BRIEF REPORT Cumulative Prognostic Score Predicting Mortality in Patients Older Than 80 Years Admitted to the ICU Dylan W. de Lange, MD, PhD,* Sylvia Brinkman, PhD,† Hans Flaatten, MD, PhD,‡§ Ariane Boumendil, PhD,¶ Alessandro Morandi, MD, MPH,k** Finn H. Andersen, MD, PhD,††‡‡ Antonio Artigas, MD, PhD,§§ Guido Bertolini, MD,¶¶ Maurizio Cecconi, MD,kk Steffen Christensen, MD, PhD,*** Loredana Faraldi, MD,††† Jesper Fjølner, MD,*** Christian Jung, MD,‡‡‡ Brian Marsh, MD,§§§ Rui Moreno, MD,¶¶¶ Sandra Oeyen, MD, PhD,kkk Christina Agvald Öhman, MD, PhD,**** Bernardo Bollen Pinto, MD,†††† Anne Marie G.A. de Smet, MD, PhD,‡‡‡‡ Ivo W. Soliman, MD, PhD,* Wojciech Szczeklik, MD, PhD,§§§§ Andreas Valentin, MD,¶¶¶¶ Ximena Watson, MD,kk Tilemachos Zafeiridis, MD,kkkk and Bertrand Guidet, MD,¶*****††††† On behalf of the VIP1 Study Group# MEASUREMENTS: Overall, 24 variables available during OBJECTIVES: To develop a scoring system model that pre- ICU admission were included as potential predictive variables. dicts mortality within 30 days of admission of patients older Multivariable logistic regression was used to identify independent than 80 years admitted to intensive care units (ICUs). predictors of 30-day mortality. Model sensitivity, specificity, and DESIGN: Prospective cohort study. accuracy were evaluated with receiver operating characteristic SETTING: A total of 306 ICUs from 24 European countries. curves. PARTICIPANTS: Older adults admitted to European ICUs RESULTS: The 30-day-mortality was 1562 (41.9%). In multi- (N = 3730; median age = 84 years [interquartile range = 81-87 variable analysis, these variables were selected as independent y]; 51.8% male). predictors of mortality: age, sex, ICU admission diagnosis, From the *Department of Intensive Care Medicine, University Medical Center, University Utrecht, Utrecht, The Netherlands; †Department of Medical Informatics, Amsterdam Public Health Research Institute, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands; ‡Department of Clinical Medicine, University of Bergen, Bergen, Norway; §Department of Anaesthesia and Intensive Care, Haukeland University Hospital, Bergen, Norway; ¶Assistance Publique - Hôpitaux de Paris, Hôpital Saint-Antoine, Service de Réanimation Médicale, Paris, France; kDepartment of Rehabilitation, Hospital Ancelle di Cremona, Cremona, Italy; **Geriatric Research Group, Brescia, Italy; ††Department of Anaesthesia and Intensive Care, Ålesund Hospital, Ålesund, Norway; ‡‡Department of Circulation and Medical Imaging, NTNU, Trondheim, Norway; §§Department of Intensive Care Medecine, CIBER Enfermedades Respiratorias, Corporacion Sanitaria Universitaria Parc Tauli, Autonomous University of Barcelona, Sabadell, Spain; ¶¶Laboratorio di Epidemiologia Clinica, Centro di Coordinamento GiViTI Dipartimento di Salute Pubblica, IRCCS - Istituto di Ricerche Farmacologiche “Mario Negri”, Ranica (Bergamo), Italy; kkSt George’s University Hospital, London, United Kingdom; ***Department of Anaesthesia and Intensive Care Medicine, Aarhus University Hospital, Denmark; †††ASST Grande Ospedale Metropolitano Niguarda, Milan, Italy; ‡‡‡Department of Cardiology, Pulmonology and Angiology, University Hospital, Düsseldorf, Germany; §§§Mater Misericordiae University Hospital, Dublin, Ireland; ¶¶¶Unidade de Cuidados Intensivos Neurocriticos e Trauma, Hospital de São José, Centro Hospitalar Universitário de Lisboa Central Nova Medical School, Lisbon, Portugal; kkkDepartment of Intensive Care 1K12IC, Ghent University Hospital, Ghent, Belgium; ****Karolinska University Hospital, Stockholm, Sweden; ††††Geneva University Hospitals, Geneva, Switzerland; ‡‡‡‡Department of Critical Care, University Medical Center Groningen, University Groningen, Groningen, The Netherlands; §§§§Intensive Care and Perioperative Medicine Division, Jagiellonian University Medical College, Kraków, Poland; ¶¶¶¶Kardinal Schwarzenberg Hospital, Schwarzach, Austria; kkkk Intensive Care Unit, General Hospital of Larissa Tsakalof Larissa, Larissa, Greece; *****Sorbonne Universités, UPMC Univ Paris 06, UMR_S 1136, Institut Pierre Louis d’Epidémiologie et de Santé Publique, Paris, France; and the †††††ICU, hospital Saint Antoine, APHP, Paris, France. Address correspondence to Dylan W. de Lange, MD, PhD, Department of Intensive Care Medicine, University Medical Center, University Utrecht, Utrecht, The Netherlands. E-mail: d.w.delange@umcutrecht.nl # List of contributors: see Acknowledgments. [Corrections added May 20, 2019, after first online publication. The author name “Bernadro Bollen Pinto” was corrected as “Bernardo Bollen Pinto”] DOI: 10.1111/jgs.15888 JAGS 67:1263–1267, 2019 © 2019 The Authors Journal of the American Geriatrics Society published by Wiley Periodicals, Inc. on behalf of The American Geriatrics Society. 