## Prognosis

Kidney Function Restoration Program

Get Instant Access

HISTORICAL PERSPECTIVE OF MEDICAL PROGNOSIS APPLIED IN ACUTE RENAL FAILURE

Criteria

Derivation

Applications

Drawbacks

Classical

Doctor's experience

Individual prognosis

Easy

Doctor's inexperience

Unmeasurable

Univariate statistical analysis

Risk stratification

Easy

Only one determinant of prognosis is considered

Present

Multivariate statistical analysis

Risk stratification

Measurable

Complexity (variable, depending on model)

Computing facilities

Individual prognosis?

Theoretically, "all" factors influencing outcome

are considered

Future

Multivariate analysis

Risk stratification

Measurable

Ideally, none

Computing facilities

Individual prognosis

"All" factors considered

Patient's quality of life evaluation

Functional prediction

### FIGURE 8-21

Estimating prognosis. The criteria for estimating prognosis in acute renal failure can be classified into four periods. The Classical or heuristic way is similar to that used since the Hippocratic aphorisms. The Traditional one based on simple statistical procedures, is not useful for individual prognosis. The Present form is more or less complex, depending on what method is used, and it is possible, thanks to computing facilities and the development of multivariable analysis. Theoretically, few of these methods can give an individual prognosis [19]. They have not been used for triage. The next step will need a great deal of work to design and implement adequate tools to stratify risks and individual prognosis. In addition, the estimate of residual renal function and survivors' quality of life, mainly for older people, are future challenges.

FIGURE 8-22

Ideally, prognosis should be established as the problem, the episode of acute renal failure (ARF), starts. Correct prognostic estimation gives the real outcome for a patient or group of patients as precisely as possible. In this ideal scenario, this fact is illustrated by giving the same surface area for the concepts of outcome and prognosis.

Mortality trends in acute renal failure (ARF). This figure shows the evolution of mortality during a 40-year period, starting in 1951. The graphic was elaborated after reviewing the outcome of 32,996 ARF patients reported in 258 published papers. As can be appreciated, mortality rate increases slowly but constantly during this follow-up, despite theoretically better availability of therapeutic armamentarium (mainly antibiotics and vasoactive drugs), deeper knowledge of dialysis techniques, and wider access to intensive care facilities. This improvement in supporting measures allows the physician to keep alive, for longer periods of time patients who otherwise would have died. A complementary explanation could be that the patients treated now are usually older, sicker, and more likely to be treated more aggressively. (From Kierdorf et al. [20]; with permission.)

Apache system

APACHE II APACHE III

ICU methods

SAPS

SAPS I

SAPS II

MPM I

MPM II

Prognostic systems used in ARF

MODS

Specific

ARF methods

SOFA

Liano

Rasmussen

Schaefer

Brivet

### FIGURE 8-24

Ways of estimating prognosis in acute renal failure (ARF). This can be done using either general intensive care unit (ICU) score systems or methods developed specifically for ARF patients. ICU systems include Acute Physiological and Chronic Health Evaluation (APACHE) [21,22], Simplified Physiologic Score (SAPS)[23,24], Mortality Prediction Model (MPM) [25,26], and Organ System Failure scores (OSF) [27]. Multiple Organ Dysfunction Score (MODS) [28] and

Sepsis-Related Organ Failure Assessment Score (SOFA) [29] are those that seem most suitable for this purpose. APACHE II used to be most used. Other systems (white boxes) have been used in ARF. On the other hand, at least 17 specific ARF prognostic methods have been developed [20,30]. The figure shows only those that have been used after their publication [31], plus one recently published system which is not yet in general use [2].

