|Year : 2014 | Volume
| Issue : 1 | Page : 29-34
Assessment of performance and utility of mortality prediction models in a single Indian mixed tertiary intensive care unit
Prachee M Sathe, Sharda N Bapat
Department of Critical Care Medicine, Grant Medical Foundation, Ruby Hall Clinic, Pune, Maharashtra, India
|Date of Web Publication||3-Mar-2014|
Sharda N Bapat
E-2/5 Girija Shankar Vihar, Karve Nagar, Pune - 411 052, Maharashtra
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Objectives: To assess the performance and utility of two mortality prediction models viz. Acute Physiology and Chronic Health Evaluation II (APACHE II) and Simplified Acute Physiology Score II (SAPS II) in a single Indian mixed tertiary intensive care unit (ICU). Secondary objectives were bench-marking and setting a base line for research.
Materials and Methods: In this observational cohort, data needed for calculation of both scores were prospectively collected for all consecutive admissions to 28-bedded ICU in the year 2011. After excluding readmissions, discharges within 24 h and age <18 years, the records of 1543 patients were analyzed using appropriate statistical methods.
Results: Both models overpredicted mortality in this cohort [standardized mortality ratio (SMR) 0.88 ± 0.05 and 0.95 ± 0.06 using APACHE II and SAPS II respectively]. Patterns of predicted mortality had strong association with true mortality (R2 = 0.98 for APACHE II and R2 = 0.99 for SAPS II). Both models performed poorly in formal Hosmer-Lemeshow goodness-of-fit testing (Chi-square = 12.8 (P = 0.03) for APACHE II, Chi-square = 26.6 (P = 0.001) for SAPS II) but showed good discrimination (area under receiver operating characteristic curve 0.86 ± 0.013 SE (P < 0.001) and 0.83 ± 0.013 SE (P < 0.001) for APACHE II and SAPS II, respectively). There were wide variations in SMRs calculated for subgroups based on International Classification of Disease, 10 th edition (standard deviation ± 0.27 for APACHE II and 0.30 for SAPS II).
Interpretation and Conclusion: Lack of fit of data to the models and wide variation in SMRs in subgroups put a limitation on utility of these models as tools for assessing quality of care and comparing performances of different units without customization. Considering comparable performance and simplicity of use, efforts should be made to adapt SAPS II.
Keywords: Acute Physiology and Chronic Health Evaluation II, India, mortality prediction, quality of care assessment, Simplified Acute Physiology Score II, standardized mortality ratio
|How to cite this article:|
Sathe PM, Bapat SN. Assessment of performance and utility of mortality prediction models in a single Indian mixed tertiary intensive care unit. Int J Crit Illn Inj Sci 2014;4:29-34
|How to cite this URL:|
Sathe PM, Bapat SN. Assessment of performance and utility of mortality prediction models in a single Indian mixed tertiary intensive care unit. Int J Crit Illn Inj Sci [serial online] 2014 [cited 2023 Mar 21];4:29-34. Available from: https://www.ijciis.org/text.asp?2014/4/1/29/128010
| Introduction|| |
In India, the health care delivery system remains largely uncontrolled with large number of private operators catering to a portion of population. With globalization, the focus of the health sector, at least of the larger set ups, is turning to standardization of protocols and projection of quality of health care. One of the tools of measurement of quality of health care is standardized mortality ratio (SMR)-ratio of observed over predicted mortality. Several models for mortality prediction have been developed to assess severity of illness and predict mortality in intensive care units (ICUs). These models are useful in assessing severity of illness, predicting mortality, standardizing research, and assessing performance of ICUs. Acute Physiology and Chronic Health Evaluation II (APACHE II) and Simplified Acute Physiology Score II (SAPS II) are two of the commonly cited models in India. ,
Several studies have reported varying performance of these models in predicting mortality in a new population and/or different case mix. These variations in performance have been attributed to variations in the population, case mix, quality of care, quality of data, and errors in the application of scores. ,,,,,, It is imperative, therefore, that the models be tested before their application to a different population and/or case mix.
