Urine Volume in Acute Kidney Injury: How Much Is Enough?

Abstract

Eight hundred and seventy-nine patients with acute kidney injury were retrospectively studied over year and eleven months for evaluation of urine volume as a risk factor for death. They were divided into five groups, according to the 24 h urine volume (UV): anuric (UV ≤ 50 mL/24 h, group 1), oliguric (UV > 50 mL/24 h and < 400 mL/24 h, group 2), and non-oliguric (UV ≥ 400 mL/24 h). Nonoliguric group was subdivided in three subgroups: UV > 400 mL/24 h and ≤ 1000 mL/24 h (group 3, reference group), UV > 1000 mL/24 h and ≤ 2000 mL/24 h (group 4), and UV > 2000 mL/24 h (group 5). Linear tendency test (Mantel extension) pointed out a significant increase in mortality with UV decrease (p < 0.001), confirmed by multivariate analysis. Anuric and oliguric patients had increased risk of respectively 95% and 76% times for death compared to controls (p < 0.05). Patients from groups 4 and 5 presented a reduced risk for death of 50% and 70%, respectively, p = 0.004 and p = 0.001. In conclusion, urine volume was a strong independent factor for mortality in this cohort of AKI patients.

Keywords:

• acute kidney injury
• urine volume
• mortality
• prognosis

INTRODUCTION

In the last five decades, the mortality rate of acute kidney injury (AKI) patients has remained very high, despite a better understanding of acute renal injury pathophysiology and significant improvements in critical care and renal replacement therapy.^[1–12] Several factors might be responsible for the maintenance of this elevated mortality, such as patients' age, presence of co-morbidities, and a prominent occurrence of AKI in a context of multiple organ failure. In fact, currently, the so-called low-mortality AKI is rarely seen, such as secondary to blood transfusion, septic abortion, or isolated drug nephrotoxicity.^{[1,6,10,13–16]} Oliguria, which is usually defined as urine volume < 400 mL/24 h, and hypervolemia are frequently described among the theoretically modifiable mortality risk factors studied for AKI.^{[7,8,13,17–20]} Although nonoliguric AKI has been frequently associated to a better survival rate,^{[7,10,19–21]} pharmacologic conversion from oliguric to non-oliguric status has not resulted in increased AKI survival.^[22–29]

In this study we evaluated the influence of urine volume in AKI mortality. Considering the higher amounts of fluid currently administered to severely ill patients, we hypothesized that urine volume above the classical nonoliguric definition might be associated to reduction in patients' mortality.

PATIENTS AND METHODS

Eight hundred and seventy-nine consecutive patients with acute kidney injury, followed for a time period of one year and eleven months at Hospital das Clínicas (a tertiary university hospital from University of Sao Paulo Medical School, Sao Paulo, Brazil), were studied in a retrospective cohort with the objective of evaluating risk factors for AKI mortality, including urine volume.

AKI was defined as serum creatinine ≥2.0 mg/dL or serum urea ≥ 50 mg/dL in patients with previously normal serum creatinine (≤1.4 mg/dL), or a 20% increase in serum creatinine in patients with serum creatinine previously elevated (≥1.5 mg/dL) associated to an identified causal factor for the renal injury. Patients with advanced chronic kidney disease, stages 4 and 5 (serum creatinine ≥6 mg/dL and renal cortex <2 cm on ultrasonography), without identifiable causal factors for AKI; infants (<1 year old); transplant recipients; and patients with missing data preventing the fulfilment of inclusion criteria were excluded.

Included patients were divided in groups according to the urine volume during the 24 h before the nephrology consultation: anuric (urine volume ≤ 50 mL/24 h, group 1), oliguric (urine volume > 50 mL/24 h and < 400 mL/24 h, group 2), and non-oliguric (urine volume ≥ 400 mL/24 h). The nonoliguric group was subdivided in three subgroups: urine volume ≥ 400 mL/24 h and ≤ 1000 mL/24 h (group 3, reference group), urine volume > 1000 mL/24 h and ≤2000 mL/24 h (group 4), and urine volume > 2000 mL/24 h (group 5).

