Bootstrap P-values Research Articles

The impact of the COVID-19 outbreak on intra- and inter-regional domestic travel: Evidence from Spain

This study empirically compares domestic tourists’ behavior before and after the Covid-19 outbreak. Specifically, official data are used to characterize the travel behavior of residents in Spain who traveled through this country for reasons of leisure, recreation, and vacations in 2019 and 2020. A discrete choice model is employed to unravel the main variables that influence the decision of being an inter-regional traveler. The bootstrap p-value method is used to detect significant changes in the marginal effect of the different variables after the Covid-19 outbreak. The estimation results demonstrate the following: (i) domestic inter- and intra-regional travelers have significant differences and, therefore, policy-makers should not design and implement “one size fits all” policies for domestic tourism; (ii) in addition to socio-economic factors, the regional push-pull factors (e.g., weather) and trip-related characteristics (e.g., type of accommodation) are also important in explaining inter-regional travel decisions; (iii) a high number of Covid-19 infections in the region of origin increases the probability of traveling inter-regionally; and (iv) the Covid-19 outbreak has caused an abrupt shift in domestic travelers’ profiles.

Effects of gene expression in 5-FU metabolic pathways in a phase III trial evaluating adjuvant S-1 therapy compared to surgery alone following curative resection for biliary tract cancer (JCOG1202A1).

732 Background: S-1, an oral fluoropyrimidine derivative, is standard adjuvant therapy in Japanese patients with resected biliary tract cancer (BTC), based on the results of JCOG1202, a phase III trial evaluating adjuvant S-1 following curative resection for BTC compared to surgery alone. The efficacy of 5-FU-based therapy is related to the expression of enzymes involved in 5-FU metabolic pathways, such as thymidylate synthase (TS), thymidine phosphorylase (TP), orotate phosphoribosyltransferase (OPRT) and dihydropyrimidine dehydrogenase (DPD). The aim of this study was to evaluate the impact of messenger RNA (mRNA) levels of the four 5-FU metabolic pathway genes on the outcomes of patients enrolled in JCOG1202. Methods: Tumor cell RNA was isolated from formalin-fixed paraffin-embedded primary BTC specimens resected in 264 of 440 randomized patients in JCOG1202. The four 5-FU metabolic pathway genes were measured in 183 patients (surgery alone: n = 94; adjuvant S-1: n = 89) who were randomly divided to training (n = 96) or validation set (n = 87). The endpoints of interest were the predictive values of the four genes for the efficacy of adjuvant S-1 on overall survival and relapse-free survival (RFS), and we here report the results regarding RFS. Cut-off levels for mRNA expression were selected in the training set which minimized the bootstrap p-values (2,000 samples) of an interaction term of treatment (surgery alone or S-1) and mRNA expression in a Cox regression model. Results: There were no obvious differences in each mRNA level and clinical characteristics between surgery alone and adjuvant S-1 groups. RFS tended to be better with adjuvant S-1 (hazard ratio [HR] = 0.790, [95% confidence interval: 0.524-1.192]) compared to surgery alone, which was maintained in the low DPD population (HR = 0.440 [0.216-0.898] in training set and 0.748 [0.334-1.675] in validation set), in the low TP population (HR = 0.709 [0.388-1.296] and 0.602 [0.287-1.262]), and in the high OPRT population (HR = 0.520 [0.152-1.779] and 0.609 [0.161-2.304]). Conclusions: The efficacy of adjuvant S-1 on RFS was representative in BTC patients with intratumoral gene expressions of low DPD, low TP, and high OPRT, which was disappeared in the population with high DPD, high TP, and low OPRT. These biomarkers might be useful to predict therapeutic benefits with S-1 containing chemotherapy for BTC.

Mutual fund performance persistence: Factor models and portfolio size

We re-examine US mutual fund performance persistence. We investigate persistence (i) using both “academic” factor models and “practitioner” index models, (ii) using decile-size recursive portfolios and also portfolios formed from smaller numbers of funds, (iii) using nonparametric bootstrap p-values as well as conventional t-tests and (iv) using both net-of-fee fund returns (net alphas) and gross alphas. Our key result is that positive net alpha performance persistence can be found using small portfolios of funds together with a holding period of 6 months or less, for both practitioner index models and academic factor models.

