Continuation Ratio Model Research Articles

Forward and Backward Continuation Ratio Models for Ordinal Response Variables

There are different types of continuation ratio (CR) models for ordinal response variables. The different model equations, corresponding parameterizations, and nonequivalent results are confusing. The purpose of this study is to introduce different types of forward and backward CR models, demonstrate how to implement these models using Stata, and compare the results using data from the Educational Longitudinal Study of 2002 (ELS:2002).

Identifying Microvascular and Neural Parameters Related to the Severity of Diabetic Retinopathy Using Optical Coherence Tomography Angiography.

PurposeTo identify microvascular and neural parameters related to the severity of diabetic retinopathy (DR) by using optical coherence tomography angiography in patients with type 2 diabetes.MethodsThis cross-sectional study included 110 eyes (63 patients) with no DR, 46 eyes (33 patients) with mild nonproliferative DR, 36 eyes (23 patients) with moderate nonproliferative DR, 36 eyes (22 patients) with severe nonproliferative DR, and 31 eyes (19 patients) with proliferative DR. The optical coherence tomography angiography images were processed to quantify the foveal avascular zone parameters, macular vessel density (VD), retinal thickness, peripapillary VD, retinal nerve fiber layer thickness, and ganglion cell complex thickness. A LASSO-based continuation ratio model was used to select the most clinically relevant parameters for predicting the stage of DR.ResultsThe regression model identified a set of regional parameters for each scanning pattern that identified the DR severity, including foveal avascular zone perimeter; FD-300; temporal perifoveal superficial capillary plexus VD and retinal thickness; temporal and nasal parafoveal deep capillary plexus VD; peripapillary VD in the temporal superior, nasal inferior, and temporal inferior sectors; temporal superior and nasal inferior retinal nerve fiber layer thickness; ganglion cell complex thickness; and FLV, which changed significantly with the progression of DR. Furthermore, two combined blocks exhibited different sensitive parameters to differentiate between the groups based on DR severity. Similar results were obtained in eyes without diabetic macular edema.ConclusionsWe identified microvascular and neural parameters related to the severity of DR using optical coherence tomography angiography, suggesting their potential clinical application for better screening and staging of DR.

Illicit opioid use following changes in opioids prescribed for chronic non-cancer pain

BackgroundAfter decades of increased opioid pain reliever prescribing, providers are rapidly reducing prescribing. We hypothesized that reduced access to prescribed opioid pain relievers among patients previously reliant upon opioid pain relievers would result in increased illicit opioid use.Methods and findingsWe conducted a retrospective cohort study among 602 publicly insured primary care patients who had been prescribed opioids for chronic non-cancer pain for at least three consecutive months in San Francisco, recruited through convenience sampling. We conducted a historical reconstruction interview and medical chart abstraction focused on illicit substance use and opioid pain reliever prescriptions, respectively, from 2012 through the interview date in 2017–2018. We used a nested-cohort design, in which patients were classified, based on opioid pain reliever dose change, into a series of nested cohorts starting with each follow-up quarter. Using continuation-ratio models, we estimated associations between opioid prescription discontinuation or 30% increase or decrease in dose, relative to no change, and subsequent frequency of heroin and non-prescribed opioid pain reliever use, separately. Models controlled for demographics, clinical and behavioral characteristics, and past use of heroin or non-prescribed opioid pain relievers. A total of 56,372 and 56,484 participant-quarter observations were included from the 597 and 598 participants available for analyses of heroin and non-prescribed opioid pain reliever outcomes, respectively. Participants discontinued from prescribed opioids were more likely to use heroin (Adjusted Odds Ratio (AOR) = 1.57, 95% CI: 1.25–1.97) and non-prescribed opioid pain relievers (AOR = 1.75, 1.45–2.11) more frequently in subsequent quarters compared to participants with unchanged opioid prescriptions. Participants whose opioid pain reliever dose increased were more likely to use heroin more frequently (AOR = 1.67, 1.32–2.12). Results held throughout sensitivity analyses. The main limitations were the observational nature of results and limited generalizability beyond safety-net settings.ConclusionsDiscontinuation of prescribed opioid pain relievers was associated with more frequent non-prescribed opioid pain reliever and heroin use; increased dose was also associated with more frequent heroin use. Clinicians should be aware of these risks in determining pain management approaches.

