Background. Calibration is the process of estimating parameters of a mathematical model by matching model outputs to calibration targets. In the presence of nonidentifiability, multiple parameter sets solve the calibration problem, which may have important implications for decision making. We evaluate the implications of nonidentifiability on the optimal strategy and provide methods to check for nonidentifiability. Methods. We illustrate nonidentifiability by calibrating a 3-state Markov model of cancer relative survival (RS). We performed 2 different calibration exercises: 1) only including RS as a calibration target and 2) adding the ratio between the 2 nondeath states over time as an additional target. We used the Nelder-Mead (NM) algorithm to identify parameter sets that best matched the calibration targets. We used collinearity and likelihood profile analyses to check for nonidentifiability. We then estimated the benefit of a hypothetical treatment in terms of life expectancy gains using different, b...

Spinal pain is a common and disabling condition with considerable socioeconomic burden. Spine pain management in the United States has gathered increased scrutiny amidst concerns of overutilization of costly and potentially harmful interventions and diagnostic tests. Conservative interventions such as spinal manipulation, exercise and self-management may provide value for the care of spinal pain, but little is known regarding the cost-effectiveness of these interventions in the U.S. Our primary objective for this project is to estimate the incremental cost-effectiveness of spinal manipulation, exercise therapy, and self-management for spinal pain using an individual patient data meta-analysis approach. We will estimate the incremental cost-effectiveness of spinal manipulation, exercise therapy, and self-management using cost and clinical outcome data collected in eight randomized clinical trials performed in the U.S. Cost-effectiveness will be assessed from both societal and healthcare perspectives using QALYs, pain intensity, and disability as effectiveness measures. The eight randomized clinical trials used similar methods and included different combinations of spinal manipulation, exercise therapy, or self-management for spinal pain. They also collected similar clinical outcome, healthcare utilization, and work productivity data. A two-stage approach to individual patient data meta-analysis will be conducted. This project capitalizes on a unique opportunity to combine clinical and economic data collected in a several clinical trials that used similar methods. The findings will provide important information on the value of spinal manipulation, exercise therapy, and self-management for spinal pain management in the U.S.

Decision makers often desire both guidance on the most cost-effective interventions given current knowledge and also the value of collecting additional information to improve the decisions made (i.e., from value of information [VOI] analysis). Unfortunately, VOI analysis remains underused due to the conceptual, mathematical, and computational challenges of implementing Bayesian decision-theoretic approaches in models of sufficient complexity for real-world decision making. In this study, we propose a novel practical approach for conducting VOI analysis using a combination of probabilistic sensitivity analysis, linear regression metamodeling, and unit normal loss integral function--a parametric approach to VOI analysis. We adopt a linear approximation and leverage a fundamental assumption of VOI analysis, which requires that all sources of prior uncertainties be accurately specified. We provide examples of the approach and show that the assumptions we make do not induce substantial bias but greatly reduce the computational time needed to perform VOI analysis. Our approach avoids the need to analytically solve or approximate joint Bayesian updating, requires only one set of probabilistic sensitivity analysis simulations, and can be applied in models with correlated input parameters.

BACKGROUND: Contralateral prophylactic mastectomy (CPM) rates have substantially increased in recent years and may reflect an exaggerated perceived benefit from the procedure. The objective of this study was to evaluate the magnitude of the survival benefit of CPM for women with unilateral breast cancer. METHODS: We developed a Markov model to simulate survival outcomes after CPM and no CPM among women with stage I or II breast cancer without a BRCA mutation. Probabilities for developing contralateral breast cancer (CBC), dying from CBC, dying from primary breast cancer, and age-specific mortality rates were estimated from published studies. We estimated life expectancy (LE) gain, 20-year overall survival, and disease-free survival with each intervention strategy among cohorts of women defined by age, estrogen receptor (ER) status, and stage of cancer. RESULTS: Predicted LE gain from CPM ranged from 0.13 to 0.59 years for women with stage I breast cancer and 0.08 to 0.29 years for those with stage II breast cancer. Absolute 20-year survival differences ranged from 0.56% to 0.94% for women with stage I breast cancer and 0.36% to 0.61% for women with stage II breast cancer. CPM was more beneficial among younger women, stage I, and ER-negative breast cancer. Sensitivity analyses yielded a maximum 20-year survival difference with CPM of only 1.45%. CONCLUSIONS: The absolute 20-year survival benefit from CPM was less than 1% among all age, ER status, and cancer stage groups. Estimates of LE gains and survival differences derived from decision models may provide more realistic expectations of CPM.

