Abstract

Sepsis is one of the leading causes of morbidity and mortality world-wide today. Prognostic categorization or stratification of patients with systemic sepsis requires repeated clinical interpretations of the patient condition, clinical assessment of tissue hypoxia and the use of severity scoring systems.  This is because the accuracy of severity scoring indices for prognostic stratification of critically ill patients is relatively poor.  Generally, such prognostic stratification depends on the severity of the septic state, ranging from systemic inflammatory response to septic shock.
Up-till-now, there is no gold standard for the clinical assessment of tissue hypoxia which can be achieved by both global and regional oxygen extractabilities, added to prognostic biologic markers as procalcitonin (PCT) and proinflammatory mediators. Because the technology used to identify the genetic make-up of the human being is rapidly advancing, the structure of 30.000 genes which make-up the human DNA bank is now known and can be downloaded from the USA National Human Genome Project Internet Site. This may allow easy prognostic stratification of critically ill patients including those suffering from sepsis.
A large number of severity scoring systems is now available for prognostic prediction of patients with systemic sepsis.  The present review spots light on those systems of widespread clinical application. For morbidity prediction, it discusses the Multiple Organ Dysfunction (MOD) score, the Sequential Organ Failure Assessment (SOFA) score, and the Logistic Organ Dysfunction (LOD) score. For mortality / survival prediction, it discusses the APACHE scores, the Therapeutic Intervention Scoring System (TISS), the Simplified Acute Physiology Score (SAPS) and the Mortality Probability Models (MPM).  As the ideal severity scoring system for prognostic stratification of patients with systemic sepsis is far from being reached, those systems should be used in conjunction with repeated clinical interpretations of the patient condition, together with the assessment of tissue hypoxia in order to attain satisfactory discriminative performance and calibration power.

Key words:
Cytokines; extraction ratio; genome; prognostic markers, stratification; sepsis; systemic inflammatory response syndrome; severity scoring systems.

Introduction

Prognostic categorization or stratification of patients with systemic sepsis may be tried through sequential repeated clinical interpretations, assessment of tissue hypoxia and the use of severity scoring systems. The major prognostic value of scoring systems is mainly to compare the effectiveness of ICU services in different centers or over time. The accuracy of scoring systems for prognostic stratification in patients with systemic sepsis is relatively poor. So, to determine patient outcome both the clinical assessment and the scoring systems are needed.
In 1991 (1), experts from a variety of disciplines met for a Consensus Conference sponsored by the American College of Chest Physicians and the Society of Critical Care Medicine and proposed new definitions for sepsis as follows:
- Systemic Inflammatory Response Syndrome (SIRS):
It denotes the systemic inflammatory response to a wide variety of severe critical insults, manifested by two or more of the following conditions:
Temperature > 38 ° or < 36°c
Heart rate > 90 beats / min
Respiratory rate > 20 breaths/min or PaCO2 < 32 mmHg
WBC > 12.000/mm3, < 4000/mm3,  or   > 10% immature forms 
Sepsis:
It denotes the systemic inflammatory response to infection.
Severe Sepsis / SIRS:
They denote sepsis or SIRS associated with organ dysfunction, hypoperfusion or hypotension. Hypotension and hypoperfusion abnormalities may include lactic acidosis, oliguria or acute alteration in mental status. Systolic blood pressure < 90 mmHg or a reduction of 40 mmHg from the baseline in the absence of other causes of hypotension notify severe sepsis or SIRS.  It is usually corrected by fluid loading.
Septic Shock or SIRS Shock:
They denote sepsis or SIRS induced hypotension not corrected by fluid loading and need inotropic and/or vasopressor support.  Perfusion abnormalities to many organs characterize the shock state.

Multiple Organ Dysfunction (MOD) Syndrome :
It represents altered organ functions in an acutely ill patient to the extent that homeostasis cannot be maintained without intervention. 
Prognostic stratification or categorization of patients with systemic sepsis depends on the severity of the critical illness state.
Sepsis is one of the leading causes of morbidity and mortality worldwide today. It is estimated that there are approximately 700.000 cases of severe sepsis annually in USA and around 400.000 patients die every year as a result of sepsis in both USA and Europe. The incidence of the various degrees of severity of sepsis is not well known but a relatively small Italian study which looked at approximately 1000 ICU admissions found the following (2) (Table 1):
Table (1): Mortality in various degrees of severity of sepsis (2)

Clinical Assessment of Tissue Hypoxia in sepsis

Tissue hypoxia is defined as a decrease in the partial pressure of oxygen in a given tissue or as a condition in which the cells of a tissue have abnormal oxygen utilization such that the tissue is experiencing anaerobic metabolism.

