Dichotomous Factor Research Articles

Comparison of four healthy lifestyle scores for predicting cardiovascular events in a national cohort study

The protective effect of different healthy lifestyle scores for the risk of cardiovascular disease (CVD) was reported, although the comparisons of performance were lacking. We compared the performance measures of CVDs from different healthy lifestyle scores among Taiwanese adults. We conducted a nationwide prospective cohort study of 6042 participants (median age 43 years, 50.2% women) in Taiwan’s Hypertensive, Hyperglycemia and Hyperlipidemia Survey, of whom 2002 were free of CVD at baseline. The simple and weighted the Mediterranean diet related healthy lifestyle (MHL) scores were defined as a combination of normal body mass index, Mediterranean diet, adequate physical activity, non-smokers, regular healthy drinking, and each dichotomous lifestyle factor. The World Cancer Research Fund and American Institute for Cancer Research cancer prevention recommended lifestyle and Life's Simple 7 following the guideline definition. The incidence of CVD among the four healthy lifestyle scores, each divided into four subgroups, was estimated. During a median 14.3 years follow-up period, 520 cases developed CVD. In the multivariate-adjusted Cox proportional hazard models, adherence to the highest category compared with the lowest one was associated with a lower incidence of CVD events, based on the simple (hazard ratio [HR] 0.43, 95% confidence interval [CI] 0.2–0.94) and weighted MHL scores (HR 0.44, 95% CI 0.28–0.68). Additionally, age played a role as a significant effect modifier for the protective effect of the healthy lifestyle scores for CVD risk. Specifically, the performance measures by integrated discriminative improvement showed a significant increase after adding the simple MHL score (integrated discriminative improvement: 0.51, 95% CI 0.16–0.86, P = 0.002) and weighted MHL score (integrated discriminative improvement: 0.38, 95% CI 0.01–0.74, P = 0.021). We demonstrated that the healthy lifestyle scores with an inverse association with CVD and reduced CVD risk were more likely for young adults than for old adults. Further studies to study the mechanism of the role of lifestyle on CVD prevention are warranted.

Serum alpha-fetoprotein and clinical outcomes in patients with advanced hepatocellular carcinoma treated with ramucirumab

BackgroundPost hoc analyses assessed the prognostic and predictive value of baseline alpha-fetoprotein (AFP), as well as clinical outcomes by AFP response or progression, during treatment in two placebo-controlled trials (REACH, REACH-2).MethodsSerum AFP was measured at baseline and every three cycles. The prognostic and predictive value of baseline AFP was assessed by Cox regression models and Subpopulation Treatment Effect Pattern Plot method. Associations between AFP (≥ 20% increase) and radiographic progression and efficacy were assessed.ResultsBaseline AFP was confirmed as a continuous (REACH, REACH-2; p < 0.0001) and dichotomous (≥400 vs. <400 ng/ml; REACH, p < 0.01) prognostic factor, and was predictive for ramucirumab survival benefit in REACH (p = 0.0042 continuous; p < 0.0001 dichotomous). Time to AFP (hazard ratio [HR] 0.513; p < 0.0001) and radiographic (HR 0.549; p < 0.0001) progression favoured ramucirumab. Association between AFP and radiographic progression was shown for up to 6 (odds ratio [OR] 5.1; p < 0.0001) and 6–12 weeks (OR 1.8; p = 0.0065). AFP response was higher with ramucirumab vs. placebo (p < 0.0001). Survival was longer in patients with an AFP response than patients without (13.6 vs. 5.6 months, HR 0.451; 95% confidence interval, 0.354–0.574; p < 0.0001).ConclusionsAFP is an important prognostic factor and a predictive biomarker for ramucirumab survival benefit. AFP ≥ 400 ng/ml is an appropriate selection criterion for ramucirumab.Clinical Trial RegistrationClinicalTrials.gov, REACH (NCT01140347) and REACH-2 (NCT02435433).