Associations between Antiretrovirals and Cognitive Function in Women with HIV.

Cognitive complications persist in antiretroviral therapy(ART)-treated people with HIV. However, the pattern and severity of domain-specific cognitive performance is variable and may be exacerbated by ART-mediated neurotoxicity. 929 women with HIV(WWH) from the Women's Interagency HIV Study who were classified into subgroups based on sociodemographic and longitudinal behavioral and clinical data using semi-parametric latent class trajectory modelling. Five subgroups were comprised of: 1) well-controlled HIV with vascular comorbidities(n = 116); 2) profound HIV legacy effects(CD4 nadir <250 cells/μL; n = 275); 3) primarily <45year olds with hepatitis C(n = 165); 4) primarily 35-55year olds(n = 244), and 5) poorly-controlled HIV/substance use(n = 129). Within each subgroup, we fitted a constrained continuation ratio model via penalized maximum likelihood to examine adjusted associations between recent ART agents and cognition. Most drugs were not associated with cognition. However, among the few drugs, non-nucleoside reverse transcriptase inhibitor (NNRTIs) and protease inhibitors(PIs) were most commonly associated with cognition, followed by nucleoside reverse transcriptase inhibitors(NRTIs) and integrase inhibitors(IIs). Directionality of ART-cognition associations varied by subgroup. Better psychomotor speed and fluency were associated with ART for women with well-controlled HIV with vascular comorbidities. This pattern contrasts women with profound HIV legacy effects for whom poorer executive function and fluency were associated with ART. Motor function was associated with ART for younger WWH and primarily 35-55year olds. Memory was associated with ART only for women with poorly-controlled HIV/substance abuse. Findings demonstrate interindividual variability in ART-cognition associations among WWH and highlight the importance of considering sociodemographic, clinical, and behavioral factors as an underlying contributors to cognition. Are antiretroviral agents a risk factor for cognitive complications in women with HIV? We examind associations between ART-agents and cognitive function among similar subgroups of women with HIV from the Women's Interagency HIV study. The patterns of associations depended on sociodemographic, clinical, and behavioral characteristics of women.

Assessing Efficient Risk Ratios: An Application to Surgical Stage Prediction in Cervical Cancer

Background: Cervical cancer remains the second most commonly diagnosed cancer and the third leading cause of cancer death in developing countries. Improving clinicians’ knowledge and understanding of surgical staging is critical in the fight against the disease. However, a systematic evaluation of different ordinal regression models based on diverse predicted outcomes has not been given its due share in literature. Objective: To systematically assess the flexibility of odds ratios for three popular ordinal regression models i.e. the Multinomial Logistic (ML) model, the Continuation Ratio (CR) model and Adjacent Category Logistic (ACL) model when applying cervical cancer data in surgical stage prediction. Method: We systematically, compared the performance of CR, ML and the ACL as the predictive mechanisms, and evaluate the most appropriate model in the cervical cancer setting. The study considered women who visited the Oncology department at the Moi Teaching and Referral Hospital’s Chandaria Cancer and Chronic Diseases Center and were diagnosed and surgically treated for cervical cancer from January 2014 to December 2018. Results and Conclusion: We presented the comparison between 3 different regression models for ordinal data within the cervical cancer setting. We found that the CR model without proportional odds yielded better results comparing Akaike Information Criterion (AIC), log likelihood ratio and residual deviance. In addition, the key prognostic factor associated with invasive cervical cancer was the (International Federation of Gynecology and Obstetrics) FIGO clinical stage which in particular, had a higher influence on the surgical Stage 2 outcomes compared to the lesser surgical stage categories. All the 5 independent features selected for classifying the patients into surgical stages were the FIGO clinical stage and partly, the presence or absence of symptomatic vaginal discharge.

