The California Automated Mortality Linkage System (CAMLIS), established in 1981 to facilitate the conduct of follow-up studies in the State of California, employs a combination of deterministic and probabilistic linkage decision criteria to perform the death clearance function. The system was evaluated against four traditional death clearance procedures and the performance of each procedure measured in terms of measures of sensitivity and specificity. Only one procedure was associated with a specificity lower than 0.99; for that one, the specificity was 0.93. There was much greater fluctuation in the observed sensitivity levels. In one of the procedures, CAMLIS demonstrated a sensitivity of 0.97 versus 0.79 for the Social Security Administration. A comparison against the National Death Index (NDI) produced sensitivities of 0.89 for CAMLIS and 0.94 for the NDI. An assessment of manual search procedures using a file of Japanese names produced a CAMLIS sensitivity measure of 0.92 compared with 0.93 for the manual search. Another manual search procedure using microfiche copies of the state death index produced a CAMLIS sensitivity of 0.97; in this evaluation, the sensitivity of the manual search was defined as 1.0. Another measure of performance of a death clearance procedure is its predictive value in identifying a person who has died; CAMLIS generated predictive values in these evaluations that ranged from 0.93 through 0.99, contrasted with the NDI value of 0.59.
[1]
G. Beebe.
Record linkage systems--Canada vs the United States.
,
1980,
American Journal of Public Health.
[2]
M. Feinleib,et al.
On the feasibility of linking census samples to the National Death Index for epidemiologic studies: a progress report.
,
1983,
American Journal of Public Health.
[3]
J. Kagawa,et al.
An Operational Approach to Record Linkage
,
1983,
Methods of Information in Medicine.
[4]
G R Howe,et al.
A generalized iterative record linkage computer system for use in medical follow-up studies.
,
1981,
Computers and biomedical research, an international journal.
[5]
H B Newcombe,et al.
Accuracies of Computer versus Manual Linkages of Routine Health Records
,
1979,
Methods of Information in Medicine.
[6]
H B NEWCOMBE,et al.
Automatic linkage of vital records.
,
1959,
Science.
[7]
J. Neaton,et al.
An evaluation of the Social Security Administration master beneficiary record file and the National Death Index in the ascertainment of vital status.
,
1983,
American journal of public health.