Validation of these scores has been extensively done in Europe, USA, and Australian continent but data from the Indian subcontinent is sparse. Most Indian studies have addressed specific subgroups like respiratory, obstetric, surgical, cancer, and so on. [10-20] To our knowledge, there are no published studies on validation of these models in mixed ICUs in Maharashtra, India.
Our aim was to assess the performance and utility of these two models in predicting mortality in a single tertiary multidisciplinary ICU. The secondary objectives were benchmarking and setting a base line for research and quality improvement.
| Materials and Methods|| |
Study design and data collection
In this observational study, data were collected prospectively for all consecutive admissions to 28-bedded multidisciplinary ICU of a private 550-bedded tertiary hospital in Western Maharashtra, India in the year 2011 (from January 1 to December 31, 2011). This hospital has separate coronary care unit, neurotrauma unit and pediatric ICU. Patients with ICU length of stay of less than 24 h, age less than 18 years and discharged against medical advice were excluded from the final analysis. For patients who were admitted to the ICU more than once during the same hospital admission, data only from the first admission were included. Approval was obtained from the local ethics committee.
ICU resident doctors were given orientation sessions to ensure high quality of collected data and were also given real-time feedback on missing data and logical errors. They were requested to double check whether the tests were obtained during the first 24 h in case of physiological parameters not filled in the forms.
Demographic data, chronic health status (organ failure/immunosuppression), diagnostic category leading to ICU admission according to International Classification of Disease, 10 th edition (ICD-10) were noted on admission on specially designed data collection forms. Worst (most deranged) values of physiological parameters required for calculation of APACHE II and SAPS II and type of admission (medical, elective/emergency surgical) were filled in by ICU registrars after 24 h. Since axillary temperatures were routinely measured in this ICU, those were noted and core temperature was assumed to be one degree higher than axillary temperature for the purpose of calculation of scores. For sedated patients, Glasgow Coma Scale (GCS) before sedation were recorded. Principal diagnostic category leading to admission was recorded as described by Knaus et al.  In case of values not recorded in the form; it was assumed that such values were within normal limits. Data were routinely cross-verified with available online laboratory reports and nursing charts including double checking missing values. Scores were calculated using online calculators for consistency. , Patients were followed-up to hospital discharge for their survival status. Data were entered into a computer program specifically designed for the study.
Student's t-test was used to compare normally distributed continuous variables. Predictive models can be tested in terms of calibration and discrimination. Calibration refers to the ability of the model to accurately predict mortality. To check the degree of association between mortality patterns predicted by the models and true mortality, records were sorted according to increasing predicted probability of death and divided into 10 subgroups (deciles) with approximately same number (154) of cases. Graphs were plotted by plotting average predicted probability of death against true risk in each decile. R 2 statistics were calculated by linear regression.
To test the predictive accuracy of the two models, the Hosmer-Lemeshow goodness-of-fit test was applied. The test statistic is obtained by applying a Chi-square test on a ×2 g contingency table. The contingency table is constructed by cross-classifying the dichotomous dependent variable with a grouping variable (with g groups) in which groups are formed by partitioning the predicted probabilities using the percentiles of the predicted event probability. In the calculation, approximately 10 groups are used (g = 10). 
SMRs were calculated by dividing the observed number of deaths by the number of deaths predicted by each model. SMRs for categories as per ICD-10 classification were calculated for subgroups consisting of at least 100 admissions.
To assess discriminating ability of the models, receiver operating characteristic (ROC) curves were constructed for each model by plotting true positives (sensitivity) against false positives (1-specificity) at several decision cutoffs. Discrimination is the ability of the model to distinguish between true positives (nonsurvivors in this case) and false positives. An important advantage of ROC analysis over traditional sensitivity-specificity analysis is that the area under the ROC curve is independent both of the cut-point criteria chosen and the prevalence of outcome of interest.
To test the stability of the models, bootstrap analysis of SMR with 1000 samples of increasing number of random cases (e.g., 100 random cases, 200 random cases, and so on) was done. Comparison of patterns of true mortality and mortalities predicted by the two models was also done by bootstrap analysis using 1000 samples of 600 random cases each and probability density chart was plotted. 