Non-HDL cholesterol level and depression among Canadian elderly—a cross-sectional analysis of the baseline data from the CLSA

To explore whether non-high-density-lipoprotein cholesterol (non-HDL-c) is associated with depression, a total of 26 819 Canadians aged 45–85 from the Canadian Longitudinal Study on Aging (CLSA) were included in analysis. Non-HDL-c, the difference between total-c and HDL-c, was categorized into five levels, i.e., &lt;2.6, 2.6 to &lt;3.7, 3.7 to &lt;4.8, 4.8 to 5.7, and ≥5.7 mmol/L. History of clinical depression was collected by questionnaire at an in-home interview, and current potential depression status was determined by CES-D10 (Center for Epidemiological Studies Depression Scale 10 questions version) score, i.e., ≥10 vs. &lt;10. Logistic continuation ratio model for ordinal data was used to estimate the odds of being at or above a higher non-HDL-c category for depression status. Compared with those without clinical depression history and currently undepressed, the adjusted odds ratios (95% CI) were 1.09 (1.02, 1.17) for those without clinical depression history but currently depressed, 1.05 (0.98, 1.12) for those had clinical depression history but currently undepressed, and 1.21 (1.10, 1.32) for those had clinical depression history and currently depressed. The average of non-HDL-c for four depression groups were 3.64, 3.71, 3.69, and 3.82 mmol/L, respectively, and group 4 was statistically higher than others ( p &lt; 0.001). In conclusion, people with both current depression and a history clinical depression are at an increased risk of having high level of non-HDL-c.

Multi-factorial age estimation: A Bayesian approach combining dental and skeletal magnetic resonance imaging

PurposeTo study age estimation performance of combined magnetic resonance imaging (MRI) data of all four third molars, the left wrist and both clavicles in a reference population of females and males. To study the value of adding anthropometric and sexual maturation data. Materials and methodsThree Tesla MRI of the three anatomical sites was prospectively conducted from March 2012 to May 2017 in 14- to 26-year-old healthy Caucasian volunteers (160 females, 138 males). Development was assessed by allocating stages, anthropometric measurements were taken, and self-reported sexual maturation data were collected. All data was incorporated in a continuation-ratio model to estimate age, applying Bayes’ rule to calculate point and interval predictions. Two performance aspects were studied: (1) accuracy and uncertainty of the point prediction, and (2) diagnostic ability to discern minors from adults (≥18 years). ResultsCombining information from different anatomical sites decreased the mean absolute error (MAE) compared to incorporating only one site (P<0.0001). By contrast, adding anthropometric and sexual maturation data did not further improve MAE (P=0.11). In females, combining all three anatomical sites rendered a MAE equal to 1.41 years, a mean width of the 95% prediction intervals of 5.91 years, 93% correctly classified adults and 91% correctly classified minors. In males, the corresponding results were 1.36 years, 5.49 years, 94%, and 90%, respectively. ConclusionAll aspects of age estimation improve when multi-factorial MRI data of the three anatomical sites are incorporated. Anthropometric and sexual maturation data do not seem to add relevant information.

Staging Clavicular Development on MRI: Pitfalls and Suggestions for Age Estimation

MRI of the clavicle's sternal end has been studied for age estimation. Several pitfalls have been noted, but how they affect age estimation performance remains unclear. To further study these pitfalls and to make suggestions for a proper use of clavicle MRI for forensic age estimation. Our hypotheses were that age estimation would benefit from 1) discarding stages 1 and 4/5; 2) including advanced substages 3aa, 3ab, and 3ac; 3) taking both clavicles into account; and 4) excluding morphological variants. Prospective cross-sectional. Healthy Caucasian volunteers between 11 and 30 years old (524; 277 females, 247 males). 3T, T1 -weighted gradient echo volumetric interpolated breath-hold examination (VIBE) MR-sequence. Four observers applied the most elaborate staging technique for long bone development that has been described in the current literature (including stages, substages, and advanced substages). One of the observers repeated a random selection of the assessments in 110 participants after a 2-week interval. Furthermore, all observers documented morphological variants. Weighted kappa quantified reproducibility of staging. Bayes' rule was applied for age estimation with a continuation ratio model for the distribution of the stages. According to the hypotheses, different models were tested. Mean absolute error (MAE) differences between models were compared, as were MAEs between cases with and without morphological variants. Weighted kappa equaled 0.82 for intraobserver and ranged between 0.60 and 0.64 for interobserver agreement. Stages 1 and 4/5 were allocated interchangeably in 4.3% (54/1258). Age increased steadily in advanced substages of stage 3, but improvement in age estimation was not significant (right P = 0.596; left P = 0.313). The model that included both clavicles and discarded stages 1 and 4/5 yielded an MAE of 1.97 years, a root mean squared error of 2.60 years, and 69% correctly classified minors. Morphological variants rendered significantly higher MAEs (right 3.84 years, P = 0.015; left 2.93 years, P = 0.022). Our results confirmed hypotheses 3) and 4), while hypotheses 1) and 2) remain to be investigated in larger studies. 1 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2020;51:377-388.

