Illness-death Model Research Articles

SCR-CUSUM: An illness-death semi-Markov model-based risk-adjusted CUSUM for semi-competing risk data monitoring

Assessing the medical quality of hospitals based on control charts has received lots of attention. However, existing control charts produce delay and false alarms when applying to semi-competing risk (SCR) data, which widely exist in biomedical and clinical fields. Some studies have suggested that SCR data are important for investigating the healthcare quality of hospitals, but there are seldom targeted control charts for monitoring the medical quality implied by them. Therefore, in this paper, we propose a risk-adjusted cumulative sum control chart based on illness-death semi-Markov model, named SCR-CUSUM, to monitor the deterioration of hospitals’ medical quality by detecting the shifts of non-terminal and terminal events simultaneously. The chart statistic of SCR-CUSUM shows good interpretability. By replacing the preset log-likelihood ratio with the generalized likelihood ratio, SCR-CUSUM become more sensitive and general. Meanwhile, we provide the theoretical control limit and verify its feasibility. Both of the results of simulation and case study prove that SCR-CUSUM works better than the comparison methods when applying to SCR data. In addition, we also analyze the causes of false and delay alarms for existing control charts using simulation data.

A population-based projection of psoriatic arthritis in Germany until 2050: analysis of national statutory health insurance data of 65 million German population

The population-based prevalence of psoriatic arthritis (PsA) is still unclear and not well described globally. The aim of this study was to conduct a population-based prevalence projection and provide long-term future estimations of PsA patients in Germany until 2050, using the illness–death model and based on historical data. We analyzed the national statutory health insurance data of 65 million population in the German Institute for Medical Documentation and Information between January 2009 and December 2012. We constructed an estimation of the PsA burden among the German population using the relevant epidemiological parameters to project the numbers of patients with PsA in Germany until 2050 under five possible scenarios by varying the incidence and mortality. The overall conservatively estimated prevalence of PsA in Germany in 2019 was 0.31% (95% CI 0.28–0.36%). Women contribute a higher prevalence than men in all five scenarios. In the assumed scenarios with increased incidence, the prevalence of PsA at 60 years of age could rise from 1% in 2019 to more than 3% in 2050 for both genders, with the increase particularly pronounced for women, reaching around 3.5%. However, in the assumed scenarios with decreasing incidence, the prevalence curve may flatten and begin a decreasing trend from 2035 to 2050 for both genders, achieving a prevalence of less than 1% in 2050. Our research is to generate assumed population-based data on PsA in Germany that can serve as a reference for public health stakeholders to prepare an optional intervention. We would expect worryingly high numbers in the coming decades if preventive strategies are not implemented. In the long term, it will be necessary to implement preventive strategies to identify predictors and treat psoriasis symptoms early in order to delay or even prevent the transition of psoriasis to PsA.

Estimating transition intensity rate on interval-censored data using semi-parametric with EM algorithm approach

Phase IV clinical trials are designed to monitor long-term side effects of medical treatment. For instance, childhood cancer survivors treated with chest radiation and/or anthracycline are often at risk of developing cardiotoxicity during their adulthood. Often the primary focus of a study could be on estimating the cumulative incidence of a particular outcome of interest such as cardiotoxicity. However, it is challenging to evaluate patients continuously and usually, this information is collected through cross-sectional surveys by following patients longitudinally. This leads to interval-censored data since the exact time of the onset of the toxicity is unknown. Rai et al. computed the transition intensity rate using a parametric model and estimated parameters using maximum likelihood approach in an illness-death model. However, such approach may not be suitable if the underlying parametric assumptions do not hold. This manuscript proposes a semi-parametric model, with a logit relationship for the treatment intensities in two groups, to estimate the transition intensity rates within the context of an illness-death model. The estimation of the parameters is done using an EM algorithm with profile likelihood. Results from the simulation studies suggest that the proposed approach is easy to implement and yields comparable results to the parametric model.

A Bayesian multivariate spatial approach for illness-death survival models.