Comparison of Composites, Dichotomous, and Latent Factor Measurement Operationalizations in Hospitality Research on Moderating Effects

Moderation testing through latent factor models is relatively underutilized in hospitality and tourism research. The purpose of this research is to highlight the differences in the treatment of measurements of reflective constructs as composite indices versus latent factors in moderating effect tests in hospitality research. For this research, we build our primer on the investigation of the differences in customer satisfaction with the perceived entertainment experience at a hospitality/tourism attraction, contingent on customers’ personality trait extraversion, borrowed from the Big-Five mini marker inventory. Our findings illustrate the consequences of the measurement conceptualization and the representation of constructs in statistical models with interaction effects. While using composites simplifies the estimation of the regression paths and provides a reasonable sense of the direction of the effect and its statistical significance, it is not always aligned with the theoretical and conceptual underpinning of the employed constructs. A statistical model with composites may underestimate an interaction effect, whereas a model with a dichotomized moderator may overestimate the interaction effect. The findings of this research draw the attention of the hospitality and tourism research community on different representations of reflective constructs in their measurement and statistical models.

Protective Effects of Healthy Lifestyle Scores Against the Risk of Cardiovascular Diseases: A Representative Cohort Study from Taiwan

Background: The protective effect of different healthy lifestyle scores against a cardiovascular disease (CVD) risk has been reported previously although a comparison of the predictive performance is lacking. We compared the performance measures of CVD risk from different healthy lifestyle scores among adult Taiwanese participants. Methods: We conducted a nationwide prospective cohort study of 6042 participants (median age 43 years, 50.2% women) in Taiwan’s Hypertensive, Hyperglycemia and Hyperlipidemia Survey, 2002, who did not have CVD at the baseline evaluation. The simple and weighted Taiwan healthy lifestyle scores were defined as a combination of normal body mass index, Mediterranean diet, adequately physical activity, not smoking and regular healthy drinking, and each dichotomous lifestyle factor. The World Cancer Research Fund (WCRF) and the American Institute for Cancer Research (AICR) have recommended lifestyle interventions and adherence to Life's Simple 7, and participants were assigned to four groups based on lifestyle scores to evaluate the incidence of cardiovascular events. Results: During a median 14.3 years follow-up period, 520 participants developed CVD events. In multivariate-adjusted Cox proportional hazard models, adherence to the highest category, compared with the lowest one, on the simple and weighted Taiwan lifestyle score showed hazard ratios of 0.43 and 0.44 (95% confidence interval [CI] 0.2–0.94 and 0.28–0.68, respectively) were significant. Age was a significant effect modifier for the protective effect of healthy lifestyle scores on CVD risk. When the simple and weighted Taiwan healthy lifestyle scores were added to the classical model, the Harrell C-statistic increased from 0.85 to 0.86 (95% CI 0.84–087; Pdiff=0.02) in both lifestyle scores. The performance measures by integrated discriminative improvement showed a significant increase after the addition of the simple and weighted Taiwan healthy lifestyle score (integrated discriminative improvement: 0.51, 95% CI 0.16–0.86, p=0.002, and 0.38, 95% CI 0.01,0.74, p=0.021, respectively). Conclusions: Healthy lifestyle scores had an inverse association with CVD risk, and the reduction of CVD risk was greater in young adults than in the older adults. Further investigations to study the mechanism of the role of lifestyle interventions in CVD prevention are warranted. Funding Statement: The authors have no funding to report. Declaration of Interests: All other authors have no potential conflicts of interest relevant to this article in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Ethics Approval Statement: Informed consent was obtained from each participant. The protocol was reviewed and approved by the Research Ethics Committee of National Taiwan University Hospital.

