Medical Administrative Database Research Articles

Evaluation of bias and gender/racial concordance based on sentiment analysis of narrative evaluations of clinical clerkships using natural language processing

There is increasing interest in understanding potential bias in medical education. We used natural language processing (NLP) to evaluate potential bias in clinical clerkship evaluations. Data from medical evaluations and administrative databases for medical students enrolled in third-year clinical clerkship rotations across two academic years. We collected demographic information of students and faculty evaluators to determine gender/racial concordance (i.e., whether the student and faculty identified with the same demographic). We used a multinomial log-linear model for final clerkship grades, using predictors such as numerical evaluation scores, gender/racial concordance, and sentiment scores of narrative evaluations using the SentimentIntensityAnalyzer tool in Python. 2037 evaluations from 198 students were analyzed. Statistical significance was defined as P < 0.05. Sentiment scores for evaluations did not vary significantly by student gender, race, or ethnicity (P = 0.88, 0.64, and 0.06, respectively). Word choices were similar across faculty and student demographic groups. Modeling showed narrative evaluation sentiment scores were not predictive of an honors grade (odds ratio [OR] 1.23, P = 0.58). Numerical evaluation average (OR 1.45, P < 0.001) and gender concordance between faculty and student (OR 1.32, P = 0.049) were significant predictors of receiving honors. The lack of disparities in narrative text in our study contrasts with prior findings from other institutions. Ongoing efforts include comparative analyses with other institutions to understand what institutional factors may contribute to bias. NLP enables a systematic approach for investigating bias. The insights gained from the lack of association between word choices, sentiment scores, and final grades show potential opportunities to improve feedback processes for students.

Renal outcomes with sodium-glucose cotransporter 2 inhibitors in Japanese people with grade 3 chronic kidney disease and type 2 diabetes: Analysis of medical administrative databases.

To evaluate whether sodium-glucose cotransporter 2 inhibitor (SGLT2i) therapy is associated with a reduction of renal events compared with other glucose-lowering drugs (oGLDs) among Japanese people with type 2 diabetes (T2D) and grade 3 (G3) chronic kidney disease (CKD) in a real-world clinical practice setting. People with T2D who were newly prescribed an SGLT2i or an oGLD from April 2014 to November 2021 (without prior use of index drugs for ≥ 1 year prior to index date) and G3 CKD (estimated glomerular filtration rate [eGFR] ≥ 30 to < 60 mL/min/1.73 m2) were selected from the Medical Data Vision database (MDV-DB) and the Real-World Data database (RWD-DB). SGLT2i and oGLD users were matched (1:1) using propensity score on patient background characteristics. The primary endpoint was a composite of the development of end-stage kidney disease or a sustained decline in eGFR of 50% or more. Hazard ratios (HRs) were estimated using the Cox proportional hazards model. Overall, 3190 (1595 per group) patients in the MDV-DB and 2572 (1286 per group) patients in the RWD-DB were included in the analyses. The composite outcome was significantly lower in the SGLT2i group than in the oGLD group in the MDV-DB (HR 0.49, 95% confidence interval [CI] 0.33 to 0.74, P < 0.001) and in the RWD-DB (HR 0.57, 95% CI 0.37 to 0.88, P = 0.011). Japanese people with T2D and G3 CKD initiating an SGLT2i had a lower risk of renal events than people initiating an oGLD.

Improved Activities of Daily Living With Adjunctive Intravenous Steroids in Bacterial Meningitis: A Nationwide, Population-Based Medical Database Study.

The benefit of using adjunctive intravenous steroids (IVS) to reduce the neurological sequelae in bacterial meningitis remains inconclusive. This study evaluated the effect of IVS on improving the subsequent Activities of Daily Living (ADL) in bacterial meningitis by analyzing data from a large nationwide administrative medical database in Japan. Data from 1,132 hospitals, covered by the administrative Diagnosis Procedure Combination (DPC) payment system from 2016 to 2022, were evaluated. The ADL levels at admission and discharge were measured using the Barthel Index (BI). Out of the cumulative 47,366,222 patients hospitalized, 8,736 were diagnosed with acute bacterial meningitis and had BI data available. The BI at discharge, adjusted for sex, age, and BI at admission, was significantly better among those treated with IVS (p<0.0001). Exploratory subgroup analyses suggested that this benefit is expected across a broad spectrum of bacterial species. In summary, the use of IVS for improving the subsequent ADL level in bacterial meningitis was suggested.