(I) Global VO2/DO2 Relationship:
The relationship between whole body DO2 and VO2 in human sepsis has been extensively studied but remains controversal.  The pathological supply dependency is an evidence of occult tissue hypoxia and has been associated with an increased incidence of MODS and poor outcomes in patients with sepsis (3).  Support for this belief comes from some clinical investigators who have demonstrated improved outcomes in patients with septic shock by pharmacologically augmenting systemic oxygenation to supranormal levels (4,5). However, other investigators thought that these clinical studies should be criticized because of methodological error from mathematical coupling because DO2 and VO2 were calculated from a common set of measured variables; cardiac output and arterial oxygen content (6).  The author of the present review was not able to find significant reduction in mortality in septic patients managed by using the supranormal hemodynamic approach (7).
However, it may be prudent to think that indices of supranormal oxygenation for management of patients with sepsis may be used for their prognostic stratification. Patients who can attain supranormal values have decreased morbidity and mortality, mostly due to better physiological reserves.  Based on this, it may be concluded that global VO2/DO2 relationship based on good oxygen extractability potentiality may denote that the extraction ratio is an excellent parameter for prognostic stratification of patients with sepsis.
Mixed venous oxygen saturation (SVO2) determination by pulmonary artery catheter is a flow-weighted average of venous effluent from all perfused vascular beds. A decrease in SVO2 can be caused by a decrease in DO2 and / or an increase in VO2.  An increased value in septic patients denotes tissue hypoxia and its improvement by normal or supranormal pharmacological interventional therapy may be used as a good prognostic marker (4).
Metabolic lactic acidosis development is one of the most important abnormalities of tissue hypoxia due to the production of hypoxic global lactate during sepsis or septic shock. Plasma lactate has been shown to be a good prognostic indicator of hypoperfusion in critically ill patients. Plasma lactate is easy to measure and can be followed sequentially to assess the prognosis of the response of septic patients to therapy (8).  The more the decrease in pH and the higher the value of base deficit, the more serious the condition of the septic patient is.
Prognostic Markers:
Procalcitonin (PCT) and proinflammatory mediators such as tumour necrosis factor - α  (TNF - α ), interleukine – 1beta (IL-1B) and interleukin – 6 (IL-6) are important clinical prognostic markers in patients with systemic sepsis (9).
There has been strong correlation between serum concentrations of proinflammatory mediators and scores of severity of illness (10).  In spite of this, these mediators are not established for clinical decision making due to their short half-life (11).
Casey et al (1993) (12) designed a biologic score for application in septic  patients. It included levels of endotoxin, 1L-IB, TNF- α and 1L-6.  It proved a strong correlation with mortality in septic patients. However, the same goal could be achieved by estimation of blood lactate level as an easier and cheaper test. 
Nylen et al (13) and Hatheril et al (14) presented the first evidence that PCT, one of the best prognostic markers of sepsis (15), may actually be a sepsis mediator and could have an integral role in the inflammatory process and its prognostic stratification or categorization.
It has been shown that in vitro and in vivo induction of cytokines leads to release of PCT (16, 17) which is released rapidly and has long half-life (18). 
Oberhofer et al, (19) showed that PCT concentration on the first day of the diagnosis of sepsis, severe sepsis or septic shock was significantly higher in non-survivors than in survivors. Ugarte et al (20) also proved a strong correlation between PCT and survival.
Using stepwise discriminant analysis, PCT was proved to be the best single predictor of outcome in patients with systemic sepsis as it allocated survivors in 95.8% and non-survivors in 83.3% of patients, with an overall prediction accuracy of 80%. (21)
There has been recent reports of altered outcome in sepsis due to the release of lipopolysaccharide binding protein, bacterial permeability inducing protein and other key proteins which may result in altered disease susceptibility and severity as heat shock protein 70 and nitric oxide synthase (NOS).
It has long been appreciated that many patients with sepsis demonstrate defects of coagulation and fibrinolytic systems.  These are manifested as anti-thrombin III, Protein C and Protein S and the consumption of fibrinogen and the appearance of disseminated intravascular coagulation (DIC).  More recently, there has been a report of a randomized multicenter trial which has examined the use of a novel human activated protein C. (22)  A total of 1690 patients were enrolled into the study; 850 patients received the protein C preparation and 840 received placebo.  All patients had severe sepsis.  The mortality rate was decreased from 30.83% in the control group to 24.71% in the active treatment group, an effect which was statistically significant.  This report may clearly have major implication for the prognostic stratification and management of patients with systemic sepsis.
The general interest in genetics culminated in the publication of the findings of the human genome project which appeared in February 2001 issue of Nature. The precise structure of 30.000 genes which make up the human DNA bank is now known and can be downloaded from the USA National Human Genome Project Internet Site.  Such knowledge will prove useful because it will increase the understanding of the etiology and pathology of many disease processes.  Because the technology used to identify the genetic make-up of individual patients is now advancing so rapidly, it will soon be possible to identify more markers in patients and will allow prognostic stratification of septic patients for future trials of new therapeutic approaches.

(II) Regional VO2 / DO2 relationship:
The technique of gastric or sigmoid tonometry measures intramucosal pH (Phi) by allowing the equilibration of CO2 pressures between fluid or air-filled balloon and the interstitial fluid of the mucosa.  Measurement of gut intramucosal CO2 is achieved through air introduced directly into the gut (balloonless air tonometry), which equilibrates with the interstitial fluid of the mucosa and is then aspirated from the stomach (23).  Measurement of CO2 content of fluid aspirated from the stomach has been also described by Mohsenifar et al,(24).  Both methods avoid the use of commercial expensive tonometry catheters costing $ 200 each (25).
Phi may decrease due to changes in blood flow to the stomach or sigmoid mucosa due to splanchnic ischemia in shocked patients. Phi appears to be useful for prognostic stratification of patients with systemic sepsis based on serial measurements (26).  The author of the present review has shown that Phi values were significantly lower in septic patients with MOD on admission to the ICU than in patients with no organ dysfunction(7).
(III) Global Versus Regional VO2/DO2 relationship:
Assessment of both global and regional VO2/DO2 relationships can combine both sides of the coin in prognostic stratification of patients with systemic sepsis.  However, there is no gold standard for the detection of tissue hypoxia.  There are no specific clinical signs and no clearcut threshold for any single laboratory test.  But a multitude of tests combined with sequential clinical evaluations of septic patients may be the best way for their prognestic stratification. So, management of patients with severe sepsis or septic shock may be through haemodynamic oriented or splanchnic directed therapy added to sequential repeated clinical interpretations. This is a gold standard for both therapy and prognostic stratification of patients with systemic sepsis.

Severity scoring systems in systemic sepsis
Severity scoring systems provide numerical scores that describe the impact of patients’ illnesses on their physiological reserves.
Most of the severity scoring systems include assessment of major organ system functions.  Two main types of scoring systems have been developed for use in ICU patients: those that focus on describing morbidity as it evolves; organ dysfunction systems, and those that focus on a single end point, survival or mortality (27). So, severity scoring systems are usually designed to help in the prognostic stratification or categorization of critically ill patients as regards their morbidity or survival (28).

I- Morbidity Prediction Systems:
Morbidity prediction systems include a large number of scoring trials by different authors, based on advanced statistical efforts for different populations of critically ill patients at various centers.  We chose to concentrate on three scoring systems that proved useful for clinical applications to be discussed in the present review, namely, the Multiple Organ Dysfunction (Mod) score, the Sequential Organ Failure Assessment (SOFA) score, and the Logistic Organ Dysfunction (LOD) score. However, other scoring systems may prove useful and the ideal prediction scoring system has not been reached yet.  It should be noted that these systems do not replace serial clinical interpretations of the patients.

The Multiple Organ Dysfunction Score:
The Multiple Organ Dysfunction (MOD) scoring system was developed by Marshall et al in 1995. (29) (table 2)

Table (2): The Multiple Organ Dysfunction (MOD) Score

a Pressure-Adjusted Heart Rate: product of the heart rate multiplied by the ratio of the right atrial pressure to the mean arterial pressure.
It included six key organ systems and a score of zero to four was given to each organ according to function (zero being normal function and four being the most severe dysfunction), with a maximum score of 24.  A mortality rate of 25% was observed for patients with a score of 9-12, 50% for a score of 13-16, 75% for a score of 17-20 and 100% for a score > 20.
A minor revision (30) of this score has abandoned the cardiovascular parameter (pressure-adjusted heart rate) in favour of a mixed cardiovascular parameter based on heart rate / min.  (table 3) as follows:
0 = heart rate < 120; 1 = heart rate > 120 and < 140; 2 = heart rate > 140; 3 = need for inotrope : (dopamine > 3 µg/kg/min), and 4 = lactate > 5 mmoL/L. The detailed analysis of the results of daily scoring demonstrated the prognostic insights gained by adopting this system (30).
Table (3): The Multiple Organ Dysfunction (MOD) Score (Revised)

The Revised MOD scoring system proved to be of value, because pressure adjusted heart rate cannot be measured in a significant proportion of ICU patients due to the absence of a central venous monitor.  In fact, approximately one half of the patients in the original Marshall et al (29) study could not have a cardiovascular component calculated.
It is recommended that the MOD score and its Revised form should be measured at the same point in time every day (first morning values).  The use of measurements at one particular time avoids capturing momentary physiological changes unrelated to patient condition.
In a small study, Jacobs et al (31) compared daily MOD scores to daily APACHE II scores in 39 septic shock patients from one Saudi Arabian ICU.  The authors (31) found that the maximum MOD score and the maximum change in the score from admission, both discriminated (the ability to predict mortality in individuals) very well between survivors and non survivors, whereas APACHE II score did not. 
To summarize, the MOD score and its Revised form can be used to represent organ dysfunction at baseline and during ICU stay.  They can also significantly contribute to the prediction of hospital or ICU mortality.

The Sequential Organ Failure Assessment (SOFA) Score:
The Sequential Organ Failure Assessment (SOFA) score (table 4) was developed in 1994 during a Consensus Conference organized by the European Society of Intensive Care and Emergency Medicine, in an attempt to provide a means of describing the degree of organ failure over time in individuals and groups of septic patients (32). 
It was initially termed the Sepsis-related Organ Failure Assessment score,  but it has been realized that it could be applied to non-septic patients as well.
It includes scores for six organ systems where a score of zero is given for normal function and a score of four is given for the most abnormal one.  The worst values on each day are recorded and organ function total score can thus be monitored over time (27).
Table (4): The Sequential Organ Failure Assessment Score.

a Adrenergic agents administered for at least 1h (doses given are in µg/kg/min)

Vincent et al (33) in 1998 working on « sepsis-related » problems published the first evaluation of the SOFA score.  They found that infected patients had more severe organ dysfunctions compared to those without infection. Antonelli et al (34) in 1999 proved that the mean total maximum SOFA score was significantly higher for non-survivors than survivors denoting a high discriminative power (the ability to predict mortality in individuals).  Because the total maximum SOFA score can be easily calculated daily for the patient, no restriction based on the patients’ ICU length of stay is necessary.  So, increasing organ dysfunction as measured by the SOFA score consistently correlates with increasing mortality. The SOFA score is also a reliable measure of organ dysfunction at ICU admission.
There are at least four recently published studies that have since examined the utility and accuracy of the SOFA score, which proved that maximum SOFA score and increasing SOFA score are highly prognostic for stratification of critically ill patients including septic patients (35-38).
The Logistic Organ Dysfunction (LOD) Score:
The Logistic Organ Dysfunction (LOD) score (table 5) was developed in 1996 using multiple logistic regressions applied to selected variables from a large database of ICU patients (39).  The score consists of six organ systems and 12 variables with a maximum of 22 points.  If no organ dysfunction is present the score is zero, rising to a maximum of five as the worst severity organ dysfunction.

Table (5): The Logistic Organ Dysfunction (LOD) Score

FiO2: Fraction of inspired oxygen; IPAP : inspiratory positive airway pressure; CPAP: continuous positive airway pressure
For maximum dysfunction of the pulmonary and hematologic systems, a maximum of 3 points can be given for the most severe levels of dysfunction and for the liver, the most severe dysfunction only receives one point.  The variables had been recorded as the worst value of each organ dysfunction in the first 24 hours of ICU admission.  A reference table converts the score to a probability of hospital mortality, the relationship being sigmoid. The score can thus discriminate between survivors and non-survivors.
The LOD score aims to achieve similar goals to the MOD score, namely, to quantitatively and qualitatively describe organ dysfunction.  The goal is to provide a tool that can itself provide a useful outcome measure (e.g., improvement/resolution of organ dysfunction) rather than merely predicting mortality.  Though, not originally described as a serial measure, it appears that the LOD score may hold the most promise for patient outcome in the future.

II- Mortality/Survival Prediction Systems
Mortality / survival prediction scoring systems include a large number of scoring trials by different authors, based on advanced statistical efforts including equations for different populations of critically ill patients at various studies.  We chose to concentrate on important examples which are useful for clinical prognostic stratification of mortality/survival of patients namely; the APACHE scores, the Therapeutic Intervention Scoring System (TISS), the Simplified Acute Physiology Score (SAPS) and the Mortality Probability Models (MPM). However, other scoring systems may prove useful and the ideal scoring system for mortality/survival prediction has not been reached yet.  It should be noted that these systems do not replace serial clinical interpretations of the patients.

The APACHE Scoring Systems:
The APACHE II scoring system was developed by Knaus et al in 1985(40) as a refinement of the original APACHE score.  It consists of: Acute Physiology Score (APS), Age points and Chronic Health points. The reduced number of physiological variables of APS from 34 in the original APACHE to 12 in APACHE II was achieved by a multivariate analysis.  The total physiological derangement score is the sum of the individual scores (0-4) for each variable, except the Glasgow Coma Scale (GCS) where the score is 15 minus the GCS.  The most deranged value in the first 24 hours of ICU admission is used as the scoring for each variable.  The total physiological derangement score is added to a score of age (0-6) and a chronic health score for patients with severe organ insufficiency (2 to 5 dependent upon admission status).  The total APACHE II score ranges between zero and 71 points.  Points of 25 or less denote less than 50% mortality while points of 35 or more denote more than 80% mortality. However, some investigators have used APACHE II scoring over time to assess the prognosis of individual patients (41).
Generally, data of the APACHE II score are computed through the following equation to deliver the final risk of hospital mortality:
(R/1-R) = -3.517 + (APACHE II x 0.146 + S + D)
where: R = Risk of hospital death.  S = Risk imposed by emergency surgery and D = Risk imposed by specific disease.
Under the APACHE II system, the predicted individual death rate is based on a decision criterion of 0.50.  Any patient with an estimated risk of death greater than 0.50 is simply expected to die (42). 
Although the APACHE II score provides valuable information about the severity of illness of patient groups, they provide little information about individual patients (43). For example, an APACHE II score of 20 does not tell whether the patient has severe renal failure or acute respiratory failure, whereas analysis of component scores of an organ dysfunction score as SOFA will provide an accurate description of the patients’ disease status.  This does not mean that organ dysfunction scores as the SOFA score should replace APACHE II score but that the two can provide different information and may be used to complement each other (27).
In 1991, Knaus et al (44) published a further refinement to their severity of illness scoring system termed APACHE III. Turning first to the APS, they added some variables and eliminated some parameters.  Additional weights were assigned to the extremes of physiological measures.  For example, the risk associated with extremely high readings is different from that associated with equally low readings. Glasgow Coma Scale variables were also refined.  The authors also re-weighted age and derived an extended chronic health co-morbidity score.  Thus, the APS in APACHE III ranged between zero and 252 points while the total score reached 299 points.
The equation of hospital prediction mortality by APACHE III differed from that of APACHE II and included a risk of location denoting the condition of transference of the patient from a previous locality, as such:
R/1-R = (APACHE III Score x 0.053) + Risk of emergency Surgery + Risk of specific disease category + Risk of patient location.
Similar to APACHE II score, the predicted death rate of the APACHE III score is based on decision criterion of 0.50 with predicted mortality if R exceeds 0.50.
Independent validation of APACHE III has been undertaken by a number of studies (45-49). which proved acceptable discrimination performance  (the ability to predict mortality in individuals as measured by the area under a receiver operating characteristic curve) and inadequate calibration power  (the ability to predict mortality in a large population as measured by a goodness-of-fit test). However, a recent evaluation of the performance of the APACHE III for 5681 admissions (Jan 1995 to Jan 2000) was conducted at the Princess Alexandra Hospital in Australia (50). The study demonstrated that the APACHE III model performs well as the standardized mortality ratio (SMR) which compares observed versus expected mortality, showed annual decreases during the study period.  Statistics showed excellent discrimination and good calibration.
A critical prognostic importance of APACHE III, may be based on the premise that the changes in APS would reflect the patient response to therapy.  The daily APS component of the risk equation would be given by the formula:
Daily risk = day 1 APS + current day APS + change in APS since yesterday
Day 1 APS is a significant predictor of hospital mortality, but its relative influence decreases dramatically over time.  The current day APS as the most important single factor should be measured retrospectively as scoring values are the most deranged in any 24 hours period. 
When the daily risk is added to the remaining patient variables included in the APACHE III score, the coefficients of each variable were established resulting in equations for days 1-7 of ICU admission.  Research is going on to extend the model beyond day 7.