Data were represented as mean ± standard deviation (SD) and significance level of 0.05 was used wherever applicable. Epiinfo, OpenEpi, Microsoft excel, IBM SPSS, and 'R' were used for data management and analysis. ,,
| Results|| |
A total of 2014 patients were admitted to the ICU in the study period. Patients with ICU stay of less than 24 h (n = 211), age less than 18 years (n = 7), discharged against medical advice (n = 163), incomplete data (n = 9), subsequent readmission to ICU (n = 81) were excluded from the final analysis. Hence, the data of 1543 patients were available for analysis. Total 273 (17.7%) patients died in the hospital, while the mortalities predicted by APACHE II and SAPS II were 311.6 (20.2%) and 288.5 (18.7%), respectively.
[Table 1] details demographic data and clinical categories of patients. Majority of patients (n = 1003, 65%) were male. Mean age of the cohort was 53.9 ± 16.8 years. Mean age (56.5 years) and mean length of ICU stay (6.3) were significantly higher in nonsurvivors than in survivors (mean age = 53.4 years, mean length of stay = 4.7). Maximum number of admissions were medical (n = 1235, 80%) followed by elective surgical (n = 231, 15%) followed by emergency surgical (n = 77, 5%). Mortality was highest in emergency surgical group (24.6%) and lowest in elective surgical group (5.6%). Mortality in medical patients was 19.5%. Total of 589 patients (38.1%) were admitted with chronic organ failure of which 529 patients (34.3%) had single, whereas 60 patients (3.8%) had multiple organ failure. Among patients with chronic organ failure, renal failure was most common (21.3%) followed by cardiovascular (9.7%), respiratory (5.9%), and liver insufficiency (5.1%). Mortality was highest in patients admitted with liver insufficiency (43.8%) followed by patients with respiratory failure (27.2%). Survival in patients with chronic cardiovascular and renal failure was 82.8% and 83.9%, respectively.
[Figure 1] and [Figure 2] give a quick visual assessment of strength of association between true and predicted mortality patterns. APACHE II overpredicted mortality below 0.5 probability of dying while underpredicted mortality above probability of 0.5. SAPS II overpredicted mortality above 0.2 probability. There was very strong association between true and predicted mortality by both models. R 2 for APACHE II and SAPS II were 0.98 and 0.99, respectively.
|Figure 1: Comparison of average risk of death predicted by APACHE II and true risk. X-axis: Deciles of approximately 154 records each in ascending risk of death. APACHE II: Acute Physiology and Chronic Health Evaluation II; SMR: Standardized Mortality Ratio|
Click here to view
|Figure 2: Comparison of average risk of death predicted by SAPS II and true risk X-axis: Deciles of approximately 154 records each in ascending predicted risk of death.SAPS II: Simplified Acute Physiology Score II; SMR: Standardized Mortality Ratio|
Click here to view
Both models performed poorly in Hosmer-Lemeshow goodness of fit test. APACHE II performed slightly better with smaller Chi-square statistics (12.8) and higher significance level (P = 0.03) as against Chi-square of 26.6 (P = 0.001) of SAPS II.
SMRs using APACHE II and SAPS II were 0.88 and 0.95, respectively.
Both models showed good discrimination by ROC Curves. APACHE II model showed a slightly better discrimination [area under the curve (AUC): 0.86 ± 0.013SE, P < 0.001, 95% confidence interval (CI): 0.83-0.88] than SAPS II (AUC: 0.83 ± 0.013 SE, P < 0.001, 95% CI: 0.81-0.86). The difference was not statistically significant (P = 0.1).
In bootstrap analysis with increasing sample size both models demonstrated reasonable stability (SD ≤ 0.05 in SMR) only above the sample size of 600.
