Now that the data are more readily available for outcomes research and the techniques to analyze that data are available, we need to use the tools to investigate the total complexity of patient care. We should no longer rely upon basic tools while ignoring sequential treatments for patients with chronic diseases or the issue of patient compliance, and we can start investigating treatments from birth to death. It is no longer possible, with these large datasets, to rely on t-tests, chi-square statistics and simple linear regression. Without the luxury of clinical trials and randomizing patients into treatment versus control, there will always be confounding factors that should be considered in the data. In addition, large datasets almost guarantee that the p-value in a standard regression is statistically significant, so other methods of model adequacy must be used. If we do not start using outcomes data, we are missing crucial knowledge that can be used to improve patient outcomes while simultaneously reducing the cost of care. If we continue to use inferential statistical methods that were not designed to work with large datasets, we will not extract the information that is readily available in the outcomes datasets.
TopBackground
As discussed in Chapter 1, data preprocessing is not given strong consideration in the medical journals. There are just a few books that discuss preprocessing generally. (Pyle, 1999; Refaat, 2006; Svolba, 2007) There are a few books that focus on analyzing the electronic medical record; however, they spend little time on the preprocessing. (Begg, Kamruzzaman, & Sarker, 2006; Weerasinghe, 2009) While it is suggested that the major problem is to find appropriate data, we suggest that it has more to do with making appropriate use of the data that are available. (Mark, Salyer, & Geddes, 1997)
The number one question remains how best to handle the patient co-morbidities as represented in ICD9, CPT, and HCPCS codes. Without techniques that can handle all of these codes to investigate the relationship of co-morbidities to codes, some of the most severe, acute conditions are omitted. We have discussed these codes in detail and given several methods to work with the data and to apply the compressed codes to linear and predictive models. Risk adjustment is important in any investigation, and that involves the use of these codes. (Garrison, et al., 2002)
Health outcomes research remains in its infancy in terms of extracting meaningful long term results of the relationship of treatment choices to patient outcomes. With more knowledge of preprocessing and of exploratory techniques, such analyses can become commonplace. While there are many national databases available as we have shown in this book, some researchers are still reluctant to use them to study patient outcomes. (Harpe & Harpe, 2009) Yet, regardless of the accuracy of the research, the results will be used more often to make health care policy. (Doherty, et al., 2004)