ACUTE RENAL FAILURE AND ERRATUM

	To the Editor:

In their study comparing initial versus delayed acute renal failure (ARF) in the intensive care unit, Guérin and colleagues observe that mortality increased in patients included after Day 3 (group B) or Day 7 (group C), when compared with patients included before Day 3 (group A) (71%, 82%, and 61%, respectively, p < 0.001) (1). However, late ARF did not predict death, a result not in accord with our study (2) or with other researchers' studies (3).

To support this discrepancy, Guérin and colleagues allege that the rates of occurrence of oliguria and the need for hemodialysis of our two groups of initial and delayed ARF were not similar. Unfortunately, they drew their conclusion from a reappraisal of our data, calculating on wrong numbers. Among 360 ARF patients, 143 were included after Day 3. Initial ARF was thus observed in 217 patients, as written in the abstract and in the text, and not in 127, as printed by mistake in Table 2. We confirm that in our series, the distribution of these two variables was not different in the two groups of initial versus delayed ARF. Thus, it invalidates the arguments developed by Guérin and coworkers in their discussion on this topic.

The major factor that may explain the differences in predictive value of late occurrence of ARF relies on definitions. Our patients presented with sCr  310 µmol/L, or BUN  36 mmol/L, or a doubling of basal value of sCr (when > 150 µmol/L). In Guérin's study, apart from sCr, the definition included oliguria or requirement for dialysis. However, hemodialysis or hemofiltration may be used to treat overhydration and has been used to treat sepsis or multiple organs dysfunction, which does not necessarily reflect severe ARF in all patients. Oliguria was the most frequent criteria (67%) in Guérin's series and might reflect other organ dysfunctions rather than primary ARF. Indeed, sCr at diagnosis in groups B and C was 277 and 238 µmol/L, respectively. These patients would not have been included in our study. Moreover, the definition of organ dysfunction based on OSF in our study was not comparable with the ODIN model used by Guérin and colleagues.

Incidentally, Table 2 shows a surprising low average number of organ dysfunctions in the overall population (2.5), whereas it was 3.0, 3.3, and 3.5 in groups A, B and C, respectively. This inconsistency also appears for all the percentages at the bottom of Table 2. Furthermore, the comparisons of mortality rates between the sample and the whole population were inappropriate (because of nonindependence), and moreover, these rates were surprisingly low for group B (22% versus 19%) and C (13% versus 13%) and inconsistent with the mortality rates observed in the whole population. This discrepancy remains unexplained.

In conclusion, Guérin and coworkers do not provide convincing evidence that late occurrence of ARF during the ICU stay is not associated with a poor prognosis.

Hôpital Antoine Béclère, Clamart, France

Philippe Loirat

Centre Médico Chirugical Foch, Suresnes, France

Paul Landais

Université Paris 5, Paris, France

1. Guérin C, Girard R, Selli JM, Perdrix JP, Ayzac L. for the Rhône-Alpes-Area Group on Acute Renal Failure. Initial versus delayed acute renal failure in the intensive care unit: a multicenter prospective epidemiological study. Am J Respir Crit Care Med 2000; 161: 872-879 [Abstract/Free Full Text].

2. Brivet FG, Kleinknecht DJ, Loirat P, Landais P, The French Study Group on Acute Renal Failure. Acute renal failure in intensive care unitscauses, outcome, and prognostic factors of hospital mortality: a prospective multicenter study. Crit Care Med 1996;24:192-198.

3. Druml W, Lax F, Grimm G, Schneeweiss B, Lenz K, Lagguer AN. Acute renal failure in the elderly 1975-1990. Clin Nephrol 1994; 41: 342-349 [Medline].

4. Levy EM, Viscoli CM, Horwiz RI. The effect of acute renal failure on mortality: a cohort analysis. JAMA 1996; 275: 1489-1494 [Abstract].

5. Liano F, Pascual J, The Madrid Acute Renal Failure Study Group. Epidemiology of acute renal failure: a prospective, multicenter, community-based study. Kidney Int 1996;50:811-818.

	From the Authors:

First of all, we would like to thank Dr, Brivet and his colleagues for their comments about our article (1) giving us the opportunity to improve it further.