Testing for Trend Specifications in Panel Data Models

This article proposes a consistent nonparametric test for common trend specifications in panel data models with fixed effects. The test is general enough to allow for heteroscedasticity, cross-sectional and serial dependence in the error components, has an asymptotically normal distribution under the null hypothesis of correct trend specification, and is consistent against various alternatives that deviate from the null. In addition, the test has an asymptotic unit power against two classes of local alternatives approaching the null at different rates. We also propose a wild bootstrap procedure to better approximate the finite sample null distribution of the test statistic. Simulation results show that the proposed test implemented with bootstrap p-values performs reasonably well in finite samples. Finally, an empirical application to the analysis of the U.S. per capita personal income trend highlights the usefulness of our test in real datasets.

Genetic resources in asparagus: diversity and relationships in a collection from different origins and breeding status

ABSTRACT Asparagus (Asparagus officinalis) is a worldwide important vegetable. Assessing the genetic diversity and relationships in a germplasm collection is a key step in the design of breeding programmes. The aim of this work was to evaluate the genetic diversity in asparagus from a wide array of origins assessed by Sequence Related Amplified Polymorphism (SRAP) markers. Twenty-four accessions of A. officinalis from 13 different origins and one accession of A. pseudoscaber were considered. Sixteen combinations of SRAP primers were tested. A total of 329 bands were obtained (98.48% polymorphism). The average polymorphic information content was 0.20. Hierarchical clustering was carried out, the dendrogram was plotted with bootstrap p-values for the different branches and the Dice’s distance matrix was represented by a Principal Coordinates Analysis. It was not possible to establish a clear origin-related grouping. The reduced cycles of selection accomplished along with the contemporaneous plant breeding and the open-pollinated nature of the species, could have precluded a marked differentiation in the species related to origins or cultivated to wild status. The genetic base of asparagus is not as narrow as previously thought. The information of the genetic relationships among asparagus germplasm could be useful to delineate crosses in order to seek potential hybrid vigour.

IL15RA and SMAD3 Genetic Variants Predict Overall Survival in Metastatic Colorectal Cancer Patients Treated with FOLFIRI Therapy: A New Paradigm.

Simple SummaryThere is an increasing scientific interest in the study of the interaction between the immune system and drugs in cancer that can affect the efficacy of an anti-cancer treatment. This study was undertaken to better understand if the genetic characteristic of a cancer patient’s immune system can predict the tumor response to the treatment and the duration of survival. The topic was studied on 335 metastatic colorectal cancer patients treated with a first-line chemotherapy (FOLFIRI regimen, irinotecan-5-fluorouracil-leucovorin). The research highlighted two markers, IL15RA-rs7910212 and SMAD3-rs7179840, significantly associated with the patient’s survival. When considering IL15RA-rs7910212 and SMAD3-rs7179840 in combination with other two genetic markers previously investigated (NR1I2-rs1054190, VDR-rs7299460), we built up a highly predictive genetic score of survival. The herein identified markers must be further validated, but still represent good candidates to understand how much a patient with a metastatic colorectal cancer can benefit from a chemotherapy with FOLFIRI regimen.A new paradigm in cancer chemotherapy derives from the interaction between chemotherapeutics, including irinotecan and 5-fluorouracil (5-FU), and the immune system. The patient’s immune response can modulate chemotherapy effectiveness, and, on the other hand, chemotherapeutic agents can foster tumor cell immunogenicity. On these grounds, the analysis of the cancer patients’ immunogenetic characteristics and their effect on survival after chemotherapy represent a new frontier. This study aims to identify genetic determinants in the immuno-related pathways predictive of overall survival (OS) after FOLFIRI (irinotecan, 5-FU, leucovorin) therapy. Two independent cohorts comprising a total of 335 patients with metastatic colorectal cancer (mCRC) homogeneously treated with first-line FOLFIRI were included in the study. The prognostic effect of 192 tagging genetic polymorphisms in 34 immune-related genes was evaluated using the bead array technology. The IL15RA rs7910212-C allele was associated with worse OS in both discovery (HR: 1.57, p = 0.0327, Bootstrap p-value = 0.0280) and replication (HR: 1.71, p = 0.0411) cohorts. Conversely, SMAD3 rs7179840-C allele was associated with better OS in both discovery (HR: 0.65, p = 0.0202, Bootstrap p-value = 0.0203) and replication (HR: 0.61, p = 0.0216) cohorts. A genetic prognostic score was generated integrating IL15RA-rs7910212 and SMAD3-rs7179840 markers with inflammation-related prognostic polymorphisms we previously identified in the same study population (i.e., PXR [NR1I2]-rs1054190, VDR-rs7299460). The calculated genetic score successfully discriminated patients with different survival probabilities (p < 0.0001 log-rank test). These findings provide new insight on the prognostic value of genetic determinants, such as IL15RA and SMAD3 markers, and could offer a new decision tool to improve the clinical management of patients with mCRC receiving FOLFIRI.

