C R Ramsay1, M K Campbell, C M Glazener. 1. Health Services Research Unit, University of Aberdeen, Polwarth Building, Foresterhill, Aberdeen AB25 2ZD.
Abstract
OBJECTIVE: To describe linking Community Health Index (CHI) records with Scottish Morbidity Records for neonates (SMR11) for the purpose of follow-up in a large clinical trial. DESIGN: A two-stage probabilistic computer match based on date of birth, surname and postcode, supplemented with hand matching. SETTING: Aberdeen Maternity Hospital, Grampian. SUBJECTS: 10,835 babies born in the two year period between 1 March 1993 and 28 February 1995. RESULTS: The computerised linkage matched 89.8% of the SMR11 records to a CHI record. All of the remaining 10.2% unlinked babies were accounted for by manual checking. There were 194 (1.8%) babies who did not have a CHI record due to the baby leaving the area or dying before allocation was possible. The sensitivity of the matching was estimated at 99.9% (95% CI is 99.8% to 100%). CONCLUSION: We were able to correctly computer match 89.7% of our babies' SMR11 records with the appropriate CHI number. We have shown that the follow-up of neonates in a clinical trial setting using record linkage between secondary and primary care systems is feasible, efficient and useful. However, this exercise would have been greatly facilitated by a unique NHS identifier common to all records.
OBJECTIVE: To describe linking Community Health Index (CHI) records with Scottish Morbidity Records for neonates (SMR11) for the purpose of follow-up in a large clinical trial. DESIGN: A two-stage probabilistic computer match based on date of birth, surname and postcode, supplemented with hand matching. SETTING: Aberdeen Maternity Hospital, Grampian. SUBJECTS: 10,835 babies born in the two year period between 1 March 1993 and 28 February 1995. RESULTS: The computerised linkage matched 89.8% of the SMR11 records to a CHI record. All of the remaining 10.2% unlinked babies were accounted for by manual checking. There were 194 (1.8%) babies who did not have a CHI record due to the baby leaving the area or dying before allocation was possible. The sensitivity of the matching was estimated at 99.9% (95% CI is 99.8% to 100%). CONCLUSION: We were able to correctly computer match 89.7% of our babies' SMR11 records with the appropriate CHI number. We have shown that the follow-up of neonates in a clinical trial setting using record linkage between secondary and primary care systems is feasible, efficient and useful. However, this exercise would have been greatly facilitated by a unique NHS identifier common to all records.
Authors: Dorothea Nitsch; Susan Morton; Bianca L DeStavola; Heather Clark; David A Leon Journal: BMC Med Res Methodol Date: 2006-03-22 Impact factor: 4.615