الفهرس 
Review of LiteratureOnly 14 pages are availabe for public view
 |  | from | 124 |  |  |
| | | from | 124 | | |
Abstract
T
he global population is aging, leading to a growing demand for specialized medical care for older adults. The Geriatric Intensive Care Unit (GICU) is a critical component of geriatric medicine, providing comprehensive care for patients with critical medical conditions. Elderly patients face unique challenges due to physiological changes, increased vulnerability, and complex medical needs. Despite the use of high-cost sophisticated devices, ICU mortality rates remain high.
In this study, the predictive performance of two commonly used severity of illness scoring systems, Mortality Probability Model (MPM0-III) and Simplified Acute Physiology Score (SAPS 3), was evaluated among 106 elderly Egyptian patients.
The study aimed to compare the performance of these models in predicting patients’ outcomes and to select a suitable model for use in GICU.
The results of the present study showed the following:
The actual patient mortality rate observed during the study period was 49.05%, with the mean age of patients being 74.8+8.34 years. Comparison between survivors and non-survivors in all demographic data revealed no significant differences.
The number of non-survivors who had pulmonary diseases and previously ICU admitted is higher when compared to survivors. The number of survivors and non-survivors with DM, cardiac diseases, renal diseases, and dementia were not significantly different.
Comparison between survivors and non-survivors regarding baseline laboratory findings revealed highly significant differences between both groups in CRP, PT, and INR (P<0.001). Serum albumin levels showed highly significant decrease in non-survivors (P<0.001). Also, a significant increase was detected in BUN levels in the group of non-survivors when compared to survivors’ group (P<0.011).
Septic shock had a highly significant difference between survivors and non-survivors (X2= 7.497, P=0.006).
Length of ICU stay was significantly longer in non-survivors (P=0.044), while length of post-ICU stay was significantly longer in survivors (P<0.001). Mortality in the ICU was recorded in 42 patients out of the 52 non-survivors, whereas 40 patients out of the 54 survivors were discharged to the ward.
Comparison between survivors and non-survivors as regards MPM0-III and SAPS 3 predictive mortality rates revealed highly statistically significant increase in the group of non-survivors (P<0.001).
Using SMR, the observed number of deaths in the study is not statistically significantly different than the predicted number of deaths by the two scores. MPM24 has the best calibration followed by MPM72, SAPS 3 and then MPM48. Logistic regression analysis was done to estimate odds of mortality using the different scores, they were highly significant for all (P<0.001). MPM24, MPM48, MPM72 and SAPS 3 scores showed fair to good discriminative power as their area under the curve (AUC) were 0.769, 0.813, 0.828, 0.778 respectively, (P<0.001 for all). MPM72 showed the best AUC.
CONCLUSION
F
rom this study we could conclude that the two severity of illness scoring systems; MPM0-III and SAPS 3 performed well and can be used to predict mortality in critically ill elderly patients, and that they both had accepted degree of discrimination and calibration in this cohort of patients.
The findings, however, emphasize the significance of calibrating and verifying severity of disease scoring systems before utilizing them in clinical practice, as well as testing their accuracy and reliability in diverse groups.
Because the model with more variables (SAPS 3) was not connected with improved discriminatory performance, a scoring system with fewer factors may be preferable.
RECOMMENDATIONS
• Prognostic models’ research in the future should concentrate on external validation, addressing the performance measures relevant for their intended use, and on models’ clinical credibility, including the incorporation of factors specific for the elderly population. External validation is essential for achieving definite evaluation of the prognostic models.
• Future research is advised to focus on the questions of whether physicians, patients, and their families value receiving more personalized prognostic information, and whether providing this information influences the process and outcome of their decision making.
• Periodic updating is highly recommended and essential for maintaining the accuracy of these predictive models because ICU populations are always changing; new diagnostic, therapeutic, and prognostic techniques become available; and the performance of predictive scoring models tends to deteriorate over time, resulting in an overestimation of mortality.
• Patient-centered scoring systems are recommended to be created because the scores were intended to be utilized in mixed groups of ICU patients and their accuracy in some patient subgroups may be questioned.
• It is proposed to estimate more precisely using local or regional equations that are created from a more uniform case mix.
• We may recommend that all patients be evaluated using a general outcome prediction model on admission or during the first 24 hours, followed by repeated organ failure scores during their ICU stay. When these approaches are integrated, they may provide a more precise picture of the degree and prognosis of the disease, which may benefit both the physician and the management in charge of resource allocation and performance evaluation.