0002-8614/19/$15.00 This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. 1264 DE LANGE ET AL. JUNE 2019–VOL. 67, NO. 6 JAGS Clinical Frailty Scale, Sequential Organ Failure Score, invasive In short, all patients older than 80 years who were acutely mechanical ventilation, and renal replacement therapy. The admitted to participating ICUs in 24 European countries were discrimination, accuracy, and calibration of the model were included in this study.4,7 The mortality 30 days after ICU good: the area under the curve for a score of 10 or higher was admission was the primary outcome. .80, and the Brier score was .18. At a cut point of 10 or higher Based on variables present at admission (eg, age, sex, rea- (75% of all patients), the model predicts 30-day mortality in son for ICU admission, the abbreviated Clinical Frailty Scale)12 91.1% of all patients who die. or treatments provided during ICU stay (eg, worst Sequential CONCLUSION: A predictive model of cumulative events Organ Failure Assessment [SOFA], (non-)invasive mechanical predicts 30-day mortality in patients older than 80 years ventilation, use of vasoactive drugs, renal replacement therapy admitted to ICUs. Future studies should include other poten- [RRT]), a multivariable logistic regression model was con- tial predictor variables including functional status, presence structed. The discrimination, accuracy, and calibration of the of advance care plans, and assessment of each patient’s model were assessed13,14 before a simple bedside model was decision-making capacity. J Am Geriatr Soc 67:1263– constructed. This simple bedside model was based on the beta 1267, 2019. of each predictor in the model as described previously.15 The total number of points assigned to each patient is Key words: critical care; prognosis; older adults; predict; called the cumulative prognostic score (CPS) and correlates model with 30-day mortality. The performance of the bedside model (the sensitivity and specificity at several cutoff points of the CPS) is assessed. RESULTS Participants M ore than 10% of the patients admitted to the intensive care unit (ICU) are 80 years and older.1 This propor- tion of “very old intensive care patients” (VIPs) is estimated to In total, 306 ICUs from 24 countries participated and included 5187 VIPs. Of these patients, 4252 were acutely admitted. Follow-up at 30 days with complete data on all variables was increase up to 36% in 2025.2,3 However, once VIPs have been obtained in 88% (3730/4252). The median number of patients admitted to the ICU for an acute medical reason, their overall recruited per country and per ICU was 104 and 12, respec- 30-day mortality is high. The frailer patients in particular have tively. Demographics of the patients included in the final ana- a high mortality.4 Despite careful patient selection before ICU lyses are presented in Table 1. admission, more than half of these VIPs will die or will experi- ence major functional deterioration in the 6 months following Table 1. Demographics of Included Nonelective Patients their admission.5,6 As a result of our current uncertainty in predicting which No. of patients % VIPs could potentially benefit from ICU treatment, we often offer them an “ICU trial.” This means admitting VIPs to an Total 3730 100 ICU, offering them life-sustaining treatment for a period of ICU mortality 1065 28.6 2 to 3 days, and then reevaluating if they show any improve- 30-day mortality 1562 41.9 Male 1932 51.8 ment. If patients deteriorate, limitations in life-sustaining ther- Age, y, median (25th percentile-75th 84 (81-87) - apies would be required.7 For ICU physicians, this ICU trial percentile) postpones the difficult ICU admission triage decision by a few CFS, median (25th percentile-75th 4 (3-6) - days, and by then, some patients have improved. However, percentile) some will still receive life-sustaining therapy, and a decision to SOFA score, median (25th 7 (4-11) - continue treatment should be discussed with the patient or his percentile-75th percentile) or her legal representatives.8 Inevitably, during such shared Intubation and mechanical ventilation 1924 51.6 decision-making