Illness-death models are a class of stochastic models inside the multi-state framework. In those models, individuals are allowed to move over time between different states related to illness and death. They are of special interest when working with non-terminal diseases, as they not only consider the competing risk of death but also allow us to study the progression from illness to death. The intensity of each transition can be modelled including both fixed and random effects of covariates. In particular, spatially structured random effects or their multivariate versions can be used to assess spatial differences between regions and among transitions. We propose a Bayesian methodological framework based on an illness-death model with a multivariate Leroux prior for the random effects. We apply this model to a cohort study regarding progression after an osteoporotic hip fracture in elderly patients. From this spatial illness-death model, we assess the geographical variation in risks, cumulative incidences and transition probabilities related to recurrent hip fracture and death. Bayesian inference is done via the integrated nested Laplace approximation.

Estimation of separable direct and indirect effects in a continuous-time illness-death model.

In this article we study the effect of a baseline exposure on a terminal time-to-event outcome either directly or mediated by the illness state of a continuous-time illness-death process with baseline covariates. We propose a definition of the corresponding direct and indirect effects using the concept of separable (interventionist) effects (Robins and Richardson in Causality and psychopathology: finding the determinants of disorders and their cures, Oxford University Press, 2011; Robins et al. in arXiv:2008.06019 , 2021; Stensrud et al. in J Am Stat Assoc 117:175-183, 2022). Our proposal generalizes Martinussen and Stensrud (Biometrics 79:127-139, 2023) who consider similar causal estimands for disentangling the causal treatment effects on the event of interest and competing events in the standard continuous-time competing risk model. Unlike natural direct and indirect effects (Robins and Greenland in Epidemiology 3:143-155, 1992; Pearl in Proceedings of the seventeenth conference on uncertainty in artificial intelligence, Morgan Kaufmann, 2001) which are usually defined through manipulations of the mediator independently of the exposure (so-called cross-world interventions), separable direct and indirect effects are defined through interventions on different components of the exposure that exert their effects through distinct causal mechanisms. This approach allows us to define meaningful mediation targets even though the mediating event is truncated by the terminal event. We present the conditions for identifiability, which include some arguably restrictive structural assumptions on the treatment mechanism, and discuss when such assumptions are valid. The identifying functionals are used to construct plug-in estimators for the separable direct and indirect effects. We also present multiply robust and asymptotically efficient estimators based on the efficient influence functions. We verify the theoretical properties of the estimators in a simulation study, and we demonstrate the use of the estimators using data from a Danish registry study.

17th International Conference on Malignant Lymphoma, Palazzo dei Congressi, Lugano, Switzerland, 13 - 17 June, 2023.

Background: In mantle cell lymphoma (MCL), time to relapse is a critical factor, and patients experiencing disease progression within 24 months from treatment start (POD24) have inferior survival. However, the cut-off at 24 months is arbitrary and potentially overly simplified. We aimed to quantify the impact of the timing of disease progression on overall survival, stratified by the most common first-line treatment concepts (Nordic-MCL, R-Bendamustin, and R-CHOP). Methods: Using the population-based Swedish lymphoma register, we identified all systemically treated MCL patients diagnosed between 2006 and 2018. We supplemented this with information on progression and relapse from medical chart reviews through 2022. Patients were categorized based on first-line treatment (Nordic MCL regimen, R-Bendamustin, R-CHOP, and others). POD was defined as a lack of response to primary therapy (progressive disease [PD]) or an initial response followed by a relapse. To assess the impact of POD timing on survival, we used an illness-death model with transition rates estimated using flexible parametric survival models to predict the five-year overall survival (OS) conditional on either being progression-free or experiencing POD as a smooth function of time since primary treatment. Results: A total of 1193 patients receiving systemic treatment were included. The median age at diagnosis was 70 years, and patients were followed for a median of 4 years (range 0–16 years, alive patients, six years). 33% of patients received R-Bendamustin (n = 387, median age 75 years), 30% received the Nordic MCL regimen (n = 351, median age 62 years), and 23% received R-CHOP (n = 276, median age 72 years). Almost half of the patients (48%, n = 571) experienced POD during follow-up. The five-year OS from treatment start was 47%, and progression-free survival was 32%. Among patients treated in the first line with either R-Bendamustin or the Nordic MCL protocol, those with a POD had substantially lower 5-year OS compared to patients who remained progression-free, also for POD occurring up to six years after primary treatment. For patients treated with R-CHOP, the impact of POD was largest if it occurred within 2–3 years after first-line therapy. Keyword: Aggressive B-cell non-Hodgkin lymphoma No conflicts of interests pertinent to the abstract.