A four-group urine risk classifier for predicting outcomes in patients with prostate cancer.

ObjectivesTo develop a risk classifier using urine‐derived extracellular vesicle (EV)‐RNA capable of providing diagnostic information on disease status prior to biopsy, and prognostic information for men on active surveillance (AS).Patients and MethodsPost‐digital rectal examination urine‐derived EV‐RNA expression profiles (n = 535, multiple centres) were interrogated with a curated NanoString panel. A LASSO‐based continuation ratio model was built to generate four prostate urine risk (PUR) signatures for predicting the probability of normal tissue (PUR‐1), D'Amico low‐risk (PUR‐2), intermediate‐risk (PUR‐3), and high‐risk (PUR‐4) prostate cancer. This model was applied to a test cohort (n = 177) for diagnostic evaluation, and to an AS sub‐cohort (n = 87) for prognostic evaluation.ResultsEach PUR signature was significantly associated with its corresponding clinical category (P < 0.001). PUR‐4 status predicted the presence of clinically significant intermediate‐ or high‐risk disease (area under the curve = 0.77, 95% confidence interval [CI] 0.70–0.84). Application of PUR provided a net benefit over current clinical practice. In an AS sub‐cohort (n = 87), groups defined by PUR status and proportion of PUR‐4 had a significant association with time to progression (interquartile range hazard ratio [HR] 2.86, 95% CI 1.83–4.47; P < 0.001). PUR‐4, when used continuously, dichotomized patient groups with differential progression rates of 10% and 60% 5 years after urine collection (HR 8.23, 95% CI 3.26–20.81; P < 0.001).ConclusionUrine‐derived EV‐RNA can provide diagnostic information on aggressive prostate cancer prior to biopsy, and prognostic information for men on AS. PUR represents a new and versatile biomarker that could result in substantial alterations to current treatment of patients with prostate cancer.

Modeling the Effect of Antipsychotic Drugs on Body Weight of Psychiatric Patients

Various studies have confirmed the significance increase in weight between baselines and follow up after commencement of antipsychotic drugs. Both Atypical Antipsychotic Drugs (AAD) and Typical Antipsychotic Drugs (TAD) have shown the tendency of inducing increase in the fat content of the body. However the modeling of the exact size of the weight increase beyond probable interval estimate is a necessity for clearer understanding; that will foster easy control of weight gain in psychiatric patient. This study thus established the claim from previous studies and presents the odds of AAD inducing weight gain relative to TAD. The study used a composite trinomial and bernoulli distribution to describe the interest variables (IV’s) body mass index (BMI) and the drug type. The composite function parameters and its standard errors were obtained using maximum likelihood estimation (MLE) method and fisher approach respectively. The study further used the continuation ratio model to determine the exact weight gain proportion due to AAD relative to TAD. The study was carried out at the Out-patient clinic of the Federal Neuropsychiatric Hospital Aro, Abeokuta Nigeria, within a period of six weeks between baseline and follow up. A total of two hundred and seventy patients participated in the study after their consent and willingness were obtained. The difficulty to control the food intake of the patient at home is a limitation to the study. Patients that have had reasons to switch were excluded from the study. A graphical representation was used to explain the exact behavioural pattern of the weight gain and its relative effect due to AAD and TAD drugs. The study thus recommends a thorough follow up of healthy eating of food rich in fibers, fruits and vegetables along with indoor or outdoor physical exercises for psychiatric patients mostly on AAD to prevent future cardiovascular related diseases.