ПРОФЕССИОНАЛЬНЫЕ ТИПЫ И ЦЕННОСТНЫЕ ОРИЕНТАЦИИ ЛИЧНОСТИ

The basic method is applied to the study of the typology of personal values. A basis is a complete, ordered, measurable set of elements. In our case, the elements are personality types. Completeness is determined by a factorial dichotomous typological set. Orderliness is determined by a single dichotomous multi-dimensional basic word structure of a set of types (a single dichotomous multi-dimensional-typological basis or «word basis»). The measurability of types is associated with a single dichotomous numerical multidimensional basic evaluation symmetric scale or with a «numerical basis». Completeness is related to the traditional dichotomous factor analysis of typological information (it resembles the factor strategy of Kettell in his typology of personality qualities). Orderliness sets a dichotomous basic multidimensionality in verbal typological information. Measurability makes it possible to create basic multidimensional methods based on a dichotomous symmetric numerical scale estimation of types. Multidimensional measuring typological basis of personality is a combination of several one-dimensional measuring typological bases with a single personal dichotomy «social-individual». In typology of information about the values of the personality is possible to allocate three levels of consistency. The first level of consistency corresponds to the traditional chaotic-eclectic dichotomous factor representation of information about personal values. The second level of consistency corresponds to functional structural coherence in the verbal dichotomous multidimensional basic regulation of the types of values of the individual. The third level of consistency corresponds to the project-dimensional dichotomous multidimensional basic implementation of the estimated-numerical measurement of the types of values of the individual. This research strategy is related to all three levels of systemicity, it is the basis of our dichotomous factor multidimensional base measurement typological concept, and it defines the strategy and stages of this research. In this article, the typology of personal values is checked for «basicity». For this purpose, based on the principle of semantic proximity, one-to-one correspondence («typological isomorphism») is established between the typology of personal values and professional personality types according to D. Holland. The article uses the method of «basic superimposition» of these two typologies, which provides a semantic community of these two typologies and generates such a psychological concept as «professional values».

753P - Prognostic and predictive value of baseline alpha-fetoprotein (AFP) in patients with advanced hepatocellular carcinoma (HCC) treated with ramucirumab from two phase III studies (REACH, REACH-2)

753P - Prognostic and predictive value of baseline alpha-fetoprotein (AFP) in patients with advanced hepatocellular carcinoma (HCC) treated with ramucirumab from two phase III studies (REACH, REACH-2)

Population kinetics of progression free survival (PFS).

e18251 Background: We assessed drug type impact on whether PFS curves could be fit by 2 phase decay models on nonlinear regression analysis (NLRA). Methods: We digitized 894 published PFS curves for incurable cancers. We used GraphPad Prism 7 for 1 phase and 2 phase decay NLRA, with constraints Y0 = 100 and plateau = 0. We defined curves as fitting 2 phase models if each subpopulation was ≥1% of the entire population and if subpopulation half-lives differed by a factor of ≥2, or if log-linear plots demonstrated unequivocal 2 phase decay. Results: PFS curves for single agents showed either high (≥75%) or low ( &lt; 30%) probability of 2 phase decay, depending on drug type (p &lt; 0.0001, Table). 11/11 PD1/ipilimumab combinations had 2 phase decay vs 36/209 curves (17%) for all other combinations. Conclusions: Drugs have either high or low probability of PFS curve 2 phase decay. Clinical trial methods or some mechanisms of acquired resistance might contribute to 2 phase decay, but 2 phase decay also could indicate a dichotomous factor (eg gene mutation/deletion or complete pathway silencing) producing 2 distinct subpopulations with differing progression rates. Drugs with high 2 phase decay could be prime candidates for RNA &amp; whole genome sequencing, pathway expression studies etc to identify dichotomous predictive factors. Further assessment is needed to better understand why some drugs behave differently when given in combinations vs as single agents. [Table: see text]

Abstract 1774: Progression-free survival curves suggest a dichotomous determinant of PD-L1 inhibitor efficacy

