Left-censored recurrent event analysis in epidemiological studies: a proposal for when the number of previous episodes is unknown

Background When dealing with recurrent events in observational studies it is common to include subjects who became at risk before follow-up. This phenomenon is known as left censoring, and simply ignoring these prior episodes can lead to biased and inefficient estimates. We aimed to propose a statistical method that performs well in this setting. Methods Our proposal was based on the use of models with specific baseline hazards. In this, the number of prior episodes were imputed when unknown and stratified according to whether the subject had been at risk of presenting the event before t = 0. A frailty term was also used. Two formulations were used for this “Specific Hazard Frailty Model Imputed” based on the “counting process” and “gap time.” Performance was then examined in different scenarios through a comprehensive simulation study. Results The proposed method performed well even when the percentage of subjects at risk before follow-up was very high. Biases were often below 10% and coverages were around 95%, being somewhat conservative. The gap time approach performed better with constant baseline hazards, whereas the counting process performed better with non-constant baseline hazards. Conclusions The use of common baseline methods is not advised when knowledge of prior episodes experienced by a participant is lacking. The approach in this study performed acceptably in most scenarios in which it was evaluated and should be considered an alternative in this context. It has been made freely available to interested researchers as R package miRecSurv.

Deep Learning Based Customer Preferences Analysis in Industry 4.0 Environment

Customer preferences analysis and modelling using deep learning in edge computing environment are critical to enhance customer relationship management that focus on a dynamically changing market place. Existing forecasting methods work well with often seen and linear demand patterns but become less accurate with intermittent demands in the catering industry. In this paper, we introduce a throughput deep learning model for both short-term and long-term demands forecasting aimed at allowing catering businesses to be highly efficient and avoid wastage. Moreover, detailed data collected from a business online booking system in the past three years have been used to train and verify the proposed model. Meanwhile, we carefully analyzed the seasonal conditions as well as past local or national events (event analysis) that could have had critical impact on the sales. The results are compared with the best performing forecast methods Xgboost and autoregressive moving average model (ARMA), and they suggest that the proposed method significantly improves demand forecasting accuracy (up to 80%) for dishes demand along with reduction in associated costs and labor allocation.

Barriers to applying for medical rehabilitation: a time-to-event analysis of employees with severe back pain in Germany

Objective: Longitudinal studies on barriers to applying for rehabilitation in Germany are lacking in light of the suspected underutilization of rehabilitation services. The aim of this study was to examine application behaviour in persons with disabling back pain and to identify relevant predictors for making an application. Design: A prospective cohort study with randomized sampling of insurants in the German Pension Insurance, using a questionnaire at baseline and follow-up with linked administrative data for 1.5 years. Subjects/patients: Employed persons (age range 45–59 years) with a high degree of limitations due to back pain and a self-reported risk of permanent work disability (not applied for disability pension, no medical rehabilitation within the last 4 years). Methods: Multivariable Cox regression was used to examine the influence of pre-selected variables on making an application in the follow-up period. Results: Of 690 persons, only 12% applied for rehabilitation. Predictors for making an application were: support from physicians (hazard ratio (HR)=2.24; 95% confidence interval (95% CI) 1.32–3.80), family, and friends (HR=1.67; 95% CI 1.02–2.73), more pain-related disability days (HR=1.02; 95% CI 1.01–1.03), and worse work ability (HR=0.86; 95% CI 0.75–0.97). An intention to apply at baseline mediated the effect of family and physician support on the application. Conclusion: The low number of applications for rehabilitation despite disabling back pain indicates access barriers to, and underuse of, medical rehabilitation.

Infectious outcomes of fibrin sheath disruption in tunneled dialysis catheters

Background and objectives: Fibrin sheath (FS) formation around tunneled central venous catheters (CVC) increases the risk of catheter-related bloodstream infections due to bacterial adherence to a biofilm. We sought to investigate whether FS disruption (FSD) at the time of CVC removal or exchange affects infectious outcomes in patients with CVC-related infections. Design, setting, participants, and measurements: Retrospective cohort study of 307 adult maintenance hemodialysis patients aged 18 years or older at a single center academic-based hemodialysis program (UHN, Toronto) who developed CVC-related infections requiring CVC removal or exchange between January 2000 and January 2019. Exposure was FSD at the time of CVC removal or exchange. Outcomes were infectious metastatic complications, recurrent infection with the same organism within 1 year, or death due to infection. We created a Markov Multi-State Model (MMSM) to assess patients’ trajectories through time as they transitioned between states. A time-to-event analysis was performed, adjusted for clinically relevant factors. Results: There was no significant relationship between FSD status at the time of CVC removal, the development of infectious complications in the multivariable model (adjusted HR = 0.71, 95% CI 0.09−5.80, p = 0.76), or mortality from infection (HR = 0.84, 95% CI 0.34−2.11, p = 0.73). Conclusions: FSD at the time of CVC removal was not associated with increased risk of infectious complications or death due to infection. Further prospective study is needed to determine whether FSD contributes to reducing CVC infectious related complications.