Gencrm: A New Command for Generalized Continuation-Ratio Models

A continuation-ratio model represents a variant of an ordered regression model that is suited to modeling processes that unfold in stages, particularly those in which a return to a previous stage is not possible (for example, educational attainment, job promotion, or disease progression). The parameters for covariates in continuation-ratio models may be constrained to be equal, vary by a set of common factors (that is, proportionality constraints), or freely vary across stages. Currently, there are three community-contributed commands that fit continuation-ratio models. Each of these commands fits some subset of continuation-ratio models involving parameter constraints, but none of them offer complete coverage of the range of possibilities. The new gencrm command expands the options for continuation-ratio models to include the possibility for some of or all the covariates to be constrained to be equal, to freely vary, or to vary by a set of common factors across stages. gencrm relies on Stata's maximum likelihood routines for estimation and avoids reshaping the data. gencrm includes options for three link functions (logit, probit, and cloglog) and supports Stata's multiple-imputation suites of commands.

Age estimation combining radiographic information of two dental and four skeletal predictors in children and subadults.

Improved age estimates may result from combining different age predictors. This study aimed to validate age estimation performances combining the radiographic development of teeth, cervical vertebrae, hand and wrist bones, skull, and mandible. The sampled consisted of 256 individuals aged between 4 and 20years. Bayes' rule with a multivariate continuation ratio model was applied for the distribution of the dental scores. The additional age information of the skeletal variables was assessed extending the dental model separately and combining the variables. The performances of all models were quantified and compared using mean error (ME), mean absolute error (MAE), and root mean squared error (RMSE). The best performance resulted combining all variables (ME - 0.04 for F and M; MAE 0.91 for F and 0.95 for M; RMSE 1.14 for F and 1.19 for M). To improve performances and minimize radiation doses, the combination of teeth and hand and wrist bones information is recommended.

Interregional inventor collaboration and the commercial value of patented inventions: evidence from the US biotechnology industry

This paper analyzes the relative importance of the spatiality of inventor collaboration links for the quality of invention. It uses patent citation and renewal data to estimate (expected) economic values for a sample of US biotechnology patents, controlling for a set of inventor and owner characteristics. The investigation extends the research by differentiating the quality of knowledge flows while at the same time refining patents as a source of innovation information. The results of the clustered logistic regression and continuation ratio models provide some support for the hypothesis that interregional (non-local) collaboration results in commercially more valuable inventions relative to patents that are product of intraregional (local) collaboration. The same empirical results also indicate that (1) inter-industry (related) knowledge flows contribute to the patent value more than inter-industry (unrelated) or intra-industry knowledge flows and (2) general-purpose technologies/products as measured by citations of patented inventions received from diverse sets of industries are indicative not only of technological/scientific quality, but also of the economic value of patented inventions. These results are robust in different citation-window designations, exclusion of self-citations, different regional definitions and different patent classifications.

Age, gender, and current living status were associated with perceived access to treatment among Canadians using a cross sectional survey

BackgroundAccess, particularly timely access, to care is the Canadian public’s most important healthcare concern. The drivers of perceived appropriateness of access to care among patients with at least one chronic health condition (CHC) are not, however, well defined. This study evaluated whether personal characteristics, self-reported health status and care received were associated with patients’ perception of effective access in managing a chronic illness.MethodsThe study population (n = 619) was drawn from a representative sample of the adult Canadian population who reported having ≥1 CHC in the 2013–2014 Health Care in Canada survey. Ordinal regression, with the continuation ratio model, was used to evaluate association of perceived level of access to treatment with socio-demographic factors, perceived health status and care utilization experience.ResultsFactors most closely associated with patients’ satisfaction with care access were: age, sex, current cohabitation, care affordability, and availability of support and information to help manage their CHCs. Individuals, particularly females, < 35 years, currently living alone, with poor access to professional support or information and who feel affordability of care has worsened over the past five years were more likely to report a poorer level of treatment access.ConclusionsIndividuals living alone, who are younger, and women may be especially susceptible to lower perceived access to care of CHCs and a sense of pessimism about things not getting better. Further evaluation of the reasons behind these findings may help develop effective strategies to assist these populations to access the care they need.