[Figure 3] presents comparison of patterns of true mortality and predicted mortalities by both models using bootstrap analysis of 1000 samples of 600 random cases.
|Figure 3: Probability Density Chart. X-axis: No. of deaths per sample of 600 cases. For any range on the X-axis, area under the curve gives probability that any given 600 people sample will have mortality within that range. APACHE II: Acute Physiology and Chronic Health Evaluation II; SAPS II: Simplified Acute Physiology Score II. Note: For the bootstrap analysis, SMR was calculated on 1000 samples of 600 random cases each picked from the original data set. It provides an assessment of underlying distribution and behavior of the models in different random samples|
Click here to view
Mean true mortality was 106.2 ± 7.5. Mean mortality predicted by APACHE II was 121.2 ± 3.9, while SAPS II predicted 112.2 ± 4.0 deaths. SMR by bootstrap analysis were 0.88 ± 0.05 and 0.95 ± 0.06 for APACHE II and SAPS II, respectively. In general, both models seemed to overpredict mortality. SAPS II seemed better but had slightly higher variance.
The most commonly missing physiological parameters were "bilirubin level" (n = 687, 44%) followed by "arterial blood gas" (n = 611, 39.5%) in the first 24 h of ICU admission.
Details of SMRs as per ICD-10 are given in [Table 2]. There were wide variations in SMR among categories (SD 0.27 and 0.30 for APACHE II and SAPS II, respectively). Both models grossly overpredicted mortality in three subgroups viz. "Genitourinary system" (SMR using APACHE II = 0.52, SAPS II = 0.51), "Circulatory system" (APACHE II 0.57, SAPS II 0.68), and "Respiratory system"(APACHE II 0.66, SAPS II 0.7). Both models underpredicted mortality in "Neoplasms" (APACHE II 1.22, SAPS II 1.26) and "Infectious and parasitic diseases"(APACHE II 1.15, SAPS II 1.16). In "Digestive system" group APACHE II overpredicted (SMR 0.86), while SAPS II underpredicted (SMR 1.36) mortality. Mortality predicted by both models in other categories was within 0.1 SD.
|Table 2: Standardized mortality ratio in subgroups based on International Classification of Disease, 10th edition|
Click here to view
| Discussion|| |
In this cohort, patterns of mortality predicted by both APACHE II and SAPS II demonstrated strong association with true mortality. Both models also showed good discriminating ability between survivors and nonsurvivors as shown by the ROC curves.
In boot strap analysis, both models overpredicted mortality. In the formal goodness-of-fit testing, both models performed poorly. There was also significant variation in SMRs calculated for subgroups based on ICD-10. It would be interesting to note that within most of these major subgroups, both models showed good fit by Hosmer-Lemeshow goodness of fit test, for example, "respiratory system" group: APACHE II (Chi-square = 8.256, P = 0.409) and SAPS II (Chi-square = 9.274, P = 0.320). This cohort covered the entire spectrum of diseases in the Knaus' classification except for "Dissecting thoracic/abdominal aneurysm" although in different proportions. Wide variation in SMRs of subgroups is partly explained by stability of the models only above the sample size of 600.
Similar results have been reported in many studies performed on different populations outside of India. Some studies from Europe/North America, Portugal, Greece, Saudi Arabia, South England, and Thailand reported acceptable discrimination but poor calibration. ,,,,,, The Greek  and the Saudi Arabian  study, however, reported underprediction of mortality, while in this cohort both models overestimated mortality. One study from Mumbai, India  reported underestimation of mortality and poor calibration of APACHE II in cancer patients. Two studies from North India reported underestimation, poor calibration as well as poor discrimination in respiratory ICUs. , The variations in the performance of the models were commonly attributed to variations in population, case mix, quality of care, quality of data, and errors in application of scores. Changing treatment modalities and other clinical and nonclinical factors not accounted for by the models may also cause the results to vary. ,,,,,,,,
During development of APACHE II, measurement of all 12 physiologic variables was mandatory, while SAPS II assumed missing variables to be within normal limits. In these data sets, the most commonly missing variables were bilirubin level and arterial blood gas. It is reasonable to believe that these tests were not obtained in the first 24 h of ICU admission because the treating physician expected the values to be within normal limits. Utmost care was taken during the study to ensure high quality of data. Being a single center study, however, some amount of bias due to differences in case mix, ICU policy, definitions, lead-time, and quality of care might have possibly occurred.