Daily Lactobacillus Probiotic versus Placebo in COVID-19-Exposed Household Contacts (PROTECT-EHC): A Randomized Clinical Trial

Background: The COVID-19 pandemic continues to pose unprecedented challenges to worldwide health. While vaccines are effective, supplemental strategies to mitigate the spread and severity of COVID-19 are urgently needed. Emerging evidence suggests susceptibility to infections, including respiratory tract infections, may be reduced by probiotic interventions. Therefore, probiotics may be a low-risk, widely implementable modality to mitigate risk of COVID-19 disease, particularly in areas with low vaccine availability and/or uptake. Methods: We conducted a randomized, double-blind, placebo-controlled trial across the United States testing the probiotic Lactobacillus rhamnosus GG (LGG) as post-COVID-19-exposure prophylaxis. We enrolled individuals > 1 year of age with a household contact with a recent (≤ 7 days) diagnosis of COVID-19. Participants were randomized to receive daily LGG or placebo for 28 days. Stool was collected to evaluate the microbiome. The primary outcome was development of symptoms of illness compatible with COVID-19 within 28 days. Findings: We enrolled 182 COVID-19-exposed participants. Intention-to-treat analysis showed that participants randomized to LGG were less likely to develop symptoms versus those randomized to placebo (26.4% vs. 42.9%, p=0.02). Further, LGG was associated with a statistically significant reduction in COVID-19 diagnosis (log rank p=0.049) via time-to-event analysis. Overall incidence of COVID-19 diagnosis was not significantly different between LGG (8.8%) and placebo (15.4%) (p=0.17). LGG was well-tolerated with no increased side effects versus placebo. Interpretation: These findings suggest that LGG probiotic may protect against the development of COVID-19 infection and symptoms when used as post-exposure prophylaxis within 7 days after exposure. Funding: This work was supported by a grant from the Duke Microbiome Center to A.D.S. and P.E.W. and private philanthropic donations to A.D.S. DSM/iHealth donated the LGG and placebo for the trial but had no role in its design, conduct, analysis, or writing. Trial registration: NCT04399252

Comparison of infectious SARS-CoV-2 from the nasopharynx of vaccinated and unvaccinated individuals

The frequency of SARS-CoV-2 breakthrough infections in fully vaccinated individuals increased with the emergence of the Delta variant, particularly with longer time from vaccine completion. However, whether breakthrough infections lead to onward transmission remains unclear. Here, we conducted a study involving 125 patients comprised of 72 vaccinated and 53 unvaccinated individuals, to assess the levels of infectious virus in in vaccinated and unvaccinated individuals. Quantitative plaque assays showed no significant differences in the titers of virus between these cohorts. However, the proportion of nasopharyngeal samples with culturable virus was lower in the vaccinated patients relative to unvaccinated patients (21% vs. 40%). Finally, time-to-event analysis with Kaplan-Myer curves revealed that protection from culturable infectious virus waned significantly starting at 5 months after completing a 2-dose regimen of mRNA vaccines. These results have important implications in timing of booster dose to prevent onward transmission from breakthrough cases.

Latent class analysis of 216 patients with adult-onset Still’s disease

Background Adult-onset Still’s disease (AOSD) is a rare systemic autoinflammatory disease which encompasses patients with heterogenous presentation and a wide range of clinical courses. In this study, we aimed to identify potential subgroups of AOSD and reveal risk factors for relapse. Methods We included a total of 216 AOSD patients who received treatment in nine hospitals between 2000 and 2019. All patients fulfilled the Yamaguchi classification criteria. We retrospectively collected information about baseline characteristics, laboratory tests, treatment, relapse, and death. We performed latent class analysis and time-to-event analysis for relapse using the Cox proportional hazard model. Results The median age at disease onset was 51.6 years. The median follow-up period was 36.8 months. At disease onset, 22.3% of the patients had macrophage activation syndrome. The median white blood cell count was 12,600/μL, and the median serum ferritin level was 7230 ng/mL. Systemic corticosteroids were administered in all but three patients (98.6%) and the median initial dosage of prednisolone was 40mg/day. Ninety-six patients (44.4%) were treated with concomitant immunosuppressants, and 22 (10.2%) were treated with biologics. Latent class analysis revealed that AOSD patients were divided into two subgroups: the typical group (Class 1: 71.8%) and the elderly-onset group (Class 2: 28.2%). During the follow-up period, 13 of 216 patients (6.0%) died (12 infections and one senility), and 76 of 216 patients (35.1%) experienced relapses. Overall and relapse-free survival rates at 5 years were 94.9% and 57.3%, respectively, and those rates were not significantly different between Class 1 and 2 (p=0.30 and p=0.19). Time-to-event analysis suggested higher neutrophil count, lower hemoglobin, and age ≥65 years at disease onset as risk factors for death and age ≥65 years at disease onset as a risk factor for relapse. Conclusions AOSD patients were divided into two subgroups: the typical group and the elderly-onset group. Although the survival of patients with AOSD was generally good, the patients often experienced relapses. Age ≥65 years at disease onset was the risk factor for relapse.

Gender Related Differences in Gastrointestinal Bleeding With Oral Anticoagulation in Atrial Fibrillation

Background: DOACs are characterized by a higher incidence of gastrointestinal bleeding and this may be different among males and females. Female patients were underrepresented in the DOAC pivotal trials. We aimed to assess real-world differences in gastrointestinal bleeding with oral anticoagulants (DOACs and VKAs) among males and females with atrial fibrillation. Methods: We performed a population-based retrospective analysis on linked administrative claims. Atrial fibrillation patients of 65 years and above were considered. Bleeding risk factors were assessed through HASBED and previous history of gastrointestinal disease. A time-to-event analysis compared gastrointestinal bleeding between males and females. Results: The overall cohort consisted of 15338 (55% female) DOAC and 44542 (50% female) VKA users. Most of the patients showed HASBED ≥2. Incidence rate of GI bleeding was higher in females as compared to males among DOAC users (0.90% vs 0.59%), and significant gender difference in GI bleeding was found, after adjustment, in the Cox regression analysis (HR 1.48, 95%CI 1.02-2.16). In the VKA group, no significant difference among genders was found in the time-to-event analysis. Conclusions: Our data suggest that female patients treated with DOACs have a higher risk of GI bleeding versus male patients; this difference is not observed in VKA patients.