Real-world management of patients with neuromyelitis optica spectrum disorder using satralizumab: Results from a Japanese claims database

BackgroundSatralizumab, a humanized anti-interleukin-6 receptor monoclonal antibody, has been approved globally for the treatment of neuromyelitis optica spectrum disorder (NMOSD), based on positive results from two randomized, double-blind, phase 3 studies: SAkuraSky (NCT02028884) and SAkuraStar (NCT02073279). There remains an unmet need to understand the real-world management of NMOSD, especially in patients undergoing tapering of concomitant therapy. We examined real-world treatment patterns, including concomitant glucocorticoids and immunosuppressants, and relapse in satralizumab-treated patients with NMOSD, using a Japanese administrative hospital claims database. MethodsWe used retrospective data from the Medical Data Vision hospital-based administrative claims database. The index date was the date of first satralizumab prescription and the study period was set between August 2018 and March 2022. Patients were included in the overall population if they had a first prescription for satralizumab between August 2020 and March 2022, an International Classification of Disease, Version10 code of G36.0 prior to March 2022, and were observable for ≥90 days prior to the index date. The primary endpoint was the percentage of patients with relapse-free reduction of oral glucocorticoids to 0 mg/day at 360 days of continued satralizumab treatment. Secondary endpoints included time to relapse, number of relapses after the index date while being on continuous satralizumab treatment, annualized relapse rate before and after the index date, and concomitant medication use. Relapse and dose reduction were identified using definition specifically developed for this study. ResultsOf the 131 patients included in the overall population, most were female (90.8 %), aged 18–65 years (75.6 %), and were prescribed oral glucocorticoids (93.1 %). Azathioprine (19.1 %) and tacrolimus, a calcineurin inhibitor (18.3 %), were the most common immunosuppressants at index date. Six (4.6 %) patients had a history of biologic use (tocilizumab, 1 [0.8 %]; eculizumab, 5 [3.8 %]). Among 111 patients observable for 360 days pre-index, there were 0.6 ± 0.8 (mean ± SD) relapses during 360 days before the index date. The median (interquartile range) duration of satralizumab exposure was 197.0 (57.0–351.0) days. Most (125/131; 95.4 %) patients were relapse-free post-index; 6 (4.6 %) patients relapsed within 90 days after the index date, of which 2 had the first relapse within 7 days after the index date. Among 21 patients with 360-day follow-up, 6 (28.6 %) patients were on 0 mg/day dose of glucocorticoid prescription without relapse 360 days post-index. Of these 6 patients, 2 had no prescription of oral glucocorticoids at the index date and remained glucocorticoid- and relapse-free 360 days after the index date. ConclusionThese real-world data support the phase 3 clinical trials. Our results, over a median duration of satralizumab exposure of 197.0 days, showed that a majority (125/131, 95.4 %) of patients were relapse-free after initiating satralizumab treatment. The number of glucocorticoid-free patients without relapse increased over time under continuous satralizumab prescription. Further studies are needed to confirm if satralizumab can be used as a potential immunosuppressant- and glucocorticoid-sparing agent.

L'avenir des registres

Medical registries play a crucial role in contemporary medicine by providing valuable information for clinical practice, research, and health policies. Systematic performance assessment, including patient input, linkage with medical-administrative databases, integration of randomized studies, and the use of artificial intelligence, mark an exciting new era for medical registries. These advancements will transform the way we understand, prevent, and treat diseases, paving the way for a more personalized, predictive, efficient, and likely cost-effective medicine.

Profiles of patients with substance-related disorders who dropped out or not from addiction treatment

This longitudinal study identified profiles of patients with substance-related disorders (SRD) who did or did not drop out of specialized addiction treatment, integrating various patterns of outpatient service use. Medical administrative databases of Quebec (Canada) were used to investigate a cohort of 16,179 patients with SRD who received specialized addiction treatment. Latent class analysis identified patient profiles, based on multi-year outpatient service use. Four patient profiles related to treatment dropout were identified: patients who did not drop out and were low service users (Profile 1); patients who did not drop out and were high service users (Profile 2); patients who dropped out and were low service users (Profile 3); patients who dropped out and were high service users (Profile 4). Profile 1 had the best health and social conditions, while Profile 4 had the worst. The risks of being frequent emergency department users, being hospitalized or dying were highest in Profile 4, followed by Profiles 3, 2 and 1. Assertive treatment programs may be suited to Profile 4 and intensive case management programs to Profile 3. Collaborative care with higher psychosocial interventions and regularity of care may be extended to Profile 2 and interventions integrating motivational treatment to Profile 1.