The Therapeutic Intervention Scoring System (TISS):
The Therapeutic Intervention Scoring System was developed by Cullen et al(51) in 1974 as the earliest severity scoring system.  It is composed of 76 monitoring and therapeutic parameters.  Each modality is assigned a weighted score, ranging from 1 to 4, depending on the intensity of intervention.  For example, a peripheral iv line or a urinary catheter is assigned one point.  A central venous line or two peripheral iv catheters are assigned two points.  A central iv line for hyperalimentation or the application of chest tubes is assigned three points.  A pulmonary artery catheter or vasoactive drug infusion is assigned four points.  Each modality is assigned to one of three categories: active therapy, ICU monitoring or standard floor care. Points are totaled and TISS score is obtained.  Patients can be then stratified into one of four classes based on the number of TISS points.  TISS is based on the premise that, regardless of the diagnosis, the amount of therapy based on the amount of monitoring reflects the degree of physiological impairment.  The TISS does not predict outcome on patient admission to the ICU.  However, trends of the score over the first three days in ICU correlate well with survival.  If the TISS points do not improve at the third day, the likelihood of death increases. So, it discriminates between survivors in whom the score falls progressively and non-survivors in whom the score remains static. Moreover, the TISS can identify those patients who require monitoring only.
The TISS is now used most frequently in conjunction with the APACHE system.  So, both together can be used to evaluate concordance between severity of illness and quantity of needed therapy. So, it is clear that either the TISS alone or in conjunction with the APACHE scoring systems can be used for prognostic stratification of patients with systemic sepsis.
Cullen et al (1994) (52) have released an intermediate therapeutic intervention scoring system as a modification of the original one with deletion of many active items and re-emphasis on the nursing input.  It has been designed for the intermediate care or high dependency care unit, where the patient needs non-invasive monitoring, a higher nurse-patient ratio than the standard ward, yet a little active intervention.

The Simplified Acute Physiology Score (SAPS) :
In 1984, Le Gall et al (53), published the Simplified Acute Physiology Score.  It was designed to overcome some of the problems of APS of the APACHE systems.  The authors selected the 13 “most easily measured” physiological variables available in 90% of patients from a previous survey employing the APS that they had conducted.  SAPS scores these variables (0-4) in an identical manner to the APS of the APACHE system, adds a score for age (0-4) and replaces respiratory rate or the P(A-a) O2 which is difficult to measure with a fixed score of 3 for patients receiving mechanical ventilation or CPAP.  The most abnormal values from the first 24 hours of ICU admission are taken as the total scoring value.
Le Gall et al (53) concluded that SAPS performed at least as well if not better than APS of the APACHE system but was more useful as it was much simpler.  They stressed that SAPS is applicable to a wide range of pathologies but that its predictive value and performance can only be applied to groups of patients, not to individual patients.
In 1993, Le Gall et al (54) published a refined version of their original SAPS termed SAPS II and the variables were 17 (12 physiological, age, type of admission and 3 chronic health diagnosis).
The main advantage of SAPS II over APACHE III is the ability to accurately predict mortality in stratified groups of patients without recourse to defining a single diagnosis, which is only possible in a minority of patients.
It is clear that SAPS II can be useful for prognostic stratification for groups of critically ill patients including those with systemic sepsis.  It can also be useful for guiding therapy, comparing the management of these patients overtime and comparing ICU performance of groups of patients in different ICU’s.