processes, the question of chances of survival Vasoactive drugs 2155 57.8 NIV 984 26.4 emerges. Most intensivists estimate a patient’s chances of out- RRT 405 10.9 come on experience and on preferences. Current severity scor- Reason for ICU admission ing systems are not tailored for VIPs,1 and proposed models Respiratory failure 900 24.1 for VIP are not precise enough.9,10 Circulatory failure 537 14.4 We hypothesize that a cumulative prognostic score can pre- Respiratory and circulatory failure 448 12 dict 30-day mortality and thus support physicians and relatives Sepsis 480 12.9 with the decision to continue care or start a new treatment. Multi-trauma without head injury 55 1.5 Multi-trauma with head injury 57 1.5 Head injury 110 2.9 Intoxication 13 0.3 METHODS Nontrauma 293 7.9 We present only a very brief discussion of methods here. A Emergency surgery 379 10.2 more elaborate description of our methods, adhering to all Other 458 12.3 Transparent Reporting of a Multivariable Prediction Model Abbreviations: CSF, Clinical Frailty Scale; ICU, intensive care unit; NIV, for Individual Prognosis or Diagnosis (TRIPOD) statements,11 noninvasive ventilation; RRT, renal replacement therapy; SOFA, Sequential is provided in Supplementary Materials S1. Organ Failure Assessment. JAGS JUNE 2019–VOL. 67, NO. 6 PREDICTING MORTALITY IN OLDER ICU PATIENTS 1265 Multivariable Logistic Regression Model the SOFA score were the two most important factors associated with 30-day mortality. Among the reasons for ICU admission, No multicollinearity was found between the variables avail- multi-trauma with head injury and nontrauma central nervous able for model development, and there was no interaction system causes were associated with a high 30-day mortality. between age and the Clinical Frailty Scale. During model devel- The sensitivity and specificity for various thresholds is opment, the following variables were selected using the Lasso based on the assigned number of points in the prediction model procedure: age, sex, reason for ICU admission categorized into (ie, CPS of the patients) (Figure 1). When all patients with a 11 options (Supplementary Materials S1), vasoactive drugs, 30-day CPS higher than 10 points are selected (corresponding Clinical Frailty Scale, SOFA score, intubation with mechanical to 76.0% of all patients), 91.8% of all patients who died dur- ventilation, and RRT. The continuous variable SOFA score ing the 30 days after ICU admission are captured (sensitivity). was included as restricted cubic spline. The final regression Of these 76.0% selected patients, 50.6% will die during the model is presented in Table 2, where the SOFA score for sim- 30 days after ICU admission (positive predicted value). plification is presented as categorical data instead of splines. Supplementary Materials S5 lists the characteristics of The overall discrimination and accuracy of the final regres- the patients with a 30-day CPS higher than 10 points and sion model that predicts mortality within 30 days after ICU the patients with a 30-day mortality score higher, lower, or admission was good; the area under the curve (AUC) was .80 equal to 10 points. (.80-.81). The Brier score was .18 (.18-.18). In the different diag- nostic subgroups, the performance of the model was also good (Supplementary Materials S2). The calibration belt of the sepsis DISCUSSION patients showed more uncertainty for the low- and high-risk patients; for the emergency surgery patients, the model is not This study demonstrates that a cumulative event model corre- able to predict high mortality risks. However, the calibration lates with outcome in acutely admitted ICU patients older than belt of the total population and all other diagnostic subgroups 80 years. This model is based on reason for ICU admission, showed no abnormalities (Supplementary Materials S3). age, sex, frailty, SOFA score (all available at admission), and the organ support during ICU stay: invasive mechanical venti- lation and RRT. Although this model discriminates between “Bedside” Model those dying and surviving 30 days after ICU admission rather accurately (AUC = .80), it remains difficult to predict which The “bedside” model, based on a point system (Supplementary patients are going to die with 100% certainty. Even if we use a Materials S4), shows parameters that influence mortality high cutoff (eg, >19 points), these patients still have a 25% 30 days after ICU admission and the weight