The following parameters were evaluated in all patients:

• AKI etiology, based on clinical diagnosis: acute tubular necrosis, prerenal, hepatorenal syndrome, urinary obstruction, vascular obstruction, glomerulonephritis, vasculitis, and acute interstitial nephritis;

• dialysis indications: uremia (signals/symptoms of uremia, or serum urea > 200 mg/dL, or serum potassium ≥5.5 mEq/L and/or serum bicarbonate ≤ 15 mEq/L resistant to clinical treatment) and/or hypervolemia (positive fluid balance with clinical manifestations);

• need of mechanical ventilation and vasopressor drugs;

• presence of electrolytic disturbance; and

• sepsis.

The primary endpoint was mortality during nephrology follow-up.

Laboratory Methods

Jaffe method was used for serum creatinine, kinetic test with urease for serum urea, COBAS INTEGRA for sodium and potassium, and gas and pH analyzer for serum bicarbonate determinations.

Statistical Analysis

Data are presented as mean ± standard deviation or proportion. Comparison of groups was done by student t-test, ANOVA, Kruskal-Wallis test, chi-square with Yates correction, or Fisher test, as appropriate. Multivariate logistic regression (BMDP-LR software) was performed to evaluate risk factors for mortality. The model tested included those variables significant in the univariate analysis, those considered clinically important, and potential confounding variables. The dependent variable was mortality. The variables included in the model were urine volume (categorized as previously described, considering 400 < urine volume ≤1000 mL/24 h as reference), age (reference < 65 years of age), sex (reference male), AKI etiology (renal, post-renal, and functional as reference), peak level of serum urea (reference < 100 mg/dL), and the following variables: reason for dialysis (uremia or hypervolemia), need for ventilatory support, need for vasoactive drugs, presence of hyperkalemia, presence of hyponatremia, metabolic acidosis, and sepsis (using their absence as reference). Backward variable selection was used to remove nonsignificant factors. The variables that, when excluded, introduced a change in parameter estimates greater than 10% were re-introduced to the model to account for confounding. Goodness-of-fit of the model was assessed by the Hosmer and Lemeshow test. Wald test was used to assess the significance of variables in the models. A second logistic regression model was tested, with the same variables of the first model, but with urine volume included as a continuous variable. Linear tendency test (Mantel extension) was done to evaluate if there was a linear association between increase in urine volume and mortality. The level of significance was established as p < 0.05.

RESULTS

One thousand three hundred fifty-one patients were followed by the AKI group of HC-FMUSP in a one year and eleven month period. Four hundred seventy-two (35%) patients were excluded: 338 of them were excluded because of stage 4 or 5 chronic kidney disease, and the remainder were children < 1 year old or patients with missing data.

Eight hundred seventy-nine AKI patients were included: 145 (16.4%) were anuric (group 1), 175 (20%) were oliguric (group 2), and 559 (63.6%) were nonoliguric. Among nonoliguric patients, 248 (28.2% of all AKI patients) had urine volume >400 ml and ≤1000 mL/24 h (group 3 or control), 247 (28.1%) had urine volume >1000 ml and ≤2000 mL/24 h (group 4), and 64 (7.3%) had urine volume >2000 mL/24 h (group 5).

There was no difference among groups regarding age, gender, and urea peak value. Male gender was more prevalent in all groups. Peak creatinine value was higher in anuric patients as compared to groups 4 and 5 (p < 0.001). Acute tubular necrosis (ATN) was the most usual etiology for AKI, found in 67% of all patients (p < 0.001 vs. others). Nonoliguric patients required less frequently mechanic ventilation (47% nonoliguric vs. 76% anuric, p < 0.001, vs. 70% oliguric, p < 0.001) and vasopressors (43% nonoliguric vs. 72% anuric, p < 0.001, vs. 69% nonoliguric p < 0.001) than the other two groups. These data are summarized in Table 1.