Abstract 5101: CAB-CAR-T: The prioritization of cell surface protein targets for conditionally active biologics to treat all solid tumors

Abstract Introduction Proteins on the surface of cancer cells represent viable targets for conditionally active biologic (CAB) CAR-T therapy that only work in the tumor microenvironment. By modeling properties of cell surface proteins from TCGA datasets we seek to identify optimal targets across all TCGA cancer cohorts that when used in CAB-CAR-T therapies will provide the greatest number of treatment options for patients across all cancer malignancies. Method Cell surface proteins(n=1086) were identified and meta data specific to each gene was organized from a variety of public databases for data modeling. Various approaches based on ideal CAB-CAR-T properties are used to rank the cell surface proteins as therapeutic targets. The 31 TCGA cohorts representing the most comprehensive collection of genomic profiled tumor samples and outcomes of all cancers is used to rank cell surface proteins for the percentage of patients in each cohort who would be eligible for treatment based on predetermined mRNA cutoffs. Various approaches were used to filter the ranked list based on ideal CAB-CAR-T properties: known antibodies that can be used for initial CAB-CAR-T development, non-receptor as a static protein structure, highly expressed in CCLE indicating mRNA expression is a feature of cancer cell lines and low expression in critical tissues like Heart, Lung, Liver, Muscle etc. Different ranked lists of cell surface proteins were used to determine the number of CAB-CAR-T products required to treat 90% of patients in TCGA cohorts. A patient with the highest mRNA expression above the mean plus one standard deviation as determine across all TCGA samples is assigned to that specific protein biomarker as eligible for treatment and removed from the list of patients still to be treated. A bootstrap p-value for the ranked lists was determined by calculating the minimum number of randomly selected cell surface proteins that would give 90% coverage of the TCGA cohort. Results It was shown that it is reasonable to find ranked list of genes with high mRNA expression in TCGA and minimum expression in off-target critical tissue that 5-7 CAB-CAR-T products could be used to treat 90% of TCGA patients. To achieve 100% treatment coverage each additional CAB-CAR-T product added to the list had minimum inclusion of additional patients for treatment. Conclusions By modeling various properties of cell surface proteins to establish future development of CAB-CAR-T products it is reasonable to expect 90% patient coverage with 10 distinct therapies. Further modeling will be performed to exam combination therapies where tumor heterogeneity is an important criteria for the ranked list to have efficacy, with a goal of maximizing complete responses (CRs) and minimize the chance of relapse in the future. Citation Format: Hiba A. Shaban, Laurence Jadin, James Onuffer, Farzad Haerizadeh, Alissa Kerner, Gregory Schreiber, Gregory Frost, Scooter Willis. CAB-CAR-T: The prioritization of cell surface protein targets for conditionally active biologics to treat all solid tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 5101.

Front-Line Ibrutinib Treatment Is Associated with Longer Time to Next Treatment, Net Total Cost Reduction, and Lower Healthcare Resource Utilization Compared to Chemoimmunotherapy in Patients with Chronic Lymphocytic Leukemia