FIGURE 8-25

Comparison of prognostic methods for acute renal failure (ARF) by ROC curve analysis [31]. A method is better when its ROC-curve moves to the upper left square determined by the sensitivity and the reciprocal of the specificity. A, ROC curves of seven prognostic methods usually employed in the ICU setting. The best curve comes from the APACHE III method, which has an area under the ROC curve of 0.74 ± 0.04 (SE). B, Four ROC curves corresponding to prognostic methods specifically developed for ARF patients are depicted. The best curve in this panel comes from the Liano method for ARF prognosis. Its area under the curve is 0.78 ± 0.03 (SE). APACHE—Acute Physiology and Chronic Health Evaluation, (II second version [21]; III third version [22]); SAPS—Simplified Acute Physiology Score [23]; SAPS-R— SAPS-reduced [33]; SAPS-E—SAPS-Extended [32]; SS—Sickness Score [33]; MPM—Mortality Prediction Model [25]; ROC curve—Receiving Operating Characteristic curve; SE—Standard Error. (From Douma [31]; with permission.)

ACUTE RENAL FAILURE: VARIABLES STUDIED WITH UNIVARIATE ANALYSIS

8 patients to chronic hemodialysis

8 patients to chronic hemodialysis n—I—I—I—r

35 40

45 50 55 60

35 40

45 50 55 60

Days of ARF evolution

Age

Hypotension

Jaundice

Catabolism

Sepsis

Hemolysis

Burns

Hepatic disease

Trauma

Kind of surgery

NSAIDs

Hyperkalemia

BUN increments

Need for dialysis

Coma

Assisted respiration

Oliguria

Site of war injuries

Obstetric origin

Disseminated intravascular coagulopathy

Malignancies

Pancreatitis

Cardiovascular disease

Antibiotics

X-ray contrast agents

Timing of treatment

Acidosis

Individual factors that have been associated with acute renal failure (ARF) outcome. Most of these innumerable variables have been related to an adverse outcome, whereas few (nephrotoxicity as a cause of ARF and early treatment) have been associated with more favorable prognosis. For a deep review of variables studied with univariate statistical analysis [34, 35]. NSAID—nonsteroidal antiinflammatory drugs; BUN—blood urea nitrogen.

### FIGURE 8-27

Duration and resolution of acute renal failure (ARF). Most of the episodes of ARF resolved in the first month of evolution. Mean duration of ARF was 14 days. Seventy-eight percent of the patients with ARF who died did so within 2 weeks after the renal insult. Similarly, 60% of survivors had recovered renal function at that time. After 30 days, 90% of the patients had had a final resolution of the ARF episode, one way or the other. Patients who finally lost renal function and needed to be included in a chronic periodic dialysis program usually had severe forms of glomerulonephritis, vasculitis, or systemic disease. (From Liano et al. [1]; with permission.)

FIGURE 8-28

Persistent hypotension

P<0.001

Yes No

Jaundice

P<0.001

10080

tor 40 20 0

10080

tor 40 M

20 0

Assisted repiration

P<0.001

Yes No

Oliguria

Yes No

Yes No

FIGURE 8-28

Precipitating condition of acute renal failure (ARF). The initial clinical condition observed in ARF patients is shown. Oliguria: urine output of less than 400 mL per day; hypotension: systolic blood pressure lower than 100 mm Hg for at least 10 hours per day independent of the use of vasoactive drugs; jaundice: serum bilirubin level higher than 2 mg/dL; coma: Glasgow coma score of 5 or less. The presence of these factors is associated with poorer outcome (see Fig. 8-29). (Data from Liano et al. [1].)

### FIGURE 8-29

Mortality associated with the presence or absence of oliguria, persistent hypotension, assisted respiration and jaundice (as defined in Fig. 8-28). The presence of an unfavorable factor was significantly associated with higher mortality. (Data from Liano et al. [1].)

FIGURE 8-30

Consciousness level and mortality. Coma patients had a Glasgow coma score of 5 or lower. Sedation refers to the use of this kind of treatment, primarily in patients with assisted respiration. Both situations are associated with significantly higher mortality (P<0.001) than that observed in either patients with a normal consciousness level or the total population. (Data from Liano et al. [1].)

