Prescription Information Research Articles

Effect of physician prescribed information on hospital readmission and death after discharge among patients with health failure: A randomized controlled trial.

In order to understand if a physician prescribed medical information changes, the number of hospital readmission, and death among the heart failure patients. A 12-month randomized controlled trial was conducted (December 2013-2014). Totally, 120 patients were randomly allocated into two groups of intervention (n = 60) and control (n = 60). Accordingly, the control group was given the routine oral information by the nurse or physician, and the intervention group received the Information Prescription (IP) prescribed by the physician as well as the routine oral information. The data was collected via telephone interviews with the follow-up intervals of 6 and 12 months, and also for 1 year after the discharge. The patients with the median age of (IQR) 69.5 years old (19.8) death upon adjusting a Cox survival model, [RR = 0.67, 95%CI: 0.46-0.97]. Few patients died during 1 year in the intervention group compared to the controls (7 vs 15) [RR = 0.47, 95%CI: 0.20-1.06]. During a period of 6-month follow-up there was not statistically significant on death and readmission between two groups. Physician prescribed information was clinically and statistically effective on the reduction of death and hospital readmission rates among the HF patients in long term follow-up.

Characteristics and Clinical Course of Diabetes of the Exocrine Pancreas: A Nationwide Population-Based Cohort Studybrief Title: Diabetes of the Exocrine Pancreas

Background: The natural course of diabetes of the exocrine pancreas (DEP) is not well established. We aimed to compare the risk of insulin initiation, diabetic complications, and mortality between DEP and type 2 diabetes. Methods: Using the Korean National Health Insurance Service–Health Screening Cohort between 2012-2017, we divided diabetic patients into those with diabetes without prior pancreatic disease (indicated type 2 diabetes, n=153 894) and diabetes following pancreatic disease (indicated DEP, n=3661). ICD-10 codes and pharmacy prescription information were used to define type 2 diabetes, DEP, and acute and chronic diabetes complications. Reverse Kaplan-Meier curves were produced to compare insulin use over time between groups. We created logistic regression models for odds of progression to diabetic complications and mortality. Findings: DEP was associated with a higher risk of insulin use than type 2 diabetes (adjusted hazard ratio 1·25 at 1 year [95% confidence interval 1·16–1·35], p<0·0001; at 5 years 1·28 [1·20–1·35], p<0·0001). Individuals with DEP showed higher risks of hypoglycaemia (odds ratio 1·82 [1·52–2·18], p<0·0001), diabetic neuropathy (1·37 [1·27–1·48], p<0·0001), nephropathy (1·34 [1·24–1·45], p<0·0001), retinopathy (1·12 [1·03–1·22], p=0·0110), cardiovascular disease (1·50 [1·39–1·61], p<0·0001), cerebrovascular disease (1·24 [1·14–1·34], p<0·0001), and peripheral artery disease (1·24 [1·15–1·33], p<0·0001). All-cause mortality was higher in those with DEP (1·52 [1·37–1·70], p<0·0001) than in those with type 2 diabetes. Interpretation: DEP is more likely to require insulin therapy than type 2 diabetes. Hypoglycaemia, micro- and macrovascular complications, and all-cause mortality events are higher in DEP compared to type 2 diabetes. Awareness of these differences in the natural history of DEP may better enable clinicians to address the challenges posed by this diabetes sub-type and better achieve glycaemic control targets, avoid hypoglycaemia, and monitor and prevent complications. Funding: This study was funded by the National Research Foundation of Korea Ministry of Science and ICT (No. NRF-2018R1C1B5044056) and Korea Medical Institute. Declaration of Interest: None to declare. Ethical Approval: This study was approved by the Institutional Review Board of Ajou University Hospital (AJIRB-MED-EXP-19-106]

Analysis of drug-food interactions in inpatient treatment: A university hospital case