Table 1  Patients' baseline characteristics: gender, age, serum urea and creatinine (mg/DL), causes of AKI,need for mechanical ventilation, and vasoactives drugs

Group (n)	1 (145)	2 (175)	3 (248)	4 (247)	5 (64)	3+4+5 (559)	Total (879)
Gender
M %	57	58	62	68	83	67	64
F %	43	42	37	32	17	33	36
Age (y)	55 ±18	55 ±19	57 ±19	51 ±19	55 ±17	55 ±19	55 ± 19
U (mg/dL)	210 ± 82	203 ± 76	219 ± 86	204 ± 84	210 ± 102	212 ± 84	210 ± 84
Cr (mg/dL)	6.6 ± 4	5.6 ± 3	5.8 ± 3	5.5 ± 3	4.9 ± 3	5.5 ± 3	5.7 ± 3
ATN (%)	71	73.7‡	66	61	70.3	64	67
PR (%)	7	8	18	21.5	14	19	14.7
HR (%)	6	5.7	3	0.4	1.5	2	3.3
UTO(%)	10││	3.4	3	7	8││	5.4	5.7
VO (%)	5││	0.6	0.4	0.8	0	0.6	1.3
GN/VASC (%)	0§││	8.5	8	4.8	3.1	6	6
AIN (%)	1	0	1.6	4.5	3.1	3	2
MV (%)	76	70	54	42	44	47││	57
VAD (%)	72	69	52	36	31	43││	53

*p < 0.05 vs. female; †p < 0.001 vs. oliguric and non-oliguric, ‡vs. other causes; §p ≤ 0.05 vs. non-oliguric; ││p < 0.001 vs. anuric and oliguric.

Abbreviations: F = female, U = urea, Cr = creatinine, ATN = acute tubular necrosis, VO = vascular obstruction, PR = pre-renal, GN/VASC = glomerulonephritis/vasculitis, HR = hepatorenal, AIN = acute interstitial nephritis, UTO = urinary tract obstruction, MV = mechanical ventilation, VAD = vasoactive drugs.

There were less dialysis indication in nonoliguric patients (40%) than in anuric (64%) and oliguric (62%) patients (p < 0.001; see Table 2). In patients with urine volume > 1000 mL/24 h (groups 4 and 5), the most usual indication for dialysis was uremia (p < 0.01). In contrast, hypervolemia was the most common indication for dialysis in oliguric (87%) and anuric (78%) patients (p < 0.001).

Table 2  Frequency of dialyzed patients among groups and indication for dialysis: comparison among groups 1 (anuric, urine volume ≤50 mL/24 h), 2 (oliguric, urine volume >50 mL/24 h and <400 mL/24 h), 3 (urine volume ≥400 mL/24 h and ≤1000 mL/24 h), 4 (urine volume >1000 mL/24 h and ≤2000 mL/24 h), and 5 (urine volume >2000 ml)

Group (n)	1 (145)	2 (175)	3 (248)	4 (247)	5 (64)	3 + 4 + 5 (559)	Total (879)
Dialysis (%)	64	62	48	34	28	40	48
Uremia (%)	69	68	84	98	78	94	82
HV (%)	87	78	60	44	22││	51	66

*p < 0.001 vs. anuric and oliguric; †p < 0.05 vs. group 3; ‡p < 0.01 vs. group 3; §p < 0.001 vs. non-oliguric; ││p = 0.05 vs. groups 3 and 4.

Abbreviation: HV = hypervolemia.

When compared to the anuric patients, the nonoliguric group had less hyperkalemia (39% vs. 54% anuric, p < 0.001), hyponatremia (28% vs. 40% anuric, p = 0.006), and metabolic acidosis (39% vs. 54% anuric, p = 0.001). There was no difference between nonoliguric and oliguric patients regarding these parameters. The infection rate was similar among groups, compromising 79% of all patients studied (see Table 3).

Table 3  Frequency of hyperkalemia, hyponatremia, acidosis, and infection among groups 1 (anuric, urine volume ≤ 50 mL/24 h),2 (oliguric, urine volume > 50 mL/24 h and < 400 mL/24 h), 3 (urine volume ≥ 400 mL/24 h and ≤ 1000 mL/24 h), 4 (urine volume > 1000 mL/24 h and ≤ 2000 mL/24 h), and 5 (urine volume > 2000 ml)

Group (n)	1 (145)	2 (175)	3 (248)	4 (247)	5 (64)	3 + 4 + 5 (559)	Total (879)
Hyper K	54	45	42	37	30	39	42
Hypo Na	40	30	34	25	17	28	30
Acidosis	54	48	42	38	31	39	43
Infection	79	82	77	78	80	78	79

*p < 0.001 vs. anuric and oliguric; †p = 0.01 vs. anuric; ‡p = 0.01 vs. oliguric; §p = 0.01 vs. non-oliguric.