1-year functional outcomes after cardiopulmonary resuscitation for older adults with pre-existing long-term care needs.

To investigate the 1-year functional outcomes after cardiopulmonary resuscitation (CPR) in adults aged ≥65years with pre-existing long-term care needs. This population-based cohort study was conducted in Tochigi Prefecture, one of 47 prefectures in Japan. We used medical and long-term care administrative databases, which included data on functional and cognitive impairment that were assessed with the nationally standardised care-needs certification system. Among individuals aged ≥65years registered between June 2014 and February 2018, patients who underwent CPR were identified. The primary outcome was mortality and care needs at 1 year after CPR. The outcome was stratified by pre-existing care needs before CPR based on the total daily estimated care minutes: no care needs, support levels 1 and 2 and care-needs level 1 (estimated care time 25-49min), care-needs levels 2 and 3 (50-89min) and care-needs levels 4 and 5 (≥90min). Among 594,092 eligible individuals, 5,086 (0.9%) underwent CPR. The 1-year mortalities after CPR in patients with no care needs, support levels 1 and 2 and care-needs level 1, care-needs levels 2 and 3 and care-needs levels 4 and 5 were 94.6% (n = 2,207/2,332), 96.1% (n = 736/766), 94.5% (n = 930/984) and 95.9% (n = 963/1,004), respectively. Among survivors, most patients had no change in care needs before and at 1 year after CPR. There was no significant association between pre-existing functional and cognitive impairment and 1-year mortality and care needs after adjusting for potential confounders. Healthcare providers need to discuss poor survival outcomes after CPR with all older adults and their families in shared decision making.

Risk of non-hypoglycemic agents for hypoglycemia-related hospitalization in patients with type 2 diabetes: a large-scale medical receipt database analysis

IntroductionHypoglycemia is listed as an adverse effect in the package inserts of not only hypoglycemic agents but also many other drugs. We aimed to clarify real-world factors related to an...

One-Year Functional Outcomes After Invasive Mechanical Ventilation for Older Adults With Preexisting Long-Term Care-Needs.

To examine 1-year functional outcomes after invasive mechanical ventilation for adults greater than or equal to 65 years with preexisting long-term care-needs. We used medical and long-term care administrative databases. The database included data on functional and cognitive impairments that were assessed with the national standardized care-needs certification system and were categorized into seven care-needs levels based on the total daily estimated care minutes. Primary outcome was mortality and care-needs at 1 year after invasive mechanical ventilation. Outcome was stratified by preexisting care-needs at the time of invasive mechanical ventilation: no care-needs, support level 1-2 and care-needs level 1 (estimated care time 25-49 min), care-needs level 2-3 (50-89 min), and care-needs level 4-5 (≥90 min). A population-based cohort study in Tochigi Prefecture, one of 47 prefectures in Japan. Among people greater than or equal to 65 years old registered between June 2014 and February 2018, patients who received invasive mechanical ventilation were identified. None. Among 593,990 eligible people, 4,198 (0.7%) received invasive mechanical ventilation. The mean age was 81.2 years, and 55.5% were male. The 1-year mortality rates after invasive mechanical ventilation in patients with no care-needs, support level 1-2 and care-needs level 1, care-needs level 2-3, and care-needs level 4-5 at the time of invasive mechanical ventilation were 43.4%, 54.9%, 67.8%, and 74.1%, respectively. Similarly, those with worsened care-needs were 22.8%, 24.2%, 11.4%, and 1.9%, respectively. Among patients in preexisting care-needs levels 2-5 who received invasive mechanical ventilation, 76.0-79.2% died or had worsened care-needs within 1 year. These findings may aid shared decision-making among patients, their families, and heath care professionals on the appropriateness of starting invasive mechanical ventilation for people with poor functional and cognitive status at baseline.

Risk of a second cancer and infection in patients with indolent B-cell lymphoma exposed to first-line bendamustine plus rituximab: A retrospective analysis of an administrative claims database.