Lack of fit in formal testing and wide variation in SMR among different subgroups put limitations on the use of these models as tools for quantifying quality of care or comparison of performance of different units with different case mix. Moreover, the models do not provide information about whether a particular patient will die or survive. , Utility of these models in an individual patient is, therefore, limited. Nevertheless, the models can be useful in assessing efficacy of quality improvement measures or medical intervention in a single unit assuming confounding factors (such as case mix, etc.) remain the same.
Considering comparable performance of both models in predicting mortality, SAPS II appears more suitable for this set up due to its simplicity of use. SAPS II requires lesser number of parameters. These parameters are also more readily available, for example, ratio of partial pressure and fraction of inspired oxygen for SAPS II as against alveolar-arterial oxygen gradient for APACHE II. Subjectivity is reduced while calculating SAPS II as against APACHE II because there is no need to choose a single primary diagnosis.  Since unique probability of death is linked to each unique SAPS II score, it can be arrived at without a statistical calculator. In case of APACHE II, probability of death is adjusted according to primary diagnosis and the APACHE II score itself is not a good indicator of severity of illness. With multicenter data collection, correction factors can be derived based on ICD-10 for SAPS II to make it suitable for local use and current treatment modalities.
Considering limited resources, especially access to computers, the older, simpler models like APACHE II and SAPS II are still relevant in developing countries like India.  Both the models under consideration have been developed by regression analysis. Other methods can be tried to develop a suitable model, for example, decision tree. The most important predictors of mortality in the decision tree drawn from this cohort were bicarbonate and bilirubin levels, hematocrit, heart rate, and need for mechanical ventilation in the first 24 h of ICU admission. Prospective data collection of such parameters from multiple centers, therefore, holds promise of developing a model suitable for Indian ICUs.
| Conclusion|| |
Lack of fit of data to the models and wide variation in SMRs in subgroups based on ICD-10 put a limitation on utility of these models as tools for assessing quality of care and comparing performances of different units without customization. Automation may facilitate development and application of more accurate models in the future. Till that time, efforts should be made to customize and apply SAPS II to aid quality improvement and research.
| Acknowledgments|| |
We are thankful to all the ICU consultants, registrars, and nursing staff who worked in the ICU during the year 2011. The study would not have been possible without their active participation in collection of data. Dr. Shilpa Kulkarni, Dr. Rahul Tambade, Dr. Sharad Yadav, Dr. Manisha Pawar, Dr. Satish Virwani, Sister Vinarasi Selvam, Sister Geeta Atre, and Mrs. Shilpa Jadhav deserve special mention. We thank Dr. Sanjeev B. Sarmukaddam, MSc, DBS, MPS, PhD for statistical inputs and statistical review of this article. We are indebted to Dr. Aniruddha Pant, PhD for his valuable statistical inputs. Last but not the least; we are thankful to the hospital administration for allowing us to conduct this study.
| References|| |
|1.||Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: A severity of disease classification system. Crit Care Med 1985;13:818-29. |
|2.||Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American Multicenter study. JAMA 1993;270:2957-63. |
|3.||Castella X, Artigas A, Bion J, Kari A. A comparison of severity of illness scoring systems for intensive care unit patients: Results of a multicenter, multinational study. The European/North American Severity Study Group. Crit Care Med 1995;23:1327-35. |
|4.||Moreno R, Morais P. Outcome prediction in intensive care: Results of a prospective, multicentre, Portuguese study. Intensive Care Med 1997;23:177-86. |
|5.||Patel PA, Grant BJ. Application of mortality prediction systems to individual intensive care units. Intensive Care Med 1999;25:977-82. |
|6.||Katsaragakis S, Papadimitropoulos K, Antonakis P, Strergiopoulos S, Konstadoulakis MM, Androulakis G. Comparison of Acute Physiology and Chronic Health Evaluation II (APACHE II) and Simplified Acute Physiology Score II (SAPS II) scoring systems in a single Greek intensive care unit. Crit Care Med 2000;28:426-32. |
|7.||Arabi Y, Haddad S, Goraj R, Al-Shimemeri A, Al-Malik S. Assessment of performance of four mortality prediction systems in a Saudi Arabian intensive care unit. Crit Care 2002;6:166-74. |
|8.||Beck DH, Smith GB, Pappachan JV, Millar B. External validation of the SAPS II, APACHE II and APACHE III prognostic models in South England: A multicentre study. Intensive Care Med 2003;29:249-56. |
|9.||Khwannimit B, Geater A. A comparison of APACHE II and SAPS II scoring systems in predicting hospital mortality in Thai adult intensive care units. J Med Assoc Thai 2007;90:643-52. |
|10.||Gupta R, Arora VK. Performance evaluation of APACHE II score for an Indian patient with respiratory problems. Indian J Med Res 2004;119:273-82. |
|11.||Nimgaonkar A, Karnad DR, Sudarshan S, Ohno-Machado L, Kohane I. Prediction of mortality in an Indian intensive care unit. Comparison between APACHE II and artificial neural networks. Intensive Care Med 2004;30:248-53. |
|12.||Tempe A, Wadhwa L, Gupta S, Bansal S, Satyanarayana L. Prediction of mortality and morbidity by simplified acute physiology score II in obstetric intensive care unit admissions. Indian J Med Sci 2007;61:179-85. |
|13.||Sam KG, Kondabolu K, Pati D, Kamath A, Pradeep Kumar G, Rao PG. Poisoning severity score, APACHE II and GCS: Effective clinical indices for estimating severity and predicting outcome of acute organophosphorus and carbamate poisoning. J Forensic Leg Med 2009;16:239-47. |
|14.||Singh VK, Gupta R, Mittal S, Sharma SC, Lakshmi B, Ray S, et al. Evaluation of APACHE II Scoring Systems to predict mortality in a Multidisciplinary Intensive Care Unit. In: 7 th ITACCS, 2003, New Delhi, India. Available from: http://openmed.nic.in/3096/[Last accessed 2012 Jun 10]. |
|15.||Chawla R, Sansarwal S, Kansal S, Jhamb T, Singh PP. A comparative study of predictive capability of critical illness scores in Indian patients. Chest 2007;132:P547a. |
|16.||Divatia J, Priya V, Ranganathan P, Chidrawar S. Evaluation of APACHE II and the ICU cancer mortality model in an Indian Cancer Hospital ICU. Crit Care 2006;10(Suppl 1):P404. |
|17.||Barreto SG, Rodrigues J. Comparison of APACHE II and Imrie scoring systems in predicting the severity of acute pancreatitis. World J Emerg Surg 2007;2:33. |
|18.||Kulkarni SV, Naik AS, Subramanian N Jr. APACHE II scoring system in perforative peritonitis. Am J Surg 2007;194:549-52. |
|19.||Aggarwal AN, Sarkar P, Gupta D, Jindal SK. Performance of standard severity scoring systems for outcome prediction in patients admitted to a respiratory intensive care unit in North India. Respirology 2006;11:196-204. |
