Standard Survival Analysis Methods Research Articles

Factors associated with heart failure readmissions from skilled nursing facilities

Background: Despite guideline-driven pharmacological therapies and careful transitional care, the rates of preventable hospital re-admission of heart failure patients and associated costs remain unacceptably high in the SNF populations. Transfer to SNF is one strategy to limit hospitalizations. As such, 25% of patients are still symptomatic at time of discharge. Purpose: The objective of this study is to identify patient factors a€ecting re-admissions of HF patients residing in SNF within 30-days. Methods: A retrospective electronic chart review was completed on patients > 65 years with HF who were admitted into large medical center between 2012 and 2014. Descriptive statistics and univariate analyses using the chi-square test or Fisher’s exact test for categorical variables and the Mann-Whitney test for continuous data was used to compare patients readmitted within 30 days vs. those who were not readmitted within 30 days. Signi‚cant factors associated with readmission in the univariate analysis (p < 0.10) were included for a multivariate logistic regression model. The analysis of LOS was accomplished by applying standard methods of survival analysis, i.e., computing the Kaplan-Meier product limit curves, where the data were strati‚ed by readmission within 30 days (Yes vs. No). No data were considered ‘censored’. The groups were compared using the log-rank test. The median rates for each group were obtained from the Kaplan-Meier/Product-Limit Estimates and their corresponding 95% con‚dence intervals were computed, using Greenwood’s formula to calculate the standard error. Unless otherwise speci‚ed, a result was considered statistically signi‚cant at the p < 0.05 level of signi‚cance. Results: Fifteen variables: creatinine, weight di€erence, CKD, Angina, Arrhythmia, VHD, Tobacco, ADL, independent in bathing, independent in the toilet, S3 Heart sounds present, HJR, AF, Nitrates, and Hydralazine, were identi‚ed for the multivariate logistic regression as potential risk factors associated with “readmission within 30 days”. Based on 23 readmissions within 30 days, our ‚nal model included only 2 predictor variables. Creatinine and ADLs were included in the ‚nal model as this subset of predictors was found to be the best for prediction of “readmission within 30 days”. Creatinine (p < 0.0087) and ADLs (p < 0.0077) were both signi‚cantly associated with readmission within 30 days in the ‚nal logistic regression model. Every 1-unit increase in creatinine was associated with an 87% increase in the odds of being readmitted within 30 days (OR = 1.87). Those patients who require assistance with ADLs were over 9 times more likely to be readmitted within 30 days (OR = 9.25) as compared to patients who are independent.

Assessing treatment effects with surrogate survival outcomes using an internal validation subsample.

In studies with surrogate outcomes available for all subjects and true outcomes available for only a subsample, survival analysis methods are needed that incorporate both endpoints in order to assess treatment effects. We develop a semiparametric estimated likelihood method for the proportional hazards model with discrete time data and a binary covariate of interest. Our proposed method allows for real-time validation of surrogate outcomes and flexible censoring mechanisms. Our proposed estimator is consistent and asymptotically normal. Through numerical studies, we showed that our proposed method for estimating a covariate effect is unbiased compared to the naïve estimator that uses only surrogate endpoints and is more efficient with moderate missingness compared to the complete-case estimator that uses only true endpoints. We further demonstrated the advantages of our proposed method in comparison with existing approaches when there is real-time validation. We also illustrated the use of our proposed method by estimating the effect of gender on time to detection of Alzheimer's disease using data from the Alzheimer's Disease Neuroimaging Initiative. The proposed method is able to account for the uncertainty of surrogate outcomes using a validation subsample of true outcomes in estimating a binary covariate effect. The proposed estimator can outperform standard semiparametric survival analysis methods and can therefore save on costs of a trial or improve power in detecting treatment effects.

