The Dose Response Multicentre Investigation on Fluid Assessment (DoReMIFA) in critically ill patients

Abstract

Background: The previously published "Dose Response Multicentre International Collaborative Initiative (DoReMi)" study concluded that the high mortality of critically ill patients with acute kidney injury (AKI) was unlikely to be related to an inadequate dose of renal replacement therapy (RRT) and other factors were contributing. This follow-up study aimed to investigate the impact of daily fluid balance and fluid accumulation on mortality of critically ill patients without AKI (N-AKI), with AKI (AKI) and with AKI on RRT (AKI-RRT) receiving an adequate dose of RRT.

Methods: We prospectively enrolled all consecutive patients admitted to 21 intensive care units (ICUs) from nine countries and collected baseline characteristics, comorbidities, severity of illness, presence of sepsis, daily physiologic parameters and fluid intake-output, AKI stage, need for RRT and survival status. Daily fluid balance was computed and fluid overload (FO) was defined as percentage of admission body weight (BW). Maximum fluid overload (MFO) was the peak value of FO.

Results: We analysed 1734 patients. A total of 991 (57 %) had N-AKI, 560 (32 %) had AKI but did not have RRT and 183 (11 %) had AKI-RRT. ICU mortality was 22.3 % in AKI patients and 5.6 % in those without AKI (p < 0.0001). Progressive fluid accumulation was seen in all three groups. Maximum fluid accumulation occurred on day 2 in N-AKI patients (2.8 % of BW), on day 3 in AKI patients not receiving RRT (4.3 % of BW) and on day 5 in AKI-RRT patients (7.9 % of BW). The main findings were: (1) the odds ratio (OR) for hospital mortality increased by 1.075 (95 % confidence interval 1.055-1.095) with every 1 % increase of MFO. When adjusting for severity of illness and AKI status, the OR changed to 1.044. This phenomenon was a continuum and independent of thresholds as previously reported. (2) Multivariate analysis confirmed that the speed of fluid accumulation was independently associated with ICU mortality. (3) Fluid accumulation increased significantly in the 3-day period prior to the diagnosis of AKI and peaked 3 days later.

Conclusions: In critically ill patients, the severity and speed of fluid accumulation are independent risk factors for ICU mortality. Fluid balance abnormality precedes and follows the diagnosis of AKI.

Keywords: AKI; Critical illness; Fluid overload; ICU; RRT.

Fig. 1

Patient flow chart. AKI acute kidney injury, Cr creatinine, RRT renal replacement therapy

Fig. 2

Fluid accumulation for N-AKI, AKI, AKI-RRT patients during the first 5 days following admission. Looking at the first 5 days of ICU stay, the cumulative fluids differed significantly each day among the three groups (N-AKI, AKI and AKI-RRT). (p and p ^* refer to the p values of the Kruskal-Wallis test and the correction for the multiple test situation with the Bonferroni test, respectively). There was progressive fluid accumulation in N-AKI and AKI patients. AKI-RRT patients accumulated a similar degree of fluid, followed by a decrease. Patients were daily assigned to the corresponding group (N-AKI, AKI and AKI-RRT). AKI patients with acute kidney injury, AKI-RRT patients with acute kidney injury treated with renal replacement therapy, FO fluid overload, N-AKI patients without AKI

Fig. 3

Delta of cumulative fluids pre and post day of AKI. The day of AKI served as a reference point. A backward calculation of cumulative fluid balance was conducted for the 3-day period pre and post day of AKI. Patients were censored on day of AKI recovery. Fluid accumulation up to 3 % occurred within a few days. AKI acute kidney injury, FO fluid overload, StdErr standard error

Fig. 4

Maximum FO (MFO) during ICU stay in N-AKI, AKI and AKI-RRT patients. MFO was higher in patients with AKI and particularly high in those treated with RRT. The horizontal axis shows the median day when MFO occurred. AKI patients with acute kidney injury, AKI-RRT patients with acute kidney injury treated with renal replacement therapy, N-AKI patients without AKI

Fig. 5

Cumulative fluid balance of AKI patients treated with renal replacement therapy. The cumulative fluid balance of AKI-RRT patients is illustrated at four different time points: (a) day of AKI diagnosis, (b) day of RRT, (c) day of maximum FO (MFO), (d) last RRT day. Cumulative fluid balances were computed also for survivors and non-survivors. The horizontal axis shows the median day when the relevant events occurred. AKI acute kidney injury, AKI-RRT patients with acute kidney injury treated with renal replacement therapy, IQR interquartile range

Fig. 6

Thirty-day survival of N-AKI, AKI, and AKI-RRT patients. The Kaplan-Meier analysis including the first 30 days of ICU stay indicated a significant survival benefit for patients without AKI (p < 0.0001). The AKI-RRT group had the lowest survival rate and AKI patients who did not receive RRT had intermediate survival rates. AKI patients with acute kidney injury, AKI-RRT patients with acute kidney injury treated with renal replacement therapy, N-AKI patients without AKI

Fig. 7

Cumulative fluid balance prior to death or discharge. This analysis includes patients who stayed in the ICU for at least 5 days. Non-survivors (n = 156) had progressive fluid accumulation in the 4 days before death whereas cumulative fluid balance decreased in survivors (n = 854). FO fluid overload, StdErr standard error

Fig. 8

Maximum fluid overload in survivors and non-survivors. Maximum fluid overload (MFO) was calculated for survivors and non-survivors among N-AKI, AKI and AKI-RRT patients. In all cohorts, non-survivors had a higher MFO. AKI patients with acute kidney injury, AKI-RRT patients with acute kidney injury treated with renal replacement therapy, N-AKI patients without AKI

Fig. 9

Impact of maximum fluid overload. The risk of death increased exponentially with the magnitude of maximum fluid overload (MFO). In this model, follow-up was limited to the median time in ICU (12 days). Circles represent the number of observations of survivors and non-survivors (at the bottom and at the top respectively)

Fig. 10

Predicted probability of death adjusted for severity of illness. The figure shows maximum FO (MFO) and predicted probability of death adjusted for APACHE II scores 1, 17.12, 27 and 53. For this model, follow-up was limited to the median time in ICU (12 days). AKI acute kidney injury, APACHE II Acute Physiology and Chronic Health Evaluation II

Fig. 11

The speed of fluid accumulation. The fluid overload slope (FO_SL) computed for three generic patients. Solid line: FO (%); dashed line: ICU admission (FO = 0 %) to maximum FO (MFO) straight line; solid bold line: FO slope