Discriminatory Capacity of Prediction Model Delirium (PREDELIRIC) in Adult Patients Hospitalized in the Intensive Care Unit University Hospital "Los Comuneros" Bucaramanga-Colombia

Monday, 9 November 2015

Claudia C. Torres Contreras, RN, MSc
Programa de Enfermeria, Universidad de Santander - UDES, Bucaramanga, Colombia
Astrid Nathalia Páez Esteban, Magister, RN
Health Faculty - nursing program, University of Santander, Bucaramanga, Colombia
Mayerli Katherine Rincón Romero, RN
Nursing Program, UNIVERSIDAD DE SANTANDER UDES- COLCIENCIAS, Bucaramanga, Colombia

The prevalence of delirium in intensive care units (ICU) is 20-80% Carrillo & Sosa, (2010). Delirium is diagnosed in only 66-84% of patients with this complication. Oumet & others (2007)

In addition, delirium is a predictor of re-intubation and increased hospital stay which has a significant impact on costs of hospitals. The development of delirium in intensive care units represents a major impact on patient management therefore directly affects and influences the hospital cost, stay and mortality. Martínez & others (2004)

The most frequent delirium in intensive care units psychiatric disorder is relevant to study the same in order to predict its onset to avoid fatal outcomes in critically ill patients. Ely & others (2004)

As regards the diagnosis of delirium several methods for diagnosis, from simple, practical and rapid methods, to more complex forms. Martinez et al, (2004).
Thus, the method most frequently used in critically ill patients, including patients on mechanical ventilation who can not communicate verbally is the CAM-ICU English confussion assesment method for the UCI, which has high sensitivity and specificity, excellent reliability and validity to identify delirium when used by nurses and ICU physicians, requires 2-3 minutes to complete. (Ely & others, 2004). This applies to patients with Sedation-Agitation Scale Richmond (English RASS Richmond Agitation Sedation Scale) -3 to +4. Sessler & others (2002).

But all the above scales allow assess the presence of delirium but none predicts the presentation. Thus, M van den Boogaard and colleagues decided to build and validate a model to predict delirium in adult patients in intensive care units. Boogaard & others (2012). Since this model of vital importance as it allows establishing preventive measures, reduce the incidence, severity and duration. Further development of this model will reduce risks of adverse effects of pharmacological prophylaxis.

This model was developed through a systematic review of known risk factors for the development of delirium. After calculating the incidence of each of the risk factors for the development of Delirium, found that 25 were the most prevalent, then eight factors were excluded due to prevalence below 10% after leaving the multivariate analysis logistic regression with other risk factors ten factors Predeliric model, easily measurable within the first 24 hours of admission to the intensive care unit. Boogaard & others (2012).

At present there is no model prediction of delirium in Spanish that allows health personnel to identify patients who should objectively implement interventions to prevent delirium. That is why in Bucaramanga Santander research group Everest nursing program, together with a group of nurse researchers from different institutions conducted a study to validate the prediction model PREDELIRIC a Spanish version, this study leave evidence the need to identify and implement nursing interventions aimed at the prevention of delirium in critically ill patients in Bucaramanga. (Torres Marquez, Vanegas & Esparza, 2014).

This research study met the Spanish process of validating the model prediction of delirium (PREDELIRIC) including: internal validity, construct validity, external validity and reproducibility. As a priority, so that this model is applied in the city of Bucaramanga - Colombia, in order to determine the discriminatory power of the prediction model PREDELIRIC adult critically ill patients within 24 hours after admission to predict the development delirium during their stay ICU, in turn facilitating the specific application of non-pharmacological preventive measures patient to determine high risk ≥50%.

Specific objectives.

 -To determine the validity and reproducibility of the model prediction of delirium in Spanish, in the patient population of the intensive care unit.

-Determine the facial validity, content, internal consistency, and reproducibility of the model prediction of delirium in Spanish.

-Bringing the urea level conversion mlmol to milligrams per deciliter (mg / dl).

-To determine the incidence of delirium in critically ill patients according to the presence of risk factors model.

-Describe the sociodemographic and clinical characteristics of patients critics three intensive care units of Bucaramanga features.

-Describe the effectiveness of the prediction model PREDELIRIC through comparison with patients who develop delirium by application of CAM - ICU.

-Design and implement a nursing intervention protocol aimed at preventing delirium in patients presenting greater percentage of prediction deliric 50%.

Methodology: A descriptive study of validation or evaluation of diagnostic technologies was conducted in two phases, the first phase of the study was cross-cultural adaptation and validation of the Spanish version of the prediction model of delirium (PREDELIRIC), the second stage evaluated the validity and reproducibility of the prediction model of delirium in Spanish, patients were recruited voluntarily, after signing informed consent and meeting inclusion criteria, in the months of February to May 2014 in "Los Comuneros" university Hospital of Bucaramanga.

Convenience sampling, entering consecutive patients who met the inclusion criteria in intensive care units of study to complete the sample size calculated was performed.

Authorization by the research committee of the UDES and The Communards Hospital Universitario de Bucaramanga
Coordination with the director of UCI
Patient recruitment written informed consent of the patient or family
Home Valuation: Scale RASS, CAM-ICU (rule delirium), PREDELIRIC
Daily monitoring (RASS, CAM-ICU)
Exit: delirium, discharge from the ICU or death

Results: The validity evaluated the discriminatory power of the model prediction of delirium in critically ill adult patients 24 hours after admission to predict the development of delirium during their ICU stay was performed by analyzing the ROC curve for short English, Receiver operating characteristic curve (ROC); thus, the sensitivity, specificity, positive predictive value, negative predictive value, test efficiency, positive likelihood ratio and negative likelihood ratio in the cut proposed by the author of the prediction model for delirium was calculated. It was also evaluated whether there was a differential behavior of the ROC curve by gender.

Data showing the validity of the criterion of ≥50% probability resulting conclusion that the prediction model delirum Predeliric has adequate discriminatory power, demonstrated by a sensitivity of 78% and specificity of 70%.  So, The discriminatory capacity of the prediction model PRE DELIRIC was good given for under the ROC curve of 81% area with confidence intervals of 95% from 70% to 93%, ie the model does predict and properly classify the outcome delirium or delirium in approximately 8 out of 10 patients.

Recommendations: It is recommended once the results with a total sample size are proposing a new cutoff Predeliric prediction model in its Spanish version.
It requires the application of the model to proactively identify and treat patients at high risk (predicted values> 50%).
This highlights the importance of implementing a program aimed at early detection and prevention of delirium in the intensive care units of Bucaramanga and Colombia