The Mortality Probability Models (MPM):
In 1985, Lemeshow et al (55) published their first attempt at an outcome prediction model.  They actually developed four models: MPM0 (probability of death from data collected at ICU admission), MPM24 (probability of death from data collected at 24 hours), MPM48 (probability of death from data collected at 48 hours) and MPM0T (probability of death “overtime” based on MPM0 and the change in probability between MPM0 and MPM24, and between MPM24 and MPM48).   Patients whose probability of mortality started high and stayed high, or increased by >10% had a very high actual mortality. It deserves mentioning that for ICU triage purposes, MPMo is the most valid model at present.
In common with APACHE and SAPS systems, MPM had low sensitivity (ability to predict those patients who are going to die) but high specificity (ability to predict those patients who are going to live).
In 1993, Lemeshow et al (56) published a revision of their MPM termed MPM II.  They employed a near identical method to that they had used in developing their original MPM.  The authors initially developed MPM II0 and MPM II24, deciding to temporarily abandon the MPM48 and MPM 0T of the original model.
MPM II0 was determined by 15 variables. Lemshow et al (56) found that patients alive but still requiring to be on ICU at 24 hours differed markedly from those who had either died or been discharged. They emphasized that MPM II24, including 13 variables, is a companion model to MPM II0 and represents a different population of patients.  The authors argue that this approach exposes one of the main weaknesses of the APACHE and SAPS models, which take the worst data from the first 24 hours of ICU admission, and failed to differentiate between the two originally observed populations. 
The following year, Lemeshow et al (57) published two further models based upon their data set, MPM II 48 and MPM II72.  Both these models use the same 13 variables as MPM II 24. They pointed out that the probability of death changes with time, while an APACHE or SAPS score is only valid at 24 hours of ICU admission.  They also emphasized that an ICU patient whose condition failed to improve day after day, was in fact deteriorating and had an increasing risk of death.  This well recognized clinical phenomenon is accurately modelled over the first 72 hours of their ICU stay by MPM II.  The same could not be said for sequential APACHE II scoring.   The authors described an on-going process to develop MPM II models for successive time points beyond 72 hours. (MPM II OT)

Overview

It is now about 20 years since the original APACHE study was published.  Tens of thousands of patients have been studied and mortality prediction models and prognostic stratification morbidity models developed on universal ICU critically ill patients.  
Severity Scoring Systems are usually designed to predict morbidity or mortality of critically ill patients. Examples of general scoring systems are APACHE III; SAPS II AND MPM II.  Examples of organ dysfunction scoring systems are MODS, SOFA and LODS.  Examples of specific severity scoring systems include acute pancreatitis and acute lung injury scores. Biological scores include measurements of serum lactate and PHi. Examples of overtime or dynamic severity scoring systems are APACHE III, MPM 24-72 and intermediate TISS.
Because general severity scoring systems are developed and validated using admission data from large ICU populations, they are most fitted to predict mortality for groups of ICU patients rather than predicting mortality for individual patients.  They are used for determining ICU proficiency (in quality assurance) and treatment efficacy (in clinical practice).  Decisions regarding ICU triage are often more dependent on values than probabilities and so, these systems should not determine the utility or futility of ICU for individuals. 
Even if a severity scoring index could perfectly predict the mortality of a septic patient from admission data, one should be cautions, because death cannot actually be predicted except just before its occurrence and by that time there would be little to be gained.
By contrast, early prediction of death might be more useful to design patient management.  It would be likely to be associated with a greater risk of a false positive result.
Outcome estimates may influence the clinical management.  The clinical awareness of the treating physician of a poor outcome for his patient may tempt him to give less than optimal therapy or to prevent ventilating him or even to withdraw active therapy.  So, inadequacy of Severity Scoring Systems is a mercy from ALLAH to the critically ill patients.
To date, however, it is almost impossible to find documented evidence of change in medical practice that have resulted from application of different prognostic scoring systems (28).  There is clearly no “best” severity scoring model, and the performance of such models varies both with time and with the population under study and so should be periodically addressed.  For this, severity scoring systems should be used in conjunction with sequential patient clinical interpretations and clinical assessment of tissue hypoxia for prognostic stratification or categorization of critically ill patients in general and septic patients in particular.

