Time-dependent Covariates Research Articles

Abstract PO5-06-01: Loss of hormone receptor expression in breast cancer is associated with increased brain tropism and accelerated progression of leptomeningeal disease

Abstract The four breast cancer subtypes—luminal A, luminal B, HER2-overexpressing, and triple-negative breast cancer (TNBC) – are classified based on gene expression profiles of hormone receptors (HRs) (estrogen receptor (ER) and progesterone receptor (PR)) and human epidermal growth factor receptor 2 (HER2/ERBB2). Negative hormone receptor status is known to be associated with more aggressive breast cancer. Compared to other breast cancer subtypes, the HR-negative TNBC and HER2-overexpressing subtypes are characterized by high rates of recurrence and a notable increase in brain metastases and in particular, leptomeningeal disease (LMD). LMD – the most aggressive form of central nervous system (CNS) metastasis, in which cancer cells infiltrate leptomeninges of the brain and spine – carries a dismal prognosis due to a rapid progression, poor detection, and no effective treatment. Several studies that investigated the link between HR status and LMD suggest that HR-negative breast cancers have a shorter time to LMD diagnosis. This shortened timeline is consistent with higher migratory capabilities reported for TNBC and HER2 breast cancer cells. However, as it is known that breast cancers exhibit subtype-specific metastatic tropism (i.e., luminal subtypes tend to metastasize to bone and liver, while TNBCs metastasize to the CNS), the longer time between primary breast cancer diagnosis and LMD diagnosis for HR-positive breast cancers may also reflect a longer metastatic route that first includes a non-CNS location. In this study, we performed a retrospective analysis on 141 breast cancer patients with LMD treated at Stanford Hospital between 2008-2020. The Stanford LMD cohort was analyzed using linear regression analysis, Cox regression analysis, and Chi-squared analysis. HR status was recorded (when available) for local recurrences and distant metastases, and was included in our regression models as a time-dependent HR covariate. We tested if: (1) HR-negative breast cancers had shorter time between primary and LMD diagnosis as well as worse survival after LMD diagnosis versus HR-positive breast cancers, and if (2) HR loss (i.e., HR status changing from positive to negative) increased likelihood of the metastasis to CNS. In agreement with published studies, the Stanford LMD cohort showed that HR-negative cancers had shorter time to development of LMD (p = 0.0004 for ER, p = 0.0001 for PR) and increased mortality following LMD (p = 0.010 for ER, p = 0.009 for PR). We found that HR-negative status at primary breast cancer diagnosis increased risk of first metastasis to CNS versus non-CNS locations (p = 0.021 for ER, p = 0.043 for PR). HR status switch (also known as hormone discordance) between primary diagnosis and metastasis occurred in 15% of patients in Stanford cohort, consistent with other studies. Importantly, accounting for HR status switch strengthened the association between HR-negative status and the risk of CNS metastasis [Hazard Ratio = 2.506 (1.395 - 4.500) for “baseline ER status” versus Hazard Ratio = 3.509 (1.890 - 6.515) for “time-dependent ER status”, Hazard Ratio = 2.420 (1.310 - 4.471) for “baseline PR status” versus Hazard Ratio = 3.407 (1.815 - 6.395) for “time-dependent PR status”]. In conclusion, HR-negative status was associated with shorter time to development of LMD, higher CNS tropism, and shorter survival following LMD diagnosis. Taking into consideration hormone discordance between the primary tumor and metastatic sites strengthened this association. Our results suggest a potential value in testing HR status at all recurrences and advanced CNS monitoring following HR loss. Citation Format: Sophia Pribus, Maxine C. Umeh-Garcia, Bo Gu, Bryanna Godfrey, Summer Han, Sophia Chernikova, Melanie Hayden Gephart. Loss of hormone receptor expression in breast cancer is associated with increased brain tropism and accelerated progression of leptomeningeal disease [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-06-01.

Abstract PO2-04-10: Effect of Alpelisib Dose Modification for AE Management on Progression-Free Survival and Treatment Duration in SOLAR-1 and BYLieve Clinical Trials