Analysis of the Bypass Angioplasty Revascularization Investigation Trial Using a Multistate Model of Clinical Outcomes

Current cardiovascular randomized trials typically use composite outcomes. We hypothesized that the Bypass Angioplasty Revascularization Investigation (BARI) outcomes and conclusions would differ using a multistate model relative to the intervention for the composite outcome of death (D) and nonfatal Q-wave myocardial infarction (MI). We used a multistate model which uses transition paths to simultaneously assess multiple end points. Using the 10-year follow-up BARI data, we post hoc analyzed outcomes according to 3 transition paths: (1) from intervention to MI; (2) from intervention to death; and (3) from MI to death. Of 1,829 patients randomized to the intervention of percutaneous transluminal coronary angioplasty or coronary artery bypass grafting (CABG), 700 (38%) experienced the composite event D/MI which included 230 (13%) nonfatal MI and 470 (26%) death without antecedent nonfatal MI, whereas 79 of 230 (34%) experienced death after nonfatal MI. Outcomes of the 3 individual transition paths were analyzed by a multistate model. In contrast to standard survival analyses, after adjustment for baseline clinical covariates, outcomes after percutaneous transluminal coronary angioplasty or CABG were not significantly different for intervention to MI (p = 0.33) or intervention to death (p = 0.23), but MI to death favored CABG (p = 0.02). Deconstruction of the BARI data using a multistate model identifies a significant difference in individual transition-stage outcomes and therefore trial conclusions in contrast to the standard methods of survival analysis. These observations suggest multistate models should be considered in the design and analysis of randomized cardiovascular trials which use composite outcomes.

754O - Impact of Node Status and Radiotherapy on Failure-Free Survival in Patients with Newly Diagnosed Non-Metastatic Prostate Cancer: Data from >690 Patients in the Control Arm of the Stampede Trial

Aim The natural history of patients (pts) with newly diagnosed high-risk non-metastatic (M0) prostate cancer receiving androgen deprivation therapy (ADT) either alone or with standard of care radiotherapy (RT) at 6 to 9 months is not well documented. Further, no RCT has tested RT in N + M0 patients; none are planned. The STAMPEDE RCT (NCT00268476; MRC PR08; CRUK/06/019) includes such pts & allows exploratory multivariate analysis of radical RT's impact. We report data from trial pts. We hypothesised that planning RT in N + M0 pts improves survival outcomes. Methods Newly diagnosed pts with confirmed M0 disease in the control arm (standard of care = ADT planned for >2yr), diagnosed <6 months pre-randomisation & on ADT for 0-12 weeks already, were identified from trial records in Jan-2014. RT is encouraged in this group, but only mandated for N0M0 pts since Nov-2011. We report failure-free survival (FFS), driven by PSA failure, & overall survival (OS); split by nodal involvement & reported RT status. Standard survival analysis methods were used, adjusting for age & PSA. Results 5272 men were recruited Oct-2005 to Jan-2014, including a cohort of 694 M0 pts with newly diagnosed disease allocated to the control arm: median age 66yr (IQR 61-71), median time from diagnosis to randomisation 2.6m (IQR 2.0-3.3) & median PSA 42ng/ml (IQR 17-88) at diagnosis. By Jan-2014, there were 34 deaths; 25 from prostate cancer. Median follow-up is short, but 2yr OS is 95% (95%CI 92, 97) & 2yr FFS is 79% (95%CI 75, 83). Median FFS is 63 months; 79% (94/119) report PSA failure-only as first FFS event. Time to FFS is worse in N+ pts (HR 1.87, 95%CI 1.29-2.72). Baseline characteristics were comparable by planned RT status, but N0M0 pts planned for RT had lower PSA at diagnosis & more WHO PS = 0. The Table shows that FFS outcomes clearly favour the planned use of RT.

MP53-20 THE PREDICTIVE VALUE OF 2-YEAR POST-TREATMENT BIOPSY AFTER PROSTATE CANCER CONFORMAL RADIOTHERAPY FOR FUTURE OVERALL SURVIVAL AND BIOCHEMICAL FAILURE. THE RESULTS OF THE UK MRC RT01 TRIAL.