Patients' nutrition during inpatient treatment can reduce the pharmacodynamics of drugs. Therefore, monitoring of drug-nutrient interactions is essential for patient safety. Pharmaceutical Data Banks (PDB) databases provide information regarding potential drug-drug, drug-food, and drug-allergy interactions. When Clinical Decision Support Systems (CDSS) are integrated with PDBs, drug-drug and drug-allergy interactions can be prevented when physicians prescribe drugs and when pharmacists evaluate those prescriptions. However, nutrition planning is done by dieticians, and it is not common practice for dieticians to use CDSSs integrated with PDB to access patient prescription information. This study aims to measure drug-food interactions in hospitals where physicians and pharmacists use CDSSs integrated with PDBs. For the most part, dieticians plan patient diets according to the patient's primary disease (diabetes, etc.) and do not access prescription data. We cooperated with a university hospital in Turkey, accredited by HIMSS in 2017 at EMRAM Stage 6, to monitor hospitalized patients for at least one week in 2018. According to the findings, it was determined that 1,451 different drugs were administered 1,620,573 times to a total of 27,455 patients. It was determined that eight (0.55%) different drugs administered to 581 (2.1%) of the patients could interact with food and that these eight drugs were prescribed 8,089 times (0.49%) during the observation period. Although some drug-nutrient interactions were documented due to the study, the number of detected and documented interactions and their severity were relatively low. Precautions taken by dieticians, such as completely removing certain nutrients, like grapefruit, from the diet list, seem to be effective in preventing common interactions. To eliminate drug-nutrient interactions, it will be beneficial for dieticians to access patients' prescribing information and use the CDSS integrated with PDB.

Unified Health Information Prescription System for Communicable Disease COVID19: Strategy to Deal with Infodemic

Unified Health Information Prescription System for Communicable Disease COVID19: Strategy to Deal with Infodemic

214. Antibiotic Use for Common Infections in British Columbia: A Review of Outpatient Prescribing from 2000 - 2018

BackgroundAntimicrobial resistance continues to jeopardize the future of modern medicine; as 92% of all antibiotics are used in the community, it is imperative to parse outpatient prescribing. In British Columbia (BC), efforts to curb the use of these essential medications have included: stewardship campaigns, practitioner guidelines, and vaccine scheduling amendments. This study reviews the trends in antibiotic use over the past two decades to identify new targets for Provincial stewardship and intervention.MethodsAntibiotic prescription information was extracted from PharmaNet, a centralized data system that links all pharmacies with prescriptions dispensed in the community setting. The Medical Service Plan records reimbursement claims submitted by physicians for services provided, including diagnostic codes. Antibiotic prescriptions were extracted from PharmaNet and then matched to the billing system using anonymized patient identifiers. Prescription rates were calculated, and trends were examined by major anatomical therapeutic chemical (ATC) classification.ResultsOur study included 3,564,258 individuals over an 18-year period, with a total of 26,108,576 antibiotic prescriptions issued, for common infections. Overall antibiotic utilization decreased 18% (from 228 to 187 prescriptions per 1000 population) over the course of the study period. This trend was reflected in both Beta-Lactam (-37%) and Macrolide (-50%) antibiotics; two of the most common classes prescribed in the outpatient setting. A significant outlier was the J01X class of Other Antibacterials, which increased by a staggering 218%, by 2018. Further analyses are currently underway to stratify these changes in magnitude by demographic variables to identify specific, new targets for stewardship.Rates of outpatient antibiotic prescriptions, for common infections, per 1000 population, by major ATC class, over time. ConclusionOutpatient antibiotic prescribing has decreased steadily since 2000. These promising results can be ascribed to the various Provincial initiatives to quell the misuse of these medications. However, many of the indications tied to these prescriptions do not warrant the use of antibiotics, and further analyses are necessary to evaluate prescribing quality to fully delineate the state of antibiotic use in BC. Next steps also include comparing BC rates with Ontario, another large province of Canada.Disclosures All Authors: No reported disclosures

EVALUATION OF PRESCRIBING INDICATORS AND PRESCRIPTION WRITING AT THE OUTPATIENT DEPARTMENT OF A PEDIATRIC TEACHING HOSPITAL IN LIBYA