Abstract Background: Alpelisib (ALP) is an α-selective phosphatidylinositol 3-kinase (PI3K) inhibitor and degrader approved with fulvestrant (FUL) for patients (pts) with phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA)-mutated, hormone receptor-positive (HR+), human epidermal growth factor receptor 2-negative (HER2−) advanced breast cancer (ABC) following progression on/after endocrine therapy-based treatment (tx). Toxicity limited ALP exposure in some pts in the SOLAR-1 and BYLieve studies. On-target effects of PI3K inhibitors, including hyperglycemia, rash, and diarrhea, are among the most common adverse events (AEs) associated with ALP. During SOLAR-1, median onset times were 15 and 13 days for grade ≥3 hyperglycemia and rash and 46 days for grade 2/3 diarrhea. ALP dose modification can help manage AEs. We report efficacy of ALP + FUL as second-line (2L) tx in pts who underwent ALP dose reduction (DR). Methods: Pts with PIK3CA mutations from SOLAR-1 and BYLieve, treated with ALP + FUL in 2L, were included. Landmark (LM) analyses were performed at 1, 2, 3, 4, 5, and 6 mo after tx initiation to evaluate the association between DR status (yes, no) with progression-free survival (PFS) and duration of tx (DoT), respectively. Pts with PFS or exposure time less than each LM time point were excluded from each LM analysis; pts were then grouped by DR (yes, no) before each LM time point. Median PFS and DoT were obtained using a modified Kaplan-Meier method. Alternatively, Cox regression analysis was used to assess the effect of DR on PFS and DoT, where 2 time-dependent variables were included separately in the model as a covariate: 1) DR status, and 2) DR groups (300 to 250 mg [1 DR] vs no DR, 300 to 250 to 200 mg [2 DR] vs no DR). Association between AEs and DR was evaluated descriptively. A matched analysis of AEs will be presented. Results: In all, the analysis included 212 pts in 2L: 77 from SOLAR-1, and 95 and 40 from BYLieve Cohorts A and C, respectively. Of these pts, 92 (43.4%) had no DR and 120 (56.6%) had DRs: 70 (58.3%) had 1 DR, 43 (35.8%) had 2, and 7 (5.8%) had other types of DR. 25 of 92 pts (27.2%) with no DR and 40 of 120 pts (33.3%) with DR had ≥12-mo exposure to ALP. Hazard ratios (HRs) for PFS and DoT in DR status for the LM analysis are shown (Table). HRs (95% CI) for PFS from the time-dependent covariate analyses were 1.28 (0.94-1.74) with DR status as the covariate, and 1.24 (0.88-1.76) and 1.21 (0.79-1.84) with DR group (1 DR vs no DR, 2 DR vs no DR) as the covariate, respectively; for DoT, these were 1.34 (1.00-1.79) for DR status, and 1.30 (0.93-1.80) and 1.30 (0.87-1.93) for DR group, respectively. During the treatment period, the incidence rate for grade ≥3 AEs was the highest for pts with 2 DR and lowest for those with no DR. Similarly, incidence of AEs leading to DR or tx discontinuation was higher in pts with 2 DR than 1 DR. The post-DR incidence rate for grade ≥3 AE was similar for pts with 2 DR or 1 DR. A matched analysis will assess the relationship between DR group and AE. Conclusions: In this pooled analysis, regardless of if/when DR occurred, PFS was similar and DoT may be sustained with ALP + FUL. ALP DR can potentially reduce AEs, which may allow pts to remain on tx. Using DR as an AE management strategy may enable pts with PIK3CA-mutated, HR+, HER2− ABC to optimize tx duration and clinical benefit of ALP in the 2L. Landmark Analyses of PFS and Tx Duration by DR Group (Yes vs No; Pooled FAS) Citation Format: Hope Rugo, Alessandra Gennari, Stephen Chia, Dejan Juric, Neil Vasan, Sherko Küemmel, Patrick Neven, Florence Lerebours, Manuel Ruíz - Borrego, Pedram Razavi, Jyotika Singh, Yogesh Chattar, Murat Akdere, Eva Ciruelos. Effect of Alpelisib Dose Modification for AE Management on Progression-Free Survival and Treatment Duration in SOLAR-1 and BYLieve Clinical Trials [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-04-10.

Trajectories of Physical Activity and Chronic Conditions Among Mid-Aged Women

IntroductionWomen's physical activity levels vary throughout adulthood. However, the associations between trajectories of physical activity and health outcomes have been little studied. The aim of this study was to examine the associations of physical activity trajectories with incident diabetes, hypertension, obesity, depression, and physical disability in mid-aged women. MethodsData were from the 1946–51 cohort of the Australian Longitudinal Study on Women's Health (n = 11,611). Mailed surveys at 3-year intervals from 1998 (age 47–52) to 2019 (age 68–73) were used to collect data on physical activity and each outcome. The analyses were conducted in 2023/24. Physical activity trajectories from age 47–61 [2–4 time-points] were identified using group-based trajectory modeling. Cox regression models with time-varying covariates were used to examine the associations of physical activity trajectories with adverse health outcomes onset from ages 56–73 [2–5 time-points]. ResultsTrajectories of physical activity were labeled as Low (59.7% of participants), Declining (5.9%), Increasing (26.3%), and High (8.1%). The median of 9-year overall physical activity (MET.minutes/week) was 450 in the Low, 1324 in the Declining, 1399 in the Increasing, and 2323 in the High group. Compared with the Low group, the risks of diabetes, obesity and physical disability were lower in the Increasing and High groups, and the risk of depression was lower in the High group. Conversely, the risk of obesity was higher in the Declining group. ConclusionsTargeting mid-aged women with declining physical activity for prevention programs could enhance health in later life.

Placental growth factor and fetoplacental Doppler indices in combination predict preterm birth reliably in pregnancies complicated by fetal growth restriction.

