Publication Date

2015-07-28

Availability

Open access

Embargo Period

2015-07-28

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PHD)

Department

Nursing (Nursing)

Date of Defense

2015-06-23

First Committee Member

Karina Gattamorta

Second Committee Member

Mary Hooshmand

Third Committee Member

Brian McCabe

Fourth Committee Member

Abel Murillo

Abstract

Estimates show that up to 3 out of 4 patients have moderate to severe postoperative pain following surgery. As a result, patients suffering from postoperative pain not only face increased rates of morbidity and mortality but also report decreased satisfaction with care. Pre-operatively predicting patients at risk for severe postoperative pain may improve postoperative pain management, patient outcomes, and patient satisfaction. A retrospective quantitative study design was conducted to examine the preoperative factors associated with the development of severe postoperative pain. The study sample was collected from the electronic health records (EHR) of all surgical patients at the University of Miami Hospital from October 2014 through April 2015. The first four months of data (N=1,794) was abstracted from the EHR to test the measurement and structural model using structural equation modeling (SEM), and to develop a prediction tool utilizing the regression coefficients. Validation utilized an independent sample (n = 1961) to confirm the prediction tools ability to predict severe postoperative pain. Preoperative predictors were gender, age, ethnicity, preoperative pain intensity, type of surgery, and baseline hemodynamic heart rate and blood pressure. Structural equation modeling (SEM) examined the factors to determine the relationship between the predictive variables and severe postoperative pain, a numerical rating scale (NRS) of ≥ 6. Results from the initial analysis supported independent links between baseline heart rate, preoperative pain intensity, expected surgical pain (low, moderate, high, highest), age, female gender, and Hispanic ethnicity to severe postoperative pain on postoperative day one (POD 1). The discrimination for the final model was fair, based on the receiver operator characteristic (ROC) curve 0.714. The validity of the prediction equation was determined through use of an independent validation dataset (n=1,961). Using a data-derived cutoff point of 0.20, obtained from the ROC curve, the predicted scores were compared to the reported scores of severe postoperative pain showing a sensitivity of 73.33% and a specificity of 74.40%. An additional cutoff pint of 0.14 was analyzed, which showed an improved sensitivity of 74.19% and a specificity of 57.33%. The study further demonstrates that severe postoperative pain can be predicted through the use of a simple modified validated prediction equation. Utilization of separate cutoff points allows for the development of intervention strategies based on the need for a higher specificity versus higher specificity. The developed prediction equation provides a tool for clinically anticipating patients at risk for the development of severe postoperative pain on postoperative day 1. Future research is needed to show if preoperative measures and interventions based on the prediction equation improve postoperative pain management, and therefore decrease the incidence of severe postoperative pain.

Keywords

Predictors; Severe; Postoperative Pain; Prediction Equation

Share

COinS