References

1. Bone RC, Blak RA, Cerra FB. Definitions of sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest 1992, 101 : 1644-1655.
2. Sola M. Intensive Care Medicine 1995 ; 21 : 5244-5249. Quoted from Nigel R. Webster: Activated Protein C in sepsis. AJAIC 2001; 4 (Suppl 1): 41-42.
3. Beal AL and Cerra FB. Multiple organ failure syndrome in the 1990s: Systemic response and organ dysfunction. JAMA 1994 ; 271 : 226-233.
4. Shoemaker WC, Appel PL, Kram HB. Prospective trial of supranormal values of survivors as therapeutic goals in high-risk surgical patients. Chest 1988; 94: 1176-1186.
5. Edwards JD, Brown GCS, Nightingale P. Use of survivors’ cardiorespiratory values as therapeutic goals in septic shock. Crit Care Med 1989; 17: 1098-1103.
6. Russell JA and Phang PT. The oxygen delivery/consumption controversy: Approaches to management of the critically ill. Crit Care Med. 1994; 149: 533-570.
7. Moemen ME, Omran SA, Rifky MA, Hasan AA. Tissue oxygenation guided resuscitation in patients with sepsis. Thesis in anaesthesia and intensive care, Faculty of medicine, Zagazig university, 2001, Egypt.
8. Bakker JM; Leon MC, Gris P, Vincent JL. Blood lactate levels are superior to oxygen-derived variables in predicting outcome in human septic shock. Chest 1991; 99: 956-962.
9. Dahaba AA, El-Awady GA, Rehak PH, List WF. Procalcitonin and proinflammatory cytokine clearance during continuous venovenous haemofiltration in septic patients.  Anaesth Intens Care 2002; 30: 269-274.
10. Monton C, Torres A, El-Ebiary M, Fillela X, Xaubert A, de la Bellacasa JP. Cytokines expression in severe pneumonia: A broncho-alveolar lavage study. Crit Care Med 1999; 27: 1745-1753.
11. Wakerfield CH, Barclay GR, Fearon KC, Goldie AS, Ross JA, Grant JS, Ramsay G, Howie JC. Proinflammatory mediator activity, endogenous antagonists and the systemic inflammatory response in intra-abdominal sepsis. Br J Surg 1998: 85; 818-825.
12. Casey LC, Balk RA, Bone RC. Plasma cytokine and endotoxin levels correlate with survival of patients with the sepsis syndrome.  Ann Intern Med 1993; 119: 771-778.
13. Nylen ES, Whang KT, Sinder RH, Steinwald PM, White JC, Becker KL. Mortality is increased by procalcitonin and decreased by an antiserum reactive to procalcitonin in experimental sepsis. Crit Care Med 1998; 26: 1001-1006.
14. Hatheri M, Tibby S, Lim E, Marsh M, Murdoch I. Procalcitonin in pediatric sepsis. Crit Care Med. 1998; 26 (suppl1): 136.
15. Assicot M, Gendral D, Carsin H, Raymond J, Guilband J, Bohoun C. High serum procalcitonin concentration in patients with sepsis and infection. Lancet 1993; 341: 515-518.
16. De-Werra I, Jaccob C, Corrdin SB. Cytokines, nitrite/nitrate, soluble tumour necrosis factor receptors and procalicitonin concentrations: Comparisons in patients with septic shock, cardiogenic shock and bacterial pneumonia. Crit Care Med 1997, 25(4): 607-613.
17. Nijsten MW, Olinga P, Koops HS, Groothuis GM, Limburg PC, Bijzet J. Procalcitonin behaves as a fast responding acute phase protein in vivo and in vitro.  Crit Care Med 2000; 28: 458-461.
18. Frederic D, Richard D, Guillaum M, Jacques B, Dominique C, Bernard A.  Alveolar and serum procalcitonin; diagnostic and prognostic value in ventilator associated pneumonia.  Anesthesiology 2002; 96: 74-79.
19. Oberhofer M, Bogee D, Meier HA, Reinhart K. ACCP / SCCM Consensus Conference definitions correlated better with procalcitonin than tumour necrosis factor and interleukine-6. Intens Care Med. 1996; 22 (Suppl1): 15.
20. Ugarte H, Silva E, Mercan D, De Mendonca A, Vincent JL.  Procalcitonin used as a marker of infection in the intensive care unit.  Crit Care Med 1999; 27: 452-453.
21. Fathia HK, Mokhtar MS, Ragab FA, Rizk AF. Role of decisive markers in diagnosis and outcome of patients with septic shock. Thesis in critical care medicine, Faculty of medicine, Cairo University, 2000.
22. Bernard GR. New Eng J Med 2001; 344: 699-709. Quoted from Nigel R. Webster: Activated protein C in sepsis. AJAIC 2001; 4 (suppl 1): 41-42.
23. Salzman AL, Strong KE, Wang H, Wollert PS, Vandermeert J, Fink MP. Intraluminal (balloonless) air tonometry: a new method for determination of gastrointestinal mucosal carbon dioxide tension. Crit Care Med 1994; 22: 126-134.
24. Mohsenifar Z, Hay A, Lewis MI, Koerner SK.  Gastric intramural pH as a predictor of success or failure in weaning patients from mechanical ventilation. Ann Inter Med. 1993; 119: 794-798.
25. Corke CF, Prisco G, Gizycki P, Selvakumaran A. A simple method for frequent monitoring of gastric carbon dioxide. Anaesth Intens Care 1996; 24: 590-593.
26. Gutierrez G, Palizas F, Dolgio C. Gastric intramucosal pH as a therapeutic index of tissue oxygenation in critically ill patients. Lancet 1992; 339 (8787): 195-199.
27. Vincent JL; Ferreira F, Moreno R. Scoring systems for assessing organ dysfunction and survival. Crit Care Med 2000; 16: 353-366.
28. Duncan YJ.  Severity scoring systems and the prediction of outcome from intensive care. Curr Opinion Anaesth 2000; 13: 203-207.
29. Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL,  Sibbald WJ. Multiple Organ Dysfunction Score: a reliable descriptor of a complex clinical outcome. Crit Care Med. 1995; 23: 1638-1652.