To assess the association between placental biomarkers (placental growth factor (PlGF) and soluble fms-like tyrosine kinase-1 (sFlt-1)/PlGF ratio) and fetoplacental Doppler indices (umbilical artery (UA) pulsatility index (PI) and uterine artery (UtA) PI) in various combinations for predicting preterm birth (PTB) in pregnancies complicated by fetal growth restriction (FGR). This was a prospective observational cohort study, performed at Mater Mother's Hospital in Brisbane, Queensland, Australia, from May 2022 to June 2023, of pregnancies complicated by FGR and appropriate-for-gestational-age (AGA) pregnancies. Maternal serum PlGF levels, sFlt-1/PlGF ratio, UA-PI and UtA-PI were measured at 2-4-weekly intervals from recruitment until delivery. Harrell's concordance statistic (Harrell's C) was used to evaluate multivariable Cox proportional hazards regression models featuring various combinations of placental biomarkers and fetoplacental Doppler indices to ascertain the best combination to predict PTB (< 37 weeks). Multivariable Cox regression models were used with biomarkers as time-varying covariates. The study cohort included 320 singleton pregnancies, comprising 179 (55.9%) affected by FGR, defined according to a Delphi consensus, and 141 (44.1%) with an AGA fetus. In the FGR cohort, both low PlGF levels and elevated sFlt-1/PlGF ratio were associated with significantly shorter time to PTB. Low PlGF was a better predictor of PTB than was either sFlt-1/PlGF ratio or a combination of PlGF and sFlt-1/PlGF ratio (Harrell's C, 0.81, 0.78 and 0.79, respectively). Although both Doppler indices were significantly associated with time to PTB, in combination they were better predictors of PTB than was either UA-PI > 95th centile or UtA-PI > 95th centile alone (Harrell's C, 0.82, 0.75 and 0.76, respectively). Predictive utility for PTB was best when PlGF < 100 ng/L, UA-PI > 95th centile and UtA-PI > 95th centile were combined (Harrell's C, 0.88) (hazard ratio, 32.99; 95% CI, 10.74-101.32). Low maternal serum PlGF level (< 100 ng/L) and abnormal fetoplacental Doppler indices (UA-PI > 95th centile and UtA-PI > 95th centile) in combination have the greatest predictive utility for PTB in pregnancies complicated by FGR. Their assessment may help guide clinical management of these complex pregnancies. © 2023 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

Role of Routine Suppressive Antibiotic Therapy After Debridement, Antibiotics, and Implant Retention for Acute Periprosthetic Joint Infections.

The first-line management strategy for acute periprosthetic joint infections (PJIs) is debridement, antibiotics, and implant retention (DAIR). Suppressive antibiotic therapy (SAT) after DAIR is proposed to improve outcomes, yet its efficacy remains under scrutiny. We conducted a multicenter retrospective study in patients with acute PJI of the hip or knee who were treated with DAIR in centers from Europe and the United States. We analyzed the effect of SAT using a Cox model landmarked at 12 weeks. The primary covariate of interest was SAT, which was analyzed as a time-varying covariate. Patients who experienced treatment failure or were lost to follow-up within 12 weeks were excluded from the analysis. The study included 510 patients with 66 treatment failures with a median follow-up of 801 days. We did not find a statistically significant association between SAT and treatment failure (hazard ratio, 1.37; 95% CI, .79-2.39; P = .27). Subgroup analyses for joint, country cohort, and type of infection (early or late acute) did not show benefit for SAT. Secondary analysis of country cohorts showed a trend toward benefit for the US cohort (hazard ratio, 0.36; 95% CI, .11-1.15; P = .09), which also had the highest risk of treatment failure. The utility of routine SAT as a strategy for enhancing DAIR's success in acute PJI remains uncertain. Our results suggest that SAT's benefits might be restricted to specific groups of patients, underscoring the need for randomized controlled trials. Identifying patients most likely to benefit from SAT should be a priority in future studies.

Multilevel Latent Differential Structural Equation Model with Short Time Series and Time-Varying Covariates: A Comparison of Frequentist and Bayesian Estimators

Continuous-time modeling using differential equations is a promising technique to model change processes with longitudinal data. Among ways to fit this model, the Latent Differential Structural Equation Modeling (LDSEM) approach defines latent derivative variables within a structural equation modeling (SEM) framework, thereby allowing researchers to leverage advantages of the SEM framework for model building, estimation, inference, and comparison purposes. Still, a few issues remain unresolved, including performance of multilevel variations of the LDSEM under short time lengths (e.g., 14 time points), particularly when coupled multivariate processes and time-varying covariates are involved. Additionally, the possibility of using Bayesian estimation to facilitate the estimation of multilevel LDSEM (M-LDSEM) models with complex and higher-dimensional random effect structures has not been investigated. We present a series of Monte Carlo simulations to evaluate three possible approaches to fitting M-LDSEM, including: frequentist single-level and two-level robust estimators and Bayesian two-level estimator. Our findings suggested that the Bayesian approach outperformed other frequentist approaches. The effects of time-varying covariates are well recovered, and coupling parameters are the least biased especially using higher-order derivative information with the Bayesian estimator. Finally, an empirical example is provided to show the applicability of the approach.

