Does Machine Learning Offer Added Value Vis-à-Vis Traditional Statistics? An Exploratory Study on Retirement Decisions Using Data from the Survey of Health, Ageing, and Retirement in Europe (SHARE)

Do machine learning algorithms perform better than statistical survival analysis when predicting retirement decisions? This exploratory article addresses the question by constructing a pseudo-panel with retirement data from the Survey of Health, Ageing, and Retirement in Europe (SHARE). The analysis consists of two methodological steps prompted by the nature of the data. First, a discrete Cox survival model of transitions to retirement with time-dependent covariates is compared to a Cox model without time-dependent covariates and a survival random forest. Second, the best performing model (Cox with time-dependent covariates) is compared to random forests adapted to time-dependent covariates by means of simulations. The results from the analysis do not clearly favor a single method; whereas machine learning algorithms have a stronger predictive power, the variables they use in their predictions do not necessarily display causal relationships with the outcome variable. Therefore, the two methods should be seen as complements rather than substitutes. In addition, simulations shed a new light on the role of some variables—such as education and health—in retirement decisions. This amounts to both substantive and methodological contributions to the literature on the modeling of retirement.

HAT2CH2 Score Predicts Systemic Thromboembolic Events in Elderly After Cardiac Electronic Device Implantation

Background: The HAT2CH2 score has been evaluated for predicting new onset atrial fibrillation, but never for adverse systemic thromboembolic events (STE) in elderly. We aimed to evaluate the HAT2CH2 score and comparing to atrial high rate episodes (AHRE) ≥24 h for predicting STE in older patients with cardiac implantable electronic devices (CIED) implantation.Methods: We retrospective enrolled 219 consecutive patients ≥ 65 years of age undergoing CIED implantation. The primary endpoint was subsequent STE. For all patients in the cohort, the CHA2DS2-VASc, C2HEST, mC2HEST, HAVOC, HAT2CH2 scores and AHRE ≥ 24 h were determined. AHRE was defined as > 175 bpm lasting ≥ 30 s. Multivariate Cox regression analysis with time-dependent covariates was used to determine variables associated with independent risk of STE.Results: The median patient age was 77 years, and 61.2% of the cohort was male. During follow-up (median, 35 months), 16 STE occurred (incidence rate, 2.51/100 patient-years; 95% CI, 1.65–5.48). Multiple Cox regression analysis showed that the HAT2CH2 score (HR, 3.405; 95% CI, 2.272–5.104; p < 0.001) was an independent predictor for STE. The optimal HAT2CH2 score cutoff value was 3, with the highest Youden index (AUC, 0.907; 95% CI, 0.853–0.962; p < 0.001). The STE rate increased with increasing HAT2CH2 score (p < 0.001).Conclusions: This study is the first to show the prognostic value of the HAT2CH2 score for STE occurrence in older patients with CIEDs.

Letter to the Editor: Response to ‘Overlooked Shortcomings of Observational Studies of Interventions in Coronavirus Disease 2019: An Illustrated Review for the Clinician‘ by Tleyjeh et al

We commented on the publication by Tleyjeh et al regarding the overlooked shortcomings of observational studies of interventions in Coronavirus Disease 2019. Although we agree with Tleyjeh and colleagues on the issue of the competing risk bias in observational studies, the recommendations on the application of the Fine-Gray model provided by the authors are incomplete. The Fine-Gray approach may not be suitable in the presence of interval time-dependent covariates, that are often the case in the studies assessing therapeutic interventions for patients with Coronavirus Disease 2019.

Joint Modelling Approaches to Survival Analysis via Likelihood-Based Boosting Techniques

Joint models are a powerful class of statistical models which apply to any data where event times are recorded alongside a longitudinal outcome by connecting longitudinal and time-to-event data within a joint likelihood allowing for quantification of the association between the two outcomes without possible bias. In order to make joint models feasible for regularization and variable selection, a statistical boosting algorithm has been proposed, which fits joint models using component-wise gradient boosting techniques. However, these methods have well-known limitations, i.e., they provide no balanced updating procedure for random effects in longitudinal analysis and tend to return biased effect estimation for time-dependent covariates in survival analysis. In this manuscript, we adapt likelihood-based boosting techniques to the framework of joint models and propose a novel algorithm in order to improve inference where gradient boosting has said limitations. The algorithm represents a novel boosting approach allowing for time-dependent covariates in survival analysis and in addition offers variable selection for joint models, which is evaluated via simulations and real world application modelling CD4 cell counts of patients infected with human immunodeficiency virus (HIV). Overall, the method stands out with respect to variable selection properties and represents an accessible way to boosting for time-dependent covariates in survival analysis, which lays a foundation for all kinds of possible extensions.