Abstract Background: PD-L1 expression varies across tumors but does not accurately predict PD-L1 inhibitor efficacy. Some negative tumors respond and some positive tumors fail. PD-L1 inhibitor progression-free survival (PFS) curve shape in non-small cell lung cancer (NSCLC) suggests that a dichotomous (present vs absent) factor might drive sensitivity rather than it being driven by a continuous variable like PD-L1 expression. PFS curves may follow first order kinetics, with a straight line if log % PFS is plotted vs time. If the population had 2 distinct subgroups with differing rates of progression then one would expect an inflection point on the log-linear curve, and the curve would fit a 2-phase decay model in nonlinear regression analysis (NLRA). A more homogeneous population would not fit a 2-phase model. Methods: We used arohatgi.info/WebPlotDigitizer/app/ to digitize published PFS curves, then GraphPad Prism5 for 2-phase NLRA, with the constraints Y0=100, plateau=0. To generate standardized 2-phase curves, we utilized 1) unselected NSCLC patients treated with epidermal growth factor receptor tyrosine kinase inhibitors (EGFR TKIs) where we expected a high proportion of curves to fit a 2-phase model as only some patients would have a sensitizing EGFR mutation; 2) NSCLC EGFR mutant and wild type (WT) patients treated with EGFR TKIs, and patients treated with platinum-based chemotherapy, taxanes or placebo/best supportive care (BSC), where we expected a lower proportion of curves to fit 2-phase models; 3) PD-L1 PFS curves in NSCLC and other tumor types. Results: With EGFR TKIs in unselected patients, 58 of 79 (73%) curves were fit by 2-phase models, vs 5 of 37 (14%) with EGFR TKIs in EGFR mutant patients (p&amp;lt;0.0001), 13 of 27 (48%) in EGFR WT patients (p&amp;lt;0.02), 4 of 37 (11%) with platinum-based therapy (p&amp;lt;0.0001), 15 of 47 (32%) with a taxane (p&amp;lt;0.0001), and 6 of 22 (27%) with placebo/BSC (p=0.0001). With PD-L1 inhibitors in NSCLC, 30 of 32 (94%) curves fit 2-phase models (p&amp;lt;0.0001 vs each of EGFR TKIs in EGFR mutants, EGFR WTs, platinum and taxane chemotherapy and placebo/BSC). In other tumor types, 27 of 32 (84%) PD-L1 curves fit 2-phase models. Conclusions: Most PD-L1 inhibitor PFS curves fit a 2-phase model. This is similar to what we observed with EGFR TKIs in unselected patients and different from EGFR TKIs in EGFR mutant and WT patients, and from chemotherapy or placebo/BSC. This leads us to hypothesize the existence of a dichotomous (present vs absent) factor such as a gene mutation, deletion or silencing that sensitizes tumors to PD-1/PD-L1 inhibitors. If found, such a dichotomous factor could prove to be a highly useful biomarker that could permit accurate prediction of PD-1/PD-L1 inhibitor efficacy. Since PD-1/PD-L1 inhibitor efficacy is higher in tumors with high PD-L1 expression, any sensitizing dichotomous factor might also drive PD-L1 expression. Citation Format: David J. Stewart, Dominick Bosse, Stephanie Brule, Andrew G. Robinson, Michael Ong, John F. Hilton. Progression-free survival curves suggest a dichotomous determinant of PD-L1 inhibitor efficacy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 1774. doi:10.1158/1538-7445.AM2017-1774

P3.02c-073 Evidence Suggesting a Dichotomous “Present vs absent” Determinant of PDL1 Inhibitor Efficacy in Non-Small Cell Lung Cancer (NSCLC): Topic: IT Biomarkers

NSCLC survival and progression-free survival (PFS) curves often follow first-order kinetics (Stewart, Lung Cancer 2011;71:217). We hypothesized that the number of curve decay phases reflects number of biologically distinct subpopulations with distinct rates of progression or death that are determined by dichotomous (present vs absent) factors. In NSCLC, some PDL1-negative patients respond to PDL1 inhibitors, while some PDL1 strongly-positive patients do not. We used “arohatgi.info/WebPlotDigitizer/app/” to digitize published NSCLC PDL1-inhibitor PFS curves and used GraphPad Prism5 for nonlinear regression analyses (one-phase and two-phase decay; constraints:Y0=100, plateau=0). 26 of 28 curves fit 2-phase decay models with R2>0.90, with distinct “fast progression” (FP) and “slow progression” (SP) subgroups. In studies with PFS curves for different PDL1-expression groups, patients with higher PDL1 tended to have a larger SP subgroup, although some with low PDL1 had slow progression and some with high PDL1 had fast progression (see Table). 1) PFS-curve nonlinear regression analysis identified 2 distinct subgroups (FP/SP) with differing degrees of benefit from PDL1 inhibitors. 2) Some PDL1-low patients had SP while some PDL1-high patients had FP, although there was a trend to more SP with PDL1-high than PDL1-low. 3) The observation of 2 distinct subpopulations leads us to hypothesize that there is a dichotomous variable (eg, gene mutation, deletion, amplification or silencing) driving PDL1 inhibitor benefit. The trend to higher benefit with high PDL1 expression, but inexact prediction of benefit by PDL1 expression is similar to the published trend seen with response vs PDL1 expression, and suggests that this dichotomous variable is linked to but distinct from PDL1 expression. 4) The published observation that high mutation burden is associated with PDL1 inhibitor benefit suggests that high mutation burden increases the probability of presence of a putative dichotomous favorable factor. We observed similar outcomes in other tumor types.