No clear influence of treatment escalation on flare prevention in serologically active clinically quiescent patients with systemic lupus erythematosus: a retrospective cohort study.

This study aimed to clarify the efficacy and safety of treatment escalation by initiating therapeutic agents in serologically active clinically quiescent (SACQ) patients with systemic lupus erythematosus (SLE). We retrospectively evaluated SACQ patients with SLE for ≥ 180days, with the introduction of a therapeutic agent for SLE defined as exposure. The efficacy endpoints included the time to flare and time to remission, whereas the safety endpoint was the incidence of adverse events. The efficacy endpoints were assessed via Cox proportional hazards model with time-dependent covariates, which included exposure, serological activity, and prednisolone dose. Among 109 SACQ patients, 24 were initiated on the following therapeutic agents for SLE: hydroxychloroquine (10 patients), belimumab (6 patients), and immunosuppressive agents (8 patients). A total of 37 patients experienced a flare (8 and 29 patients during exposure and nonexposure periods, respectively). The time to flare was comparable between the exposure and control groups. Among 68 patients who were not in remission at the start of observation, 27 patients achieved remission (5 and 22 patients during exposure and nonexposure periods, respectively). Although both groups had a similar time to remission, the exposure group treated with belimumab had a significantly higher rate of remission than the control group. The adverse events were more frequent during the exposure period than during the nonexposure period. Thus, this study did not reveal a clear influence of treatment escalation on flare prevention and remission achievement.

Uptake of SGLT2i and Outcomes in Patients with Diabetes and Heart Failure: A Population-Based Cohort and a Specialized Clinic Cohort

BackgroundSodium/glucose cotransporter 2 inhibitors (SGLT2i) are effective in adults with diabetes mellitus (DM) and heart failure (HF) based on randomized clinical trials. We compared SGLT2i uptake and outcomes in two cohorts: a population-based cohort of all adults with DM and HF in Alberta, Canada and a specialized heart failure clinic (HFC) cohort. MethodsThe population-based cohort was derived from linked provincial healthcare datasets. The specialized clinic cohort was created by chart review of consecutive patients prospectively enrolled in the HFC between February 2018 and August 2022. We examined the association between SGLT2i use (modeled as a time-varying covariate) and all-cause mortality or deaths/cardiovascular hospitalizations. ResultsOf the 4,885 individuals from the population-based cohort, 64.2% met the eligibility criteria of the trials proving the effectiveness of SGLT2i. Utilization of SGLT2i increased from 1.2% in 2017 to 26.4% by January 2022. In comparison, of the 530 patients followed in the HFC, SGLT2i use increased from 9.8% in 2019 to 49.1 % by March 2022. SGLT2i use in the population-based cohort was associated with fewer all-cause mortality (aHR 0.51, 95% CI 0.41–0.63) and deaths/cardiovascular hospitalizations (aHR 0.65, 95% CI 0.54–0.77). However, SGLT2i usage rates were far lower in HF patients without DM (3.5% by March 2022 in the HFC cohort). ConclusionsDespite robust randomized trial evidence of clinical benefit, the uptake of SGLT2i in patients with HF and DM remains low, even in the specialized HFC. Clinical care strategies are needed to enhance the use of SGLT2i and improve implementation.

Diverticulitis Is Associated with Increased Risk of Colon Cancer-A Nationwide Register-Based Cohort Study.

Background: An association between diverticulitis and colon cancer has been proposed. The evidence is conflicting, and the guidelines differ regarding recommended follow-up with colonoscopy after an episode of diverticulitis. To guide regimes for follow-up, this study aimed to investigate if patients with diverticulitis have an increased risk of colon cancer. Methods: This study is reported according to the RECORD statement. We performed a cohort study with linked data from nationwide Danish registers. The inclusion period was 1997-2009, and the complete study period was 1995-2013. The primary outcome was the risk of developing colon cancer estimated using a Cox regression analysis with time-varying covariates. We performed a sensitivity analysis on a cohort of people with prior colonoscopies, comparing the risk of colon cancer between the diverticulitis group and the control group. Results: We included 29,173 adult males and females with diverticulitis and 145,865 controls matched for sex and age. The incidence proportion of colon cancer was 2.1% (95% confidence interval (CI) 1.9-2.3) in the diverticulitis group and 1.5% (95% CI 1.4-1.5) in the matched control group (hazard ratio 1.6; 95% CI 1.5-1.8). The risk of having a colon cancer diagnosis was significantly increased in the first six months after inclusion (hazard ratio 1.7; 95% CI 1.5-1.8), and hereafter there was a lower risk in the diverticulitis group compared with controls (hazard ratio 0.8; 95% CI 0.7-0.9). This protective effect lasted eight years. The increased risk of colon cancer during the first six months after diverticulitis was also found in the cohort with prior colonoscopies. Conclusions: The risk of a colon cancer diagnosis was significantly increased for patients with diverticulitis 0-6 months after the diagnosis of diverticulitis. Hereafter, we found a protective effect of diverticulitis until eight years later, possibly due to a screening effect. We recommend a follow-up colonoscopy after the first diagnosis of diverticulitis.

