Development and validation of a multivariable risk score for prolonged length of stay in the surgical intensive care unit.
Aged
Anti-Infective Agents
/ administration & dosage
Cohort Studies
Comorbidity
Critical Illness
Enteral Nutrition
Female
Humans
Intensive Care Units
Length of Stay
Logistic Models
Male
Middle Aged
Parenteral Nutrition
Predictive Value of Tests
Psychomotor Agitation
Respiration, Artificial
Risk Assessment
Sensitivity and Specificity
Serum Albumin
/ analysis
Switzerland
Journal
Swiss medical weekly
ISSN: 1424-3997
Titre abrégé: Swiss Med Wkly
Pays: Switzerland
ID NLM: 100970884
Informations de publication
Date de publication:
09 09 2019
09 09 2019
Historique:
entrez:
1
10
2019
pubmed:
1
10
2019
medline:
30
7
2020
Statut:
epublish
Résumé
Chronically critical illness is highly relevant in intensive care units, but the definitions in literature vary greatly. The timely detection of prolonged intensive care unit length of stay could support care planning for chronically critical ill patients. To develop and validate a risk score for predicting prolonged length of stay in the surgical intensive care unit. This single centre cohort study formed part of a nursing-led project in one surgical intensive care unit. We examined the performance of seven predefined predictive factors of prolonged (>20 days) intensive care unit length of stay in adults on the seventh day of stay in intensive care to develop (n = 304) and validate (n = 101) a risk score. Candidate variables (Charlson Comorbidity Index, Simplified Acute Physiology Score II, minimum plasma albumin, need for anti-infective drugs, time of mechanical ventilation, main feeding method and score on the Sedation-Agitation Scale) were analysed using multiple logistical regression analysis. Our risk score assigned different points to the following conditions: Charlson Comorbidity Index >2, minimum albumin <20 g/l between days 1 and 7, mechanical ventilation >14 hr on day 7 and the need for parenteral nutrition on day 7. For a validation data set (n = 101), the area under the receiver operating characteristic curve was 0.89 (95% confidence interval 0.770.87). At a cut-off value of 100 points, the degree of sensitivity was 88%, the specificity 75%, the positive predictive value 53%, the negative predictive value 95%, and the model fit R2 0.40. Our model allowed the timely detection of prolonged intensive care unit length of stay with four candidate predictive factors. The timely identification of patients with prolonged intensive care unit length of stay is possible and could influence the person-centred prevention of chronically critical illness and adequate resource allocation.
(Trial registration no DRKS 00017073)
Sections du résumé
BACKGROUND
Chronically critical illness is highly relevant in intensive care units, but the definitions in literature vary greatly. The timely detection of prolonged intensive care unit length of stay could support care planning for chronically critical ill patients.
AIM
To develop and validate a risk score for predicting prolonged length of stay in the surgical intensive care unit.
METHODS
This single centre cohort study formed part of a nursing-led project in one surgical intensive care unit. We examined the performance of seven predefined predictive factors of prolonged (>20 days) intensive care unit length of stay in adults on the seventh day of stay in intensive care to develop (n = 304) and validate (n = 101) a risk score. Candidate variables (Charlson Comorbidity Index, Simplified Acute Physiology Score II, minimum plasma albumin, need for anti-infective drugs, time of mechanical ventilation, main feeding method and score on the Sedation-Agitation Scale) were analysed using multiple logistical regression analysis.
RESULTS
Our risk score assigned different points to the following conditions: Charlson Comorbidity Index >2, minimum albumin <20 g/l between days 1 and 7, mechanical ventilation >14 hr on day 7 and the need for parenteral nutrition on day 7. For a validation data set (n = 101), the area under the receiver operating characteristic curve was 0.89 (95% confidence interval 0.770.87). At a cut-off value of 100 points, the degree of sensitivity was 88%, the specificity 75%, the positive predictive value 53%, the negative predictive value 95%, and the model fit R2 0.40.
CONCLUSIONS
Our model allowed the timely detection of prolonged intensive care unit length of stay with four candidate predictive factors. The timely identification of patients with prolonged intensive care unit length of stay is possible and could influence the person-centred prevention of chronically critical illness and adequate resource allocation.
(Trial registration no DRKS 00017073)
Identifiants
pubmed: 31568554
doi: 10.4414/smw.2019.20122
pii: Swiss Med Wkly. 2019;149:w20122
doi:
pii:
Substances chimiques
Anti-Infective Agents
0
Serum Albumin
0
Types de publication
Journal Article
Validation Study
Langues
eng
Sous-ensembles de citation
IM