assigned to these chance of survival in the first 30 days after ICU admission. variables. In theory, the minimum and maximum scores a patient can obtain are 0 and 26 points, respectively. The mini- mum and maximum scores obtained in the study population 1.0 CPS > 6 were 1 and 26, respectively. The reason for ICU admission and CPS > 8 CPS > 10 Table 2. Final Multivariable Regression Model 0.8 CPS > 12 Covariate Odds ratio 30-day mortality CPS > 14 Age 1.05 (1.03-1.07) 0.6 sensitivity Sex, male 1.26 (1.09-1.47) Mechanical ventilation 2.07 (1.72-2.50) CPS > 16 RRT 1.70 (1.33-2.18) CFS 1.19 (1.14-1.25) 0.4 SOFA score <4 Reference CPS > 18 ≥4 and <7 1.80 (1.37-2.37) 0.2 ≥7 and <10 2.70 (2.02-3.60) ≥10 4.40 (3.26-5.92) CPS > 20 Reason for ICU admission Respiratory failure 3.58 (.73-17.48) 0 Circulatory failure 3.74 (.76-18.39) Respiratory and circulatory failure 4.55 (.92-22.43) 0.8 0 0.2 0.4 0.6 1.0 Sepsis 3.20 (.65-15.77) Multi-trauma without head injury 3.62 (.66-19.77) 1 - specificity Multi-trauma with head injury 5.06 (.94-27.23) Isolated head Injury 4.30 (.84-22.09) Figure 1. Sensitivity and specificity of the simple bedside model Intoxication Reference for 30-day mortality. When a patient has >10 points on the Nontrauma CNS causes 5.45 (1.10-27.01) Cumulative Prognostic Score (CPS), that patient has a 50.6% Emergency surgery 1.87 (.38-9.27) positive predictive value (PPV) on 30-day mortality and a nega- Other 2.0 (.40-9.88) tive predictive value (NPV) of 85.8%. The other values are as Abbreviations: CFS, Clinical Frailty Scale; CNS, central nervous system; follows: at >12 points, a PPV of 55.9% and an NPV of 81.1%; ICU, intensive care unit; RRT, renal replacement therapy; SOFA, Sequential at >16 points, a PPV of 67.6% and an NPV of 69.2%; and at Organ Failure Assessment. >20 points, a PPV of 77.0% and an NPV of 60.5%. 1266 DE LANGE ET AL. JUNE 2019–VOL. 67, NO. 6 JAGS Previous research identified variables associated with a this means that 25% of those patients do survive the first poor outcome in this very old patient group: age,1,4 sex,4,9 30 days. A longer follow-up will undoubtedly show there is mechanical ventilation,16 circulatory shock,17 acute kidney substantial additional mortality in these very old patients. For injury,18 and the presence of comorbidities.19 However, up to this reason a new study (the so-called Very old Intensive care now, only a few studies have tried to build a model from these Patient study 2 [VIP2]) will look at mortality at day 180 after variables to predict outcome after ICU admission. Ball et al ICU admission. Third, we combined data from admission with developed a model based on age, serum creatinine, Glasgow treatment data from the subsequent treatment days. This Coma Scale, and pH.10 The data in that study were collected simplifies and strengthens prognostication but prevents the during the first day of ICU admission and did not include the model being used for triage purposes before admission. Fourth, response to active treatment for which the ICU trial is we only looked at survival, although we know that many intended. Heyland et al made a model to predict functional patients do not fully recover, and many older adult patients pri- outcome 1 year after ICU admission.9 Although that model oritize “quality of life” above “quantity of life.” Future research had a good performance (AUC = .81), the selected variables should focus on functional outcome and quality of life. And, were less easily collected at the bedside because they were finally, we did not include data on nutritional status, functional derived from more complex scores, for example, the Acute status (activities of daily living and instrumental activities of Physiology, Age, Chronic Health Evaluation II score and the daily living), cognitive impairment, dementia, delirium, depres- Charlson Comorbidity Index. sion, and comorbidities (eg, active cancer). These variables will It is very difficult to predict outcome in acutely ill patients undoubtedly influence outcome. Our CPS can become more older than 80 years, and many are admitted to the ICU for an comprehensive if such variables are included. ICU trial. However, such an admission should be reevaluated in In conclusion, this relatively simple cumulative events a shared decision-making conference with the patient or, when model can help assess the chances of 30-day mortality in the patients lack decision-making capacity, with the family or very old patients who