Burden of Renal Cysts Imaging: A Survey of Patients among the Greater Plains Collaborative

<b><i>Purpose:</i></b> Renal cysts are a frequent incidental finding on cross-sectional radiographic imaging. While most cysts are indolent, individuals with such cysts are frequently monitored for interval growth and potential malignant transformation, which is ultimately rare. In this study, we aimed to assess patients’ values and preferences (believes and attitudes) about renal cysts. <b><i>Methods:</i></b> We deployed a cross-sectional survey to a random sample of patients with a diagnosis of renal cysts who were identified by billing code and self-identification. We collected data about demographics, insurance status, family history and overall health, and characteristics of patients with renal cysts. We performed a binary regression analysis (adjusted for age, gender, family history of cancer and kidney disease, and treatment plan for renal cysts) to determine anxiety predictors in patients with renal cysts. <b><i>Results:</i></b> We included 301 respondents in whom billing code and self-identification corresponded; of these, 138 had renal cysts and 163 did not. In an adjusted regression analysis, there was a suggestion that a clear management plan (OR = 0.49, 95% CI [0.22–1.11]) (<i>p</i> value 0.08) may be associated with less anxiety and a family history of renal disease may be associated with more anxiety (OR = 1.94 [0.76–4.94]) (<i>p</i> value 0.17). Family history of cancer also did not significantly predict anxiety (OR = 0.54 [0.24–1.19]) (<i>p</i> value 0.13). All these results were not statistically significant and had wide confidence intervals of the effect estimates make the results imprecise. <b><i>Conclusion:</i></b> Findings of this pilot study suggest a clear management plan for the renal cyst(s) management may be associated with a lower level of anxiety, thereby by emphasizing the importance of good communication, patient engagement and evidence-based guidance. More definitive, adequately powered studies are needed to evaluate this finding further. In addition, further studies exploring differences in imaging practices, patient symptomatology and patient engagement by different provider types would be insightful. Ultimately, tools to improve shared decision-making are needed to provide more patient-centered care.

Time modifier billing code - an interrupted time series analysis

Time modifier billing code: Interrupted time series analysis. Terrence McDonald, Brendan Cord Lethebe, Alistair McGuire, Lee Green Background: Alberta has the highest percentage of fee-for-service Family Physicians in Canada at over 80%. In 2019 as part of a cost containment strategy, the Alberta government proposed a policy change to eliminate the most used fee code that compensates family physicians for extended visit times (16-25 minutes). Optimal length for patient visit times varies throughout the world and countries with health systems that place emphasis on relational continuity demonstrate a trend towards longer appointment times. In Canada, the relationship between visit length and outcomes is not known. Implementation: What would be the likely consequences of eliminating the extended visit code? We examined this question using two different observational methods, to improve confidence in our findings: a retrospective longitudinal cohort (time series) around the time the code was introduced in 2009, and a cross-sectional cohort at current time. We explored the usage patterns of that fee code, its association with the outcomes of emergency department visits and hospitalizations, along with physician billings. Results: We found rates of emergency department visits decreased after the time-modifier code was implemented starting in 2010. This effect was maintained in the years that followed. A similar but less pronounced effect was observed in the hospitalization rates. The cross-sectional analysis had to include an interaction term because family physicians selectively extend visits for patients at risk, but when that is accounted for, the same effect is observed as in longitudinal results. The code was not used ubiquitously among primary care providers, especially in rural areas. Female physicians used it more often. Users use it for an average of 40% of 03.03A office visits. Non-users of the code earned more income than their user-colleagues. Conclusion: We believe our findings will fill an important gap in informing the importance of an extended time service billing code in a fee-for-service system in reducing ED visits and hospitalizations. Advice and Lessons Learned: The fee-for-service time-modifier code, introduced in 2009, resulted in reduced ED visits and hospitalizations. It is likely that discontinuing the code would result in increased ED and hospital utilization, costing much more than removing the code would save. Usage of the time-modifier code was not uniform among primary care. Users of the code had different practice patterns and provider demographics. Our next step is to model the uptake of the code by primary care providers and explore the health system utilization and down-stream costs between users and non-users of the code.