Bendamustine has a potent immunosuppressive effect because it causes T-cell lymphopenia, which might lead to a second primary malignancy (SPM) and would increase the risk of infection. Using the Medical Data Vision administrative claims database, we compared the cumulative incidence of SPM, infections within 6months, and overall survival (OS) among untreated patients with indolent B-cell lymphomas (iBCL) who received rituximab-based chemotherapy between 2009 and 2020. Patients with grade 3b follicular lymphoma or a previous history of malignancy were excluded. Eligible 5234 patients were assigned to three cohorts: rituximab monotherapy (N=780), RCHOP/RCVP/RTHPCOP (doxorubicin replaced with pirarubicin) (N=2298), or bendamustine/rituximab (BR) (N=2156). There were 589 recorded SPMs, of which myelodysplastic syndromes were the most common (1.7%). The cumulative incidence of SPM was significantly higher in patients treated with BR than in those treated with rituximab monotherapy (p<0.01) or RCHOP/RCVP/RTHPCOP (p<0.0001): the 5-year cumulative incidence function was 18.1%, 12.5%, and 12.9%, respectively. In the Fine-Gray subdistribution hazards model, BR showed a significantly higher cumulative incidence of SPM than RCHOP/RCVP/RTHPCOP (subhazard ratio, 1.33; 95% confidence interval [CI], 1.10-1.61). Furthermore, in sensitivity analysis, a nested case-control study using an entire cohort showed consistent results: the SPM odds ratios (95% CI) of first-line bendamustine, bendamustine after first-line, and any-line bendamustine were 1.43 (1.14-1.78), 1.26 (0.96-1.64), and 1.33 (1.09-1.62), respectively. Regarding infections, adjusted odds ratios (95% CI) of BR compared to RCHOP/RCVP/RTHPCOP were as follows: cytomegalovirus infection, 13.7 (4.88-38.4); bacterial pneumonia, 0.63 (0.50-0.78); and pneumocystis pneumonia, 0.24 (0.11-0.53). There was no significant difference in OS between RCHOP/RCVP/RTHPCOP and BR in patients with follicular, mantle cell, marginal zone, or lymphoplasmacytic lymphomas. In conclusion, treatment strategies that consider the risk of SPM and infections after chemotherapy are warranted in patients with iBCL.

Profiles, correlates, and risk of death among patients with mental disorders hospitalized for psychiatric reasons

This study identified profiles of hospitalized patients with mental disorders (MD) based on their 3-year hospitalization patterns and clinical characteristics and compared sociodemographic profiles and other service use correlates as well as risk of death within 12 months after hospitalization. Quebec (Canada) medical administrative databases were used to investigate a 5-year cohort of 4,400 patients hospitalized for psychiatric reasons. Latent class analysis, chi-square tests and survival analysis were produced. Three profiles of hospitalized patients were identified based on hospitalization patterns and other patient characteristics. Profile 3 patients had multiple hospitalizations and early readmissions, worst health and social conditions, and used the most outpatient services. Profiles 2 and 1 patients had only one hospitalization, of brief duration in the case of Profile 2 patients, who had mainly common MD and made least use of psychiatric care. All Profile 1 patients were hospitalized for serious MD but received least continuity of physician care and fewest biopsychosocial interventions. Risk of death was higher for Profiles 3 and 2 versus Profile 1 patients. Interventions like early follow-up care after hospitalization for Profile 3, collaborative care between general practitioners and psychiatrists for Profile 2, and continuous biopsychosocial care for Profile 1 could be greatly improved.

Relative survival after meningioma surgery. A French nationwide population-based cohort study