Reading English with Japanese in mind: Effects of frequency, phonology, and meaning in different-script bilinguals

Previous priming studies suggest that, even for bilinguals of languages with different scripts, non-selective lexical activation arises. This lexical decision eye-tracking study examined contributions of frequency, phonology, and meaning of L1 Japanese words on L2 English word lexical decision processes, using mixed-effects regression modeling. The response times and eye fixation durations of late bilinguals were co-determined by L1 Japanese word frequency and cross-language phonological and semantic similarities, but not by a dichotomous factor encoding cognate status. These effects were not observed for native monolingual readers and were confirmed to be genuine bilingual effects. The results are discussed based on the Bilingual Interactive Activation model (BIA+, Dijkstra &amp; Van Heuven, 2002) under the straightforward assumption that English letter units do not project onto Japanese word units.

Circulating Plasma Cells By Routine Complete Blood Count Identify Patients With Similar Outcome As Plasma Cell Leukemia

IntroductionPlasma cell leukemia (PCL) is a plasma cell dyscrasia that predicts for a shortened survival. World Health Organization (WHO) criteria require the presence of ≥20% circulating plasma cells (CPC) in the peripheral blood, and an absolute plasma cell count ≥2,000/µl (microliter). These criteria for PCL were established by Noel and Kyle in 1974, but subsequent studies to further validate this cut off are lacking. Since the presence of any plasma cells in the peripheral blood is now known to be a poor-risk feature, we sought to compare the overall survival of patients (pts) with any CPC observed on routine complete blood count (CBC) and differential with those that meet the WHO criteria of PCL. MethodsWe evaluated pts that received hyper-CVAD (cyclophosphamide /vincristine/doxorubicin/dex); CBAD (cyclophosphamide/bortezomib/doxorubicin/ dex), and DT-PACE (dexamethasone/thalidomide/ cisplatin /etoposide / doxorubicin/ cyclophosphamide) chemotherapy at The University of Texas M. D. Anderson Cancer Center for multiple myeloma from 2003-2012. We included pts with circulating blasts, plasma cells, or plasmacytoid cells regardless of the percentage (%) and absolute count. Pts were excluded if blasts or plasma cells were seen transiently in the setting of leukopheresis, growth factor support, or during severe sepsis. Overall survival (OS) was defined from the first detection of CPC until death due to any cause. We treated the % of plasma cells in the CBC as a continuous variable in the range from 1% to 20%. Pts with plasma cells below the cut off were defined as the “low” group and their survival was compared to the pts with plasma cells above the cut off that was defined as the “high” group. We conducted a log rank test for difference in survival curves between the two groups for each cut off, and the results were verified with a Cox model and both results were identical. Results85 pts were identified with presence of any CPC or PCL diagnosed based on WHO criteria; 19 had primary PCL (pPCL) and 66 pts had secondary PCL (sPCL). At the time of diagnosis, 83 (2 with unknown values) pts had a median CPC of 4% and median absolute plasma cells of 178/µL. 73 pts had cytogenetic abnormalities by florescence in situ hybridization studies, including deletion/monosomy 13 in 32 pts, IgH gene rearrangement in 24 pts and TP53 deletions in 12 pts. 17/19 pts with pPCL received bortezomib based therapy and 14/19 underwent stem cell transplant.26 pts met the WHO criteria of PCL at some point during their disease, including 11 who met criteria when plasma cells were first detected, and 15 who had CPC < 20% initially, but subsequently met WHO criteria. Of the remaining 57 pts that did not meet the WHO criteria for presence of both absolute plasma cell count and % plasma cells, 13 met one of the criteria during the course of their disease. 44 pts had CPC but did not meet either of the WHO criteria.Survival data were available for 79/85 pts. Pts with pPCL (either by CPC or met at least one of the WHO criteria) had a median OS of 18.5 months. pPCL pts who only had CPC and did not meet either of the WHO criteria (n=5) had a median OS of 19 months; pPCL patients who met either or both of the WHO criteria (n=13) had a median OS of 18 months. Pts with sPCL (either by CPC or met at least one of the WHO criteria) had a median OS of 5 months (n=61). sPCL pts who only had CPC and did not meet either of the WHO criteria (n=39) had a median OS of 5 months; sPCL patients who met either or both of the WHO criteria (n=22) had a median OS of 4 months.Based on evaluating plasma cell % as a dichotomous factor with cut-off ranging from 1% to 20%, there was no significant difference in overall survival based on specific levels of plasma cell percent, and verified with a Cox model, however with many cut-offs, the sample size was small potentially limiting analysis. ConclusionIn our analysis of OS of pts having plasma cells in their peripheral blood, we did not find any statistically significant difference based on their degree of % plasma cells or absolute plasma cells at the time of diagnosis. This supports the hypothesis that, irrespective of quantity, the presence of any plasma cells in the CBC is a poor prognostic indicator, with a similar natural history and a uniformly poor OS. While these findings will need to be validated in a larger, independent dataset, we would propose that these initial data support redefining the diagnostic criteria of PCL to include any pt with CPC observed in a routine CBC. Disclosures:No relevant conflicts of interest to declare.