An accelerated failure time regression model for illness-death data: A frailty approach.

This work presents a new model and estimation procedure for the illness-death survival data where the hazard functions follow accelerated failure time (AFT) models. A shared frailty variate induces positive dependence among failure times of a subject for handling the unobserved dependency between the nonterminal and the terminal failure times given the observed covariates. The motivation behind the proposed modeling approach is to leverage the well-known interpretability advantage of AFT models with respect to the observed covariates, while also benefiting from the simple and intuitive interpretation of the hazard functions. A semiparametric maximum likelihood estimation procedure is developed via a kernel smoothed-aided expectation-maximization algorithm, and variances are estimated by weighted bootstrap. We consider existing frailty-based illness-death models and place particular emphasis on highlighting the contribution of our current research. The breast cancer data of the Rotterdam tumor bank are analyzed using the proposed as well as existing illness-death models. The results are contrasted and evaluated based on a new graphical goodness-of-fit procedure. Simulation results and data analysis nicely demonstrate the practical utility of the shared frailty variate with the AFT regression model under the illness-deathframework.

Comment on: Chen et al. Trends and ethnic differences in stroke recurrence and mortality in a biethnic population, 2000–2019: a novel application of an illness-death model. Annals of Epidemiology, in press

Comment on: Chen et al. Trends and ethnic differences in stroke recurrence and mortality in a biethnic population, 2000–2019: a novel application of an illness-death model. Annals of Epidemiology, in press

The Imperative for Population-based Cancer Registration of All Metastatic Cancers

Metastases are the main cause of morbidity and mortality from solid tumors. Surprisingly, population-based cancer registries in various countries, including the National Cancer Institute's Surveillance, Epidemiology, and End Results program in the United States, only capture data on individuals diagnosed with cancers that are metastatic at diagnosis (M1). Metastatic recurrences of previously diagnosed, initially nonmetastatic tumors are missed. Devasia and colleagues specify an illness-death model for chronic disease and estimate that in prostate cancer, which has a large pool of primary disease that may or may not progress to metastases, about half of all metastatic cancers arise as recurrences from initially nonmetastatic disease. Capturing all incident metastatic cancer cases across all tumor types in population-based cancer registries, not only based on initial stage at diagnosis, would be critical to better understand the disparities in metastatic disease burden and the effectiveness of primary prevention, screening, and therapies for primary and metastatic disease. See related article by Devasia et al., p. 659.

Trends and ethnic differences in stroke recurrence and mortality in a biethnic population, 2000–2019: a novel application of an illness-death model

PurposeTo estimate temporal trends in post-stroke outcomes in Mexican Americans (MAs) and non-Hispanic whites (NHWs). MethodsWe included first-ever ischemic strokes from a population-based study in South Texas (n = 5343, 2000–2019). We applied an illness-death model with three jointly specified Cox-type models to estimate ethnic differences and ethnic-specific temporal trends in recurrence (first stroke to recurrence), recurrence-free mortality (first stroke to death without recurrence), recurrence-affected mortality (first stroke to death with recurrence), and postrecurrence mortality (recurrence to death). ResultsMAs had higher rates of postrecurrence mortality than NHWs in 2019 but lower rates in 2000. One-year risk of this outcome increased in MAs and decreased in NHWs, resulting in ethnic differences changing from −14.9% (95% CI −35.9%, −2.8%) in 2000 to 9.1% (1.7%, 18.9%) in 2018. For recurrence-free mortality, lower rates were observed in MAs until 2013. Ethnic differences in 1-year risk changed from −3.3% (95% CI −4.9%, −1.6%) in 2000 to −1.2% (−3.1%, 0.8%) in 2018. For stroke recurrence and recurrence-affected mortality, significant ethnic disparities persisted over the study period. ConclusionsAn ethnic disparity in postrecurrence mortality was newly identified, driven by the increasing trend in MAs but a decreasing trend in NHWs.

