This study aims to compare traditional time-to-first- event model with approaches that take into account multiple cardiovascular burden and/or recurrency in the context of a prospective cohort study- the Cardio- vascular Health Study (CHS). In order to provide direct comparison of the methods and derive a real-world esti- mate of the extent of understated cardiovascular risk by using time-to-first-event approaches, we selected to use a class of parametric survival models that is amenable to different approaches and then applied the various ap- proaches to a single significant epidemiologic data set – the CHS. The selected parametric survival approaches for recurrent and multiple event-types relied heavily on random effects models. As a result, the direct effects es- timated using different approaches cannot be immedi- ately compared. The fixed effect for smoking, in the random effect model, for example, represents an esti- mate that is conditional on the random effect. One way to solve the problem is to “marginalize” the mixed ef- fects models – i.e., first estimate the mixed effects model and then integrate out the random effects based on the estimated parameters from the previous step [20,21]. To summarize, instead of proposing new epide- miologic methods for multiple event-types and recur- rent events, our goal here is two-folded: (1) to present results from applying comparable models representing different approaches for handling time-to-event data, and (2) to assess the level of discrepancies in traditional risk reporting on a collection of well-known cardiovas- cular risk factors of their effects on overall cardiovascu- lar burden.
symptom expression in the medical and epide- miological domains; however there are docu- mented examples of its application within ecology. Muenchow (1986) advocated the use of such models in ecology and proposed a number of ecological questions that could be phrased in terms of ‘‘time until an event occurs.’’ Examples include flower visitation events by insects (Muenchow 1986), time of fish passage in rivers (Zabel et al. 2014), tree mortality (Woodall et al. 2005) and duration of tarantula fighting (Moya-Lora ˜no and Wise 2000). Most of these authors note the novelty of the application of survivalanalysis within their specific fields. One particular area of expansion of its applica- tion has been in plant pathology (see Scherm and Ojiambo 2004 for summary) and it is here that we found the only example of the use of recurrent event survivalanalysis within ecology (Thomson and Copes 2009). We were particu- larly interested in the variant of these recurrent event survival models called the marginal stratified Cox proportional hazards model. In the marginal approach each event is considered as a separate process (i.e., there is no condition on events being progressive, such as in a disease where symptoms have to occur in a certain order), and different response events can repre- sent different response types that may occur in the same subject (Kleinbaum and Klein 2005).
Survivalanalysis of time to first event, hospitalisation for recurrent stroke or death, was undertaken using both the Kaplan-Meier and cumulative incidence meth- ods . In both methods survival time is censored at death after stroke (a competing risk event) but the Kaplan-Meier method ignores the competing event in calculation of survival probabilities whereas the cumula- tive incidence method takes the competing event into account. When there are competing events it aids inter- pretation if results are shown for both the event of interest and the competing event(s) . For this reason, we show results for both hospitalisation for recurrent stroke and death. Complete five-year follow-up was obtained for all patients. Descriptive statistics of the demographics of the group of individuals hospitalised with incident stroke and unadjusted risk of first events are presented for all years as well as for periods at the start and end of the study period (1986 to 1989 and 1998 to 2001). This allows assessment of how demo- graphics and risks have changed over time. We used regression on the cumulative incidence functions  to model the temporal trends of first event with adjust- ment for age, sex, socioeconomic status and comorbid- ity. We used restricted cubic splines  to determine the best fitting relationship between the survivalevents and study year. A significance level of 0.05 was used throughout. All analyses were carried out using the sta- tistical package R.
