ABSTRACT
Background: Immunization clinical decision support (CDS) systems provide much needed guidance for clinicians in interpreting immunization guidelines and determining when vaccinations are due. In most circumstances, patients should be vaccinated according to the standard Advisory Committee on Immunization Practices (ACIP) schedule. However, for some patients the standard recommendations encoded into the CDS fail to address individual health. For example, underspecfication in ICD-10 codes for immunocompromised status led to confusion about whether a patient should or should not be given vaccines. Additionally, the system we currently used only allowed for limited 'on' and 'off' designations, resulting in some families choosing to self-select as “no to all vaccines” even there were some vaccines they were interested in their child receiving such as school required vaccines. As some new vaccines have included a shared decision making recommendation, including the meningitis B vaccine and the COVID vaccines, we recognized the need for a more finite level of control in our immunization CDS. Methods: We overhauled our immunization CDS backend to provide increased flexibility in rule interpretation and recommendation presentation. First, we switched from a binary interpretation of rules (due or not due) to a three-level interpretation (due and default selected, due but not default selected, or not due). Diagnoses that previously had shut off notices that a vaccine were due were re-interpreted to note that the vaccine would be due, but that a specific diagnosis in the chart indicated it should not be defaulted as selected. We developed a provider portal accessible by the clinician in real time to update rule recommendations at the patient level allowing a clinician to control for a patient what immunizations would be recommended at the antigen level. To facilitate ease of use for the clinician, we included pre-populated selections for reasons a parent may elect to defer on default immunizations. Additionally, we provided links to content on supporting shared decision making for immunizations. Results: The updated immunization CDS successfully allowed clinicians to modify recommended vaccinations at the patient and antigen level. Multiple scenarios were tested including personal history of varicella, which can be driven by problem list entries, and history of hepatitis A infection with natural immunity, for which diagnosis data is insufficient for altering the recommendations. In all cases clinicians were able to unselect default recommendation for immunizations. Conclusion: Immunization CDS, which previously only allowed for default selection of immunizations was successfully modified to support personal preferences, finite recommendations related to previously inaccessible information within the EHR, and to allow for future implementation of shared decision making recommendation level immunizations. The interphase developed allowed for clinician updating of immunization recommendations in realtime at the patient level and antigen level. After the rules engine has determined which vaccines are potentially due, presentation of products occurs through this order screen. Products with a shared decision making recommendation status as per ACIP are not defaulted. Where possible, patient data from the EHR is used to support recommendation evaluated (e.g. for varicella). Where notpossible due to underspecification of ICD-10 codes, clinician entered information is used to enhance the specificity of alerts. Narrative text describing all modifications to recommendations, including the source of the modification, is displayed. This is the user interface used by clinicians to modify recommendations at the patient level. With a single click clinicians can add or remove personalizations to the routinely recommended immunizations. By default all antigens are recommended. When a clinician unselected a check box products containing that antigen are still listed as due, but no longer defaulted. Any combination products containing that an igen are not recommended so that the system can have other portions of the combination product still be default ordered.