Prosthetic outcomes after amputation and the impact of mobility level on survival

ObjectiveLower extremity amputation continues to be necessary in a significant number of patients with peripheral vascular disease. The 5-year survival following lower limb loss is markedly reduced. Many of these patients are never fitted with a prosthesis, and there is a dearth of knowledge regarding the barriers to prosthetic attainment. The goal of this study was to identify the risk factors for not receiving a prosthesis and the effect of mobility level on survival following major amputation. MethodsThis was a retrospective analysis of all patients that underwent lower extremity amputation by surgeons in our practice from January 1, 2010, to December 31, 2019. Abstracted data included: age, sex, race, body mass index, comorbidities, American Society of Anesthesiologists score, statin use, level of amputation, stump revision, fitting for prosthesis, type of prosthesis, and the United States’ Medicare Functional Classification Level, also called K level. Survival was determined using a combination of sources, including the Social Security Death Master File, searches of multiple genealogic registries, and general internet searches. Multivariable logistic regression was used to determine risk factors associated with prosthesis attainment. Multivariable Cox proportional hazard regression with time-dependent covariates was performed to assess risk factors associated with 5-year mortality. ResultsA total of 464 patients were included in this study. The mean age was 65 years, and mean body mass index was 27 kg/m2. The majority of patients were male (68%), White (56%), diabetic (62%), and hypertensive (76%), and underwent below-the-knee amputation (69%). Prosthetic attainment occurred in 185 (40%). On multivariable analysis, age >81 years and current tobacco use were associated with no prosthetic fitting. Overall 5-year survival was 41.9% (95% confidence interval [CI], 37.6%-46.6%) (below-the-knee amputation, 47.7% [95% CI, 42.5%-53.5%]; above-the-knee amputation, 28.7% [95% CI, 22.1%-37.2%]). On multivariable analysis, age >60 years, congestive heart failure, above-the-knee amputation, and no prosthetic attainment were associated with decreased survival. Increasing K level was incrementally associated with improved survival. ConclusionsThis study has identified several patient factors associated with prosthetic attainment, as well as multiple factors predictive of reduced survival after amputation. Being referred for prosthetic fitting was associated with improved survival not explained by patient characteristics and comorbidities. The Medicare Functional Classification Level K level predicts survival. More research is needed to determine the barriers to prosthetic attainment and if improving a patients K level will improve survival.

Peritoneal Dialysis Technique Survival: A Cohort Study

Rationale & ObjectiveReasons for transfer from peritoneal dialysis (PD) to hemodialysis (HD) remain incompletely understood. Among incident and prevalent patients receiving PD, we evaluated the association between prior treatment with HD and PD technique survival. Study DesignRetrospective cohort study. Setting & ParticipantsAdults who initiated PD at a Dialysis Clinic, Inc. (DCI) outpatient facility between January 1, 2010 and September 30, 2019. ExposuresThe primary exposure of interest was timing of PD start, categorized as PD-first, PD-early, or PD-late. Other covariates included demographics, clinical characteristics, and routine laboratory results. OutcomesModality switch from PD to HD sustained for more than 90 days. Analytical ApproachMultivariable Fine-Gray models with competing risks and time-varying covariates, stratified at 9 months to account for lack of proportionality. ResultsAmong 5224 patients who initiated PD at a DCI facility, 3174 initiated dialysis with PD (“PD-first”), 942 transitioned from HD to PD within 90 days (“PD-early”), and 1108 transitioned beyond 90 days (“PD-late”); 1472 (28%) subsequently transferred from PD to HD. PD-early and PD-late patients had higher risk of transfer to HD as compared to PD-first patients [adjusted hazard ratio (aHR) 1.51 (95% CI: 1.17-1.96) and 2.41 (1.94-3.00), respectively, in the first 9 months and aHR 1.16 (0.99-1.35) and 1.43 (1.24-1.65), respectively, after 9 months]. More peritonitis episodes, fewer home visits, lower serum albumin, lower residual kidney function, and lower peritoneal clearance calculated with weekly Kt/V were additional risk factors for PD-to-HD transfer. LimitationsMissing data on dialysis adequacy and residual kidney function, confounded by short PD technique survival. ConclusionsInitiating dialysis with PD is associated with greater PD technique survival, though many of those who initiate PD late in their dialysis course still experience substantial time on PD. Peritonitis, lower serum albumin, and lower Kt/V are risk factors for PD-to-HD transfer that may be amenable to intervention.

