in critically ill patients with AKI where neither urea generation rate nor volume of distribution can be clearly defined. In patients receiving continuous renal replacement therapy (CRRT), clearance of small uncharged molecules such as urea and creatinine is essentially equal to the delivered effluent rate. Therefore, the effluent rate roughly corresponds to the prescribed replacement or dialysis rate and is often used as a surrogate of urea clearance. Following the landmark study by Ronco et al. in 2000, it was suggested to index the flow rate to the patient’s body weight [3]. As a result, the dose of CRRT is often expressed as the amount of dialysis/hemofiltration flow delivered to the patient in ml/kg per hour. Whether to use actual body weight or ideal body weight is unclear [3–8].

Degree of ‘pre-dilution’ and ‘filter-down’ time are important factors which reduce the effective dose. They need to be taken into account when prescribing and reviewing the dose of RRT. Infusion of replacement solution as pre-dilution will reduce effective effluent dose by the degree to which the plasma is diluted. The final dilution effect is dependent on circuit blood flow, replacement fluid rate, and haematocrit. Furthermore, discrepancies between prescribed doses and measured creatinine clearance increase with higher doses over time as demonstrated for predilution continuous veno-venous hemodiafiltration (CVVHD) [9]. Premature circuit clotting or need for investigations outside the ICU are common reasons for unintended interruptions in treatment which can lead to reduced clearance [10]. It is therefore necessary to review regularly whether the delivered dose of RRT matches the prescribed target and to adjust the prescription if necessary.

Of note, effluent rate only represents clearance of small solutes but not larger molecules or molecules with high protein binding. In addition, there are many other important aspects of RRT which need to be considered when prescribing a dose, like acid–base homeostasis, nutritional support and, perhaps most importantly, fluid balance.

Since 2000, there have been eight randomized controlled trials (RCTs) on intensity of RRT in AKI [3–8, 11, 12]. Two studies evaluated RRT doses in patients receiving intermittent haemodialysis (IHD) [11, 12], five studies in patients on CRRT [3–5, 7, 8] and one study enrolled patients on IHD, slow extended dialysis (SLED) and CRRT [6]. Two single center studies showed better outcomes with increased intensity of small solute clearance [3, 7]. In contrast, the two largest multi-center RCTs, the ATN study (n = 1,124) and the RENAL study (n = 1,464), showed no benefit in survival or recovery of renal function with higher doses of RRT [4, 6]. A subsequent meta-analysis of all eight RCTs concluded that higher intensity RRT did not reduce mortality rates or improve renal recovery in patients with AKI [13]. Current international guidelines recommend delivering an effluent volume of 20–25 ml/kg/h for post dilution CRRT in AKI [14, 15]. Increasing the dose beyond 20–25 ml/kg/h has not been shown to be beneficial and may potentially result in losses of important solutes including phosphate and antibiotics, and heat. In patients receiving CRRT in pre-dilution mode, the target dose should be increased to 25–30 ml/kg/h in order to achieve a delivered dose of 20–25 ml/kg/h. For patients receiving intermittent RRT for AKI, international recommendations differ. While the guideline by the Kidney Disease Improving Global Outcome (KDIGO) expert group recommends a target Kt/V of 3.9 per week for intermittent or extended acute RRT [15], the European Renal Best Practice (ERBP) guideline recommends not to use Kt/V as a marker of adequacy and to adapt the duration of IHD to metabolic and volume status [14].

It is important to acknowledge that in the RCTs mentioned earlier, the RRT dose was not adjusted for severity of disease or degree of catabolism. Instead, patients were treated with a fixed dose indexed to the patient’s body weight. Finally, the above mentioned RCTs evaluated different doses of RRT but did not study possible pleiotropic effects of RRT in patients with sepsis or the effect of fluid balance on outcome.

There is increasing evidence that fluid overload is detrimental to both renal outcome and survival in critically ill patients with AKI, including patients treated with RRT [16, 17]. In a retrospective analysis of the RENAL study the authors found that a negative daily fluid balance during the treatment period was independently associated with a shorter ICU and hospital stay and lower 90 day mortality [18]. The multicenter observational FINNAKI study demonstrated an association between fluid overload at RRT initiation and increased 90-day mortality, which remained significant after adjustment for common risk factors [18, 19].

The relationship between fluid accumulation, AKI and outcome is complex. Fluid overload may be a marker of the severity of AKI but may also be causing harm as a result of interstitial edema, visceromegaly and secondary organ dysfunction. Based on existing data, it would be advisable to target a negative fluid balance in patients on RRT, as soon as the patient is adequately resuscitated and hemodynamic status allows.

It is not possible to recommend a general net ultrafiltration rate. Instead, the ultrafiltration rate should be tailored to the patients’ needs and haemodynamic and fluid status.

It is rather inopportune that studies in the literature have defined high-volume hemofiltration (HV-HF) by ultrafiltrate rates of 35–200 ml/kg/h. In RCTs in animals HV-HF was only beneficial when using very high ultrafiltrate rates (>100 mL/kg/h) and initiating hemofiltration early (i.e., before or very early after the septic challenge) [20, 21]. The number of RCTs in humans is limited and their size is small [22–26]. Important differences among animal and human studies include the later initiation of HV-HF, the lower ultrafiltrate rates and the use of antibiotics in humans [20]. The two largest recent RCTs were negative: a Chinese study in 280 patients comparing high volume (50 ml/kg/h) versus very high volume (80 ml/kg/h) [26] and the IVOIRE study in 140 patients comparing 35 ml/kg/h versus 70 ml/kg/h [25]. In a recent systematic review and meta-analysis HV-HF was defined as continuous high-volume treatment with an effluent rate of 50–70 ml/kg/h (for 24 h per day) or intermittent very high volume treatment with an effluent rate of 100–120 ml/kg/h for a 4–8 h period followed by conventional renal dose hemofiltration [27]. Four studies (470 participants) were included and the authors concluded that HVHF, compared with standard renal dose had no significant impact on short-term mortality, kidney recovery, improvement in hemodynamic profile, or reduction in ICU or hospital length-of-stay. Another more extensive systematic review and meta-analysis including 7 RCTs (558 patients) using the same cut-off of 50 ml/kg/h but additionally investigating pulse HV-HF with 85–100 ml/kg/h over 8 h confirmed the lack of effect on relevant endpoints such as renal recovery as well as vasopressor requirements or cytokine clearance [28]. The application of continuous HV-HF hemofiltration or pulse very HV-HF in severe sepsis and septic shock cannot be recommended based on the results of human studies.