Impact of misspecifying the distribution of a prognostic factor on power and sample size for testing treatment interactions in clinical trials.

BackgroundInteraction in clinical trials presents challenges for design and appropriate sample size estimation. Here we considered interaction between treatment assignment and a dichotomous prognostic factor with a continuous outcome. Our objectives were to describe differences in power and sample size requirements across alternative distributions of a prognostic factor and magnitudes of the interaction effect, describe the effect of misspecification of the distribution of the prognostic factor on the power to detect an interaction effect, and discuss and compare three methods of handling the misspecification of the prognostic factor distribution.MethodsWe examined the impact of the distribution of the dichotomous prognostic factor on power and sample size for the interaction effect using traditional one-stage sample size calculation. We varied the magnitude of the interaction effect, the distribution of the prognostic factor, and the magnitude and direction of the misspecification of the distribution of the prognostic factor. We compared quota sampling, modified quota sampling, and sample size re-estimation using conditional power as three strategies for ensuring adequate power and type I error in the presence of a misspecification of the prognostic factor distribution.ResultsThe sample size required to detect an interaction effect with 80% power increases as the distribution of the prognostic factor becomes less balanced. Misspecification such that the actual distribution of the prognostic factor was more skewed than planned led to a decrease in power with the greatest loss in power seen as the distribution of the prognostic factor became less balanced. Quota sampling was able to maintain the empirical power at 80% and the empirical type I error at 5%. The performance of the modified quota sampling procedure was related to the percentage of trials switching the quota sampling scheme. Sample size re-estimation using conditional power was able to improve the empirical power under negative misspecifications (i.e. skewed distributions) but it was not able to reach the target of 80% in all situations.ConclusionsMisspecifying the distribution of a dichotomous prognostic factor can greatly impact power to detect an interaction effect. Modified quota sampling and sample size re-estimation using conditional power improve the power when the distribution of the prognostic factor is misspecified. Quota sampling is simple and can prevent misspecification of the prognostic factor, while maintaining power and type I error.

A Monte Carlo study comparing PIV, ULS and DWLS in the estimation of dichotomous confirmatory factor analysis

We conducted a Monte Carlo study to investigate the performance of the polychoric instrumental variable estimator (PIV) in comparison to unweighted least squares (ULS) and diagonally weighted least squares (DWLS) in the estimation of a confirmatory factor analysis model with dichotomous indicators. The simulation involved 144 conditions (1,000 replications per condition) that were defined by a combination of (a) two types of latent factor models, (b) four sample sizes (100, 250, 500, 1,000), (c) three factor loadings (low, moderate, strong), (d) three levels of non-normality (normal, moderately, and extremely non-normal), and (e) whether the factor model was correctly specified or misspecified. The results showed that when the model was correctly specified, PIV produced estimates that were as accurate as ULS and DWLS. Furthermore, the simulation showed that PIV was more robust to structural misspecifications than ULS and DWLS.