### FIGURE 8-31

Outcome of acute renal failure (ARF). Two groups of factors play a role on ARF outcome. The first includes factors that affect the patient: 1) previous health condition; 2) initial disease—usually, the direct or indirect (eg, treatments) cause of kidney failure; 3) the kind and severity of kidney injury. While 1 is a conditioning element, 2 and 3 trigger the second group of factors: the response of the patient to the insult. If this response includes a systemic inflammatory response syndrome (SIRS) like that usually seen in intensive care patients (eg, sepsis, pancreatitis, burns), a multiple organ dysfunction syndrome (MODS) frequently appears and consequently outcome is associated with a higher fatality rate (thick line). On the contrary, if SIRS does not develop and isolated ARF predominates, death (thin line, right) is less frequent than survival (thick line).

### FIGURE 8-32

INDIVIDUAL SEVERITY INDEX Individual severity index (ISI). The ISI was published in its second version in 1993 [36]. The ISI estimates the probability of death. Nephrotoxic indicates an ARF of that origin; the other variables have been defined in preceding figures. The numbers preceding these keys denote the contribution of each one to the prognosis and are the factor for multiplying the clinical variables; 0.210 is the equation constant. Each clinical variable takes a value of 1 or 0, depending, respectively, on its presence or absence (with the exception of the age, which takes the value of the patient's decade). The parameters are recorded when the nephrologist sees the patient the first time. Calculation is easy: only a card with the equation values, a pen, and paper are necessary. A real example is given.

ISI=0.032 (age-decade) - 0.086 (male) - 0.109 (nephrotoxic) + 0.109 (oliguria) + 0.116 (hypotension) + 0.122 (jaundice) + 0.150 (coma) — 0.154 (consciousness) + 0.182 (assisted respiration) + 0.210

### Case example

A 55-year-old man was seen because of oliguria following pancreatic surgery. At that moment he was hypotensive and connected to a respirator, and jaundice was evident. He was diagnosed with acute tubular necrosis. His ISI was calculated as follows:

ISI=0.032(6) — 0.086 + 0.109 + 0.116 + 0.122 + 0.182 + 0.210 = 0.845

ATN

Acute GN

6 -

-No recovery-

- 11

24 -

No recovery

31

— Partial recovery —

- 32

47

35

Partial recovery

24

63

— Total recovery —

_ 57

41

Total recovery

" 29

1 yr

Acute TIN

5 yr

1 yr

HUS/ACN

5 yr

No recovery

- 8

25 _

Partial recovery

25

63

No recovery

75 _

91

Total recovery

67

27

Partial recovery

- 9

1 yr

5 yr

1 yr

Outcome of acute renal failure (ARF). Long-term outcome of ARF has been studied only in some series of intrinsic or parenchymatous ARF. The figure shows the different long-term prognoses for intrinsic ARF of various causes. Left, The percentages of recovery rate of renal function 1 year after the acute episode of renal failure. Right, The situation of renal function 5 years after the ARF episode. Acute tubulointerstitial nephritis (TIN) carries the better prognosis: the vast majority of patients had recovered renal function after 1 and 5 years. Two thirds of the patients with acute tubule necrosis (ATN) recovered normal renal function, 31% showed partial recovery, and 6% experienced no functional recovery. Some patients with ATN lost renal function over the years. Patients with ARF due to glomerular lesions have a poorer prognosis; 24% at 1 year and 47% at 5 years show terminal renal failure. The poorest evolution is observed with severe forms of acute cortical necrosis or hemolytic-uremic syndrome. GN—glomerulonephritis; HUS— hemolytic-uremic syndrome; ACN—acute cortical necrosis. (Data from Bonomini et al. [37].)

174 \

113

L 50 \

\ Alive J

Alive

Alive

225

\ 143 y

53

< 65 yr

65-79 yr

> 80 yr

(n = 399)

(n = 256)

Age as a prognostic factor in acute renal failure (ARF). There is a tendency to treat elders with ARF less aggressively because of the presumed worse outcomes; however, prognosis may be similar to that found in the younger population. In the multicenter prospective longitudinal study in Madrid, relative risk for mortality in patients older than 80 years was not significantly different (1.09 as compared with 1 for the group younger than 65 years). Age probably is not a poor prognostic sign, and outcome seems to be within acceptable limits for elderly patients with ARF. Dialysis should not be withheld from patients purely because of their age.