Background:Irrational drug prescribing and inappropriate prescription writing have a serious medical and economicimpacton patients and society. Information on pediatric prescribing and quality of prescription writing in Libya are lacking. The aim of this study was to evaluate the extent of rational prescribing using the WHO indicatorsand to assessthe completeness of prescriptions recorded informationat the outpatient department of Benghazi Childrens Hospital. Results:A total of 603 prescriptions were sampled and analyzed. Concerning the prescribing indicators, the average number of medicines per encounter was 1.52, 5.47% of drugs were prescribed by generic name, 20.56% of encounters had at least one antibiotic prescribed, 25.87% of encounters were prescribed one injection or more and 61.27% of the prescribed drugs were from the Essential Drugs List. As regards the completeness of the recorded data on prescriptions, the patients name, gender, age and diagnosis were recorded in 99.34%, 10.78%, 85.74% and 29.85% of prescriptions, respectively. None of the prescriptions included the patients contact details and only 0.66% mentioned the patients weight. Prescribers information such as the name, signature and contact details were present in more than half of prescriptions while only 3.98% were stamped.Medication details like the dosage form, dose and frequency were written for more than 82% of the prescribed drugs whereas the strength and treatment duration were the least recorded drug information, 46.28% and 56.67%, respectively. Conclusion:Some forms of irrational drug prescribing as well as suboptimal recording of prescriptions information were observed at the outpatient department of Benghazi Childrens Hospital.

Rapid Antigen Detection Test Using Rates of Family Physicians, Test Results and Its Impact on Their Prescription Behaviours

Aim: The aim of this study is to determine the rates of rapid antigen detection test use, test results and its relationship with Modified Centor Score and investigate the effect of rapid antigen detection test use on prescribing behaviour. Methods: This research was conducted by scanning the family physicians’ information system retrospectively. From the Family Medicine Information System, where polyclinics and prescription records of the family health centers where 12 family physicians work in Kayseri are located, the number and results of rapid antigen detection tests performed between February 2017 and April 2019, the Modified Centor Score automatically calculated by the system, the number of prescribed antibiotics and symptomatic treatment information was recorded. Results: The positivity of rapid antigen detection test was considered, cut-off value of Modified Centor Score was &gt;2 (ROC EAA:63.9, Cl:0.95,0.61-0.66). Sensitivity rate of the test for diagnostic score was 86% and selectivity rate was 48.8%. Conclusion: Family physicians should be encouraged to use rapid antigen detection test. The fact that antibiotics are prescribed in rapid antigen detection testnegative patients may indicate that confidence level of the test is less than physical examination and clinical findings. Keywords: sore throat, primary care, score

Text Formatting Based on Keyword Detection

Adverse drug effects are a major cause of death across the world each year because of prescription errors. Many of such errors involve the administration of the wrong drug or dosage by care givers to patients due to indecipherable handwritings, drug interactions, confusing drug names etc. The adoption of voice-based prescription project could eliminate some of these errors because they allow prescription information to be captured and heard through voice response rather than in the physician’s handwriting. Our project will generate an electronic prescription using a “Speech to Text converter” (Perceptual Linear Prediction (PLP)) and capture the data from the keywords spoken by doctor(s). There won’t be any need to carry paper prescriptions on revisiting doctors. A patient will be able to share his historic medical records to a new doctor. This project also provide facility to sign the prescription and send to the patient directly on his phone and email id. The System enables the patient to manage the privacy of their personal health record. This project is proposed to target those doctors and clinics that are still using paper-based handwritten prescriptions

Factors Impacting Pharmaceutical Prices and Affordability: Narrative Review.

Increasing prescription drug pricing often reflects additional work stress on medical professionals because they function as financial advisors for patients and help them manage out-of-pocket expenses. Providers or prescribers wish to help patients with prescription costs but often lack related information. Healthcare plan providers try to display prescription and drug cost information on their websites, but such data may not be linked to electronic prescription software. A mark-up is defined as the additional charges and costs that are applied to the price of a product for the purpose of covering overhead costs, distribution charges, and profit. Therefore, the policies implemented in the pharmaceutical distribution chain might include the regulation of wholesale and retails mark-ups and pharmaceutical remuneration. If mark-ups are regulated, countries are highly recommended to use regressive mark-ups rather than fixed percentage mark-ups. This narrative review provides insights into the framework of pharmaceutical mark-up systems by describing different factors impacting pharmaceutical prices and affordability. These include the interplay of medicine pricing and the supply chain, the impact of pertinent laws and regulation and out-of-pocket expenditure.