Modeling repeated measures of dichotomous data: Testing whether the within-person trajectory of change varies across levels of between-person factors

In this paper, we consider the following question for the analysis of data obtained in longitudinal panel designs: How should repeated-measures data be modeled and interpreted when the outcome or dependent variable is dichotomous and the objective is to determine whether the within-person rate of change over time varies across levels of one or more between-person factors? Standard approaches address this issue by means of generalized estimating equations or generalized linear mixed models with logistic links. Using an empirical example and simulated data, we show (1) that cross-level product terms from these models can produce misleading results with respect to whether the within-person rate of change varies across levels of a dichotomous between-person factor; and (2) that subgroup differences in the rate of change should be assessed on an additive scale (using group differences in the effects of predictors on the probability of disease) rather than on a multiplicative scale (using group differences in the effects of predictors on the odds of disease). Because usual approaches do not provide a significance test for whether the rate of additive change varies across levels of a between-person factor, sample differences in the rate of additive change may be due to sampling error. We illustrate how standard software can be used to estimate and test whether additive changes vary across levels of a between-person factor.

The effect of reduction on the processing of flaps and /g/ in isolated words

This study investigates the comprehension of words with reduced and unreduced flaps and /g/s by means of an auditory lexical decision task. The experimental stimuli are compared to words produced in various speech styles confirming their naturalness and illustrating the distribution of the experimental stimuli. The experiment shows that reduced words are more difficult to process than unreduced words, and that listeners' response latencies for unreduced words are longest for the highest and lowest frequencies. Listeners' error rates are highest for the lowest frequency items. A comparison of the flap stimuli to the /g/ stimuli reveals no difference in processing between the two phonemes. When an acoustic measure (change in intensity) is used as a predictor, instead of a dichotomous factor contrasting reduced with unreduced, items on the extreme ends of the acoustic measure (very reduced or very unreduced) have longer response latencies than the middle items. The results of the second analysis supports the view that listeners optimize comprehension for intermediate degrees of reduction. The data also suggests that production frequency, or predictability (e.g., the likelihood of a reduced flap) affects the comprehension of reduced and unreduced words, such that highly predictable productions have shorter response latencies.

Further development of YBOCS dimensions in the OCD Collaborative Genetics Study: Symptoms vs. categories

Despite progress in identifying homogeneous subphenotypes of obsessive-compulsive disorder (OCD) through factor analysis of the Yale Brown Obsessive-Compulsive Scale Symptom Checklist (YBOCS-SC), prior solutions have been limited by a reliance on presupposed symptom categories rather than discrete symptoms. Furthermore, there have been few attempts to evaluate the familiality of OCD symptom dimensions. The purpose of this study was to extend prior work by this collaborative group in category-based dimensions by conducting the first-ever exploratory dichotomous factor analysis using individual OCD symptoms, comparing these results to a refined category-level solution, and testing the familiality of derived factors. Participants were 485 adults in the six-site OCD Collaborative Genetics Study, diagnosed with lifetime OCD using semi-structured interviews. YBOCS-SC data were factor analyzed at both the individual item and symptom category levels. Factor score intraclass correlations were calculated using a subsample of 145 independent affected sib pairs. The item- and category-level factor analyses yielded nearly identical 5-factor solutions. While significant sib–sib associations were found for four of the five factors, Hoarding and Taboo Thoughts were the most robustly familial ( r ICC ≥ 0.2). This report presents considerable converging evidence for a five-factor structural model of OCD symptoms, including separate factor analyses employing individual symptoms and symptom categories, as well as sibling concordance. The results support investigation of this multidimensional model in OCD genetic linkage studies.

