Abstract
Background: Availability of longitudinal data from the electronic medical record (EMR) provides opportunities to monitor trends in the incidence and treatment of chronic diseases such as asthma. The Centers for Disease Control (CDC)-funded PASS studies developed, evaluated and implemented an algorithm for using the EMR to identify incident cases of asthma in the Kaiser Permanente Northern California (KPNW) region over a 5-year period.
Methods: The original PASS study was carried out in two phases. In phase I candidate algorithms based on medication dispensings and asthma diagnoses from doctor visits were developed and piloted. A range of options for medication dispensing criteria, health plan eligibility, and length of the surveillance period were considered. The resulting algorithm was arrived at through a collaborative decision process involving the CDC and investigators from two field centers. In phase II the algorithm was validated by recruiting a subset of 219 randomly-selected patients identified by the algorithm as having incident asthma and collecting clinical and self-report measures from them, including pulmonary function tests. Two KPNW pulmonologists reviewed all information relating to each of these cases to arrive at a ‘gold standard’ asthma classification for these patients.
Results: The final PASS asthma algorithm requires the following criteria be met: either two dispensings of asthma medication, or two or more visits at which asthma was noted as a diagnosis. The ‘at-risk’ population is identified by excluding those who meet these criteria in the 4 years prior to the year of interest and those who do not meet membership eligibility. Individuals in the resulting ‘at-risk’ population who meet the asthma criteria in the year of interest are identified as incident for asthma. The validation phase resulted in KPNW physicians rating 85% of the 219 charts reviewed as probable/possible asthma and 15% as unlikely. This represents a predictive value of 86%. This algorithm has subsequently been implemented in the 5-year PASS2 study which aims to, among other things, carry out asthma surveillance to determine the incidence and prevalence of physician-diagnosed asthma and develop and maintain a research database of incident asthma cases. The next step is to evaluate the scalability of this algorithm in other EMR-based systems.
Conclusion: The PASS studies have demonstrated that it is feasible to develop and implement an EMR-based algorithm for asthma surveillance. Similar methods can be employed for other chronic diseases.
- Received September 11, 2008.
