An assessment of the reliability of race, Hispanic ethnicity, birthplace, and tobacco history data in the Massachusetts cancer registry, 2005-2009.

The Massachusetts Cancer Registry (MCR) reviewed the medical charts of 5,438 randomly selected breast and colorectal cancer cases diagnosed from 2005-2009 in part to assess the reliability of the race, Hispanic ethnicity, birth country, and tobacco history variables. The kappa statistic was used to assess the reliability between the originally reported variable and the reabstracted variable. There was strong agreement of kappa score for race among whites, blacks, and other (Asian, Native American), indicating a good quality of race data. The agreement for birth country was strongest among those not born in the United States with a statistically significantly higher kappa score compared to the other categories. Agreement for Hispanic ethnicity was strongest for non-Hispanics, Puerto Ricans, Central Americans, and Dominicans, groups that represent the majority of Hispanics in Massachusetts. The agreement for tobacco history was strongest among current tobacco users. This study provided useful information on the reliability of the race, Hispanic ethnicity, and birthplace variables, which are frequently used in MCR reports. It also provided heretofore unknown data on the reliability of the tobacco history variable. All categories of the race variable were very reliable as were the categories of Puerto Rican, Dominican, and Central American. Hispanic, NOS was not as reliable due to the large Portuguese population with Hispanic sounding surnames. Birth country was not as reliable due to the paucity of the data in many of the larger facilities in the state.