Health Information Prescription System For Non Communicable Diseases: A Systematic Review And Thematic Analysis

Objective ― Prescribing health information is very important to empower informed patients. The goal of present study is to recognize challenges for developing health information prescription on non-communicable diseases. Material and Methods ― Six data bases related to health information prescription were investigated. They included Web of Science, Scopus, PubMed, Google Scholar, Ovid and EMBASE. The investigated studies were published from 2000 to 2019. The language of the articles was English and the access to full text was one of the inclusion criteria. The research was evaluated by Prisma checklist and critical apprising. Different dimensions of health Information prescription system were recognized by thematic analysis. Results ― 54 studies were recognized based on the inclusion criteria. The results showed that there were three main concepts related to primary dimensions of the system in designing health information prescription system: determination of system functional goals, recognition of implementation barriers and recognition of developmental sub-structures. There were 16 subcategories including primary goals for accessibility, the concerns related to patients information confidentiality, individual differences and interests and personalizing the process of information prescription, the lack of integrity in health Information system for providing pattern of health Information system related to diabetic patients. Conclusion ― The goals, implementing barriers and functional substructures of health information prescription system should be recognized in order to improve self-care behaviors of diabetic patients in clinic. It is recommended that the future investigations focus on research gaps in personalizing health information prescription and integration of health information prescription process in health care system.

Testosterone Therapy and Cardiovascular Risk: A Critical Analysis of Studies Reporting Increased Risk

BackgroundTreatment of “adult-onset hypogonadism” (AOH) with exogenous testosterone therapy (TTh) to raise serum testosterone (T) levels may influence cardiovascular (CV) risk factors in patients with AOH, whereas low endogenous T levels are associated with an increased CV risk and mortality. AimTo critically evaluate studies reporting increased CV risk associated with TTh and to provide an overview of the risks and benefits of restoring T levels through exogenous TTh. MethodsA review of publications focusing on the association between TTh and increased CV risk was conducted, and the study methodologies and conclusions of each were critically evaluated. Further, recent clinical and epidemiological studies associating AOH or TTh with a change in CV risk, and pertinent hematologic and vascular effects noted in animal studies and in vitro, as well as in clinical practice were also reviewed. OutcomesA review of the literature shows that untreated testosterone deficiency and/or low T is associated with an increase in CV risk and adverse outcomes, with numerous studies and meta-analyses to support a positive association between exogenous TTh and an improvement in CV risk factors in men with AOH. ResultsNumerous studies in the literature demonstrate the positive benefits of using TTh; however, since 2013, some publications have suggested a link to increased CV risk associated with TTh. A number of these studies retrospectively analyzed insurance claims databases using diagnosis codes, procedures codes, and prescription information. Many reviews published since have pointed out the methodological flaws and debatable conclusions of these studies. Clinical ImplicationsA careful assessment of the patient's current health status and CV risk factors should be weighed against the benefits and possible risks resulting from TTh, and consideration should be given to deferring treatment pending resolution or stabilization of CV disease or risk factors. Strengths & LimitationsIn this review, we provide an in-depth analysis of studies reporting increased CV risk with TTh. Many of the studies were not well-designed, randomized, double-blind, prospective clinical trials but rather post hoc analyses of cohort data. These studies may reflect bias in how treatment and nontreatment decisions are made or reflect conclusions based on widely cited methodological flaws. ConclusionAppropriate patient selection supported by low pre-treatment T levels and monitoring T levels during treatment with the goal of achieving and maintaining physiologic levels all contribute to the safe and effective use of TTh in men with AOH.Khera M, Miner M, Jaffe J, et al. Testosterone Therapy and Cardiovascular Risk: A Critical Analysis of Studies Reporting Increased Risk. J Sex med 2021;18:83–98.