Three-part joint modeling methods for complex functional data mixed with zero-and-one-inflated proportions and zero-inflated continuous outcomes with skewness.

We take a functional data approach to longitudinal studies with complex bivariate outcomes. This work is motivated by data from a physical activity study that measured 2 responses over time in 5-minute intervals. One response is the proportion of time active in each interval, a continuous proportions with excess zeros and ones. The other response, energy expenditure rate in the interval, is a continuous variable with excess zeros and skewness. This outcome is complex because there are 3 possible activity patterns in each interval (inactive, partially active, and completely active), and those patterns, which are observed, induce both nonrandom and random associations between the responses. More specifically, the inactive pattern requires a zero value in both the proportion for active behavior and the energy expenditure rate; a partially active pattern means that the proportion of activity is strictly between zero and one and that the energy expenditure rate is greater than zero and likely to be moderate, and the completely active pattern means that the proportion of activity is exactly one, and the energy expenditure rate is greater than zero and likely to be higher. To address these challenges, we propose a 3-part functional data joint modeling approach. The first part is a continuation-ratio model to reorder the ordinal valued 3 activity patterns. The second part models the proportions when they are in interval (0,1). The last component specifies the skewed continuous energy expenditure rate with Box-Cox transformations when they are greater than zero. In this 3-part model, the regression structures are specified as smooth curves measured at various time points with random effects that have a correlation structure. The smoothed random curves for each variable are summarized using a few important principal components, and the association of the 3 longitudinal components is modeled through the association of the principal component scores. The difficulties in handling the ordinal and proportional variables are addressed using a quasi-likelihood type approximation. We develop an efficient algorithm to fit the model that also involves the selection of the number of principal components. The method is applied to physical activity data and is evaluated empirically by a simulation study.

How to Test for Goodness of Fit in Ordinal Logistic Regression Models

Ordinal regression models are used to describe the relationship between an ordered categorical response variable and one or more explanatory variables. Several ordinal logistic models are available in Stata, such as the proportional odds, adjacent-category, and constrained continuation-ratio models. In this article, we present a command (ologitgof) that calculates four goodness-of-fit tests for assessing the overall adequacy of these models. These tests include an ordinal version of the Hosmer–Lemeshow test, the Pulkstenis–Robinson chi-squared and deviance tests, and the Lipsitz likelihood-ratio test. Together, these tests can detect several different types of lack of fit, including wrongly specified continuous terms, omission of different types of interaction terms, and an unordered response variable.

Radiomics features of ground glass nodules tailored different pathological grades of lung adenocarcinoma.

e23127 Background: Different pathological subtypes as well as different grades of adenocarcinoma based on the IASLC/ATS/ERS classification had been proven to be stage-independent predictor of survival. Radiomics features, as a novel analytic method, has been increasingly applied in variety cancer research and may be a potential predictor for preoperatively differentiating pathological grades of adenocarcinoma. Methods: Patients (pts) with radiological proved as solitary ground glass nodule were eligible in this study. Radiomics features derived from computed tomography (CT) images were extracted by Chinese Academy of Science. All pts will be categorized into three groups with lepidic predominance as low-grade, acinar and papillary predominance as intermediate-grade, micropapillary and solid predominance as high-grade. We used L1 penalized constrained continuation ratio model to select relevant radiomics features, and corresponding radiomics signature was constructed. Association between the radiomics signature and pathological grades of adenocarcinoma was explored using the Kruskal-Wallis test and C-index was performed to test the efficacy of differentiating. Results: 82 pts were included in this study. Low-grade, intermediate-grade and high-grade contained 15 (18.3%), 53 (64.6%), 14 (17.1%) pts respectively. 475 radiomics features were extracted from thin section CT image and 10 of them selected through L1 penalized constrained continuation ratio model composed radiomics signature which significantly associated with pathological grades (P &lt; 0.0001). C-index for radiomics signature were 0.813 (95%CI 0.793-0.833). Since clinical characters including gender, age, smoking status, NSE, CEA and CYFRA21-1 were not associated with different grades of adenocarcinoma, we could not establish nomogram based on the radiomics signature and correlated clinical characters. Conclusions: Radiomics features only can be a potential predictor for preoperatively differentiating pathological grades of adenocarcinoma, which may be a more applicable clinical predictor for patients’ survival. Yet large sample sizes are warranted to confirm the results.