Introduction: Novel targeted therapies greatly improved outcomes of patients with chronic lymphocytic leukemia (CLL). Among these, ibrutinib is the only one approved in the front-line (1L) setting for patients with or without del(17p) or TP53 mutations. This study aims to compare healthcare resource utilization (HRU), costs, and time to next treatment (TTNT), following initiation of 1L ibrutinib single agent versus CIT and in particular bendamustine/rituximab (BR), in real-world.Methods: The Optum Clinformatics® Extended DataMart De-Identified Databases (01/2004-06/2017) was used to identify adults with CLL diagnosis initiating 1L ibrutinib single agent or CIT (index date) on 02/12/2014 or after. For CIT, only combinations of chemotherapy and CD20 antibody agents recommended by the National Comprehensive Cancer Network guidelines were considered. Patients were also required to have ≥12 months and ≥30 days of continuous eligibility pre- and post-index date, respectively. For each comparison, patients' baseline characteristics were balanced using inverse probability of treatment weighting (IPTW). HRU, costs, and TTNT, were evaluated over 1L therapy. Mean monthly cost differences (MMCD) and rate ratios (RR) obtained from generalized linear models were used to compare costs and HRU, respectively, along with 500-sample bootstrap confidence intervals and P-values. TTNT was defined as the time from initiation of 1L treatment to the initiation of a new antineoplastic agent that was not part of the 1L regimen. TTNT comparisons were performed using Kaplan-Meier (KM) curves with log-rank tests. The hazards of initiating a new treatment were compared between the cohorts using hazard ratios (HRs) obtained from Cox models.Results: Of 1,161 1L patients, 322 initiated ibrutinib and 839 initiated CIT, including 455 on BR. After IPTW, cohorts had comparable baseline characteristics. CIT patients had significantly more days with outpatient (OP) services (RR=0.75; P<0.05), mainly driven by antineoplastic drug administration-related OP visits (Fig.1; P<0.0001). Ibrutinib patients incurred significantly higher pharmacy costs (MMCD=$6,849; P<0.0001) that were fully offset by lower medical costs (MMCD=-$10,615; P<0.0001), resulting in net monthly total cost reduction of $3,766 (P<0.0001; Fig. 2) vs. CIT patients. While no significant difference was observed in the number of days with other OP visits (i.e., not related to antineoplastic treatment), the corresponding cost was significantly lower in ibrutinib patients (MMCD=-$1,002; P=0.0120).Ibrutinib patients also had fewer OP visits compared to BR patients, mainly due to drug administration-related OP visits (Fig. 1). Relative to BR, ibrutinib patients' higher pharmacy costs (MMCD=$7,002; P<0.0001) were fully offset by lower medical costs (MMCD=-$12,571; P<0.0001), yielding monthly reduction in the total cost of care of $5,569 (P<0.0001; Fig. 2).KM analysis of TTNT revealed that, at 24 months post-index, 88.6% of ibrutinib patients did not initiate a new treatment compared to 75.9% among CIT patients (log-rank P=0.0347; Fig. 3A). The hazard of initiating a new treatment was also lower for ibrutinib when compared to CIT, with significant HRs of 0.30 (P=0.0045), 0.59 (P=0.0415), and 0.54 (P=0.0171) at 3, 18, and 24 months post-index, respectively (Fig. 3A). The proportion of patients who did not initiate a new treatment 24 months post-index was significantly higher in ibrutinib cohort than in BR cohort (89.0% versus 79.0%; log-rank P=0.0421). The hazard of initiating a new treatment was also lower for ibrutinib compared to BR, with significant HRs of 0.40 (P=0.0449), 0.54 (P=0.0410), 0.55 (P=0.0360), 0.57 (P=0.0346), and 0.54 (P=0.0219) at 3, 9, 12, 18, and 24 months post-index, respectively (Fig. 3B).Conclusion: During 1L CLL treatment, ibrutinib patients had higher pharmacy costs that were fully offset by lower medical costs, resulting in net total cost reduction compared to CIT or BR patients. In addition to fewer OP visits and lower OP costs related to the antineoplastic treatment, ibrutinib was also associated with lower cost in other OP services. TTNT was significantly longer among ibrutinib patients vs. CIT or BR patients. Furthermore, consistent with the high rates of progression-free survival observed among ibrutinib patients in clinical trials, nearly 90% of patients did not initiate a new treatment after 24 months of follow-up. [Display omitted] DisclosuresWang:Janssen Scientific Affairs, LLC: Employment, Equity Ownership. Emond:Janssen Scientific Affairs, LLC: Research Funding. Romdhani:Janssen Scientific Affairs, LLC: Research Funding. Lefebvre:Janssen Scientific Affairs, LLC: Research Funding. Sundaram:Johnson & Johnson: Employment, Equity Ownership, Other: Travel; AbbVie: Employment, Equity Ownership, Other: Travel. Mato:TG Therapeutics: Research Funding; Regeneron: Research Funding; Pharmacyclics: Consultancy, Honoraria, Research Funding; Janssen: Consultancy, Honoraria; Sunesis: Honoraria, Research Funding; Prime Oncology: Speakers Bureau; Abbvie: Consultancy; Acerta: Research Funding; AstraZeneca: Consultancy; Celgene: Consultancy.

