ABSTRACT
Introduction: The Food and Drug Administration Center for Biologics Evaluation and Research conducts post-market surveillance of biologic products to ensure their safety and effectiveness. Studies have found that common vaccine exposures may be missing from structured data elements of electronic health records (EHRs), instead being captured in clinical notes. This impacts monitoring of adverse events following immunizations (AEFIs). For example, COVID-19 vaccines have been regularly administered outside of traditional medical settings. We developed a natural language processing (NLP) algorithm to mine unstructured clinical notes for vaccinations not captured in structured EHR data. Methods: A random sample of 1,000 influenza vaccine administrations, representing 995 unique patients, was extracted from a large U.S. EHR database. NLP techniques were used to detect administrations from the clinical notes in the training dataset [80% (N = 797) of patients]. The algorithm was applied to the validation dataset [20% (N = 198) of patients] to assess performance. Full medical charts for 28 randomly selected administration events in the validation dataset were reviewed by clinicians. The NLP algorithm was then applied across the entire dataset (N = 995) to quantify the number of additional events identified. Results: A total of 3,199 administrations were identified in the structured data and clinical notes combined. Of these, 2,740 (85.7%) were identified in the structured data, while the NLP algorithm identified 1,183 (37.0%) administrations in clinical notes; 459 were not also captured in the structured data. This represents a 16.8% increase in the identification of vaccine administrations compared to using structured data alone. The validation of 28 vaccine administrations confirmed 27 (96.4%) as "definite" vaccine administrations; 18 (64.3%) had evidence of a vaccination event in the structured data, while 10 (35.7%) were found solely in the unstructured notes. Discussion: We demonstrated the utility of an NLP algorithm to identify vaccine administrations not captured in structured EHR data. NLP techniques have the potential to improve detection of vaccine administrations not otherwise reported without increasing the analysis burden on physicians or practitioners. Future applications could include refining estimates of vaccine coverage and detecting other exposures, population characteristics, and outcomes not reliably captured in structured EHR data.
ABSTRACT
INTRODUCTION: Although prognostic differences between screen-detected, interval and symptomatic breast cancers are known, factors associated with wait times to diagnosis among these three groups have not been studied. METHODS: Of the 16,373 invasive breast cancers diagnosed between January 1, 1995 and December 31, 2003 in a cohort of Ontario women aged 50 to 69, a random sample (N = 2,615) were selected for chart abstraction. Eligible women were classified according to detection method; screen-detected (n = 1181), interval (n = 319) or symptomatic (n = 406). Diagnostic wait time was calculated from the initial imaging or biopsy to breast cancer diagnosis. Logistic regression analysis examined associations between diagnostic wait times dichotomized as greater or less than the median and demographic, clinical and prognostic factors separately for each detection cohort. RESULTS: Women who underwent an open biopsy had significantly longer than median wait times to diagnosis, compared to women who underwent a fine needle aspiration or core biopsy; (screen-detected OR = 2.76, 95% CI = 2.14-3.56; interval OR = 2.56, 95% CI = 1.50-4.35; symptomatic OR = 5.56, 95% CI = 3.33-9.30). Additionally, screen-detected breast cancers diagnosed with stage II and symptomatic cancers diagnosed at stage III or IV had significantly shorter diagnostic wait times compared to those diagnosed at stage 1 (OR = 0.66 95% CI = 0.50-0.87 and OR = 0.46, 95% CI = 0.25-0.85 respectively). CONCLUSIONS: Our study is consistent with expedited diagnostic work-up for breast cancers with more advanced prognostic features. Furthermore, women who had an open surgical biopsy had a greater than the median diagnostic wait time, irrespective of detection method.
ABSTRACT
BACKGROUND: Longer times from diagnosis to breast cancer treatment are associated with poorer prognosis. This study examined factors associated with wait times by phase in the breast cancer treatment pathway. METHODS: There were 1760 women eligible for the study, aged 50-69 diagnosed in Ontario with invasive breast cancer from 1995-2003. Multivariate logistic regression examined factors associated with greater than median wait times for each phase of the treatment pathway; from diagnosis to definitive surgery; from final surgery to radiotherapy without chemotherapy and from final surgery to chemotherapy. RESULTS: The median wait times were 17 days (Inter Quartile Range (IQR) = 0-31) from diagnosis to definitive surgery, 44 days (IQR = 34-56) from final surgery to postoperative chemotherapy and 75 days (IQR = 57-97) from final surgery to postoperative radiotherapy. Diagnosis during 2000-2003 compared to 1995-1999 was associated with significantly longer wait times for each phase of the treatment pathway. Higher income quintile was associated with longer wait time from diagnosis to surgery (OR = 1.47, 95% CI = 1.05-2.06) and shorter wait times from final surgery to radiotherapy (OR = 0.60, 95% CI = 0.37-0.96). Greater stage at diagnosis was associated with shorter wait times from diagnosis to definitive surgery (stage III vs I: OR = 0.49, 95% CI = 0.34-0.71). CONCLUSIONS: While diagnosis during the latter part of the study period was associated with significantly longer wait times for all phases of the treatment pathway, there were variations in the associations of stage and income quintile with wait times by treatment phase. Continued assessment of factors associated with wait times across the breast cancer treatment pathway is important, as they indicate areas to be targeted for quality improvement with the ultimate goal of improving prognosis.