Background Survival after meningioma surgery is often reported with inadequate allowance for competing causes of death. Methods We processed the Système National des Données de Santé, the French administrative medical database to retrieve appropriate patients’ case of surgically treated meningiomas. The Pohar Perme relative survival (RS) method was implement. Results A total of 28,778 patients were identified between 2007 and 2017 of which 75% were female. Median age at surgery 59 years. Cranial convexity was the most common (24.7%) location and, benign meningioma represented 91.5% of all meningioma. Median follow-up was 3.5 years interquartile range [3.4–3.5]. At data collection, 2,232 patients were dead. The five-year survival relative to the expected survival of an age- and gender-matched French standard population was 96.2% 95% confidence interval (CI)[95.7–96.8]. Meningioma absolute excess risk of death was 973/100,000 person-years 95%CI[887–1068] (p< .001). The related standardised mortality ratio was 1.8 95%CI[1.7–1.9] (p< .001). In the adjusted model, male gender (hazard ratio [HR] =1.39, 95%CI[1.27–1.54], p< .001), age at surgery (HR=0.97, 95%CI[0.97–0.97], p < .001), type 2 neurofibromatosis (HR=2.95, 95%CI[1.95–4.46], p < .001), comorbidities HR=1.39, 95%CI[1.36–1.42], p < .001), location (HR=0.8, 95%CI[0.67–0.95], p= .0111), pre-operative embolization, (HR=1.3, 95%CI[1.08–1.56], p= .00507), cerebro-spinal fluid shunt, (HR=2.48, 95%CI[2.04–3.01], p < .001), atypical (HR=1.3, 95%CI [1.09–1.54], p= .00307) or malignant histology (HR=1.86, 95%CI[1.56–2.22], p< .001), redo surgery (HR=1.19, 95%CI[1.04–1.36], p= .0122) and radiotherapy (HR=1.43, 95%CI[1.26–1.62], p < .001) were established as independent predictors of RS. Conclusion This unique study highlights the excess mortality associated with meningioma disease. Many factors such as gender, age, location, histopathological grading, redo surgery influence the RS.

Predictors of Length of Hospitalization and Impact on Early Readmission for Mental Disorders.

Length of hospitalization, if inappropriate to patient needs, may be associated with early readmission, reflecting sub-optimal hospital treatment, and translating difficulties to access outpatient care after discharge. This study identified predictors of brief-stay (1-6 days), mid-stay (7-30 days) or long-stay (≥31 days) hospitalization, and evaluated how lengths of hospital stay impacted on early readmission (within 30 days) among 3729 patients with mental disorders (MD) or substance-related disorders (SRD). This five-year cohort study used medical administrative databases and multinomial logistic regression. Compared to patients with brief-stay or mid-stay hospitalization, more long-stay patients were 65+ years old, had serious MD, and had a usual psychiatrist rather than a general practitioner (GP). Predictors of early readmission were brief-stay hospitalization, residence in more materially deprived areas, more diagnoses of MD/SRD or chronic physical illnesses, and having a usual psychiatrist with or without a GP. Patients with long-stay hospitalization (≥31 days) and early readmission had more complex conditions, especially more co-occurring chronic physical illnesses, and more serious MD, while they tended to have a usual psychiatrist with or without a GP. For patients with more complex conditions, programs such as assertive community treatment, intensive case management or home treatment would be advisable, particularly for those living in materially deprived areas.

Evaluation of Bias in Medical Student Clinical Clerkship Evaluations Using Natural Language Processing.

Purpose: There is increasing interest in understanding potential bias in medical education. Prior studies have demonstrated potential disparities in the words used to describe medical students in their evaluations, such as variations by gender/race. 1,2 For example, natural language processing (NLP) has been used to determine narrative differences between medical clerkship evaluations on the basis of gender and under-represented minority status. 2 Females were more commonly described using words for personal attributes, whereas males were more likely to be described using words related to professional competencies. 1 In this study, we used NLP to evaluate potential bias among third-year clinical clerkship evaluations. Method: This study was conducted at the University of California San Diego School of Medicine. Data was extracted from medical evaluation and administrative databases for medical students enrolled in third-year clinical clerkship rotations across 2 academic years (2019–2020 and 2020–2021). For each evaluation, we collected information regarding demographics of both the student and the faculty evaluator. These were used to determine gender and racial concordance (i.e., whether the student and faculty identified in the same group). We extracted the full narrative text of each evaluation, as well as numerical evaluation scores. Narrative text was processed using a Python NLP package, which assigned a sentiment score for each evaluation. Word clouds were generated for various demographic groups to exhibit the most commonly used words. We analyzed the distributions of sentiment scores and clerkship grades to determine any differences by gender or race/ethnicity. We used multinomial logistic regression to model final clerkship grades, using predictors such as numerical evaluation scores, gender and racial concordance between faculty and students, and sentiment scores. Statistical analyses were performed using R. Statistical significance was defined as P < .05. Results: We analyzed 963 evaluations from 198 students, with 109 (55%) females. Ninety-two (47%) identified as Caucasian, 77 (39%) Asian, 13 (7%) African American, and 16 (9%) other/unspecified or declined to state race. Ten (5%) identified as Hispanic. Females (190/534, 34%) were more likely to receive honors than males (113/420, 26%) (P = .02). Significantly more students received honors or near-honors grades in the 2020–2021 school year in comparison with the 2019–2020 school year (P < .00001). Sentiment scores for evaluations did not vary significantly by student gender, race, or ethnicity (P = .88, .64, and .06, respectively). Word choices were similar across faculty and student demographic groups. Similarly, in the multinomial logistic regression model of final clerkship grades, the narrative evaluation sentiment score was not predictive of an honors grade (odds ratio [OR] 1.23, P = .58). However, the numerical evaluation average (OR 1.45, P < .001) and gender concordance between faculty and student (OR 1.32, P = .049) were significant predictors of receiving honors. Discussion: There was no clear evidence of bias in medical student evaluations. No differences were found in sentiment scores or in word choices by gender or race/ethnicity. Of note, sentiment scores from narrative evaluations were not significantly associated with final grades, while numerical evaluation scores were. Narrative feedback tended to be positive for all students regardless of final grade, which may not be constructive to students for understanding how to improve. Significance: The lack of disparities in our study contrasts prior findings from other institutions. Ongoing efforts include comparative analyses with other institutions to understand what institutional factors (e.g., geographic location) may contribute to bias. NLP enables a systematic approach for investigating bias. The insights gained from the lack of association between word choices, sentiment scores, and final grades show potential opportunities to improve feedback processes for students.