Temporal Trends and Drivers of Heart Team Utilization in Transcatheter Aortic Valve Replacement: A Population‐Based Study in Ontario, Canada

Background The multidisciplinary Heart Team (HT) is recommended for management decisions for transcatheter aortic valve replacement (TAVR) candidates, and during TAVR procedures. Empiric evidence to support these recommendations is limited. We aimed to explore temporal trends, drivers, and outcomes associated with HT utilization. Methods and Results TAVR candidates were identified in Ontario, Canada, from April 1, 2012 to March 31, 2019. The HT was defined as having a billing code for both a cardiologist and cardiac surgeon during the referral period. The procedural team was defined as a billing code during the TAVR procedure. Hierarchical logistical models were used to determine the drivers of HT. Median odds ratios were calculated to quantify the degree of variation among hospitals. Of 10 412 patients referred for TAVR consideration, 5489 (52.7%) patients underwent a HT during the referral period, with substantial range between hospitals (median odds ratio of 1.78). Utilization of a HT for TAVR referrals declined from 69.9% to 41.1% over the years of the study. Patient characteristics such as older age, frailty and dementia, and hospital characteristics including TAVR program size, were found associated with lower HT utilization. In TAVR procedures, the procedural team included both cardiologists and cardiac surgeons in 94.9% of cases, with minimal variation over time or between hospitals. Conclusions There has been substantial decline in HT utilization for TAVR candidates over time. In addition, maturity of TAVR programs was associated with lower HT utilization.

Abstract P118: Machine Learning Algorithm For Improving Classification Of Myocardial Infarction Across A Diverse Population Using US Veteran Electronic Health Record Data

Introduction: The majority of population-based studies of myocardial infarction (MI) rely on billing codes for classification. Classification algorithms employing machine learning (ML) increasingly used for phenotyping using electronic health record (EHR) data. Hypothesis: ML algorithms integrating billing and information from narrative notes extracted using natural language processing (NLP) can improve classification of MI compared to billing code algorithms. Improved classification will improve power to compare risk factors across population subgroups. Methods: Retrospective cohort study of nationwide Veterans Affairs (VA) EHR data. MI classified using 2 approaches: (1) published billing code algorithm, (2) published phenotyping pipeline incorporating NLP and ML. Results compared against gold standard chart review of MI outcomes in 308 Veterans. We also tested known association between high density lipoprotein cholesterol (HDL-C) and MI outcomes classified using the 2 approaches among Black and White Veterans, stratified by sex and race; prior study showed HDL-C less protective for Black compared to White individuals. Results: We studied 17,176,658 million Veterans, mean age 69 years, 94% male, 12% self-report Black, 71% White. The billing code algorithm classified MI at positive predictive value (PPV) 0.64 compared to the published ML approach, PPV 0.90; the latter classified a modestly higher percentage of non-White Veterans. Using ML algorithm for MI, we replicated a reduced protective effect of HDL-C in Black vs White male and female Veterans (Table); with the billing code algorithm no association was observed between low density lipoprotein cholesterol (LDL-C) or HDL-C with MI among Black female Veterans. Conclusions: Using nationwide VA data, application of an ML approach improved classification of MI particularly among non-White Veterans, resulting in improved power to study differences in association for MI risk factors among Black and White Veterans.