Future number of people with diagnosed type 1 diabetes in Germany until 2040: an analysis based on claims data

IntroductionWe aim to project the number of people with diagnosed type 1 diabetes in Germany between 2010 and 2040.Research design and methodsWe first estimate the age-specific and sex-specific incidence and...

How Traumatic Brain Injury History Relates to Brain Health MRI Markers and Dementia Risk: Findings from the 3C Dijon Cohort.

The long-term effects of traumatic brain injury (TBI) with loss of consciousness (LOC) on magnetic resonance imaging (MRI) markers of brain health and on dementia risk are still debated. To investigate the associations of history of TBI with LOC with incident dementia and neuroimaging markers of brain structure and small vessel disease lesions. The analytical sample consisted in 4,144 participants aged 65 and older who were dementia-free at baseline from the Three City -Dijon study. History of TBI with LOC was self-reported at baseline. Clinical Dementia was assessed every two to three years, up to 12 years of follow-up. A subsample of 1,675 participants <80 years old underwent a brain MRI at baseline. We investigated the associations between history of TBI with LOC and 1) incident all cause and Alzheimer's disease (AD) dementia using illness-death models, and 2) neuroimaging markers at baseline. At baseline, 8.3% of the participants reported a history of TBI with LOC. In fully-adjusted models, participants with a history of TBI with LOC had no statistically significant differences in dementia risk (HR = 0.90, 95% CI = 0.60-1.36) or AD risk (HR = 1.03, 95% CI = 0.69-1.52), compared to participants without TBI history. History of TBI with LOC was associated with lower white matter volume (β= -4.58, p = 0.048), but not with other brain volumes, white matter hyperintensities volume, nor covert brain infarct. This study did not find evidence of an association between history of TBI with LOC and dementia or AD dementia risks over 12-year follow-up, brain atrophy, or markers of small vessel disease.

Survival by First-line Treatment Type and Timing of Progression Among Follicular Lymphoma Patients: A National Population-based Study in Sweden.

In follicular lymphoma (FL), progression of disease ≤24 months (POD24) has emerged as an important prognostic marker for overall survival (OS). We aimed to investigate survival more broadly by timing of progression and treatment in a national population-based setting. We identified 948 stage II-IV indolent FL patients in the Swedish Lymphoma Register diagnosed 2007-2014 who received first-line systemic therapy, followed through 2020. Hazard ratios (HRs) with 95% confidence intervals (CIs) were estimated by first POD at any time during follow-up using Cox regression. OS was predicted by POD using an illness-death model. During a median follow-up of 6.1 years (IQR: 3.5-8.4), 414 patients experienced POD (44%), of which 270 (65%) occurred ≤24 months. POD was represented by a transformation in 15% of cases. Compared to progression-free patients, POD increased all-cause mortality across treatments, but less so among patients treated with rituximab(R)-single (HR = 4.54, 95% CI: 2.76-7.47) than R-chemotherapy (HR = 8.17, 95% CI: 6.09-10.94). The effect of POD was similar following R-CHOP (HR = 8.97, 95% CI: 6.14-13.10) and BR (HR = 10.29, 95% CI: 5.60-18.91). The negative impact of POD on survival remained for progressions up to 5 years after R-chemotherapy, but was restricted to 2 years after R-single. After R-chemotherapy, the 5-year OS conditional on POD occurring at 12, 24, and 60 months was 34%, 46%, and 57% respectively, versus 78%, 82%, and 83% if progression-free. To conclude, POD before but also beyond 24 months is associated with worse survival, illustrating the need for individualized management for optimal care of FL patients.