The Impact of 90-day Physician Follow-up Care on the Risk of Readmission Following a Psychiatric Hospitalization.

This study measures the impact of 90-day physician follow-up care after psychiatric hospitalization among 3,311 adults and youth, with risk of subsequent readmission within six months. A 5-year investigation was conducted based on Quebec (Canada) medical administrative databases. Cox proportional-hazards regression was performed, with 90-day follow-up care as the main independent variable, controlling for various sociodemographic, clinical, and other service use variables. Within the 90-day follow-up period after patient discharge, or in the first 30 days, receiving at least one consultation per month as opposed to no consultation was associated with a reduced risk of psychiatric readmission. Women showed an increased readmission risk compared to men, while those living in less materially deprived areas a decreased risk as opposed to more deprived areas. Patients hospitalized for suicide attempt or schizophrenia spectrum and other psychotic disorders, and those with co-occurring mental and substance-related disorders or chronic physical illnesses, especially illnesses high on the severity index, also presented a heightened risk of hospitalization. Patients hospitalized for personality disorders or receiving a high continuity of physician care showed a reduced risk of readmission. This study demonstrates that follow-up care, if provided within the first 30 days of discharge or monthly during the 90-day follow-up period, decreased the risk of readmission, as did having a high continuity of physician care prior to and within the 90-day follow-up period. However, few patients in this study had received such high-quality care, indicating that the Quebec system needs to considerably improve its discharge planning processes.

This Month in Adult Urology

This Month in Adult Urology

The Association Between Telehealth Utilization and Policy Responses on COVID-19 in Japan: Interrupted Time-Series Analysis.

BackgroundTelehealth using telephones or online communication is being promoted as a policy initiative in several countries. However, there is a lack of research on telehealth utilization in a country such as Japan that offers free access to medical care and regulates telehealth provision—particularly with respect to COVID-19.ObjectiveThe present study aimed to clarify telehealth utilization, the characteristics of patients and medical institutions using telehealth, and the changes to telehealth in Japan in order to support the formulation of policy strategies for telehealth provision.MethodsUsing a medical administrative claim database of the National Health Insurance and Advanced Elderly Medical Service System in Mie Prefecture, we investigated patients who used telehealth from January 2017 to September 2021. We examined telehealth utilization with respect to both patients and medical institutions, and we determined their characteristics. Using April 2020 as the reference time point for COVID-19, we conducted an interrupted time-series analysis (ITSA) to assess changes in the monthly proportion of telehealth users to beneficiaries.ResultsThe number of telehealth users before the reference time point was 13,618, and after the reference time point, it was 28,853. Several diseases and conditions were associated with an increase in telehealth utilization. Telehealth consultations were mostly conducted by telephone and for prescriptions. The ITSA results showed a sharp increase in the proportion of telehealth use to beneficiaries after the reference time point (rate ratio 2.97; 95% CI 2.14-2.31). However, no apparent change in the trend of increasing or decreasing telehealth use was evident after the reference time point (rate ratio 1.00; 95% CI 1.00-1.01).ConclusionsWe observed a sharp increase in telehealth utilization after April 2020, but no change in the trend of telehealth use was evident. We identified changes in the characteristics of patients and providers using telehealth.