Panel kink regression with an unknown threshold

In this paper, we extend the kink regression model with an unknown threshold in Hansen (2017) to the panel data framework, where the cross-sectional dimension (N) goes to infinity and the time period (T) is fixed. Following the literature of threshold regressions, we propose an estimator based on the within-group transformation. Under fixed threshold effect assumption, we establish that the slope and threshold estimators are jointly normally distributed with the same convergence rate OpN−1∕2 and a non-zero asymptotic covariance. We also suggest a sup-Wald test for the presence of kink effect, and derive its limiting distribution. A bootstrap procedure is proposed to obtain the bootstrap p-values to improve the finite sample performance of the test. Monte Carlo simulations show that the FE estimator and the sup-Wald test perform quite well in estimating the unknown parameters and testing for kink effect, respectively.

The formal relationship between analytic and bootstrap approaches to parametric inference

Two routes most commonly proposed for accurate inference on a scalar interest parameter in the presence of a (possibly high-dimensional) nuisance parameter are parametric simulation (‘bootstrap’) methods, and analytic procedures based on normal approximation to adjusted forms of the signed root likelihood ratio statistic. Under some null hypothesis of interest, both methods yield p-values which are uniformly distributed to error of third-order in the available sample size. But, given a specific dataset, what is the formal relationship between p-values calculated by the two approaches? We show that the two methodologies give the same inference to second order in general: the analytic p-value calculated from a dataset will agree with the bootstrap p-value constructed from that same dataset to O(n−1), where n is the sample size. In practice, the agreement is often startling.

Bootstrap approach to test the homogeneity of order restricted mean vectors when the covariance matrices are unknown

ABSTRACTTesting homogeneity of multivariate normal mean vectors under an order restriction when the covariance matrices are unknown, arbitrary positive definite and unequal are considered. This problem of testing has been studied to some extent, for example, by Kulatunga and Sasabuchi (1984) when the covariance matrices are known and also Sasabuchi et al. (2003) and Sasabuchi (2007) when the covariance matrices are unknown but common. In this paper, a test statistic is proposed and because of the main advantage of the bootstrap test is that it avoids the derivation of the complex null distribution analytically, a bootstrap test statistic is derived and since the proposed test statistic is location invariance the bootstrap p-value defined logical and some steps are presented to estimate it. Our numerical studies via Monte Carlo simulation show that the proposed bootstrap test can correctly control the type I error rates. The power of the test for some of the p-dimensional normal distributions is computed by Monte Carlo simulation. Also, the null distribution of test statistic is estimated using kernel density. Finally, the bootstrap test is illustrated using a real data.

Accurate p-values for adaptive designs with binary endpoints.

Adaptive designs encompass all trials allowing various types of design modifications over the course of the trial. A key requirement for confirmatory adaptive designs to be accepted by regulators is the strong control of the family-wise error rate. This can be achieved by combining the p-values for each arm and stage to account for adaptations (including but not limited to treatment selection), sample size adaptation and multiple stages. While the theory for this is novel and well-established, in practice, these methods can perform poorly, especially for unbalanced designs and for small to moderate sample sizes. The problem is that standard stagewise tests have inflated type I error rate, especially but not only when the baseline success rate is close to the boundary and this is carried over to the adaptive tests, seriously inflating the family-wise error rate. We propose to fix this problem by feeding the adaptive test with second-order accurate p-values, in particular bootstrap p-values. Secondly, an adjusted version of the Simes procedure for testing intersection hypotheses that reduces the built-in conservatism is suggested. Numerical work and simulations show that unlike their standard counterparts the new approach preserves the overall error rate, at or below the nominal level across the board, irrespective of the baseline rate, stagewise sample sizes or allocation ratio. Copyright © 2017 John Wiley & Sons, Ltd.