Abbreviations: Hyper K = hyperkalemia; Hypo Na = hyponatremia; acidosis = metabolic acidosis.

The mortality rate was significantly lower in all nonoliguric groups (p < 0.001) as compared to the overall, oliguric, and anuric group mortalities. Table 4 shows the risk factors found by bivariate analysis. The risk of death was significantly higher in anuric and oliguric patients and significantly lower in patients who had urine volume >1000 ml and ≤2000 ml and urine volume >2000 mL/24 h. Linear tendency test (extension Mantel test) pointed out an extremely significant tendency of mortality increase with urine volume decrease (0.0000). The risk of death was also significantly higher in patients on mechanic ventilation, using vasopressors, or with infection.

Table 4  Univariate analysis: risk factors for death

Urine volume	n	Death (%)	OR	CI 95%	p
≤ 50	143	83	3.56	2.1–6.1	<0.00001
50 < UV < 400	174	78	2.54	1.6–4.0	<0.00001
400 ≤ UV ≤ 1000	241	59	1.00	—	—
1000 < UV ≤ 2000	243	38	0.46	0.3–0.7	<0.00001
>2000	63	29	0.28	0.2–0.5	<0.00001
Total	864	58	—	—	—
MV	493	83	13.56	9.6–19.2	<0.0000
VAD	461	85	13.76	9.7–19.5	<0.0000
Hyperkalemia	368	61	1.24	0.93–1.65	0.146
Hyponatremia	258	64	1.36	0.99–1.86	0.055
Acidosis	377	61	1.23	0.92–1.64	0.157
Infection	683	63	2.62	1.84–3.73	<0.000001
<65 yrs.	504	55	—	—	—
≥65 yrs.	330	63	1.42	1.05–1.91	0.018

Abbreviations: OR = odds ratio, CI = confidence interval, UV = urine volume mL/24 h, MV = mechanical ventilation, VAD = vasoactive drug.

In the final logistic regression model, the following factors were identified as risk factors for death: anuria, oliguria, uremia, mechanical ventilation, and use of vasopressor, as shown in Table 5. The multivariate analysis confirmed the inverse relationship between urine volume and mortality. Anuric and oliguric patients had an increase of 95% and 76% in the risk for death comparing to controls, respectively (p < 0.005). Patients who urine volume >1000 ml and ≤2000 mL (group 4) and urine volume >2000 ml (group 5) decreased their risk for death in 50% and 70% (p = 0.004, p = 0.001, respectively). When urine volume was assessed as a continuous variable, increments of 100 ml were associated to a 7.2% decrease in the mortality risk (< 0.001). If patients younger than 18 years old were excluded, the same results were obtained, except that hypervolemia was now identified as an independent risk factor for mortality (OR 1.623; CI 1.0–2.6; p = 0.045; see Table 6).

Table 5  Multivariate (logistic regression): independent riskfactors for death

	Adjusted OR for death	CI 95%	p
UV ≤ 50	1.95	1.06–3.60	0.032
50 < UV < 400	1.76	1.02–3.03	0.042
400 ≤ UV ≤ 1000	1.00	—	—
1000 < UV ≤ 2000	0.50	0.31–0.80	0.004
UV > 2000	0.30	0.14–0.62	<0.001
Uremia	1.64	1.12–2.39	0.010
MV	4.93	3.18–7.65	<0.001
VAD	4.60	2.96–7.15	<0.001

OR = odds ratio, CI = confidence interval, MV = mechanical ventilation, VAD = vasoactive drug, UV = urine volume (mL/24 h).

