Paruk F, Bhagwanjee S, Becker P, Muckart DJJ, Moodley J. Predicting outcomes in critically ill obstetric and gynecological patients. Critically Ill Obstetric and Gynecologic Patients: Development and Validation of an Outcome Prediction Model.
CHAPTER INDEX
CHAPTER THREE: RESULTS
OUTCOME PREDICTION
The art of predicting subjective outcome is not entirely accurate as human judgment is prone to error55,56. The ultimate goal of objective outcome predictions is to eliminate the variability that comes with subjective decision making.
ETHICAL CONSIDERATIONS IN OUTCOME PREDICTION
Modern medicine tends towards a utilitarian social approach, in an attempt to benefit the entire population. In such situations, a highly predictive outcome prediction model has the power to play an important additional role in medical decision making by adding more information to the subjective but scientifically informed opinion of the intensivist.
DEVELOPMENT OF OUTCOME PREDICTION MODELS
Continuous revision of the forecasting model for its efficiency, effectiveness and usability. Model accuracy is determined by assessing discrimination (the ability of the model to predict the outcome in an individual patient) and calibration (the ability of the model to predict the outcome across the entire range of risk).
OUTCOME PREDICTION TOOLS
- GENERAL OUTCOME PREDICTION MODELS
- THE APACHE SYSTEM
- SIMPLIFIED ACUTE PHYSIOLOGICAL SCORE (SAPS)
- MORTALITY PREDICTION MODEL (MPM)
- THERAPEUTIC INTERVENTION SCORING SYSTEM (TISS)
- ORGAN FAILURE
- ORGAN SYSTEM FAILURE {OS F)
- THE SEQUENTIAL ORGAN FAILURE ASSESSMENT (SOFA) SCORE
- THE SYSTEMIC INFLAMMATORY RESPONSE SYNDROME (SIRS) The progressive pathophysiological derangement, which precedes the
Organ System Failure, Multiple Organ Dysfunction Score, Sequential Organ Failure Score, and Logistic Organ Dysfunction Score were used to assess organ dysfunction. Organ dysfunction criteria are not specified and therefore depend on the interpreting ability of the attending physician.
PREDICTION MODELS IN THE CONTEXT OF OBSTETRICS AND GYNAECOLOGY
- OBSTETRIC PATIENTS
- General obstetric population
- GYNAECOLOGY PATIENTS
While this is perfectly acceptable, one should be aware of the potential limitations of using such a strategy in critically ill obstetrics and gynecology patients. The aim of the study is to stratify the severity of the disease, rather than predict the outcome.
DATABASE
An obstetric survey was defined in the context of a gestational state from ~16 weeks gestation to 6 weeks post delivery. APACHE II, SOFA score, Organ System Failure and SIRS were assessed in the entire cohort.
THE APACHE II SYSTEM
They felt that given the lack of control over patient care (after discharge from intensive care) in these circumstances, this could inadvertently lead to a greater adverse outcome.
ORGAN SYSTEM FAILURE
THE SYSTEMIC INFLAMMATORY RESPONSE SYNDROME (SIRS) This was assessed in terms of the definitions proposed by the American
Breakage and contamination of a normally sterile site with microorganisms SIRS was documented on admission and daily during the patients' leu stay.
THE OBSTETRIC AND GYNAECOLOGY OUTCOME PREDICTION (OGOP) MODEL
Significant outcome predictors were identified and an outcome prediction model (Obstetric and Gynecology Outcome Prediction. model) was developed. Obstetric hypertension (Group I, n= 144) and obstetric non-hypertension (Group II) accounted for 66.1% and 33.9% of the obstetric admissions, respectively. There was no significant difference in mean pregnancy, intensive care stay and parity in survivors compared to non-survivors for each of the subgroups (p>0.05 for all comparisons within each group).
There was no significant difference in duration of ventilation comparing Groups I and II (p>0.05).
THE APACHE II SYSTEM
Figures I through III illustrate the receiver operating characteristic (ROC) curves for each of the subgroups. The sensitivity, specificity and predictive value of the APACHE II system at a relative risk (RR). Group Lila showed a sensitivity of 38% and a specificity of 93%, while for Group IIlb a sensitivity of 75% and a specificity of 80% were observed.
The sample sizes were too small (Group Ilia and IlIb) and the estimates too unstable for the construction of ROC curves.
ORGAN SYSTEM FAILURE
It illustrates that mortality associated with failure of each organ varied by group. In the IIb group, cardiovascular and renal failure were associated with higher mortality compared with the same organ failure in the lilac group. Figures 6–10 illustrate the general trend of increased mortality as day 1 organ failure score increased.
The minimum number and duration of organ failure associated with a 100% mortality rate for Groups lila and IIlb was the same.
THE SEQUENTIAL ORGAN FAILURE ASSESSMENT SCORE (SOFA) SCORE
The mean values of the different components of the SOFA score were significantly higher in non-survivors than in survivors (Table 19), in all categories except coagulation failure. The graphs illustrate that in each group there were variable degrees of dysfunction for each of the different types of organ failure. Figures 22-27 illustrate the SOFA scores on a daily basis for the different subgroups and for the entire group.
Scores were consistently higher in nonsurvivors compared to survivors (p<0.001) for each group (Table 15).
Mean SOFA Score
THE OBSTETRIC AND GYNAECOLOGY OUTCOME PREDICTION (OGOP) MODEL
The data from the entire Development Group (consisting of Groups I, II and III) were subjected to multiple logistic regression analysis. Age, temperature, mean arterial pressure, respiratory rate, pH and the Glasgow Coma Scale (GCS) were identified as significant independent predictors of outcome (p<0.05) as documented in Table 27. The estimate was unstable and the biostatistician could not integrate the data into ' a model
A sensitivity of 93%, specificity of 83% and correct classification rate of 85% was observed for the entire validation cohort (n=54).