HAT2 CH2 Score Performance Predicting Neurologic Events After Cardiac Implantable Electronic Device

The HAT2CH2 score has been evaluated for predicting new-onset atrial fibrillation in several clinical conditions, but never for adverse neurologic events. We aimed to evaluate the effectiveness of HAT2CH2 score in predicting neurologic events in patients with cardiac implantable electronic device (CIED), comparing with atrial high-rate episodes (AHRE). This case-control study enrolled 314 consecutive patients aged 18 years or older with CIED implantation between January 2015 and April 2021. Patient data were analyzed retrospectively. The primary endpoint was subsequent neurologic events (NE) after implantation. AHRE was defined as > 175 bpm (Medtronic®) lasting ≥ 30 seconds. Variables associated with independent risk of NE were identified using multivariate Cox regression analysis with time-dependent covariates. Patients’ median age was 73 years and 61.8% of them were male. During follow-up (median 32 months), 18 NE occurred (incidence rate 2.15/100 patient-years, 95% CI 1.32-4.30). Multiple Cox regression analysis showed that the HAT2CH2 score (HR 2.972, 95% CI 2.143-4.123, p < 0.001) was an independent predictor for NE. Optimal HAT2CH2 score cutoff value was 3 with highest Youden index (AUC, 0.923; 95% CI, 0.881–0.966; p < 0.001). Significant increase was observed in NE occurrence rates using the HAT2CH2 score (p < 0.001). The HAT2CH2 score and episodes of AHRE lasting ≥ 1 minute are independent risk factors for NE in patients with CIED.

HATCH Score and Left Atrial Size Predict Atrial High-Rate Episodes in Patients With Cardiac Implantable Electronic Devices

Background: Patients with sustained atrial high-rate episodes (AHRE) have a high risk of major adverse cardio/cerebrovascular events (MACCE). However, the prediction model and factors for the occurrence of AHRE are unknown. We aimed to identify independent factors and various risk models for predicting MACCE and AHRE.Methods: We retrospectively enrolled 314 consecutive patients who had cardiac implantable electronic devices (CIEDs). The primary endpoint was MACCE after AHRE ≥3, 6 min, and 6 h. Atrial high-rate episodes was defined as &gt;175 bpm (Medtronic®) lasting ≥30 s. Multivariate Cox and logistic regression analysis with time-dependent covariates were used to determine variables associated with independent risk of MACCE and occurrence of AHRE ≥3 min, respectively.Results: One hundred twenty-five patients (39.8%) developed AHRE ≥3 min, 103 (32.8%) ≥6 min, and 55 (17.5%) ≥6 h. During follow-up (median 32 months), 77 MACCE occurred (incidence 9.20/100 patient years, 95% CI 5.66–18.39). The optimal AHRE cutoff value was 3 min for MACCE, with highest Youden index 1.350 (AUC, 0.716; 95% CI, 0.638–0.793; p &lt; 0.001). Atrial high-rate episodes ≥3 min−6 h were independently associated with MACCE. HATCH score and left atrial diameter were independently associated with AHRE ≥3 min. The optimal cutoff for HATCH score was 3 and for left atrial diameter was 4 cm for AHRE ≥3 min.Conclusion: Patients with CIEDs who develop AHRE ≥3 min have an independently increased risk of MACCE. Comprehensive assessment using HATCH score and echocardiography of patients with CIEDs is warranted.

Atrial high-rate episodes predict major adverse cardio/cerebrovascular events in patients with cardiac implantable electrical devices

Patients with atrial high-rate episodes (AHRE) have a high risk of neurologic events, although the causal role and optimal cutoff threshold of AHRE for major adverse cardio/cerebrovascular events (MACCE) are unknown. This study aimed to identify independent factors for AHRE and subsequent atrial fibrillation (AF) after documented AHRE. We enrolled 470 consecutive patients undergoing cardiac implantable electrical device (CIED) implantations. The primary endpoint was subsequent MACCE after AHRE ≥ 6 min, 6 h, and 24 h. AHRE was defined as > 175 beats per minute (bpm) (Medtronic®) or > 200 bpm (Biotronik®) lasting ≥ 30 s. Multivariate Cox regression analysis with time-dependent covariates was used to determine variables associated with independent risk of MACCE. The patients’ median age was 76 year, and 126 patients (26.8%) developed AHRE ≥ 6 min, 63 (13.4%) ≥ 6 h, and 39 (8.3%) ≥ 24 h. During follow-up (median: 29 months), 142 MACCE occurred in 123 patients. Optimal AHRE cutoff value was 6 min, with highest Youden index for MACCE. AHRE ≥ 6 min ~ 24 h was independently associated with MACCE and predicted subsequent AF. Male gender, lower body mass index, or BMI, and left atrial diameter were independently associated with AHRE ≥ 6 min ~ 24 h. Patients with CIEDs who develop AHRE ≥ 6 min have an independently increased risk of MACCE. Comprehensive assessment of patients with CIEDs is warranted.

Additive rates model for recurrent event data with intermittently observed time-dependent covariates

Various regression methods have been proposed for analyzing recurrent event data. Among them, the semiparametric additive rates model is particularly appealing because the regression coefficients quantify the absolute difference in the occurrence rate of the recurrent events between different groups. Estimation of the additive rates model requires the values of time-dependent covariates being observed throughout the entire follow-up period. In practice, however, the time-dependent covariates are usually only measured at intermittent follow-up visits. In this paper, we propose to kernel smooth functions involving time-dependent covariates across subjects in the estimating function, as opposed to imputing individual covariate trajectories. Simulation studies show that the proposed method outperforms simple imputation methods. The proposed method is illustrated with data from an epidemiologic study of the effect of streptococcal infections on recurrent pharyngitis episodes.