You have accessJournal of UrologyProstate Cancer: Detection & Screening II1 Apr 2014MP53-20 THE PREDICTIVE VALUE OF 2-YEAR POST-TREATMENT BIOPSY AFTER PROSTATE CANCER CONFORMAL RADIOTHERAPY FOR FUTURE OVERALL SURVIVAL AND BIOCHEMICAL FAILURE. THE RESULTS OF THE UK MRC RT01 TRIAL. Antoine Kass-iliyya, Gordana Jovic, Matthew Sydes, Cyril Fisher, Claire Murphy, David Nicol, and David Dearnaley Antoine Kass-iliyyaAntoine Kass-iliyya More articles by this author , Gordana JovicGordana Jovic More articles by this author , Matthew SydesMatthew Sydes More articles by this author , Cyril FisherCyril Fisher More articles by this author , Claire MurphyClaire Murphy More articles by this author , David NicolDavid Nicol More articles by this author , and David DearnaleyDavid Dearnaley More articles by this author View All Author Informationhttps://doi.org/10.1016/j.juro.2014.02.1651AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookTwitterLinked InEmail Introduction and Objectives There is still an uncertainty regarding the significance of prostate biopsy status performed at 2 years after conformal radiotherapy (CFRT) used for treatment of localized prostate cancer. The objective of this analysis is to determine whether prostate biopsy status at around 2 years is predictive of overall survival (OS) and long-term biochemical progression free survival (bPFS). Methods Between January 1998 and December 2001, 843 consenting men with localized prostate cancer were randomized in to the MRC RT01 trial to receive either Control-64Gy or Escalated-74Gy of CFRT. Patients were stratified according to the risk of seminal vesicles (SV) involvement. As part of the trial’s protocol, post-CFRT prostate biopsies were requested at two years from the start of RT. Exclusion criteria for this analysis included biopsies performed before 18 months or after 36 months of starting radiotherapy and any evidence of biochemical failure or progression before or on the date of the biopsy. Analyses were performed using standard survival analysis methods with a landmark approach; hazard ratios (HR) are presented. Results 223 patient biopsies were taken and reviewed by a central pathologist. 46 of these patients met the exclusion criteria, leaving 177 patients for the main analysis. Baseline and patients characteristics were comparable in both groups. Median follow up was 7.8 years. In total, 64 bPFS events occurred: 46/145 in patients with negative biopsies, 6/18 in suspicious and 12/14 in positive. p: <0.001. HR: 4.69 (95%CI 2.44-9.01). Figure (1) Four deaths secondary to prostate cancer occurred: 1 (0.7%) in patient with negative biopsy and 3 (21.4%) in patients with positive biopsies. Twenty-seven deaths from any cause occurred: 21/145 (14%) in patients with negative biopsies, 2/18 (11%) in suspicious and 4/14 (29%) in positive. p: 0.410. HR: 1.58 (95%CI 0.53-4.71). Conclusions The risk of biochemical failure in patients with positive biopsies at 2 years is almost five times higher than in patients with negative or suspicious biopsies and the risk of death from any cause is about 1.6 times higher. Post-treatment prostate biopsies are a strong predictor of future biochemical failure. © 2014FiguresReferencesRelatedDetails Volume 191Issue 4SApril 2014Page: e596-e597 Advertisement Copyright & Permissions© 2014MetricsAuthor Information Antoine Kass-iliyya More articles by this author Gordana Jovic More articles by this author Matthew Sydes More articles by this author Cyril Fisher More articles by this author Claire Murphy More articles by this author David Nicol More articles by this author David Dearnaley More articles by this author Expand All Advertisement Advertisement PDF downloadLoading ...

Clinic-Based Outpatient Palliative Care Before Hospice Is Associated With Longer Hospice Length of Service

ContextOutpatient nonhospice palliative care has been shown to provide many benefits to patients facing advanced illness, but such services remain uncommon in the U.S. Little is known about the association between clinic-based outpatient palliative care consultation and the timing of hospice enrollment. ObjectivesTo determine whether there are differences in hospice length of service (LOS) between patients who were seen vs. patients who were not seen in an outpatient palliative care clinic before enrollment in hospice. MethodsUsing a retrospective study of medical records, a “prior palliative care clinic” group was formed of those hospice patients who had had a nonhospice clinic-based outpatient palliative care consult before hospice admission (n = 354). For those patients, “control” hospice patients without prior clinic-based palliative care were chosen who were matched by age, gender, median income of their zip code, and diagnostic group. Both groups were restricted to patients who died while enrolled in hospice. LOS for these two groups was compared using standard statistical methods of survival analysis. ResultsPrior palliative care clinic patients had a median LOS of 24 days, whereas control patients had a median LOS of 15 days (95% CI for difference between the medians 5–13 days). The difference between the LOS distribution curves was statistically significant by the log-rank test (P < 0.001). ConclusionHospice patients who had clinic-based outpatient palliative consults before hospice enrollment tended, on average, to have a longer LOS in hospice than patients who did not.

