A latent variable approach to jointly modeling longitudinal and cumulative event data using a weighted two-stage method.

Ecological momentary assessment (EMA), a data collection method commonly employed in mHealth studies, allows for repeated real-time sampling of individuals' psychological, behavioral, and contextual states. Due to the frequent measurements, data collected using EMA are useful for understanding both the temporal dynamics in individuals' states and how these states relate to adverse health events. Motivated by data from a smoking cessation study, we propose a joint model for analyzing longitudinal EMA data to determine whether certain latent psychological states are associated with repeated cigarette use. Our method consists of a longitudinal submodel-a dynamic factor model-that models changes in the time-varying latent states and a cumulative risk submodel-a Poisson regression model-that connects the latent states with the total number of events. In the motivating data, both the predictors-the underlying psychological states-and the event outcome-the number of cigarettes smoked-are partially unobservable; we account for this incomplete information in our proposed model and estimation method. We take a two-stage approach to estimation that leverages existing software and uses importance sampling-based weights to reduce potential bias. We demonstrate that these weights are effective at reducing bias in the cumulative risk submodel parameters via simulation. We apply our method to a subset of data from a smoking cessation study to assess the association between psychological state and cigarette smoking. The analysis shows that above-average intensities of negative mood are associated with increased cigarette use.

Predictors of Acute and Late Toxicity in Patients Receiving Chemoradiation for Unresectable Pancreatic Cancer.

Purpose: Patients with pancreatic cancer undergoing chemoradiation therapy may experience acute and chronic side effects. We conducted an exploratory analysis of patients with locally advanced pancreatic cancer (LAPC) undergoing definitive chemoradiation to identify factors influencing the occurrence of gastrointestinal (GI) bleeding, short-term radiation side effects, patterns of failure, and survival. Methods and Materials: Under an institutional review board-approved protocol, we retrospectively studied patients with LAPC treated with chemoradiation. Statistical models were used to test associations between clinical characteristics and outcomes, including upper GI bleeding, radiation treatment breaks, and weight loss during therapy. Results: Between 1999 and 2012, 211 patients were treated with radiation for pancreatic cancer. All patients received concurrent chemotherapy with either gemcitabine (174) or 5-fluorouracil (27), and 67 received intensity modulated radiation therapy (IMRT). Overall, 18 patients experienced an upper GI bleed related to treatment, with 70% of bleeds occurring in the stomach or duodenum, and among those patients, 11 (61%) patients had a pancreatic head tumor and 17 (94%) patients had a metallic biliary stent. IMRT was associated with decreased risk of postradiation nausea (odds ratio, 0.27 [0.11, 0.67], P = .006) compared with 3-dimensional conformal radiation. Regarding long-term toxicities, patients with a metallic biliary stent at the time of radiation therapy were at a significantly higher risk of developing upper GI bleeding (unadjusted hazard ratio [HR], 15.41 [2.02, 117.42], P = .008), even after controlling for radiation treatment modality and prescribed radiation dose (adjusted HR, 17.38 [2.26, 133.58], P = .006). Furthermore, biliary stent placement was associated with a higher risk of death (HR, 1.99 [1.41, 2.83], P < .001) after adjusting for demographic, treatment-related, and patient-related variables. Conclusions: Metallic biliary stents may be associated with an increased risk of upper GI bleeding and mortality. Furthermore, IMRT was associated with less nausea and short-term toxicity compared with 3-dimensional conformal therapy.

Comparing Individualized Survival Predictions From Random Survival Forests and Multistate Models in the Presence of Missing Data: A Case Study of Patients With Oropharyngeal Cancer.

Background: In recent years, interest in prognostic calculators for predicting patient health outcomes has grown with the popularity of personalized medicine. These calculators, which can inform treatment decisions, employ many different methods, each of which has advantages and disadvantages. Methods: We present a comparison of a multistate model (MSM) and a random survival forest (RSF) through a case study of prognostic predictions for patients with oropharyngeal squamous cell carcinoma. The MSM is highly structured and takes into account some aspects of the clinical context and knowledge about oropharyngeal cancer, while the RSF can be thought of as a black-box non-parametric approach. Key in this comparison are the high rate of missing values within these data and the different approaches used by the MSM and RSF to handle missingness. Results: We compare the accuracy (discrimination and calibration) of survival probabilities predicted by both approaches and use simulation studies to better understand how predictive accuracy is influenced by the approach to (1) handling missing data and (2) modeling structural/disease progression information present in the data. We conclude that both approaches have similar predictive accuracy, with a slight advantage going to the MSM. Conclusions: Although the MSM shows slightly better predictive ability than the RSF, consideration of other differences are key when selecting the best approach for addressing a specific research question. These key differences include the methods' ability to incorporate domain knowledge, and their ability to handle missing data as well as their interpretability, and ease of implementation. Ultimately, selecting the statistical method that has the most potential to aid in clinical decisions requires thoughtful consideration of the specific goals.