Estimating disease incidence rates and transition probabilities in elderly patients using multi-state models: a case study in fragility fracture using a Bayesian approach

BackgroundMulti-state models are complex stochastic models which focus on pathways defined by the temporal and sequential occurrence of numerous events of interest. In particular, the so-called illness-death models are especially useful for studying probabilities associated to diseases whose occurrence competes with other possible diseases, health conditions or death. They can be seen as a generalization of the competing risks models, which are widely used to estimate disease-incidences among populations with a high risk of death, such as elderly or cancer patients. The main advantage of the aforementioned illness-death models is that they allow the treatment of scenarios with non-terminal competing events that may occur sequentially, which competing risks models fail to do.MethodsWe propose an illness-death model using Cox proportional hazards models with Weibull baseline hazard functions, and applied the model to a study of recurrent hip fracture. Data came from the PREV2FO cohort and included 34491 patients aged 65 years and older who were discharged alive after a hospitalization due to an osteoporotic hip fracture between 2008-2015. We used a Bayesian approach to approximate the posterior distribution of each parameter of the model, and thus cumulative incidences and transition probabilities. We also compared these results with a competing risks specification.ResultsPosterior transition probabilities showed higher probabilities of death for men and increasing with age. Women were more likely to refracture as well as less likely to die after it. Free-event time was shown to reduce the probability of death. Estimations from the illness-death and the competing risks models were identical for those common transitions although the illness-death model provided additional information from the transition from refracture to death.ConclusionsWe illustrated how multi-state models, in particular illness-death models, may be especially useful when dealing with survival scenarios which include multiple events, with competing diseases or when death is an unavoidable event to consider. Illness-death models via transition probabilities provide additional information of transitions from non-terminal health conditions to absorbing states such as death, what implies a deeper understanding of the real-world problem involved compared to competing risks models.

Estimating distribution of length of stay in a multi-state model conditional on the pathway, with an application to patients hospitalised with Covid-19.

Multi-state models are used to describe how individuals transition through different states over time. The distribution of the time spent in different states, referred to as 'length of stay', is often of interest. Methods for estimating expected length of stay in a given state are well established. The focus of this paper is on the distribution of the time spent in different states conditional on the complete pathway taken through the states, which we call 'conditional length of stay'. This work is motivated by questions about length of stay in hospital wards and intensive care units among patients hospitalised due to Covid-19. Conditional length of stay estimates are useful as a way of summarising individuals' transitions through the multi-state model, and also as inputs to mathematical models used in planning hospital capacity requirements. We describe non-parametric methods for estimating conditional length of stay distributions in a multi-state model in the presence of censoring, including conditional expected length of stay (CELOS). Methods are described for an illness-death model and then for the more complex motivating example. The methods are assessed using a simulation study and shown to give unbiased estimates of CELOS, whereas naive estimates of CELOS based on empirical averages are biased in the presence of censoring. The methods are applied to estimate conditional length of stay distributions for individuals hospitalised due to Covid-19 in the UK, using data on 42,980 individuals hospitalised from March to July 2020 from the COVID19 Clinical Information Network.

An indirect assessment of individual-level association between disease-free survival (DFS) and overall survival (OS) for resectable esophageal or gastroesophageal junction cancer (EC/GEJC): A pooled analysis of randomized controlled trials (RCTs) by an illness-death model.