BACKGROUND: Simulation models designed to evaluate cancer prevention strategies make assumptions on background mortality-the competing risk of death from causes other than the cancer being studied. Researchers often use the U.S. life tables and assume homogeneous other-cause mortality rates. However, this can lead to bias because common risk factors such as smoking and obesity also predispose individuals for deaths from other causes such as cardiovascular disease. METHODS: We obtained calendar year-, age-, and sex-specific other-cause mortality rates by removing deaths due to a specific cancer from U.S. all-cause life tables. Prevalence across 12 risk factor groups (3 smoking [never, past, and current smoker] and 4 body mass index [BMI] categories [<25, 25-30, 30-35, 35+ kg/m(2)]) were estimated from national surveys (National Health and Nutrition Examination Surveys [NHANES] 1971-2004). Using NHANES linked mortality data, we estimated hazard ratios for death by BMI/smoking using a Poisson regression model. Finally, we combined these results to create 12 sets of BMI and smoking-specific other-cause life tables for U.S. adults aged 40 years and older that can be used in simulation models of lung, colorectal, or breast cancer. RESULTS: We found substantial differences in background mortality when accounting for BMI and smoking. Ignoring the heterogeneity in background mortality in cancer simulation models can lead to underestimation of competing risk of deaths for higher-risk individuals (e.g., male, 60-year old, white obese smokers) by as high as 45%. CONCLUSION: Not properly accounting for competing risks of death may introduce bias when using simulation modeling to evaluate population health strategies for prevention, screening, or treatment. Further research is warranted on how these biases may affect cancer-screening strategies targeted at high-risk individuals.

BACKGROUND: Treatment options for colorectal cancer (CRC) have improved substantially over the past 25 years. Measuring the impact of these improvements on survival outcomes is challenging, however, against the background of overall survival gains from advancements in the prevention, screening, and treatment of other conditions. Relative survival is a metric that accounts for these concurrent changes, allowing assessment of changes in CRC survival. We describe stage- and location-specific trends in relative survival after CRC diagnosis. METHODS: We analyzed survival outcomes for 233965 people in the Surveillance Epidemiology and End Results (SEER) program who were diagnosed with CRC between January 1, 1975, and December 31, 2003. All models were adjusted for sex, race (black vs white), age at diagnosis, time since diagnosis, and diagnosis year. We estimated the proportional difference in survival for CRC patients compared with overall survival for age-, sex-, race-, and period-matched controls to account for concurrent changes in overall survival using two-sided Wald tests. RESULTS: We found statistically significant reductions in excess hazard of mortality from CRC in 2003 relative to 1975, with excess hazard ratios ranging from 0.75 (stage IV colon cancer; P < .001) to 0.32 (stage I rectal cancer; P < .001), indicating improvements in relative survival for all stages and cancer locations. These improvements occurred in earlier years for patients diagnosed with stage I cancers, with smaller but continuing improvements for later-stage cancers. CONCLUSIONS: Our results demonstrate a steady trend toward improved relative survival for CRC, indicating that treatment and surveillance improvements have had an impact at the population level.