VARIABLES ASSOCIATED WITH PROGNOSIS: MULTIVARIATE ANALYSIS (16 STUDIES)

PROGNOSIS IN ACUTE RENAL FAILURE

 Assisted respiration 11 Hypotension or inotropic support Age 10 8 Cardiac failure/complications 6 Jaundice 6 Diuresis volume 5 Coma 5 Male sex 4 Sepsis 3 Chronic disease 3 Neoplastic disease 2 Other organ failures 2 Serum creatinine 2 Other conditions 12 Summary Clinical variables 20 Laboratory variables 6

1960-1969

P

1980-1989

No.

119

124

Mortality (%)

51

NS

63

Mean age (y)

50.9

< 0.0001

63

Median APACHE II score

32

< 0.0001

35

Range

(22-45)

Prognosis in acute renal failure (ARF). This figure shows the utility of a prognostic system for evaluating the severity of ARF over time, using the experience of Turney [38]. He compared the age, mortality, and APACHE II score of ARF patients treated at one hospital between 1960 and 1969 and 1980 and 1989. In the latter period there were significant increases in both the severity of the illness as measured by APACHE II and age. Although there was a tendency to a higher mortality rate in the second period, this tendency was not great enough to be statistically significant.

### FIGURE 8-35

Outcome of acute renal failure (ARF). A great number of variables have been associated with outcome in ARF by multivariate analysis. This figure gives the frequency with which these variables appear in 16 ARF studies performed with multivariable analysis (all cited in [30]).

70 6050 t 40

1 30-Mor 20

10 0

Dialysis patients

68

42

22 ± 6

— Apache II score —►

22 ± 6

Nondialysis patients r

Nondialysis patients

 Time Nonsurvivors Survivors Admission in ICU 24 22 Before dialysis 22 22 24 h after dialysis 25 22 48 h after dialysis 24 APACHE score. The APACHE II score is not a good method for estimating prognosis in acute renal failure (ARF) patients. A, Data from Verde and coworkers show how mortality was higher in their ICU patients with ARF needing dialysis than in those without need of dialysis, despite the fact that the APACHE II score before dialysis was equal in both groups [39]. B, Similar data were observed by Schaefer's group [40], who found that the median APACHE II score was similar in both the surviving or nonsurviving ARF patients treated in an intensive care unit. Recently Brivet and associates have found that APACHE II score influences ARF prognosis when included as a factor in a more complex logistic equation [2]. Although not useful for prognostic estimations, APACHE II score has been used in ARF for risk stratification. P<0.001 0.57 Dialysis P<0.001 0.35 No dialysis FIGURE 8-38 Analysis of the severity and mortality in acute renal failure (ARF) patients needing dialysis. This figure is an example of the uses of a severity index for analyzing the effect of treatment on the outcome of ARF. Looking at the mortality rate, it is clear that it is higher in patients who need dialysis than in those who do not. It could lead to the sophism that dialysis is not a good treatment; however, it is also clear that the severity index score for ARF was higher in patients who needed dialysis. Severity index is the mean of the individual severity index of each of the patients in each group [36]. (Data from Liano et al. [1].) FIGURE 8-39 Causes of death. The causes of death from acute renal failure (ARF) were analyzed in 337 patients in the Madrid ARF Study [1]. In this work all the potential causes of death were recorded; thus, more than one cause could be present in a given patient. In fact, each dead patient averaged two causes, suggesting multifactorial origin. This could be the expression of a high presence of multiple organ dysfunction syndrome (MODS) among the nonsurviving patients. The main cause of death was the original disease, which was present in 55% of nonsurviving patients. Infection and shock were the next most common causes of death, usually concurrent in septic patients. It is worth noting that, if we exclude from the mortality analysis patients who died as a result of the original disease, the corrected mortality due to the ARF episode itself and its complications, drops to 27%. GI—gastrointestinal; DIC—disseminated intravascular coagulation.