Obstetric Complications and Birth Outcomes After Antenatal Coronavirus Disease 2019 (COVID-19) Vaccination.

To evaluate the association between antenatal messenger RNA (mRNA) coronavirus disease 2019 (COVID-19) vaccination and risk of adverse pregnancy outcomes. This was a retrospective cohort study of individuals with singleton pregnancies with live deliveries between June 1, 2021, and January 31, 2022, with data available from eight integrated health care systems in the Vaccine Safety Datalink. Vaccine exposure was defined as receipt of one or two mRNA COVID-19 vaccine doses (primary series) during pregnancy. Outcomes were preterm birth (PTB) before 37 weeks of gestation, small-for-gestational age (SGA) neonates, gestational diabetes mellitus (GDM), gestational hypertension, and preeclampsia-eclampsia-HELLP (hemolysis, elevated liver enzymes, and low platelet count) syndrome. Outcomes in individuals vaccinated were compared with those in propensity-matched individuals with unexposed pregnancies. Adjusted hazard ratios (aHRs) and 95% CIs were estimated for PTB and SGA using a time-dependent covariate Cox model, and adjusted relative risks (aRRs) were estimated for GDM, gestational hypertension, and preeclampsia-eclampsia-HELLP syndrome using Poisson regression with robust variance. Among 55,591 individuals eligible for inclusion, 23,517 (42.3%) received one or two mRNA COVID-19 vaccine doses during pregnancy. Receipt of mRNA COVID-19 vaccination varied by maternal age, race, Hispanic ethnicity, and history of COVID-19. Compared with no vaccination, mRNA COVID-19 vaccination was associated with a decreased risk of PTB (rate: 6.4 [vaccinated] vs 7.7 [unvaccinated] per 100, aHR 0.89; 95% CI, 0.83-0.94). Messenger RNA COVID-19 vaccination was not associated with SGA (8.3 vs 7.4 per 100; aHR 1.06, 95% CI, 0.99-1.13), GDM (11.9 vs 10.6 per 100; aRR 1.00, 95% CI, 0.90-1.10), gestational hypertension (10.8 vs 9.9 per 100; aRR 1.08, 95% CI, 0.96-1.22), or preeclampsia-eclampsia-HELLP syndrome (8.9 vs 8.4 per 100; aRR 1.10, 95% CI, 0.97-1.24). Receipt of an mRNA COVID-19 vaccine during pregnancy was not associated with an increased risk of adverse pregnancy outcomes; this information will be helpful for patients and clinicians when considering COVID-19 vaccination in pregnancy.

Partial-linear single-index transformation models with censored data.

In studies with time-to-event outcomes, multiple, inter-correlated, and time-varying covariates are commonly observed. It is of great interest to model their joint effects by allowing a flexible functional form and to delineate their relative contributions to survival risk. A class of semiparametric transformation (ST) models offers flexible specifications of the intensity function and can be a general framework to accommodate nonlinear covariate effects. In this paper, we propose a partial-linear single-index (PLSI) transformation model that reduces the dimensionality of multiple covariates into a single index and provides interpretable estimates of the covariate effects. We develop an iterative algorithm using the regression spline technique to model the nonparametric single-index function for possibly nonlinear joint effects, followed by nonparametric maximum likelihood estimation. We also propose a nonparametric testing procedure to formally examine the linearity of covariate effects. We conduct Monte Carlo simulation studies to compare the PLSI transformation model with the standard ST model and apply it to NYU Langone Health de-identified electronic health record data on COVID-19 hospitalized patients' mortality and a Veteran's Administration lung cancer trial.

Customer Churn Analysis in Telecom Industry using Logistics Regression in Machine Learning with Kaplan–Meier and Cox Proportional Hazards Model

Customer churn poses a significant challenge for companies in the telecom industry, impacting revenue and profitability. In this paper, we present a comprehensive survey of customer churn analysis methodologies, focusing on the application of logistic regression, Kaplan-Meier survival analysis, and the Cox Proportional Hazards Model. Logistic regression serves as a fundamental technique for binary classification, enabling the prediction of churn based on customer attributes and behaviors. Kaplan-Meier survival analysis offers insights into customer survival probabilities over time, facilitating the identification of high-risk cohorts. Additionally, the Cox Proportional Hazards Model provides a flexible framework for analyzing the impact of predictor variables on churn timing, accounting for censoring and time-varying covariates. Through a review of existing literature, case studies, and practical applications, we explore the strengths and limitations of these methodologies in customer churn prediction. We also discuss challenges in data collection, feature engineering, and model evaluation, along with future research directions to enhance churn analysis effectiveness. By leveraging logistic regression with Kaplan-Meier and the Cox model, telecom companies can develop robust churn prediction models to improve customer retention strategies and mitigate revenue loss. Keywords: Customer churn analysis, Telecom industry, Logistic regression, Kaplan-Meier survival analysis, Cox Proportional Hazards Model, Retention strategies.