Stop losses momentum strategy: From profit maximization to risk control under White’s Bootstrap Reality Check

Abstract This paper proposes a new approach to decision making processes for investors to focus on factor investing and stock selection strategies on the national stock market by capturing the momentum effect (when two portfolios of past relative winners and past losers continue to beat a given benchmark for a certain period of time in the future). Our approach is based on ranking all the combinations of strategy design (5184 strategies) and the disclosure of the momentum effect with two criteria (mean return and risk) controlling for momentum return probability distribution. It is new perspective on the momentum effect permitting its analysis from a comprehensive view taking into account the strategy’s different design elements and observing how the disposition of investor preferences has changed depending on the criterion. We distinguish two criteria: (1) maximizing the mean return of the investment with control t-statistics and the author’s innovation with the Bootstrap p-value, (2) minimizing the risk approximated by the number of drawdowns. This paper conducts the first comprehensive examination of the momentum effect on the Russian stock market.

The Direct Economic Burden of Gout in an Elderly Canadian Population.

To estimate the direct healthcare cost and resource use from the public payer perspective between patients with incident gout and matched gout-free patients in Ontario. Patients with incident gout aged ≥ 66 with uninterrupted Ontario Health Insurance Plan (OHIP) coverage in the 1-year baseline period were included in the study. Patients with gout were indexed at first gout diagnosis or prescription over the study period April 1, 2008, to March 31, 2014. Gout-free patients with no gout diagnosis within history were matched (up to 5:1) to each patient with gout. Linked medical records were analyzed until end of study, death, or OHIP ineligibility. Bang and Tsiatis adjusted healthcare costs and resource use were compared using bootstrap p-values and 95% CI. A total of 29,894 patients with gout and 148,231 gout-free patients were included in the study. Patients were 56% male, had a median Adjusted Clinical Group healthcare resource use band of moderate morbidity, and had a median age of 75-79 years. Baseline comorbidities were similar between groups except for renal disease. Analyzing 5-year total healthcare costs, patients with gout ($44,297) incurred a significantly higher average healthcare cost compared to gout-free patients ($33,965), for an incremental cost of $10,332 (95% CI $9617-$11,039; p < 0.01). Similar trends were observed in all individual healthcare component cost and use metrics. Following onset of gout, patients in Ontario incur significantly greater healthcare costs and resource use compared to matched gout-free patients. Alternative gout management strategies should be investigated to reduce the incremental burden of gout borne by the Ontario healthcare system.

A spatial scan statistic for beta regression

Spatial Scan Statistics have been developed for geographical cluster detection in different types of models, for example, Bernoulli, multinomial, Poisson, Exponential, Weibull and Normal. However, some data are continuous in the interval (0,1) such as rates, proportions and indices, or are limited in the interval (a,b), a<b. In this paper, we propose a spatial scan statistic for a Beta regression model. The test statistic is based on a likelihood ratio test and evaluated using Bootstrap p-value. The proposed method is illustrated using index of basic education and infant mortality in the Brazilian Amazon. The statistical power, sensitivity and positive predicted value of the test are examined through a simulation study.

A Chi-Square Goodness-of-Fit Test for Autoregressive Logistic Regression Models with Applications to Patient Screening

We propose a chi-square goodness-of-fit test for autoregressive logistic regression models. General guidelines for a two-dimensional binning strategy are provided, which make use of two types of maximum likelihood parameter estimates. For smaller sample sizes, a bootstrap p-value procedure is discussed. Simulation studies indicate that the test procedure satisfactorily approximates the correct size and has good power for detecting model misspecification. In particular, the test is very good at detecting the need for an additional lag. An application to a dataset relating to screening patients for late-onset Alzheimer’s disease is provided.

Specification test for panel data models with interactive fixed effects

In this paper, we propose a consistent nonparametric test for linearity in a large dimensional panel data model with interactive fixed effects. Both lagged dependent variables and conditional heteroskedasticity of unknown form are allowed in the model. We estimate the model under the null hypothesis of linearity to obtain the restricted residuals which are then used to construct the test statistic. We show that after being appropriately centered and standardized, the test statistic is asymptotically normally distributed under both the null hypothesis and a sequence of Pitman local alternatives by using the concept of conditional strong mixing that was recently introduced by Prakasa Rao (2009). To improve the finite sample performance, we propose a bootstrap procedure to obtain the bootstrap p-value. A small set of Monte Carlo simulations illustrates that our test performs well in finite samples. An application to an economic growth panel dataset indicates significant nonlinear relationships between economic growth, initial income level and capital accumulation.