were acutely admitted to the ICU. designated surrogate. Our model can be used to estimate, in a This model may assist both intensivists and family members more objective way than just subjective clinical intuition, what during the shared decision-making process to estimate the the chances of 30-day mortality will be. We believe this informa- otherwise subjective chances of survival. tion can help both the intensivist and the family members to put treatment into perspective.20 Indeed, although almost all intensi- vists claim that they value the opinions of surrogates (eg, rela- ACKNOWLEDGMENTS tives, family, legal representatives, and caregivers),21 these family opinions on ICU admissions are, in reality, rarely sought.8,22 This study could not be performed without the help of One of the reasons why this is omitted is the uncertainty that many contributors. These VIP1 study contributors are listed intensivists feel during the prognostication of patients.22-25 in Supplementary Materials S6. And yet, a poor prognosis is one of the most important Conflict of Interest: The authors have declared no con- reasons to implement limitations in life-sustaining thera- flicts of interest for this article. pies.26,27 Indeed, family members reported that what was most Author Contributions: All authors contributed to important to them was that the “patient should be comfortable obtaining ethical clearance for their countries, including patients, and suffer as little as possible.” The belief that “life should be and corrected the final draft of the manuscript. Statistical ana- preserved at all costs” was their least important value consid- lyses: Brinkman, Soliman, Bertolini, and Boumendil. Writing the ered in making treatment decisions. A substantial proportion first draft: de Lange. of the caregivers (24%) reported that “comfort care without Sponsor’s Role: There was no sponsor for this article. life support” was their most preferred treatment goal, but The Society of Intensive Care Medicine (ESICM) endorsed 14% were “unsure about their treatment preferences.”28 Care- the study and was awarded a scientific prize. givers and surrogates who had a shared decision-making con- ference with their intensivist experienced less decisional conflicts than family members who had not talked to a physi- REFERENCES cian.28 Strikingly, despite family members only prioritizing “comfort measures,” 83.7% of these patients still received life- 1. Flaatten H, Artigas A, de Lange DW, et al. The status of ICM research and a future agenda for very elderly patients in the ICU. Intensive Care Med. 2017; sustaining treatments, and approximately 20% received such 43:1319-1328. treatments for more than a week.28 Our bedside model for 2. Laake JH, Dybwik K, Flaatten HK, Fonneland IL, Kvåle R, Strand K. Impact 30-day mortality can assist both intensivists and family mem- of the post-World War II generation on intensive care needs in Norway. Acta bers in the decision-making process to continue or cease fur- Anaesthesiol Scand. 2010;54(4):479-484. ther treatment. 3. Angus D. Admitting elderly patients to the intensive care unit—is it the right The strong feature of this study is the international inclu- decision? JAMA. 2017;318:1443-1444. 4. Flaatten H, De Lange DW, Morandi A, et al. The impact of frailty on ICU sion of more than 3700 very old patients and the high follow-up and 30-day mortality and the level of care in very elderly patients (≥ rate at 30 days after ICU admission. However, some limitations 80 years). Intensive Care Med. 2017;43:4940-4948. need to be discussed. First, we did not collect data on the timing 5. Boumendil A, Angus DC, Guitonneau AL, et al. Variability of intensive care of discussions to withhold or withdraw life-sustaining treat- admission decisions for the very elderly. PLoS One. 2012;7:e34387. ment. For example, some patients had advanced directives not 6. Guidet B, Leblanc G, Simon T, et al. Effect of systematic intensive care unit triage on long-term mortality among critically ill elderly patients in France: a to instigate certain treatments that might have influenced out- randomized clinical trial. JAMA. 2017;318:1450-1459. come. A second limitation of the model is that it does not predict 7. Guidet B, Flaatten H, Boumendil A, et al. Withholding or withdrawing of life patients who will have a definite poor outcome (100% mortal- sustaining therapy in very elderly patients (≥ 80 years) admitted to the inten- ity). The predicted mortality at more than 19 points is 75%, but sive care unit. Intensive Care Med. 2018;44:1027-1038. JAGS JUNE 2019–VOL. 67, NO. 6 PREDICTING MORTALITY IN OLDER ICU PATIENTS 1267 8. Le Guen J, Boumendil A, Guidet B, Corvol A, Saint-Jean O, Somme D. Are 23. Teno JM, Fisher E, Hamel MB, et al. Decision-making and outcomes of pro- elderly patients’ opinions sought before admission to an intensive care unit? longed ICU stays in seriously ill patients. J Am Geriatr Soc. 2000;48(S1): Results of the ICE-CUB study. Age Ageing. 