Using Cure Models for Analyzing the Influence of Pathogens on Salmon Survival

AbstractParasites and pathogens influence the size and stability of wildlife populations, yet many population models ignore the population‐level effects of pathogens. Standard survival analysis methods (e.g., accelerated failure time models) are used to assess how survival rates are influenced by disease. However, they assume that each individual is equally susceptible and will eventually experience the event of interest; this assumption is not typically satisfied with regard to pathogens of wildlife populations. In contrast, mixture cure models, which comprise logistic regression and survival analysis components, allow for different covariates to be entered into each part of the model and provide better predictions of survival when a fraction of the population is expected to survive a disease outbreak. We fitted mixture cure models to the host–pathogen dynamics of Chinook Salmon Oncorhynchus tshawytscha and Coho Salmon O. kisutch and the myxozoan parasite Ceratomyxa shasta. Total parasite concentration, water temperature, and discharge were used as covariates to predict the observed parasite‐induced mortality in juvenile salmonids collected as part of a long‐term monitoring program in the Klamath River, California. The mixture cure models predicted the observed total mortality well, but some of the variability in observed mortality rates was not captured by the models. Parasite concentration and water temperature were positively associated with total mortality and the mortality rate of both Chinook Salmon and Coho Salmon. Discharge was positively associated with total mortality for both species but only affected the mortality rate for Coho Salmon. The mixture cure models provide insights into how daily survival rates change over time in Chinook Salmon and Coho Salmon after they become infected with C. shasta.

AB0579 Abatacept in rheumatoid arthritis: Clinical and imaging considerations

ObjectivesTo investigate the efficacy and safety of abatacept in patients with rheumatoid arthritis (RA) and to describe abatacept survival rates and reasons for discontinuation. Additionally, to investigate the effectiveness of...

Long-term outcome of cats and dogs with acute kidney injury treated with intermittent hemodialysis: 135 cases (1997–2010)

To determine the long-term outcome for small animal patients with acute kidney injury (AKI) treated with intermittent hemodialysis (IHD). Retrospective case series. 42 cats and 93 dogs treated with IHD for AKI. Medical records of cats and dogs treated with IHD for AKI from January 1997 to October 2010 were reviewed. Standard methods of survival analysis with Kaplan-Meier product limit curves were used. The log-rank, Mann-Whitney, and Kruskal-Wallis tests were used to determine whether outcome, number of IHD treatments, or duration of hospitalization was different when dogs and cats were classified according to specific variables. The overall survival rate at the time of hospital discharge was 50% (21/42) for cats and 53% (49/93) for dogs. The overall survival rate 30 days after hospital discharge was 48% (20/42) for cats and 42% (39/93) for dogs. The overall survival rate 365 days after hospital discharge was 38% (16/42) for cats and 33% (31/93) for dogs. For all-cause mortality, the median survival time was 7 days (95% confidence interval, 0 to 835 days) for cats and 9 days (95% confidence interval, 0 to 55 days) for dogs. Cats and dogs with AKI treated with IHD have survival rates similar to those of human patients. Although there was a high mortality rate prior to hospital discharge, those patients that survived to discharge had a high probability of long-term survival.

Survival from high-grade localised extremity osteosarcoma: combined results and prognostic factors from three European Osteosarcoma Intergroup randomised controlled trials

Neoadjuvant chemotherapy improves outcome in osteosarcoma. Determination of optimum regimens for survival, toxicity and prognostic factors requires randomised controlled trials to be conducted. Between 1983 and 2002, the European Osteosarcoma Intergroup recruited 1067 patients with localised extremity osteosarcoma to three randomised controlled trials. Standard treatment in each was doxorubicin 75 mg/m(2) and cisplatin 100 mg/m(2). Comparators were addition of methotrexate (BO02/80831), a multidrug regimen (BO03/80861) and a dose-intense schedule (BO06/80931). Standard survival analysis methods were used to identify prognostic factors, temporal and other influences on outcome. Five- and 10-year survival were 56% (95% confidence interval 53% to 59%) and 52%, respectively (49% to 55%), with no difference between trials or treatment arms. Median follow-up was 9.4 years. Age range was 3-40 years (median 15). Limb salvage was achieved in 69%. Five hundred and thirty-three patients received the standard arm, 79% completing treatment. Good histological response to preoperative chemotherapy, distal tumour location (all sites other than proximal humerus/femur) and female gender were associated with improved survival. Localised osteosarcoma will be cured in 50% of patients with cisplatin and doxorubicin. Large randomised trials can be conducted in this rare cancer. Failure to improve survival over 20 years argues for concerted collaborative international efforts to identify and rapidly test new treatments.