Improving Advance Care Planning Documentation Using Reminders to Patients and Physicians: A Longitudinal Study in Primary Care

Objective: Documenting advance care planning (ACP) in primary care requires multiple triggers. New Medicare codes make it easier for providers to bill for these encounters. This study examines the use of patient and provider reminders to trigger advance care planning discussions in a primary care practice. Secondary outcome was billing of new ACP billing codes. Methods: Patients 75 years and older scheduled for a primary care appointment were screened for recent ACP documentation in their chart. If none was found, an electronic or mail message was sent to the patient, and an electronic message to their provider, about the need to have discussion at the upcoming visit. Chart review was performed 3 months after the visit to determine if new ACP discussion was documented in the chart. Results: In the 3 months after the reminder had been sent to patients and providers, new ACP documentation or billing was found in 28.8% of the patients. Most new documentation was health care decision maker (75.6% of new documentation) with new DNR orders placed for 32.3% of these patients. The new Medicare billing code was filled 10 times (7.8%). Conclusion: Reminders sent to both patients and providers can increase documentation of ACP during primary care visits, but rarely triggers a full ACP conversation.

A Rare Case of Pneumomediastinum after Eustachian Tube Dilation

Eustachian tube dilation (ETD) is a relatively new intervention for the treatment of eustachian tube dysfunction. Though it previously had no assigned billing code, the American Medical Association recently accepted a new Category I Current Procedural Terminology code application for ETD to be effective in January 2021. Reported complications are uncommon and usually minor. Herein, we present a rare case of massive pneumomediastinum following this procedure. Such major complications are critical to report as ETD becomes a more commonly practiced procedure.

Is There an Association between Herpetic Infections and Giant Cell Arteritis? A Population-Based Study

Recent data suggests that herpes zoster (HZ) and herpes simplex virus (HSV) may be one of the underlying immunological triggers for giant cell arteritis (GCA). However, there is limited population-based data to support this. Our goal was to determine if herpetic infections increase the likelihood of GCA in the British Columbia (BC) population. The background prevalence of GCA was compared to the prevalence of GCA in subjects with HZ and HSV using diagnostic billing code data from an online BC database (BC Data ScoutTM). BC residents ≥30 years old at the time of diagnosis from January 2000 to January 2019 were included. The relevant International Classification of Disease codes was used to identify patients with GCA, HZ, and HSV. Comparisons were made using two-sample Z tests. There were 4315 GCA diagnoses, from a total population of 3,026,005 subjects. The prevalence of GCA was 143 per 100,000 people. In terms of herpetic infections, 850 GCA cases were identified in 249,900 subjects with HZ versus 310 diagnoses of GCA in 163,170 subjects with HSV. The prevalence of GCA in subjects with HZ (0.340%) was significantly higher than the prevalence of GCA (0.143%) in the general population (p < 0.00001). The prevalence of GCA in HSV subjects (0.190%) was also significantly higher (p < 0.00001) than the population prevalence but lower than (p < 0.00001) the GCA with HZ prevalence. The likelihood of GCA appears to increase with herpetic infections, more significantly with HZ.

Exploring the Privacy-Preserving Properties of Word Embeddings: Algorithmic Validation Study

Background Word embeddings are dense numeric vectors used to represent language in neural networks. Until recently, there had been no publicly released embeddings trained on clinical data. Our work is the first to study the privacy implications of releasing these models. Objective This paper aims to demonstrate that traditional word embeddings created on clinical corpora that have been deidentified by removing personal health information (PHI) can nonetheless be exploited to reveal sensitive patient information. Methods We used embeddings created from 400,000 doctor-written consultation notes and experimented with 3 common word embedding methods to explore the privacy-preserving properties of each. Results We found that if publicly released embeddings are trained from a corpus anonymized by PHI removal, it is possible to reconstruct up to 68.5% (n=411/600) of the full names that remain in the deidentified corpus and associated sensitive information to specific patients in the corpus from which the embeddings were created. We also found that the distance between the word vector representation of a patient’s name and a diagnostic billing code is informative and differs significantly from the distance between the name and a code not billed for that patient. Conclusions Special care must be taken when sharing word embeddings created from clinical texts, as current approaches may compromise patient privacy. If PHI removal is used for anonymization before traditional word embeddings are trained, it is possible to attribute sensitive information to patients who have not been fully deidentified by the (necessarily imperfect) removal algorithms. A promising alternative (ie, anonymization by PHI replacement) may avoid these flaws. Our results are timely and critical, as an increasing number of researchers are pushing for publicly available health data.