The main purpose of this example is to illustrate the application of recurrent event techniques and pro- vide comparisons between techniques, rather than giv- ing estimates for recurrence of respiratory illnesses. Two characteristics, parents reported smoking (1=yes, 0=no) and family history of asthma (1=yes, 0=no) were included as two univariate analysis in each of the models studied. We chose these two covariates because they are poten- tially related with the outcome as indicated in the medical literature and also because they are binary variables as we used in our simulation study. Table 3 contains the results of three different models: AG, PWP and WLW, fitted to the respiratory data. The β values estimated by the AG and PWP models are lower than those estimated by the WLW model for both covariates, smoking and asthma. But all of them show a negative effect in the outcome. When we analyze the standard errors, the robust standard errors are larger in all models than those estimated by the naive method. And again, the WLW model has the biggest standard error in comparison to the other models. These results are coherent with the previous simulation results.
process of the multivariate counting process. Prentice et al. (1981), Wei et al. (1989), Lawless and Nadeau (1995), Sun et al. (2004), among others considered marginal rates (or mean) model to evaluate the effect of the risk factors on recurrent event data. Some examples include the occurrence of new tumours in patients with superficial bladder cancer (Byar, 1980), recurrent seizures in epileptic patients (Albert 1991), rejection episodes in patients receiving kidney transplants (Cole et al. 1994), repeated infections in HIV-patients (Li and Lagakos 1997) and repeated cardiovascular events in patients (Cui et al. 2008). The two primary frameworks to study the association between the risk factors and the disease recurrence are the additive and multiplicative rates models. Most modern analysis of survival data address multiplicative models for relative risk (rates) using proportional rates models, mostly due to desirable theoretical properties along with the easy interpretation of results (Pepe and Cai 1993, Lawless 1995, Lin et al. 2000, Schaubel et al. 2006, Kang and Cai 2009a). However, researchers may be interested in the risk (rate) difference, rather than the relative measure, attributed to the exposure. Further, the risk difference is more relevant to the public health as it translates directly to the number of disease cases that may be avoided by eliminating the exposure (Kulich and Lin 2000). Consequently, the additive rates models can be considered as an alternative to the multiplicative model (Schaubel et al. 2006, Yin and Cai 2004, Zeng and Cai 2010, Liu et al. 2013, Kang et al. 2013, He et al. 2013).
Patient charts were reviewed, and the following data were collected: demographic information, possible etiology of HCC, imaging appearance of the tumor, extent and location of the extrahepatic disease, laboratory data before and after the Y90 RMS procedure, dosage and duration of treatment with sorafenib before and after treatment with Y90 RMS, Y90 RMS treatment details, Response Evaluation Criteria in Solid Tumors (RECIST) and European Association for the Study of the Liver (EASL) responses according to imaging, other concurrent therapies given, adverse events occurring after treatment with Y90 RMS according to Common Terminology Criteria for Adverse Events (CTCAE), sorafenib-related toxic effects, and OS duration, which was defined as the time interval between the start of the treatment with sorafenib and the date of death or last follow-up. The data obtained were also analyzed to determine the PFS duration for each patient, which was defined as the time interval between the day of initiation of sorafenib and the date of disease progression or death. We also determined PFS specifically for hepatic disease and extrahepatic disease.
Abstract: Background: Relapse in osteosarcoma is associated with very poor prognosis. The prognostic value of time to relapse in recurrent osteosarcomas remains controversial. Hence, a meta-analysis was conducted to investigate the effect of relapse-free interval (RFI) on post-relapse survival (PRS). Methods: From inception to November 2015, we searched for cohort studies using the following databases: PubMed, Cochrane Library, EBSCO and Science Direct. Eligible studies should provide the number of patients with a short and long RFI (24-months as a cutoff usu- ally) and their corresponding 5-year PRS. The pooled relative risk (RR) with 95% confidence interval (95% CI) was used to assess the impact of RFI on PRS. Results: A total of 11 studies published between 2003 and 2014 were found to be in accordance with our inclusion criteria. 1692 cases of recurrent osteosarcomas were enrolled in this meta-analysis. Approximately 60% relapses occur early, usually within 24 months. The primary meta-analysis indi- cated that recurrent patients with a short RFI had a worse PRS (n=11, RR=2.63, 95% CI: 2.17 to 3.21, P<0.001). By subgroup analyses, we further found that 24-months was a common and alternative cutoff value (n=7, RR=2.37, 95% CI: 1.84 to 3.06, P<0.001). No significant differences were obtained after stratified by age range, sample size, and geographic region. And a complete surgical remission (CR) of primary tumor correlated with a higher probability of a long RFI (subgroup difference: P=0.01). Conclusion: The findings of our meta-analysis suggest that late relapses fares a better PRS in recurrent osteosarcomas. We recommend 24-months as a clinically alternative cutoff value and long-term follow-up. Complete surgical resection, if feasible, may be prerequisite for late relapse.