Survival after recurrent osteosarcoma: Data from 3 European Osteosarcoma Intergroup (EOI) randomized controlled trials

BackgroundRecurrence after osteosarcoma usually leads to death; thus prognostic factors for survival are of great importance. MethodsBetween 1983 and 2002, the European Osteosarcoma Intergroup accrued 1067 patients to 3 randomized controlled trials of pre- and post-operative chemotherapy for patients with resectable non-metastatic high-grade osteosarcoma of the extremity. Control treatment in all trials was doxorubicin 75mg/m2 and cisplatin 100mg/m2. The comparators were additional high-dose methotrexate (BO02), T10-based multi-drug regimen (BO03) and G-CSF intensified-DC (BO06). Post-recurrence survival (PRS) was investigated on combined data with standard survival analysis methods. ResultsMedian recurrence-free survival was 31months; 8 recurrences were reported more than 5years after the diagnosis. In 564 patients with a recurrence (median 13months post-randomisation), there was no difference in post-relapse survival between treatment arms. Patients whose disease recurred within 2years after randomization had a worse prognosis than those recurring after 2years. Patients with good initial histological response to pre-operative chemotherapy had a better overall survival after recurrence than poor responders. Local relapse was more often reported after limb-saving procedures (2 versus 8%; amputation versus limb-saving), independent of the primary tumour site. Site of first recurrence (local 20%, lung 62%, “other” 19%) affected survival, as patients recurring with non-lung distant metastases only or any combination of local relapse, lung metastases and non-lung metastases (=group “other”) had significantly worse overall survival (local 39%, lung 19%, “other” 9% at 5years). ConclusionsThese data describing a large series of patients with recurrent extremity osteosarcoma confirm the relationship between early recurrence and poor survival. There was better PRS in patients after good histological response to pre-operative chemotherapy, or with local-only recurrence.

Statistical Analysis of Competing Risks: Overall Survival in a Group of Chronic Myeloid Leukemia Patients

Background: Survival analysis is a collection of statistical methods for inference on time-to-event data. If several causes of failure occur and the occurrence of one event precludes the occurrence of the other events, the situation is known as competing risks. Since the competing risks violate the fundamental assumption of independent censoring, specific methods for inference are needed. Objectives: The aim of this paper is to recall the competing risks model and statistical methods for nonparametric analysis, and to illustrate the competing risks methods on a real data set of 118 Chronic Myeloid Leukemia (CML) patients from the Clinic of Haemato-oncology of the University Hospital in Olomouc. Methods: The overall survival probability and risk factors of two types of failure (death due to CML and death from other causes) are assessed. Predicted probabilities of the two types of failure with stratification based on the risk factors (Sokal score, haematological response to treatment) are shown. Results: Outcomes of the specific methods designed for the competing risks analysis are compared with the outcomes of the standard survival analysis methods. The effect of the Sokal score classification is found ambiguous. While the score should identify high- and low-risk CML patients, it seems to be predictive only for the failure due to other causes than CML. Conclusions: The importance of careful censoring and the need of using proper methods of analyses of competing risks data is shown. The use of the Sokal score for classification of the CML patients should be considered more thoroughly.

Expected long-term survival of patients diagnosed with acute myeloblastic leukemia during 2006–2010

BackgroundTreatment of acute myeloblastic leukemia (AML) has evolved over the past several decades. Therefore, currently available estimates of long-term survival, which are based on survival for patients treated with potentially now obsolete protocols, may not pertain to patients currently diagnosed. MethodsUsing data from the 1973–2005 database of the Surveillance, Epidemiology, and End Results Program, we empirically validated a novel model-based method to project 5- and 10-year relative survival of AML patients and we applied the method to project relative survival of AML patients in the United States diagnosed during 2006–2010. ResultsEmpirical evaluation indicated that the modeling approach provides more accurate estimates of currently diagnosed patients than standard methods of survival analysis, such as cohort analysis or period analysis, in the majority of cases. Projected figures for 2006–2010 show 5- and 10-year relative survival estimates of 21.4% and 18.7% for all ages combined, 62.2% and 57.4% for ages 25–34, and 60.6% and 58.1% for ages 35–44. These estimates are substantially higher than the most up-to-date estimates obtained by standard survival analysis. ConclusionPatients diagnosed with AML during 2006–2010 at younger ages have much higher long-term survival expectations than indicated by previously available survival statistics.