ABSTRACT
Background. With the onset of the coronavirus disease 2019 (COVID-19) pandemic, pediatric primary care delivery changed rapidly. Prior studies have demonstrated a reduction in ambulatory encounters and antibiotic prescriptions with the pandemic onset;however, the durability of these reductions in pediatric primary care in the United States has not been assessed. Methods. We conducted a retrospective cohort study to assess the impact of the COVID-19 pandemic and associated public health measures (e.g. social distancing, masking, school closures, and increased availability of telemedicine) on antibiotic prescribing and encounter volume in 27 pediatric primary care practices, and the duration of these changes. Patients under age 19 with an encounter from January 1, 2018 through December 31, 2020 were included. The primary outcome was monthly antibiotic prescriptions per 1000 patients, in the overall population and a subset of encounters with infectious diagnoses, including respiratory tract infections (RTIs). Interrupted time series (ITS) analysis was performed. Results. There were 60,562 total antibiotic prescriptions from April to December in 2019 and 14,605 antibiotic prescriptions during the same months in 2020, a 76% reduction. The reduction in RTI encounter prescriptions accounted for 91.5% of the overall reduction in prescriptions from 2019 to 2020. Using ITS analysis, there was an immediate decrease from 31.6 to 7.4 prescriptions/1000 patients (predicted means) in April 2020 (-24.2 prescriptions/1000 patients;95% CI: -31.9, -16.4) (Figures 1 and 2). This was followed by a stable rate of antibiotic prescriptions that remained flat through December 2020. For RTI encounters, a similar pattern was seen, with a decrease by 21.8 prescriptions/1000 patients;95% CI: -29.5, -14.2) (Figures 1 and 2). Encounter volume also decreased immediately, and while overall encounter volume began returning to a pre-pandemic baseline volume toward the end of the study period, RTI encounter volume remained persistently lower through December 2020 (Figure 3). RTI = respiratory tract infection;UTI = urinary tract infection;SSTI = skin and soft tissue infection. Months are numbered sequentially, starting with January (number 1). Dashed line indicates first full month of the pandemic, April 2020. Interrupted time series analysis for antibiotic prescriptions per 1000 patients by month from January 2018 to December 2020 for (A) all antibiotics as well as antibiotics prescribed at encounters with (B) respiratory tract infections (RTIs), (C) urinary tract infections (UTIs), and (D) skin and soft tissue infections (SSTIs) Intervention starts in April 2020 (dashed line). Months are numbered sequentially, starting with January (number 1). Dashed line indicates first full month of the pandemic, April 2020. Antibiotic prescriptions per 1000 billed encounters by month from January 2018 to December 2020 for (A) all encounters, as well as antibiotics prescribed at encounters with (B) respiratory tract infections (RTIs), (C) urinary tract infections (UTIs), and (D) skin and soft tissue infections (SSTIs) Months are numbered sequentially, starting with January (number 1). Conclusion. Dramatic reductions in antibiotic prescribing in pediatric primary care during the COVID-19 pandemic were sustained through 2020, primarily driven by reductions in RTI encounters.
ABSTRACT
Rationale: The COVID-19 pandemic dramatically changed daily routines as well as healthcare utilization and delivery patterns in the United States. We sought to identify changes in pediatric asthma-related healthcare utilization and levels of air pollution i.e. particulate matter (PM2.5, PM10) and gaseous chemicals (NO2, O3) during the COVID-19 pandemic in Philadelphia. We hypothesized that declining utilization of asthma care and changed pollution levels during the early stages of the pandemic rebounded after the relaxation of COVID-19-related public health measures. Methods: For the time period Mar 17 to Dec 17 during the years 2015-2020, asthmarelated encounters and weekly summaries of respiratory viral testing data were extracted from Children's Hospital of Philadelphia (CHOP) electronic health records. Daily average estimates of PM2.5, PM10, O3, and NO2 for the same time period were obtained from AirData, an EPA resource that provides quality-assured summary air pollution measures collected from outdoor regulatory monitors across the United States. Patterns in encounter characteristics and viral testing in Philadelphia from Mar 17 to Dec 17, 2020, were compared to data from 2015-2019 as a historical reference. Encounter pattern results were summarized as percentage changes. Controlled interrupted time series regression models were created to identify statistically significant differences in pollution levels that differed in 2020 compared with historical time periods. Results: We present data on asthma encounters, viral testing, and air pollution from Mar 2020 through Dec 2020. Contrary to the early stages of the pandemic when in-person asthma encounters decreased by 87% (outpatient) and 84% (emergency + inpatient), asthma-related encounters rebounded with the relaxation of COVID-19-related public health measures. During the initial months of the pandemic, the daily average of PM2.5, PM10, and NO2 levels decreased by 29.0% (2.17 μg/m3), 18.2% (3.13 μg/m3), and 44.1% (6.75 ppb), respectively, whereas ozone levels increased by 43.4% (10.08 ppb), changes that were not statistically significantly different compared to historical trends. Levels of all pollutants considered remained similar during subsequent 2020 months compared to the 2015-2019 reference period. Conclusion: The COVID-19 pandemic in Philadelphia was accompanied by initial decreases in pediatric asthma healthcare activity. Concurrent with the relaxation of COVID-19-related public health measures, there was a subsequent increase in asthma healthcare activity. No substantial change in air pollution levels compared with historical patterns was observed during the time period considered, suggesting that other factors influenced changes in asthma trends during the COVID-19 pandemic.