Examining lack of referrals to heart valve specialists as mechanisms of potential underutilization of aortic valve replacement

BackgroundRecent studies suggest that aortic valve replacement (AVR) remains underutilized. AimsInvestigate the potential role of non-referral to heart valve specialists (HVS) on AVR utilization. MethodsPatients with severe aortic stenosis (AS) between 2015 and 2018, who met class I indication for intervention, were identified. Baseline data and process-related parameters were collected to analyze referral predictors and evaluate outcomes. ResultsAmong 981 patients meeting criteria AVR, 790 patients (80.5%) were assessed by HVS within six months of index TTE. Factors linked to reduced referral included increasing age (OR: 0.95; 95% CI: 0.94-0.97; P < .001), unmarried status (OR: 0.59; 95% CI: 0.43-0.83; P = .002) and inpatient TTE (OR: 0.27; 95% CI: 0.19-0.38; P < .001). Conversely, higher hematocrit (OR: 1.13; 95% CI: 1.09-1.16; P < .001) and eGFR (OR: 1.01; 95% CI: 1.00-1.02; P = .003), mean aortic valve gradient (OR: 1.03; 95% CI: 1.01-1.04; P < .001) and preserved LVEF (OR: 1.59; 95% CI: 1.02-2.48; P = .04), were associated with increased referral likelihood. Moreover, patients assessed by HVS referral as a time-dependent covariate had a significantly lower two-year mortality risk than those who were not (aHR: 0.30; 95% CI: 0.23-0.39; P < .001). ConclusionA substantial proportion of severe AS patients meeting indications for AVR are not evaluated by HVS and experience markedly increased mortality. Further research is warranted to assess the efficacy of care delivery mechanisms, such as e-consults, and telemedicine, to improve access to HVS expertise.

Estimating high-dimensional Hawkes process with time-dependent covariates

The Hawkes process models have been recently become a popular tool for modeling and analysis of neural spike train data. Despite this popularity, existing methodological and theoretical work has mainly focused on models without assumption of time-varying covariates, such as stimulus drive. In this article, motivated by neuronal spike trains study in the presence of stimuli, we propose a new Hawkes process model, where covariates are included in the intensity function. We consider the problem of simultaneous variable selection and estimation for the Hawkes process in the high-dimensional regime. Estimation of the intensity function of the high-dimensional point process is considered within a nonparametric framework, applying B-splines and the SCAD penalty for matters of sparsity. We apply the Doob-Kolmogorov inequality to establish the consistency of the resulting estimators. Finally, we illustrate the performance of our proposal through simulation and demonstrate its utility by applying it to the neuron spike train data set.

A Symmetric Kernel Smoothing Estimation of the Time-Varying Coefficient for Medical Costs

In longitudinal studies, subjects are repeatedly observed at a set of distinct time points until the terminal event time. The time-varying coefficient model extends the parametric method and captures the dynamic trajectories of time-dependent covariate effects, thus enabling it to describe the potential relationship between the longitudinal variable and the observed time points. In this study, we propose a novel approach to the estimation of medical costs using a symmetric kernel smoothing method in the time-varying coefficient joint model. A smooth function of medical costs is derived by weighting the values of longitudinal data at all distinct observed time points via the combination of the kernel method and the inverse probability weighting method. For the simulation study, we first set up the true functions of time-varying coefficients; we then generated random samples for covariates and censored survival times. Subsequently, the longitudinal data of response variables could be produced. Further, numerical simulation experiments were conducted by using the proposed method and applying R code to the generated data. The estimated results for the parameters and non-parametric functions were compared with different settings. The numerical results illustrate that as the sample size increases, the bias and model-based standard errors decrease, and the performance improves with larger sample sizes. The estimates of functions in the model almost coincide with the true functions, as shown in the figures of the simulation study. Furthermore, the consistency of the obtained estimator is demonstrated via theoretical analysis, and a numerical simulation is performed to illustrate the performance of the proposed estimators. The proposed model is applied to a real-world data set acquired from a multicenter automatic defibrillator implantation trial (MADIT).