Table 6  Multivariate (logistic regression): independent risk factors for death in patients > 18 years old

	Adjusted OR for death	CI 95%	p
UV ≤ 50	1.98	1.06–3.67	0.03
50 < UV < 400	1.97	1.1–3.4	0.017
400 ≤ UV ≤ 1000	0.50	—	—
1000 < UV ≤ 2000	0.52	0.33–0.85	0.08
UV > 2000	0.28	0.14–0.61	0.001
Hypervolemia	1.62	1.0–2.6	0.045
Uremia	1.67	1.14–2.45	0.010
MV	4.96	3.18–7.7	0.000
VAD	3.9	2.5–6.13	0.000

OR = odds ratio, CI = confidence interval, MV = mechanical ventilation, VAD = vasoactive drug, UV = urine volume (mL/24 h).

DISCUSSION

The present study was performed in a large retrospective AKI patient's cohort, and aimed to assess the influence of urine volume in AKI mortality. RIFLE criteria were not used because the study was developed before its publication.^[30,31]

The group of AKI patients studied comprised of critically ill individuals. They were old, almost 50% required dialysis, more than 50% needed mechanical ventilation or vasopressor drugs, and 79% had infection. Most of them were men, and acute tubular necrosis was the most usual cause for AKI. These characteristics are consistent with previous publications on AKI.^{[1,5,6,10,13–15,31,32]}

Although the majority of our AKI patients were nonoliguric (63.6%), mortality rate was very high (58%). Univariate analysis showed a progressive decrease of mortality while urine volume increased.

One limitation of the study is that we do not have patients' severity of disease scores, such as APACHE or ATN-ISS; thus, it is possible that other factors besides urine volume have influenced mortality. However, when logistic regression was done to evaluate the influence of urine volume as an independent variable in mortality rate, a clear inverse and linear correlation between urine volume and risk of death was found. Nonoliguric patients who had urine volume above 1000 mL/24 h and specially above 2000 mL/24 h had a lower risk of death, as compared to patients with urine volume > 400 ml but lower than 1000 ml, although all of them are considered nonoliguric by the classical definition. In fact, the pool of patients defined as nonoliguric comprise different kinds of AKI patients, using diuretics or not, volume overloaded or volume depleted. When urine volume was assessed in a second logistic regression, as a continuous variable, even small increments of 100 ml were significantly associated to lower mortality. In agreement, several recent articles have showed urine volume as an independent factor of mortality.^[9,13,33]

Although the patients' fluid balance was not recorded in this study, a smaller urine volume might be associated with positive fluid balance. Recent studies in adult and pediatric critically ill patients showed the negative effects of positive fluid balance on prognosis.^[33–37] In a secondary analysis of the SOAP (Sepsis Occurrence in Acutely Ill Patients), Payen et al. studied 3,147 critically ill patients, and 36% of them (1,120) developed AKI. Mortality at 60 days was higher for those with AKI (36% vs. 16%; p < 0.01). Average daily fluid balance was significantly higher in non-survivors than in survivors (1,000 ml vs. 150 ml; p < 0.001). On multivariate analysis, a positive fluid balance was an independent factor for 60-day mortality (OR 1.21, 95% confidence interval 1.13 to 1.28; p < 0.001).^[34] Considering the extreme severity and complexity of current AKI patients, frequently receiving elevated volume infusions due to resuscitation protocols, parenteral and enteral nutrition, antibiotics, and continuous sedation, urine volume <1000 ml/24 h is probably insufficient for maintaining an adequate fluid balance. Therefore, it is possible that a more adequate fluid balance was one of the factors associated with the better prognosis found in the group of patients with urine volume over 1000 mL/24 hours. Further studies are certainly necessary for explore this possibility.

In conclusion, urine volume was identified as an independent strong predictor of mortality for critically ill AKI patients. Urine volumes over 1000 mL/24 and 100 ml increments in urine volume were associated with decreased mortality. These results suggest that the urine volume usually used in the classical definition of oliguria is probably inadequate for current AKI patients, and that urine volume should be analyzed as a continuous variable instead of a discrete one.

ACKNOWLEDGMENTS

The authors report no conflicts of interest. The authors alone are fully responsible for the content and writing of the paper.