Design considerations for a remote randomized multi-site clinical trial evaluating an e-health self-management program for chronic pain patients receiving opioid therapy

Based on the adverse consequences and inadequate evidence of effectiveness for long-term opioid therapy (LOT), the CDC developed recommendations to decrease the use of LOT and morphine equivalent dose (MED) for patients receiving LOT. However, the majority of these patients report that opioid medication is significantly beneficial for pain management and are hesitant to reduce/decrease its use. Compounding the problem is poor access to non-pharmacologic therapies for many patients due to insurance reimbursement structures and limited pain-service availability. EMPOWER is an intent-to-treat, two-arm, open-label, randomized controlled trial evaluating a web-based self-management chronic pain program (E-Health) that has been found to reduce self-reported MED, while also decreasing pain, in two randomized controlled trials. Approximately 400 chronic pain patients receiving LOT at a daily average prescribed MED ≥ 20 mg at one of two U.S. healthcare systems, located in North Carolina and Ohio, will be randomized in a 1:1 ratio to treatment as usual (TAU) or TAU plus E-Health (E-Health+). TAU consists of LOT from a prescribing clinician. E-Health+ participants are provided with a 4-month E-Health subscription (active treatment phase). All participants will complete web-based self-report measures at baseline, the end of the active treatment phase, and 6-months post-active treatment. Opioid prescription information will be collected from the participants' electronic health record (EHR) from baseline through 6 months post-active treatment. This paper describes design considerations for this unique trial which is conducted completely remotely, with no in-person visits, and utilizes the EHR for participant identification and primary outcome collection.

Linked administrative data’s role in Victoria’s first social impact investment, journey to social inclusion, working to end chronic homelessness

IntroductionIndigenous people worldwide are overrepresented and adversely effected by diabetes and its complications. Optimal glycemic control and lipid monitoring is fundamental to the management of diabetes. This study linked population level data to assess monitoring, treatment and control of blood sugars and lipids in First Nations’ people in Ontario. Objectives and ApproachWe linked 17 Ontario population-based health administration datasets at the individual level with the Indian Register dataset. The latter provides information on all registered or Status First Nations people in Canada . Age and sex-adjusted rates of HbA1c and lipid monitoring were calculated for each 12-month period from April 1, 1995, to March 31, 2015.). We assessed the proportion of individuals with diabetes whose HbA1c and lipid values were controlled. To capture prescriptions for antidiabetic drugs, we used the Drug Identification Number database to identify all antidiabetic drugs and linked these to the Ontario Drug Benefit database to capture prescription information. ResultsCompared with other people in Ontario, First Nations people with diabetes are monitored less for key indicators of diabetes control. In 2014/15, 37.0% of First Nations people with diabetes living in First Nations communities had their blood sugar levels monitored compared to 45.0% of other people in Ontario. A similar pattern was shown for lipid level monitoring, with 48.3% of First Nations people living in First Nations communities, and 65.8% of other people in Ontario having recorded lipid measurements. Conclusion / ImplicationsEarly screening for complications and screening for hemoglobin A1c is strongly recommended.

Prescription patterns of antiepileptic drugs for adult patients with newly diagnosed focal epilepsy from 2006 to 2017 in Japan

ObjectiveThis study aimed to examine prescription patterns of antiepileptic drugs (AEDs) for adult patients with newly diagnosed focal epilepsy in Japan and whether these patterns adhere to the 2010 Japanese Society of Neurology Guidelines of Epilepsy Treatment. MethodsData from the JMDC Claims Database were obtained for patients aged between 20 and 65 years with newly diagnosed focal epilepsy who were prescribed AEDs between 2006 and 2017. Available prescription information up to the patient’s first year was recorded and longitudinal descriptive statistics, Cochran Armitage Trend (CAT) tests, and annual percentage change (APC) were used to analyze AED trends and overall guideline adherence. In addition, logistic regression analyses were used to compare these results across different health facilities. ResultsA total of 6024 adult patients with newly diagnosed focal epilepsy were enrolled. The prescription of new AEDs increased significantly (CAT, p < 0.001, APC = 28.74 %) up to 36.8 % of all prescriptions in 2017 when compared to 2006. Among new AEDs, prescriptions for levetiracetam increased most rapidly and were followed by lamotrigine. In contrast, prescriptions for older AEDs, especially valproate, decreased over this same time period. The average guideline adherence rate from 2010 to 2017 was 75.3 %, and was not significantly different over time (CAT, p = 0.55). Health facilities with either more than 500 beds or between 20–499 beds had higher odds of prescribing new AEDs and improved guideline adherence when compared to facilities with 0–19 beds. ConclusionPrescription patterns of AEDs for adult patients with newly diagnosed focal epilepsy exhibited a trend from older to new AED classes between 2006 and 2017, with consistent, high guideline adherence from 2010 to 2017. Health facilities with 0–19 beds were less likely to prescribe new AEDs and completely adhere to proposed guidelines.