The Natural History of Atrial Functional Mitral Regurgitation

BackgroundThe natural history of moderate/severe atrial functional mitral regurgitation (AFMR) is unknown. ObjectivesThe authors sought to study the incidence of left ventricular (LV) systolic dysfunction (LVSD), progression or regression of ≥mild-moderate AFMR, and impact on mortality. MethodsAdults with left atrial (LA) volume index ≥40 mL/m2, ≥mild-moderate AFMR, and follow-up echocardiogram were followed for incident LVSD (ejection fraction <50% and ≥10% lower than baseline), progression of mild-moderate/moderate AFMR to severe, and persistent regression of AFMR to no/trivial. Relation of AFMR progression or regression as time-dependent covariates with all-cause mortality was studied. Incidence of LVSD was compared with patients with no/mild AFMR matched on age, sex, comorbidities and ejection fraction. Patients were followed until mitral intervention, myocardial infarction, or last follow-up. ResultsA total of 635 patients (median age 75 years, 51% female, 96% mild-moderate/moderate AFMR, 4% severe AFMR) were included. Over a median 2.2 years (Q1-Q3: 1.0-4.3 years), incidence rates per 100 person-years were 3.2 for LVSD (P = 0.52 vs patients with no/mild AFMR), 1.9 for progression of AFMR, and 3.9 for regression. Female sex and larger LA volume index were independently associated with progression, whereas younger age, male sex, absent atrial fibrillation, and higher LA emptying fraction were independently associated with regression. Neither AFMR progression nor regression was independently associated with mortality. Instead, independent risk factors for mortality included older age, concentric LV geometry, and higher estimated LV filling and pulmonary pressures. ConclusionsIn patients with predominantly mild-moderate/moderate AFMR, regression of MR was more common than progression, but neither was associated with mortality. Instead, diastolic function abnormalities were more important. Over a median 2-year follow-up, LVSD risk was not increased.

Patient-reported outcome measures in early psychosis: A cross-cultural, longitudinal examination of the self-reported health and self-reported mental health measures in Chennai, India and Montreal, Canada

ObjectiveDespite their acknowledged value, patient-reported outcome measures (PROMs) are infrequently used in psychosis, particularly in low-and middle-income countries. We compared ratings on two single-item PROMs, Self-Rated Health (SRH) and Self-Rated Mental Health (SRMH), of persons receiving similar early psychosis services in Chennai, India and Montreal, Canada. We hypothesized greater improvements in SRH and SRMH in the Chennai (compared to the Montreal) sample. MethodsParticipants (Chennai N = 159/168 who participated in the larger study; Montreal N = 74/165 who participated in the larger study) completed the SRH and SRMH during at least two out of three timepoints (entry, months 12 and 24). Repeated measures proportional odds logistic regressions examined the effects of time (baseline to month 24), site, and relevant baseline (e.g., gender) and time-varying covariates (i.e., symptoms) on SRH and SRMH scores. ResultsSRH (but not SRMH) scores significantly differed between the sites at baseline, with Chennai patients reporting poorer health (OR: 0.33; CI: 0.18, 0.63). While Chennai patients reported similar significant improvements in their SRH (OR: 7.03; CI: 3.13; 15.78) and SRMH (OR: 2.29, CI: 1.03, 5.11) over time, Montreal patients only reported significant improvements in their SRMH. Women in Chennai (but not Montreal) reported lower mental health than men. Higher anxiety and longer durations of untreated psychosis were associated with poorer SRH and SRMH, while negative symptoms were associated with SRH. ConclusionsAs hypothesized, Chennai patients reported greater improvements in health and mental health. The marked differences between health and mental health in Montreal, in contrast to the overlap between the two in Chennai, aligns with previous findings of clearer distinctions between mind and body in Western societies. Cross-context (e.g., anxiety) and context-specific (e.g., gender) factors influence patients' health perceptions. Our results highlight the value of integrating simple PROMs in early psychosis.

Developing deep learning-based strategies to predict the risk of hepatocellular carcinoma among patients with nonalcoholic fatty liver disease from electronic health records

ObjectiveThe accuracy of deep learning models for many disease prediction problems is affected by time-varying covariates, rare incidence, covariate imbalance and delayed diagnosis when using structured electronic health records data. The situation is further exasperated when predicting the risk of one disease on condition of another disease, such as the hepatocellular carcinoma risk among patients with nonalcoholic fatty liver disease due to slow, chronic progression, the scarce of data with both disease conditions and the sex bias of the diseases. The goal of this study is to investigate the extent to which the aforementioned issues influence deep learning performance, and then devised strategies to tackle these challenges. These strategies were applied to improve hepatocellular carcinoma risk prediction among patients with nonalcoholic fatty liver disease. MethodsWe evaluated two representative deep learning models in the task of predicting the occurrence of hepatocellular carcinoma in a cohort of patients with nonalcoholic fatty liver disease (n = 220,838) from a national EHR database. The disease prediction task was carefully formulated as a classification problem while taking censorship and the length of follow-up into consideration. ResultsWe developed a novel backward masking scheme to deal with the issue of delayed diagnosis which is very common in EHR data analysis and evaluate how the length of longitudinal information after the index date affects disease prediction. We observed that modeling time-varying covariates improved the performance of the algorithms and transfer learning mitigated reduced performance caused by the lack of data. In addition, covariate imbalance, such as sex bias in data impaired performance. Deep learning models trained on one sex and evaluated in the other sex showed reduced performance, indicating the importance of assessing covariate imbalance while preparing data for model training. ConclusionsThe strategies developed in this work can significantly improve the performance of hepatocellular carcinoma risk prediction among patients with nonalcoholic fatty liver disease. Furthermore, our novel strategies can be generalized to apply to other disease risk predictions using structured electronic health records, especially for disease risks on condition of another disease.