Integrated Survival Estimates for Cancer Treatment Delay Among Adults With Cancer During the COVID-19 Pandemic.

Importance: Cancer treatment delay has been reported to variably impact cancer-specific survival and coronavirus disease 2019 (COVID-19)-specific mortality during the severe acute respiratory syndrome coronavirus 2 pandemic. During the pandemic, treatment delay is being recommended in a nonquantitative, nonobjective, and nonpersonalized manner, and this approach may be associated with suboptimal outcomes. Quantitative integration of cancer mortality estimates and data on the consequences of treatment delay is needed to aid treatment decisions and improve patient outcomes. Objective: To obtain quantitative integration of cancer-specific and COVID-19-specific mortality estimates that can be used to make optimal decisions for individual patients and optimize resource allocation. Design, Setting, and Participants: In this decision analytical model, age-specific and stage-specific estimates of overall survival pre-COVID-19 were adjusted by the probability of COVID-19 (individualized by county, treatment-specific variables, hospital exposure frequency, and COVID-19 infectivity estimates), COVID-19 mortality (individualized by age-specific, comorbidity-specific, and treatment-specific variables), and delay of cancer treatment (impact and duration). These model estimates were integrated into a web application (OncCOVID) to calculate estimates of the cumulative overall survival and restricted mean survival time of patients who received immediate vs delayed cancer treatment. Using currently available information about COVID-19, a susceptible-infected-recovered model that accounted for the increased risk among patients at health care treatment centers was developed. This model integrated the data on cancer mortality and the consequences of treatment delay to aid treatment decisions. Age-specific and cancer stage-specific estimates of overall survival pre-COVID-19 were extracted from the Surveillance, Epidemiology, and End Results database for 691 854 individuals with 25 cancer types who received cancer diagnoses in 2005 to 2006. Data from 5 436 896 individuals in the National Cancer Database were used to estimate the independent impact of treatment delay by cancer type and stage. In addition, data from 275 patients in a nested case-control study were used to estimate the COVID-19 mortality rate by age group and number of comorbidities. Data were analyzed from March 17 to May 21, 2020. Exposures: COVID-19 and cancer. Main Outcomes and Measures: Estimates of restricted mean survival time after the receipt of immediate vs delayed cancer treatment. Results: At the time of the study, the OncCOVID web application allowed for the selection of up to 47 individualized variables to assess net survival for an individual patient with cancer. Substantial heterogeneity was found regarding the association between delayed cancer treatment and net survival among patients with a given cancer type and stage, and these 2 variables were insufficient to discriminate the net impact of immediate vs delayed treatment. Individualized overall survival estimates were associated with patient age, number of comorbidities, treatment received, and specific local community estimates of COVID-19 risk. Conclusions and Relevance: This decision analytical modeling study found that the OncCOVID web-based application can quantitatively aid in the resource allocation of individualized treatment for patients with cancer during the COVID-19 global pandemic.

Biopsy Cell Cycle Proliferation Score Predicts Adverse Surgical Pathology in Localized Renal Cell Carcinoma.

Active surveillance (AS) is an accepted management strategy for some patients with renal cell carcinoma, but limited tools are available to identify optimal AS candidates. While renal mass biopsy provides diagnostic information, risk stratification based on biopsy is limited. In a retrospective, multi-institutional cohort that underwent renal mass biopsy followed by surgery, we assessed the ability of the cell cycle proliferation (CCP) score from clinical biopsy specimens to predict adverse surgical pathology (ie, grade 3-4, pT stage ≥3, metastasis at surgery, or papillary type II). Of 202 patients, 98 (49%) had adverse surgical pathology. When added to a baseline model including age, sex, race, lesion size, biopsy grade, and histology, CCP score was significantly associated with adverse pathology when modeled as a binary (odds ratio [OR]: 2.44 for CCP score >0, p = 0.02) and a continuous (OR: 1.72 per one unit increase, p = 0.04) variable. Discriminative performance measured by the area under the curve (AUC) improved from 0.73 in the baseline model to 0.75 and 0.76 in models including the CCP score. In the subgroup of patients with nephrectomy CCP score available (n = 67), the biopsy-based model outperformed the nephrectomy-based model (AUC 0.78 vs 0.75). These data support prospective assessment of biopsy CCP score to confirm clinical validity and assess potential utility in AS-eligible patients. PATIENT SUMMARY: In patients with localized renal cell carcinoma who underwent renal mass biopsy followed by surgery, the cell cycle proliferation score from clinical biopsy specimens could predict adverse surgical pathology.