Functional clustering methods for longitudinal data with application to electronic health records.

We develop a method to estimate subject-level trajectory functions from longitudinal data. The approach can be used for patient phenotyping, feature extraction, or, as in our motivating example, outcome identification, which refers to the process of identifying disease status through patient laboratory tests rather than through diagnosis codes or prescription information. We model the joint distribution of a continuous longitudinal outcome and baseline covariates using an enriched Dirichlet process prior. This joint model decomposes into (local) semiparametric linear mixed models for the outcome given the covariates and simple (local) marginals for the covariates. The nonparametric enriched Dirichlet process prior is placed on the regression and spline coefficients, the error variance, and the parameters governing the predictor space. This leads to clustering of patients based on their outcomes and covariates. We predict the outcome at unobserved time points for subjects with data at other time points as well as for new subjects with only baseline covariates. We find improved prediction over mixed models with Dirichlet process priors when there are a large number of covariates. Our method is demonstrated with electronic health records consisting of initiators of second-generation antipsychotic medications, which are known to increase the risk of diabetes. We use our model to predict laboratory values indicative of diabetes for each individual and assess incidence of suspected diabetes from the predicted dataset.

Impact of the zero-mark-up drug policy on drug-related expenditures and use in public hospitals, 2016–2018: an interrupted time series study in Shaanxi

ObjectiveThe aim of this study was to measure the impact of zero-mark-up drug policy (ZMDP) on drug-related expenditures and use in urban hospitals.DesignThis was a retrospective observational study of trends...

ESSD Commentary on Dysphagia Management During COVID Pandemia.

Since the World Health Organization declared the COVID-19 pandemic a Global Public Health Emergency, experts in swallowing are seeking guidance on service delivery and clinical procedures. The European Society for Swallowing Disorders provides considerations to support experts in swallowing disorders in clinical practice. During the COVID-19 pandemic, assessment and treatment of patients with oropharyngeal dysphagia should be provided, while at the same time balancing risk of oropharyngeal complications with that of infection of patients and healthcare professionals involved in their management. Elective, non-urgent assessment may be temporarily postponed and patients are triaged to decide whether dysphagia assessment is necessary; instrumental assessment of swallowing is performed only if processing of the instruments can be guaranteed and clinical assessment has not provided enough diagnostic information for treatment prescription. Assessment and management of oropharyngeal dysphagia is a high-risk situation as it must be considered an aerosol-generating procedure. Personal protective equipment (PPE) should be used. Telepractice is encouraged and compensatory treatments are recommended.

Testing the Utility of Electronic Health Records and Health Information Network Data to Determine Potential Health Impacts Associated with Opportunities for Exposure to Environmental Contaminants