369 Background: Validation of intermediate endpoints such as DFS as surrogates for overall survival (OS) in RCTs at individual-level remains as a challenge due to difficulty of collection of individual-level patient data. This study explores the individual-level correlation between DFS and OS indirectly using reconstructed Kaplan-Meier (KM) curves from the RCTs assessing therapies for (neo)adjuvant and perioperative treatment of EC/GEJC. Methods: An illness-death model previously validated in estimating individual-level association between DFS and OS for four correlation meta-analyses, two of which were published by the GASTRIC group, in early and late stage treatment of gastric cancer was employed. Reconstructed DFS and OS data from 25 RCTs published between 1994 - 2020 was pooled separately for each arm and each endpoint. Pooled data was analyzed to estimate the correlation between DFS and OS, as measured by Pearson’s r, Spearman’s rho, and Kendall’s tau. Predictive accuracy of the approach was evaluated by comparing its OS and restricted mean survival time (RMST) predictions to their counterparts obtained from the reported OS data from the RCTs. Sensitivity of correlation measures were assessed with respect to pre-specified variations in pre-recurrence death probability, post-recurrence mortality rate, and choice of the survival model used to extrapolate reported DFS curves from the RCTs. Results: Across all RCTs, the primary cancers covered were EC (n=13), EC/GEJC(n=9), GEJC (n=2) and GEJC/gastric cancer (n=1). According to the histology, 10 RCTs studied only squamous cell carcinoma (SCC) and 9 RCTs studied only adenocarcinoma (AC) of the disease, whereas the remaining 6 RCTs included varying mixtures of SCC and AC patients. The follow-up duration ranged between 27 and 72 months across RCTs. Comparator and intervention arms of the pooled data had 2499 and 2490 patients, respectively. Predicted Pearson’s r, Spearman’s rho, Kendall’s tau were 0.67, 0.79, and 0.70, respectively with negligibly narrow 95% CIs. Across both arms of the pooled data, on average, predicted OS curves laid within the 95% CIs of the pooled OS KM-curves 92% of the time. Average deviation between the RMSTs under the model-predicted OS curves and pooled OS KM curves was 2%. In sensitivity analyses, the ranges for r, r, and t were (0.56-0.88), (0.69-0.89), and (0.61-0.80), respectively. Conclusions: Results indicate moderate-to-strong correlation between DFS and OS in early-stage treatment of EC/GEJC providing supportive evidence for the validation of DFS as a surrogate for OS. Our approach mitigates the lack of patient level data in endpoint correlation assessment but may be prone to changes in the predicted value of pre-recurrence death probability.

Prognostic factors of readmission and mortality after first heart failure hospitalization: results from EPICAL2 cohort.

We aimed to identify prognostic individual factors in patients with first acute heart failure (HF) hospitalization, considering both death and readmission as part of the natural history of HF. We used data from the observational, prospective, multicentre EPICAL2 cohort study from which we selected incident cases of acute HF alive at discharge. We relied on an illness-death model to identify prognostic factors on first readmission and on mortality before and after readmission. In 451 patients hospitalized for first acute HF, we observed within the year after discharge, 23 (5.1%) deaths before readmission and 270 (59.9%) first readmissions, of which 60 (22.2%) were followed by death of any cause. First, among patient characteristics, only Charlson index≥8 was associated with first readmission [adjusted hazard ratio (aHR)=1.6, 95% confidence interval (CI) (1.1-2.3), P=0.011]. Second, Charlson index≥8 [aHR=4.2, 95% CI (1.2-14.8), P=0.025], low blood pressure (BP) [aHR=12.2, 95% CI (1.9-79.6), P=0.009], high BP [aHR=6.9, 95% CI (1.3-36.4), P=0.023], and prescription of recommended dual or triple HF therapy at index discharge [aHR=0.2, 95% CI (0.1-0.7), P=0.014] were associated with mortality before any readmission. Third, Charlson index≥8 [aHR=2.4, 95% CI (1.1-5.6), P=0.037] and the time to first readmission (per 30days additional) [aHR=1.2; 95% CI (1.1-1.4), P=0.007] were associated with mortality after readmission. Regardless of the prognostic state considered, we showed that comorbidities are of critical prognostic value in a real-world cohort of incident HF cases. This argues in favour of multidisciplinary care in HF.

CO18 Exploring Endpoint Correlation Between Progression Free Survival (PFS) and Overall Survival (OS) in Previously Untreated Metastatic Melanoma Using Pseudo Individual Patient Data (IPD)

Lack of IPD poses challenges in estimating the correlation between two endpoints. We proposed a novel approach to parameterize an illness-death model with pseudo-IPD and applied it to estimate the correlation between PFS and OS in previously untreated metastatic melanoma.

Trajectories of physical and mental functioning over 25years before onset of frailty: results from the Whitehall II cohort study.

