Patient-assessed health outcomes in a cycle of continuous quality improvement : a case coronary artery bypass graft surgery

Thesis argues that patient assessed outcomes and data and records used in health information systems can be used to improve subsequent patient outcomes in a cycle of continous quality improvement.. The problem addressed is whether patient-assessed outcomes can be used to improve subsequent patient outcomes through altering procedures and protocols. Despite a growing awareness amongst some health professionals of the need for "real time science in medicine' -that is, routinely linking patient-assessed outcomes to the processes of care -there are no such studies reported in the literature. A continuous quality improvement (CQI) model for surgical procedures incorporating patient-assessed outcomes is proposed. The principles of CQI are demonstrated with a model formulated, tested and refined with data collected for coronary artery bypass graft surgery (CABG) at a major Australian teaching hospital. This database records pre-surgery, peri-surgery and post-surgery details for 3979 CABG between 1984 and 1993. The perfu~ion process is described with emphasis on the main measures -blood gases, blood pressure and temperature -under the control of the perfusionist. A follow-up survey administered to patients one year after surgery has two aggregate scores of patient health outcomes -one assesses neurological outcome (NSUM) whilst the other assesses physical outcome (PSUM). Descriptive, univariate and multivariate analyses of these data have been conducted to detennine associations with explanatory variables. Graphical procedures are used to illustrate associations between outcome measures and physiological variables. The univariate analyses has been carried out using SPSSX for MAC and the multivariate analyses uses the SAS LOGIST procedure on a VAX mainframe. As NSUM and PSUM are classified into ordered categorical data, a multivariate statistical model to explain variations using ordered polytomous logistic regression is employed to analyse these data. The variables that have association with increasing NSUM or neurological deficit and increasing PSUM or physical deficit are presented and discussed. The results of these statistical analyses are used in a CQI model. The process of improvement is demonstrated with improved neurological and physical outcomes over time. A general model for other surgical procedures is described. Various ways of presenting the relationship between outcome and process are presented for different methods of initiating change in CABG surgery practice and other surgical practice. Areas of research significance are: multivariate statistical techniques (ordered categorical polytomous logistic regression) applied to a large database demonstrates the attribution of patient-assessed outcomes to surgical procedure.; validation of self-assessed measures (ordinal and categorical) provides an important contribution to the outcomes movement; and an outcome management process, where specific postdischarge measures of outcome have been used to continually develop protocols for the surgical process. All of the above represent significant contribution to both health outcome and health information systems research.