It was hypothesized that health care claims (HCC) and electronic health records (EHRs) might provide the opportunity to access more timely and geographically refined data, and more information than is typically available from traditional public health surveillance systems; for example, measurements of risk factors (such as high cholesterol), co-diagnosis, medication prescription information and laboratory test results. We designed a pilot study in the state of Delaware (DE) to explore the utility of these data as a surveillance tool. The city of New Castle, DE was selected because: (1) it had levels of contaminants in drinking water that exceeded federal advisory levels and (2) it has universal claims and access to EHR data through the Delaware Health Information Network (DHIN).We included residents who lived in New Castle City during the period 2013 to 2018. We selected the following outcomes to focus on based on the contaminants of interest: cardiovascular disease; thyroid disease; kidney cancer; and testicular cancer. We are in the process of defining the inclusion and exclusion criteria for de identified laboratory and clinical diagnosis data on these outcomes, to identify HCC and EHR datasets for analysis. We will analyze the data by zip code and census tract, and include demographic information (age, race/ethnicity, and gender) for each outcome of interest. We will also compare rates of health outcomes obtained using HCC data to those obtained using EHR data.The overall goal is to determine the value of conducting ecologic analyses using HCC and EHR data to inform the design of larger, long term epidemiological studies evaluating exposure to environmental contaminants and potential health impacts. The need to identify such surveillance tools is critical as existing disease registries are often not timely enough due to reporting requirements and disease latency and/or are limited by small numbers of events.

Challenges Of The Clinical Machine Learning Implementation For The Prospective System For High Intensity Evaluation During Radiation Therapy (SHIELD-RT) Study

Patients undergoing outpatient cancer radiation therapy (RT) or chemoradiation (CRT) may require evaluation in the emergency department (ED) or hospitalization. We previously developed a machine learning (ML) algorithm utilizing electronic health records (EHR) to predict risk of such events, and this algorithm was used to stratify high risk patients for randomization as part of the recently completed prospective System for High Intensity Evaluation During Radiation Therapy (SHIELD-RT) study. This study demonstrated that ML-directed mandatory supplemental evaluations reduced the rate of acute care visits during RT and CRT. Here we describe the workflow report on the operational challenges encountered while using the oncology information system (OIS) ARIA (Varian Medical Systems, Palo Alto) for clinical implementation of the ML algorithm. In this single institution randomized QI study (NCT04277650), all outpatient adult courses of RT and CRT started from January 7, 2019 to June 30, 2019 were evaluated during their first week of treatment by the ML algorithm. New courses of treatment in the prior week were identified via the OIS, using a preexisting OIS designation of patient appointments as “new start” (new patient beginning new course), “old start” (previously treated patient starting new course) or “final treatment” (either final fraction of multi-fraction treatment or single fraction treatment). To ensure that the ML algorithm was run on newly started courses, “old start” and “final treatment” appointments were checked to verify the appointment corresponded to new treatment courses or to new single fraction treatment courses, respectively. The final patient list was used to query pre-treatment EHR histories and treatment prescription information. Weekly time requirements by the physics staff were documented over the course of the study. Manual review of patient and treatment prescription data from the OIS required on average 4.9 hours per week. Patient appointments labeled “old start” sometimes corresponded to the start of a new treatment plan (i.e. boost plans) during a treatment course that was already underway and previously evaluated for the study. At the time of the treatment start, some sequential boost prescriptions were entered as drafts that required systematic confirmation. Ideally, extraction of patient and radiation therapy treatment plan data would happen on a rolling basis as patients started treatment, but the need for review of the data limited data extraction to once weekly. The workflow for implementation will be presented at the conference. Our study demonstrated a successful clinical implementation of an ML-based algorithm, but manual review of treatment course identification for algorithm assessment was required. Full operationalization beyond the SHIELD-RT study would necessitate changes to data management practices in the OIS centering on identifying new starts.

Quantifying the Causal Effects of Conversational Tendencies

Understanding what leads to effective conversations can aid the design of better computer-mediated communication platforms. In particular, prior observational work has sought to identify behaviors of individuals that correlate to their conversational efficiency. However, translating such correlations to causal interpretations is a necessary step in using them in a prescriptive fashion to guide better designs and policies. In this work, we formally describe the problem of drawing causal links between conversational behaviors and outcomes. We focus on the task of determining a particular type of policy for a text-based crisis counseling platform: how best to allocate counselors based on their behavioral tendencies exhibited in their past conversations. We apply arguments derived from causal inference to underline key challenges that arise in conversational settings where randomized trials are hard to implement. Finally, we show how to circumvent these inference challenges in our particular domain, and illustrate the potential benefits of an allocation policy informed by the resulting prescriptive information.