A Comparison of Logistic Regression Models for DIF Detection in Polytomous Items: The Effect of Small Sample Sizes and Non-Normality of Ability Distributions

This study investigated the effectiveness of logistic regression models to detect uniform and non-uniform DIF in polytomous items across small sample sizes and non-normality of ability distributions. A simulation study was used to compare three logistic regression models, which were the cumulative logits model, the continuation ratio model, and the adjacent categories model. The results revealed that logistic regression was a powerful method to detect DIF in polytomous items, but not useful to distinguish the type of DIF. Continuation ratio model worked best to detect uniform DIF, but the cumulative logits model gave more acceptable type I error results. As sample size increased, type I errors increased at cumulative logits model results. Skewness of ability distributions reduced power of logistic regression to detect non-uniform DIF. Small sample sizes reduced power of logistic regression.

Parametric Dose Standardization for Optimizing Two-Agent Combinations in a Phase I-II Trial with Ordinal Outcomes.

A Bayesian model and design are described for a phase I-II trial to jointly optimise the doses of a targeted agent and a chemotherapy agent for solid tumors. A challenge in designing the trial was that both the efficacy and toxicity outcomes were defined as four-level ordinal variables. To reflect possibly complex joint effects of the two doses on each of the two outcomes, for each marginal distribution a generalised continuation ratio model was assumed, with each agent's dose parametrically standardised in the linear term. A copula was assumed to obtain a bivariate distribution. Elicited outcome probabilities were used to construct a prior, with variances calibrated to obtain small prior effective sample size. Elicited numerical utilities of the 16 elementary outcomes were used to compute posterior mean utilities as criteria for selecting dose pairs, with adaptive randomisation to reduce the risk of getting stuck at a suboptimal pair. A simulation study showed that parametric dose standardisation with additive dose effects provides a robust, reliable model for dose pair optimisation in this setting, and it compares favourably with designs based on alternative models that include dose-dose interaction terms. The proposed model and method are applicable generally to other clinical trial settings with similar dose and outcome structures.

Developing Empirically Informed Policies for Sexual Assault Kit DNA Testing: Is It Too Late to Test Kits Beyond the Statute of Limitations?

A growing body of research indicates that there are thousands of sexual assault kits (SAKs) in police property storage facilities that have never been submitted for DNA forensic testing. Some of these rape kits may be quite dated, and the statute of limitations (SOL) for prosecution of the case may have expired. Whether testing such kits could still provide useful information for criminal justice system personnel is unknown. To address this gap in the literature and to inform policy regarding rape kit testing, we randomly sampled 700 previously untested SAKs from Detroit, MI: 350 were presumed to be beyond the SOL for prosecution (based on the date the SAK was collected), and 350 were still within the SOL. All SAKs were submitted for DNA testing, and then we quantified and compared the forensic testing outcomes. At issue was whether these older SAKs would yield DNA profiles that were eligible for entry into Combined DNA Index System (CODIS), the federal DNA forensic database, and whether these profiles would match (“hit”) to other criminal offenses catalogued in CODIS. Rates for presumed SOL-expired SAKs and unexpired SAKs were compared via a continuation-ratio model and equivalence tests. The rates of CODIS-eligible DNA profiles, CODIS hits, and serial sexual assault CODIS hits were statistically equivalent in the SOL-expired and SOL-unexpired groups. Testing older SAKs has potential utility to the criminal justice system because these kits produced DNA matches to other crimes, including other sexual assault crimes, at a rate equivalent to current, SOL-unexpired SAKs.