Skip to main content

Table 2 Risk stratification tools

From: A systematic review of tools for predicting severe adverse events following patient discharge from intensive care units

Author,
Tool, Country
Year
Tool Components/
Variables
(Weighting/Points)
Prediction Outcome (Follow-up) Tool Development
(# patients)
Tool Validation
(# patients/#ICU)
Sensitivity
(%)
Specificity
(%)
LR+ LR- AUROC
(95% CI)
Gajic
'SWIFT Score'
USA, 2008
Source of ICU admission
(Other than ED: 8 pt)
ICU length of stay
(2 to 10 d: 1 pt, >10 d: 4 pt)
Last measured PaO2/FiO2 ratio
(150 to 399: 5 pt, 100 to 149: 10 pt, <100: 13 pt)
GCS at ICU discharge
(11 to 14: 6 pt, 8 to 10: 14 pt, <8: 24 pt)
Last PaCO2 (>45 mmHg: 5 pt)
Readmission
(7 days)
Multivariate
(1,131)
Internal
(783/1)
External
(708/1)
56
27
83
87
3.09
2.13
0.56
0.84
0.75
(0.70 to 0.80)
0.70
(0.64 to 0.76)
Frost
Australia, 2010
Age (years: 0 to 8 pt)
Male (2 pt)
Elective admission (12 pt)
Admission source
(ED: 9 pt, Other hospital: 10 pt, Ward: 15 pt)
APACHE II score (0 to 20 pt)
ICU length of stay >7 days (17 pt)
After hours discharge (4 pt)
Renal failure (10 pt)
Readmission
(Hospital Discharge)
Multivariate
(14,952)
Internal
(14,952/1)
n/a n/a n/a n/a 0.66
(n/a)
Reini
Sweden, 2012
Pulse rate (0 to 3 pt)
Respiratory rate (0 to 3 pt)
Systolic blood pressure (0 to 3 pt)
Level of consciousness (0 to 3 pt)
Temperature (0 to 2 pt)
Readmission
(72 hours)
Existing Score External
(518)
15δ 85δ 1.01δ 0.99δ OR 0.98
(0.69 to 1.37)
Badawi
USA, 2012
23 variablesλ Readmission
(48 hours)
Multivariate
(469,967)
Internal
(234,976/219)
6 to 96ψ 19 to 99ψ 1.19 to 5.72ψ 0.19 to 0.95ψ 0.71
(0.71 to 0.71)
  26 variables§ Mortality
(48 hours)
Multivariate
(469,967)
Internal
(234,976/219)
47 to 82ψ 87 to 99ψ 6.44 to 55ψ 0.20 to 0.53ψ 0.92
(0.92 to 0.92)
Daly
United Kingdom,
2002
β coefficients
Age per year (0.0532)
Chronic Health Points (0.2501)
ICU length of stay per day (0.0447)
Acute Physiology points (0.1556)
Cardiothoracic surgery (-2.104)
Constant (-4.5821)
Mortality
(Hospital Discharge)
Multivariate
(5,475)
Internal
(1,136/1)
External
(7,313/19)
74 71 2.55 0.37 0.80
(0.79 to 0.81)θ
Fernandez
'Sabadell Score' Spain,
2006 and 2010
Subjective intensive care physician scoring:
Good Prognosis (0)
Poor long term prognosis, >6 months (1)
Poor short term prognosis, <6 months (2)
Death expected within hospitalization (3)
Mortality
(Hospital Discharge)
Existing Score
Modified
Internal
(1,521/1)
External
(3,587/31)
23 to 87
26 to 85
79 to 99
71 to 99
4.14 to 23
2.93 to 26
0.16 to 0.78
0.21 to 0.75
0.88
(0.84 to 0.93)
0.84
(0.81 to 0.87)
Ouanes
'MIR'
France,
2012
β coefficients
SAPS II (admission) (0.017)
Central venous catheter (0.74)
SIRS (max) (0.61)
SOFA (discharge) (0.19)
Discharge at night (0.92)
Constant (-5.59)
Readmission or mortality
(7 days)
Multivariate
(3,462)
Internal
(3,462/4)
50 to 96α 19 to 82α 1.19 to 2.78α 0.21 to 0.61α 0.74
(0.68 to 0.79)
  1. APACHE, Acute Physiology and Chronic Health Evaluation; AUROC, area under the receiver operating characteristic curves; ED, emergency department; GCS, Glasgow Coma Scale; ICU, Intensive Care unit; LR, likelihood ratio; MIR, Minimizing ICU Readmission; N/A, Not Available; SAPS, Simplified Acute Physiology Score; SIRS, Systemic Inflammatory Response Syndrome; SOFA, Sequential Organ Failure Assessment score; SWIFT, Stability and Workload Index for Transfer; USA, United States of America.
  2. ‡ Resampling using bootstrap techniques.
  3. δ Calculated using the Modified Early Warning Score on admission to ICU of <6 vs. > 6.
  4. ¶ Odds ratio for readmission to ICU within 72 hours of ICU discharge reported for each one point increase in the Modified Early Warning Score at the time of ICU discharge. The receiver operating characteristic curve not reported.
  5. λ Readmission model variables: Admission characteristics (age), Elective surgery, ICU type, Admission diagnosis category, Admit source, ICU visit number, Body mass index, ICU interventions (number of lactate values in 24 hours, ICU length of stay), Last day labs, (serum bicarbonate, white blood cell count, serum creatinine, hemoglobin), Last day physiology (heart rate, respiratory rate, diastolic blood pressure, systolic blood pressure, percent oxygen, most recent Glasgow coma scale score).
  6. ψ Range of sensitivities and specificities reported for four different thresholds of predicted readmission or mortality following ICU discharge ranging from 1% to 10%.
  7. § Mortality model variables: Admission characteristics (age, body mass index), Operative diagnosis (elective surgery), ICU interventions (ICU length of stay, ventilation status), Last day labs (serum lactate, serum creatinine, white blood cell count, serum glucose), Last day physiology (diastolic blood pressure, heart rate, mean arterial pressure, respiratory rate, percent oxygen saturation, most recent Glasgow coma scale score).
  8. † β coefficients reported from multivariable regression model.
  9. θ AUROC reported for combined data for internal and external validation cohorts.
  10. α Range of sensitivities, specificities and likelihood ratios reported for four different thresholds of the MIR score.