Occurrence of Lymphedema Continues To Increase Twelve to Thirty Months after Breast Cancer Treatment.

Abstract Introduction:Breast cancer (BC) survivors are at lifetime risk for developing lymphedema (LE), the accumulation of protein-rich fluid in the interstitial spaces of the affected body part due to a blockage or malfunction in the lymph system. This is different than swelling which may occur immediately after surgery and may be present at the post-op visit, and the physical and psychological aspects of the condition greatly impact the daily lives of LE patients (Geller et al., 2003, Hull, 1998). Of those affected by breast cancer, up to 40% will develop LE, potentially comprising 1 to 5 million survivors. However, all breast cancer survivors are considered at risk for the condition (American Cancer Society, 2007).While numerous studies have reported LE incidence during the first 12 months following breast cancer treatment, little is known regarding long-term LE diagnosis. Very few studies have examined LE incidence past 1 year post-treatment, and many that have are retrospective or cross-sectional, not prospective, in nature. In fact, in one analysis of existing literature the study with the shortest follow-up (12 months) reported the lowest LE incidence (Ball et al., 1992) and the study with the longest follow-up (11 years) reported the highest incidence (Schunemann and Willich, 1997).Methods:Participants were enrolled following diagnosis of BC but before treatment and followed every 3 months for 12 months, then every 6 months until 2.5 years post-surgery. Limb volume changes (LVC) were measured using: (a) circumferences via tape measure at 4 cm intervals, (b) infra-red laser perometry, and (c) symptom experience (SS) via interview. Four diagnostic criteria for LE were used: (i) 2 cm circumferential change; (ii) 200 mL perometry LVC; (iii) 10% perometry LVC; and (iv) report of limb heaviness/swelling, 'now' or 'in the past year'. Standard survival analysis methods were applied to identify when the LE criteria were met.Results:At 30 months post-treatment, LE incidence using the four criteria ranged from 41%- 91% and had continued to increase over time, with 2 cm being the highest estimation method and SS the lowest.Conclusions:These preliminary findings provide additional evidence that BC survivors continue to be at risk for developing LE beyond the first year following treatment. LE identification, regardless of the criteria used, continued to increase past the initial 12 months post-treatment. From month 12 to month 30, LE identification increased by an additional 10-27%, depending on the criteria used. While identification of LE via symptom experience in the initial 12 months occurred in 31% of participants, only an additional 10% met the criteria at 30 months, by far the lowest incidence rate among all criteria. Overall, this analysis finds 2 cm criteria as the most liberal definition of LE (91%), while self-report of heaviness and swelling, along with 10% LVC, represent the most conservative definitions (41% and 45%, respectively). Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 2070.

Lymphedema in Breast Cancer Survivors: Incidence, Degree, Time Course, Treatment, and Symptoms

To examine the incidence, degree, time course, treatment, and symptoms of lymphedema in breast cancer survivors. We conducted a 5-year, population-based prospective study of 631 randomly selected Philadelphia and Delaware County, Pennsylvania female residents with incident breast cancer who were diagnosed from 1999 to 2001. Using a questionnaire previously validated against physical therapists' measurement-based clinical criteria, we assigned a score indicating the degree of lymphedema (none, mild, or moderate/severe) to each month of follow-up based on the respondent's perceived differences in hand/arm size. Standard survival analysis methods permitted maximum use of follow-up. Five-year cumulative incidence of lymphedema was 42 (42%) per 100 women. Among the 238 affected women, lymphedema first occurred within 2 years of diagnosis in 80% and within 3 years in 89%. Among 433 women observed for 3 years, 23% reported no more than mild lymphedema, 12% reported moderate/severe lymphedema, and 2% reported chronically moderate/severe lymphedema. Women with mild lymphedema were more than three times more likely to develop moderate/severe lymphedema than women with no lymphedema. Thirty-seven percent of women with mild lymphedema and 68% with moderate/severe lymphedema received treatment. Increasing proportions of women with increasing degree of lymphedema reported symptoms (eg, jewelry too tight, tired/thick/heavy arm). Symptoms present before the first occurrence of lymphedema were associated with a higher probability of later lymphedema (eg, hazard ratio for jewelry too tight = 7.37; 95% CI, 4.26 to 12.76). Lymphedema after breast cancer is common but mostly mild. Subtle differences in self-reported hand/arm size and symptoms can be early signs of progressing lymphedema.