Our main finding is that an increased life event load, taking into account the number and threat of life events, impacts both first and recurrent admissions in bipolar patients. This has also been found in previous studies (Bender and Alloy 2011; Hunt et al. 1992; Kessing et al. 1998, 2004), but it was hypothesized that this might be due to life events occurring as a consequence of the dis- ease (Kessing et al. 2004). We now extended these previ- ous findings by showing that the effect of life events on admissions did not change when events related to the disorder were excluded from the analyses. This suggests that the effect of life events is independent of life events occurring in relation to the disorder. We consider this robust influence of life events on first and recurrent ad- missions an important finding, as exposure and re- sponses to life events are potentially modifiable. A better
Physical examination revealed blurred margins of both optic discs. Her Karnofsky Performance Status (KPS) score was 50. The ANP treatment was started at the dosage of 1.23 g/kg/day of IV A10 and 0.15 g/kg/day of AS2-1. Dosage of A10 and AS2-1 was gradually increased, according to protocol, to 10.48 g/kg/day and 0.17 g/kg/day, respectively. ANP was discontinued after 18.5 months due to CR. The only adverse drug experiences (ADE) possibly related to treatment was grade 1 skin rash, which resolved spontaneously, and grade 1 hyperna- tremia. The adverse events for all three cases were graded according to the Common Terminology Criteria for Adverse Events Version 3.0 (CTCAE v.3). After stopping treatment with IV ANP, the patient was converted to capsules of ANP A10 and AS2-1 (0.5 g), initially 0.05 g/kg/day of both. The dosage of both A10 and AS2-1 was gradually increased to 0.14 g/kg/day. All therapy was discontinued after 31 months from treatment start. During the course of treatment, the patient made substantial recovery. At 9 months into the treatment with IV ANP, she was able to walk with the help of a walker and made progress in her speech therapy. She continued to have fur- ther improvement, as indicted by her scheduled physical examinations. Her KPS increased to 80. During her examination at 4 years after treatment completion, she was able to walk with a cane. Her baseline MRI revealed a right temporal enhancing nodule measuring 1.3 × 1.2 cm. There was continuous improvement on follow-up MRIs. Beginning 4 months after start of ANP, MRI images showed a CR. Follow-up MRIs, repeated every 8 weeks to 3 months, with the latest one performed in October 2013, did not show any recurrent tumor (Figure 1).