Dr. Brivet and his colleagues provide the explanation of the discrepancy of the statistical results of oliguria and dialyzed patients between those shown in their Table 2 (2) and our own calculations with the same data. In Table 2 of their paper (2), Dr. Brivet and coworkers state that the number of patients with acute renal failure was 127, whereas elsewhere in their paper they state that the number was 217. We based our calculations on 127 patients, which led erroneously to a significant difference for the percentages of oliguria and dialyzed patients between initial and delayed acute renal failure. Thus, part of our discussion (p. 877, right column, lines 7-9 and lines 15-18), based on our calculations, is false.

We agree that the definition of ARF differs between the two studies (1, 2), and it generally true that there are as many definitions of ARF as there are studies in this field. We used the definition of ARF given by Fagon and colleagues (3) for internal consistency with the other organ failure definitions. Whether this difference in the definition for ARF explains the difference observed for the impact of delayed ARF on patient outcome between the two studies remains to be established.

Dr. Brivet and his colleagues detected some numerical errors in our article, and we would like to apologize for them and to provide corrections. In Table 2, for the whole population, at the first episode of ARF, the rate of cardiovascular dysfunction is 67%, not 58% as originally given; the rate of respiratory dysfunction is 79%, not 78%; the rate of hepatic dysfunction is 16%, not 12%; the rate of hematological dysfunction is 19%, not 18%; the rate of neurological dysfunction is 33%, not 31%; the rate of mechanical ventilation is 77%, not 74%; the mean ± SD number of organ dysfunction is 3.1 ± 1.2, not 2.5 ± 1.3; and the rate of infection is 41%, not 37%. The p values at the time of ICU admission are < 0.05, not < 0.01, between group A versus group B for respiratory dysfunction; < 0.05, not NS, between group B versus group C for hematological dysfunction; < 0.05, not NS, between group A versus group B for neurological dysfunction; NS, not < 0.05, between group A versus group C for neurological dysfunction; < 0.001, not < 0.01, between group A versus group C for mechanical ventilation, and NS, not < 0.05, between group A versus group B for infection. The p values for hepatic dysfunction at the first episode of ARF are NS, not < 0.001, between group A versus group B and < 0.0001, not NS, between group A versus group C. The other values in Table 2 have been checked and are correct. As a result of these corrections, a new Table 2 is now provided above.

In the text (p. 875, in the section "In-hospital mortality," line 23), the sentence should read as follows: "Between sample and population, the mortality rates were 64% versus 61%, 68% versus 71% and 83% versus 82% (p = NS) in Groups A, B, and C, respectively."

In Table 3, for the severity of the first episode of acute renal failure, the correct rates for hemodialysis (last line of Table 3) are 52% instead of 54% in the whole population, 49% instead of 51% in Group A, 57% instead of 58% in Group B, and 63% instead of 64% in Group C. It must be pointed out that the rates of hemodialysis in Table 3 correspond to the first episode of ARF, whereas the numbers of hemodialysis in Table 5 are for hemodialysis performed at any time during the ICU stay.

We would also like taking advantage of this reply to clarify an apparent discrepancy regarding the numbers of patients who did not receive hemodialysis. On p. 875, left column, lines 21-23, the numbers correspond to the whole population, whereas in the DISCUSSION section (p. 878, right column, lines 6-7) the numbers are for the 510 randomly selected patients.

Finally, we disagree with the conclusion of Brivet and his colleagues about the relationship between delayed ARF and patient outcome in our study. In our study, although the patients with delayed ARF had a higher fatality rate, the results of the logistic regression analysis suggest that this was not due to the delayed occurrence of ARF per se, but was more likely to be due to other factors.

Hôpital de la Croix-Rousse, Lyon, France

Raphaele Girard, and Louis Ayzac

Centre Hospitalier Lyon Sud, Pierre Bénite, France

1. Guérin C, Girard R, Selli J-M, Perdrix J-P, Ayzac L. for the Rhône-Alpes-Area Study Group on Acute Renal Failure. Initial versus delayed acute renal failure in the intensive care unit: a multicenter prospective epidemiological study. Am J Respir Crit Care Med 2000; 161: 872-879 .

2. Brivet FG, Kleinknecht DJ, Loirat P, Landais P. the French Study Group on Acute Renal Failure. Acute renal failure in intensive care units causes, outcome, and prognostic factors of hospital mortality: a prospective multicenter study. Crit Care Med 1996; 24: 192-198 [Medline].

3. Fagon JY, Chastre J, Novara A, Medioni P, Gilbert C. Characterization of intensive care unit patients using a model based on the presence or absence of organ dysfunctions and/or infection: the ODIN model. Intensive Care Med 1993; 19: 137-144 [Medline].