30. Cook R, Cook D, Tilley J, Lee K, Marshall J. Multiple Organ Dysfunction baseline and serial component scores. Crit Care Med 2001; 29: 2046-2050.
31. Jacobs S, Zuleika M, Mphansa T. The Multiple Organ Dysfunction Score as a descriptor of patient outcome in septic shock compared with two other scoring systems. Crit Care Med 1999; 27: 741-744.
32. Vincent JL, Moreno R, Takala J.  The SOFA (Sepsis – related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis- Related Problems of the European Society of Intensive Care Medicine. Intens Care Med 1996; 22: 707-710.
33. Vincent JL, de Mendonco A, Cartraine F. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units : results of a multicenter,  prospective study. Working group on sepsis-related problems of the European Society of Intensive Care Medicine.  Crit Care Med 1998 ; 26 : 1793-1800.
34. Antonelli M, Moreno R, Vincent JL. Application of SOFA score to trauma patients. (Sequential Organ Failure Assessment). Intens Care Med 1999 ; 25 : 389-394.
35. Janseens U, Graf C, Graf J. Evaluation of the SOFA score: a single-center experience of a medical intensive care unit in 303 consecutive patients with predominantly cardiovascular disorders. Intens Care Med 2000 ; 26 : 1037-1045.
36. Oda S, Hirasawa H,  Sugai T. Comparison of Sepsis-related Organ Failure Assessment (SOFA) Score and CIS (Cellular Injury Score) for scoring of severity for patients with multiple organ dysfunction syndrome. Intens Care Med 2000; 26: 1786-1793.
37. Ferreira FL, Bota DP, Bross A, Melot C, Vincent JL. Serial evaluation of the SOFA score to predict outcome in critically ill patients. JAMA 2001; 286: 1754-1758.
38. Moreno R, Miranda DR, Matos R, Fevereiro. Mortality after discharge from intensive care: the impact of organ system failure and nursing workload use at discharge. Intens Care Med 2001; 27: 999-1004.
39. Le Gall JR, Klar J, Lemeshow S. The Logistic Organ Dysfunction system: A new way to assess organ dysfunction in the intensive care unit. JAMA 1996;276:802-810.
40. Knaus WA, Draper EA, Wagner DP. APACHE II: A severity of disease classification system. Crit Care Med 1985; 13: 818.
41. Bion JFD, Aitchison TC, Edlin SA. Sickness scoring and response to treatment as a predictor of outcome from critical illness. Intens Care Med 1988; 14: 167.
42. Rafkin H.S. Assessing the critically ill patient. In: Critical Care Practice (the American Society of Critical Care Anaesthesiologists), 1991; Chap. 1, W.B., Saunders, Page 1.
43. Chang RW. Individual outcome prediction models for intensive care units. Lancet 1989; 2: 143.
44. Knaus WA, Wagner DP, Draper EA. The APACHE III prognostic system.  Risk prediction of hospital mortality for critically ill hospitalized adults. Chest 1991; 100: 1619-1636.
45. Beck DH, Taylor BL, Millar B, Smith GB. Prediction of outcome from intensive care: A prospective cohort study comparing Acute Physiology and Chronic Health Evaluation II and III prognostic systems in a United  Kingdom intensive care unit. Crit Care Med 1997; 25: 9-15.
46. Pappachan JV, Millar B, Bennett ED, Smith GB. Comparison of outcome from intensive care admission after adjustment for case mix by the APACHE III prognostic system. Chest 1999, 115: 802-810.
47. Livingston BM, Mackirdy FN, Howie JC, Jones R, Norrie JD. Assessment of the performance of five intensive care scoring models within a large Scottish database. Crit Care Med 2000; 28: 1820-1827.
48. Zimmerman JE, Wagner DP, Draper EA, Wright L, Alzola C, Knaus WA. Evaluation of acute physiology and chronic health evaluation III predictions of hospital mortality in an independent database. Crit Care Med 1998; 26: 1317-1326.
49. Sirio CA,k Shepardson LB, Rotondi AJ. Community-wide assessment of intensive care outcomes using a physiologically based prognostic measure: implications for critical care delivery from C level and Health Quality choice. Chest 1999; 115: 793-801.
50. Cook DA, Joyce CT, Barnett RJ, Birgan SP, Playford H, Cockigns JGL, Hurford RW. Prospective independent validation of APACHE III models in an Australian Tertiary Adult Intensive Care unit. Anaesth Intens Care 2002; 30: 308-315.
51. Cullen DJ, Civetta JM, Briggs BA, Ferrara LC. Therapeutic Intervention Scoring System: A method for quantitative comparison of patient care.  Crit Care Med 1974; 2: 57-60.
52. Cullen DJ, Nemeskal RR, Zaslavsky AM.  Intermediate TISS: a new therapeutic intervention scoring system for non- ICU patients.  Crit Care Med 1994; 22-1406-1411.
53. Le Gall JR, Loirat P, Alperovitch A. A simplified acute physiology score for ICU patients. Crit Care Med 1984; 12: 975-977.
54. Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American mulitcenter study. JAMA 1993; 270: 2957-2963.
55. Lemeshow S, Teres D, Pastides H, Avrunin JS, Steingrub JS. A method for predicting survival and mortality of ICU patients using objectively derived weights. Crit Care Med 1985; 13: 519-525.
56. Lemeshow S, Teres D, Klar J, Avrunin JS, Gehlbach SH, Rapoport J. Mortality Probability Models (MPM II) based on an international cohort of intensive care unit patients. JAMA 1993; 270: 2478-2486.
57. Lemeshow S, Klar J, Teres D. Mortality probability models for patients in the intensive care unit for 48 or 72 hours: a prospective, multicenter study. Crit Care Med 1994; 22: 1351-1358.