recurrentevents, and conditions on having had previous events. In contrast, the marginal model treats the events as unordered, and all subjects are at risk for any event. In dif- ferent trials the outcomes of interest and validity of assumptions will differ. Our guidelines for reporting results from an RCT involving a recurrent event suggest statistical methods which correspond to the objectives of the trial, such as addressing the study question of interest, assessing comparable groups and estimating effect size. First, the average event rate by intervention group is a measure of the average number of events accrued per per- son-time. These event rates serve an important role in determining sample size and follow-up time for the design of future RCTs involving recurrentevents . Sec- ond, the MCF by intervention group provides a measure of the average number of events experienced per subject within a certain time. The MCF allows us to determine how many events per subject the intervention would pre- vent, on average, compared to the control group . Third, the common rate ratio, as measured by the gamma- Poisson and independent-increment models, quantifies the average rate of event in the intervention group relative to the control group. This rate ratio provides an estimate of the common effect size, thereby indicating whether the intervention had an impact on the event occurrence. Fourth, conditional event-specific rate ratios, which quan- tify the rate of the kth event in the intervention relative to the control, conditional on experiencing preceding events, should be reported. These rate ratios allow us to evaluate how the effect of intervention changes, if at all, on subsequent events. Lastly, we suggest reporting the marginal event-number-specific rate ratios, which repre- sent the rate of transitioning to higher-order events from the start of treatment in the intervention group relative to the control group. These rate ratios allow us to evaluate the overall protective effect of intervention. For methods used in the assessment of goodness of fit for each model we refer the reader to the corresponding papers [17,27]. It has been argued that the average event rate might have little relevance in the context of recurrentevents because this measure does not acknowledge dependence between
SA was originally developed in bio-informatics to organize, classify, and parse protein and DNA sequence data (Durbin, Eddy, Krogh, & Mitchison, 1998). In the social sciences, Abbott introduced the use of SA in life course analysis in the mid-1980s (Abbott, 1983; Abbott, 1995; Abbott & Tsay, 2000). The basic idea in SA is to measure the distance or dissimilarity of two sequences consisting of the succession of categorical states describing the trajectories. Two major issues are essential for SA. The first concerns the composition of sequences: how many and what type of states? The second issue is related to determining the dissimilarities between the sequences: which dissimilarity measure to use and, for some measures, how to assign the “cost” of converting one state to another? Typical steps in SA include the following: 1) creating sequences using a finite set of states; 2) choosing and implementing a method for computing pairwise dissimilarities between sequences; 3) analysing the dissimilarities (e.g. cluster analysis and/or multi-dimensional scaling); 4) graphical illustration and examination of sequence data.
Abstract Two new methods of analyzing dialogue interactions are outlined. One method depends on abstract representations of dialogue events as symbols in a for- mal language. This method invites analysis of the expressivity requirements of dia- logue grammar, as well as distribution analysis of dialogue event symbol sequences. The method is presented in relation to a temporal construction of regular languages, one which supports increasingly fine granularity of temporal analysis. The other method proposed is also temporally oriented. It also depends on dialogue events and dialogue states, and proposes to analyze causal relations among dialogue events through survivalanalysis. These methods are suggested as additions to the extant repertoire of approaches to understanding the structure and temporal flow of nat- ural dialogue. Additional methods of analysis of natural dialogue may contribute to deeper understanding of the phenomena. With deeper undertanding of natural dialogue one may hope to more fully inform the construction of believable artifi- cial systems that are intended to engage in dialogue with a manner close to human interaction in dialogue.
in the patient’s body. Surgical intervention is difficult and diffuse intrinsic pontine glioma (DIPG) is considered first with poor prognosis in brain tumors. Prior statistics provided various incidences of brainstem glioma (BSG). The most recent Central Brain Tumor Registry of the United States (CBTRUS) reports only 3.6% distribution among malignant primary brain and central nervous system (CNS) tumors by site (N = 112,458) . There is increased incidence of these tumors in childhood (ages 0 - 19) with 10.4% among 21,512 patients . BSG be- comes rare again in young adults with incidence of 2.6% (N = 27,899) . These tumors are typically diagnosed by magnetic resonance imaging (MRI) -. Tissue diagnosis was seldom obtained due to technical difficul- ties, but now is more common. Approximately 80% are DIPG, which carry the worst prognosis with less than 7% of patients surviving beyond 2 years . The other types include focal, exophytic, cervicomedullary and mid- brain tumors. Except for DIPG, the other subcategories of pediatric BSG, as well as BSG in adults, have better prognoses, and reach the overall survival rate at 5 years over 45%  -. Based on the most recent studies, pediatric DIPG forms a distinct group that is characterized by mutations in the histone H3.3 (H3F3A gene) . Antineoplastons are peptides and amino acid derivatives that inhibit the growth of neoplastic cells without growth inhibition of normal cells . Antineoplaston (ANP) A10 injection is a 4:1 mixture of phenylacetylglu- taminate (PG) and phenylacetylisoglutaminate (isoPG) sodium, and ANP AS2-1 injection is a 4:1 mixture of phenylacetate sodium (PN) and PG . The study of the affect of PG and PN on human U87 glioblastoma (GBM) cells indicated that PG and PN interrupt signal transduction in RAS/MAPK/ERK and PI3K/AKT/PTEN pathways, interfere with cell cycle, decrease metabolism and promote apoptosis in GBM cells. The effect on multiple cellular pathways and over 100 targets suggests that ANP is a promising candidate for clinical studies in malignant brain tumors .
