Statistical Analysis

Crude cancer incidence rates were calculated as the total number of patients with cancer during the follow-up period divided by the person-years at risk. The person-years at risk was the sum of time for all patients from SARD diagnosis to the initial cancer diagnosis, NHI program dropout, death, or December 31, 2012 (whichever came first). This study did not deal with a second or recurrent cancer. For age-specific rates, the age shifted over time to reflect the risk in different age groups. Individuals contributed his/her observation time to several 5-year age categories during the follow-up time. The total number of person-years was the sum of the amount of observation time contributed by all patients to given 5-year age categories. These person-years were the denominators of the rates for the 5-year age categories.¹³ The numerators of the rates for the 5-year age categories were the number of patients with cancer based on age at cancer diagnosis, not SARD diagnosis. The SIR is the ratio of the number of patients with cancer in this study to the expected number of patients with cancer based on age- and time-specific incidence rates in 5-year age intervals in the general Taiwanese population recorded in the Taiwan National Cancer Registry (1996–2000, 2001–2005, and 2006–2010).¹⁴ The data consist of cases and population data for specific age groups over a certain calendar time period, in the form of a matrix of 5-year age groups with 5-year intervals. This quinquinquennium method (5-year age group intervals × 5 calendar year cancer rates) could coincide with the national cancer incidence rate that is typically published by 5-year age intervals every 5 calendar years.¹³ SIR confidence intervals (CIs) were computed based on the Poisson distribution of observed cancer cases.¹⁵