The relationship between the five-factor model and latent Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition personality disorder dimensions

This study compared the latent structure of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition personality disorders (PDs) with the 5-factor model (FFM) of general personality dimensions. The subjects in the study were 742 community-residing individuals who participated in the Hopkins Epidemiology of Personality Disorders Study. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition PD traits were assessed by psychologists using the International Personality Disorder Examination, and PD dimensions were derived previously using dichotomous factor analysis. The Revised NEO Personality Inventory, a measure of the FFM, was administered to all subjects. The relationship between the 2 sets of personality-related constructs was examined using a construct validity framework and also using Pearson correlation coefficients, multiple linear regression models, and spline regression models. The 5 PD factors each exhibited small to moderate correlations with several NEO dimensions; together, the NEO domain and facet scores explained a fifth to a third of the variance in PD dimensions. Examples of nonlinear relationships between the personality dimensions were identified. There is a modest correspondence between the PD dimensions and FFM traits, and the traits of FFM only partially explain the variance of the PDs. Dimensional measures of general personality may be a suitable alternative to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. Whether additional maladaptive traits would better define the domain of PDs remains an important objective for future research.

Evaluation of diagnostic criteria for panic attack using item response theory: Findings from the National Comorbidity Survey in USA

Background The dichotomous diagnostic systems such as the Diagnostic and Statistical Manual of Mental Disorders (DSM) and International Classification of Diseases (ICD) lose much important information concerning what each symptom can offer. This study explored the characteristics and performances of DSM-IV and ICD-10 diagnostic criteria items for panic attack using modern item response theory (IRT). Methods The National Comorbidity Survey used the Composite International Diagnostic Interview to assess 14 DSM-IV and ICD-10 panic attack diagnostic criteria items in the general population in the USA. The dimensionality and measurement properties of these items were evaluated using dichotomous factor analysis and the two-parameter IRT model. Results A total of 1213 respondents reported at least one subsyndromal or syndromal panic attack in their lifetime. Factor analysis indicated that all items constitute a unidimensional construct. The two-parameter IRT model produced meaningful and interpretable results. Among items with high discrimination parameters, the difficulty parameter for “palpitation” was relatively low, while those for “choking,” “fear of dying” and “paresthesia” were relatively high. Several items including “dry mouth” and “fear of losing control” had low discrimination parameters. Limitations The item characteristics of diagnostic criteria among help-seeking clinical populations may be different from those that we observed in the general population and deserve further examination. Conclusions “Paresthesia,” “choking” and “fear of dying” can be thought to be good indicators of severe panic attacks, while “palpitation” can discriminate well between cases and non-cases at low level of panic attack severity. Items such as “dry mouth” would contribute less to the discrimination.

A dynamic block-randomization algorithm for group-randomized clinical trials when the composition of blocking factors is not known in advance

We present an algorithm for randomizing units in blocks for controlled trials when the composition of blocking factors is not known in advance. For example, suppose the desired goal of an intervention study is to randomize units to one of two interventions while blocking on a dichotomous factor (e.g., gender), but the total number of units, and therefore number or composition, of males and females among those units assembled for randomization cannot be determined in advance. This situation arises in randomized trials when subjects are scheduled to be randomized as a group, but not all of the subjects show up for the visit. Since investigators do not know which of the scheduled subjects will or will not show up, a dynamic randomization scheme is required to accommodate the unknown composition of the blocking factor once a group of subjects (units) is assembled for randomization. These settings are further complicated when there is more than one blocking factor. In this paper, we present an algorithm that ensures the integrity of the randomization process in these settings.

A Computer Program to Estimate Power and Relative Efficiency of Flexibly Matched Case-control Studies

We recently introduced the concept of flexible matching strategies with varying proportions of a dichotomous matching factor among controls to increase power and efficiency of case-control studies. We now present a method and a computer program to calculate power and relative efficiency compared to an unmatched design varying the proportion of the matching factor in controls over all possible values from 0 to 100 percent. For all these values, the program calculates the expected variance of the combined Mantel-Haenszel odds ratio and determines the power using the standard error of the expected combined Mantel-Haenszel odds ratio under the null hypothesis as derived from the Mantel-Haenszel test statistic without continuity correction. Thereby, the program allows estimating the optimal prevalence of the matching factor in selected controls for a given scenario which often differs from the prevalence in cases. It furthermore allows to estimate loss in power and efficiency compared to optimal matching by suboptimal matching. Estimations like these are helpful with respect to the decision when to stop efforts to optimize the degree of matching during the recruitment of controls. Our program will strongly facilitate assessing the benefits of flexible matching strategies.