A record linkage protocol for a diabetes registry at ethnically diverse community health centers

Community health centers serve ethnically diverse populations that may pose challenges for record linkage based on name and date of birth. The objective was to identify an optimal deterministic algorithm to link patient encounters and laboratory results for hemoglobin A1c testing and examine its variability by health center site, patient ethnicity, and other variables. Based on data elements of last name, first name, date of birth, gender, and health center site, matches with >= 50% to < 100% of a maximum score were manually reviewed for true matches. Match keys based on combinations of name substrings, date of birth, gender, and health center were used to link encounter and laboratory files. The optimal match key was the first two letters of the last name and date of birth, which had a sensitivity of 92.7% and a positive predictive value of 99.5%. Sensitivity marginally varied by health center, age, gender, but not by ethnicity. An algorithm that was inexpensive, accurate, and easy to implement was found to be well suited for population-based measurement of clinical quality.