THE OGOP MODEL COMPARED TO THE APACHE II SYSTEM
THE SOFA SCORE, ORGAN FAILURE SCORE AND THE OGOP MODEL IN RELATION TO THE SYSTEMIC INFLAMMATORY RESPONSE
In the current series, Group III accounted for 16.2% of total obstetric and gynecology admissions. This can be attributed to the older age and higher incidence of sepsis in the gynecology surveys. The mortality in the current setting has increased compared to the 21.3% reported by Platteau (17 of 80 patients) et al in 19974.
The lower mortality in Group I compared to Group II may be attributed to the rapid resolution of disease after delivery of the feto-placental unit in hypertensive inpatients.
THE APACHE II SYSTEM
A striking observation was that the APACHE II scores for all 3 groups (survivors and non-survivors) were similar, however. The APACHE II system was not designed with the obstetrics and gynecology patient in mind35. However, the APACHE II system remains the most common outcome prediction tool in critically-related reports.
To evaluate the outcome, the majority of studies use the APACHE II score (a raw score) instead of the calculated prediction equation (APACHE II system).
RGAN SYSTEM FAILURE
First, one can determine the presence or absence of organ failure based on strictly defined criteria (for the different organs). The number of organ system failures is then added to obtain a score such as the Organ Failure Score79. Furthermore, organ failure was associated with higher mortality (for each of the organ failures) in Group III compared to Groups I and II.
The different organ failure profile was attributed to the difference in the disease profiles of the two cohorts9.
THE SEQUENTIAL ORGAN FAILURE ASSESSMENT SCORE
Karnad et al (2004) from Mumbai also found that the SOFA score was higher in non-survivors than in survivors25. This was surprising as Muckart et al (1997) documented an increase in mortality following a change in SIRS48 response. Afessa et al (2002) observed in their cohort of 74 patients over an eight-year period that a SIRS response occurred in 59% of their patients.
Zakalik et al (2005) showed that 30.5% of their patients were diagnosed with severe sepsis or septic shock.
THE OBSTETRIC AND GYNAECOLOGY OUTCOME PREDICTION MODEL
There are no data related to the development of an outcome prediction model for this population. There is no specific tool for predicting outcome for obstetric and gynecological patients. The OGOP model represents the first step in predicting outcome for this unique population.
It should be emphasized that outcome prediction models are not designed for individual use.
THE OBSTETRICS AND GYNAECOLOGY OUTCOME PREDICTION MODEL COMPARED TO THE ACUTE PHYSIOLOGY AND CHRONIC
THE SEQUENTIAL ORGAN FAILURE ASSESSMENT SCORE, ORGAN SYSTEM FAILURE AND THE OBSTETRICS AND GYNAECOLOGY
It is clearly evident that in the current series, critically ill obstetrics and gynecology patients represent a significant proportion of the intensive care population at this center. The results confirmed that critically ill obstetrics and gynecology hospitals are not a completely homogeneous population and that subgroups should be perceived as separate groups. SIRS has been shown to occur in the majority of critically ill obstetric and gynecological patients.
Although the entity of critical care medicine is nearly 50 years old, "critical care obstetrics and gynecology" is only now being recognized as a discipline with more questions than answers.
RESEARCH AGENDA
This study sought to evaluate existing outcome prediction tools and further develop and validate an outcome prediction model for this unique population. However, it must be recognized that this study has only addressed the tip of the iceberg. Verification of the results regarding organ failure score, SOFA score and SIRS in other critically ill obstetric and gynecological patients.
Using the above outcome prediction tools to "risk stratify" patients for different therapeutic or interventional studies.
RECOMMENDATIONS
This may assist in the development and refinement of management protocols for issues of specific importance in the critically ill obstetric and gynecological patient. Certainly from a South African perspective it is hoped that this study will create an impetus to formalize critical obstetric care in the country. The high utilization of critical care services by obstetric patients in the study raises the question of what constitutes the ideal location for such a service.
The geographical location and the absence of an on-site obstetrician often result in reluctance to transfer prenatal patients.
26.Dao B, Rouamba A, Ouedraogo 0, et al: Transfer of obstetric patients to the intensive care unit: Eighty-two cases in Burkina Faso. Tang LC, Kwok AC, Wong AY, et al: Intensive care in obstetric patients: an eight-year review. In Shoemaker WC, Thompson WL, Holbrook PH et al (eds): Textbook of Critical Care, 2nd edition 1988; Philadelphia, WB Saunders, p.1442.
Gilbert TT, Smulian JC, Martin AA, et al: Obstetric intensive care unit admissions: Outcomes and severity of illness.
APACHE II SCORE (A+B+C BELOW)
Acute Physiology Score
Age Points
Chronic Health Points
ORGAN FAILURE SCORE
SOFA SCORE
Sepsis associated with hypotension, despite adequate fluid resuscitation together with the presence of perfusion abnormalities as listed for severe sepsis. Infection was diagnosed in the context of a positive microbial culture, contamination of the gastrointestinal tract or the presence of overt sepsis. Infection: Microbial phenomenon characterized by an inflammatory response to the presence of microorganisms or the invasion of normally sterile host tissue by those organisms.
Hypotension: a systolic blood pressure of < 90 mmHg or a decrease in blood pressure of > 40 mmHg from baseline, in the absence of other causes of hypotension.
GLASGOW COMA SCALE