INTRODUCTION In the 1990s, several organizations began recommending evaluation of > 12 lymph nodes during colon resection because of its association with improved survival. We examined practice implications of multispecialty quality guidelines over the past 20 years recommending evaluation of ≥ 12 lymph nodes during colon resection for adequate staging. MATERIALS AND METHODS We used the 1988 to 2009 Surveillance, Epidemiology, and End Results program to conduct a retrospective observational cohort study of 90,203 surgically treated patients with colon cancer. We used Cochran-Armitage tests to examine trends in lymph node examination over time and multivariate logistic regression to identify patient characteristics associated with guideline-recommended lymph node evaluation. RESULTS The introduction of practice guidelines was associated with gradual increases in guideline-recommended lymph node evaluation. From 1988 to 1990, 34% of patients had > 12 lymph nodes evaluated, increasing to 38% in 1994 to 1996 and to > 75% from 2006 to 2009. Younger, white patients and those with more-extensive bowel penetration (T3/4 nonmetastatic) and high tumor grade saw more-rapid increases in lymph node evaluation (P < .001). Multivariate analyses demonstrated a significant interaction between year of diagnosis and both T stage and grade, indicating that those with higher T stage and higher grade were more likely to receive guideline-recommended care earlier. CONCLUSION The implementation of lymph node evaluation guidelines was accepted gradually into practice but adopted more quickly among higher risk patients. By identifying patients who are least likely to receive guideline-recommended care, these findings present a starting point for promoting targeted improvements in cancer care and further understanding underlying contributors to these disparities.

BACKGROUND: Over the past 20 years, surgical practice organizations have recommended the identification of >/=12 lymph nodes from surgically treated colon cancer patients as an indicator of quality performance for adequate staging; however, studies suggest that significant variation exists among hospitals in their level of adherence to this recommendation. We examined hospital-level factors that were associated with institutional improvement or maintenance of adequate lymph node evaluation after the introduction of surgical quality guidelines. RESEARCH DESIGN: Using the 1996-2007 SEER-Medicare data, we evaluated hospital characteristics associated with short-term (1999-2001), medium-term (2002-2004), and long-term (2005-2007) guideline-recommended (>/=12) lymph node evaluation compared with initial evaluation levels (1996-1998) using chi tests and multivariate logistic regression analysis, adjusting for patient case-mix. RESULTS: We identified 228 hospitals that performed >/=6 colon cancer surgeries during each study period from 1996-2007. In the initial study period (1996-1998), 26.3% (n=60) of hospitals were performing guideline-recommended evaluation, which increased to 28.1% in 1999-2001, 44.7% in 2002-2004, and 70.6% in 2005-2007. In multivariate analyses, a hospital's prior guideline performance [odds ratio (OR) (95% confidence interval (CI)): 4.02 (1.92, 8.42)], teaching status [OR (95% CI): 2.33 (1.03, 5.28)], and American College of Surgeon's Oncology Group membership [OR (95% CI): 3.39 (1.39, 8.31)] were significantly associated with short-term guideline-recommended lymph node evaluation. Prior hospital performance [OR (95% CI): 2.41 (1.17, 4.94)], urban location [OR (95% CI): 2.66 (1.12, 6.31)], and American College of Surgeon's Oncology Group membership [OR (95% CI): 6.05 (2.32, 15.77)] were associated with medium-term performance; however, these factors were not associated with long-term performance. CONCLUSIONS: Over the 12-year period, there were marked improvements in hospital performance for guideline-recommended lymph node evaluation. Understanding patterns in improvement over time contributes to debates over optimal designs of quality-improvement programs.

BACKGROUND / OBJECTIVE: Modelers lack a tool to systematically and clearly present complex model results, including those from sensitivity analyses. The objective was to propose linear regression metamodeling as a tool to increase transparency of decision analytic models and better communicate their results. METHODS: We used a simplified cancer cure model to demonstrate our approach. The model computed the lifetime cost and benefit of 3 treatment options for cancer patients. We simulated 10,000 cohorts in a probabilistic sensitivity analysis (PSA) and regressed the model outcomes on the standardized input parameter values in a set of regression analyses. We used the regression coefficients to describe measures of sensitivity analyses, including threshold and parameter sensitivity analyses. We also compared the results of the PSA to deterministic full-factorial and one-factor-at-a-time designs. RESULTS: The regression intercept represented the estimated base-case outcome, and the other coefficients described the relative parameter uncertainty in the model. We defined simple relationships that compute the average and incremental net benefit of each intervention. Metamodeling produced outputs similar to traditional deterministic 1-way or 2-way sensitivity analyses but was more reliable since it used all parameter values. CONCLUSIONS: Linear regression metamodeling is a simple, yet powerful, tool that can assist modelers in communicating model characteristics and sensitivity analyses.