|20.||Juneja D, Singh O, Nasa P, Dang R. Comparison of newer scoring systems with the conventional scoring systems in general intensive care population. Minerva Anestesiol 2012;78:194-200. |
|21.||Online APACHE II calculator. Available from: http://www.opus12.org/APACHE_II_Calculator.html/[Last accessed on 2012 May 24]. |
|22.||Online SAPS II calculator. Available from: http://www.opus12.org/SAPS_II.html/[Last accessed on 2012 May 24]. |
|23.||Hosmer DW, Lemeshow S. Applied Logistic Regression. 2 nd ed. New York: John Wiley and Sons Ltd; 2000. |
|24.||Efron B. Bootstrap methods: Another look at the jackknife. Ann Stat 1979;7:1-26. |
|25.||Epi Info™ 3.4.3-software developed at the Centers for Disease Control and Prevention (CDC). Available from: http://www.cdc.gov/epiinfo/[Last accessed on 2012 Mar 13]. |
|26.||Dean AG, Sullivan KM, Soe MM. OpenEpi: Open Source Epidemiologic Statistics for Public Health, Version 2.3.1. Available from: http://www.openepi.com/OE2.3/Menu/OpenEpiMenu.htm [Last accessed on 2012 Mar 13]. |
|27.||The R project for Statistical Computing. Available from: http://www.r-project.org/[Last accessed on 2012 May 25]. |
|28.||Piacentini E, Ferrer C. Scoring prognostic system: To predict or not to predict. Minerva Anestesiol 2012;78:149-80. |
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2]
|This article has been cited by|
||Modelling kidney outcomes based on MELD eras - impact of MELD score in renal endpoints after liver transplantation
| ||Paulo Ricardo Gessolo Lins, Roberto Camargo Narciso, Leonardo Rolim Ferraz, Virgilio Gonçalves Pereira, Ben-Hur Ferraz-Neto, Marcio Dias De Almeida, Bento Fortunato Cardoso Dos Santos, Oscar Fernando Pavão Dos Santos, Júlio Cesar Martins Monte, Marcelino Souza Durão Júnior, Marcelo Costa Batista |
| ||BMC Nephrology. 2022; 23(1) |
|[Pubmed] | [DOI]|
||Comparison of mNUTRIC-S2 and mNUTRIC scores to assess nutritional risk and predict intensive care unit mortality
| ||So Jeong Kim, Hong Yeul Lee, Sun Mi Choi, Sang-Min Lee, Jinwoo Lee |
| ||Acute and Critical Care. 2022; 37(4): 618 |
|[Pubmed] | [DOI]|
||Clinical outcome and comparison of burn injury scoring systems in burn patient in Indonesia
| ||Risa Herlianita,Edi Purwanto,Indri Wahyuningsih,Indah Dwi Pratiwi |
| ||African Journal of Emergency Medicine. 2021; 11(3): 331 |
|[Pubmed] | [DOI]|
||Pediatric Simplified Acute Physiology Score II: Establishment of a New, Repeatable Pediatric Mortality Risk Assessment Score
| ||Stefan Irschik, Jelena Veljkovic, Johann Golej, Gerald Schlager, Jennifer B. Brandt, Christoph Krall, Michael Hermon |
| ||Frontiers in Pediatrics. 2021; 9 |
|[Pubmed] | [DOI]|
||EVALUATION OF APACHE- IV & SAPS- II SCORING SYSTEMS AND CALCULATION OF STANDARDISED MORTALITY RATE IN SEVERE SEPSIS AND SEPTIC SHOCK PATIENTS- A PROSPECTIVE OBSERVATIONAL STUDY
| ||Siddhartha Chakraborty,Sarbari Swaika,Rajat Choudhuri,Suchismita Mallick |
| ||Journal of Evolution of Medical and Dental Sciences. 2019; 8(12): 843 |
|[Pubmed] | [DOI]|
||Performance of critical care prognostic scoring systems in low and middle-income countries: a systematic review
| ||Rashan Haniffa,Ilhaam Isaam,A. Pubudu De Silva,Arjen M. Dondorp,Nicolette F. De Keizer |
| ||Critical Care. 2018; 22(1) |
|[Pubmed] | [DOI]|
||Cost-Utility in Medical Intensive Care Patients. Rationalizing Ongoing Care and Timing of Discharge from Intensive Care
| ||Kurien Thomas,John Victor Peter,Jony Christina,Anna Revathi Jagadish,Amala Rajan,Prabha Lionel,Lakshmanan Jeyaseelan,Bijesh Yadav,George John,Kishore Pichamuthu,Binila Chacko,Priscilla Pari,Thilagavathi Murugesan,Kavitha Rajendran,Anu John,Sowmya Sathyendra,Ramya Iyyadurai,Sudha Jasmine,Rajiv Karthik,Alice Mathuram,Samuel George Hansdak,Kundavaram Paul P. Abhilash,Shuba Kumar,K. R. John,Thambu David Sudarsanam |
| ||Annals of the American Thoracic Society. 2015; 12(7): 1058 |
|[Pubmed] | [DOI]|