Monitoring vaccine safety using the vaccine safety Datalink: Assessing capacity to integrate data from Immunization Information systems.

BACKGROUND: The Vaccine Safety Datalink (VSD) uses vaccination data from electronic health records (EHR) at eight integrated health systems to monitor vaccine safety. Accurate capture of data from vaccines administered outside of the health system is critical for vaccine safety research, especially for COVID-19 vaccines, where many are administered in non-traditional settings. However, timely access and inclusion of data from Immunization Information Systems (IIS) into VSD safety assessments is not well understood. METHODS: We surveyed the eight data-contributing VSD sites to assess: 1) status of sending data to IIS; 2) status of receiving data from IIS; and 3) integration of IIS data into the site EHR. Sites reported separately for COVID-19 vaccination to capture any differences in capacity to receive and integrate data on COVID-19 vaccines versus other vaccines. RESULTS: All VSD sites send data to and receive data from their state IIS. All eight sites (100%) routinely integrate IIS data for COVID-19 vaccines into VSD research studies. Six sites (75%) also routinely integrate all other vaccination data; two sites integrate data from IIS following a reconciliation process, which can result in delays to integration into VSD datasets. CONCLUSIONS: COVID-19 vaccines are being administered in a variety of non-traditional settings, where IIS are commonly used as centralized reporting systems. All eight VSD sites receive and integrate COVID-19 vaccine data from IIS, which positions the VSD well for conducting quality assessments of vaccine safety. Efforts to improve the timely receipt of all vaccination data will improve capacity to conduct vaccine safety assessments within the VSD.

An independently validated, portable algorithm for the rapid identification of COPD patients using electronic health records.

Electronic health records (EHR) provide an unprecedented opportunity to conduct large, cost-efficient, population-based studies. However, the studies of heterogeneous diseases, such as chronic obstructive pulmonary disease (COPD), often require labor-intensive clinical review and testing, limiting widespread use of these important resources. To develop a generalizable and efficient method for accurate identification of large COPD cohorts in EHRs, a COPD datamart was developed from 3420 participants meeting inclusion criteria in the Mass General Brigham Biobank. Training and test sets were selected and labeled with gold-standard COPD classifications obtained from chart review by pulmonologists. Multiple classes of algorithms were built utilizing both structured (e.g. ICD codes) and unstructured (e.g. medical notes) data via elastic net regression. Models explicitly including and excluding spirometry features were compared. External validation of the final algorithm was conducted in an independent biobank with a different EHR system. The final COPD classification model demonstrated excellent positive predictive value (PPV; 91.7%), sensitivity (71.7%), and specificity (94.4%). This algorithm performed well not only within the MGBB, but also demonstrated similar or improved classification performance in an independent biobank (PPV 93.5%, sensitivity 61.4%, specificity 90%). Ancillary comparisons showed that the classification model built including a binary feature for FEV1/FVC produced substantially higher sensitivity than those excluding. This study fills a gap in COPD research involving population-based EHRs, providing an important resource for the rapid, automated classification of COPD cases that is both cost-efficient and requires minimal information from unstructured medical records.

The sensitivity of single air parcel trajectory calculations to starting elevation.

Trajectory models are frequently used to characterize the atmospheric transport pathways for airborne gases and aerosols. Users of these models must specify a starting elevation for their calculations. The variation of wind with altitude causes trajectory models to be sensitive to the starting elevation, particularly when single trajectories rather than Lagrangian particle dispersion simulations are used to characterize atmospheric transport. In this work we systematically investigate and quantify the sensitivity of single trajectory calculations to the starting elevation. The analysis was based on an eight-year database of daily, 48-h back-trajectories calculated for ten sites. Trajectories were calculated at four different starting elevations, and the horizontal difference between endpoints was determined for five upwind travel times. Trajectory model calculations were found to be strongly sensitive to starting elevation. A 500 m difference in starting elevation leads to an average horizontal separation of 326 km after 48 h. Mean horizontal separations of 627 km and 886 km were found for starting elevation differences of 1000 m and 1500 m, respectively. A seasonal dependence of the sensitivity was found, with the smallest separations occurring during the summer, the largest during winter, and intermediate values during the fall and spring. A linear relationship was observed between trajectory model sensitivity and difference in starting elevation. Empirical equations were presented to approximate this relationship.