Premorbid Functional Status as a Predictor of 1-year Mortality and Functional Status in Intensive Care Patients Aged 80 Years or Older

Pietiläinen L, Hästbacka J, Bäcklund M et al.

Intensive Care Medicine. 2018; 44: 1221-9


Aim of Study

To explore to what extent premorbid functional status predicts 1-year mortality in very old (> 80 years) ICU patients.

Stated Hypothesis: A poor premorbid functional status is associated with a poor 1-year outcome.

Design and Location

Prospective observational cohort study


Cohort: All patients over 80 admitted to 25 Finnish ICUs over a 1-year period (May 2012 – April 2013)

Data gathered:

  • Age, gender, length of stay
  • Type of ICU admission: elective/emergency surgery/medical
  • Diagnosis
  • Pre-morbid functional status
  • Severity of illness (SAPS II score/SOFA score)
  • Therapeutic intensity (TISS score)
  • Assessment of premorbid functional status
  • Accommodation type
  • Performance status
  • Activities of daily living – Physical activities (getting out of bed, moving indoors, dressing, climbing stairs, walking 400m)

Categorised patients into “good” or “poor” functional status:

  • “Good functional status” definition – Independent in ADLs and can climb stairs
  • “Poor functional status” definition – Dependent for 1 or more ADLs; or can’t climb stairs

Created a “functional status score”

Score out of 5 – 1 point for each of the physical activities

Information obtained by interview of patient/family during ICU admission

1-year functional status assessed using questionnaire

Primary Outcomes

  • Hospital Mortality
  • 1-year mortality post-ICU
  • Functional status 1-year post-ICU

Secondary Outcomes

  • Treatment intensity on ICU
  • Orders to restrict treatment on ICU


  • Chi-square – categorical variables
  • t-test – continuous variables
  • Mann-whitney – nonparametric continuous variables
  • Univariate regression analysis to assess association of baseline variable with hospital and 1-year mortality
  • Multivariate regression used significant variables from univariate analysis to assess independent association with:

1. Hospital mortality

2. 1-year mortality

Created 2 models to predict probability of hospital and 1-year mortality

First model: Using age, gender, admission type and SAPS II score

Second model: Same as above, but with premorbid functional status data

Compared the models using the area under the receiver operating characteristic curve (AUROC)


1827 (11.1%) patients over 80 were admitted to the ICUs studied over the 1 year period

In patients over 80 years old: 43.3% had poor functional status

Poor premorbid functional state increased risk of in hospital death (OR 1.50, 95% CI 1.07-2.10) and 1-year mortality (OR 2.18, 95% CI 1.67 – 2.85)

Hospital mortality (p<0.001)

  • Good functional status: 13.7%
  • Poor functional status: 23.5%

1-year mortality (p<0.001)

  • Good functional status: 25.5%
  • Poor functional status: 47.1%

Poor functional status increased risk of 1-year mortality in:

  • Medical patients (OR 1.82, 95% CI 1.30 – 2.54)
  • Surgical patients (OR 3.55, 95% CI 2.31 – 5.45)

Predictive models

Adding functional status to the models improved prediction of 1 year mortality (AUROC 0.789 vs 0.772; p=0.002)

But it did not improve prediction of hospital mortality (AUROC 0.833 vs 0.830; p = 0.169)

Orders to restrict treatment intensity

  • 3% of elective surgical admissions
  • 25.4% of emergency surgical admissions
  • 33% of medical admissions
    • Poor functional status – 32.3%
    • Good functional status – 13.5%
    • Difference was significant (p<0.001)

Hospital mortality was 55.6% in those with treatment restriction vs 10.8 for those without (p<0.001)

Treatment intensity

Mean daily TISS score (p<0.001)

  • good functional status = 32.0
  • poor functional status = 29.2

Good function patients were treated more intensively

Function in 1-year survivors

Complete data available for 61.1% of patients, partial data for 81.6%

84.2% lived at home at 1 year vs 88.3% prior to admission

Functional status score

  • Score 4/5 or more in 84.4% at baseline
  • Score 4/5 or more in 79.6% at 1 year

77.8% of 1 year survivors scored the same or better at 1 year compared to baseline

  • 84.9% of elective surgery patients
  • 79.3% of emergency surgery patients
  • 70.0% of medical patients


  • Poor premorbid functional status is associated with increased 1-year mortality in patients over the age of 80
  • Poor functional status doubles the odds of death at 1 year
  • This association was stronger in surgical patients (OR 3.5)
  • Most patients that survived to 1 year returned to their baseline functional status

The association of poor premorbid function and poor outcomes was in agreement with other similar studies. However 1 year survival was 50% in other studies (38.2% in this study), and only 50% returned to premorbid function (77.8% in this study).

They attributed the difference in outcomes to the fact that in the other study, elective surgical patients and those admitted for less than 24h were excluded, whereas they were included in this study.

Outcome prediction tools for ICU could be improved by incorporating markers of physiological age, not just chronological age.

Stated Limitations from the Study

High number of elective surgical patients included – 26.5%

Study is prone to selection bias – many elderly patients would not have been referred or admitted to ICU. The study population represents patients who were thought likely to do well in ICU.

Decisions to limit treatment intensity would likely have affected outcomes – more frail patients would had lower treatment intensity on ICU, and this may account for some of their worst outcomes.

Discussion from Journal Club Meeting (?Change of Practice)

This study is unlikely to be generalisable to our ICU patients as the study cohort appear less sick that our patients – we anticipate we would probably see higher mortalities.

The median length of stay was just 1.48 days in patients over 80 – probably different to our LOS.

The study shows that good outcomes are achievable in well selected patients – one difficulty comes in trying to set the bar of what is a good function and what is poor function

The “good function” definition in the study set quite a high standard – the majority of over 80s would not reach this.

Anecdotally ICU admissions are becoming older, and it is becoming more acceptable to admit older patients, especially if they have good premorbid function.

This study does emphasise the need of a good assessment of functional status when considering an ICU admission.

Summary by Dr J Durrant. Journal Club Meeting 30 August 2018.

Comments are closed.

Blog at

Up ↑

%d bloggers like this: