Acute renal failure (ARF) is a life threatening illness with a high mortality despite advances in supportive care 1234567891011121314. Although there is a strong and direct relation between multiorgan failure and in-hospital mortality, severity of illness does not explain the variation in outcomes among patients with acute renal failure 5910. There is an increased cost in terms of patient prognosis, financial and clinical management. In a study, 30% of patients did not recover completely their renal function 15 and, in other studies, progression to chronic renal failure in many patients is suggested 916. In regard to these concerns, prompt recognition and early consultation with a nephrologist have been postulated to improve the outcome of patients with acute renal failure 7111217. Nevertheless, the published evidence is scarce.

The absence of a consensus definition of ARF has made research about ARF difficult 61819. Recently, the Acute Dialysis Quality Initiative (ADQI) published a uniformed definition called the RIFLE (Risk, Injury, Failure, Loss, End Stage) criteria 20. The RIFLE criteria has been adopted by the Acute Kidney Injury Network (AKIN) 6.

Diagnosis of ARF based upon changes in serum creatinine may be delayed due to the fact that, in the non-steady-state conditions of ARF, as GFR falls creatinine secretion is increased 11. Large changes in GFR are initially manifested as small quantitative changes in serum creatinine in the first 24–48 hours after renal injury 11. After these one or two days, the degree of serum creatinine changes will reflect the change in GFR. Finally, the serum creatinine is stabilized, and that takes about 7 days 1120. Therefore, it is almost impossible to exactly determine the onset of the ARF; and also calculating the exact time lapsed until the nephrology consultation. However, measuring serum creatinine level is a practical approach for discovering short-term alteration in renal function, despite its limitations, because it is readily used in clinical practice and it is specific for renal function 2021. In patients with stable renal function, those levels are constant, with a daily variability of 8% 2223, so increasing levels might suggest the first stage of ARF.

We conducted this study to identify and quantify a correlation between acute serum creatinine changes in ARF, measuring them at the time of nephrology consultation, and mortality and recovery of renal function. We hypothesized that, at the same levels of severity of renal and comorbid illnesses, the higher the acute serum creatinine reached with respect to its baseline, before the nephrologist first began to treat the patient, the worse the outcome will be.

Six hundred and forty-six patients were evaluated, among them one hundred and eighty-eight died in-hospital. The demographics and baseline clinical characteristics are shown in Table 1. Subjects in the ≥ 101% increment of creatinine group were more likely to be female (p = 0.01), had impaired prior food intake (p = 0.02), and required more renal replacement therapy (p < 0.001). The ≥ 101% increment of creatinine group was associated with high prevalence of community-acquired ARF (p = 0.02), and worse Karnofsky scoring (p = 0.015). The group with ≥ 101% increment of creatinine was also associated with higher Liano scoring at the time of nephrology consultation (p < 0.001), including higher prevalence of hypotension (p < 0.001), oliguria (p < 0.001), jaundice (p = 0.008), coma (p = 0.027), and were on mechanical ventilation (p = 0.002). RIFLE classes reached during the ARF were more severe in subjects in the ≥ 101% increment of creatinine group (p < 0.001). There was a higher prevalence of diabetes (p = 0.037) in this group.

The ≥ 101% increment group was associated with significant increased total inhospital mortality (35.6% vs. 22.6%, P < 0.001), with an adjusted odds ratio of 1.81 (95% CI: 1.08–3.03) (Table 2). Moreover, those patients in the ≥ 101% increment group who required continuous renal replacement therapy presented a higher increase of inhospital mortality (62.7% vs 46.4%, P = 0.048), with an adjusted odds ratio of 2.66 (95% CI: 1.00–7.21) (Table 3). We noticed that patients in the ≥ 101% increment group had a higher mean serum creatinine level in comparison to their baseline level (114.72% vs. 37.96%) at hospital discharge (Table 4). In the ≥ 101% increment group, creatinine was an adjusted 48.92% (95% CI: 13.05–84.79) higher than the < 101% increment group.

The overall mortality rate was 337 deaths (52.2%) in 6 months. According to the Kaplan-Meier plot, the ≥ 101% increment group also had a statistically significant higher mortality rate in this period as compared with the < 101% increment group (Log Rank test = 0.012) (Figure 1). During the first 60 days of follow-up 175 deaths (27.1%) occurred.

