Font Size: a A A

Identification of factors associated with postoperative pnuemonia using a data mining approach

Posted on:2006-09-23Degree:Ph.DType:Dissertation
University:Boston CollegeCandidate:Berger, Anne M. DunnFull Text:PDF
GTID:1458390008972240Subject:Health Sciences
Abstract/Summary:
Postoperative pneumonia is currently a leading postoperative complication, occurring in roughly 18% of all patients undergoing surgery. It has the highest mortality rate, with fatalities ranging from 20% to 46%, and it is significantly more costly to treat than other nosocomial infections. Over the past decade, substantial gains have been made in the treatment of postoperative pulmonary complications; however, prevention is paramount, particularly with increasing antibiotic resistance. Few factors have been identified as established predictors of postoperative pneumonia. Using Donabedian's Structure, Process, Outcome (SPO) framework; this study used a data mining approach to examine structure and process variables related to the outcome of postoperative pneumonia. Since human health is influenced by numerous structural and process attributes, it is important to evaluate the effects of the multiple factors that create risk beyond that of each individual factor alone.; Data mining methods allowed maximum use of clinical data, and the ability to examine numerous variables and their interactions in relationship to the outcome of postoperative pneumonia. An objective of this study was to identify the combination of characteristics that appeared to indicate risk for the development of this outcome. The identification of factors associated with the outcome of postoperative pneumonia provides the clinical evidence necessary to ground infection control practices and the care of surgical patients in scientific knowledge. This knowledge is essential before nurses can develop effective health promotion strategies and preventive interventions.; This study identified twenty variables associated with the outcome of postoperative pneumonia. It confirmed two established predictors, eight likely predictors, seven questionable predictors, and identified seven new factors associated with postoperative pneumonia. It also resulted in the development of a Naive Bayes predictive model that is 72% accurate in the prediction of positive postoperative pneumonia cases. The identification of patients who are at risk for this complication also provides the groundwork for future nursing research studies to assess the effectiveness of nursing assessment, tailored interventions and care delivery to prevent this serious complication.
Keywords/Search Tags:Postoperative, Factors associated, Data mining, Complication, Identification
Related items