2016;45:303-309. S70-S74. 9. Heyland DK, Stelfox HT, Garland A, et al. Predicting performance status 24. Sinuff T, Adhikari NK, Cook DJ, et al. Mortality predictions in the intensive 1 year after critical illness in patients 80 years or older: development of a care unit: comparing physicians with scoring systems. Crit Care Med. 2006; multivariable clinical prediction model. Crit Care Med. 2016;44:1718-1726. 34:878-885. 10. Ball IM, Bagshaw SM, Burns KE, et al. A clinical prediction tool for hospital 25. Soliman IW, Cremer OL, de Lange DW, et al. The ability of intensive care mortality in critically ill elderly patients. J Crit Care. 2016;35:206-212. unit physicians to estimate long-term prognosis in survivors of critical illness. 11. Collins GS, Reitsma JB, Altman DG, Moons KG. Transparent reporting of a J Crit Care. 2018;43:148-155. multivariable prediction model for individual prognosis or diagnosis 26. Hoel H, Skjaker SA, Haagensen R, Stavem K. Decisions to withhold or with- (TRIPOD): the TRIPOD statement. BMJ. 2015;350:g7594. draw life-sustaining treatment in a Norwegian intensive care unit. Acta 12. Rockwood K, Song X, MacKnight C, et al. A global clinical measure of fit- Anaesthesiol Scand. 2014;58(3):329-336. ness and frailty in elderly people. CMAJ. 2005;173:489-495. 27. Keenan SP, Busche KD, Chen LM, McCarthy L, Inman KJ, Sibbald WJ. A retro- 13. Steyerberg EW, Bleeker SE, Moll H, et al. Internal and external validation of spective review of a large cohort of patients undergoing the process of withholding predictive models: a simulation study of bias and precision in small samples. or withdrawal of life support. Crit Care Med. 1997;25:1324-1331. J Clin Epidemiol. 2003;56:441-447. 28. Heyland DK, Dodek P, Mehta S, et al. Admission of the very elderly to the 14. Finazzi S, Poole D, Luciani D, Cogo PE, Bertolini G. Calibration belt for intensive care unit: family members’ perspectives on clinical decision-making quality-of-care assessment based on dichotomous outcomes. PLoS One. from a multicenter cohort study. Palliat Med. 2015;29:324-335. 2011;6:e16110. 15. Brandenburg R, Brinkman S, de Keizer NF, Kesecioglu J, Meulenbelt J, de Lange DW. The need for ICU admission in intoxicated patients: a prediction SUPPORTING INFORMATION model. Clin Toxicol (Phila). 2017;55:4-11. 16. Schortgen F, Follin A, Piccari L, et al. Results of noninvasive ventilation in Additional Supporting Information may be found in the very old patients. Ann Intensive Care. 2012;21:5. online version of this article. 17. Biston P, Aldecoa C, Devriendt J, et al. Outcome of elderly patients with cir- culatory failure. Intensive Care Med. 2014;40:50-56. Supplementary Appendix S1. Supplementary Materials. 18. Chronopoulos A, Rosner MH, Cruz DN, Ronco C, et al. Acute kidney injury in elderly intensive care patients: a review. Intensive Care Med. 2010; Supplementary Materials S1. Expanded methods. 36:1454-1464. Supplementary Materials S2. Performance of multivari- 19. Flaatten H, Oeyen S, de Lange DW. Predicting outcomes in very old ICU ate logistic regression model. patients: time to focus on the past? Intensive Care Med. 2018;44:1344-1345. Supplementary Materials S3. GiViTi calibration belt 20. Guidet B, de Lange DW, Flaatten H. Should this elderly patient be admitted to the ICU? Intensive Care Med. 2018;44:1926-1928. for the developed prediction model. 21. Guidet B, De Lange DW, Christensen S, et al. Attitudes of physicians Supplementary Materials S4. Prediction model based towards the care of critically ill elderly patients—a European survey. Acta on a point system. Anaesthesiol Scand. 2018;62:207-219. Supplementary Materials S5. Demographics for patients 22. Heyland DK, Barwich D, Pichora D, et al. Failure to engage hospitalized elderly patients and their families in advance care planning. JAMA Intern with ≤10 and >10 Cumulative Prognostic Score. Med. 2013;173:778-787. Supplementary Materials S6. VIP1 study contributors.