Jared Sturgeon and Dr. Ray Merrill, Health Science
Despite the generally lower burden of cancer experienced among LDS than non-LDS residents, prostate cancer incidence and mortality rates tend to be higher in Utah than the rest of the country.1 Reasons for this difference are unknown but may be related to differences in diet, screening, and/or treatment.
Currently prostate cancer is the leading cause of cancer incidence and death in men in the United States.2 Understanding if LDS men are more likely to be diagnosed with and die from the disease than non-LDS men may provide greater insight into risk factors, screening behaviors, and the influence of treatment for this disease.
The data being used for this research comes from three primary sources. First, the Utah State Death Records were used, complete with information on cause as well as other identifying information. The second source is the Church of Jesus Christ of Latter-day Saints. The LDS church gave us death records from 1985-1999, which contain basic identifying information. Lastly, we received data from the Utah Cancer Registry. This information contains individuals with prostate cancer, and details on grade, histology, and medical care.
These three data sets were combined into one through a probabilistic linkage. As one of the contributing professors has had extensive use with LinkPro and found it quite adequate for the data we were examining, we chose to continue with LinkPro. We first linked the church to state data. In previous work by Dr. Sterling Hilton very successful linkages between data sets similar to these had been conducted using eleven identifying variables. These are listed below:
· soundex first name
· soundex middle name
· soundex last name
· first name
· middle name
· last name
· birth day
· birth month
· birth year
· county
· marital status
These variables yielded a remarkable 96.7% linkage in our data. The final step, which was delayed for several months, was the linkage of our combined data to the Utah Cancer Registry. This linkage occurred in the summer of 2000, and linked at approximately 94% with the addition of social security number. The Utah Cancer Registry representative went to great lengths to ensure confidentiality, and the final working data set contains no personal identifying information that could be inappropriately used.
As stated above, our goal of examining the patterns of mortality incidence between LDS and non-LDS males in Utah would lead to insight concerning prostate cancer. However, those goals have not yet been realized due to unforeseen delays in data acquisition and analysis. These delays come from two sources. The first was academic policy that traced our data through a University of Utah professor who allegedly had not gone through the proper University of Utah approval. Though this inquiry did not concern us directly, it did indirectly halt our project while we independently secured the data. Second, justifiable concerns about confidentiality led to unexpected delays in getting data from the Utah Cancer Registry. These concerns were resolved by going through all proper committees, and having a Cancer Registry representative present when their data was being linked.
The current data set is presently being examined for any potential flaws that could make it biased. Barring any problems, the actual analysis of the data will begin in the fall of 2000. This analysis will answer the original questions we posed, and provide the information for a journal article.
At this point the ORCA project is nearing completion. The primary lessons we have learned deal more with data set management and academic confidentiality than prostate cancer. Fortunately the exciting parts lie ahead.
References
- Utah Cancer Registry Cancer Statistics Annual Report.
- Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics, 1999 [see comments]. CA Cancer J Clin 1999;49:1, 8-31. 370