Branching Processes as Models of Progenitor Cell Populations and Estimation of the Offspring Distributions

We consider two new models of reducible age-dependent branching processes with emigration in conjunction with estimation problems arising in cell biology. Methods of statistical inference are developed using the relevant embedded discrete branching structure. Based on observations of the branching process with emigration, estimators of the offspring probabilities are proposed for the hidden unobservable process without emigration, which is of prime interest to investigators. The problem under consideration is motivated by experimental data generated by time-lapse video recording of cultured cells, which provides abundant information on their individual evolutions and thus on the basic parameters of their life cycle in tissue culture. Some parameters, such as the mean and variance of the mitotic cycle time, can be estimated nonparametrically without resorting to any mathematical model of cell population kinetics. For other parameters, such as the offspring distribution, a model-based inference is needed. Age-dependent branching processes have proven to be useful models for that purpose. A special feature of the data generated by time-lapse experiments is the presence of censoring effects due to the migration of cells out of the field of observation. For the time-to-event observations, such as the mitotic cycle time, the effects of data censoring can be accounted for by standard methods of survival analysis. No methods are available to accommodate such effects in the statistical inference on the offspring distribution. Within the framework of branching processes, the loss of cells to follow-up can be modeled as a process of emigration. Incorporating the emigration process into a pertinent branching model of cell evolution provides the basis for the proposed estimation techniques. Statistical inference on the offspring distribution is illustrated with an application to the development of oligodendrocytes in cell culture.

Expected Long-Term Survival of Patients Diagnosed with Multiple Myeloma in 2006–2010

Background: New therapeutic options have led to substantial increases in survival expectations of patients with multiple myeloma (MM) in recent years. In the past, the impact of these innovations on long-term survival of MM patients on the population level has been disclosed only with substantial delay, partly due to delayed availability of data from population-based cancer registries.Methods: Using data from the 1973–2005 database of the Surveillance, Epidemiology, and End Results (SEER) Program, we employed a novel model-based projection method to project 5-and 10-year relative survival expectations of MM patients in the United States diagnosed in 2006 to 2010, after conducting an empirical evaluation of the method using data from patients diagnosed between 1996 and 2000 (the last cohort for which 5-year follow up is available through SEER) and earlier years.Results: Preliminary empirical evaluation of the method using historical data indicates good performance in projection of age specific and overall 5- and 10-year relative survival. Model based projection of survival produced survival estimates that matched observed survival estimates more closely than cohort analysis or period analysis, the standard methods of survival analysis, in 21 out of 24 cases. The average difference from observed was −1.27 percentage units for model based projection, −3.71 percentage units for period analysis, and −4.23 percentage units for cohort analysis. Five-year relative survival projections for patients diagnosed in 2006–2010 at ages 15–49 and 50–59 are 61.4% and 52.0% (see table), respectively, and exceed the most up-to-date estimates obtained from traditional cohort analysis by 9.5 and 6.9 percent units, respectively. Ten-year relative survival projections for patients in these age groups are 47.1% and 30.4%, exceeding the most up-to-date estimates from traditional cohort analysis by 13.5 and 9.6 percent units, respectively. By contrast, survival projections remain much lower and hardly exceed estimates from traditional survival analysis for older patients.Conclusions: The use of model based projection results in more up-to-date estimates of survival than standard methods of survival analysis. Patients diagnosed with multiple myeloma in 2006–2010 have much higher long-term survival expectations than suggested by previously available survival statistics.Table: Most up-to-date estimates of 5-year and 10-year relative survival of patients diagnosed with multiple myeloma obtained by different methods.AgeCohort estimates*Period estimates for 2001–2005 periodProjections for patients diagnosed in 2006–2010PESEPESEPESE5-year relative survival15+32.10.734.20.736.11.015–4951.92.356.42.261.43.150–5945.11.748.01.652.02.360–6933.81.436.51.436.72.070–7927.41.226.71.227.91.780+13.21.414.11.416.41.910-year relative survival15+14.10.616.60.618.00.815–4933.62.240.32.447.13.250–5920.81.627.01.730.42.360–6912.41.015.01.215.41.470–798.81.08.90.99.91.180+6.41.65.91.46.61.4* The most up-to-date cohort estimates of 5-year survival pertain to patients diagnosed in 1996–2000, and the most up-to-date cohort estimates of 10-year survival pertain to patients diagnosed in 1991–1995.