This longitudinal study shows that a relative increase in baseline serum creatinine by ≥ 101% compared with the levels measured at the time of nephrology consultation is an independent predictor of mortality. Such an increase was also associated with worse functional renal recovery irrespective of the degree of ARF severity or presence of comorbid illnesses.

Due to the heterogeneous nature of ARF 11, it was important to adjust for the severity of illness with Liano score, and for the acute kidney injury with the RIFLE criteria. We have chosen the Liano score because of its ability to discriminate mortality from survival and its ability to calibrate the observed mortality rate with the expected mortality in ARF 111427. The RIFLE consensus criteria gives a standard definition and a level of classification of severity in ARF 20, a request often made by the experts 61819. Working in this manner may allow continuing further investigation in ARF, and is one of the main strengths of this study. Patients did not do worse because they had more severe ARF or comorbid illnesses. Instead, patients presenting the same RIFLE criteria and Liano scores, but with higher serum creatinine increments prior to the nephrology consultation, showed higher mortality rates.

Using multivariate logistic regression analysis, we found that an increment in creatinine with respect to baseline level of 101% at the time of nephrology consultation started is a factor independently and strongly predictive of death in this cohort. Doubling the basal creatinine level presented an odds ratio of 1.81 for mortality among general population. This increased to a threefold (OR = 2.66) mortality in patients who needed continuous renal replacement therapy. These results might suggest that the increase of serum creatinine level at the moment when the ARF began to be treated by an expert was related to the outcome. However, we should concede the possibility that the more abrupt increase in serum creatinine was, the worse the prognosis. Because of the observational nature of the present study, our findings need to be interpreted with caution. We cannot determine with certainty that finding a higher increase in the percentage of creatinine was a consequence of a delayed nephrology consultation. It could be that among patients who received the nephrology intervention at the same time, some of them presented a higher increase in the percentage of creatinine because of an abrupt deterioration of renal function. Nevertheless, after adjusting for severity of ARF and comorbid illnesses, patients who had had higher increments in percentage of creatinine prior to being treated by a nephrologist did present higher mortality rates. We might stand for prompt expert consultation because these ARF patients at risk would benefit from an intensive and specialized treatment, and further kidney damage could be minimized.

The cohort included a heterogeneous distribution of patients, in terms of unit, nature, or source of admission; but with an important number of oncology patients (35.4%) since our institution is a well-known center in Spain for the treatment of oncologic diseases. For this reason, we adjusted adding the oncology factor to the logistic regression in an extra analysis. However, because of its longitudinal design, we could obviate the Berkson bias (to take a falsely typical population of patients since the group of people being studied has no form of control over whether to participate) 28. Also, it could be argued that this was a single center study, with its limits to apply to other hospitals. We have tried to minimize any bias using objetive data collected by a single investigator following strict criteria, and reviewing the data. Urine output data were not available; therefore it was not possible to estimate the RIFLE criteria according to this value. We started to collect the data before the RIFLE criteria was formulated, and most of the ward patients had no urine output measured on a six hourly basis. We assume that we might underestimate some cases according to the RIFLE criteria; but only 21 and 62 patients from groups R and I respectively had oliguria. Although we cannot rule out the possibility of residual confounding, it is unlikely that it can fully explain the observed strong associations. Moreover, when we adjusted for RIFLE criteria, an increase of our estimations occurred in the multivariate model; so misclassification of this covariate would most likely bias the odds ratios toward the null value.

We calculated the MDRD equation to obtain a baseline serum creatinine in patients with no previous history of renal failure because a true baseline is often unknown. We solved the equation assuming a GFR of 75 ml/minute/1,73 m², which has been reported to estimate a lower limit of a normal GFR 41320. Because GFR was assumed to be 75 ml/min, using MDRD for patients with unknown pre-morbid renal dysfunction may underestimate severity of ARF according to the RIFLE criteria 2930. This lack of reliability is one of the controversial issues regarding the RIFLE criteria, but it is still unresolved 31. Although the MDRD formula may be less accurate at normal GFR, resulting in overdiagnosis of ARF, it should not be a disadvantage in a clinical setting where ARF is taken seriously, and the MDRD also has the advantage of not requiring body weight data 32 that could be wrongly estimated in fluid retention settings.

