Hospital readmissions for asthma: a feasibility study comparing strategies for linking hospital morbidity data

To investigate whether the variables in the National Hospital Morbidity Database (NHMD) provide sufficient information to validly link hospital admission records for the same person so that persons admitted and re-admissions for a specified disease can be enumerated.

Records of hospital admissions where asthma was the principal diagnosis were extracted from the New South Wales Inpatient Statistics Collection for the period July 2000 to June 2003 and linked using several strategies. The optimal' strategy applied probabilistic record linkage, using many demographic and administrative variables. A range of restricted strategies, using only those variables that were available with the NHMD (sex, date of birth, and either postcode or statistical local area of residence) and linking them deterministically, were evaluated and their validity for quantifying readmission for asthma within 28 days was assessed relative to the optimal strategy.

The optimal and restricted linkage strategies obtained similar estimates of readmissions. Approximately 95% of readmissions within 28 days identified using the optimal strategy were also identified using the restricted strategies.

Linking hospital records where asthma was the principal diagnosis using only those variables available in the NHMD enabled reliable identification of hospital readmissions tor asthma.

This methodology may have useful applications for monitoring the rate of readmissions for asthma and other chronic diseases nationally.