Recently, Tawakol et al. have demonstrated that emotional stress such as depression can increase amygdalar activity in the brain, which could be examined as tremors in the amygdala by Functional Magnetic Resonance Imaging (fMRI) [12-14]. The activation of amygdala was associated with increased bone marrow activity and inflammation with increased hsCRP . A recent case study revealed that the incidence of depression was three-fold higher among ACS patients compared to the control group (Table 1). Depression is one of the most important mental disorders that have adverse effects on the brain as well as on the cardiovascular system. A large number of systematic review articles as well as meta-analyses of epidemiological studies reported the relation between depression and the risk of Coronary Artery Disease (CAD), which had some limitations [4-11]. In a meta-analysis comprising of 893,850 subjects (59,062 CAD cases), the period of follow-up ranged from minimum 2 years to 37 years. The Relative Risks (RRs) were significant 1.30 (95% CI: 1.22-1.40) for CAD and 1.30 (95% CI: 1.18-1.44) for myocardial infarction. Further analysis of subgroup by duration of follow up, the RR of CAD was 1.36 (95% CI: 1.24- 1.49) for less than 15 years follow-up, and 1.09 (95% CI: 0.96-1.23) for equal to or more than 15 years follow-up . These findings indicate that depression may have independent association with a significantly greater risk of CAD and myocardial infarction. A more recent meta-analysis assessed the association between depression and the risk of myocardial infarction and coronary death . Table 1: Frequency of depression and its characteristics in patients with acute coronary syndrome and controls.
If children with no outcome data at either 2 or 5 years of age were considered to have a disability or be dead, rather than without a disability and alive, or if the analyses were limited to only those who were assessed at 5 years of age, then few statistical con- clusions were altered. The minor exception was that antenatal steroid therapy was no longer associated significantly with survival free of disability on day 1 (OR: 1.5; 95% CI: 0.92, 2.5). In the 210 children in the preterm group who were assessed at both 2 and 5 years of age, the classification of major disability or not was the same at both 2 and 5 years in 190 (90.5%). The level of agreement between the 2- and 5-year assessments was highly significant ( ⫽ 0.691, t ⫽ 10.0, P ⬍⬍ .0001). There was no significant bias in the assessment of major disability at 2 years of age; dis- ability at 5 years of age was underestimated in 9 children and overestimated in 11 children at 2 years of age (McNemar test, P ⫽ .82, NS).
Conclusion: Comparative evaluation finds that Qol in cancer patients was measured by European Organization for Research and Treatment of Cancer (EORTC), Quality of Life-score 30 (QLQ-C30), Functional living index cancer (FLIC), Traditional Chinese medicine index (TCMI), karnofsky performance index (KPI), GLQ-8 and Spitzer’s uniscale. In all studies adverse events related to mistletoe extract treatment were local reaction at the injection site, chill and muscle pain, allergic skin reaction and fever. Survivalanalysis done by using Wilcoxon paired sample test and cox proportion hazard model and hazard ratio was reported to 0.36 to 1.32.