Dementia Associated with Anticholinergic Drugs Used for Overactive Bladder: A Nested Case-Control Study Using the French National Medical-Administrative Database.

We analyzed the relationship between use of anticholinergic drugs to treat overactive bladder (OAB) and risk of incident dementia in older patients, overall and for each drug separately. We conducted a nested case-control study using the French National Medical-Administrative Database. We identified incident dementia cases and controls from January 1, 2013 to December 31, 2018 in individuals aged ≥60 years. Controls were matched 5:1 to cases by date of case diagnosis (index date), age, sex, and income. We set a 5-year exposure period ending 2 years before the index date (lag-time period to avoid protopathic bias). We quantified cumulative exposure to flavoxate, oxybutynin, solifenacin, trospium, and fesoterodine using defined daily doses (DDDs). We performed conditional logistic regression analyses adjusted for factors known to be associated with OAB and/or dementia including obesity, diabetes, stroke, coronary heart disease, and psychotic disorders. We analyzed 4,810 cases and 24,050 matched controls with a median age of 82 years. OAB anticholinergic use was associated with an increased risk of dementia (adjusted OR [aOR]=1.23, 95% CI 1.10-1.37) with a cumulative dose-response: aOR=1.07 (95% CI 0.91-1.25) for 1-90 DDDs, aOR=1.29 (1.05-1.58) for 91-365 DDDs and aOR=1.48 (1.22-1.80) for >365 DDDs. Considering each OAB anticholinergic separately showed a particularly marked increased risk of dementia for oxybutynin and solifenacin, but no increased risk for trospium. When treating OAB in older patients, OAB anticholinergics should be used with caution, taking into account the patient's cognitive status, the anticholinergic load, and the different therapeutic options.

"Book Music" Representation for Temporal Data, as a Part of the Feature Extraction Process: A Novel Approach to Improve the Handling of Time-Dependent Data in Secondary Use of Healthcare Structured Data.

Book music is extensively used in street organs. It consists of thick cardboard, containing perforated holes specifying the musical notes. We propose to represent clinical time-dependent data in a tabular form inspired from this principle. The sheet represents a statistical individual, each row represents a binary time-dependent variable, and each hole denotes the "true" value. Data from electronic health records or nationwide medical-administrative databases can then be represented: demographics, patient flow, drugs, laboratory results, diagnoses, and procedures. This data representation is suitable for survival analysis (e.g., Cox model with repeated outcomes and changing covariates) and different types of temporal association rules. Quantitative continuous variables can be discretized, as in clinical studies. The "book music" approach could become an intermediary step in feature extraction from structured data. It would enable to better account for time in analyses, notably for historical cohort analyses based on healthcare data reuse.

Risk factors for early mortality of lung cancer patients in France: A nationwide analysis.

Despite therapeutic advances, lung cancer remains the first cause of death from cancer. The main objective of this study was to identify risk factors associated with death within 3-months of the first hospitalization for lung cancer in France. This analysis included patients with a first hospitalization for lung cancer (between January 1, 2016 and December 31, 2018) according to diagnosis-related groups entered into the French national medical-administrative database. Clinical and socioeconomic parameters and characteristics of that first hospitalization were analyzed. A model predictive of early mortality was developed based on those variables. The 144,087 included patients were 67% men; median age of 68 [interquartile range 60-76] years; 47% had metastatic disease at diagnosis; and 34% and 23%, respectively, had received systemic treatment or undergone curative surgery. The 3-month mortality was 19%, and significantly higher for those ≥70 versus <70 years old (OR 1.33, 1.22-1.45), men versus. women (OR 1.50, 1.44-1.55), those with metastatic disease at diagnosis (OR, 3.30, 3.18-3.43), first hospitalization via the emergency room (OR 1.65 1.59-1.71) and first hospitalization lasting >30 days (OR, 1.58 1.49-1.68). In contrast, no socioeconomic characteristic was associated with early mortality. Almost 1 in 5 patients diagnosed with lung cancer in France died within 3months post-diagnosis. Improving survival requires diagnosis at an earlier stage and better organization of diagnosis and specific care pathways.