Research on frailty, a major contributor to heterogeneity in health, is undertaken on older adults although the processes leading to frailty are likely to begin earlier in the life course. Using repeat data spanning 25years, we examined changes in physical and mental functioning before the onset of frailty, defined using Fried's frailty phenotype (FFP). Functioning was measured using the Short-Form 36 General Health Survey (SF-36) on nine occasions from 1991 (age range 40-63years) to 2015 (age range 63-85years). The poorest of four FFP scores from 2002, 2007, 2012 and 2015 was used to classify participants as frail, pre-frail, or robust. We used linear mixed models with a backward timescale such that time 0 was the person-specific date of frailty classification for frail and pre-frail participants and the end of follow-up for robust participants. Analyses adjusted for socio-demographic factors, health behaviours, body mass index and multi-morbidity status were used to compare SF-36 physical (PCS) and mental (MCS) component summary scores over 25years before time 0 as a function of FFP classification, with estimates extracted at time 0, -5, -10, -15, -20 and -25years. We also used illness-death models to examine the prospective association between SF-36 component summary scores at age 50 and incident FFP-defined frailty. Among 7044 participants of the Whitehall II cohort study included in the analysis [29% female, mean age 49.7 (SD=6.0) at baseline in 1991], 2055 (29%) participants remained robust, and 4476 (64%) became pre-frail and 513 (7%) frail during follow-up. Frail compared with robust participants had lower SF-36 scores at t=-25 before onset of frailty with a difference of 3.4 [95% confidence interval (CI) 1.6, 5.1] in PCS and 1.8 (-0.2, 3.8) in MCS. At t=0, the differences increased to 11.5 (10.5, 12.5) and 9.1 (8.0, 10.2), respectively. The differences in SF-36 between the robust and pre-frail groups, although smaller [at t=0, 1.7 (1.2, 2.2) in PCS and 4.0 (3.4, 4.5) in MCS], were already observed 20 and 25years, respectively, before the onset of pre-frailty. Prospective analyses showed that at age 50, scores in the bottom quartiles of PCS [hazard ratio (HR) compared with the top quartile=2.39, 95% CI 1.85, 3.07] and MCS [HR=1.49 (1.15, 1.93)] were associated with a higher risk of FFP-defined frailty at older ages. Differences in trajectories of physical and mental functioning in individuals who developed physical frailty at older ages were observable 25years before onset of FFP-defined frailty. These findings highlight the need for a life course approach in efforts to prevent frailty.

A bayesian approach to model the underlying predictors of early recurrence and postoperative death in patients with colorectal Cancer

ObjectiveThis study aimed at utilizing a Bayesian approach semi-competing risks technique to model the underlying predictors of early recurrence and postoperative Death in patients with colorectal cancer (CRC).MethodsIn this prospective cohort study, 284 patients with colorectal cancer, who underwent surgery, referred to Imam Khomeini clinic in Hamadan from 2001 to 2017. The primary outcomes were the probability of recurrence, the probability of Mortality without recurrence, and the probability of Mortality after recurrence. The patients ‘recurrence status was determined from patients’ records. The Bayesian survival modeling was carried out by semi-competing risks illness-death models, with accelerated failure time (AFT) approach, in R 4.1 software. The best model was chosen according to the lowest deviance information criterion (DIC) and highest logarithm of the pseudo marginal likelihood (LPML).ResultsThe log-normal model (DIC = 1633, LPML = -811), was the optimal model. The results showed that gender(Time Ratio = 0.764: 95% Confidence Interval = 0.456–0.855), age at diagnosis (0.764: 0.538–0.935 ), T3 stage (0601: 0.530–0.713), N2 stage (0.714: 0.577–0.935 ), tumor size (0.709: 0.610–0.929), grade of differentiation at poor (0.856: 0.733–0.988), and moderate (0.648: 0.503–0.955) levels, and the number of chemotherapies (1.583: 1.367–1.863) were significantly related to recurrence. Also, age at diagnosis (0.396: 0.313–0.532), metastasis to other sites (0.566: 0.490–0.835), T3 stage (0.363: 0.592 − 0.301), T4 stage (0.434: 0.347–0.545), grade of differentiation at moderate level (0.527: 0.387–0.674), tumor size (0.595: 0.500–0.679), and the number of chemotherapies (1.541: 1.332–2.243) were the significantly predicted the death. Also, age at diagnosis (0.659: 0.559–0.803), and the number of chemotherapies (2.029: 1.792–2.191) were significantly related to mortality after recurrence.ConclusionAccording to specific results obtained from the optimal Bayesian log-normal model for terminal and non-terminal events, appropriate screening strategies and the earlier detection of CRC leads to substantial improvements in the survival of patients.