Electronic Supplementary Mterial (ESM)

Case-mix-adjusted length of stay AND Mortality IN twenty-three finnish ICUs

Electronic supplementary material (ESM)

Table E1. Distribution of disease categories according to APACHE III (8) in the development sample (n= 37 718) and co-efficients for each disease category in models predicting ICU length of stay (ICU LOS) and hospital death.

Expected ICU LOS and risk of hospital death can be calculated as follows:

Expected ICU LOS (days) =

10 ** (-0.385 - 0.001 * age + 0.005 * SAPS II - 0.0000235*SAPSII2+ 0.021 * single highest TISS score + 0.495 * ICU death

- 0.013 * (SAPSII * ICU death) +0.063*readmission -0.052*other ICUs in the hospital + coefficient for disease category)

Risk of hospital death =

1 / (1 + e ** (5.146 - 0.070 * SAPSII - 0.009 * age- coefficient for disease category)) =

In the equations above, “ICU death” and “other ICUs in the hospital” are dichotomous variables (0,1) and “age” is years. When calculating risk of hospital death, readmissionwere excluded.

Example 1: A 70 year-old patient is admitted to ICU from the emergency department (=nonoperative) with bacterial pneumonia. He/ she receives 30 SAPS II points. The maximum TISS points during the ICU stay are 50. She is discharged from the ICU alive this first ICU admission. There is also another ICU in the same hospital.

Expected ICU LOS (days) =

10 ** (-0.385 - 0.001 * age + 0.005 * SAPS II - 0.0000235*SAPSII2+ 0.021 * single highest TISS score + 0.495 * ICU death

- 0.013 * (SAPSII * ICU death) +0.063*readmission -0.052* other ICUs in the hospital + coefficient for disease category)

= 10 ** (-0.385 - 0.001 * 70 + 0.005 * 30 - 0.0000235 * 302+ 0.021 * 50 + 0.495 * 0 - 0.013 * 30 * 0 + 0.063 * 0-0.052*1+ 0.180)

=10 ** 0.85185 = 7.11

Risk of hospital death =

1 / (1 + e ** (5.146 - 0.070 * SAPSII - 0.009 * age- coefficient for disease category)) =

1 / (1 + e ** (5.146 - 0.070 * 30 - 0.009 * 70- 0.36)) =

1 / (1 + e ** 2.056) =

1 / (1 + 2.71828 ** 2.056) = 0.11

The expected ICU LOS is 7.11 days and the risk of hospital death is 11%.

Example 2: A 40 year-old patient is admitted to ICU from the operating theatre (=operative) with head trauma. This is her/his second ICU admission during this hospital stay. He/ she receives 45 SAPS II points. The maximum TISS points during the ICU stay are 55. She dies in the ICU during this admission. There are no other ICUs in the same hospital.

Expected ICU LOS (days)

= 10 ** (-0.385 - 0.001 * age + 0.005 * SAPS II - 0.0000235*SAPSII2+ 0.021 * single highest TISS score + 0.495 * ICU death

- 0.013 * (SAPSII * ICU death) +0.063*readmission -0.052* other ICUs in the hospital + coefficient for disease category)

= 10 ** (-0.385 - 0.001 * 40 + 0.005 * 45 - 0.0000235*452+ 0.021 * 55 + 0.495 * 1 - 0.013 * (45 * 1) +0.063*1 -0.052* 0)

=10**0.88 = 7.59

Risk of hospital death =

1 / (1 + e ** (5.146 - 0.070 * SAPSII - 0.009 * age- coefficient for disease category)) =

1 / (1 + e ** (5.146 - 0.070 * 45 - 0.009 * 40) =

1 / (1 + e ** 1.636) =

1 / (1 + 2.71828 ** 1.636) = 0.16

The expected ICU LOS is 7.59 days and the risk of hospital death is 16%.

Table E2.Demographic data of the ICUs (validation sample, n=25 586). *n=21 502

ESM Figure.ICU LOS (Days)

Figure E1. Distribution of length of stay in the ICU (ICU LOS) in the development sample (n=37718). Mean 3.3; standard deviation 5.4; median 1.6; maximum 140