We chose to calculate the percentage change in creatinine respect to its baseline because serum creatinine is modulated by muscle bulk. Therefore, smaller changes as 0.3 or 0.5 mg/dL will have distinct significance for patients with different genders and/or ages, whose normal baseline levels are different 18. For this reason, we found that a relative increase in creatinine was more accurate than an absolute increase one.

The higher presence of chronic renal failure in the < 101% increment group might be explained because concern about it could have induced earlier nephrology consultation, even if the patient did not present signs of ARF in that time. However, we must admit that 74 chronic renal failure patients (48.68%) might have presented an unstable serum creatinine value at hospital admission, thus underestimating the severity of ARF for them. Chronic renal failure was included in this study to represent the common clinical practice, and thereafter it was adjusted for in the analysis because of expecting different outcomes than the general population.

On the other hand, the higher proportion of cases of ARF developed in the community in the ≥ 101% increase group might suggest that nephrology consultation may have been delayed because of time spent before hospital admittance, giving serum creatinine an extra time to reach a higher value.

Our results about mortality in the Renal Replacement Therapy sub-group are consistent with previously reported 93334, and it is of interest the higher mortality rate in patients in the ≥ 101% increase group. What we may learn here is that this group of patients is most likely to benefit from a more intensive treatment.

The degree of renal injury is likely to affect renal recovery. Patients where consultation was started with ≥ 101% creatinine increase had significant higher creatinine levels at discharge compared to their baseline levels, and many of them would be expected to develop chronic renal malfunction. Adjustment for confounding factors in the analysis was done, too.

We analyzed 6-months mortality separately, since the time when patients are discharged from hospital might vary between admission units. Although it could be argued that all-cause mortality at six months is non specific, we found that a high number of deaths (175) occurred in the first 60 days. These findings agree with previous reports and recommendations about optimal follow-up time in patients with established ARF 202635.

It is of interest to note that 306 patients from the original population of 1008 would not have been diagnosed of ARF had the RIFLE criteria been used then. They did not present oliguria nor enough creatinine increment, but were diagnosed of ARF according to another criteria at that time. Some experts might prefer to use more inclusive criteria because of concern about under-recognizing ARF 6.

Finally, had our patients presented other mortality rates, if they would have been treated by a nephrologist before reaching serum creatinine levels more than twice their baseline levels, is a question beyond this study and needs further research.

At the same degree of severity of kidney injury and comorbid illnesses, while controlling for other potential confounders, ARF patients with a relative increase of ≥ 101% in the baseline creatinine at the time of the nephrology consultation was an independent predictor of mortality and worse prognosis. The calculation of relative serum creatinine increase could be used as a precise and useful tool in the daily general medical practice, and may be an effective tool to identify those patients who would benefit from an intensive therapy. However, further investigation about timing in nephrology consultation is needed to provide robust knowledge about treatment and prognosis in ARF.

1. Acute renal failure is a condition with a high risk of mortality.

2. At the same degree of acute renal failure and comorbid illnesses severity, and receiving similar treatment, patients who had presented an increment of creatinine ≥ 101% with respect to baseline at the time of nephrology consultation showed increased mortality rates.

3. Those patients were discharged from hospital with more deteriorated renal function.

4. The calculation of relative serum creatinine increase might be used as a precise and useful tool in the general medical practice. Predicting the outcome helps to identify those patients who would benefit from a intensive therapy, and also to reassure patients and their relatives.

ARF: Acute Renal Failure.

RIFLE: Risk, Injury, Failure, Loss, End Stage Classification for acute kidney injury.

ADQI: Acute Dialysis Quality Initiative.

GFR: Glomerular Filtration Rate.

AKIN: Acute Kidney Injury Network.

MDRD: Modification of Diet in Renal Disease

OR: Odds Ratio.

CI: Confidence Intervals.

The author(s) declare that they have no competing interests.

All the authors conceived the research question. JRPV and MBR wrote the first draft and conducted the statistical analysis with advice from JdI, PM and JL. PM and JL coordinated the research. JL contributed to data collection. Subsequent revisions were carried by JRPV and MBR with input from all authors. The final version was approved by all authors.