Comparisons of patient and tumor characteristics were performed using the χ 2 test or two-sample t-test. Survival curves were constructed using Kaplan–Meier method and tested by log-rank test. Multivariate adjusted hazard ratios (HRs) with 95 % confidence intervals (CIs) were calculated using the Cox proportional hazards model. The Mann– Whitney test was used to test gene expression differences. To analyze the combined results, we employed a two-step approach . At first, the individual participant data from each study were analyzed separately (i.e. to obtain the results of each cohort). Then, the results were syn- thesized in the second step using a suitable model for meta-analysis of aggregate data. The meta-analysis was conducted in adherence to the standards of quality . To pool the proportions, we used the command “meta- prop_one” in Stata. According to a previous study , the score methods are recommended for proportion interval estimates and in our study the Wilson score confidence intervals were computed. We also assessed the heterogeneity among cohorts by using Cochran χ 2 Q statistics and I 2 statistics. If P values <0.05 or I 2 > 25 % were obtained, we determined that there was a significant
Recurrentevents data analysis is quite common in bio- medicine, such as low back pain, sick leave from work, sporting injuries, hospital readmissions and episodes of infectious diseases such as malaria [1-7]. Literature review indicates that most statistical models applied to such data are often based on naive techniques. Such naive tech- niques are characterized by either ignoring the existence of recurrentevents, or ignoring the fact that the recurrentevents within subjects are correlated [1,2]. Even when taking into account the non-independence of recurrentevents within subjects, data analyses are mostly done with continuous risk interval models [2-5], which may not be relevant for health conditions with discontinuous risk . In the medical field, it is quite common to encounter recurrent health conditions with such discontinuous risk intervals, e.g. in cases with persistent treatment effect. Examples include infections, such as malaria, disability episodes, hospitalizations, and nursing home admissions [7-12]. When subjects have a disability episode, they are not at risk of the second episode of disability until they have recovered from the first episode. To obtain unbiased estimates of incidence rates, the person-time period when the subject is not at risk should be excluded from the risk set. When analysing recurrent time-to-event outcomes with discontinuous risk intervals the subject is not at risk of another event while a previous one is ongoing or if the subject is under treatment. Appropriate models for analys- ing recurrentevents data include marginal models or frailty or random coefficient analysis models [8-13], which take into account the non independence assumption of events within the subject.
All statistical analyses were carried out in R . Baseline characteristics among groups were assessed using Pearson’s chi-square test and one way ANOVA. Significance of non-parametric data was assessed by Mann-Whitney U and categorical data by χ 2 . Survival analyses were performed in R, using the Surv func- tion from the survival library  and the npsurv and survplot functions from the rms library . Graphs are plotted with 95% confidence intervals. Cox proportional hazards ratios are calculated using the coxph function from the survival library, with testing of assumptions in cox.zph. The Cox model included donor category, age category, dialysis vin- tage, peak PRA, gender, recurrence, and total ische- mic time. A time interaction variable was introduced for GN recurrence in the cox model for death cen- sored graft survival.
group. One hundred and nine patients received at least one dose of treatment. All enrolled patients were included in the efficacy and safety analysis. Two patients in the irinotecan plus S-1 group and one in the S-1 monotherapy group were lost to follow-up. The baseline characteristics of the 123 treated patients are listed in Table 1. The median age of the enrolled patients was 58.5 years (range 39.1–70.0 years). One hundred and twelve enrolled patients (91.1%, 112/123) had an ECOG performance status score of 0–1. All patients received platinum-based or taxane-based chemotherapy in the first-line setting. Forty-seven patients (38.2%, 47/123) had surgery and 64 (52.0%, 64/123) had received local radiation in previous treat- ments. Additionally, a large proportion of patients had at least one of the poor prognostic factors, including poorly differentiated tumors (46.3%, 57/123) and 3 or more metastatic sites (16.3%, 20/123). The baseline characteristics were generally well balanced between the two treatment groups.