Prognostic value of circulating tumor cells’ reduction in patients with extensive small-cell lung cancer

ObjectivesCirculating tumor cells (CTCs) have been hypothesized to be a prognostic factor in small-cell lung cancer (SCLC), and different cutoffs have been proposed to identify patients at high risk. We assessed the prognostic value of CTCs in patients with extensive SCLC. Materials and methodsCTCs were assessed with the CellSearch system in 60 extensive SCLC patients. CTC count at baseline or after one cycle of chemotherapy (cycle-1) or as change after chemotherapy were analyzed separately. Primary outcome was overall survival. The accuracy of prognostic role was assessed by Harrell's c-index. “Optimal” cutoffs were derived by bootstrap resampling to reduce the overfitting bias; accuracy improvement was estimated by calculating the difference of c-indexes of models including clinical variables with or without CTCs. ResultsCTCs were identified in 90% (54/60) of patients at baseline, in which CTC count ranged from 0 to 24,281. CTC count was strongly associated with the number of organs involved. The prognostic accuracy was only marginally increased by the addition to clinical information of “optimal” CTC cutoffs at baseline and after cycle-1. Conversely, a reduction of CTC count higher than 89% following chemotherapy significantly improved prognostic accuracy (bootstrap p-value=0.009) and was associated with a lower risk of death (HR 0.24, 95% CI 0.09–0.61). When previously proposed cutoffs were applied to our cohort, they showed only marginal improvement of the prognostic accuracy. ConclusionCTCs have useful prognostic role in extensive SCLC, but only the change of CTC count after the first cycle of chemotherapy provides clinically relevant information. Previously reported CTC cutoffs were not prognostic in our cohort of patients.

Specification Test for Panel Data Models with Interactive Fixed Effects

In this paper, we propose a consistent nonparametric test for linearity in a large dimensional panel data model with interactive fixed effects. Both lagged dependent variables and conditional heteroskedasticity of unknown form are allowed in the model. We estimate the model under the null hypothesis of linearity to obtain the restricted residuals which are then used to construct the test statistic. We show that after being appropriately centered and standardized, the test statistic is asymptotically normally distributed under both the null hypothesis and a sequence of Pitman local alternatives by using the concept of conditional strong mixing that was recently introduced by Prakasa Rao (2009). To improve the finite sample performance, we propose a bootstrap procedure to obtain the bootstrap p-value. A small set of Monte Carlo simulations illustrates that our test performs well in finite samples. An application to an economic growth panel dataset indicates significant nonlinear relationships between economic growth, initial income level and capital accumulation.

Bicycle helmet wearing is not associated with close motor vehicle passing: a re-analysis of Walker, 2007.

ObjectivesTo re-analyse bicycle overtaking data collected by Walker (2007) with a view to assess factors associated with close passing (<1 m), to adjust for other observed factors in a multivariable analysis, and to assess the extent to which the sample size in the original analysis may have contributed to spurious results.MethodA re-analysis of 2,355 motor vehicle passing events recorded by Walker that includes information on cyclist's distance to the kerb, vehicle size and colour, city of observation, time of day, whether the event occurred while in a bikelane and helmet wearing. Each variable was considered for a final, multivariable model using purposeful selection of variables. The analysis was repeated using multiple logistic regression with passing distance dichotomised by the one metre rule. Bootstrap p-values were computed using sample sizes computed from conventional values of power and effect size.ResultsThe previously observed significant association between passing distance and helmet wearing was not found when dichotomised by the one metre rule. Other factors were found to be significantly associated with close passing including cyclists' distance to the kerb, vehicle size and city of observation (Salisbury or Bristol, UK). P-values from bootstrap samples indicate the significance of helmet wearing resulted from an overly large sample size.ConclusionsAfter re-analysis of Walker's data, helmet wearing is not associated with close motor vehicle passing. The results, however, highlight other more important factors that may inform effective bicycle safety strategies.