Table 2

Existing scales or screening tools to predict risk of death and their domains

CrisTAL Inclusion criteria and comments
Year/AuthorScale name and scoringComponentsReadily availableClinical judgementValue judgementSufficient for prognosis
1949 Clark53 and modified by Péus55Karnofsky Performance Score (KPS)Quality of life across the spectrum of health from 0=normal to 100=terminal
Administered face to face or by phone; involves value judgements; poor inter-rater reliability; does not cater for preadmission functional status
1981 Addington-Hall54Spitzer Quality of Life IndexFive dimensions of quality of life: activity, daily living, general health, support of family and friends, and outlook
High clinician's acceptability as it takes 1 min to administer but has not proven accurate in predicting death within 6 months in individuals
1985 Knaus64
1992 McMahon65
1995 DelBufalo66
2006 Zimmerman67
2013 Sharif47
Acute Physiology and Chronic Health Evaluation
The point score is calculated from 11 ICU physiological measurements + age:
Temperature (rectal), Mean arterial pressure, pH arterial, Heart rate,
Respiratory rate, Sodium (serum)
Potassium (serum), Creatinine,
Haematocrit, White cell count,
Glasgow Coma Scale
Used to predict hospital mortality in ICU. Unsuitable for admissions unit
1987 Charlson68
1988 Pompei69
Charlson Comorbidity Index (CCI)Includes 19 categories of comorbidity and ach condition is assigned with a severity score of 1, 2, 3 or 6 depending on the risk of dying associated with this condition. Higher scores indicate greater comorbidity (patients with a score >5 have a 100% risk of dying at 1 year)
Complex calculation. Many adaptations attempted to improve predictive accuracy of 10-year mortality. Some capability for predicting short-term mortality. Does not cater for functional status or immediate risk of death, that is, physiological risk
1993 Le Gal70SAPS IIAge, heart rate, Systolic BP,
Temp, GCS, CPAP Y/N, PaO2, FIO2, urine
Output, BUN, K, Bicarbonate, WCC, Chronic diseases, medical/surgical admission
Validated in 12 countries and the results were encouraging even in the absence of a primary diagnosis but high reliance on sophisticated testing not routinely conducted outside ICU
1996 Anderson 71
2008 Virik and Glare72
Palliative Performance Scale (PPS)Assessment of observed ambulation, activity, evidence of disease, self-care, intake, level of physical activity and level of consciousness.
Score 0=death
Score 70=bed bound
Score 100=full health and ambulation
Validated in Canada and Australia. However, the original intention of developers was not to use PPS for prognostication.71 Subjective observations do not contribute to standardisation of assessment
Recent validation showed a PPS of 10 was associated with over 90% in-hospital mortality whereas a PPS of 70 was associated with 0% deaths
1998 Elixhauser 73
2009 Van Walraven 18
2013 Austin74
Elixhauser comorbidity IndexRelies on administrative databases to retrieve diagnostic items for 30 coexisting disease groups and applies weights to severity
Data items which are incomplete and not detailed enough to provide a clinically precise time of diagnosis. Complex to calculate, not too accurate on predicting mortality, more useful for researchers than clinicians at predicting length of stay
2001 Subbe75MEWS
Scores of 5 or more were associated with increased risk of death
Good predictive ability for risk of death in busy acute services
2004 Glare76
2012 Maltoni77

Palliative Prognostic Score (PaP)
Karnofsky Performance Status plus
White cell counts
Clinician's weighted prediction of survival
Validated in Italy, Australia and England. Good association with short-term mortality but predictive value of tool affected by less experienced doctors

2013 Kuo-H 79
Rapid Emergency Medicine Score (REMS)Blood pressure, respiratory rate, Glasgow Coma Scale, peripheral oxygen saturation,
Effective in predicting risk of death in hospital in conjunction with other clinical parameters including surgical treatment within 24 h. However, it has little relevance for elderly patients with chronic disease seeking hospital care
2005 Rockwood58CSHA Clinical Frailty ScaleScores of 1 (very fit) to 7 (severely frail) assigned by physician on the basis of qualitative definitions incorporating physical functioning and presence of comorbidities
Each 1-category increment of the frailty scale increased the risk of mortality. Largely subjective or reliant on clinical and value judgements
2006 Paterson78SEWSRespiratory rate, oxygen saturation, temperature, blood pressure, heart rate and conscious level
Score correlated both with in-hospital mortality and length of stay
2006 Kellet 46
2012 Kellett80
Simple Clinical Score
Weighted cores derived from 16 independent variables: age, pulse, systolic blood pressure, respiratory rate, temperature, oxygen saturation, breathless on presentation, abnormal ECG, diabetes, coma, altered mental status, new stroke, unable to stand unaided, nursing home resident, daytime bed rest prior to current illness
Most items available and some easily obtainable. Successfully validated for 30-day and 1-year prediction but limited generalisability for many chronic conditions
2008 Groarke81EWS

Pulse, systolic blood pressure, respiratory rate, oxygen saturation and neurological status. Increases in score indicate risk of complication or death
Used to identify physiological deterioration in patients on admission. Good predictor of transfer to high dependency care
2008 Stone82Palliative Prognostic Index (PPI)PPS +
Oral intake
Dyspnoea at rest
Developed for Japanese patients with advanced cancer in hospices and validated in Ireland in hospitals, hospices and the home. Prediction of positive predictive value of 86% for survival of less than three weeks PPV of 91% for survival of less than six weeks. Not generalisable to other conditions or longer term mortality predictions
2008 Glare83Clinical Prediction of Survival (CPS)Combines clinical experience with performance assessment
More accurate closer to death, overestimates survival if patient–doctor relationship is stronger
2010 Prytherch 84ViEWSApplies paper-based EWS score to a Vital Signs database and uses known relationship between deteriorated physiological measures and clinical outcomes such as in-hospital mortality with 24 h of the observations
It appears to predict immediate mortality well but vital signs databases are not widely available in many health systems
2012 & 2013
Rothman44 56
Rothman IndexNurse-led assessment of whether minimum standards for each of 8 body systems, food intake, pain, risk of falls and 1 psychosocial (adequate support system)criteria are met or not met
Based on well-defined minimum standards as documented by nurses in electronic medical records in one USA hospital; independent of expert opinion; data not routinely available in other hospitals
  • APACHE, Acute Physiology and Chronic Disease Evaluation; CSHA, Canadian Study of Health and Aging; EWS, early warning score; ICU, intensive care unit; MEWS, modified early warning score; SAPS II, Simplified Acute Physiology Score II; SEWS, standardised early warning scoring system; ViEWS, VitalPAC™ early warning score.