Early studies tended to use logistic
regression methods for binary mortality status at the end of the follow-up period. These
models report odds ratios (OR). Most recent studies use more sensitive Cox proportional
hazards methods for analysis of time to death. These models report hazard ratios (HR). The
method of analysis does not appear to affect the significance of the results.
become an extremely well-replicated finding in a very short time, considering that the
studies are prospective and have quite long follow-up periods. Most or all of the studies
would be considered "secondary analysis" of data (the data were collected
initially for another purpose). Because self-ratings of health are so commonly used in
surveys, there were many available data files with the necessary elements for the
analysis, and hence a large number of studies have appeared.