Including Injured Workers Without Compensated Time Loss in Cox Regression Models: Analyzing Time Loss Using All Available Data

Cox proportional hazards regression is commonly used to analyze time loss duration, but statistical packages conventionally exclude cases with no recorded follow-up time. For this and other substantive reasons, many researchers limit time loss analyses to the subset of workers who received time loss compensation. This can exclude both injured workers who missed no work days and those missing up to a week of work. For some research questions, excluding cases where injury is reported but no time loss is recorded may result in significant ascertainment bias. We present a novel technique based on standard survival analysis methods to allow for the inclusion of all cases when appropriate. A simple technique to allow standard statistical software to include both medical-only and time loss claims in Cox regression is illustrated by example and compared with a two-part model using a time-varying step function to allow regression effects to change over time. We showed that a pooled analysis is obtained by simply adding a small constant to the time loss duration variable. This technique produced appropriate estimates while accounting for censoring when a suitable method was used for tied event times. Using a formal statistical framework, the combined model was justified as a special case of the more standard two-part model approach. When it is desirable to have a single pooled outcome estimate for injured workers with both medical-only and time loss claims, all claims can be combined into one statistical model. This may have particular utility for research questions where the risk factor or intervention of interest would be expected to affect time loss duration beginning upstream of claim filing or statutory compensation waiting periods. This novel alternative modeling strategy expands the tool kit available for analyzing time loss data.

Survival after recurrent osteosarcoma: Data from three European Osteosarcoma Intergroup (EOI) randomized controlled

10505 Background: Between 1983 and 2002, the EOI accrued to three randomized controlled trials of pre- and postoperative chemotherapy for patients (pts) with resectable nonmetastatic high-grade osteosarcoma of the extremity: MRC-BO02, BO03 and BO06 (also known as EORTC-80831, 80861 and 80931) with n=179, 391 and 497 pts randomised, respectively. Control treatment in all trials was doxorubicin 75 mg/m2 + cisplatin 100 mg/m2 (DC). The comparators were additional high-dose methotrexate (BO02), T10-based multi- drug regimen (BO03) and G-CSF intensified-DC (BO06). No statistically significant differences between treatments were reported in the original trial reports. Methods: Survival after recurrence was investigated with standard survival analysis methods, including all 1,067 pts from the 3 trials. Results: Median recurrence-free survival was 31 months (m); only 4 recurrences were reported later than 5 years. In 564 pts with recurrence (median 13m post-randomisation), there was no difference in 5-year survival between first-line treatment arms (13–19%). Pts with good histologic response to preoperative chemotherapy had better overall survival after recurrence than poor responders (p=0.015). Pts whose disease recurred within 2 years after randomization had worse prognosis than those recurring after 2 years (14 vs 35% at 5 years; p<0.001). Local relapse was more often reported after limb saving procedures (5 vs 25%), independent of the primary tumor site. Site of first recurrence (in total: local 20%, lung 62%, other 18%) affected survival as pts recurring with metastases other than lung had significantly worse overall survival (local 27%, lung 18%, other 10% at 5 years, p=0.001). Conclusions: These data describe the largest series of pts with recurrent osteosarcoma. They confirm the relationship between early recurrence and poor survival. New findings include worse survival in pts with a recurrence after poor histologic response to preoperative chemotherapy. No significant financial relationships to disclose.

Influence of Psychiatric Comorbidity on Recovery and Recurrence in Generalized Anxiety Disorder, Social Phobia, and Panic Disorder: A 12-Year Prospective Study

Objective:The authors sought to observe the long-term clinical course of anxiety disorders over 12 years and to examine the influence of comorbid psychiatric disorders on recovery from or recurrence of panic disorder, generalized anxiety disorder, and social phobia. Method: Data were drawn from the Harvard/Brown Anxiety Disorders Research Program, a prospective, naturalistic, longitudinal, multicenter study of adults with a current or past history of anxiety disorders. Probabilities of recovery and recurrence were calculated by using standard survival analysis methods. Proportional hazards regression analyses with timevarying covariates were conducted to determine risk ratios for possible comorbid psychiatric predictors of recovery and recurrence. Results: Survival analyses revealed an overall chronic course for the majority of the anxiety disorders. Social phobia had the smallest probability of recovery after 12 years of follow-up. Moreover, patients who had prospectively observed recovery from their int...